Handbook of Defense Economics

INTRODUCTION TO THE SERIES

The aim of the Handbooks in Economics series is to produce Handbooks for various branches of economics, each of which is a definitive source, reference, and teaching supplement for use by professional researchers and advanced graduate students. Each Handbook provides self-contained surveys of the current state of a branch of economics in the form of chapters prepared by leading specialists on various aspects of this branch of economics. These surveys summarize not only received results but also newer developments, from recent journal articles and discussion papers. Some original material is also included, but the main goal is to provide comprehensive and accessible surveys. The Handbooks are intended to provide not only useful reference volumes for professional collections but also possible supplementary readings for advanced courses for graduate students in economics. KENNETH J. ARROW and MICHAEL D. INTRILIGATOR

ACKNOWLEDGEMENTS

We are indebted to the many colleagues who have directly and indirectly contributed to this Handbook. Michael Intriligator was centrally involved in initiating the volume. Our distinguished contributors responded willingly and enthusiastically to our various comments and suggestions and were a pleasure to work with. Colleagues whom we have worked with on a variety of defense economics issues include Jon Cauley, John A.C. Conybeare, Walter Enders, John E Forbes, Nick Hooper, Harvey Lapan, Dwight Lee, Stephen Martin and James Murdoch. Typing and correcting manuscripts requires care and patience. Roberta Blackburn, Margaret Cafferky, Anne Hrbek, Eileen Mericle and Sue Streeter excelled themselves in producing the final version of the manuscript. Our thanks also to the staff at North-Holland. At a personal level, we owe much to the support of our wives (Winifred and Jeannie) and our children (Cecilia, Lucy, Adam and Tristan). Keith Hartley's research was supported in part by the Economic and Social Research Council and the Leverhulme Trust. York, England Ames, Iowa May 1995

Keith Hartley Todd Sandler

Chapter I

INTRODUCTION KEITH HARTLEY University of York TODD SANDLER Iowa State University

Contents Abstract Keywords 1. Introduction 2. A brief history of the field 3. On the nature of defense economics 4. Importance of defense economics today 5. Organization of the book References

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

2 2 3 4 6 7 8 11

2

K. Hartley and T7Sandler

Abstract This chapter defines defense economics and indicates how it differs from other subfields of economics. The nature and topics of defense economics are presented. A brief history of the field is also reviewed. The current importance of defense economics in the post-cold war era is discussed. Finally, the organization of the book and its component chapters are presented.

Keywords alliance burden sharing, arms races, arms trade, contract design, conversion, disarmament, discretionary budget, econometrics, externality, game theory, history of defense economics, industrial policies, insurrections, military manpower, Nash equilibrium, procurement, public choice, public goods

Ch. 1:

Introduction

3

1. Introduction Defense economics applies the tools of economics to the study of defense and defenserelated issues, including disarmament and peace. When applying these tools, defense economists must tailor economic methods, both theoretical and empirical, to defense issues and policies, while taking account of institutional aspects that characterize the defense sector. For example, the study of procurement practices in the United States must include the relevant participants - the Congress, the defense contractors, and the military - and how they interact and sequence their actions in making equipment choices. Relevant defense issues include an analysis of alliance burden sharing, the impact of contract design on the efficiency of procurement, the study of arms races and stability, the effects of defense expenditure on economic growth and development, the allocative implications of disarmament treaties, the impact of industrial policies on the defense sector, the employment implications of conversion, the study of insurrections and conflict, and policies to regulate the arms trade. The field of defense economics is broad in its scope and intersects with a variety of economic subfields, including public economics, public choice, microeconomics and macroeconomics, regional economics, and international economics. In recent years, defense economics has grown in importance, as documented by the large number of articles in general journals, the appearance of numerous books, and the publication of a field journal, Defence and Peace Economics (formerly Defence Economics). Contributions in defense economics have made fundamental advances to game theory [e.g., Thomas Schelling (1960), The Strategy of Conflict], procurement theory [Cummins (1977), Laffont and Tirole (1993), McAfee and McMillan (1986), Tirole (1986)], and the econometrics of military manpower [Ash, Udis and McNown (1983)]. The study of defense economics is of importance to a large audience that includes researchers, teachers, practitioners, taxpayers and students, who are concerned with allocative, distributional, dynamic, and stabilization influences of defense expenditures and policies. Although the share of gross domestic product (GDP) devoted to defense varies greatly among nations, in 1990 and on average nations devoted approximately five percent of GDP to defense [Hartley et al. (1993, p. 17)]. However, between 1989 and the early 1990s, world military expenditure declined [Deger (1993)]. In many countries, such as the United States, defense expenditure is a sizable share of the discretionary budget of the central government. As a consequence, the study of resource allocation and economic policy is incomplete unless defense issues are analyzed. The Handbook of Defense Economics is intended to take stock of the myriad contributions of the last three and half decades. To accomplish this purpose, we have assembled many of the leading scholars who have authored some of the seminal pieces in the field. The nineteen chapters of the Handbook present the fundamental concepts and analyses in a form that should prove useful to researchers and students. Thus, authors have presented key tools and analyses of the field in a way that emphasizes intuition and significance. Theoretical analysis, econometric techniques, and policy

4

K. Hartley and T Sandler

issues are addressed. The chapters fall into two essential categories: surveys (chapters by Smith, Murdoch, Brito and Intriligator, Enders and Sandler, Ram, Deger and Sen, Rogerson, Warner and Asch, Lichtenberg, Hartley, and Anderton) and conceptual studies (McGuire, Brzoska, Hirshleifer, Grossman, Dunne, Braddon, and Fontanel). Survey chapters present a synthesis, interpretation, and evaluation of the literature for particular subfields of defense economics, whereas the conceptual chapters elucidate the analysis of specific topics. Both types of chapters provide directions for future research. Novel ideas and insights are contained throughout the volume, insofar as authors use their chapters to give an overview and to push the analysis forward. We have provided editorial guidance so that authors wrote chapters that present an upto-date statement of a line of inquiry, while emphasizing some pedagogical tools. As with earlier Handbooks, we intend that the chapters serve researchers and find their way onto student reading lists.

2. A brief history of the field Significant interest in defense economics as an emerging field coincided with the appearance of three now-classic works in 1960: Charles J. Hitch and Roland McKean (1960), The Economics of Defense in the Nuclear Age; Lewis F Richardson (1960), Arms and Insecurity; and Thomas C. Schelling (1960), The Strategy of Conflict. The pioneering book by Hitch and McKean applied economic concepts of allocative efficiency to the defense sector and inspired researchers for the next generation to investigate defense economic issues. The interest in defense economics blossomed during the Cold War, sparked, in part, by Richardson's (1960) arms race model. A large literature grew that applied economic analysis to the study of arms races and their stability (see the Brito and Intriligator chapter). This theoretical literature was accompanied by a parallel literature that attempted to document the empirical existence of arms races [see, e.g., McGuire (1977) and Linden (1991)]. Finally, Schelling (1960) introduced game-theoretic notions to the study of conflict and defense. The book served to clarify the meaning of deterrence, compellent threats and promises, and strategic moves. Although Schelling's work did not emphasize formal modelling, many of his ideas have been subsequently formalized with recent advances in game theory - e.g., the notion of a perfect equilibrium in which noncredible threats are culled from feasible equilibria. There were three further pioneering contributions in the 1960s. A US study by Peck and Scherer (1962) on The Weapons Acquisition Process analyzed the nonmarket characteristics of weapons acquisition, the nature of competition for defense contracts, and the results of weapons programs. A companion volume by Scherer appeared in 1964: The Weapons Acquisition Process:Economic Incentives. This volume examined the competitive and contractual incentives in US weapons acquisition programs, including a theoretical and empirical analysis of different types of contracts (e.g., fixed price; cost plus). In 1966, Olson and Zeckhauser applied the theory of public

Ch. 1: Introduction

5

goods to the study of burden sharing within military alliances. Subsequent papers analyzed burden sharing, allocative efficiency, and the allies' demand for defense expenditure [see the Murdoch chapter and Sandler (1993)]. During the 1970s, the literature of defense economics emphasized at least five issues or subfields. First, arms races received much attention with important pieces by Brito (1972) and Intriligator (1975) that provided a strategic foundation to the Richardsonian arms race equations. Second, the appearance of Benoit's (1973) book began a controversy over whether military expenditure is growth promoting or growth inhibiting. This controversy continues to the present day (see the chapters by Ram, and Deger and Sen). Third, economists devoted attention to the economics of alternative systems of military recruitment: conscription versus the all-volunteer army (see the Warner and Asch chapter). Fourth, defense economists investigated the defense industries in terms of profitability, procurement practices, conversion impacts, competition, and industrial policies (see the chapters by Dunne, Hartley, Lichtenberg, and Rogerson). Fifth, regional impacts of defense expenditure were studied with a variety of tools (see the Braddon chapter). In the 1980s and 1990s, these earlier topics were pursued along with some new ones. Advances in the study of asymmetric information and game theory were applied to the study of incentive contracts and procurement practices, as shown in the Rogerson chapter. Game theory was also used in the study of terrorism and the analysis of negotiation strategy between terrorists and the authorities (see the Enders and Sandler chapter). Modern general equilibrium analysis was employed to examine insurrections and appropriative behavior as highlighted by the Grossman chapter. Game-theoretic concepts and insights drawn from rent-seeking analysis were used by Jack Hirshleifer to theorize about the notion of conflict. Within the last fifteen years, other topics of defense economics included an analysis of defense R&D (the Lichtenberg chapter), arms trade (the Anderton chapter), disarmament (the Fontanel chapter), and conversion (the Fontanel and Hartley chapters). The modem study of defense economics employs the latest theoretical and empirical tools. Game-theoretic concepts are used in a number of subfields of defense economics, including the study of alliances, arms races, incentive contracting, terrorism, insurrections, and conflict. A crucial game-theoretic notion of equilibrium, which shows up in a number of places in the volume, is that of a Nash equilibrium. If no player or agent would unilaterally want to change his or her strategic choice, then a Nash equilibrium is achieved. At such an equilibrium, each player is optimizing his or her choice variable(s) subject to the best-response choices of his or her counterpart(s). In dynamic settings where interactions are repeated, a multiplicity of Nash equilibria exist, some of which rely on noncredible threats that may hurt the threateners sufficiently that they are unwilling to carry them out. Refinements to the Nash equilibrium concepts have been developed to remove such noncredible threats and to maintain a greater degree of rationality on behalf of agents. Subgame perfection is but one of many refinements. An equilibrium is subgame perfect if it contains a Nash equilibrium for all component subgames, so that no player would unilaterally

6

K. Hartley and T Sandler

want to change his or her strategy at any point in the game. Current analysis of defense economics employs principal-agent analysis, rational expectations, dynamic optimization, and other theoretical advances. Modern defense economics also uses the latest refinements in econometrics. In recent years, time-series analysis and its refinements have figured prominently in defense economics. For example, the technique of vector autoregressive analysis (VAR), whereby the interrelationships of multiple time series are studied, has been used to analyze terrorism and the impact of defense expenditure on growth and employment. Other time-series techniques - causality tests, intervention analysis, and cointegration tests - have been used throughout defense economics. Military manpower studies have applied the latest techniques of regression analysis to investigate recruitment, retention, military labor supply, and military labor demand. Time-series cross-sectional methods have been recently applied to the growth and defense controversy [Macnair et al. (1995)]. Some of these techniques are more advanced than the data available for satisfactory empirical work. Often sophisticated econometric techniques use available statistics, failing to recognize the limitations of the data arising from its aggregative nature and the use of different definitions of military expenditure (see the Brzoska chapter).

3. On the nature of defense economics Defense economics is the study of resource allocation, income distribution, economic growth, and stabilization applied to defense-related topics. As such, defense economics involves an investigation of the impact of defense expenditures, both domestically and internationally, on macroeconomic variables such as employment, output, and growth. It also has a microeconomic dimension involving analysis of the defense industrial base, collaborative programs, offsets, the pricing and profitability of military contracts, and the regulation of contractors. Defense economics draws from a variety of economic subfields. In particular, public economics is important, because the provision of defense (security) can be viewed as a public good, whose benefits are nonrival and nonexcludable within a nation and among allies. Another market failure, germane to defense economics, is that of an externality, which arises when the action of one agent influences the well-being of another and no means of compensation exists. For defense economics, the arming of one nation may yield positive (on allies) or negative (on adversaries) externalities on another nation. However, measuring the "output" of defense spending is fraught with difficulties and the problem is often ignored, or assumed away, or presented in generalizations such as security, protection, and deterrence. Industrial organization is another branch of economics that figures crucially in the study of the defense industrial base (the Dunne chapter), procurement (the Rogerson chapter), defense R&D (the Lichtenberg chapter), and industrial policies such as collaboration, licensed production and offsets (the Hartley chapter). Other relevant

Ch. 1:

Introduction

7

subfields include labor economics, international economics, public choice, economic growth, and macroeconomics. Public choice is particularly appropriate to defense economics, because it focuses on nonmarket decision-making. By modelling the behavior of voters, political parties, governments, bureaucracies, and other interest groups, it provides a framework for analyzing the "military-industrial complex". Defense economics encompasses aspects and topics from peace science and conflict studies. Thus, the economics of disarmament and conversion (the Fontanel and Hartley chapters) are relevant topics. There is no ideological presumption in defense economics; that is, defense economists are not inclined towards military expenditures and armed conflict. They are, instead, concerned with understanding the processes and dynamics of arms expenditures, conflict and its resolution, and any economic aspects associated with the defense sector. Defense economics has a strong policy orientation; analysis is often undertaken to make policy recommendations regarding publicly financed defense outlays. And, in the last resort, economists cannot ignore the opportunities for applying economic theory and empirical techniques to a sector which is a major user of scarce resources and which has the potential to protect or destroy civilization. A further discussion and analysis of the nature and scope of defense economics is taken up in the McGuire chapter. In recent years, the scope of defense economics has expanded to consider a broader range of security matters that include nonmilitary issues such as the protection of the environment from transnational pollution (e.g., ozone shield depletion, global warming, acid rain). We have taken a narrower and more traditional view of defense economics so as to preserve the field's well-defined identity.

4. Importance of defense economics today Even though the Cold War has ended and the superpower confrontation has declined greatly, defense economics is still of great importance. In fact, the events of the late 1980s and early 1990s may have increased the need and importance of defense economics, especially in regards to resource allocation and the application of economics methods to security-related issues. First, the downsizing of military budgets in the first half of the 1990s highlights a host of economic challenges. Falling budgets and rising equipment costs will force nations to seek efficiency improvements in the acquisition of weapons and the provision of armed forces. Nations will be more willing to buy available equipment off-theshelf from overseas suppliers rather than paying the price of supporting a national defense industrial base. Within the armed forces, there will be pressures to substitute, say, equipment for manpower, reserves for regulars, women for men, and civilians for military personnel (e.g., via contractorizing activities traditionally undertaken "inhouse" by the military, such as catering, repair, and maintenance). Unit costs may also be higher on new weapon systems when scale economies are present unless an

8

K. Hartley and T: Sandler

arms producer is able to resort to foreign sales, which may imply other kinds of risks and costs to the supplier nation. Faced with procurement cuts, defense firms have merged to take advantage of economies of scale and economies of scope. In so doing, national defense industries have become more concentrated and less competitive. This downsizing trend may have bottomed out. In 1994, the Clinton administration gave an emergency allocation to the military to meet readiness requirements. The Anderton chapter mentions that the arms trade may be on the rise again. Second, the breakup of the Soviet Union and Eastern Europe has unleashed ethnic conflicts that have erupted into civil war once central power was diminished. In addition, the breakup of the Soviet Union has raised the risk of nuclear weapons getting into the hands of terrorists and the proliferation of nuclear-weapon nations. Third, since these trouble spots can create conflict externalities for neighboring states, the role of peacekeeping forces and their financial support have grown in importance. Fourth, the Gulf War of 1991 points to a source of conflict in the future - that is, wars fought over disputed and/or scarce resources, such as common oil pools. Fifth, recent defense treaties raise a host of allocative concerns as the elimination of weapons creates expenditures on their disposals, environmental cleanup, verification, and the development of alternative classes of weapons. Peace as well as confrontation have their costs. Recent reallocations of resources have had regional and national impacts on employment and output. Sixth, nonconventional conflict in the form of terrorism and insurrections presents exigencies that have allocative and distributional concerns. The provision of defense is still an important activity that requires huge resource allocations to meet a variety of contingencies and uncertainties. Economic aspects of the defense sector continue to be important and to require study. Interest continues to grow in the economics field for applying economic methods to defense issues.

5. Organization of the book The book contains eighteen additional chapters that include all essential subfields of defense economics. Coverage varies among chapters owing to the background literature: some areas (e.g., arms races) have a large literature, while others (insurrections) have a much smaller literature. Chapters 2 (M.C. McGuire) and 3 (M. Brzoska) set the stage for the study of defense economics. McGuire introduces the notion of defense economics, its origin and meaning. In the chapter by Brzoska, the reader learns about the various data sets available to study world military expenditure and their limitations. The next three chapters concern the demand and supply of military expenditure within the nation, among allies, and between adversaries. Chapter 4 (R. Smith) examines the theoretical and empirical issues surrounding the estimation of a nation's demand for military expenditure. In chapter 5 (J.C. Murdoch), the economic theory of alliances is presented, based on the theory of pure and impure public goods. Empirical tests are also reviewed. Chapter 6 (D.L. Brito and M.D. Intriligator) focuses on arms races, beginning with Richardson's (1960) formulation. Dynamic methods and optimization

Ch. 1: Introduction

9

techniques are used to display a variety of models. This chapter also applies the models and their recent extensions to study nuclear weapon proliferation and other relevant policy issues. Using modem tools from public economics, J. Hirshleifer (chapter 7) presents some theoretical representations to analyze conflict among agents. The modem theories of rent seeking and tournaments are behind some of the models in this provocative presentation. Chapter 8 (H.I. Grossman) focuses on appropriative behavior, whereby a nation's ruler must allocate resources to maintain his or her ability to live off of the population, while elements within the population attempt to capture the ruler's accumulated assets. Tools of general equilibrium are employed. Grossman clearly distinguishes his theoretical paradigms from other treatments of rebellion. In chapter 9, W. Enders and T. Sandler present rational-actor models of terrorist behavior based on choice-theoretic and game-theoretic approaches. A host of empirical techniques are displayed for testing the effectiveness of anti-terrorist policies. Chapters 10 (R. Ram) and 11 (S. Deger and S. Sen) are devoted to investigating the impact of defense expenditure on growth and development, and are motivated, in part, by Benoit's (1973) surprising finding that defense was growth-enhancing in some developing countries. In the chapter by Prof. Ram, theoretical and empirical aspects of the Feder-Ram model for analyzing the influence of defense on growth are discussed. The Feder-Ram model has been applied to developed and underdeveloped countries. In the subsequent chapter, Drs. Deger and Sen review the issues surrounding the impacts of defense on developing countries. An emphasis is given to displaying empirical studies of this impact. Defense inputs (capital, labor, and R&D), the defense industry, industrial policies, and the regional impacts of defense spending are the subjects of the chapters 12-17. These chapters are devoted to both demand- and supply-side factors. Modern tools of microeconomic analysis are applied by W.P. Rogerson (chapter 12) to study the incentive aspects of defense procurement and R&D. This chapter stresses principalagent analysis and game theory, while also reviewing some empirical evidence. Military manpower issues - recruitment, enlistment, retention, the draft versus the all-volunteer army, and other labor supply and demand concerns - are analyzed by J.T. Warner and B. Asch in chapter 13. Theoretical and empirical issues are presented in their survey, but, in keeping with the literature, the emphasis is on empirical techniques and analysis. This study is followed chapter 14 (P. Dunne) on the defense industrial base, including discussions of defense equipment markets, the competitiveness of defense industries, and cost considerations in weapons production. A portion of this chapter is concerned with reviewing the definitions of the defense industrial base and military industrial complex.

The Feder-Ram model is a supply-side model based on the analysis in Feder (1983), Ram (1986), and Biswas and Ram (1986).

10

K. Hartley and T Sandler

Prof. Lichtenberg (chapter 15) analyzes a host of issues surrounding the economics of defense R&D, including mechanisms designed to encourage firms to pursue defense R&D, the private and social returns from military R&D, and the ability of defense contractors to shift R&D costs from commercial to defense R&D. Social returns to defense R&D are investigated with a production function/productivity growth framework. Chapter 16 (K. Hartley) carries the analysis of the defense industry further by reviewing industrial policies used in Europe and the United States. Policy options include the use of international collaboration involving the joint development and production of weapons, the implementation of licensed production and co-production between two or more nations, the encouragement of trade flows in weapons sales among allies, and the reliance on offset agreements for purchasing governments to recover some of the purchase price. Conversion issues are also considered as part of industrial adjustment policies. In chapter 17, Dr. Braddon then investigates the regional impact of the defense industries and sector on employment, output, and growth. This chapter presents alternative theoretical tools (i.e., input-output analysis, regional multiplier techniques) and econometric models for identifying regional impacts. Selected case studies are also reviewed. The final two chapters are devoted to peace science aspects of defense economics. Prof. Anderton (chapter 18) surveys the cumulative work on the economics of arms trade and concludes that this subfield is "ripe for foundational contributions" that develop a theoretical underpinning for arms trade analysis. Anderton furthers this goal by presenting some theoretical representations concerning the economic aspects of arms trade. This chapter also examines whether arms sales are peacepromoting or peace-inhibiting. Furthermore, recent trends in arms trade are indicated. In chapter 19 (J. Fontanel), both the theoretical and empirical economic consequences of disarmament are examined. The influences of disarmament on stability, weapon proliferation, and economic development are also investigated. The chapter concludes by reviewing public policies for conversion. It is our intention that this Handbook will serve as a reference source, a teaching tool, and a stimulant for future research. Certainly, there is no shortage of research opportunities embracing theoretical, empirical, and policy issues. For example, there is a need to monitor the impacts of disarmament at the national, local, and industrial levels and to study conversion to identify the features of successful and failed conversions. Uncertainty means that economists need to focus on the costs and benefits of alternative methods of reconstituting armed forces and national defense industries to meet future emergencies and conflicts (e.g., via reserve forces; dual-use technologies; technology demonstrators). The efficiency with which the military uses its resources remains a vastly under-researched field embracing such issues as budgeting, substitution possibilities, employment contracts, incentives, and performance indicators. Finally, efforts to achieve and maintain a more peaceful world are likely to focus on the United Nations as a peace-making and peace-keeping organization with opportunities for applying economic models of alliances and burden sharing.

Ch. 1:

Introduction

11

References Ash, C., B. Udis and R.F. McNown, 1983, Enlistments in the all-volunteer force: A military personnel supply model and its forecasts, American Economic Review 73, 144-155. Benoit, E., 1973, Defense and economic growth in developing countries (D.C. Heath, Boston, MA). Biswas, B., and R. Ram, 1986, Military expenditures and economic growth in less developed countries: An augmented model and further evidence, Economic Development and Cultural Change 34, 361-372. Brito, D.L., 1972, A dynamic model of an armament race, International Economic Review 13, 359-375. Cummins, J.M., 1977, Incentives contracting for national defense: A problem of optimal risk sharing, Bell Journal of Economics 8, 168-185. Deger, S., 1993, World military expenditure, in: World armaments and disarmament: Stockholm International Peace Research Institute (SIPRI) yearbook 1993 (SIPRI, Stockholm) 337 397. Feder, G., 1983, On exports and economic growth, Journal of Development Economics 12, 59 73. Hartley, K., et al., 1993, Economic aspects of disarmament: Disarmament as an investment process (United Nations, New York). Hitch, C.J., and R.N. McKean, 1960, The economics of defense in the nuclear age (Harvard University Press, Cambridge, MA). Intriligator, M.D., 1975, Strategic considerations in the Richardson model of arms races, Journal of Political Economy 83, 339-353. Laffont, J.-J., and J. Tirole, 1993, A theory of incentives in procurement and regulation (MIT Press, Cambridge, MA). Linden, M., 1991, The dynamics and the instability of the Middle East military expenditures in years 1955 1984, Defence Economics 2, 199-208. Macnair, E.S., J.C. Murdoch, C.-R. Pi and T. Sandler, 1995, Growth and defense: Pooled estimates for the NATO alliance, 1951-1988, Southern Economic Journal 61, 846-860. McAfee, R.P., and J. McMillan, 1986, Incentives in government contracting (University of Toronto Press, Toronto). McGuire, M.C., 1977, A quantitative study of the strategic arms race in the missile age, Review of Economics and Statistics 59, 328-339. Olson, M., and R. Zeckhauser, 1966, An economic theory of alliances, Review of Economics and Statistics 48, 266-279. Peck, M.J., and F.M. Scherer, 1962, The weapons acquisition process (Harvard University Press, Cambridge, MA). Ram, R., 1986, Government size and economic growth: A new framework and some evidence from cross-section and time-series data, American Economic Review 76, 191-203. Richardson, L.F., 1960, Arms and insecurity: A mathematical study of the causes and origins of war (Homewood, Pittsburgh, PA). Sandler, T., 1993, The economic theory of alliances: A survey, Journal of Conflict Resolution 37, 446-483. Schelling, T.C., 1960, The strategy of conflict (Harvard University Press, Cambridge, MA). Scherer, F.M., 1964, The weapons acquisition process: Economic incentives (Harvard University Press, Cambridge, MA). Tirole, J., 1986, Procurement and renegotiation, Journal of Political Economy 94, 235-259.

Chapter 2

DEFENSE ECONOMICS AND INTERNATIONAL SECURITY MARTIN C. McGUIRE* University of California, Irvine

Contents Abstract Keywords 1. Introduction 1.1. What is defense economics? 1.2. Some examples where economics has made central contributions to security studies 1.2.1. Defense strategy and resource allocations 1.2.2. Analysis of deterrence 1.2.3. Economic models of alliances 1.2.4. National power, economic survival, and international trade 1.2.5. Arms races and strategic interactions 1.2.6. Economic ecology and international conflict

2. Interactions between economics and defense 2.1. Levels of interaction 2.2. Historical evolution

3. Themes in the history of Defense Economics 3.1. World War II 3.2. Early Cold War 3.3. Later Cold War 3.3.1. Alliances, burden sharing, and erosion of US hegemony 3.3.2. Long run sustainability of defense and relative decline of the US 3.3.3. Economic warfare/defense: a game of economic ruin 3.3.4. Economic equity and defense: military manpower and conscription 3.3.5. Defense industry analysis, acquisition and contracting 3.4. Post Cold War 3.4.1. The emergence of positive theories of defense 3.4.2. New directions in the normative analysis of defense

4. Agenda for Defense Economics 4.1. Continuing/inherited policy concerns 4.2. Normative issues arising from world economic change and the collapse of Communism

* The author thanks the Heinz Chair Endowment for research support. Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T7 Sandler © 1995 Elsevier Science B. V All rights reserved

14 14 15 15

16 17 17 18

18 19

20

20 20 21

21 21 23 24 25 26 27 28 28 29 29 30

31 32 34

M.C. McGuire

14 4.2.1. Economics of transition 4.2.2. New definition of national/international security 4.2.3. New economic instruments for security 4.2.4. New concepts of alliances and replacement for bi-polar world hegemony 4.3. Positive analysis of international power equilibria

5. Concluding remark References

34 35 35 36 36

39 39

Abstract Defense economics derives from and is embedded in the multi-dimensional array of issues each country must address when providing for its national security. Applying economic concepts and methods, it attempts to evaluate this great diversity of security related questions, and to understand how each country's security interacts and fits in with the security of all nations in the international system. Included in Defense Economics are such overarching questions as: definition of what security actually is; how resource scarcity, distribution, and stage of economic development influences the security obtainable by each nation in the international system; relationships between defense sectors and national economies within and across countries; efficiency in provision of security; incentive structures which promote or resolve conflict; institutional arrangements which promote or retard peace, stability, and equity.

Keywords alliances, arms races, arms trade, conflict resolution, defense budgets, deterrence, disarmament, economic development, economic warfare, foreign aid, international conflict, national security, military strategy, nuclear proliferation, peace, war

Ch. 2: Defense Economics and InternationalSecurity

15

"National Security, from the point of view of an economist, may be said to depend on three things: (1) the quantity of national resources available, now and in the future; (2) the proportion of these resources allocated to national security purposes; and (3) the efficiency with which the [Hitch and McKean (1960, p. 4)]. resources so allocated are used."

1. Introduction 1.1. What is defense economics? The study of war, in all its diverse aspects - avoidance, causes, preparations, initiation, conduct, termination, consequences, etc. - has engaged scholarly energies from every conceivable origin and discipline. Where does economics - economics as concerned with the consequences of resource scarcity, and the necessity to allocate among alternatives - fit in with philosophy, history, biology, psychology, law, political science, mathematics, religion and so on? The family of economic problems studied over the past half-century in the cause of defense and security is indeed dauntingly vast. A partial list of these would include the items in Table 1. Table 1 Themes in economics of defense and international security Deterrence, war avoidance, war initiation and termination Strategic interactions, arms races, arms control Alliance formation, resource allocation, and behavior Defense macro-economy interactions during war, peace, disarmament, and conversion Command vs market economies as resources for defense Mobilization, war recovery, and reconstitution of forces Optimization and efficiency in force level and composition Capital-labor utilization: manpower, conscription, and volunteerism Military readiness, strategic materials, and defense industry policy Procurement, acquisition, and defense contract analysis Trade dependence, export control, trade sanctions, and economic warfare Foreign aid, developing economies, and defense in the third world International management of nuclear materials, non-proliferation Extragovernmental threats, terrorism, drugs, refugees, ethnic and religious fanaticisms

Although lengthy, this list provides little clue as to the structure or evolution of "Defense Economics". Some readers approaching this volume might have thought the term meant the study of how to provide and manage military forces efficiently, but it clearly reaches far beyond that important subject. Others, while admitting a larger set of topics, may see them as a disjointed set of issues somehow relating to the military

16

M.C. McGuire

affairs of a country which happen to have proven of interest to economists. Still other readers may construe the "Defense" in "Defense Economics" exclusively as defense of the West. But that perspective has been overtaken by world events and can no longer be sustained. For purposes of this essay "Defense" must include the many diverse protective and aggressive activities undertaken by nations and their governments to define, advance, and preserve their interests among the tribe of nations. So economics of defense must include the effects of resource scarcity both as placing limits upon such aggressive or protective activities, and as giving rise to them at the same time. Moreover, because nations function within an international system of other states, and numerous other non-governmental actors, the economics of defense must incorporate interactions among the various nation-states all seeking security in the larger regional and international economies of which they form a part. In short, Defense Economics must recognize the impact defense activities impose upon the larger international economy, and the effect of that economy-system upon defense. This is a demanding program, the more so given that problems of defense/security have mutated continuously over recent generations to a degree that in 1995 the meaning of national, international, or global security seems elusive, and the question "defense of whom from whom and what?" more germane. Clearly to produce some integrated idea of what Defense Economics is, requires more than a simple enumeration of topics. 1.2. Some examples where economics has made central contributions to security studies For some of the topics in Table 1, economists have provided the central clues which have defined the problem. In other instances the security problem has served as a paradigm context for application of previously known principles. And in still other cases consistent application of economic principles has produced unexpected insight. As examples, I list six cases below, in which economic insights have been crucial in understanding a security or defense phenomenon.

For each of these topics the economic way of thinking has proven highly useful and has engaged economists over prolonged periods. Nevertheless, as pointed out by Intriligator (1990) "Defense Economics" or "Economics of International Security" has been slow to develop into a recognized field supported by graduate and undergraduate course curricula, research departments, and specialty research journals. Why this is so may be due to the fact that economists must share the study of Defense Economics with other fields, and may not have a decisive voice in policy determination. Because the rapidly evolving technology is crucial to a country's defense effectiveness, and because political, social, and psychological factors can be so central in defense studies, economists necessarily only contribute a fraction to ongoing policy and academic debates. Another reason may have been a normative quality which has informed defense studies producing political and ideological distinctions among economists. But this normative aspect has diminished rather steadily with the passage of the decades.

Ch. 2:

Defense Economics and InternationalSecurity

17

1.2.1. Defense strategy and resource allocations Application of widely accepted notions of efficiency in resource allocation to problems of national security is perhaps the most widely held and minimalist definition of Defense Economics. The concept of government as a producer of quantifiable (if difficult to measure) outputs which use scarce inputs available at various prices, at one time revolutionary, has become commonplace. The idea of quantifying benefits, valuing resource inputs, incorporating intangibles, and therefore determining cost efficient defenses grew out of earlier operations research efforts - for example in submarine search programs, or bomber target selection criteria [US Strategic Bombing Survey (1945-46), Hitch (1958)]. Subsequently, with the rise of systems analysis in defense, economists would come to argue that "strategy, technology, and economy are not three independent "considerations" ... but interdependent elements of the same problem. Strategies are ways of using budgets or resources to achieve military objectives. Technology defines the possible strategies ... The economic problem is to choose that strategy which is most efficient ... or economical." [Hitch and McKean (1960, p. 3)] Similarly, economics identified the public good nature of defense and, therefore, the requirement for a collective logic of efficient overall provision, which demands that the sum of marginal benefits equal marginal costs at the true optimal amount of defense. Thus defining "MB" as the marginal benefit of defense and "MC" as the marginal cost the formula E MB = MC - in principle - solves the elusive problem of commensurability between dollars and security 2 . 1.2.2. Analysis of deterrence Another important application of the economic way of thinking to defense arose in developing an understanding of deterrence. Two economists stand out for first uncovering how deterrence is a matter of incentives parties impose upon each other by conditioning each others' expectations. These were Schelling and Ellsberg. "Deterrence", as Schelling (1960) explained "... is concerned with influencing the choices that another party will make ... by influencing his expectations of how we will behave. It involves confronting him with evidence for believing that our behavior will be determined by his behavior". Before the explosion of game theory into economics at all levels, economists introduced the crucial importance of credibility of a deterrent threat as captured in forward looking anticipation of future incentives 3 and measured by a critical risk [Ellsberg (1956)]. To deter an opponent both effectiveness of retaliation and credibility that it will be employed are necessary. The idea of critical The summation is taken over all individuals within "our" group whether they benefit from defense or not, but excludes those outside "our" group who disbenefit, such as the enemy. 3 This work anticipated the forward looking backward induction repeated move equilibria now so commonplace in game theoretic concepts of equilibrium. 2

18

M.C McGuire

risk can be illustrated using V, to denote the utility of country i, under situation m,n. If both countries i and j refrain from attack then m =NA, n=NA. But if i attacks (m = A) then j may retaliate (n = R) with probability p, or not retaliate (n = NR) with probability (1 -p). Country j's objective should be to make the risk p to i that it, j, will retaliate so great that: VNA,NA >pV,R + (1 -p)

VA,NR

(1)

Ellsberg's required critical risk then is a value of p that insures that an expected value maximizing country, i, will prefer to refrain from attack. Further analyses of such deterrence incentive structures incorporated asymmetry of information, manipulations of payoffs (and thus meta-games), control of agenda, promises, threats, commitments, and the advantages or disadvantages of a last move into deterrence analysis. The implications of this theory for crisis behavior (e.g. city evacuation in nuclear crisis), and nuclear force design/operation were developed in a literature on efficacy of secure, survivable but expensive weapons, versus cheap vulnerable ones [Kent and Thaler (1989)]. 1.2.3. Economic models of alliances A third example in which economics effectively defined a crucial defense problem is the economic theory of alliances. Here Olson and Zeckhauser (1966) perceived that the defense provisions of each member of an alliance was really a voluntary contribution to the public consumption good of the group. The implications of this idea swiftly followed; that an alliance will provide its members with an inefficient, suboptimal level of defense. The reason for the inefficiency is that in equilibrium under voluntary provision of a public good E MB > MC; here MB and MC are as defined above and the summation is taken over the alliance group members. Also, it is inferred that larger (richer) members have a tendency to provide disproportionately large shares (necessarily so if preferences and the costs of public good supply are identical among members). An explosion of the literature ensued applying the voluntary public good provision model to military alliances and other international groupings [Sandler and Cauley (1975)]. 1.2.4. Nationalpower, economic survival, and internationaltrade A fourth example of interaction between economics and security originates with the observation that protection of domestic industries against import competition may be justified when a country is subject to a risk of trade interruption. In earlier days of the Cold War such realization produced arguments for strategic material stockpiling, standby production capabilities, and "warm" production base maintenance whether to enhance survival from nuclear attack [Winter (1963)] or preserve a war mobilization [Lincoln (1954)] and fighting capability in less extreme conditions.

Ch. 2: Defense Economics and InternationalSecurity

19

This has been elaborated in more formal insurance models in recent years by economists concerned with the trade and supply interruption (e.g. oil embargo, or critical material supply disruption) [Mayer (1977), Tolley and Wilman (1978), Srinivasan (1987), McGuire (1990a)]. To understand the general logic of this argument, consider a small country which exports good x, and imports y, at given world prices of px in terms of y. When trade is disrupted, this country can and must consume only what it produces; in ordinary times it can export/import. Suppose that the country must choose a single production program both for "war" and "peace" because resources cannot be shifted across industries during war. The chance of "war" or trade disruption is given as 7r. Then for r= 0, the country should choose a program such that its marginal rate of transformation, "MRT", between x and y equals its marginal rate of substitution, "MRS", and both equal Px/Py. If yr = 1 then disruption is certain and the country should select (x, y) such that MRS = MRT at autarky along the transformation curve. But if 0 < jr < 1 then there is a definite risk of trade disruption, and domestic production should be subsidized but not so much as to support the autarkic maximum. Here the object of an enemy attack and of self defense (or deterrence) may be the economy itself. Recognition that efficacious defense (defined as that which improves the chance of peace at some cost in resources) and trade may be connected, has led to more general explorations of theoretical and empirical relationships between the two [Polachek (1980), McGuire and Shibata (1988), Ihori (1994)], including the prospect that trade dependence may actually increase the risks that trade will be disrupted. 1.2.5. Arms races and strategic interactions Another example of the utilization of economics in defense and security studies was the elaboration of L.F Richardson's (1960) ecological models of arms races a massive literature which includes several important insights from economics. For example it has been shown that the mechanistic reaction curves he derived from postulated differential equation processes among states, could also be derived from economic optimizing behavior of nations, confronted with a missile duel technology [Boulding (1962), McGuire (1965), Brito (1972), Intriligator (1975)]. Thus the broad range of allocations processes, solution concepts and techniques inherent in Cournot, Stackelberg, and other modern models of market organization became available to analyze the strategic interactions among countries, to suggest equilibrium outcomes, and to measure progress toward arms limiting equilibria [Shubik (1987), Fontanel (1994)]. This has generated an extensive empirical literature estimating actual arms interactions between the USA and the Soviet Union [McGuire (1977)], between NATO and the Warsaw Pact [Murdoch and Sandler (1982)], between the Soviet Union and Japan plus the USA [Okamura (1991)], and with respect to regional arms races such as between Israel and its Arab neighbors [McGuire (1982)]. Secondly it has been shown that arms race incentives and processes may be logically linked to a country's motives to begin, conduct, and terminate war; and these in turn derived from the tensions between aggression and deterrence and the payoffs a country

20

M.C. McGuire

perceives from such behavior [Brito and Intriligator (1985), Wittman (1991)]. Such models linking arms accumulation processes, with incentives to use armaments also found application in the strategic debates surrounding missile defense and "star wars" programs [Kent (1964), Bailey (1973), McGuire (1987a, 1992), Schelling (1967)]. 1.2.6. Economic ecology and internationalconflict Boulding (1962) was among the first to perceive that the population dynamics and survival/extinction mechanisms modeled by earlier ecologists could be applied to international struggle conceived as the result of economic conflicts. Thus Boulding adapted models of price wars and spatial competition to explain equilibrium in the frontier configurations among States or absorption of one nation by another. The notion that the configuration of nation states and national borders was an equilibrium has also been propounded by Thompson (1974) and by Friedman (1977). Insightful extensions to these approaches have been suggested by Hirshleifer (1987, 1991) in work which draws on elements from the economics of property rights, and of theft and rent-seeking to build economic-ecologic models of resource allocation between production and war/theft in systems of nations - the economics of "ongoing conflict". As the Cold War terminated, borders dissolved, and ethnic conflicts escalated, the practical significance of this work has increased apace.

2. Interactions between economics and defense Does this small drawing of examples suggest any structure of interdependence between economics and defense or security beyond naive enumeration? Here one might pursue two approaches, a strictly conceptual or taxonomic, and a historical approach. 2.1. Levels of interaction Economics figures in the security nexus at least at these four levels of interdependence: (i) Economic management for effectiveness and efficiency in defense, including orchestration of policies deriving from all military and economic policy tools available to the modern State. (ii) The national economy as a source of resource support for security, and a target through which to weaken or to punish enemies. Military capability is only one of many defense instruments supported by economic strength. Other instruments include trade policies, aid programs, financial leverage, debt posture, immigration regulation and leverage, and other economic policy tools of all sorts. (iii) Macro-effects of defense on national economies and interactions among national economies within the international system, including the incidence of defense on economic growth, stability, and prosperity.

Ch. 2:

Defense Economics and InternationalSecurity

21

(iv) Economics as an explanation or a source of the security problems of nations: Are the security threats which diverse nations or regions pose for one another based on economic factors? More specifically, are such threats ultimately based on quarrels over property and wealth, and how to acquire or produce it? Historically, American interests certainly included such factors as safety of US citizens and property at home or abroad, discrimination against US exports or financial interests, control over colonies and possessions, together with a generally stable world order, which fostered liberty and free commerce. This level of interdependence between economics and security is premised on the notion that, at bottom, international security concerns the economic incentives nations face to acquire wealth. Preferences, technology, and wealth distribution may be such as to make conquest, colonization, and property seizure most cost effective, or instead may encourage production, savings, investment, and trade. A systemic equilibrium configuration of nations would balance these various forces. 2.2. Historicalevolution Just as a taxonomic or conceptual approach may impose order on the diversity of topics included in "economics of security" a historical approach can help to identify topics temporarily eclipsed, trends or cycles. More particularly, a historical account will reveal how the focus of Defense Economics has shifted in recent years toward the last two of the foregoing categories. Table 2 (overleaf) presents a summary of the more prominent elements of each of the above categories as they evolved over time. For purposes of exposition recent history is divided into four time periods. The table describes how the economics/security nexus over these periods of time has shifted among categories, and gives selected important issues in the economics of defense at the various periods.

3. Themes in the history of Defense Economics 3.1. World War II Defense Economics as a modern study evolved from highly practical problems of assembling and utilizing the human and material resources in the Second World War, that is out of war production, logistics management [Lincoln (1954)] and operations (or operational) research. Having derived from a policy need, Defense Economics, during this period, was primarily normative. The main macro-economic themes of war production, and material management, asked basically how to manage a command economy with tolerable inflation, how most efficiently to cripple an enemy's economy and whether this could be accomplished by military force [see US Strategic Bombing Survey (1945-46)], or whether "economic warfare" could effectively disable an

22

M.C. McGuire Table 2 Historical evolution of defense economics

Economically efficient defense management

The economy as a resource base for defense

Effects of defense on nat'l economies and the int'l system

Economics as an explanation of the need for defense

Examples of defining issues

Historical period: World War II and Korea 1942-1953 Logistics management

Economy as a support for military strength

Inflation, debt, and war finance

Control of world's resources

Effectiveness of strategic bombing

Operations research

War mobilization and production

Post-war reconstruction

Territorial integrity

Strategic materials and stockpiling

Disarmament and deflation

World order

Marshall Plan

Manpower management

Historical period: early Cold War 1953-1970 Cost effectiveness analysis and program budgeting

The economy an instrument of national security

Sustainability of defense

Containment

Recovery from nuclear attack

Strategic interactions, arms race analysis

Trade, aid, development and defense

Effect of defense on investment, growth, stability

Control of world resources

Military/economic aid, alliances, trade control

Military manpower, R&D, weapon procurement

Economic isolation of Eastern Bloc

East vs. West growth

World order, nuclear survival, domino theory

Missile build-up, arms control

Historical period: later Cold War 1971-1990 Burden shares, cost incentives, and alliance efficiency

Economic sanctions and warfare: unfriendly economies as targets

Defense a source of competitive decline

Preserve int'l boundaries

Missile defense

Force readiness and defense industry policy

Economic punishment of East Bloc countries

Defense as a cause of underdevelopment

Protect Western economies

Int'l arms trade; non-proliferation terrorism

Defense acquisition and contracting management

Long run competition with USSR

Defense a game of economic ruin

World order, nuclear survival

Missile build-down

continued on next page

23

Ch. 2: Defense Economics and InternationalSecurity

Table 2, continued Economically efficient defense management

The economy as a resource base for defense

Effects of defense on nat'l economies and the int'l system

Economics as an explanation of the need for defense

Examples of defining issues

Historical period: post Cold War 1990Defense conversion and downsizing

Investment and trade becomes a substitute for war/conquest

Resource allocation in conflict: trade versus conquest

Security a world wide public good

International trade in nuclear materials

Disarmament

Incentives for nations to enrich themselves peacefully

Defense, Form of governance, and economic performance

Equilibrium distribution of property

Role of NGO's in int'l security

Management of non-traditional peacekeeping missions

Management of violent means of property determination

Endogenous alliance formation

Configurations of nation states in int'l system

Manage boundary revisions, contain local conflicts; migrations, human rights, refugees

opponent's economic capacity to wage war [Knorr (1956)] and finally how to convert to peace-economy without deflation. 3.2. Early Cold War As the national security problem of the US deepened into the Cold War, the policy requirement mutated from how to win a hot war into how to prevail in a protracted conflict. In this environment economics of defense developed along several paths. At the micro-level, operations research led to the development of cost-effectiveness and cost-benefit analysis. The subtleties and paradoxes of deterrence attracted the attention of economists. And increasingly, other policy tools in addition to the military became recognized as instruments to advance national security. Thus, the concept that governments - through effective use of the economy as a policy implement in addition to the military - could induce or sustain a variety of strategic objectives was early established as a doctrine of national security [Knorr (1956), Hirschman (1945)]. The idea that the economy itself is an instrument of defense policy, while economic principles would govern effective organization and application of such instruments, was elaborated during this era. Among the economic instruments for rewarding friends and penalizing enemies for strategic policy purposes one must include: (1) trade policy in its broadest terms - including strategic export control, boycotts, concessionary terms of trade, quota allocations, economic sanctions [Wan (1961)]; (2) foreign economic, technical, agricultural and military assistance; (3) foreign intelligence, information,

24

M. C. McGuire

propaganda, and cultural outreach. These all came to be instruments directed toward a common goal of US national security, and orchestrated by principles of economic effectiveness and efficiency. The early stages of this period being characterized by an optimism over survival from nuclear attack were marked by numerous economic studies of how to manage after an exchange [Winter (1963)]. Simultaneously, realism over the irreversible growth in weapons inventories and staggering costs of defense prompted intense study of the macro-economic consequences of the Cold War itself [Klein and Mori (1973), Weidenbaum (1974, 1992), Leontieff and Hoffenberg (1961)], and particular focus on the capacities of socialist versus free enterprise capitalist economies to sustain a 39% deflection of GNP into defense indefinitely. Although we now know that the real Socialist System was unable to sustain these costs over more than a generation, the record was by no means obvious in the 1950's and 60's. Nor are the ultimate costs to the United States obvious - costs which continue to fall due into the present. The central theme of Defense Economics in this period - as in the preceding period - was normative. Nevertheless, with a widening perception of the uses of the economy as an instrument of security, interest in its positive role in the international security system grew apace. The roots of positive economic analysis of defense arise from early ecological modeling of warlike behavior of nations. The game theoretic structure implicit in Richardson models was exploited to yield positive explanation of international conflict as an economic process. Such economists as Schelling (1960), and Boulding (1962) applied these tools to generalized models of international conflict 4 . Boulding, for instance, recognized that the sources of conflict, might be based solely on preferences - ethnic or national hatreds, inherited perhaps - or might have their basis in quarrels over ownership or use of property, or markets, resources or other items of economic value. In doing so he anticipated current concerns by a generation. On a less sweeping general level it was recognized that arms accumulations by competing countries might follow a pattern of economic interaction similar to oligopoly and that oligopoly models, therefore, could apply to competitions between attack and defense technologies [Kent (1964), McGuire (1967)], and in understanding how deterrence models and arms race models fit together to generate a lesser or greater risk of war itself. 3.3. Later Cold War As the Cold War matured, the range of issues to which economists would address themselves in the name of security continued to increase. Several of these merit

4 Boulding recognized that the basis for conflict, might be based solely on preferences ethnic or national hatreds, inherited perhaps, or might have their basis in quarrels over ownership or use of property, or markets, resources or other items of economic value; in doing so he anticipated current concerns by a generation.

Ch. 2: Defense Economics and InternationalSecurity

25

special attention. Although the subjects to follow by no means exhaust the inventory of relevant topics they represent a fair sampling.

3.3.1. Alliances, burden sharing, and erosion of US hegemony The study of group formation and behavior in defensive alliances drew special attention with the wider appreciation of game theory throughout economics. The agenda setting study of Olson and Zeckhauser (1966) had predicted proportional disparities in the distribution of individual country contributions in favor of the poorer countries in an alliance. With impressive results, this model has been elaborated by many, has been subjected to wide test, and has been extended to the analysis of the voluntary provision of a variety of international public goods (such as foreign aid). Special attention also has been given to the effects of impurity of public goods, or of alliance goods with both a public and a private (joint supply) dimension [Sandler (1977), McGuire (1990b)] extending to less formal alliances than NATO or the Warsaw Pact [Dudley (1979)] and to merging the public consumption good model of alliances with international trade concepts of trading blocks, customs unions, and free trade areas [Wong (1991)]. As in the study of arms races, a large policy and empirical literature grew up to estimate the structure of alliance allocative interactions. This literature estimates the effects of impurity of public goods but also attempts to identify alternative allocation processes - whether driven by Nash-Cournot behavior of individual countries or other more cooperative protocols [Denoon (1985), McGuire and Groth (1985), Murdoch and Sandler (1982, 1984), Murdoch, Sandler and Hansen (1991), Sandler and Murdoch (1990)]. Extensions of the basic model to include impure public goods models of alliance defense provision have been particularly suggestive. "Impurity" may take several forms. One of these is congestion wherein the average cost of providing a unit of public good varies with the size of membership of the alliance and possibly also with the quantity or quality of public defense provided to the group. For example operational coordination requirements or equipment, training, or communication compatibility requirements may increase simply with membership size. In this case, when defense may lie on a spectrum between "public" and "private", one task of empirical economics is to identify its location on that spectrum. Another example of imperfection or impurity of public good may arise when resources allocated to the public good also create a joint private good. Any individual country's contribution to defense against a common threat may simultaneously generate strictly "private" or exclusive benefits for the country making the contribution. Thus forces contributed for defense may serve for internal security needs as well. Still another form of imperfection in the defense public good may arise because the spillovers created by different countries' contributions may not be perfect substitutes for each other. For example, operationally equivalent German and American air wings might not be weighted identically in the defense objective

26

M.C. McGuire

function of the British Defence Ministry. In the conventional pure public good case we would write the utility function of any single country, i, as follows:

Ui [(wi-gi),

Z

gi

(2)

,

where wi denotes country i's income or wealth, g' denotes its contribution to common defense with unit cost assumed to be constant at $1, and E gJ denotes the aggregate provision of defense by all alliance members. Now to incorporate impurity of public good one might write

Ui [(wi- {p(n)gi}), (Z

/gi ),fi(gi),

(3)

where the congestion effect is introduced as p(n) with "p" being the price or average cost of g, and "n" the number of alliance members; the joint product effect is included as f'(g'), and the differential spillover effect is represented by the parameters 7 to indicate the coefficient i attaches to public good provision by j. Most recent empirical estimates of the original formulation as well as of more complex versions thereof, include Conybeare, Murdoch and Sandler (1994) and references therein. Further comparative static analyses of the properties of equilibrium in the voluntary provision model have produced insights of direct relevance to Defense-Economic issues: most particularly the neutrality of income redistribution among alliance members suggests that an alliance's aggregate expenditure may be independent of the distribution of wealth among its members - thus directly challenging the meaning of "exploitation of the great by the small". Extensions of this theme analyzing the effects of average cost differentials among alliance partners have demonstrated that an alliance partner with higher average costs of contributing will benefit in NashCournot equilibrium from displacing supply onto other members of the alliance. The implication that individual countries may gain from being inefficient, high cost providers of defense in an alliance, has caused these effects to be termed "adverse cost control incentives" [Jack (1991)]. 3.3.2. Long run sustainability of defense and relative decline of the US Parallel to these academic developments, as the Cold War ground on and Japan and Germany prospered, Defense Economic policy became increasingly concerned about how alliances divide the costs of common defense - the issue of burden sharing and how division of burdens were affected by work force skill/education levels, wage rates, productivity levels, technological sophistication, levels of research and development and other measures of a nation's competitive success. Underlying such concern was awareness that defense shares of Japan and NATO had dramatically lagged

Ch. 2: Defense Economics and International Security

27

their economic growth. To what extent was the decline in American competitiveness due to a generation of resource focus on defense? The theoretical possibility that defense deflects technological advance, capital investment, and creative human capital with highest skills for technical innovation away from growth had long interested economists [Weidenbaum (1974, 1992)]. The proposition that defense deflected and retarded growth of Western economies argued forcefully by Smith (1980), while others prophetically pointed to the still more onerous burden of the Soviet "Empire" [Rowen and Wolf (1990)]. Similarly, development economists were concerned with analyses of Benoit (1978) and such critics as Deger and Smith (1983), Deger (1986), or Faini, Annez and Taylor (1984), arguing whether defense efforts retarded development by reducing/deflecting growth inducing investment and skimming off technology and trained manpower to the same effect, or instead stimulated growth via manpower skill acquisition benefits of military experience. During these later stages of the Cold War the reality of US relative decline and loss of competitiveness led to fears of an ultimate eclipse of US geopolitical predominance. Academic analyses bearing on strategic overreach [Olson (1982)] combined with growing realizations of the economic constraint on strategy [Denoon (1985)] led to corresponding defense policy concerns over how a country with declining real wages, having lost its technical preeminence, could over a very long period - possibly a generation - sustain big defense to protect high tech, high wage allies. Had America's comparative advantage in defense turned merely to provision of "cheap" labor? This issue remains of crucial interest to defense economics. 3.3.3. Economic warfare/defense: a game of economic ruin Such concerns of this late Cold War period about the very long term consequences of high defense regimes were reflected in several analyses such as Wolfson and Farrell (1987), exploring the concept that the East-West struggle had evolved into a game of economic punishment and endurance leading (whether intended or not) to eventual economic ruin. Thus the military instrument itself was analyzed essentially as a method of economic warfare. The mechanism of economic punishment proposed was that self-protection from a military threat forced countries to divert resources from technological advance and investment with inevitable long run destruction of their ability to compete, of their consumption standards, military effectiveness, and finally even the sustainability of regimes. The idea that imposition of economic costs on an adversary might be militarily effective had been the object of economic analysis before. For example, Schelling (1967) had explored how raising the cost to an adversary of achieving his objectives may be of benefit even when we are indifferent about those objectives per se. If the enemy's demand elasticity structure is such that he will reduce effort directed against an objective we value, then such "a strategy of inflicting cost" can be efficacious. Moreover the idea that one country might use its economy and financial system as instruments of economic war had a significant history from the inter-war period, which

28

M.CC McGuire

in turn had formed a foundation for economic theories of trade and of financial warfare. Wan (1961), using international trade modelling, had developed a subtle and comprehensive analysis of how export/import tariffs-subsidies, as well as State intervention via dumping, boycotts, and embargoes can be used to benefit or harm a "target" nation's economy. This and a small parallel literature on use of financial instruments to harm a target country, form conceptual foundations for a theory of economic sanctions which has received little notice, possibly because repeated analyses of sanctions themselves [Hufbauer and Schott (1985)] has shown them to be frequently ineffective. Arguably, use of the economy as an instrument to drive an adversary to ruin was deliberately and concretely applied in the Reagan Administration with the consequence that the Cold War is now history. History has yet to show however whether strategic competition was a game of ruin for both protagonists. 3.3.4. Economic equity and defense: military manpower and conscription Armies are labor intensive. How to obtain manpower to staff them has been a permanent question throughout history, and economists have been concerned with manpower procurement before World War II. Prominent Cold War studies of efficient and equitable methods for acquiring manpower particularly include Oi (1967) and A. Fisher (1969). Primary issues of concern have included the (a) effects of military manpower acquisition upon the economy, (b) budgetary versus true economic costs of conscription, (c) distributive consequences of conscription and (d) retirement versus retention incentive systems for volunteers. An aftermath of the Vietnam War was to raise acutely the question of equity of conscription. Economics contributed significantly to this issue with the insight that conscription really is the economic equivalent of a special tax upon the draftee. Viewed in these terms, the equity argument against conscription becomes the more convincing, though still to be balanced against arguments that all citizens ought to share equally in risks associated with military service, or that stratification of armed forces by economic or social class is undemocratic and undesirable. In this context, economics made important contributions to the US defense establishment's conversion to volunteer armed forces in the 1970's, and management of military manpower continues to be an important economic issue [see Olvey, Golden and Kelly (1984)]. 3.3.5. Defense industry analysis, acquisition and contracting Next to manpower costs, resources required to arm military forces are second. Fifty years of defense procurement (averaging 1% to 2% of US G.D.P and directed by policy makers from the top of officialdom) has unsurprisingly created a thriving field for analysis with strong connections to industrial organization. Since early work of Peck and Scherer (1962) and Williamson (1967), this field has focussed on contract incentives, principal agent issues, budgetary and decentralization structures, and research development productivities. Studies in this field, both positive and normative,

Ch. 2: Defense Economics and InternationalSecurity

29

illuminate a variety of issues as to how governments can procure required equipment at satisfactory costs to the society, tolerable profits to the private suppliers, and sufficient incentives to producers. These early efforts have evolved in the context of more general developments in industrial organization emphasizing information asymmetries between the government principal and the contracting firm/agent, the game-theoretic structure of interactions among agencies and firms; and these in turn have led to emphasis on auction and bidding mechanisms, single versus multi-source contract design, and the multi-dimensional nature of government's objective function [see Rogerson (Ch. 12)]. Behind such technical implementation problems lay complex issues of the national need for a self-sufficient defense production base, trade-offs between stockpiling and "warm" production base maintenance, and dependability of foreign suppliers in time of emergency or war. Are governments actually better at forecasting risks of emergency or conflict than are private producers? If the private sector can anticipate probabilistic demands which would arise in time of emergency, war, or limited conflict, would the profit motive lead it to make adequate provision for emergencies? Can one depend only on domestic suppliers in this context or might foreign suppliers - whose "excess windfall" profits could not be expropriated - be more likely to maintain needed "excess" capacities? Further up the chain of reasoning one finds considerations of the types of weapon systems required in an evolving framework of international conflict. Economics has assisted in clarifying this question by applying ideas from production theory to defense provision. Thus, studies of efficient capital-labor utilization undergird the force structures of the larger countries, as does analysis of multi-use versus single use weapons, study of reliability versus replaceability, and the host of issues encompassed under the term "logistics". 3.4. Post Cold War The recent upheavals in international politics have again altered/extended the range of topics central to Defense Economics. The end of the Cold War and the dissolution of the Soviet Union have produced a profound alteration in the rationale for defense, the objects of defense and, therefore, the means to achieve them - an upheaval which has evoked beginnings of a re-examination of the security question from top to bottom. Possibly of still greater significance, however, the end of the Cold War has seen growth in the positive study of defense as an economic endeavor pursued in the name of survival by countries and/or their rulers. Ultimately, as a superior positive understanding of this phenomenon arises, and of the equilibrium system of nations generated by national behaviors, it will shape both the objectives of each country's defense strivings, and the instruments favored to influence results. 3.4.1. The emergence ofpositive theories of defense Perhaps acceleration in study of this intriguing development in Defense Economics was evoked by the fragmentation of the USSR and East Bloc countries. Once national

30

M.C. McGuire

borders and the number of countries into which the Soviet Union would divide became a concrete issue, the economic principles for determining the equilibrium and/or optimum of such configurations would become important. Whatever the practical application, a defining element in the post Cold War study of Defense Economics is a new literature which attempts to characterize the configuration of national boundaries as explicable in economic terms, and the result of economic forces. One source for a positive theory returns again to Lanchester (1956)/Richardson (1960) models of ecological struggle and survival as adopted to international behavior. But those more or less mechanistic analyses of international conflict and defense lack micro-foundations for a country's or its ruler's rationale for allocating resources to defense; therefore, they do not exploit the strength of economics in building from individual, self serving, maximizing behavior to systems evolution. The post World War II origins of this literature in positive economics derive foremost from earlier foundations of the economics of property rights and rent seeking, most notably that of Tullock (1974) and followers. If property ownership is prior to production of economic benefit, then property is at risk of expropriation and may be worth protecting from theft. In short, property must be defended in anarchic systems, and defense requires allocations of resources. Thus, the economic concept now emerging as a positive explanation for conflict and security is that ultimately the organization of the world into nation-states is the result of struggle and eventual balance between resources allocated to offense - that is to property right capture, conquest, and control - and resources devoted to defense - that is to protection from, defeat of, or neutralization of the attack. At the margin, mutual neutralization of resource allocations to weapons or systems of conquest and those of defense becomes a necessary characteristic of an equilibrium configuration of nations, each controlling its own territory/property. (Some resource allocations or weapons may serve both offensive and defensive purposes.) At the forefront in analyses of these political economy foundations of the international structure is Hirshleifer in a series of papers [Hirshleifer (1987, 1991, Ch. 7)]. Inspired by insights from economic ecology and evolutionary game theory, Hirshleifer has postulated economies in which conflict is ongoing, and the opportunity cost of protecting one's production is that one produces less and therefore has less to protect. This work also draws on earlier analyses of Thompson (1974), and Friedman (1977), as well as Tullock (1974). Of special interest is how defensive and acquisitive resources offset each other when competing for property rights, and how the protocol of play or competition (Cournot behavior, Stackelberg leadership etc.) influence the outcome. Others, including Wittman (1991), have begun to model a spatial element in these equilibria, in a fashion related to Boulding's (1962) concept of a country's loss of strength gradient - the postulate that a country's capacity to control others declines with distance from its center of power. 3.4.2. New directions in the normative analysis of defense Ultimately, economics will provide an adequate positive explanation for and shared understanding of how geography, technology, costs, and mechanisms for coalition

Ch. 2: Defense Economics and InternationalSecurity

31

formation, generate opportunities for peaceful enrichment through investment, production and trade, as well as opportunities for acquisition through conquest and expropriation. Such insights, one might expect, will profoundly reshape how individual nations define and pursue their individual security. Already the precursors of such an understanding with their implications for the fragmentation of countries following upon the 1990's revolution in Eastern Europe have considerably extended the role of economics in security studies. Most notably, new definitions of security are now perceived as appropriate, which emphasize economic security as having merged with territorial security of States, which focus on the economic behavior of states and populations as the source of security threats (migration, pollution and other externalities, resource depletions, restraint of trade), and which regard territorial threats as deriving essentially from economic motives and forces. This development shifts the definition of security away from individual protection from unilateral territorial invasion, and toward universal issues of management of collective risks. In this context, the planet-wide security interest in national border determination, including recognition of what constitutes a country, is seen as having greater weight than before, and the necessarily myopic interests of individual groups less weight. If security is essentially, largely, or in part, perceived to be a world wide collective good issue - which represents an enormous change in perspective - a major share of the burden of national security falls on solutions to the free rider problem. Thus, a critical element in post Cold War security involves alliance formation and the implied collaboration and compromise. Although the essentials of security seem to have become globalized, a residue of old fashioned parochial national security endures. Defense against physical conquest, occupation, or the exaction of tribute, will always be a vital national interest for each individual country and a global vital interest where the boundaries at issue are those of the great powers. The growing conundrum of international security concerns how to conceptualize a large range of frequently collective problems to face the increasingly numerous tribe of nations. While admitting that deforestation, or global warming, refugee migrations, or massive human rights transgressions may be critical international collective problems, are they security problems in the sense of inducing a role for armed forces? Similarly for narcotics trade, terrorism, conventional arms, and large scale ethnic or religiously motivated barbarisms [see Sandler (1992)]. Borders may be violated not only by foreign troops, but no less by pollution, drugs, crime, refugee migrations etc., but does this mean that such invasions have become the modern national security "threats"? The new twist in the challenge to economics posed by such military or social threats is to devise incentive structures which can make domestic production, saving, and foreign trade more advantageous to potential transgressors than efforts toward territorial expansion, or exploitation of externalities. 4. Agenda for Defense Economics As the range of topics treated in this volume testifies, the tasks for "Defense Economics" are indeed challenging. Economics is summoned to illuminate an ever

32

M.CC McGuire

larger range of pressing issues concerning conflict among states, between groups not organized into states, between these two classes, and concerning the preservation and evolution of the security among them. These security problems are increasingly seen as economic in their source, outcomes, and solutions. Moreover, the spectacle of policy makers floundering in "ad hockery", should provide ample incentive to economists to tackle the tough issues. How might we organize these to gain some insight ourselves as to the likely future focus of this field of study? There are, I think, three broad avenues along which Defense Economics will evolve in the future. These may interleave one another but be distinguishable nevertheless. (a) First, is a group of inherited, continuing concerns which have long occupied Defense Economics. Some questions from the earlier repertoire may have lapsed such as how to "prevail" in a thermonuclear war. Many other inherited issues, however, continue as important as ever. (b) Second, is a broad range of new policy issues deriving from the economic and political evolution which produced the end of the Cold War, dissolution of the USSR, and collapse of Communism. (c) And third, is a set of conceptual issues, both positive and normative, concerning the economic incentive structures in the international system and their implications for war, security, and peace. I take these areas up in the order mentioned. 4. 1. Continuing/inheritedpolicy concerns With respect to these issues, "a principal task of defence economics ... is to develop an analytic framework that takes into account the specialized institutions of the defence sector in order to deal with the standard economic problems of choice, efficiency, and growth in a realistic and policy-relevant manner, while remaining firmly rooted in the intellectual traditions of economics in order to exploit advances made in the larger field" [Reppy (1991, p. 270)]. These represent extrapolations of the conventional analyses which has evolved to become defense policy. Some of these may be quite straightforward, others requiring highly innovative new analysis. I will mention only a few from the huge existing menu. Arguably most important of the inherited problems of security concerns nuclear issues, particularly the imperative to preserve nuclear deterrence and avoid nuclear weapons use between or among powers with large weapon inventories and delivery systems. This category encompasses questions of proliferation of nuclear weapons, including disposal of retired weapons and effects of increasing membership in a nuclear club. With respect to the latter how does increased membership influence the likelihood of abstinence from use, stability and dynamics of deterrence, and the opportunities for confrontation and compellence [Schelling (1966), Intriligator and Brito (1981, 1984)]. Management of nuclear downsizing involves important issues in cost effectiveness and the avoidance of destabilizing incentive structures for use of

Ch. 2:

Defense Economics and InternationalSecurity

33

such weapons. Safety within the cone of mutual deterrence [Intriligator (1975)] may involve not merely numbers of weapons (or megatonnage) but also specific sequencing of build down [Kent and Thaler (1989)]. The unavoidable, if regrettable, truth that potential thermonuclear war is a permanent feature of the landscape of international politics insures the lasting importance of economic analysis of such issues. Included especially are the incentives, costs, and resulting (in)stabilities which arise in control of strategic nuclear defense, particularly avoidance of surprise developments in missile defense ([McGuire (1987a, 1992)]. The fact that the nuclear threat persists in a more amorphous, less direct and structured, less understood strategic environment of multiple interests and overlapping coalitions, makes this problem more treacherous than ever. Aside from nuclear questions, included in the "inherited" or "continuing" category is a large set of parallel ongoing issues concerning management of technological improvements in defense particularly great power conventional forces. The advance of conventional "smart" weapons technology making "surgical" or "sanitized" warfare closer and closer to reality would demand serious re-examination of conditions for use of force, and the wide array of traditional issues as to efficiency in their provision, including defense procurement and contracting, defense industrial policy and preparedness. Among the more intriguing insights to emerge from current ongoing analyses of these questions are the incentives which countries create for themselves by "conservatively" preparing for the worst. As pointed out by Wolfson (1991) if readiness causes some of the costs of war to be borne before the decision to go to war, then at the time of decision, the true costs of war will in part be "sunk" and thereby diminish incentives to compromise. Is this an instance of myopia and inconsistency in multiperiod planning - so prominent a feature of many other areas of multi-period decision analysis? Or is it a case of success in making a commitment by spending in advance to reduce the incentive to abandon the present promise? Perhaps equally arresting and insightful analogues await further study of the interactions between defense and national economies, defense and development, or other traditional subjects. With outbreak of major war between the superpowers now less an imminent threat, and its avoidance less preoccupying, numerous other "traditional" issues will come into stronger focus. For example, the relationships between defense and development, and the closely related questions of the impact of foreign aid on both. This nexus of problems promises to be of greater concern not only because the lapse of the Cold War has re-ordered our priorities as to which questions most deserve energetic study. The end of the Cold War has also produced changes in behavior which tend to elevate the importance of defense in developing countries. Thus, for example, ethnic violence, once suppressed under the monolith regime of East-West rivalry, has come to the fore. Clearer understanding of the effects of defense on the political economy of nation building may derive from new theoretical endeavors which focus on the relationships between defense spending, the form and durability of government, and its allocative behavior [Palda (1993), Grossman (1992)]. These studies - if synthesized with the burgeoning literature which concerns governmental form and development - promise

34

M.C. McGuire

significant new insights into the connections between development and defense. Related to the interconnected political economy of development and defense are recent analyses concerning the effects of foreign aid on the incentive structures of democratic versus autocratic regimes which illuminate unexpected relationships between economic or military assistance and regime durability. This approach could be merged with earlier work on aid, defense and development [Dudley (1979), Smith, Humm and Fontanel (1987), McGuire (1987b)] to produce better understanding of the Third World as a new active player in the international security drama - no longer a passive stage for great power rivalry as in the Cold War (terminology due to economist Charles Wolf, Jr.). The spread of and potential for ethnic conflict suggests that the literature on development in conflict ridden contexts [Hendry (1962), Dacy (1983)] will find renewed application. And though statistical evidence relating defense, development, and growth is ambiguous to muddled, these problems may yield to careful case study. Another of the traditional areas of Defense Economics to carry forward from the earlier era, and be an object of closer study must be the economics of international arms trade and arms transfer. Exponential growth in this sector would insure closer scrutiny [Levine, Sen and Smith (1994)]. Moreover, the arms trade is of special interest for three trends which it embodies. (1) The increasing relevance of the decisions of non-governmental bodies and organizations to national/international security. Arms suppliers are often private firms, sometimes multinational corporations themselves. (2) The world's growing dependence on international economic integration. (3) The augmented arena for external economies/diseconomies as an international free rider issue. 4.2. Normative issues arisingfrom world economic change and the collapse of Communism The traditional concerns of Defense Economics - some very few of which have been enumerated above - all must fit in with the exciting if confusing developments following the autumn of 1989. More than this, the demise of the Cold War has ushered in a new range of normative issues sweeping in their scope with economics at their center. These issues can be usefully organized into several categories. 4.2.1. Economics of transition The primary security problem generated specifically by the end of the Cold War is that of a successful transition of East Bloc countries toward peaceful and stable growth. Inasmuch as such success is so largely economic, "Defense Economics" has a major input. Here one must include such issues as transition from command to market economies, the management of defense industries in economies under such stressful change, utilization of defense capital, labor, and technologies in economies under transition. Nor, of course, is it solely former Communist states which must manage transitional arms reduction and resource redirection challenges. The economics of

Ch. 2:

Defense Economics and International Security

35

defense reductions within the US and other NATO countries will continue to attract strong attention from economists [Fontanel (1994), Seiglie (1992)]. 4.2.2. New definition of national/internationalsecurity As the threat to territorial integrity of Great Powers has receded, the issue of what now constitutes security has risen to the fore and particularly whether national security is mostly just economic security plain and simple. Just as lessened threats of territorial war have allowed and encouraged greatly increased interdependence among economies of the world, the ever growing economic linkages make individual countries increasing vulnerable to unanticipated economic crises [Cooper (1986)]. Such crises may stem from sudden monetary phenomena, including debt default and liquidity crises, exchange rate or other price catastrophes, or from slowly building decay of competitive institutions. They may originate in natural disasters such as famine, epidemic, overpopulation, or massive human migrations as examples; or from neglect or free riding behavior such as pollution and overexploitation of resources; and they might originate in deliberate economic acts of resource owners, or they might derive from deliberate political acts, possibly unanticipated effects of political realignments. All such interdependencies and their breakdowns will display complex components, such as unexpected institutional failures, technical unknowns, natural physical discontinuities, and/or abrupt divergences in individual human action and collective political behavior. It is not difficult to imagine threats deriving from religious fanaticisms which would combine all these components. Economists will also be alert to the likelihood that the definition of "security" should expand if the costs of controlling undesired developments has declined. That is, whereas it may once have been too expensive to undertake corrections of human rights abuse, excesses of religious fanaticism, or refugee starvation, now with the political costs of corrections diminished, new incentives may arise to include such mishaps under the aegis of security. But then this raises the question of how far to carry such extended definition: all the way to include world poverty, income inequality, health, and environmental safety? Encompassed in this category are crucial new questions of conditions under which conflicts within a country can reach a stage that international intervention is justified, even required; that is, norms for international interference in the "internal" affairs of a country [see Kaysen and Reed (1993)]. Reevaluation of the acceptable limits of national sovereignty will ultimately rest upon economic concepts such as externalities, spillovers, public goods, private property rights, social and individual welfare, free riding, and transactions costs, as well as other criteria in the realm of the political, legal, moral, psychological, and purely pragmatic. 4.2.3. New economic instrumentsfor security Greater interdependence among world economies implies an increase in the consequences of economic mishap. Corresponding to the benefits from increased

36

M.C. McGuire

specialization and interdependence are increased risk of disruption. Accordingly, an important new goal for Defense Economics will be to further understanding of how economic instruments may be used to deal with such risks. Some of these risks may be common or collective, requiring extensions of models of insurance, and self-protection to the realm of collective action and the relations among states [see for example McGuire and Shibata (1988), Ihori (1994)]. Other types of economic instruments may be employed individually by specific countries or groups of countries and thus be "private" rather than public goods; for example stockpiling policy as an instrument to reduce effects of trade interruptions may be an example of a private response, while contingent agreements for multi-country energy sharing would represent a collective response. The uses of economic sanctions, and all other manner of economic policy instruments to achieve security or other political goals will deserve fresh re-examination. Of special interest here may be analysis of the relative impact of economic rewards versus punishments, the role of credit/debt policies and threat of default, new uses of economic and military aid, trade and factor migration controls broadly conceived. 4.2.4. New concepts of alliances and replacementfor bi-polarworld hegemony The last sub-category under the umbrella title of "new issues post-Communism" concerns security management in a world of evolving great power coalitions. Can Defense Economics contribute to a resolution of the puzzle over what is to replace bipolar hegemony. Such an endeavor is by no means the exclusive realm of economics, but economists do have a special contribution to make especially insofar as the adhesive to bind alliances together may no longer be simply a common defense against conquest. Instead a complex web of exchanges of all types of property, including goods, services, people, capital, financial promises, physical externalities also binds "allies". These create mutual expectations of self and others, and complex agreements and practices to provide more or less common collective goods adds to the binding force. A major problem for economics is to understand how collective goods such as common defense protection, and individual goods such as benefits from specialization and exchange combine in producing incentives for alliance formation/functioning [see McGuire (1989), Wong (1991), Ihori (1993)]. And because management of the free rider problem is so central to "alliance" effectiveness, the economics of voluntary behavior in small groups will itself continue to be central to security questions, and should illuminate such important issues as developing regional alliances, power balances, and conflicts. Such analysis might encompass applications of honest preference revealing techniques and self-enforcing commitment mechanisms to group adhesion in defense environments. 4.3. Positive analysis of internationalpower equilibria Although the details of an ultimate positive understanding of defense are far from clear, and even the outlines obscure, one can see various micro-economic elements which

Ch. 2:

Defense Economics and InternationalSecurity

37

will have to be integrated into an overall model. These micro-economic elements which must be combined, comprise at least the following: property rights and the underlying appropriability of "production", which is to say the technology of conflict versus peaceful competition; costs of politico-military conquest versus defense; institutions of collaboration and coalition formation; and principles of systems equilibrium. We can briefly describe each of these components in turn to obtain some small idea of how they might fit together. Property rights and appropriability: The centrality of the concept of property, property rights, rent seeking behavior and theft being stipulated, a most important component for positive understanding of Defense Economics concerns when and why one group or country finds it advantageous to conquer or otherwise expropriate the product of another group, and when it finds it more advantageous to save, invest, labor, and trade. In other words what determines the opportunity cost of expropriative activity. Part of the answer to this question will lie in the costs of political or military dominance compared with the costs of defending against such dominance. We will turn to these momentarily, but more fundamental may be the inherent availability or vulnerability of a country's production to coerced seizure. This should depend upon a country's stage of economic and political development. Natural resources may be easy to seize and expensive to defend, while expropriation of sophisticated, knowledge intensive products far more difficult because their production or delivery requires willing cooperation of a myriad of technical specialists. And not only may technically advanced, highly networked economies be inherently less vulnerable to conquest, the same is likely to be true of politically advanced countries. It is one thing to seize control of a country from a dictator or oligarchy who already had established a system of expropriation from its own population; it is quite another to devise institutions which will effect such expropriations from a free market of a previously competitive population. Costs of conquest versus defense: The economic explanation of international structure centers on the notion that ultimately the organization of the world into nation-states is the result of struggle and eventually a balance between resources allocated to capture and resources devoted to defense. Such mutual neutralization of resource allocations is then a characteristic of an equilibrium configuration of nations. In this environment, the technology of offense versus defense with special reference to scale economies must be highly influential, so a significant part of the problem of understanding the equilibrium in the international system will be to understand those technologies of capture vs. protection. The forces of theft and protection play themselves out over geography, which thus mediates crucial components of the relevant technology. Original analysis of aspects of this geo-technology, are due to Friedman (1977) indicating the effects of distance on market dominance, and by Boulding (1962) of the effects of distance on military power projection as measured by a "loss of strength gradient". At the margin, however, wherever forces compete and the struggle between nations or coalitions is in balance, this will involve a sort of neutralization of defensive and offensive strength.

38

M.CC McGuire

Consequently, the geography over which to expect conquest to dominate defense must depend on the relative technical capabilities of the two activities, returns to scale (both geographic scale and resource input scale), resource availabilities of competitors and other factors. Suppose one could measure a country's vulnerability to or risk of being conquered, annexed, colonized, etc. This risk will depend on the weapons and forces threatening such a country and the defensive weapons it commands through resource allocations to defense and protection. Now consider the marginal ratio of defense forces to offensive forces which will just offset or neutralize each other, keeping the defender at the same level of vulnerability. Such a ratio is a measure of the relative efficacy of defense against offense. How does such a ratio depend on the absolute level of risk? Is it also a function of absolute resource levels? How does it depend on geography? Measures of defensive versus offensive effectiveness such as these have been an important element in strategic analysis and operations research analyses of missile duels and missile defense [Bailey (1973)]. But now it appears that similar concepts would seem to have wider application to the entire problem of systems equilibrium of nations. Hirshleifer also has explored these tradeoffs at a theoretical level (Ch. 7 of this volume) and he argues more generally for a redirection of intellectual focus toward conflict ridden processes ("the dark side") throughout economic study. Other economists have delved more specifically into the micro-foundations of allocations between offense (property acquisition or conquest) activities and defensive ones (property protection, offensive weapon defeat). [see Grossman (1991), Skaperdas and Syropoulos (1994), and Garfinkel (1990)]. Institutions of collaboration: The flourishing economic theory of games will doubtless be central to an evolving understanding of international security and its political economy. Game theory one might hope would provide insights into whether the system of interacting nations has a "core" and the implications of a positive or negative answer. Such conjectures will certainly build on knowledge which presently is accumulating with respect to how institutions influence warlike proclivities of nations. Evidence is growing as to institutional foundations for a peaceful world; first that the greater the volume of international trade between countries, the less likely are they to have recourse to war to settle disputes [Polachek (1980)]. Second, democratic regimes are less likely to go to war with each other than either autocratic regimes with each other or with democracies. Future empirical and conceptual progress in understanding these issues may help in evaluating the present trend toward increased fragmentation of nation states. Is the monopolistic model of a single world government in effect to be rejected by a competitive system of hundreds and hundreds of countries? Or is an oligopolistic system of countries a possibility? Or consider the economic model of a price leadership in which one or a few big participants in a market interact so as to determine variables which are accepted as parameters by numerous smaller participants. Can this be transplanted into this environment where unlike the concentration of industry example, the goods provided to the market are not private goods but public?

Ch. 2:

Defense Economics and InternationalSecurity

39

Principles of system equilibrium: Thompson (1974) first made a suggestion as to how the incentives which apply to countries individually would fit together in an international security equilibrium. In the tradition of general equilibrium in resource allocations his suggestion has great appeal. Thompson's conjecture was that a distribution of property among states must be such that when each country mounts its optimal defense, any other country after expending the resources necessary to overcome that optimal defense if possible, would find that conquest of the defending country cannot produce a net gain. Such an equilibrium condition would presumably determine the number of competing countries, every country's overall allocation between production and security, and within each country's security budget its allocation between defense and offense. Further, the effects of technological change, increased wealth, capital and labor accumulation, and population change would seem in principle to be deducible from such condition. Thompson's equilibrium criterion may not be ultimately satisfactory. Possibly explicit inclusion of public good and externality effects would alter equilibria qualitatively. This criterion ignores asymmetries in information, and effects of expectational factors and commitment issues. A full application of the idea should require a multi time period context, and consideration of coalition possibilities, but the object of understanding the system-wide implications of these security and defense strivings will constitute a central goal of future research in Defense Economics. 5. Concluding remark This essay has briefly surveyed the course of Defense Economics as a field of study in the post World War II era. The field has expanded continuously in the scope of the questions it addresses, in the economic tools employed in its analyses, and in the criteria of intellectual rigor and social usefulness by which it is to be judged. Amidst this surging growth in Defense Economics as a field of study, one trend stands out. Where the focus of study originated in normative concerns for national security within competing power blocks, the field has evolved as one of positive study of the individual elements within the international system, and with normative analysis increasingly reflective of a globalized point of view. References Bailey, M.J., 1973, Strategic interaction inforce structure planning, unpublished Working Paper (CarnegieMellon University, Graduate School of Industrial Administration, Pittsburgh, PA). Benoit, E., 1978, Growth and defense in developing countries, Economic Development and Cultural Change 26, 271-280. Boulding, K.E., 1962, Conflict and defense: A general theory (Harper and Row, New York). Brito, D.L., 1972, Adynamic model of an armaments race, International Economic Review 13, 271-280. Brito, D.L., and M.D. Intriligator, 1985, Conflict war and redistribution, The American Political Science Review 79, 943 957.

40

M.C. McGuire

Conybeare, J.A.C., J.C. Murdoch and T. Sandler, 1994, Alternative collective-goods models of military alliances: Theory and empirics, Economic Inquiry 32, 525-542. Cooper, R., 1986, Economic policy in an interdependent world (MIT Press, Cambridge, MA). Dacy, D.C., 1983, Foreign aid, war, and economic development (Cambridge University Press, Cambridge). Deger, S., 1986, Economic development and defense expenditure, Economic Development and Cultural Change 35, 179-196. Deger, S., and R. Smith, 1983, Military expenditure and growth in less developed countries, Journal of Conflict Resolution 27, 335-353. Denoon, D.B.H., 1985, Constraints on strategy (Pergamon-Brassey, Washington, DC). Dudley, L., 1979, Foreign aid and the theory of alliances, Review of Economics and Statistics 61, 564-571. Ellsberg, D., 1956, Theory of the reluctant duelist, American Economic Review 56, 909-923. Faini, R., P. Annez and L. Taylor, 1984, Defense spending, economic structure and growth: Evidence among countries and over time, Economic Development and Cultural Change 32, 487-498. Fisher, A., 1969, The cost of the draft and the cost of ending the draft, American Economic Review 59, 239-254. Fontanel, J., 1994, The economics of disarmament: A survey, Defence and Peace Economics 5, 87-120. Friedman, D., 1977, The size and shape of nations, Journal of Political Economy 85, 59-77. Garfinkel, M., 1990, Arming as strategic investment in cooperative equilibrium, American Economic Review 80, 50-68. Grossman, H.I., 1991, A general equilibrium model of insurrections, American Economic Review 81, 912-921. Grossman, H.I., 1992, Foreign aid and insurrection, Defence Economics 3, 275 288. Hendry, J.B., 1962, Economic development under conditions of guerilla warfare: The case of Vietnam, Pacific Affairs, 2, 275 288. Hirschman, A.O., 1945, National power and the structure of foreign trade (University of California, Berkeley, CA). Hirshleifer, J., 1987, Economic behaviour in adversity (University of Chicago Press, Chicago, IL). Hirshleifer, J., 1991, The paradox of power, Economics and Politics 3, 177-200. Hitch, C.J., 1958, Economics and military operations research, Review of Economics and Statistics 40, 199-209. Hitch, C.J., and R.N. McKean, 1960, Economics of defense in the nuclear age (Harvard University Press, Cambridge, MA). Hufbauer, G., and J. Schott, 1985, Economic sanctions reconsidered (Institute of International Economics, Washington, DC). Ihori, T., 1993, International public goods and risks of emergency, unpublished mimeograph (Department of Economics, Tokyo University). Ihori, T., 1994, Alliance protection against emergency and welfare, unpublished mimeograph (Department of Economics, -Tokyo University). Intriligator, M.D., 1975, Strategic considerations in the Richardson model of arms races, Journal of Political Economy 83, 339 353. Intriligator, M.D., 1990, On the nature and scope of defence economics, Defence Economics 1, 3 11. Intriligator, M.D., and D.L. Brito, 1981, Nuclear proliferation and the probability of nuclear war, Public Choice 27, 247-260. Intriligator, M.D., and D.L. Brito, 1984, Can arms races lead to the outbreak of war? Journal of Conflict Resolution 28, 63-84. Jack, B., 1991, International public goods: The economics of their provision and cost-control under the Cournot Nash hypothesis, unpublished dissertation (University of Maryland, College Park, MD). Kaysen, C., and L.W. Reed, eds., 1993, Emerging norms of justified intervention (American Academy of Arts and Sciences, Cambridge).

Ch. 2:

Defense Economics and InternationalSecurity

41

Kent, G.A., 1964, Damage limiting: A rationale for the allocation of resources by the U.S. and U.S.S.R. (US Department of Defense, Defense Research and Engineering, Washington, DC). Kent, G.A., and D.E. Thaler, 1989, First strike stability and strategic defenses (The RAND Corporation, Santa Monica, CA). Klein, L.R., and K. Mori, 1973, The impact of disarmament on aggregate economic activity, in: B. Udis, ed., The economic consequences of reduced military spending (Lexington Books, Lexington) 59-77. Knorr, K., 1956, The war potential of nations (Princeton University Press, Princeton). Lanchester, F.W., 1956, Aircraft in warfare: The dawn of the fourth arm, in: J. Newman, ed., The world of mathematics, Vol. 4 (Simon and Shuster, New York) 2138-2148. Leontieff, W., and M. Hoffenberg, 1961, The economic effects of disarmament, Scientific American 204, 47-55. Levine, P., S. Sen and R. Smith, 1994, A model of the international arms market, Defence and Peace Economics 5, 1-18. Lincoln, G.A., 1954, Economics of national security (Prentice Hall, Englewood Cliffs, NJ). Mayer, W., 1977, The national defense tariff argument reconsidered, Journal of International Economics 7, 363-377. McGuire, M.C., 1965, Secrecy and the arms race (Harvard University Press, Cambridge, MA). McGuire, M.C., 1967, The structure of choice between deterrence and defense, in: R.N. McKean, ed., Issues in defense economics (National Bureau of Economic Research, New York) 129-149. McGuire, M.C., 1977, A quantitative study of the strategic arms race in the missile age, Review of Economics and Statistics 59, 328-339. McGuire, M.C., 1982, U.S. foreign assistance, Israeli resource allocation, and the arms race in the Middle East, Journal of Conflict Resolution 26, 199-235. McGuire, M.C., 1987a, Economic considerations in the comparison between assured destruction and assured survival, in: C. Schmidt and E Blackaby, eds., Peace, defense, and economic analysis (Macmillan, London) 122-150. McGuire, M.C., 1987b, Foreign assistance, investment and defence: A methodological study with application to Israel, 1960-1979, Economic Development and Cultural Change 35, 847-873. McGuire, M.C., 1989, Alliance protection against national emergency: Prevention, preparedness, and insurance, Pew studies in international security (University of Maryland, College Park, MD). McGuire, M.C., 1990a, Coping with foreign dependence: The simple analytics of stockpiling versus protection, Discussion paper 70 (Institute of Southeast Asian Studies, Singapore). McGuire, M.C., 1990b, Mixed public-private benefit and public good supply with an application to the NATO alliance, Defence Economics 1, 17 36. McGuire, M.C., 1992, The new strategic environment and economic factors in the future of nuclear defense, in: W. Isard and C. Anderton, eds., The economics of arms reduction and the peace process (North-Holland, Amsterdam) 143-157. McGuire, M.C., and C.H. Groth Jr, 1985, A method for identifying the public good allocation process within a group, Quarterly Journal of Economics 100, 915-934. McGuire, M.C., and H. Shibata, 1988, Protection of domestic industries versus defense against trade disruptions: Pew studies in international security (University of Maryland, College Park, MD). Murdoch, J.C., and T. Sandler, 1982, A theoretical and empirical analysis of NATO, Journal of Conflict Resolution 26, 237-263. Murdoch, J.C., and T. Sandler, 1984, Complementarity, free riding, and the military expenditures of NATO allies, Journal of Public Economics 25, 83-101. Murdoch, J.C., T. Sandler and L. Hansen, 1991, An econometric technique for comparing median voter and oligarchy choice models of collective action: The case of the NATO alliance, Review of Economics and Statistics 73, 624-631. Oi, W., 1967, The economic cost of the draft, American Economic Review 57, 39-62. Okamura, M., 1991, Estimating the impact of the Soviet Union's threat on the United States-Japan alliance: A demand systems approach, Review of Economics and Statistics 75, 624-631.

42

M.C. McGuire

Olson, M., 1982, The rise and decline of nations (Yale University Press, New Haven). Olson, M., and R. Zeckhauser, 1966, An economic theory of alliances, Review of Economics and Statistics 48, 266 79. Olvey, L.D., J. Golden and R. Kelly, 1984, The economics of national security (Avery Publishing, Wayne, NJ). Palda, F., 1993, Can repressive regimes be moderated through foreign aid? Public Choice 77, 535-550. Peck, M.J., and FM. Scherer, 1962, The weapons acquisition process: An economic analysis (Harvard Graduate School of Business Administration, Boston, MA). Polachek, S.W., 1980, Conflict and trade, Journal of Conflict Resolution 24, 55-78. Reppy, J., 1991, On the nature and scope of defense economics: A comment, Defense Economics 2, 269-271. Richardson, L.F., 1960, Arms and insecurity: A mathematical study of the causes and origins of war (The Boxwood Press, Chicago, IL, and Quadrangle Books, Pittsburgh, PA). Rowen, H., and C. Wolf Jr, 1990, The impoverished superpower: Perestroika and the Soviet military burden (ICS Press, San Francisco). Sandler, T., 1977, Impurity of defense: An application to the economics of alliances, Kyklos 30, 443 460. Sandler, T., 1992, On terrorism, guerilla warfare, and insurrections, Defence Economics 3, 259-262. Sandler, T., and J. Cauley, 1975, On the economic theory of alliances, Journal of Conflict Resolution 19, 330-348. Sandler, T., and J.C. Murdoch, 1990, Nash-Cournot or Lindahl behavior?: An empirical test for the NATO allies, Quarterly Journal of Economics 105, 875-894. Schelling, T.C., 1960, The strategy of conflict (Harvard University Press, Cambridge, MA). Schelling, T.C., 1966, Arms and influence (Yale University Press, New Haven, CT). Schelling, T.C., 1967, The strategy of inflicting costs, in: R.N. McKean, ed., Issues in defense economics (Columbia University Press, New York) 105-127. Seiglie, C., 1992, Determinants of military expenditures, in: W. Isard and C. Anderton, eds., Economics of arms reduction and the peace process (North-Holland, Amsterdam) 183 202. Shubik, M., 1987, The uses, value and limitations of game theoretic methods in defense analysis, in: C. Schmidt and F. Blackaby, eds., Defense and economic analysis (St Martin's Press, New York) 53-84. Skaperdas, S., and C. Syropoulos, 1994, Competing for claims to property, unpublished mimeograph (University of California-Irvine, CA). Smith, R., 1980, Military expenditure and investment in OECD countries, 1954-1973, Journal of Comparative Economics 4, 19 32. Smith, R., A. Humm and J. Fontanel, 1987, Capital labour substitution in defence provision, in: S. Deger and R. West, eds., Defence, security, and development (Frances Pinter, London) 69 80. Srinivasan, T.N., 1987, The national defense argument for government intervention in foreign trade, in: R. Stern, ed., U.S. trade policies in a changing world economy (MIT Press, Cambridge, MA) 337-363. Thompson, E.A., 1974, Taxation and national defense, Journal of Political Economy 82, 755 782. Tolley, G., and J. Wilman, 1978, The foreign dependence question, Journal of Political Economy 85, 323-347. Tullock, G., 1974, The social dilemma: Economics of war and revolution (Center for Study of Public Choice, University of Virginia, Charlottesville). US Strategic Bombing Survey, 1945-46, The effects of strategic bombing on the German (Japan's) war economy (US Department of the Air Force, Washington, DC). Wan, H.Y., 1961, A Contribution to the Theory of Trade Warfare, unpublished dissertation (Massachusetts Institute of Technology, Cambridge, MA). Weidenbaum, M., 1974, The Economics of Peacetime Defense (Praeger Publishers, New York).

Ch. 2:

Defense Economics and InternationalSecurity

43

Weidenbaum, M., 1992, Small wars, big defense: Paying for the military after the Cold War (Oxford University Press, New York). Williamson, O.E., 1967, The economics of defense contracting: Incentives and performance, in: R.N. McKean, ed., Issues in defence economics (Columnbia University Press, New York) 217-278. Winter Jr, S., 1963, Economic viability after thermonuclear war: RM-3436-PR (The RAND Corporation, Santa Monica, CA). Wittman, D., 1991, Nations and states: Mergers and acquisitions; dissolutions and divorce, American Economic Review Papers and Proceedings 81, 126-129. Wolfson, M., 1991, Essays on the cold war (Macmillan, London). Wolfson, M., and J.P. Farrell, 1987, Economic warfare between the superpowers, in: C. Schmidt and F. Blackaby, eds., Peace, defense and economic analysis (Macmillan, London) 155-184. Wong, K., 1991, Foreign trade, military alliance, and defence-burden sharing, Defence Economics 2, 83-103.

Chapter 3

WORLD MILITARY EXPENDITURES MICHAEL BRZOSKA Bonn International Centerfor Conversion

Contents Abstract Keywords 1. Introduction 2. Concepts and definitions 2.1. Functional versus institutional approaches 2.2. Direct, indirect and intangible costs and benefits 2.3. Current versus comprehensive accounting 2.4. Standard definitions 3. Intertemporal consistency 4. International comparisons 5. Relative measures 6. Data sources 7. International arms transfers 8. Conclusion References

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

46 46 47 48 49 50 51 51 52 54 57 59 63 64 65

46

M. Brzoska

Abstract Military expenditure is difficult to define. Major issues are functional versus institutional approaches to defense, indirect and intangible costs and benefits and current versus comprehensive accounting. Authoritative institutions have adopted standard definitions but national governments are free to use their own definitions. Specific inflation pressures complicate the creation of real time series of military expenditures. International comparisons are influenced by the choice of exchange rates. For some countries, no credible data are available. Data series on military expenditures and arms transfers must be used with caution as the publishing institutions have only limited resources to deal with the numerous conceptual and practical problems.

Keywords arms transfers, conscription, costs of wars, data limitations, data sources, deflators, exchange rates, indirect costs, inflation, international comparisons, military aid, military strength, national accounts, opportunity costs, productivity, purchasing power parity, secrecy, social benefits

Ch. 3:

World Military Expenditures

47

1. Introduction At first glance, military expenditures seem to be a straightforward measure: the costs of maintaining a military establishment in war and peace. But there are a number of conceptual problems with military expenditure data. In addition, data on military expenditures are of unequal quality and availability. Military expenditures are an overburdened concept both in academic analysis and in international politics. Military expenditures are basically an input measure: the aggregation of payments for soldiers and other persons concerned with the regular armed forces of a particular country, for goods purchased by the armed forces, and services bought from civilians, over a period of time, usually a year. Related to other economic aggregates, such as Gross Domestic or National Product (GDP; GNP) or the Central Government Expenditure (CGE) it is a measure of the relative burden of the military sector. The input costs of the military sector can also be thought of as their opportunity costs for expenditures in other, civilian, sectors. Similarly to other government expenditures whose productivity is difficult to measure, military expenditures are often used as an output measure. There is no readily available indicator of military strength, so military expenditures often serve this purpose. Of course, this is highly problematic. Strength is basically a stock measure, dependent on the military equipment available: for an effort to measure military capital see Hildebrandt (1990). Fighting performance in war and the ability to prevent the outbreak of war are difficult to measure and depend on a number of non-pecuniary factors, such as training, motivation and leadership [Wiberg (1984), Goertz and Diehl (1986)]. The use of military expenditures as a measure of military strength is especially numerous in international comparisons. Some methods have been developed to deal with the major deficiencies, e.g., the "building block" approach used to measure Soviet military expenditures during the Cold War. Another use of military expenditures occurs in military alliances in order to measure the "contribution" of members to the common cause. Again, the nature of military expenditures as an input measure requires interpretation or corrective calculations, e.g. to deal with voluntary versus draft recruitment. The level and composition of military expenditures are often treated as state secrets, even in states with open political and parliamentary systems. Sizeable portions of the military budget, so called "black programmes" are not open to public scrutiny. The availability of military expenditures is low in states with highly autocratic regimes and in regions in conflict, for instance, the Middle East. Governments in Europe are required to reveal military expenditures to each other though not to the public - in the context of the Organization for Security and Cooperation in Europe (OSCE) 1. A number of states report data to the UN on a According to an agreement within the predecessing organization, the Conference on Security and Cooperation in Europe (CSCE), the Vienna Document 1990 of the Negotiations on Confidence and Security Building Measures convened in accordance with the relevant provisions of the Concluding Document of the Vienna Meeting of the CSCE, article 1,paragraphs 14 16, reprinted in SIPRI Yearbook (1991, p. 476-477).

48

M. Brzoska

voluntary basis. Comprehensive worldwide statistics of military expenditures are not available from any authoritative source; instead there exists a number of interested institutional bodies that issue series of military expenditures. They are only capable of correcting some of the major deficiencies of military expenditure data. 2. Concepts and definitions Governments are basically free to define military expenditures according to their own wishes and purposes. The chosen definition is reflected in the national budget or statement of expenditure. Sometimes the liberty to define is used for purposes of deception, or because of specific tradition. But there are serious issues where differing definitions can be justified (see also Table 1). Table 1 Definitions of military expenditures Possible items of military expenditures

Expendituresfor militaryforces and their support 1. Pay to soldiers, officers 2. Salaries of technicians, bureaucrats etc. within armed forces or connected to military organization 3. Medical services, tax benefits, social benefits to above (including relatives) 4. Pensions 5. Military schools, hospitals etc. 6. Current procurement expenditures on weapons (incl. arms imports) 7. Infrastructure construction, housing etc. 8. Operation and maintenance 9. Procurement of other goods 10. Military research and development Other expenditures with military/defense/strategicpurpose 11. Stockpiling of strategic goods 12. Mothballing of weapons, production lines, etc. 13. Arms production subsidies/conversion subsidies 14. Military aid to other countries 15. Contributions to international organizations (military alliances, UN peacekeeping, etc.) 16. Civil defense Expendituresfor past militaryforces/action 17. Veteran benefits, etc. 18. Service of war debts

Items h in definitions issued by NATO IMF UN

X X

X X

X X

X

X

X

X X X

X X

X ? X

X X X X

X X X X

X X X X

Xc C

X

X

X X X

X

-

X

X

X

X

Ch. 3:

49

World Military Expenditures

Expenditures on otherforces 19. Paramilitary/Gendarmery 20. Border/Customs Guards 21. Police

Xd Xd

Xd

Xd

Xd

Xd -

Xd -

X X

X

Incomes from 24. Military schools, hospitals, companies 25. Civilian use of military infrastructure 26. VIP transport 27. Sale of patents, know-how 28. Repayment of production subsidies 29. Military aid from other countries

Y Y Y Y Y -

Y Y Y Y Y -

Obligationsfor future spending 30. Procurements on credit

X

X

Chargeable to other accounts 22. Humanitarian/disaster relief 23.

UN peacekeeping

Y Y Y Y Y Xe

Sources: Ball (1988, pp. 402-04), UN (1977, 1981), NATO DPQ(90)FIN&ECON, Part 1 (reprinted in Deutscher Bundestag, Drucksache 11/7373, 28). b Symbols: X, should be included in military expenditures; -, should not be included in military expenditures; Y, should be budgeted as income. If managed and financed by defense organization. d When judged to be trained, equipped and available for military operations. Double-counting has to be considered before aggregations. a

2.1. Functional versus institutionalapproaches Aggregation of expenditures belonging to military functions, such as military deterrence, defense and warfighting, is preferable. But government budgets are predominantly organized along bureaucratic lines. These may not coincide with functional purposes. For instance, paramilitary forces may come under the budget of the Ministry of Domestic Affairs, military aid under the budget of Foreign Affairs, and procurement under the budget of special "Ministries of Machinery". Especially for unstable developing countries it may be very misleading to equate military expenditures with the budget of the Ministry of Defense (or War, or the Armed Forces). In such cases expenditures for internal "regime security", for police and paramilitary organizations can be very large [Ball (1988)]. Herrera (1994, p. 42) gives data for the not untypical example of Tunisia. Between 1985 and 1991, Ministry of Defense expenditures rose from 180 million to 224 million dinars. At the same time, expenditures for "public order and security", mostly contained in the budget of the Ministry of Interior, increased from 100 million to 238 million dinars. There are no clear limits to the functions of the military. Is civil defense such a function? Do pensions to veterans belong here? What is the difference between a

50

M. Brzoska

well-armed police force and lightly armed infantry? Is it the function of the military to provide for meteorological services, air traffic control, etc. as is the case in many countries? Is disaster relief something military? Is military aid to distant countries helpful for the national military purpose? In a number of countries, the armed forces run enterprises, often to produce weapons, but sometimes also to manufacture or sell civilian goods and services. There is no uniform way in which costs and incomes of such activities are budgeted. In the case of Indonesia, for instance, they seem to be fully outside the official budget [see e.g. FarEastern Economic Review (1994), June 9, p. 12]. A specific problem is presented by armed opposition groups. Their expenditures obviously are not part of the official state budget. Considering that in the early 1990s more than 40 such groups were active the total expenditures of armed opposition groups may be quite substantive. Unfortunately, there are no data available. 2.2. Direct, indirect and intangible costs and benefits Budgets are basically statements of direct costs. Sometimes they include indirect costs, such as destruction during training exercises, when compensation is paid out by military departments. But often the use of civilian infrastructure etc. is not recorded. Environmental impacts of the activities of armed forces are seldom budgeted. Likewise there is often no attribution of costs to civilians even though they use military equipment, personnel or infrastructure. Thus, it is frequent practice that Air Forces provide VIP support out of the military budget. Armed forces are often exempted from indirect taxes and customs duties, sometimes even direct taxes. Different rules exist for the compensation of the armed forces for disaster relief, search and rescue operations, medical treatment and the like. A special case of indirect costs results from the practice of conscription. No country budgets the opportunity cost of using conscripts with low pay instead of professional soldiers. The argument is made frequently in international comparisons of military expenditures, especially for instance by governments within the North Atlantic Treaty Organization (NATO). Just to suggest orders of magnitude: the US government could save up to 25% of its military expenditure if it substituted conscripts for volunteers in the manner the German government does. Conscription, on the other hand, is an indirect cost to the economy since the supply of labor is reduced. The opportunity cost of a conscripted soldier is difficult to measure, especially in economies with high rates of unemployment in the relevant age groups. Another problem is presented by military aid. This can be given directly to the armed forces of another country or channelled through the budget of the recipient country. Practices differ. For instance, German assistance to foreign military forces is budgeted by the Foreign Office but administratively handled by the Ministry of Defense. US security assistance, which comes from the Pentagon's budget, consists of new weapon systems, surplus weapons of the US armed forces, training in the USA and in foreign countries, credits and cash to foreign governments.

Ch. 3:

World Military Expenditures

51

2.3. Current versus comprehensive accounting The functional approach generally excludes the consideration of pensions, veteran benefits, etc. as well as debts incurred in prior wars. It can be argued that all social benefits should be excluded, as well as payments to children, relatives, etc. In practice, though, military bureaucracies often take care of such social benefits. In the other direction of time, a functional accounting of expenditures extending into the future, such as procurement on credit terms, would imply use of a depreciation stream. In budgetary practice this is not often found. Sometimes, actual expenditures are recorded for the years in which payments of interest and principal occurs. But in many cases debt service for procurement is aggregated with other debt service in a special budget account and not recognizable any more as expenditure connected with military activity. This is one of the reasons why detailed studies of military budgets in countries with resource gaps fail to find much expenditure on weapons [Ball (1984a, 1988)]. Military expenditures are especially difficult to ascertain in times of war. Countries at war are notoriously missing from data series on military expenditures. Also, there are typical costs of wars that are of interest such as destruction of property, loss of lives or loss of output. While not military expenditures proper, they are part of the larger picture of the economic consequences of military activity. 2.4. Standarddefinitions There may be good reasons for national practices of defining and budgeting military expenditures but they make international comparisons very difficult. Nicole Ball, who has made a detailed study of the budgets of a large number of developing countries has reported frequent changes in budgeting practices, so that intertemporal comparisons are also often very difficult [Ball (1984a)]. Standard definitions of military expenditures have been proposed by international organizations involved in data collection (Table 1). Three definitions are widely used, namely those by NATO, the International Monetary Fund (IMF) and the United Nations (UN). The NATO definition is first and foremost for internal purposes, but has also been adopted as the basis of other data series, such as those of the Stockholm International Peace Research Institute (SIPRI) and the United States Arms Control and Disarmament Agency (US ACDA). The NATO definition is a fairly comprehensive measure of the financial burden of the operation of armed forces. The IMF definition provides the basis for various data products published by the World Bank (such as World Development Report, World Tables) and the IMF (such as Government Finance Statistics). It is an element in the IMF Government Finance Statistics and therefore less specifically oriented towards military considerations than the NATO definition. The UN definition was developed specifically for the purposes of international comparison. From the 1970s, UN expert committees have worked on the problem of defining and reporting military expenditure. A detailed reporting scheme distinguishing major components of military expenditures and force groups (a total of 588 data entry

52

M. Brzoska

fields) was developed [UN (1977, 1981)]. The number of countries reporting data slowly increased, numbering close to 40 in the early 1990s [US ACDA (1993)]. The most detailed data have tended to come from the Western industrialized states for which disaggregated national data are publicly available anyway while only few Eastern or developing countries have reported even aggregate figures. The UN definition is also used in the OSCE framework. The UN definition is the most comprehensive of the definitions. In fact, it is so detailed that a number of governments justify failure to report with the inability to conform to the UN framework. The differences between the three standard definitions do not look very striking (Table 1). Nevertheless, figures can be quite different in cases where there are large pension payments or "strategic" expenditures. Differences between national and standard definitions can be quite large. German military expenditures for 1993 according to NATO criteria were put at DM million 63 854 while the German Defense Budget was only DM million 49 6022. Official Chinese defense expenditures for 1993 amounted to 42.5 billion yuan or $7.3 billion. Adding such categories as production subsidies to defense industries and military research and development expenditures on People's Armed Police and militia brings the figure up to something like $34 billion. Pension and demobilization expenditures, which fall under the NATO, but not the IMF definition add another $2 billion or so. In addition, the People's Liberation Army has large incomes from arms sales and commercial enterprises. These may be as high as $9 billion. Most of this income is not budgeted centrally and spent by the unit which generated it [all estimates follow Shambaugh (1994)]. In total, then, Chinese military expenditures following the NATO or UN definitions are something like six times the official figure. In addition to definitional differences, there are outright manipulations of data. It is widely assumed, though seldomly documented, that transactions are made outside of the budget, for instance through special accounts [Brzoska (1982), Ball (1984b), Sen (1992)]. The income for these funds usually comes from the exports of raw materials, as in the cases of Chile (copper) and Iran (oil) in the late 1970s. Sometimes, budgetary control is in such a bad shape that it is impossible to establish military expenditure data, as was the case in Argentina in the early 1980s. A World Bank team estimated that official data were underreporting true military expenditures by at least 50% in the 1970s and early 1980s [Herrera (1994, p. 25)]. 3. Intertemporal consistency Deflation is a major source of problems with international military expenditure data. There are only a few countries that have specific military price deflators [Sk6ns (1983),

Data from NATO Brief, April 1994, p. 33 and Bundesministerium der Verteidigung, WeiBbuch 1994, Bonn 1994, p. 76. Differences are due to payments from other budgets, such as contributions to the Gulf War, military aid and pensions. 2

Ch. 3: World Military Expenditures

53

UN (1986), Fontanel (1982, 1986)]. These are regularly larger than comparable civilian deflators. A major problem in the calculation of inflation in the military sector is the correction for increases in product quality. For some purposes, such as opportunity cost measurements, it is not necessary to have specific military deflators. No other country provides as many resources to the calculation of military deflators as the United States, at least if judged by what is known in the public sphere. The Pentagon publishes an annual series containing information on prices for weapons systems under procurement [DoD (annual)]; there are Pentagon aggregate estimates of deflators for weapon categories and force groups and there is a specific deflator for national account purposes, prepared from Pentagon raw data in the US Department of Commerce (monthly). Disaggregated military deflators are also available for a number of European NATO member states [Sk6ns (1983), UN (1986)]. Two basic methods are used to construct these deflators. One is the use of weighted civilian deflators, such as labor costs, purchases of consumer goods, purchases of investment goods, etc. with weights corresponding to the composition of military expenditures. Another is to record the changes in input costs of specific military items, such as pay for soldiers and weapon systems, and to correct these for increases in productivity and product quality. In both cases, the correction of quality change is a major problem. Reported military deflators are usually larger than GDP or industrial production deflators but this may be due to insufficient calculation of quality improvement [Sk6ns (1983)]. On the other hand, there are some reasons for assuming that military inflation might be above average. For instance, military inputs tend to come overproportionally from sectors with above average inflation, such as high salary groups and high-technology industrial sectors. Also, there is a tendency for "gold-plating" weapons and the arms market is often monopsonistic or, at best, oligopolistic [Albrecht et al. (1979), Albrecht (1982), Smith (1985)]. The rate of cost increase in reported military deflators - for military expenditures or weapon categories - above civilian inflation is sometimes called the "escalation rate". It is sometimes interpreted as a sign of secular overinflation in the military sector that has to be remedied through above average inflation compensation in military budgets. But it can also be argued that "escalation" reflects product improvement in the military sector above civilian average, otherwise the mix of inputs would not have been chosen [Smith (1985)]. Going back to a well-known study from the early 1970s comparing both inter- and intra-generational cost rises for a number of weapon systems, a "military escalation" rate of 4.5 per cent per annum is often quoted [US General Accounting Office (1973)]. But given the difficulty of comparing weapons for their quality, any such number is arbitrary. Based on such escalation rates, Norman Augustine, later chairman of Martin Marietta, predicted in the mid 1970s that the entire US defense budget would only buy one ship, one tank and one aircraft by 2054 [Augustine (1975)]. One additional problem with military deflators for weapon systems is that they tend to be distorted by procurement cycles. Unit weapon costs are often higher in

54

M. Brzoska

the early phases of procurement than in later phases. They also tend to vary with the number of items bought in a particular year, because of the high share of fixed costs. For aggregate measures, the usual quality correction problems of deflation arise. The quality of weapon systems of successive generations is notoriously difficult to value. Military deflators are of no relevance when the object is to measure the opportunity cost of military expenditures. Important here are the changes in the price levels of the goods and services foregone because of the allocation of expenditures to the military. Assumptions about the counter-factual have to be made: for instance that, if available for civilian purposes, resources would be spent on consumption, or in the same way as current GDP. In the first case, the consumer price deflator would be the proper one, in the second case, the GDP deflator. Inflation shapes another problem for international comparisons. Many countries have fiscal years that differ from financial years. With high rates of inflation, correct recalculations to a standard time frame, such as the calendar year, depend on some knowledge of the timing of expenditures within a fiscal year.

4. International comparisons Military expenditures are mostly expenditures in national currencies, for salaries and operation. Other parts, such as procurement of weapons, often are spent in international currencies, for instance US dollars. Basic data are reported in national budgets in national currencies. For purposes of international comparisons this presents problems of currency conversions. The most frequently used conversion rates for military expenditures - as for all other economic aggregates - are annual average market exchange rates as for instance regularly published in the IMF's International Financial Statistics. They basically reflect the comparative value of international economic transactions, such as trade, services, foreign investment and other capital transfers, but are also influenced by government interference in international transactions, expectations about future levels of interest rates and future political decisions. The more closed an economy, the higher is the probability that domestic prices diverge from international prices calculated with market exchange rates [Kravis et al. (1978)]. For economies with highly regulated foreign exchange markets, market rates are misleading or not even available. This was, for instance, the case in the former socialist countries. For these cases, a wide spectrum of exchange rates can be used, ranging from official rates set by governments to experts' estimates. The latter are also often used to estimate exchange rates of economies with very high rates of inflation. There are various efforts to address some of the problems associated with market exchange rates. For instance, the UN's price adjusted rates of exchange (PARE) and the World Bank Atlas conversion rate are "corrected" market exchange rates. PARE uses a base year, judged to be reflective of relative prices, which is made current with GDP implicit price deflators. For many of the entries in the World Bank Atlas, an

Ch. 3:

World Military Expenditures

55

average of the current rate and two rates using earlier base years is calculated. Neither of these measures solves the structural problems of the comparability of relative price levels. A radical departure from the international transaction approach is the use of Purchasing Power Parities (PPPs). PPPs are usually calculated from the comparison of common baskets of goods and services of pairs of countries with the US dollar as the common currency. The most extensive effort to calculate PPPs for a large number of economies and to estimate real exchange rates for almost all countries has been the International Comparison Project (ICP) based at the University of Pennsylvania [Kravis et al. (1978), Summers and Heston (1988)]. Kravis, Heston, Summer and associates have developed PPPs for GDP and parts of GDP such as government expenditures. PPPs are also used for the World Economic Outlook published jointly by the IMF and the World Bank, and by the OECD. The use of alternative currency conversion rates produces differing data sets of international military expenditures (Table 2). With exceptions, PPPs tend to value goods and services in poor countries higher than market exchange rates; therefore conversion of military expenditures with PPPs produces larger numbers for such countries [West (1987), Sen (1992)]. A rough estimate is that world military expenditures would be reported at about 125 per cent of their current level if converted with PPPs. The major increases in numbers are for countries such as India and China with large military expenditures and very low market exchange rates compared with PPPs. Such PPP conversions reflect opportunity cost measures of international military expenditures. For an output oriented measure of military expenditures, specific PPPs for the military sector have to be calculated. Some limited work has been done on specific military PPPs. A group of experts studied the issue in the framework of the UN Reduction of Military Expenditure exercise [UN (1986), Fontanel (1986), Cars and Fontanel (1987)]. Sufficiently detailed data are only available for a few OECD member countries. Since these countries have fairly similar price structures, differences with figures calculated at market exchange rates are not very large [UN (1986, Table 12)]. Building on a combination of the strictly military PPPs from the UN expert group and economic PPPs from the ICP project, Heston and Aten (1993) have estimated real military expenditures for a large number of countries (excluding socialist countries). The base year was 1980 with further estimations up to 1985. They first estimated the shares of personnel, international procurement (arms imports) and other expenditures from a number of sources. They then applied differing PPPs to these categories. For some procurement items, such as food and construction, civilian PPPs were used. For purchases of military weapons and salaries of personnel specific PPPs based on the UN expert group's work were estimated. Since it makes a large difference whether there is draft or voluntary conscription of what military expenditures buy in terms of output, they estimated full opportunity cost salaries for conscripts. Heston and Aten's military PPP estimates are different from data based on market exchange rates as well as from GDP PPPs (Table 2). One problem with the data is that

56

M. Brzoska Table 2 Comparison of military expenditures using alternative currency conversion methods', 1985

Expenditures in local currencies Country Currency Amount Algeria Bolivia Hungary Israel Nigeria Myanmar USA

m dinars m bolivanos m forint m new shekels m nairas m kyats b dollars

4793 95 37700 4055 976 1973 264

MER 953 226 752 3439 1094 233 264

Expenditures in million US dollarsb PPP PARE WBA RMP 926 351 n.a. 5081 1008 887 264

760 164 821 3465 699 282 198

953 216 752 3439 1091 234 264

1411 133 n.a. 5028 1160 200 264

Sources: SIPRI (1989, appendix 5A), UN (1993, Table A2), Heston and Aten (1993, Table 19.2). Abbreviations: MER, Market Exchange Rates [IMF according to UN (1993)]; PPP, Purchasing Power Parities [ICP Project according to UN (1993)]; PARE, Base Year 1970 absolute pares [UN (1993)]; WBA, World Bank atlas rates [World Bank according to UN (1993)]; RMP, Real Military Expenditures [Heston and Aten (1993)]; m, million; b, billion; n.a., not available. a

arms imports are a large component of military expenditures in the data Heston and Aten used 3 . Another is the number of data gaps that have to be filled with the help of regressions. Such data gaps are especially large and worrisome in the cases of East European and Middle Eastern countries and China, where more than 40% of global military expenditures occur. Until more data become available, military PPP estimates are not very reliable. Though they were not named this way, PPP type estimates of military expenditures were the predominant method to estimate data for East European socialist countries during the Cold War. The most important of these was the Central Intelligence Agency's (CIA) "building block method" for the Soviet Union. First, an extensive and labor intensive "count" of all physical items in the Soviet military was made. The physical numbers were than multiplied with prices for equipment and personnel. To these aggregates, cost estimates for other categories, such as research and development and operation and maintenance were added. US prices were used to produce a "dollar" estimate, rouble prices (from a variety of sources, including intelligence) for a "rouble" estimate. The CIA estimates were criticized for a variety of reasons. For instance, the use of either dollar or rouble estimates raises index number problems. Also, dollar and rouble estimates were surprisingly similar, given the large differences in pay for soldiers and workers in the arms industries in the Soviet Union and the USA [Holzman (1982)].

Most of these data are for industrialized countries. An extrapolation of these data to developing countries is problematic, as Ball (1984a, 1988) has demonstrated. 3

Ch. 3:

World Military Expenditures

57

A number of alternative approaches were proposed and discussed [Jacobsen (1987)]. This is now largely a matter of historical interest, although a lively debate continues [Gonchar (1994)].

5. Relative measures One way to avoid some of the mentioned problems of intertemporal and international comparisons is to use ratios of military expenditures in relation to other financial aggregates. Unfortunately, new problems arise. However, depending on the purpose for which military expenditure data are to be used, relative measures can be an attractive alternative to military expenditures in local or international currency. The most often used relation is military expenditures as a share of GNP or GDP. It is often called the "military burden measure" because it indicates the importance of the military sector in the spending of the national income. The "military burden" can be calculated without the use of deflators or currency conversion. It makes for easy comparisons between small and large countries as well as between rich and poor countries. As the name suggests, it is most useful for economic purposes. It has little relevance for purposes of measuring relative military strength and the like. The main problem with the "military burden" measure stems from the difficulties in establishing reliable data on national income. Especially in countries without developed markets, be they centrally planned or poor countries, there are large estimation problems. Since defense is a territorial concept, GDP, the production of goods and services in a territory, is preferable as a denominator over GNP, the income of the inhabitants of a territory. Another relative indicator based on a revision of GDP data has been proposed by Lock (1979). He argued that the "burden measure" was skewed between rich and poor countries since in poor countries a larger share of GDP is subsistence production and can, or at least, should not be used to finance military expenditures. He attempted to calculate military expenditures as a share of Hypothetical Maximum Surplus (HMS). This is defined as national income beyond minimum expenditures necessary to satisfy the population's basic needs. It was calculated by subtracting from GDP the product of the population with the dollar value for the absolute poverty level. For developed countries there is no large difference between the "burden measure" and military expenditures as a share of HMS. For developing countries, especially very poor ones, the HMS based indicator shows much higher values. For Africa in 1989, for instance, the share of military expenditures in HMS was 50% while it was only 4.4% of GDP [Brzoska (1994b, p. 54)]. There are important conceptual problems with HMS. There is no guarantee that people will receive a minimum income that will cover their basic needs, especially in countries with authoritarian governments. HMS is thus truly hypothetical, though educational in highlighting differences between developing and developed countries. In addition, there are large conceptual and data problems in the calculation of absolute

58

M. Brzoska

poverty levels. There is no regular publication of military expenditures as a share of HMS. There are also relative measures designed to be more reflective of military strength, defense capability and the like. None of these have received much attention. Military expenditures per square kilometer, per kilometer of border line, per inhabitant, etc., can all provide some insight into specific aspects of international comparison, but none are capable of offsetting the basic problems of military expenditures as a measure of military strength mentioned above. Military expenditures as share of GDP may be a good measure of the "burden" for the economy but not of the priority given to the military sector in decisionmaking over the use of available resources. Governments, as providers of resources for the military sector, only command part of the national income. If the purpose is to establish national political priorities, it makes more sense to use government expenditures as the denominator rather than the full national income. The share of military expenditures in government expenditures is a good indicator of the relative importance of the military sector in a given country. In practice, the value of this indicator is limited by conceptual and data problems for government expenditures. There are centralized and decentralized governments, governments that budget most expenditures for health, education and pensions directly and governments that leave such functions to private corporations, lean governments and governments that tax heavily and subsidize heavily, etc. The most often used denominator is the Central Government Expenditure series as provided by the IMF. Obviously, it is skewed in cases of very centralized, or decentralized, governments (compare France and Switzerland in Table 3). Recently, the United Nations Development Programme (UNDP) in its Human Development Report has highlighted another indicator, the relationship of military expenditures to expenditures on health and education. The indicator is not new: it can be found in early editions of US ACDA and in Sivard, and can easily be constructed from IMF data [IMF (annual); see also Abdallah (1995)]; in fact, it provides much of the rationale of Sivard's publication 4 . UNDP puts additional emphasis on the indicator by declaring health and education expenditures a measure of priority on "human development". The comparison of military expenditures with human development expenditures was judged to be a good indicator of priorities. Although there are obvious problems with comparisons based on health and education expenditures - the shares of private and government provision differ, labor costs are very sensitive to specific labor supply curves, for instance for doctors and teachers, and the marginal costs of health and education tend to increase with national income - the UNDP data had a remarkable policy impact in the early 1990s. Together with the burden measure it is, for instance, an important instrument in the discussion about economic aid [Ball (1993)].

4 Sivard was in charge of the US ACDA data in the early 1970s when it was decided to discontinue

the publication of social data.

Ch. 3:

59

World Military Expenditures

Table 3 Military expenditures as share of GDP and Central Government Expenditures (CGE) and in relation to social expenditures (selected countries, 1991 or latest year available) Country GDP Algeria Bolivia Brazil China Colombia France Hungary Iran Iraq Israel India Nigeria Myanmar (Burma) Soviet Union/Russia Switzerland Uganda USA World average Developed Developing a

Military expenditures in relation to (%) CGE Social expenditures

1.8

7 .7 b

2.4 1.3 3.3 2.6 3.6 2.1 5.7 74.9 8.1 2.7 0.8 5.6 10.3 1.9 2.6 4.9 4.2 4.1 4.5

12.8 2.1 b 17.4 20.5 8.1 4.1 b 24.9 n.a. 20.5 16.4 2.8b 13.8 43.1 b 19.4 21.8 19.6 15.0 14.2 18.3

Sources: US ACDA (1994), UNDP (1994).

11

57 23 114 57 29 18 38 271 106 65 33 222 132 14 18 46 37 33 60

Data for 1990.

Table 3 illustrates some of the differences between named measures for selected examples. Developing countries generally have higher shares of military expenditures in relation to CGE than developed countries. Part of the difference comes from comparatively low spending on health and education expenditures in developing countries as highlighted by the data in the third column of Table 3. Obviously, the variances in the data series increase with the decreases in the values of the denominators relative to military expenditures. While the general trend in the three series is similar, there are important differences due to country specifics with respect to CGE and social expenditures.

6. Data sources The general interest in military expenditures data, for whatever purpose, has impelled a number of institutions to publish such data series. Unfortunately, there is a large

60

M. Brzoska

amount of cross citation that is not always recognized by users. It is a time-consuming task to bring available data in line with a chosen definition and to fill in the gaps of poor or missing data. The UN military expenditure data exercise has already been mentioned. Unfortunately, only a limited number of countries respond to the questionnaires sent out to them annually. There is no check to the data applied by the UN. The data are reported in annual publications by the UN in national currencies. The IMF sends out questionnaires to member countries for the purposes of its various statistical series. Following a review in the early 1990s, questionnaires are now more specific and detailed. Member countries are under an obligation to provide statistics. The IMF provides governments with its definition of military expenditures but accepts the data sent in as long as it does not expressly conflict with the IMF definition or internal accounting logic. Because of bureaucratic practice in many countries, it is likely that the IMF under-represents expenditures outside of the budget of the Ministries of Defense. The data are published in a number of readily available IMF/World Bank sources, in most detail in the Government Finance Statistics Yearbook (GFSY). In 1993, military expenditure was reported by 94 of the 179 IMF member countries. For those countries involved in extended credit programmes, IMF economists usually have a closer look at the statistics, including military expenditure data. Such data, often more conformant to the IMF definition, do not find their way into published statistics if they are not explicitly authorized by national governments. They have been used in an aggregated way in studies by IMF authors, and published in aggregated form in at least one instance in the World Economic Outlook [Arora and Bayoumi (1993)]. In some of its standard series, such as UN National Account Statistics, the UN also publishes data supplied by national authorities. For special publications, such as the UNDP's Human Development Report, data are collected from a variety of sources, including IMF, Sivard and SIPRI. Agencies of the United States government collect, and for some countries of special interest, generate military expenditure data. The data are used for a number of purposes including Congressional requirements to report on countries receiving US aid. Agencies involved include US embassies around the world, the US Agency for International Development (AID) for developing countries, the CIA and the Defense Intelligence Agency (DIA) for countries such as the former Soviet Union, Russia and China. The data are published in a number of places, such as reports to Congress that are difficult to access. The most useful of the US government's publications is the annual World Miltary Expenditures and Arms Transfers prepared by the US ACDA. It is the most comprehensive of all available data sources. The data presented in this publication come from a number of sources within the US government but also from the IMF [Fei (1979), Ball (1988)]. There are some peculiarities. For instance, for a number of countries arms imports, as estimated by US government sources, are added to estimates of military expenditures. Although it is probably true that a large number

Ch. 3:

World Military Expenditures

61

of governments do not record any expenditures on arms imports in official military expenditures, there is a danger of double counting for military aid and a mixture of actual and estimated expenditure [see below; for a similar procedure see Hewitt (1993)]. Also, such procedures are not in line with the NATO definition that is reportedly chosen as the basis. Finally, the criteria for selection of countries for which the procedure is adopted are not clear. US ACDA reports data in current and constant US dollars. The constant dollar series is created by converting national data deflated with a CPI inflation index with market exchange rates for some base year. The current dollar series is calculated by applying the US CPI deflator. National and US deflators are thus mixed for the current dollar series. At SIPRI information on military expenditures is collected from a large number of publicly available sources, such as newspapers, special military journals, government publications and budget data. The preferred definition is a simplified NATO definition but SIPRI only has very limited means to correct national data [Huisken (1973), Blackaby and Ohlson (1982), Sen (1992)]. For a number of countries, published IMF data are used, but still data have to be estimated for some countries. Because of misgivings about some of the data, estimation of a figure for the Soviet Union was discontinued in 1987. The data are presented both in national currencies at current prices and at constant prices, using CPI deflators and market exchange rates. Ruth Sivard has for some time been publishing brochures including military expenditure data. Her data are mostly drawn from other sources, such as published IMF data, SIPRI and US ACDA, but for many countries, including the Soviet Union, estimates are made. Each publication includes aggregate data for a number of years and detailed data in US dollars for one year. The London based International Institute of Strategic Studies (IISS) presents data on military budgets in its Military Balance. The figures are usually projected budget expenditures for the current fiscal year as reported to the IISS by national governments. There is no effort to bring them to a common definitional standard or to check whether projections correspond with actual expenditures. Differences among the sources can be large in individual cases, as shown by a random sample in Table 4. Among the reasons for the differences are deflation methods, the use of currency conversion, the extent of data manipulation to reflect definitions, corrections for differences in fiscal and financial years and others. Rough trends, for instance for regional totals, are similar [Brzoska (1982), Herrera (1994), Happe and Wakeman-Linn (1993, 1994)]. But rankings of individual countries, growth rates of military expenditures for individual countries and levels of military expenditures are source dependent. An example is the number for global military expenditures (Table 5). The main reason for the difference between US ACDA and Sivard is the valuations of Soviet/Russian and Chinese military expenditures. US ACDA, based on CIA estimates, puts very high values on these, while Sivard reports very low numbers.

62

M. Brzoska

Table 4 Military expenditures - differences among international sources (millions constant US dollars of 1988) Country

Year

SIPRI

US ACDA

IISS

GFSY

Sivarde

Algeria

1989 1991 1989 1991 1989 1991 1989 1991 1989 1991 1989 1991 1989 1991 1989 1991 1989 1991 1989 1991 1989 1991

1045 1177 181 178 n.a. n.a. 1944 1354 3830 3909 9609 9033 247 275 292 429 n.a. n.a. 290 593 264383 628189 671805 111 81

1084 644 145 103 42900 43 500 3888 1112 5975 4402 6412 6129 113 215 754 1129 288 000 221000 290938 247197 1023445 915500 122 115

929 734 n.a. 139 6350 16 200 792b 660 4961' 4543d 8500 6970 264 208 325 338 111 500 114700 277898C 247146 744628 825727 116 103

n.a. n.a. 80 103 n.a. n.a. 582 n.a. 5630 4302 8353 6100 n.a. n.a. 266 820 n.a. n.a. 278 440 248220 496887 423456 72 45

1184

Bolivia China Hungary Israel India Nigeria Myanmar (Burma) Soviet Union/Russia USA Totals, 1989 Totals, 1991 Number of countries in source

124 13 140 556 5429 9905 258 539 78300 295 770 706088 132

Sources: SIPRI (1992), US ACDA (1994), Sivard (1993), IMF GFSY (1994), IISS (1991-1993), Happe and Wakeman-Linn (1994); deflation rate: US ACDA (1994). Abbreviations: US ACDA, US Arms Control and Disarmament Agency; SIPRI, Stockholm International Peace Research Institute; IISS, International Institute for Strategic Studies; GFSY, International Monetary Fund Government Finance Statistics Yearbook. d Fiscal Year, including US military aid. b As announced by government. Fiscal Year. C 1990 or earlier, as available. a

Table 5 World military expenditures Data sourcea 1980 Sivard US ACDA SIPRI

549 850 777

Billion US dollars, 1987 prices 1989 1990 690 986

676 960

b

_b

Sources: Sivard (1993, p. 42), US ACDA (1994, Table 1), SIPRI (1986, p. 231). b No world total estimated after 1986. a

1991 655 882

Ch. 3:

63

World Military Expenditures

Table 6 World military spending relative to GDP, 1991 a (in percent of GDP, averages over countries in sourceb) Region/country World total Western industrial countries

US ACDA

SIPRI

IISS

GFSY

WEO

4.9 2.7

3.3 2.7

4.1 2.4

3.1 2.9

3.4c 3.3

Eastern Europe Soviet Union/Russia

4.6 10.3

1.6 n.a.

3.6 n.a.

3.3 n.a.

3.8 6.3

Developing countries Asia Middle East North Africa Sub-Saharan Africa Western Hemisphere

5.6 3.4 18.8 4.3 3.3 1.8

3.8 3.2 10.0 3.6 1.8 1.6

4.8 3.8 12 4.4 3.7 2.5

3.2 2.5 8.4 2.0 2.7 1.3

2.6 2.4 7.2 2.2 d 1.1

a Sources: Arora and Bayoumi (1993), Happe and Wakeman-Linn (1993). Abbreviations: US ACDA, US Arms Control and Disarmament Agency; SIPRI, Stockholm International Peace Research Institute; IISS, International Institute for Strategic Studies; GFSY, International Monetary Fund-Government Finance Statistics Yearbook; WEO, International Monetary Fund/World Bank, World Economic Outlook. b Data for WEO are for 1992. Own estimate including Russia. d Africa total.

Large differences are also visible in the data of military expenditures as share in GDP (Table 6). The relations between the various world regions are similar, though not identical. Available data sources have to be used with great caution. 7. International arms transfers For developing countries, more than half of the value of procurement is the import of weapons and other military goods. As mentioned earlier, the treatment of arms imports in many countries' military expenditures is difficult to track, leading some experts into the practice of adding arms imports to military expenditures [US ACDA, Hewitt (1993)]. Unfortunately, available data on the arms trade are not well suited for such correction of military expenditure data, even in cases where national budgeting practice would allow such adjustment. Available data - from SIPRI, US government and the UN - are output oriented. They are constructed with the primary goal of measuring the transfer of military power [Brzoska (1981)]. SIPRI researchers collect public information on transfers of major weapons 5 and then multiply them with a price (called SIPRI unit

5 Aircraft, armored vehicles and artillery, ships, missiles and stand-alone radar and guidance systems.

64

M. Brzoska

value) that is considered to reflect values regardless of whether this price is actually paid or not. The US government collects data on individual arms transfer deals, recording the number of items transferred and prices paid. But if actual prices seem to diverge from imputed values or no money is paid, average prices from similar deals are imputed. The data are regularly published in the reports by US ACDA and the Congressional Research Service [Grimmett (annually)]. At the UN, a voluntary arms transfer register has been in operation since 1993. A large percentage of exporters and a fair amount of importers report transfers of major weapons. Some cross-checking is therefore possible. No financial data are published [Laurance, Wulf and Wezeman (1993)]. Especially during the Cold War, when large numbers of weapons were transferred for free or at subsidized prices, there were large discrepancies between arms transfer data as published by international sources and what was actually paid for imported weapons. In the case of the Soviet Union, for instance, less than one third of income from arms sales reported in standard sources actually occurred. The rest was delivery on credit terms where repayment was insecure or not expected at all [Brzoska (1983, 1994a)]. On the other hand, because of the widespread practice of purchasing weapons on credit, a sizeable amount of military induced debt was accumulated in the late 1970s and 1980s. Debt service on weaponry imported earlier may be higher since the late 1980s than the cost of current arms transfers [Brzoska (1992)]. These costs are regularly not part of reported military expenditures. If they were included, arms transfer data would be increased by one fourth to one third [Brzoska (1994a)].

8. Conclusion For academic analysis, military expenditures may be the best indicator of military activity but available data have to be used with a great deal of caution, for instance in regression analysis [Goertz and Diehl (1986)]. For interpretation of all kinds it has to be taken into account that there is no general agreement on the concept nor its translation into definitions. Data accuracy and validity is not sufficient to support weak statistical results. Exceptions from this statement are only acceptable in the case of OECD countries. The data are good for some countries outside of the OECD, such as India, but very poor for a large number of others. The ready availability of data in some well-known sources should not detract from the numerous data problems. None of the institutions collecting and openly publishing data are able to put much effort into correcting military expenditures data. Compared with the resources available to national or international statistical offices, the labor input is marginal. At SIPRI, Sivard, and US ACDA, timely publication of the series depends on the initiatives of single persons. Unfortunately, the UN has not succeeded in bringing a sufficient number of countries to report detailed data and the IMF does not publish its more standardized series.

Ch. 3:

World Military Expenditures

65

The available data are more suited for trend analysis than for international comparisons. The range of definitions is large, and corrections according to a standard definition are limited. Currency conversion problems are serious. Deflation is another large problem. Above all, for the majority of countries, the availability of data remains as the outstanding issue. Military expenditures are still a highly political issue in many countries, although the most controversial case, Soviet military expenditures, has lost its relevance. Many governments continue to treat military expenditures as a national secret even if they do not engage in manipulation of such data. It is to be hoped that the end of the Cold War will allow a continuation of the UN's efforts at reporting military expenditures or another major effort by another international organization. The success of the arms transfer register is a good sign. More transparency of military expenditures is also one of the goals of recent activities by donors of development aid [Ball (1993), Kan (1993)].

References Abdallah, W.K., 1995, GFS Yearbook highlights trends in fiscal balance, military spending, IMF Survey, March 6. Albrecht, U., 1982, Ristung und Inflation, in: P. Sonntag, ed., Rfistung und Okonomie (Haag und Herchen, Frankfurt) 209-236. Albrecht, U., et al., 1979, A short research guide on arms and armed forces (Croom Helm, London). Arora, VB., and T.Bayoumi, 1993, Econbmic benefits of reducing military expenditure, World Economic Outlook, October, 104-112. Augustine, N., 1975, One plane, one tank, one ship: Trend for the future?, Defense Management Journal, April, 34-40. Ball, N., 1984a, Third world security expenditure: A statistical compendium, FOA Report C 10250-MS (F6rsvarets Forskningsanstalt, Stockholm). Ball, N., 1984b, Measuring Third World security expenditure: A research note, World Development 12, 157-164. Ball, N., 1988, Security and economy in the Third World (Princeton University Press, Princeton, NJ). Ball, N., 1993, Pressing for reform (Overseas Development Council, Washington, DC). Blackaby, E, and T. Ohlson, 1982, Military expenditure and the arms trade: Problems of data, Bulletin of Peace Proposals 13, 291-308. Brzoska, M., 1981, The reporting of military expenditures, Journal of Peace Research 18, 373-384. Brzoska, M., 1982, Arms transfer data sources, Journal of Conflict Resolution 26, 77-108. Brzoska, M., 1983, The military related debt in Third World countries, Journal of Peace Research 19, 271-277. Brzoska, M., 1992, Military trade, aid and debt, in: G. Lamb and V.Kallab, eds., Military expenditure and economic development. A symposium on research issues (World Bank Discussion Papers 185, Washington, DC) 79-112. Brzoska, M., 1994a, The financing factor in military Trade, Defense and Peace Economics 5, 67-80. Brzoska, M., ed., 1994b, Militarisierungs- und Entwicklungsdynamik (Deutsches Ubersee-Institut, Hamburg). Cars, H.C., and J. Fontanel, 1987, Military expenditure comparisons, in: C. Schmidt and F. Blackaby, eds., Peace, defense and economic analysis (Macmillan, London) 250-265.

66

M. Brzoska

DoD (US Department of Defense), annual, Programme acquisition costs by weapon system (US Department of Defense, Washington, DC). Fei, E.T., 1979, Understanding arms transfer and military expenditure data problems, in: S.G. Neumann and R.E. Harkavy, eds., Arms transfers in the modem world (Praeger, New York) 37-46. Fontanel, J., 1982, La comparaison de dpense militaire, Defense Nationale, 38, 107-121. Fontanel, J., 1986, A note on the international comparison of military expenditures, in: C. Schmidt, ed., The economics of military expenditures (Macmillan, London) 29-46. Goertz, G., and P. Diehl, 1986, Measuring military allocations: A comparison of different approaches, Journal of Conflict Resolution, 30, 553-581. Gonchar, K., 1994, Military spending in the USSR: A reconsideraton, Journal of Peace Research, 31, 219-228. Grimmett, R., annual, Conventional arms transfers to the Third World (Library of Congress, Congressional Research Service, Washington, DC). Happe, N., and J. Wakeman-Linn, 1993, Military expenditure and arms Trade: Alternative data sources, Part I, Peace Economics, Peace Science and Public Policy 1, 3-38. Happe, N., and J. Wakeman-Linn, 1994, Military expenditure and arms trade: Alternative data sources, Part II, Peace Economics, Peace Science and Public Policy 2, 10-23. Herrera, R., 1994, Statistics on military expenditure in developing countries: Concepts, methodological problems and sources (OECD Development Centre, Paris). Heston, A., and B. Aten, 1993, Real world military expenditures, in: J. Brauer and M. Chatterji, eds., Economic issues of disarmament (New York University Press, New York) 257-287. Hewitt, D.P., 1993, Military expenditures worldwide: Determinants and trends, Journal of Public Policy 12, 105-152. Hildebrandt, G.F., 1990, Services and wealth measures of military capital, Defence Economics 1, 159176. Holzman, F.D., 1982, Soviet military spending: Assessing the numbers game, International Security 6, 78-101. Huisken, R., 1973, The meaning and measurement of military expenditure (SIPRI Research Report, Stockholm). IISS (International Institute for Strategic Studies), annual, Military Balance (IISS, London). IMF (International Monetary Fund), monthly, International financial statistics (IMF, Washington, DC). IMF (International Monetary Fund) GFSY, annual, Government finance statistics yearbook (IMF, Washington, DC). Jacobsen, C.G., ed., 1987, The Soviet defence enigma. Estimating costs and burdens (Oxford University Press, London). Kan, S., 1993, Military expenditures by developing countries: Foreign aid policy issues, Report 93-999F (US Library of Congress, Congessional Research Service, Washington, DC). Kravis, I.B., et al., 1978, International comparisons of real product and purchasing power (The Johns Hopkins University Press, Baltimore, MD). Laurance, E.J., H. Wulf and S. Wezeman, 1993, Arms watch (Oxford University Press, London). Lock, P., 1979, New international order and armaments, Vierteljahresberichte der Friedrich-Ebert-Stiftung, 77, 14-28. Sen, S., 1992, Military expenditure data for developing countries: Methods and measurement, in: G. Lamb and V Kallab, eds., Military expenditure and economic development. A symposium on research issues (World Bank Discussion Papers 185, Washington, DC) 1-18. Shambaugh, D., 1994, World military expenditures, China, in: SIPRI yearbook 1994, 441-448. SIPRI (Stockholm International Peace Research Institute), annual, SIPRI yearbook (Oxford University Press, London). Sivard, R.L., annual, World military and social expenditures (World Priorities, Leesburg, VA). Sk6ns, E., 1983, Military prices, in: SIPRI yearbook 1983 (SIPRI, Stockholm) 195-211.

Ch. 3:

World Military Expenditures

67

Smith, R., 1985, Defence costs, in: J. Roper, ed., The future of British defence policy (Gower, London) 143-158. Summers, R., and A. Heston, 1988, A new set of international comparsions of real products and prices: Estimates for 130 countries, The Review of Income and Wealth 34, 1-25. UN (United Nations), 1977, Reduction of military budgets: Measurement and international reporting of military expenditures, Report of the Secretary-General, A/31/222/Rev. 1 (UN Centre for Disarmament, New York). UN (United Nations), 1981, Reduction of military budgets. International reporting of military expenditures, Report of the Secretary-General, A/35/479 (UN Centre for Disarmament, New York). UN (United Nations), 1986, Reduction of military budgets. Construction of military price indexes and purchasing power parities for comparison of military expenditures, Report of the Secretary-General, A/40/421 (UN Department for Disarmament Affairs, New York). UN (United Nations), 1993, Trends in international distribution of gross world product (Department for Economic and Social Information and Policy Analysis, Statistical Division, New York, Special Issue of National Accounts Statistics, ST/ESA/STAT/SER.X/18. UNDP (United Nations Development Programme), annual, Human development report (Oxford University Press, London). US ACDA (US Arms Control and Disarmament Agency), annual, World military expenditures and arms transfers (US Government Printing Office, Washington, DC). US Department of Commerce, monthly, Bureau of Economic Analysis, Survey of current business (US Department of Commerce, Washington, DC). US General Accounting Office (GAO), 1973, Cost growth in major weapon systems (US General Accounting Office, Washington, DC). West, R., 1987, Improved measures of the defense burden in developing countries, in: S. Deger and R. West, eds., Defense, security and development (Francis Pinter, London) 19-48. Wiberg, H., 1984, Measuring military expenditures: Purposes, methods, sources, Cooperation and Conflict, 18, 161-177. World Bank, annual, World development report (Oxford University Press, London).

Chapter 4

THE DEMAND FOR MILITARY EXPENDITURE RON SMITH Birkbeck College, London

Contents Abstract Keywords 1. Introduction 2. The framework 3. Interpretation 4. Theoretical specification 5. Data 6. Econometric issues 7. Some examples 8. Conclusion References

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

70 70 71 71 74 76 78 80 83 84 85

70

R. Smith

Abstract This chapter provides an analytical survey of models explaining the level of military expenditure, emphasizing empirical econometric models based on optimizing behavior. After setting out the standard neo-classical model in which nation states are represented as rational agents maximizing a welfare function depending on security and economic variables subject to a budget constraint, the chapter considers: criticisms of this model, alternative theoretical specifications, data and measurement issues, alternative econometric approaches and reviews selected studies.

Keywords demand, military expenditures, econometric methods, optimizing models

Ch. 4:

The Demandfor Military Expenditure

71

1. Introduction This chapter provides an analytical survey of the large literature on models which attempt to explain the level of military expenditure. It emphasizes empirical econometric models based on optimizing behavior, but tries to confront some of the more fundamental methodological issues raised by these models. Section 2 sets out the standard neo-classical model in which nation states are represented as rational agents who maximise a welfare function depending on security and economic variables, subject to a budget constraint which includes military spending, and a function which determines its security in terms of its own and other countries military forces. The state then balances the welfare benefits of extra security derived from military expenditure against its opportunity costs in terms of foregone civilian output, and solves this optimization problem to give a derived demand for military spending. Section 3 considers the fundamental questions about this approach. Firstly, who (median voter or rational state) is maximizing what and why? Secondly, can this model be regarded as a plausible or useful summary of the political and bureaucratic process actually generating military spending. Third, is the fundamental assumption of this framework that military expenditure is motivated by security, rather than domestic political or economic concerns, justified? Section 4 considers the detailed theoretical specification of such models. The issues discussed include the choice of welfare function, budget constraint, and functional forms; the role of prices; the form of any dynamics and the arguments of the security function. These issues are also discussed in the arms-race and alliance chapters. Section 5 examines the data and measurement issues. Section 6 examines alternative econometric approaches used to model military spending. In particular, it contrasts theory based with data based models and single equation with systems estimators. Section 7 reviews selected studies and Section 8 draws some conclusions. Many of the issues discussed in this chapter are also covered in chapter 3 of Sandler and Hartley (1995).

2. The framework The standard neo-classical model of the demand for military expenditure assumes that there is a national state which maximizes welfare W, which is a function of security S, economic variables such as total consumption C, population N, and other variables, say ZW, which parameterize shifts in the welfare function (e.g. the politics of the party in power). For instance: W = W(S, C, N, ZW).

(1)

Population is included because while defense may be a public good, consumption probably is not, so it is per capita consumption that matters for welfare. This

72

R. Smith

maximization is subject to a budget constraint and a security function. The simplest budget constraint is Y = pcC +pmM,

(2)

where Y is nominal aggregate income and Pm and Pc are the prices of real military spending M and consumption C. Security can be thought of as perceived freedom from threat of attack. Like utility or welfare, security is unobservable to the econometrician and is replaced by some set of quantifiable variables, such as the military forces of the country and other countries, together with other strategic variables, ZS, which parameterize shifts in the security environment S = S(M, M1, ... , M,, ZS).

(3)

Some of the other countries may be allies in which case there is a "spillin" from their expenditure which raises your security, others may be potentially hostile in which case their forces pose a threat to your security. There are a variety of ways that the forces of the allies may be aggregated: for instance total alliance military effectiveness may depend on the sum of the forces; the minimum force (the "weakest link in the chain" determines effectiveness) or the maximum force (the "best shot" determines effectiveness). Conybeare, Murdoch and Sandler (1994) test between alternative models of the aggregation technology. The forces of each country will be determined in a similar way. This joint determination of the set of forces is the focus of much arms race and alliance literature. Here we will focus on the partial equilibrium determination of one country's forces, given those of the others. This optimization problem is then solved to give a derived demand for the level of military spending: M = M(pm/P, Y, N, M1 ... , M, ZW, ZS).

(4)

For estimation, this demand function will need to be supplemented by auxiliary assumptions about the appropriate forms of the functions. To be more specific, consider a very simple example in which we ignore N, ZW and ZS and assume that the welfare function is of the Stone-Geary form: W = a log(C) + ( - a) log(S).

(5)

A minimum level of consumption is often allowed for C*, with C- C* appearing instead of C in the welfare function. Let us suppose that the state is not aggressive but faces one threatening neighbor, with military forces M 1, and neither has allies. Its security is given by: S = M -M* =M - (o

+ M), 1

(6)

where M* is the forces the country would need to resist an attack by its neighbor. M* is determined partly by fixed elements unrelated to the opponents forces (for instance, if

Ch. 4:

73

The Demandfor Military Expenditure

there are natural strategic defenses &o would be negative; if surprise attack would give the neighbor an advantage io0 would be positive) and partly by the size of the opponents forces with 31, the Lanchester coefficient, measuring the relative effectiveness of the forces in combat [see, for instance, Anderton (1990)]. Using the budget constraint above, the Lagrangean is then: L = a log(C) + (1 - a) log(M - M*) +2,(Y -pcC -pmM).

(7)

The first order conditions are: dL aC

a C

L

1-a

MM

M -- M*

dL

i.e.

a C =-A,

2M PM =0 O;

i.e.

pc = 0;

Y

- --

(8) --a

C-PmM =

M=

p

+M*,

.

(9) (10)

Eliminating the Lagrange multiplier using 1

Y -pmM*

(11)

gives the two demand equations of the Linear Expenditure System (LES) M=

Pm

Y + a(/3o + fi1M 1),

(12)

and a C = -[Y -m(fi0 + Pc

Mi)]

(13)

which determine real consumption and military spending as a function of income, prices, preference parameters (a), strategic parameters (/3i) and the other countries military spending. This system satisfies the usual requirements of adding up, homogeneity, symmetry and negativity of the Slutsky matrix. Much of the interest in demand systems has focussed on testing these implications of theory [see, for instance, Chapter 3 of Deaton and Muellbauer (1980)]. Dunne, Pashardes and Smith (1984) show that a complete dynamic demand system for four categories of UK government consumption including military spending does satisfy homogeneity and symmetry. However, such testing is quite rare, partly because of the lack of data on military prices. The model can be easily generalized to allow for the more realistic assumption that security depends on stocks of military forces, rather than flows of military

R. Smith

74

expenditures. The stock of military forces (equipment plus the human capital embodied in experienced personnel) can be defined as the depreciated sum of past expenditures (14)

Kt = (1 - 6)Kt- 1 +Mt,

where 6 is the depreciation rate, a parameter to be estimated. Depreciation here is a peace-time concept, not destruction during war. If there is a similar stock measure for the other country, security is then given by St = Kt - (o + 31K,) = M, - Mt,

(15)

where (16)

Mt = fPo+/J31[Mt + (1 -- )K 1tl] This gives a demand function a)( Mt = (I - P+

(17)

aM.

If stocks are not observed, which is the usual case, they can be substituted out using K,

(18)

M 1- (1- (5)L'

where L is the lag operator such that Lx, =xt-l. Multiplying through by (1-(1-6)L) and rearranging gives M,= a3 0 +(1 a)

+ aiMlt + (1 - a)(l -)

-

,

(19)

which does not include the unobserved stocks, but compared to Equation (12) introduces dynamics into the demand equation (lagged values of military expenditure and output are included on the right hand side). The rate of depreciation can be estimated from this equation and one could test whether stocks or expenditures should appear in the security function by testing whether 6 = 1 (i.e. that the coefficient of the difference between lagged military expenditure and income was insignificant). One could also test the theoretical implication that the coefficients of the lagged variables should be equal and opposite. Of course, since this model is very restrictive, rejection of the theoretical restrictions could be a result of various other forms of misspecification. 3. Interpretation The framework set out in the previous section has been widely used, but it raises a series of fundamental questions. Military expenditures are chosen by governments in the light of a multitude of specific circumstances and influences. Thus, one

Ch. 4: The Demandfor Military Expenditure

75

might not expect those choices to satisfy any simple structural relations of the sort set out above. To justify the possibility of such structural explanation requires an argument that governmental choices are constrained in particular ways which limit their potential freedom and make them behave in predictable ways. It also requires that those constraints can be empirically or theoretically identified, allowing us to make conditional predictions of the level of military expenditure. While one can hypothesise a large number of potential sources of such constraints, their plausibility needs to be justified. In addition, the constraints, while real, may not be binding. Clearly, a government cannot spend more on military expenditure than is available to it in real resources from its own economy and abroad. But this is rarely a binding constraint which determines choices. One source of constraint, for democratic governments, is the electorate, and a number of studies [e.g. Dudley and Montmarquette (1981)] have interpreted the demand for military expenditure as that of the median voter. If there is an onedimensional issue space (e.g. for the level of military expenditure) in which the amount is determined by a direct vote, then the median voter is decisive. If there are two parties, who compete only along this single dimension and the distribution of preferences is unimodal and symmetric and if each voter votes for the party closest to his or her preferred position, then electoral competition will drive the parties to the position favored by the median voter. To make these, somewhat restrictive, assumptions operational it is also usually assumed that preferences are aligned with income. Then the person with median income determines the level of military spending. While there are some areas where median voter theory probably does have explanatory power (e.g. in Swiss Cantons where direct referenda are common) it seems questionable that it is applicable to the determination of national military expenditures. The issue space in most elections is highly multi-dimensional and the preference distribution is as likely to be bimodal (hawks and doves) as unimodal. Murdoch, Sandler and Hansen (1991) use military expenditure by NATO allies to test between median voter and oligarchy models. There is also a range of political science literature which examines the role of electoral cycles in influencing the defense budget [see, for instance, the articles in Part I of Mintz (1992)]. Economists have emphasized the constraints on choice imposed by rationality. Given the ends to which action is directed (the welfare function) and the means available (the budget constraint), rationality restricts choice in ways which may make it predictable. Such "rational actor" models of military expenditure are subject to many types of criticism. Governments are not unified rational actors but complex coalitions operating in a political and bureaucratic environment. Inter-service rivalry, lobbying, and classstruggle may all play a role in the determination of defense policy. Smith (1977, 1978) and Griffin, Wallace and Devine (1982) discuss class interests; Sandler and Hartley (1995) review a range of other bureaucratic and interest group models. The choices which arise from such bargaining and log-rolling are unlikely to satisfy rationality requirements. Even if governments could be regarded as individual agents, rational action may not be well defined given the complexity of the problems and the constraints

76

R. Smith

on information processing capability. In any event, any optimal strategies would be conditional on objectives and opportunities about which the observer is poorly informed, making prediction difficult. For instance, a variety of people have argued that the main factor driving US military expenditure is the need to offset a tendency to stagnation [see, for instance, Cypher (1987), Pivetti (1992) and the response to Pivetti by Smith and Dunne (1994)]. This is essentially a "rational" model of military Keynesianism, but because the objective function is different from that used in neo-classical, or, for that matter, traditional Marxist models, it has quite different predictions of the determinants of military spending than the model of the previous section. There are also non-optimizing, but structural, models, based on bureaucratic, interest-group, or satisficing behavior. In these, governments are constrained to follow certain types of standard operating procedures or rules of thumb, which introduce a degree of predictability. Many quantitative political science models are of this sort [e.g. Ostrom (1978), Cusack and Ward (1981), Kamlet and Mowery (1987)]. But these constraints may not apply to military spending directly. For instance, Stoll (1992) argues that the use of weapon counts and simple rules of thumb can account for much of behavior that would not be revealed using optimizing models and total military expenditures. One implication of these models is incrementalism: the main determinant of this years budget is last years budget [e.g. Rattinger (1975)]. However, a range of other models, including optimizing ones discussed below, would also suggest that last years expenditures should be a determinant of this years spending. The literature on organizational reaction models is critically reviewed by Williams and McGinnis (1992a). While it may be plausible to assert that rationality, bureaucracy or political pressures impose constraints which reduce the freedom of choice of governments in setting their military expenditure and thus provide structure and predictability on choices, it is less plausible to assert that these constraints are constant. But variation in the constraints will produce structural instability and loss of predictability. Some of the more general issues involved in the dynamics of arms acquisition are discussed in Gleditsch and Njolstad (1990). Given these objections it would be difficult to argue that the demand functions used to explain military expenditure are more than a simple metaphor for a poorly understood process. However, simple metaphors can be useful. Making the theoretical model more formal and explicit ensures logical consistency within the argument, allows interpretation of the parameters, identifies crucial assumptions, and enables the results to be integrated into a wider body of knowledge. Models are deliberately simplified representations of reality designed to provide a framework for thought, within which the data can be organized and the questions of interest posed and this is the way that we will use them in the rest of this chapter. 4. Theoretical specification Within the optimizing framework, the main issues are the forms chosen for the budget constraint, the welfare function and the security function. A wide variety of forms

Ch. 4:

The Demandfor Military Expenditure

77

have been used. McGuire (1982, 1987) uses a Stone-Geary objective function to give a Linear Expenditure System (LES) of the sort used in Section 2 and allows for the effect of military aid in the Israeli Budget constraint. Smith (1980, 1987) uses a Constant Elasticity of Substitution welfare function and a Cobb-Douglas security function and estimates the first order condition rather than the demand function. Dunne, Pashardes and Smith (1984) estimate a complete system of equations for the components of government expenditure, use a Deaton-Muellbauer (1980) flexible functional form and write the budget constraint in terms of total government expenditure rather than national income. Okamura (1991) uses a generalized indirect translog utility function, which gives the Linear Logarithmic expenditure system, of which the LES is a special case. In addition, there are a large number of different ways of specifying how threats and spillins from allies will impact on your own security and thus your military spending and how decisions are made by groups of allies. For instance, Sandler and Murdoch (1990) test between Nash-Cournot and Lindahl behavior by NATO allies. Probably the most widely applied demand function is that introduced in Murdoch and Sandler (1984, 1986), who use a general functional form in their theoretical work and a linear (or log-linear) demand function for estimation for groups of allies. They estimate a system of demand equations for the US, the UK, France, Belgium, the Netherlands, Denmark, Norway, West Germany and Italy for 1961-1979, allowing for cross-equation covariances. The equations have the form Mit = ai + Yif +(Yi + 2iDt)Ai,t- 1 + it, N Ai= A: ,mjMi,

(20) (21)

j=1

where A is the alliance spillin from the other NATO countries, D is a dummy that is zero before 1974 and one after to allow for the shift to flexible response. This is an example of the strategic shift variable, ZS, of Section 1. Income had a positive effect and the change in doctrine caused a reduction in the spillins, which differed between countries. They also found no significant difference between the equations for Britain and France, a result also found by Smith (1989) with a very different model. Many of the models that have been estimated have been static, though there are many ways that dynamics can be introduced, including setting the model up in terms of stocks as in Section 2. A natural generalization is to allow the agent to maximise a forward looking function, which is the discounted value of expected future welfare 00

Vt = E

yiEt[ Wt+i(Ct,i, St+i, Nt+i, ZWt+i)],

(22)

i-o

where y is the discount factor, and E, denotes expectations formed at time t. In addition, GDP could be made a function of capital stock, with military expenditure crowding out investment, and thus feeding back on capital stock and output. Such

78

R. Smith

dynamic features are quite common in theoretical models, [e.g. Levine and Smith (1995)], but not in applied models. At an informal level, one might expect unconstrained forward-looking optimization, with no costs of adjustment and rational expectations, to suggest that military expenditures should follow a random walk: current military expenditure takes account of the discounted value of anticipated future determinants, thus all that can cause it to change is unanticipated events. Assuming rational expectations, these unanticipated events will be random: unpredictable given the current information set. Of course, the assumption of instantaneous costless adjustment is implausible, given the lags involved in military procurement programs, which can take decades to complete. Forward looking models with costs of adjustment will lead to dynamic equations of the partial adjustment or error correction forms [see, for instance, Nickell (1985) or Alogoskoufis and Smith (1991)]. Thus, there are a range of theoretical considerations (security depends on stocks not military spending; bureaucratic incrementalism; adjustment costs are important; and governments are forward looking) that suggest that military expenditure should show quite complex dynamics, even in the absence of full inter-temporal optimization.

5. Data Estimating a real demand function for defense faces an almost insuperable difficulty: there is virtually no good information on prices. Strictly, most of the estimated equations cited in this survey are not really demand functions since they do not include price as an explanatory variable. Even where there are published price indexes (e.g. the UK and the USA) the practical and conceptual difficulties associated with their construction make them potentially very unreliable. Common assumptions made to avoid having to use prices are either that the relative price of military and civilian goods is constant (in which case the effect is picked up by the intercept in a linear equation) or that price and income elasticities are unity (in which case the share of military expenditure in GDP is the relevant dependent variable). Testing whether these assumptions are adequate would require better data on prices than are now available. Although there is little evidence on the price elasticity of total military spending, components of it are clearly price sensitive. The mix between personnel and equipment, for instance, differs systematically across countries, with high wage countries choosing more capital intensive postures than low wage countries [see Smith, Humm and Fontanel (1987)]. Data on nominal military spending is itself suspect. In many cases, national governments may try to distort the figures for political purposes (usually to underestimate the total) but even with honest reporting there are major ambiguities. Although NATO definitions are widely used there is no general agreement about the right way to treat: conscripts; paramilitary forces like the French Gendarmerie; pensions; dual-use nuclear and space programmes and various other items. There

Ch. 4:

The Demandfor Military Expenditure

79

are standard international sources for military expenditure data, in particular the Stockholm International Peace Research Institute, SIPRI, and the US Arms Control and Disarmament Agency, ACDA. But for countries that do not use a standard NATO definition, quite different results can be obtained from the SIPRI and US ACDA figures. Given that total military expenditure is a very heterogenous aggregate, definition makes a lot of difference to explanation as is indicated by the fact that the components, namely spending on equipment and on personnel are determined rather differently. This is a common pattern in studies which disaggregate the total [e.g. Fontanel and Smith (1990) for France]. Soviet military spending presented particular problems. The official Soviet figure was almost certainly too low and it is unclear whether anyone in the Soviet system actually knew the real cost of their military programme. The CIA "building block figure" was also suspect, though there was dispute as to whether it was too high or too low. The CIA aggregated Soviet forces using US prices to obtain a dollar figure (a measure of the physical amount of forces available to the Soviets) and then used an exchange rate to convert it into a rouble figure that could be compared with Soviet output to estimate the defense burden (a measure of the opportunity cost or the economic resources required to maintain the military). The CIA estimate of Soviet military spending as a share of GDP was increased sharply in the mid 1970s, not because they changed their judgement about Soviet military forces, but because they changed their judgement about the efficiency of Soviet industry and thus the appropriate exchange rate for converting the dollar estimate to roubles. Ostrom and Marra (1986) discuss the issues in more detail. When using the figures for Soviet military spending to model the "threat" and explain NATO military spending, it does not matter what the Soviets actually spent. What matters is what the NATO countries thought Soviet spending was at the time they made their decisions, and this may be more easily measured. Whenever military expenditures are being compared between countries they need to be converted into a common currency. Use of market exchange rates leads to large fluctuations in measured spending unrelated to military realities, so it is common to express the figures in terms of the prices and exchange rate of a single year. The results can be sensitive to the base-year chosen. In principle, purchasing power parity exchange rates could be used, but since we do not have good price indexes for military spending, there is probably no large advantage to this. One advantage of using shares of military expenditure in output is that they can be compared internationally without having to use an exchange rate, but whether they are the appropriate measure depends on the particular theoretical question. A wide variety of different types of data have been used. Although time-series studies for individual countries have been most common, there are also cross-section studies like Dudley and Montmarquette (1981), Tait and Heller (1982) and Maizels and Nissanke (1986). Cross-sections provide more variation in the independent variable, may provide estimates of long-run effects and allow the effect of variables which are relatively constant for any one country to be investigated. There are also studies using

80

R. Smith

panel data like Chan, Hsiao and Keng (1982), which have the advantage that they can combine both the cross-section and time-series dimensions of the data. The relative advantages and disadvantages or time-series, cross-section and panel data for military studies are discussed in Smith (1994). The data are undoubtedly poor and within the empirical literature there seem to be two responses to this. Some argue that given the known limitations of the data only the simplest models should be used. Others argue that only the most sophisticated techniques will be able to separate the signal from the background noise that dominates the data. But this dispute raises more general questions of econometric technique, which are discussed in the next section.

6. Econometric issues The quantitative regression approach, like the optimizing approach, is controversial. Among those hostile to such methods, they are widely dismissed as useless or misleading: the work of "analysts preoccupied with technique" who have "shown insufficient regard for the plausibility of the explanations to which their statistical manipulations have led them" [Greenwood (1987) p. 99]. There are various motivations for the statistical manipulations. Some give the models only descriptive status: as convenient atheoretical ways to extrapolate for forecasting purposes or to summarise certain characteristics of the data. Others give the models structural or causal status, in which case the plausibility of the theory is crucial. In the atheoretical approach, Vector Autoregressions (VARs) and tests for Granger (1969) causality have been widely used. For instance Chowdhury (1991) using timeseries data for 55 less-developed countries (LDCs) found the share of military expenditures, M, and growth, G, to be independent for 30, interdependent for 3, and with Granger causality going just from M to G in 15 and just from G to M in 7. Thus, only a quarter of the 110 tests reject the null of no association. This is in line with most other studies [e.g. Georgiou and Smith (1983), Dunne and Smith (1990), Payne and Ross (1992)] which have found no clear pattern of Granger causality between economic variables and military spending. The argument for the atheoretical approach is that because of the complexity of the process, structural or institutional models of the arms race are unlikely to be effective. Williams and McGinnis (1992a,b) make this argument strongly and use atheoretical frequency domain methods to investigate the patterns determining military spending. VARs need not be inherently atheoretical, since they can also be interpreted as the reduced form of traditional structural models or the decision rules resulting from certain optimization processes. The argument against the atheoretical approach is that the theory, to the extent that it provides valid restrictions, increases the efficiency of estimation, allows the coefficients to be interpreted within a wider intellectual and institutional framework, and enables theoretical propositions to be tested. The debate between those who believe that the estimating equation should be determined by a tightly specified theoretical model and

Ch. 4:

81

The Demandfor Military Expenditure

those who think it should be determined by the best fit to the data continues. In this context fit means more than R2 : all the statistical assumptions of the model need to be tested. A compromise position is to start from a loose general theoretical specification and allow the data to decide the exact form. The problem with this approach is that since the model was designed to fit the data, the same data cannot be used to test it. These econometric issues are discussed in more detail in Smith (1989), where a general to specific model that passes all the diagnostic tests is developed on UK data and then tested on French data. The reliability of econometric estimates and inferences depend on the validity of the statistical assumptions. These assumptions include a correctly specified linear relationship with constant parameters, exogenous regressors and normally distributed, serially independent, homoskedastic disturbances. Failure of normality, serial independence and homoskedasticity leave OLS estimates unbiased, though not fully efficient, while the reported standard errors and test statistics are wrong, making inference invalid. There are a range of diagnostic tests available to check for failure of these assumptions, though they have not been widely used in this literature. Most of the studies report tests for serial correlation and this seems to be a problem for many of the static models, where it is usually corrected by Cochrane-Orcutt type transformations. However residual serial correlation may be an indication of more fundamental dynamic misspecification. Suppose /3and xt are kx 1 vectors and the model is a static equation with serially correlated errors Mt = ac+ 'x, + ut;

Ut =put

+ Vt.

(23)

Noting that ut = p(Mt-1 -

a-

- xt ) + t

and substituting for ut, Equation (23) can be written Mt = a(l - p) + 'xt + pyt_ - plp'xt

+ vt

(24)

which is a special case of the dynamic linear regression model M, = ao +/bOxt + alyt i -

Ixt1l + Vt

(25)

subject to the k (common factor) restrictions Po3al = -i. These restrictions, which residual serial correlation factor restrictions are not inconsistent. While static

(26) are imposed by Cochrane-Orcutt, are rarely tested. If the is an indicator of dynamic misspecification (the common true) Cochrane-Orcutt type estimators will be biased and models are implausible in that they imply that short and

82

R. Smith Table 1 Autoregressions explaining log of real military expenditure 1960-1985

Quantity

Country US

constant lnM(-l) InM(-2) t (x100) R2 for level of M R2 for change in M SCpb ADF(1)

0.94 (2.02)a 1.47 (8.59) -0.66 (3.69) 0.25 (1.58) 0.89 0.42 0.51 -2.03

France 0.47 (1.82) 1.19 (6.36) -0.36 (1.92) 0.40 (1.84) 0.98 0.20 0.31 -1.79

Germany 1.20 (3.05) 0.56 (2.78) 0.04 (0.37) 0.55 (1.86) 0.89 0.33 0.12 -2.92

UK 0.59 (1.07) 0.84 (3.92) -0.05 (0.19) 0.33 (1.52) 0.89 0.11 0.30 -1.07

a Figures in parentheses are absolute t ratios.

b SC p is the p value of the F version of an LM test for first order serial correlation. ADF(l) is the Augmented Dickey Fuller Statistic including one lagged change, its critical value -3.59.

long-run effects are the same, they can provide estimates of the long-run effects if the variables are I(1) (integrated of order one, i.e. stationary after being differenced once) and cointegrated. In fact, both the theory discussed above and the data strongly suggest that the dynamics are important. Table 1 gives estimates of the following simple model for the four largest NATO countries 1960-1985 ln(Mt) = Po +plln(M 1)+p 21n(Mt- 2) + rt + Et.

(27)

The data, which are in 1980 US $ bn, were taken from Table A7.1 of Murdoch and Sandler (1990). In the case of a random walk, pi = 1 and P2=O; a first order autoregression in the growth rates is the special case where pi +P2 = 1. In either case there is a unit root in the process. In all four cases, the first order augmented Dickey Fuller statistic, ADF(1), does not reject the unit root hypothesis that the series is I(1). None of the equations shows significant serial correlation and they indicate that at least 89% of the variation in real military expenditure can be explained "atheoretically", purely from univariate dynamics. Of course, this high R2 is misleading as is indicated by the much lower value for the proportion of the change in M explained. In fact, for France and Britain, the hypothesis that real military expenditure is just a random walk with drift, cannot be rejected. Given the importance of both theory and dynamics, integrating the theoretical static and atheoretical dynamic approaches is an important avenue for future research. This is, of course, also a major research agenda in mainstream economics.

Ch. 4:

The Demandfor Military Expenditure

83

In many cases demand equations for military expenditure have been estimated as components of complete systems. The systems may include the other components of demand in a particular country (as in traditional complete demand systems) or the demand equations for other countries as in alliance or arms race models. As an example, Hilton and Vu (1991) estimate a 3 equation demand system for each of a set of NATO allies by Full Information Maximum Likelihood. There are three main reasons for using systems estimators. Firstly, systems allow cross-equation restrictions to be imposed (e.g. symmetry in complete demand system) which increases estimation efficiency and permits hypothesis testing. Secondly, systems can allow for crossequation disturbance covariances, also improving efficiency. Thirdly, systems can allow for simultaneity, where explanatory and dependent variables are jointly determined; enabling consistent estimates of the structural parameters to be obtained when the system is identified. Simultaneous systems arise in alliance or arms race models when the military expenditures of sets of countries are jointly determined and in macroeconomic models where the main focus of the exercise is to estimate the economic effects of military spending but equations for military expenditure are included to deal with potential simultaneity biases.

7. Some examples One of the most striking features of this literature is its variety. This is well illustrated by the studies in Hartley and Sandler (1990). The authors of this collection were asked to estimate their country's demand for military expenditure, using a standard model in which military expenditure was determined by GDP, the threat, spillins from allies and other variables such as governing party or a change in strategy. Despite being provided with a standard framework, the authors managed to produce a wide range of equations. In many cases the final equation was the result of a specification search. Looney and Mehay (1990) regressed US real defense spending (1965-1985) on its lagged value, anticipated and unanticipated Soviet defense spending, inflation, deviation from trend in federal revenues, lagged federal deficit, a Vietnam dummy, deviations from trend in other NATO spending and a detente dummy, correcting for ARI serial correlation, by Cochrane-Orcutt. In a number of their specifications the coefficient of the lagged dependent variable was very close to unity. Smith (1990) regressed the change in the UK share of military spending (19491987) on the change in the US share, the difference between the UK share and the Soviet share in the previous period and dummies for the Korean War, defense reviews, and the NATO 3% commitment, using OLS. Schmidt, Pilandon and Aben (1990) regressed the SIPRI measure of French defense spending in constant dollars (19651978) on GDP, Soviet military spending, other NATO expenditures, Allied African expenditures correcting for AR2 serial correlation. They also discuss using French measures (which do not correspond to the NATO definition used by SIPRI).

84

R. Smith

Fritz-AB1mus and Zimmermann (1990) regress German defense spending in constant dollars (1961-1987) on GDP (which had a negative sign), the lagged level in French and other NATO countries defense spending, lagged Soviet defense spending, and a dummy for government composition, where the coefficients of the spillin terms were allowed to shift with the change in NATO strategy in 1974. Murdoch and Sandler (1990) use a log-log equation to explain real Swedish military expenditure in US dollars (1958-1985) by GDP, population, the lagged real military expenditures of Norway, lagged Soviet spending (which was negative but insignificant) and a fiscal dummy; allowing coefficients of the Norwegian spillin terms to shift with the change in NATO strategy. They experimented with a variety of other spillin measures. Deger and Sen (1990) estimate equations explaining India and Pakistan's real military spending by their own lagged spending, current or lagged values of the others spending, GDP, arms imports, arms production and the ratio of government spending to output. A priori, it seems unlikely that the structure of the processes generating military expenditure are so different in each of these countries. It seems more likely that the differences are the products of the prior beliefs and the modelling techniques of the authors. Such proliferation of competing models is not uncommon in applied econometrics and there is considerable methodological dispute as to what the appropriate response should be. The issues being very similar to that within the military literature: the relative weight to be given to firm theoretical foundations or an adequate representation of the data.

8. Conclusion The studies discussed in this chapter indicate that it is possible to achieve equations that explain military expenditure reasonably well in terms of R2 and significant t statistics. However, for integrated time-series subject to the danger of spurious regression these may not be very stringent criteria. In a regression with two unrelated integrated series, R 2 goes to unity and the t statistics to non-zero values as the sample size increases. This suggests that more demanding statistical criteria are required: tests for cointegration, lack of serial correlation, functional form, normality, and heteroskedasticity are now standard in econometrics packages and may help in distinguishing between models. In addition, tests for structural stability, whether the parameters are constant over time, should also be informative in future work given the large changes in the strategic environment over the last few years. It would also be useful if authors could make their data sets more widely available, so that the performance of different models on the same data could be compared. There is also scope for improving the theory. There are a range of theoretical armsrace models which treat countries as strategically interacting, forward-looking agents maximizing inter-temporal welfare functions subject to constraints. However, these have not yet been widely used in empirical work and there is scope to investigate the

Ch. 4:

The Demandfor Military Expenditure

85

empirical implications of such models. Ideally, one would like a generally accepted dynamic inter-temporal optimizing model which nested the variety of theoretical and atheoretical forms for the demand for military expenditure. This model could then be used as a framework to assess the theoretical consistency and statistical adequacy of the competing equations. While this seems unlikely to be achieved, it does provide an interesting research agenda. References Alogoskoufis, G., and R. Smith, 1991, On error correction models: Specification, interpretation, estimation, Journal of Economic Surveys 5, 97-128. Anderton, C.H., 1990, The inherent propensity towards peace or war embodied in weaponry, Defence Economics 1, 197-220. Chan, M.W.L., C. Hsiao and C.WK. Keng, 1982, Defence expenditures and economic growth in developing countries: A temporal cross-sectional analysis, in: O.D. Anderson and M.R. Perryman, eds., Applied time series analysis (North-Holland, Amsterdam) 53-64. Chowdhury, A.R., 1991, A causal analysis of defence spending and economic growth, Journal of Conflict Resolution 35, 80-97. Conybeare, J.A.C., J.C. Murdoch and T. Sandler, 1994, Alternative collective-goods models of military alliances: Theory and empirics, Economic Inquiry 32, 525-542. Cusack, T.R., and M.D. Ward, 1981, Military spending in the United States, Soviet Union and the People's Republic of China, Journal of Conflict Resolution 25, 429-469. Cypher, J., 1987, Military spending, technical change, and economic growth: A disguised form of industrial policy? Journal of Economic Issues 21, 33-59. Deaton, A., and J. Muellbauer, 1980, Economics and consumer behaviour (Cambridge University Press, Cambridge). Deger, S., and S. Sen, 1990, Military security and the economy: defence expenditure in India and Pakistan, in: K. Hartley and T. Sandler, eds., The economics of defence spending: An international survey (Routledge, London) 189-227. Dudley, L., and C. Montmarquette, 1981, The demand for military expenditures: An international comparison, Public Choice 37, 5-31. Dunne, J.P., and R. Smith, 1990, Military expenditure and unemployment in the OECD, Defence Economics 1, 57-73. Dunne, J.P, P. Pashardes and R. Smith, 1984, Needs, costs and bureaucracy: The allocation of public consumption in the UK, Economic Journal 94, 1-15. Fontanel, J., and R. Smith, 1990, The impact of strategy and measurement on models of French military expenditure, Defence Economics 1, 261-273. Fritz-A3mus, D., and K. Zimmermann, 1990, West German demand for defence spending, in: K. Hartley and T. Sandler, eds., The economics of defence spending: An international survey (Routledge, London) 118 147. Georgiou, G., and R. Smith, 1983, Assessing the effect of military expenditure on OECD countries, Arms Control 4, 3 15. Gleditsch, N.P., and O. Njolstad, eds., 1990, Arms races: Technological and political dynamics (Sage Publications, London). Granger, C.WJ., 1969, Investigating causal relations by econometric models and cross-spectral models, Econometrica 37, 424-438. Greenwood, D., 1987, Note on the impact of military expenditure on economic growth and performance, in: C. Schmidt, ed., The economics of military expenditures (Macmillan for the International Economic Association, London) 98-103.

86

R. Smith

Griffin, L.J., M. Wallace and J. Devine, 1982, The political economy of military spending: Evidence from the United States, Cambridge Journal of Economics 6, 1-14. Hartley, K., and T. Sandler, eds., 1990, The economics of defence spending: An international survey (Routledge, London). Hilton, B., and A. Vu, 1991, The McGuire model and the economics of the NATO alliance, Defence Economics 2, 105-21. Kamlet, M.S., and D.C. Mowery, 1987, Influences on Executive and Congressional budgetary priorities 1955-1981, American Political Science Review, 81, 155-178. Levine, P., and R. Smith, 1995, The arms trade and arms control, Economic Journal 105, 471-484. Looney, R.E., and S.L. Mehay, 1990, United States defence expenditures: trends and analysis, in: K. Hartley and T. Sandler, eds., The economics of defence spending: An international survey (Routledge, London) 13-40. Maizels, A., and M.K. Nissanke, 1986, Determinants of military expenditures in developing countries, World Development 14, 1125-1140. McGuire, M.C., 1982, U.S. foreign assistance, Israeli resource allocation, and the arms race in the Middle East, Journal of Conflict Resolution 26, 199-235. McGuire, M.C., 1987, US foreign assistance, Israeli resource allocation and the arms race in the Middle East, in: C. Schmidt, ed., The economics of military expenditures (Macmillan for the International Economic Association, London) 197-238. Mintz, A., ed., 1992, The political economy of military spending in the United States (Routledge, London). Murdoch, J.C., and T. Sandler, 1984, Complementarity, free riding and the military expenditure of NATO allies, Journal of Public Economics 25, 83-101. Murdoch, J.C., and T. Sandler, 1986, The political economy of Scandinavian neutrality, Scandinavian Journal of Economics 88, 583-603. Murdoch, J.C., and T. Sandler, 1990, Swedish military expenditure and armed neutrality, in: K. Hartley and T. Sandler, eds., The economics of defence spending: An international survey (Routledge, London) 148-176. Murdoch, J.C., T. Sandler and L. Hansen, 1991, An econometric technique for comparing median voter and oligarchy choice models of collective action: The case of the NATO alliance, Review of Economics and Statistics 63, 624-631. Nickell, S., 1985, Error Correction, partial adjustment and all that: an expository note, Oxford Bulletin of Economics and Statistics 47, 119-129. Okamura, M., 1991, Estimating the impact of the Soviet Union's threat on the United States-Japan alliance: A demand system approach, Review of Economics and Statistics 73, 200-207. Ostrom Jr, C.W., 1978, A reactive linkage model of the US defence expenditure policy-making process, American Political Science Review 72, 941-957. Ostrom Jr, C.W., and R.F. Marra, 1986, US defence spending and the Soviet estimate, American Political Science Review 80, 819-842. Payne, J.E., and K.L. Ross, 1992, Defense spending and the macroeconomy, Defence Economics 3, 161-168. Pivetti, M., 1992, Military spending as a burden on growth: An "underconsumptionist" critique, Cambridge Journal of Economics 16, 373-384. Rattinger, H., 1975, Armaments detente and bureaucracy: The case of the arms race in Europe, Journal of Conflict Resolution 19, 571-595. Sandler, T., and K. Hartley, 1995, The economics of defense (Cambridge University Press, Cambridge). Sandler, T., and J.C. Murdoch, 1990, Nash-Cournot or Lindahl behaviour: An empirical test for the NATO allies, Quarterly Journal of Economics 105, 875-94. Schmidt, C., L. Pilandon and J. Aben, 1990, Defence spending in France: the price of independence, in: K. Hartley and T. Sandler, eds., The economics of defence spending: An international survey (Routledge, London) 93-117.

Ch. 4:

The Demandfor Military Expenditure

87

Smith, R., 1977, Military expenditure and capitalism, Cambridge Journal of Economics 1, 61-76. Smith, R., 1978, Military expenditure and capitalism: a reply, Cambridge Journal of Economics 2, 299-304. Smith, R., 1980, The demand for military expenditure, Economic Journal 90, 811-820. Smith, R., 1987, The demand for military expenditure: A correction, Economic Journal 97, 989-90. Smith, R., 1989, Models of military expenditure, Journal of Applied Econometrics 4, 345-359. Smith, R., 1990, Defence spending in the United Kingdom, in: K. Hartley and T. Sandler, eds., The economics of defence spending: An international survey (Routledge, London) 76-92. Smith, R., 1994, Measuring the effects of military spending: Cross-sections or time-series? in: M. Chatterji, H. Jager and A. Rima, eds., The economics of international security (Macmillan, Basingstoke) 232-244. Smith, R., and J.P. Dunne, 1994, Is military spending a burden? A marxo-marginalist response to Pivetti, Cambridge Journal of Economics 18, 515-521. Smith, R., A. Humm and J. Fontanel, 1987, Capital labour substitution in defence provision, in: S. Deger and R. West, eds., Defence, security and development (Frances Pinter, London) 69-80. Stoll, R.J., 1992, Steaming in the dark? Rules, rivals and the British Navy 1860-1913, Journal of Conflict Resolution 36, 263-283. Tait, A.A., and P.S. Heller, 1982, International comparisons of government expenditure, Occasional Paper No. 10 (International Monetary Fund, Washington, DC). Williams, J.T., and M.D. McGinnis, 1992a, Expectations and the dynamics of US defense budgets: A critique of organisational reaction models, in: A. Mintz, ed., The political economy of military spending in the United States (Routledge, London) 282-304. Williams, J.T., and M.D. McGinnis, 1992b, The dimension of superpower rivalry: A dynamic factor analysis, Journal of Conflict Resolution 36, 86-98.

Chapter 5

MILITARY ALLIANCES: THEORY AND EMPIRICS JAMES C. MURDOCH University of Texas at Dallas

Contents Abstract Keywords 1. Introduction 2. Theoretical contributions 2.1. The pure public good model 2.2. The joint product model 2.3. Other contributions 2.3.1. Technology of supply 2.3.2. The allocation process

3. Empirical contributions 3.1. Tests based on the pure public good model 3.2. Tests based on the joint product model 3.3. Other empirical contributions

90 90 91 91 92 95 98 98 99

100 100 103 104

4. Conclusion

106

References

107

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T2Sandler © 1995 Elsevier Science B. V All rights reserved

90

J.C. Murdoch

Abstract This chapter provides an overview of the main theoretical and empirical findings in the economics of military alliances. The pure public and joint product models are presented along with the empirical methods used to test them. Issues concerning burden sharing and strategic doctrine in the NATO alliance are discussed. Comparisons between the Nash and Lindahl allocation processes are made by presenting the empirical work that has attempted to distinguish between them. Attention is also given to the median-voter model and to alternative technologies of public good supply.

Keywords alliance, best shot, burden sharing, collective action, joint-product model, Lindahl equilibrium, median-voter model, military alliance, NATO, Nash equilibrium, Nash reaction paths, pure public good, strategic doctrine, weakest link

Ch. 5:

Military Alliances: Theory and Empirics

91

1. Introduction Since the seminal study of Olson and Zeckhauser (1966), economists and political scientists have applied the tools of public economics to the study of military alliances. In this chapter, a survey of these studies is presented with particular emphasis on the theoretical and empirical behavior of individual nations that belong to a military alliance. By definition, a military alliance is a group of nations that are bound to provide protection to all members from aggression by common enemies [Olson and Zeckhauser (1966)]. From the perspective of the individual ally, the level of protection, which is also called security or deterrence, is a public good. Regardless of whether defense is purely or impurely public, the nonexclusion and nonrivalry properties generate several testable hypotheses about the behavior of the individual allies. The purpose of this chapter is to highlight these hypotheses and indicate the empirical methods used to test them. Much of the literature follows in the tradition of Olson and Zeckhauser (1966) and concerns burden sharing and allocative efficiency. With respect to burden sharing, the discussion begins with an "exploitation hypothesis" [Olson (1965)], in which the burdens for defending the alliance will be shared unequally, with the large wealthier allies contributing relatively more to the defense. Empirical tests of this hypothesis generally study the association between measures of burden and measures of wealth. While related to burden sharing, the analysis of allocative efficiency generally concerns the behavioral assumptions for the allies. If the allies follow Nash behavior, then we can expect some free riding and a sub-optimal level of provision when compared to the Pareto-optimal standard. Over time, however, the allies may adopt cooperative behavior by, for example, bargaining [Palmer (1990)], and we may see a reduction in free riding and improved allocative efficiency. The empirical tests regarding allocative efficiency depend on measuring the extent of free riding and distinguishing Nash behavior from cooperative behavior. The primary body of this chapter is organized into two sections - theoretical and empirical contributions. Within each of these sections, I address the literature relating to (i) the pure public good model, (ii) the joint product model, and (iii) other perspectives. The final section of the paper offers some summary remarks.

2. Theoretical contributions There is a close relationship between the theoretical literature addressing military alliances and the private provision of public goods literature [e.g., Warr (1983)]. This relationship is interesting primarily because the origin of some models is unclear. For example, the joint product model appears to have its roots in military alliances. It was then ported to other applications in public sector economics. In this chapter, I am not concerned with these issues and, therefore, attribute the theoretical arguments primarily to the military alliances literature.

92

J.C. Murdoch

2.1. The pure public good model The pure public good model characterizes the behavior of an ally who belongs to an ncountry alliance. The ith ally is assumed to allocate its national income, Ii , between a military activity, q', and a private nondefense numeraire good, yi. The military activity produces a pure public good for all allies, which I call deterrence, zi . The production relationship is governed by z i =f(qi),

(1) 2

2

where df/dq' =f' > 0 and d f/dqi =f" < O. Alternative specifications of Equation (1) are employed in the literature. The most common is to simply set z=qi [Sandler (1993)]. The qualitative conclusions of the model are basically unaffected by these restrictions and we maintain Equation (1) for more generality and to set up the joint product model. Total deterrence, Z, is nonrival and nonexcludable; thus, all allies receive Z regardless of how much they contribute. The technology of public supply in the basic, pure public good model is additive with unit weights on contributions: zi.

Z =

(2)

i=l

This simple aggregation implies perfect substitutability of zi among allies. As discussed below, this technology may not be appropriate for some alliances. From the ith country's perspective, total deterrence is i

i

Z = z +2 ,

(3)

where Z' is the level of deterrence that "spills in" to country i. Using Equations (1)-(3), we can express Z as a function of the qi's: Z =f(qi)+

f(qj) =f(qi) + h(Oi),

(4)

j ,i

with h' > 0 and h" < 0. l The budget constraint for ally i is = yi +pq',

(5)

where p is the price per unit of the military activity and the price of the numeraire has been normalized to one. The preferences of the ith ally are assumed to be represented by an unspecified agent with a utility function, U = Ui(y, Z, Ei),

(6)

i

where E represents (possibly) a vector of exogenous environmental influences that shift the utility function. In many treatments, E/ represents, in part, the military strength ' We are implicitly assuming a functional form for the f(.) function such that h(Q') = Ejif(qj).

Ch. 5: Military Alliances: Theory and Empirics

93

of the alliance's common enemies, referred to as threat. Other influences included in Ei may be the price of oil or periods of time. Rather than denoting the preferences of an agent for the entire nation, the utility function may represent other decision makers. For example, Dudley and Montmarquette (1981) and Murdoch, Sandler and Hansen (1991) test the median voter hypothesis by redefining Equation (5) to reflect the budget of the median voter. In this case, the optimal choice of the median voter is assumed to be translated into national action. Substituting Equation (4) into (6), we specify the Nash problem for ally i as ma x { Ui(yi,f(qi) + h(i'),E') Ii = y +pq'i}. y', q,

(7)

A Nash equilibrium requires a vector q* = (ql*,..., q"n) and y*= (y*, .. ,yn*) such that Equation (7) is simultaneously satisfied for every ally. The first-order conditions require satisfaction of the budget constraint and f' MRSy = p,

(8)

where MRSzy denotes the marginal rate of substitution of deterrence for the private good. Equation (8) illustrates that each ally will provide qi until their marginal valuation of deterrence, weighted by the productivity of q' in producing deterrence, is equal to the marginal cost of q'. The allocative inefficiency of the Nash equilibrium is evident in Equation (8), since a Pareto optimal level of Z would require that f' Ej MRSy = p. The Nash solution generates a provision level that is less than the Pareto level, because each ally does not consider the marginal valuation of the other allies that its defense provision conveys. At a Nash equilibrium, the first-order conditions of Equation (7) imply a set of "demand" equations for the military activity: qi = qi(Ii,p

i, E'),

(9)

illustrating that the demand for military activity depends on the national income, price per unit of activity, the total military activity of a nation's allies, and various shift measures. The demand equations facilitate hypotheses about the relationships between q' and the exogenous variables. For example, comparative static results suggest a direct relationship between qi and Ii and an inverse relationship between q' and p, assuming that the military activity is a normal good [Murdoch and Sandler (1984)]. Moreover, the comparative static relationship between qi and Oi, (Oq'i/dQ), is expected to be negative and is called the slope of the "reaction path". Much of what distinguishes the alliance literature concerns this reaction path; i.e., how one ally responds to the actions of the other allies. In contrast, arms-race models concern how an ally responds to the actions of an enemy over time; hence the analysis focuses on dqilOEi in a dynamic model. 2 2

Smith's Chapter 4 reviews the alternative empirical forms of equations like (9).

94

. Murdoch

One way to see the burden sharing and efficiency implications of the pure public good model is to analyze an algebraic example [see Cornes and Sandler (1986) and Sandler (1993) for geometric treatments]. Consider a 3-country alliance, where each ally differs only by income; e.g., let I 1 > 2 > 3. If the utility functions are identical and of the Cobb-Douglas family (U =yaQt, Q = q + q2 + q3 ) and f' = , then the demand equations are ql = a(II/p) - b(q2 +q3),

(10a)

q2 = a(I2 /p) - b(ql +q3),

(O10b)

q3 = a(i 3/p) - b(ql +q2),

(10c)

where a = fi/(a + 3), b = a/(a +/f),and Ei is dropped to simplify the presentation. This particular example nicely illustrates the hypothesized relationships between income, price, the spillins and the military activity. Moreover, the parameters of the model are easily identified in an empirical exercise. The Nash equilibrium is the simultaneous solution to the demand equations. In this example, the Nash or "subscription equilibrium" [Warr (1983)] solution is ql* = a[I + b(I -_12 _ 3)]/c,

(1 la)

q2* = a[I 2 + b(I 2 -_I - I3 )]/c,

(1 lb)

q3* = a[i3 + b(I3 _ 2 _ Il)]/c.

(1lc)

In Equations (1 la)-( lc), each q* > 0 and c = (1 + b - 2b 2)p. An important result of the pure public good model is that the provision level depends on the relative size of income. For example, let I 1= 100, 2 = 90, 3 = 80 and a = b = 0.5. Additionally, assume that the defense price equals one, giving c = 1. Then, ally 3 will make a contribution because its income is almost as large as ally 1 and 2. The equilibrium ally provisions are ql* = 32.5, q2 * = 22.5, and q3 * = 12.5 for a total provision level of 67.5. On the other hand, if II = 100, 2 = 60, and 3 = 40, ally 3 will not make a contribution. The equilibrium values are ql* =46.67, q2* = 6.67, and q3* = 0 for a total provision of 53.33. Bergstrom, Blume and Varian (1986) and Andreoni and McGuire (1993) discuss more general procedures for identifying the group of contributors when incomes differ. Both numerical examples illustrate the exploitation hypothesis. In the first example, the burdens (calculated as qi/i) are 0.325, 0.25, and 0.156, respectively for allies 1, 2, and 3. Similarly, in the second example, the burdens are 0.466, 0.066, and 0.0, respectively. In either case, the larger allies shoulder larger burdens and we see a high correlation between economic size and military burden. The exploitation is more pronounced in the second example where the income differences are greater. This fact is the basis for the interest in the relative income differentials in NATO discussed in Section 3, below [see Olson (1986) and Oneal and Elrod (1989).]

Ch. 5: Military Alliances: Theory and Empirics

95

The inefficiency of the Nash solution is also readily apparent in these numerical examples. When I = 100, j2 = 90 and 13 = 80, the Pareto optimal levels of provision are q =55, q2=45, and q3 =35 for a total of 135. Alternatively, under the second income distribution, the optimal levels are q =66.67, q2 =26.67 and q3= 6.67 for a total of 1003. In either case, we see that the Nash levels are less than the Pareto optimal levels. 2.2. The joint product model The pure public good paradigm of Olson and Zeckhauser (1966) is the seminal economic model for understanding military alliances. However, it does not explain certain observable facts. In particular, Russett (1970) noted that by the late 1960s, the correlation between economic size and defense burden for a sample of NATO allies was essentially insignificant. This fact was counter to the main hypothesis of the pure public good model and contrary to the correlations of just a few years earlier. What could explain this result? A partial answer was proposed by van Ypersele de Strihou (1967), who argued that defense burden sharing must be influenced by the "private benefits" of defense 4 By private, he meant that some defense activities, for example, protection of colonial interests and drug traffic interdiction, provide benefits to the nation undertaking such activities, but not to the other members of the alliance. The benefits may be purely public to the citizens of the providing nation, but they can be withheld from the citizens of the other nations in the alliance. van Ypersele de Strihou (1967) concluded that the burden sharing is more likely to reflect the benefits-received principle when private benefits are a significant proportion of the defense benefits received by allies. Sandler (1977) extended the analysis by proposing a joint product model, for which the defense activity of an ally jointly produces a vector of outputs that may be purely public (e.g., deterrence), impurely public (e.g., damage limitation), and private (e.g., policing terrorist activities at home). The model is a straightforward extension of the pure public good model and facilitates modelling the complexity of modern military arsenals. A nation like the USA, for example, maintains an arsenal of nuclear and conventional forces that may be used to produce a host of various public and private benefits 5.

3 These calculations are very simple, given the assumptions of the example. One way to find the optimal levels is to recognize that the MRSfy =y/Z V i. Because the sum of the MRS's must equal 1, we can set each equal to 1/3. Then, with the three budget constraints, we have six equations in six unknowns. Alternatively, we can recognize that the Pareto optimal level must be 50% of the total income of the group because of homothetic preferences and unit prices, because a=i. See Knorr (1985), Oneal (1990a,b), Russett (1970), and Sandler (1977). s The joint product model has many applications in public economics [See Comes and Sandler (1986,

1994)]. A formal derivation of the primary results of the model as they apply to ally behavior first appeared in Murdoch and Sandler (1984).

J. Murdoch

96

In the basic joint product model, the military activity, q', produces the pure public good (zi), as described above, and a private good x', which enters the ally's utility function. The joint product relationships are xi = g(q'),

z' =f(q'),

(12,13)

while the utility function is U i = U(yix i , Z, Ei),

(14)

where g' >0, g" < 0, andf is as described above. Substituting Equations (12) and (4) into (14), we derive the Nash problem for the ith ally as max Ui(yi , g(qi),f(qi) + h(Qi), Ei ) I

i

=yi +pq} ,

(15)

which is the same as Equation (7) except for the addition of the private defense output, x'. Once again, the Nash equilibrium requires vectors q* andy* such that Equation (15) is simultaneously satisfied for all allies. The first-order conditions imply that g'MRS'y +f'MRSy = p.

(16)

Hence, each ally equates the weighted sum of its MRS for the private defense good and the MRS for the public defense good to the price, where the weights are the marginal products of the defense activity in producing the defense outputs. As in the pure public good model, the equilibrium is still suboptimal in a Pareto sense, because each ally only considers its MRSzy in the provision decision; Pareto optimality would require a summing over all MRSzy's. The joint product model provides a plausible explanation to the phenomenon noted by Russett (1970): if the military activities of individual nations generate more private benefits, then the level of qi must increase to satisfy Equation (16), making the burdens more equitably shared. The joint product model implies a demand equation that appears the same as that implied by the pure public good model. Thus, the first-order conditions of Equation (15) (by the implicit function theorem) imply qi = qi(P,p, Qi,Ei).

(17)

However, the relationship between q and i' in Equation (17) is potentially quite different when compared to the same relationship in Equation (9). Murdoch and Sandler (1984) noted that Oqil0/Q may be positive or negative in the joint product model; a key determinant in signing this derivative is the complementarity of the joint products. When the joint products are Hicksian complements, the derivative may be

Ch. 5:

97

Military Alliances: Theory and Empirics

positive - the opposite of the predicted sign in the pure public good model 6. Murdoch and Sandler argued that the different theoretical results have important implications for understanding NATO because the alliance switched from providing protection via a nuclear deterrent (the doctrine of mutually assured destruction) to protection via a doctrine of flexible response. According to the flexible response doctrine, NATO should respond to external threats in a variety of ways; e.g., conventional ground forces, air power, or limited nuclear strikes. Hence, the latter doctrine enhanced the complementarity of the private and public defense goods 7 . The result of this effect, over time, is to generate the more equalized burdens in NATO. Because the reduced-form demand equations (9) and (17) look identical, it is not obvious how to distinguish between the pure public good model and the joint product model. One means is to simply estimate an equation with data on q, i', p, Q, and Ei and examine the estimated relationship between q' and O'. A positive relationship would certainly be evidence against the pure public good model, but a negative sign is not inconsistent with either model. This approach is, consequently, weak because the two models are not measured against each other. Sandler and Murdoch (1990) found that by expressing the demand equations in their "full income" form, a simple nested test could be derived to distinguish between the models. The full income is the sum of an ally's income (Ii) and the value of the spillins (pQi) [Bergstrom, Blume and Varian (1986)]. Adding pQi to each side of Equation (4) we get Fi = I i +pQi =yi +pQi,

(18) i

in whichpq' + p' = pQi. In the full income model, the choice is overy and Qi', rather than yi and qi. The utility function in the pure public good model becomes Ui(yi,f(qi) + h(Qi), E) = vi(yi, q', Qi, E),

(19)

while in the joint product model we have Ui(yi, g(Qi - 0i),f (Qi - Qi) + h(Qi), Ei )

(yi, , Q, QiEi).

(20)

For the pure public good model, the full income demand equation is Qi= Qi(Fi,p,E)

(21)

when Equation (19) is maximized subject to Equation (18), while, for the joint product model, the full income demand equation is Qi = Q'(F, Qi,p, E i)

(22)

when Equation (20) is maximized subject to Equation (18). In both models, a Nash equilibrium requires that Qi=Qj for every i j. Equations (21) and (22) provide a See Cornes and Sandler (1994) for an extensive treatment of the comparative statics of the joint product model. Also, the neutrality results are quite different in the joint product model, see Sandler (1993). 6

7

This argument is challenged by Oneal and Elrod (1989) and defended by Murdoch and

Sandler (1991).

98

JC. Murdoch

theoretical distinction between the two models that is empirically testable: the joint product model can be estimated and then the significance of the Qi effect can be tested. If this effect is restricted to zero, the pure public good model results. 2.3. Other contributions 2.3.1. Technology of supply Until recently, the theoretical literature assumed a simple summation technology, such as that displayed in Equation (2), for the total supply of the public good 8. Hirshleifer (1983) and McGuire (1990) offer alternative technologies that are particularly relevant to the economics of military alliances. Hirshleifer suggested the "weakest-link" and "best-shot" technologies. Under the weakest-link, Z = min(zl,... ,z),

(23)

while under the best-shot, Z = max(z',...,z).

(24)

The weakest-link supply equation may apply, for example, to an alliance that fortifies a perimeter in order to keep the enemy from breaking through, so that the security of the nations within the perimeter is determined by the strength of the weakest-link nation. The best-shot technology, on the other hand, may apply to an alliance wherein the security is provided by the nuclear deterrent of a single nation, so that the level of provision is determined by the contribution of the strongest nation. For both the best-shot and weakest-link models, the efficiency and burden sharing implications are different when compared to the summation technology. These differences are discussed by Conybeare, Murdoch and Sandler (1994), who derive the demand and efficiency conditions under both the best-shot and weakest-link technologies for a pure public good and joint product model. An interesting finding is that, in a joint product model, the demand equations differ depending on the supply technology and the identification of the best shooter or weakest-link. To see this, assume that the supply is weakest-link and that country 1 is the weak-link nation, making the demand for the military activity in 1 q t = q(1 1,p,El).

(25)

For the other allies (0'= 2,.. ., n), the demand is qi = q/(I,p, q, E).

(26)

The demand equations are similar for a best-shot alliance, except for the interpretation of ally 1. In the best-shot alliance, let ally 1 denote the best shooter; then Equations 8 This section only covers a small subset of the theoretical literature. See Sandler (1993) and Sandler and Hartley (1995) for more exhaustive surveys.

Ch. 5:

Military Alliances: Theory and Empirics

99

(25) and (26) will represent the reduced-form demand equations. The identification of these equations enabled Conybeare, Murdoch and Sandler (1994) to devise a testing procedure for identifying the actual supply technology. McGuire (1990) offered a generalization of the summation technology. In his model, the aggregation technology is Qi = qi +

wJq,

(27)

ji where the weights (wi) are all positive. An interesting feature of the model is that the weights reflect the consumer (i.e., receiving nation) and the producer nation. Thus, w4, for example, is the spillover from ally 4 to ally 2. McGuire defines several specific cases with restrictions on Equation (27). He also derives the basic efficiency and burdensharing implications of his model. 2.3.2. The allocation process The pure public and joint product models assume a noncooperative Nash allocation mechanism, which is a natural point of departure for modelling the behavior of agents who contribute to the supply of a public good. However, it is conceivable that a relatively small military alliance will behave cooperatively, especially over time as political pressure to contribute comes to bear on the free riders. Unfortunately, as outside observers, we have no way of knowing the "true" allocation process - we only observe the data generated from the process. McGuire and Groth (1985) speculated that the data patterns would be different under a Nash allocation scheme when compared to a Lindahl process. In particular, with unit prices and Cobb-Douglas utility functions, the relationship between wealth and defense expenditures should lie on a ray out of the origin under a Lindahl model. Under the Nash model, in contrast, the relationship should lie on a straight line with a negative y-intercept. McGuire and Groth went on to derive several econometric specifications for testing the underlying allocation process with observed data. Sandler and Murdoch (1990) operationalized the McGuire-Groth approach in order to test the allocation mechanism using NATO data. The essential theoretical contribution of Sandler and Murdoch was to derive reduced-form demand equations under the Lindahl and Nash hypotheses that are comparable empirically. Let 0i denote the Lindahl cost share for ally i, then Qi = Qi(Ii Oip,Ei)

(28)

is the ith ally's demand for the total military activity in the alliance. Note that a Lindahl equilibrium requires a set of cost shares that fully finance a level of Q= Q* such that Qi = Q* V i. The Lindahl Equation (28) differs from the Nash pure public good model,

100

J.C Murdoch

Equation (21), and from the Nash joint product model presented in Equation (22), illustrating that it is theoretically possible to identify the allocation mechanism. 3. Empirical contributions The earliest empirical studies focused on the relationship between measures of burden and economic size, using cross-sectional data from members of the NATO alliance. These studies sought primarily to investigate the inequality of defense burdens among allies at a particular point in time. In the 1960s when Olson and Zeckhauser (1966) proposed their theory of alliances, there were too few annual observations to consider a time-series methodology. In the latter half of the 1970s, however, the empirical methods switched so as to take advantage of the time-series variation in data from the individual NATO allies. The object of the analysis also changed from the burden sharing question to estimating demand equations like those presented in Equations (9), (17) and (22). This new focus facilitated an investigation into why the sharing of the burdens of defense might change as well as several direct tests of the public goods models. There are numerous empirical studies on military alliances [see Sandler (1993)] and just as many ways to categorize their results. Below, a small subset of the empirical findings are presented that, hopefully, represent the range of empirical issues and methods addressed in this literature. To organize the findings, I have categorized empirical exercises according to their antecedent paradigm, making three subsections tests derived from either the pure public good or the joint product model, and other interesting empirical tests. The scope of the presentation is limited in order to complement other chapters in this volume, especially Chapter 4, by Smith. 3.1. Tests based on the pure public good model Olson and Zeckhauser (1966, p. 274) hypothesized that within an alliance and all else equal, larger nations, in terms of economic size, should "devote larger percentages of their national incomes to defense than do the smaller nations." To test this hypothesis, they examined the Spearman rank correlation coefficient between GNP and the defense budget as a percent of GNP in 1964 for fourteen NATO nations. The null hypothesis of no association was rejected in favor of the hypothesis of positive association. This simple test provided strong evidence to support the pure public good model; i.e., it appeared as though the small NATO allies were able to exploit the larger allies. Interestingly, Olson and Zeckhauser's definition of defense burden (i.e., the percentage of national income devoted to defense) became the de facto definition for future research, and little attention was paid to alternative measures of burden until the 1980s [Sandler and Forbes (1980)]. 9 9 Russett (1970) and Oneal and Elrod (1989) noted that the military budgets across nations may not be comparable, owing to the conscription practices of different nations. Oneal and Elrod provide some corrected defense burden measures but they do not affect the basic conclusion of Olson and Zeckhauser.

Ch. 5:

Military Alliances: Theory and Empirics

101

Olson and Zeckhauser were reluctant to take a parametric approach by, for example, regressing national income on defense burden. The primary problem with regression methods is that within an alliance there are typically some very large (USA) and small (Luxembourg) nations, meaning that a fitting technique that weights observations according to squared deviations from the mean, such as ordinary least squares (OLS), gives misleading parameter estimates. OLS gives so much weight to the very large nation that the fitted line is forced through that point regardless of the general pattern of the other data points. Dudley and Montmarquette (1981) circumvented this problem by formulating a median-voter model. Their approach allowed them to pool countries from NATO and other alliances into one cross-sectional sample. To summarize the median-voter model estimated by Dudley and Montmarquette, define the following notation where all expenditures are in real terms 10: MEi = military expenditures in nation i, Ni

=

population in nation i,

GNPi

= gross national product in nation i,

SPILLi

= sum of the MEi of nation i's allies that are in the sample,

OSPILLi

= sum of the MEi of nation i's allies that are not in the sample,

Ei

= auxiliary variables that may shift the relationship. Then, using a Stone-Geary utility function, they derive an empirical approximation to the demand for defense as ME= Ni

0

+

(GNPi) +2( )i +

I

°N--i

) +3 (SPILLi+ OSPILLi)

+ 4 Ei +

,

(29)

Ni

where the O3's are unknown parameters and ei is a random disturbance term. The inverse population measure (1/Ni a) facilitates a test of the publicness of defense in consumption, since as a approaches zero, defense approaches a pure public good in the model . Dudley and Montmarquette (1981) used data from 38 countries and several alliances. The parameters were estimated with data from 1960, 1970, and 1975 using a fullinformation maximum likelihood (FIML) routine. The FIML estimator was necessary because of the simultaneity between the MEi and SPILLi. The results generally supported the pure public good model with the median-voter as the decision maker; i.e., /3 was positive and significant, /3 was negative and significant, and a was not significantly different from zero. Moreover, the Stone-Geary utility function setup allowed Dudley and Montmarquette to solve for the underlying parameters of the

'° In " In

Some studies presented in this section use gross domestic product (GDP) to measure national income. those cases, I will use GDPj to denote the variable. The linearity of the form and the test for publicness are artifacts of the Stone-Geary utility function. ageneral model, it is difficult to identify the a parameter.

102

JC.Murdoch

model. They found that defense was a superior good (income elasticity greater than one) and that the elasticity with respect to the tax price of defense was somewhat elastic. As noted in Section 2, the equalizing of the defense burdens in NATO resulted in the formulation of the joint product model. In a series of empirical papers, Oneal (1990a, 1990b) and Oneal and Elrod (1989) argued that, with respect to NATO, the basic predictions of the pure public good model held true well into the 1980s. Additionally, they argued that the potential for complementarity between the joint products, owing to the doctrine of flexible response, was actually quite limited 12. Thus, NATO even in the 1980s looked much like a "uniquely privileged group" with the USA shouldering the bulk of the defense for the alliance. There are many dimensions to their argument; however, the essential empirical evidence in its favor is that, after appropriate controls, the estimated relationship between measures of defense burden and measures of economic size is positive and significant. The problem with the simple rank correlations, which, as indicated above, showed that the relationship was not significant after the mid-1960s, is that they do not hold the other influences constant. Oneal (1990a) used data from 15 NATO countries for the years 1950-1984 to estimate the following relationship: ME GDPit

( -

i

GDPit GDPit

2

/32E .

(30)

Thus, the economic size variable is stated relative to the total economic size of NATO, where size is gross domestic product (GDP). Oneal includes a measure of Soviet defense burdens, European interdependence, and two dummy variables, one indicating Greece, Turkey, and Portugal and one indicating the other European nations. The parameters are estimated using a pooled, time-series, cross-sectional estimator. The estimate on /31 is positive and significant and robust to alternative samples and transformations of the size measure. Okamura (1991) provided a unique and interesting application of the pure public good model to the USA-Japan alliance. Okamura employed a Linear Logarithmic Expenditure System (LLES) to derive the Nash reaction functions for the USA and Japan. The real defense expenditures of the Soviet Union (THREAT) entered both reaction functions, while the defense expenditures of the NATO allies, excluding the USA (MEN), entered the US equation. For the USA, the reaction function was ME =PM(YMo + yM1 THREAT) +[GDP- PM(YMO + YMI THREAT) - Pc(yco + Ycl THREAT)]

(31)

x [al + PI ln(PM) + /32 ln(Pc)] + [al + Pi ln(PM) + /32 ln(Pc) - 1] x (MEj + MEN), where PM and Pc denote the price deflators of military expenditures and consumption expenditures, respectively, and MEj denotes the military expenditures of Japan. The 12

See Murdoch and Sandler (1991) for critical comments on this point.

Ch. 5: Military Alliances: Theory and Empirics

103

equation for Japan was similar except that MEN did not enter the model. The parameters of the two-equation system (y's, a's, and /3's) were estimated using the expenditure share equations. The LLES estimates were generally consistent with the pure public good model. Both nations responded positively to GDP and Soviet defense expenditures. The USA responded negatively to Japan's defense expenditures, while Japan responded negatively to the defense spending in the USA. Okamura reported a stable Nash equilibrium with the USA spending $150.79 billion and Japan spending $51.64 billion in 1982 dollars. Despite the fact that the price term did not perform in the expected fashion in Japan's equation, the study stands out for including measures of the price of defense in the equation.

3.2. Tests based on the joint product model Sandler and Forbes (1980) extended the empirical methodology of Olson and Zeckhauser to test the pure public good model and some predictions from the joint product model using data from NATO allies. Their first test was to compute year-byyear Kendall rank correlation statistics () between GDPi and MEi/GDPi.As expected, they found generally significant correlations before 1967 but insignificant correlations afterwards. They then computed the first- and second-order partial 's in an attempt to hold constant the private benefits of defense. In the first-order , the GDP per capita was held constant, while in the second-order , both GDP per capita and exposed border were held constant. Even the higher-order statistics were generally insignificant after 1967, suggesting that the pure public good model must be rejected. Sandler and Forbes next looked for evidence in favor of the joint product model. They argued that, under the joint product model, the private benefits of defense should predict the defense burden sharing; i.e., relative economic size is not the only predictor. Thus, they compared the relative defense burdens (MEi/ Zi MEi) to a private benefit measure, finding that by 1975 the private benefits did in fact predict the defense burdens better, when compared to 1960. The private benefit measure was an average of the GDP share, the exposed border share, and the population share. It is interesting to note that the questions addressed by Oneal (1990a) and Sandler and Forbes are similar. However, the two studies used quite different statistical methods and drew essentially opposite conclusions. As discussed below, the weight of the evidence now favors the conclusions drawn by Sandler and Forbes. Two papers by Murdoch and Sandler (1982, 1984) changed the empirical methods of alliance studies from cross-sectional to time series and emphasized country-by-country comparisons of the "demand for military expenditures". The demand interpretation was derived from the reduced-form demand equation [e.g., Equation (17)]. With the price of the military activity set equal to 1, the military expenditures are the best measure of qi.

104

JC. Murdoch

The Murdoch-Sandler approach involved estimating MEit = Pio +Ail GDPit + i2SPILLit-l + /i 3Eit + it

(32)

for each i in an alliance. Thus, they generally presented an estimated equation for each ally that was based on the time-series variation in the data. Various measures have been considered in Eit, including the ratio of the USA's strategic defense force to that of the Soviet Union's, force thinning, the price of oil, a dummy variable for the Viet Nam war, and Soviet defense expenditures. The simultaneity between MEi and SPILLi was addressed by lagging the spillin measure. Murdoch and Sandler (1984) found that for some of the NATO allies, the response to spillins (i2) was positive (5 out of 10), while for others it was negative. The negative response is not inconsistent with the pure public good or joint product model. However, the positive responses suggest rejecting the pure public good model. To devise a cleaner test of the joint product model, they argued that the doctrine of flexible response resulted in increased complementarity between the joint products and, therefore, a shift in the /i2 coefficients toward the positive direction. Their estimates of Equation (32) with a slope-shifter on SPILLit after 1974 indicated that Pi2 was more positive after the doctrine of flexible response in 7 out of the 10 allies in their sample. Hence, the evidence seemed to favor the joint product model. Two other tests of the applicability of the joint product model to the NATO alliance are worthy of mention. The first is by Hansen, Murdoch and Sandler (1990) who disaggregated the military expenditures in NATO into expenditures on conventional forces and strategic forces. This disaggregation allowed them to identify complementarity between conventional forces and substitutability between conventional forces and strategic forces. Thus, free riding in NATO was only possible on the strategic forces provided by the nuclear allies. The second test of the joint product model was the implementation of Equation (22) by Sandler and Murdoch (1990). Using time-series data for NATO allies, they estimated E

MEit = io + f3il (GDPit +SPILLit) + i2SPILLit + 3i3Eit + it.

(33)

Note that in Equation (33) the dependent variable is the same for each ally. From Equation (21), the pure public good model implies /32 =0. Using data from 1956 to 1987 and a two-stage least squares estimator, Sandler and Murdoch rejected this restriction for all 10 of the allies in their study. These parametric results, coupled with the nonparametric findings by Sandler and Forbes (1980), provide convincing evidence in favor of the joint product model. 3.3. Other empirical contributions In this section, three studies that offer unusually interesting empirical contributions are reviewed. These studies are chosen not so much for their insights into military

Ch. 5:

105

Military Alliances: Theory and Empirics

alliances, but because they illustrate how the study of military alliances has generated empirical techniques that may be useful for a host of public economic questions. The first study is by Murdoch, Sandler and Hansen (1991) who used data on the NATO allies to test between the "oligarchy" and the median-voter model. In the oligarchy model, the budget set constrains the choices of a decision-making body that acts on behalf of the entire nation, while under the median-voter model, the defense allocation decision is made by the median voter; hence, the appropriate budget constraint is that of the median voter's. With the exception of Dudley and Montmarquette (1981), most military alliance studies have implicitly or explicitly assumed an oligarchy choice perspective. Murdoch, Sandler and Hansen (1991) showed that restrictions on

MEt I 3+fi

it

( )

i2 (

G

it) +/i3SPILLit + Eit

(34)

are consistent with the oligarchy or median-voter model. Hence, the specification provides an empirical method for discriminating between the two models. In Equation (34), Lit represents the employed labor force, but in another application it could be equal to Nit. For A= 1, the specification is consistent with the median-voter model, while for Ai = 0 and Pil = 0, the specification is consistent with the oligarchy choice model. Interestingly, these authors found that some of the NATO allies generate median-voter-like data, while others appear to follow the oligarchy choice or neither paradigm. Specifically, the median-voter model could not be rejected for Belgium, the Netherlands and the United Kingdom, while the oligarchy model fitted the data from West Germany, France, Italy, and the USA. Neither model appeared correct for Canada, Denmark, and Norway. The second study is by Sandler and Murdoch (1990) who tested the Nash allocation mechanism against the Lindahl mechanism, again using data from the NATO allies. The question of which allocation mechanism is an important one because it addresses the efficiency question; i.e., a Lindahl allocation implies Pareto efficiency. The Sandler-Murdoch test requires estimates from four equations for each nation. The first is derived from the Nash joint product model: E

MEit = io + fil (GDPit +SPILLit) + Pi2SPILLit + fi 3Eit + it,

(35)

while the second represents Lindahl

2d MEit =

3

io + fiil (GDPit) + 13i2

ME

/ E+ I Elit. it +

(36)

The third and fourth equations are "joint" or combinations of Equations (35) and (36) that serve as auxiliary equations in order to implement the J test developed

106

JC. Murdoch

by MacKinnon, White and Davidson (1983). The J tests computed by Sandler and Murdoch provided some evidence to support the Nash assumption in five of the smaller allies, while the five largest allies could not be classified. Because no support for the Lindahl model was found, the authors suggested that NATO provided a suboptimal level of defense over the time period under investigation. The last study addresses the technology of public supply. Conybeare, Murdoch and Sandler (1994) tested the best-shot and weakest-link models in four military alliances - the Triple Alliance (1880-1914), the Triple Entente (1880-1914), the Warsaw Pact (1963-1987), and NATO (1961-1987). The authors show, assuming the existence of joint products, that the technology of public supply can be empirically determined by a recursive system of equations. They find support for the best-shot technology in the Triple Alliance, limited support for the weakest-link technology in the Triple Entente, and generally inconclusive findings in the Warsaw Pact and NATO.

4. Conclusion Theoretical research in military alliances closely parallels research in the theory of public goods. For over twenty-five years, economists have made significant contributions to our understanding of military alliances by modelling the behavior of individual allies as rational members of a collective that provides a public good. In addition to insights concerning burden sharing and allocative efficiency, our ability to explain and predict changes in ally behavior has increased over time. Probably the most important theoretical paradigm to date has been the development and application of the joint product model, which repeatedly appears to explain actual data better than the pure public good model. Perhaps the most important empirical development has been the application of annual time-series data to test the theoretical propositions. The gain in flexibility from having individual country data has, unquestionably, led to several creative tests, some of which have been applied to other public good problems. Future research should concentrate in three areas. First, methods must be engineered to test allocative processes other than Nash and Lindahl. One possibility is to extend the hybrid models suggested by McGuire and Groth (1985), so that they could be estimated with time-series data. Alternatively, a repeated game structure might facilitate other allocation mechanisms and provide some structure for modelling actions between alliances. The second area of future research concerns membership size, which, especially with respect to NATO, has been ignored. However, recent political events indicate that NATO will soon need to decide if some of the Eastern European nations should be admitted to the alliance. Lastly, more research must be done on the technology of public supply. The techniques developed by Conybeare, Murdoch and Sandler (1994) should be extended to other alliances and supply functions. It may prove particularly fruitful to integrate a geographic approach to the technology of public supply by modelling the spatial interaction between allies.

Ch. 5: Military Alliances: Theory and Empirics

107

References Andreoni, J., and M.C. McGuire, 1993, Identifying the free riders: A simple algorithm for determining who will contribute to a public good, Journal of Public Economics 51, 447-454. Bergstrom, T.C., L. Blume and H. Varian, 1986, On the private provision of public goods, Journal of Public Economics 29, 25-49. Conybeare, J.A.C., J.C. Murdoch and T. Sandler, 1994, Alternative collective-goods models of military alliances: Theory and empirics, Economic Inquiry 32, 525-542. Comes, R., and T. Sandler, 1986, The theory of externalities, public goods, and club goods (Cambridge University Press, New York). Comes, R., and T. Sandler, 1994, The comparative static properties of the impure public good model, Journal of Public Economics 54, 403-421. Dudley, L., and C. Montmarquette, 1981, The demand for military expenditures: An international comparison, Public Choice 37, 5-31. Hansen, L., J.C. Murdoch and T. Sandler, 1990, On distinguishing the behavior of nuclear and non-nuclear allies in NATO, Defence Economics 1, 37-56. Hirshleifer, J., 1983, From weakest-link to best-shot: The voluntary provision of public goods, Public Choice 41, 371-386. Knorr, K., 1985, Burden-sharing in NATO: Aspects of U.S. policy, Orbis 29, 517-36. MacKinnon, J.G., H. White and R. Davidson, 1983, Tests for model specification in the presence of alternative hypotheses: Some further results, Journal of Econometrics 21, 53-70. McGuire, M.C., 1990, Mixed public-private benefit and public good supply, with an application to the NATO alliance, Defence Economics 1, 17-35. McGuire, M.C., and C.H. Groth Jr, 1985, A method for identifying the public good allocation process within a group, Quarterly Journal of Economics 100, 915-934. Murdoch, J.C., and T. Sandler, 1982, A theoretical and empirical analysis of NATO, Journal of Conflict Resolution 26, 237-263. Murdoch, J.C., and T. Sandler, 1984, Complementarity, free riding, and the military expenditure of NATO allies, Journal of Public Economics 25, 83-101. Murdoch, J.C., and T. Sandler, 1991, NATO burden sharing and the forces of change: Further observations, International Studies Quarterly 35, 109-114. Murdoch, J.C., T. Sandler and L. Hansen, 1991, An econometric technique for comparing median voter and oligarchy choice models of collective action: The case of the NATO alliance, Review of Economics and Statistics 73, 624-631. Okamura, M., 1991, Estimating the impact of the Soviet Union's threat on the United States-Japan alliance: A demand system approach, Review of Economics and Statistics 73, 200-207. Olson, M., 1965, The logic of collective action: Public goods and the theory of groups (Harvard University Press, Cambridge, MA). Olson, M., 1986, A theory of the incentives facing political organizations, International Political Science Review 7, 165-179. Olson, M., and R. Zeckhauser, 1966, An economic theory of alliances, Review of Economics and Statistics 48, 266-279. Oneal, J.R., 1990a, Testing the theory of collective action: Defense burdens, 1950-1984, Journal of Conflict Resolution 34, 426-448. Oneal, J.R., 1990b, The theory of collective action and burden sharing in NATO, International Organization 44, 379-402. Oneal, J.R., and M.A. Elrod, 1989, NATO burden sharing and the forces of change, International Studies Quarterly 33, 435-456. Palmer, G., 1990, Corralling the free rider: Deterrence and the Western alliance, International Studies Quarterly 34, 147-164.

108

J.C. Murdoch

Russett, B.M., 1970, What price vigilance? The burdens of national defense (Yale University Press, New Haven, CT). Sandler, T., 1977, Impurity of defense: An application to the economics of alliances, Kyklos 30, 443 460. Sandler, T., 1993, The economics theory of alliances: A survey, Journal of Conflict Resolution 37, 446-483. Sandler, T., and J.E Forbes, 1980, Burden sharing, strategy, and the design of NATO, Economic Inquiry 18, 425-444. Sandler, T., and K. Hartley, 1995, The economics of defense (Cambridge University Press, Cambridge). Sandler, T., and J.C. Murdoch, 1990, Nash-Cournot or Lindahl behavior?: An empirical test for the NATO allies, Quarterly Journal of Economics 105, 875-894. van Ypersele de Strihou, J., 1967, Sharing the defense burden among Western allies, Review of Economics and Statistics 49, 527-536. Warr, P.G., 1983, The private provision of a public good is independent of the distribution of income, Economics Letters 13, 207-211.

Chapter 6

ARMS RACES AND PROLIFERATION* DAGOBERT L. BRITO Rice University MICHAEL D. INTRILIGATOR University of California, Los Angeles

Contents Abstract Keywords 1. Introduction and overview of arms races, proliferation, and the outbreak of war 2. The Richardson model of the arms race 3. The dynamics of arms accumulation: the rational actor model 4. The dynamics of nuclear war: strategy and outcomes 5. Arms races and stability 6. A general model of an arms race 7. Power, bargaining, and war 8. The role of nuclear weapons 9. Nuclear weapons in a formal model of bargaining 10. Nuclear proliferation 11. Proliferation and the probability of war 12. The future of the arms race and of proliferation 13. Appendix 13.1. 13.2. 13.3. 13.4.

Proof of Proof of Proof of Proof of

proposition proposition proposition proposition

1 2 4 7

References

110 110 111 114 117 121 126 128 130 144 146 150 151 155 160 160 160 160 161 161

* Our work was supported by the Baker Institute at Rice University, The Center for International and Strategic Affairs at UCLA, the Ford Foundation, The Mershon Center at The Ohio State University, and the National Science Foundation. We would like to thank Radu Filimon, Jonathan Hamilton, Jack Hirshleifer, Steven Slutsky, Erica Worth, and Dina Zinnes for their help over the years. Todd Sandler and Keith Hartley made invaluable contributions as editors to this chapter. Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

110

D.L. Brito and M.D. Intriligator

Abstract Previous analyses of arms races and proliferation are integrated and extended, building from a treatment of the behavioral foundations of weapons acquisitions to a general theory of arms races, with implications for the role of negotiations, the balance of power, the timing of crises, and nuclear proliferation. Recent developments in economic theory are also applied here to the problems of the arms race, nuclear proliferation, and the outbreak of war, yielding a deeper treatment of these phenomena by directly or indirectly treating asymmetric information in bargaining, repeated games that involve threats, and principal-agent problems in decisions on technology and weapons accumulation.

Keywords accidental nuclear war, arms races, asymmetric information, bargaining, crisis stability, deterrence, equilibrium, nuclear proliferation, nuclear war model, outbreak of war, precommitment, probability of war, repeated games, rational choice models, reaction curves, Richardson model, stability, stock adjustment model, timing of crises, warfighting

Ch. 6: Arms Races and Proliferation

111

1. Introduction and overview of arms races, proliferation, and the outbreak of war In the nuclear era the potential implications of arms races and the proliferation of weapons of mass destruction for life and death are staggering. Their political implications in terms of power and influence and their economic implications in terms of resources mobilized for weapons production are of overwhelming importance. Nevertheless, arms races have received relatively little attention from social scientists, particularly economists, in recent years. The interpretive guide to the literature on conflict theory in Intriligator (1982), which organizes the literature by eight analytic approaches and eight areas of application, shows that most of the work in this field has been conducted by political scientists and that there tends to be a clustering of work on certain key approaches and models, such as the Richardson model of arms races, neglecting many important areas of potential inquiry and many possible alternative methodologies. While economists did some fundamental and relatively early work on these and related topics, as in Morgenstern (1959), Schelling (1960, 1966), and McGuire (1965), relatively few economists have conducted research in this area. This neglect by economists is both surprising and unfortunate because many of the essential features of arms races, proliferation, and related phenomena, such as arms control and strategic choices in warfare, involve resource allocation, oligopolistic interactions, the production of public goods, bureaucratic and organizational behavior, and externalities, which are subjects of central concern to economists. The purpose of this chapter is to integrate and to extend previous analyses of arms races and proliferation, building from a treatment of the behavioral foundations of weapons acquisition to a general theory of arms races. Some recent developments in economic theory are also applied here to the problems of the arms race, nuclear proliferation, and the outbreak of war. These new approaches yield a deeper treatment of these phenomena by directly or indirectly treating such salient aspects of an arms race as asymmetric information in bargaining, repeated games that involve threats, and principal-agent problems in decisions on technology and weapons accumulation. Various approaches have been used to study arms races, including the Richardson model, as discussed in the next section, and models based on dynamic optimization l. Much of the previous theoretical work on the arms race is in the Richardson (1960) tradition, which explains the arms race descriptively and mechanically. The Richardson model uses a coupled pair of differential equations to explain the change in levels of weapons in each of two nations as a function of the weapons held by both sides. More recent work in this area, by contrast, has used rational choice models to explain the arms race in terms of the behavior of decision makers, making use of the postulates of rationality and maximizing behavior. In our prior separate work, for example, in Brito

For a survey of the arms race literature see the book edited by Gleditsch and Njolstad (1990). See also Sandler and Hartley (1995).

112

D.L. Brito and M.D. Intriligator

(1972) and in Intriligator (1975), we used an optimal control approach of optimization in a dynamic setting to explain choices of arms acquisitions over time. We each obtained a general set of equations describing an arms race of which the Richardson model is one special case. The value of arms in such a framework is a consequence of the behavioral decisions of defense decision makers on both sides in seeking either to deter or to attack the other side. As in the case of general equilibrium theory, a common denominator in much of the literature on arms races is concern over the existence and stability of equilibrium. Equilibrium, in the case of an arms race, as elsewhere, refers to a situation in which there is no change over time in certain variables characterizing the system. From a behavioral standpoint, an equilibrium is a Nash equilibrium in which none of the decision makers has an incentive to change their decisions, given the choices of the other decision makers. In certain cases it is possible to obtain functional dependencies giving the equilibrium levels of weapons on each side as functions of technical and strategic parameters. Another common denominator in much of this chapter is the concept of stability, which is an essential ingredient in any study of arms races and related phenomena. Stability is a concept that occurs in various and often conflicting guises in this chapter. First, there is the conventional notion of the stability of the quantitative arms race, involving determining whether nations will engage in the proverbial behavior of piling arms upon arms or whether this process will stop at some stable equilibrium, as in the Richardson model with certain restrictions on its parameters. Second, there is the stability of the qualitative arms race, involving determining whether nations engage in competitive research and development to acquire a technological advantage. Another concept of stability is that of crisis stability: under what circumstances will a crisis between rival nations, such as the Cuban missile crisis of 1962 between the United States and the Soviet Union, lead not to the outbreak of war but rather to some type of peaceful resolution? This is a fundamental question of international relations that has been only partly answered by political theorists. Some wars have occurred due to irrational behavior, accidents, or imperfections in the process of decision making. In such cases, paradoxically, the stability of the arms race in the conventional sense of low levels of weapon accumulations may be inconsistent with crisis stability. Low levels of weapons accumulations may in some cases even increase the probability of war. All of these aspects of stability must be treated when considering arms races that involve multiple countries and thereby nuclear proliferation, referring to the possibility of their acquiring nuclear weapons, other weapons of mass destruction, and delivery systems 2. As other nations acquired nuclear weapons, such as the United Kingdom, France, and China, it was necessary to treat them as additional participants. Regional conflicts have also involved the acquisition or potential acquisition of nuclear weapons, including the Arab-Israeli conflict, the conflict between India and Pakistan,

2

For a survey of the proliferation literature see the book edited by Davis and Frankel (1993).

Ch. 6:

Arms Races and Proliferation

113

that between Iran and Iraq, that between North and South Korea, and others. The possibility of such nuclear proliferation can imply that the results about equilibrium and the various stability properties of the bipolar arms race may no longer be valid. Most previous work on proliferation has assumed, explicitly or implicitly, that additional nuclear powers would be destabilizing in the sense of increasing the probability of war. By contrast, Waltz (1981, 1983) concluded that more nations with nuclear weapons would reduce the chance of conflict. In Intriligator and Brito (1981) and Brito and Intriligator (1983) we analyzed the consequences of additional nuclear weapons states on crisis stability. We concluded that nuclear proliferation may have different qualitative effects on the probability of nuclear war depending on the number of existing nuclear nations and the level of weapons held by the various countries involved. We found that, in certain circumstances, but not universally, in contrast to Waltz, having an additional nuclear power in the system may reduce rather than increase the probability of deliberate nuclear war. This effect occurs when the restraining influence of the additional nuclear weapon state on the prior states with nuclear weapons offsets the destabilizing influence of there being an additional state able to start a nuclear war. However, this whole line of research is open to question since we have been able to show (see below) that qualitative predictions about the change in the probability that individual nations will initiate a nuclear war as the number of nuclear powers increases are not sufficient to make predictions about the change in the aggregate probability of a nuclear war. A third influence that is always destabilizing is that of the potential for accidental nuclear war, implying the value of introducing safeguards against accidental or inadvertent use of nuclear weapons. In the absence of such safeguards in the new nuclear nations, the probability of an accidental nuclear war will increase and, as a result, the aggregate probability of war will likely increase. The explicit treatment of these three different ways in which proliferation may change the probability of war outbreak is more useful in analyzing or formulating policy in this area than either implicitly assuming that proliferation is always destabilizing, as in the traditional approaches, or explicitly assuming that it is always stabilizing, as in the Waltz approach. A fundamental question of international relations that has been only partly answered by political theorists is that of when and under what circumstances conflict between nations leads to the outbreak of war. Some wars have occurred due to irrational behavior, accidents, or imperfections in the process of decision making. Other wars have occurred due to the use of force in an attempt to change the allocation of rights between nations, such as the Iraqi invasion of Kuwait in 1990 and the subsequent Gulf War to restore the previous situation. If countries are in conflict over the rights to an asset, such as land, population, or oil, then war can occur as a result of an attempted redistribution of such rights. Thus, yet another aspect of stability in an arms race is that of a stable distribution of wealth and power. Some of the models that we have developed provide a way of analyzing this type of war. There are three factors that are clearly relevant to the outbreak of war. The first is imperfect information, where the

114

D.L. Brito and M.D. Intriligator

relevant decision makers may be uninformed about salient aspects of the problem. The second is the redistributionofresources, that can act as an equilibrating mechanism and thus reduce the likelihood of war. The third is risk taking, in which nations may engage in a game of brinkmanship in order to gain or to avoid losing resources. The history of war suggests that all three factors have played important and, in some instances, crucial roles in explaining why wars did - or did not - occur. The important idea is that there is a relationship between the distribution of power and the distribution of wealth. Conflict can occur if these two relationships are not in equilibrium. We have studied war outbreak and related topics, such as the influence of an arms race on the likelihood of war, using a two-period model in Brito and Intriligator (1985). In the first period the two countries choose between consumption and investment in arms, so this period involves a potential arms race. In the second period, the countries bargain over the reallocation of resources, possibly using force or the threat of force to attain their goals, so this period involves a potential war. Our model builds on recent work on bargaining with asymmetric information and employs some recent developments in models of sequential equilibria. The model implies that if both countries are fully informed as to the parameters of the problem, then there will be no war but rather a voluntary redistribution of resources. If, however, there is a situation of asymmetric information in which one of the countries is informed about all parameters of the problem but the other is not so informed, then war can occur if the uninformed country adopts a strategy in which it precommits itself to a positive probability of war in order to prevent bluffing by the informed country. There would be no conflict if the uninformed country finds it optimal not to attempt to prevent such bluffing or, alternatively, if either country has very few resources and thus chooses not to participate. If each of the countries is fully informed about the characteristics of the other country, then when transfers occur, there is no conflict. There exists an initial distribution of resources that leads to no transfers and the two countries would invest in weapons to deter rather than to engage in a war. A surprising result is that a Nash equilibrium in pure strategies does not exist in the fully informed case. However, a Nash equilibrium in the space of mixed strategies does exist, where some of the realizations of the mixed strategies might lead to a crisis. Deterrence requires the use of mixed strategies, with countries choosing positive probabilities of both a war-deterring strategy and a war-fighting strategy. A model in which nations are rational actors and in which threats and deterrence play a major role in the distribution of resources yields significant insights into why deterrence may fail in the real world. Nations are not single rational actors, however, and rational behavior within a bureaucratic and hierarchical organization may lead to undesired consequences. 2. The Richardson model of the arms race The Richardson arms race model constitutes one of the best known models of arms race phenomena and, at the same time, one of the most influential formal models

Ch. 6: Arms Races and Proliferation

115

in all of the international relations literature 3 . Different types of models have been used in the study of arms races, including descriptive and normative models. The Richardson model is an example of a descriptive model, one with neither an explicit objective nor an assumption of maximizing behavior. Another such descriptive model of an arms race is the stock adjustment model. Both of these models are intended primarily to describe arms race phenomena. Another type of model is the normative model, with an explicit goal and the assumption of maximizing behavior. This type of model is intended to explain the underlying motivations for arms races in terms of goal-directed behavior. One such normative model is the optimal resource allocation model of Brito (1972). Another is the differential game model developed by Simaan and Cruz (1975, 1976), which applies the Nash solution to the basic resource allocation model proposed by Brito. We begin by defining our terms and assumptions. By an arms race we mean the dynamic process of interaction between countries in their acquisitions of weapons. Like Richardson, we treat only two countries, labeled 1 and 2, thus avoiding such issues as proliferation, alliance formation, multicountry stability, etc. From an empirical standpoint, much of the observed interaction in arms races is in fact between two countries or two alliances. In nuclear weapons, there was the US-USSR superpower arms race interaction (or, more broadly, NATO vs. Warsaw Pact) with only relatively insignificant impacts of other nuclear powers. Conventional weapons arms races include the Arab-Israeli arms race, the Iran-Iraq arms race, and many others. We also assume, again as in Richardson, that there is a single homogeneous weapon, where country 1 has wl weapons and country 2 has w2 weapons. The Richardson model can be extended to the more complex case of several weapons types, but both theoretical and empirical arguments justify this assumption. A model with several types of weapons would considerably complicate the basic theory as it would be necessary to address the problem of portfolio selection. It is important to start with the simpler analysis of a single type of weapon. It is possible, from an empirical standpoint, to aggregate either nuclear weapons or conventional weapons (but not the two together) into an overall measure of military capability. The theory to be developed can be considered applicable to one of these aggregates. The Richardson model is summarized by two differential equations describing the rate of change over time of weapon stocks in each of the two countries. For country 1, if wl(t) is the stock of weapons at time, then iwl = dwl/dt is the rate of change of weapon stocks in country 1 at time t. According to the Richardson model, this change is described as the sum of three separate influences. The first is the defense term, where the accumulation of weapons is influenced positively by the stock of 3 Basic references for the Richardson model are Richardson (1939, 1951, 1960) and Rapoport (1957, 1960). Lewis Fry Richardson's seminal work became available in 1947 1949 in microfilm and was published in book form in 1960, as Richardson (1960). Rapoport (1957, 1960) provides a review and appreciation of this work. See Intriligator and Brito (1976b) and Isard and Anderton (1985); see Anderton (1985) for an extensive survey and bibliography.

116

D.L. Brito and M.D. Intriligator a2w -a w1

blW +b3 1 b2

-w.

Figure 1.

weapons of the opponent w2, representing the need to defend oneself against the opponent. The second is the fatigue term, where the accumulation of weapons is influenced negatively by one's own stock of weapons, representing the economic and administrative burden of the arms race. The third is the grievance term, representing all other factors influencing the arms race, whether historical, institutional, cultural, or others. In the Richardson model these terms are independent, additive, and linear, resulting in the two coupled linear differential equations: Vwl = alw2 - a2 wl + a3

(al and a2 > 0),

(1)

¢2 = blWl - b2 w2 + b3

(bl and b2 >0).

(2)

Equation (1) for country I determines weapons acquisitions as a function of the constants al, a2, a3, and the weapons stocks in both countries. According to the theory, al and bl are positive since the two countries are opponents. Also according to the theory, a2 and b2 are positive since maintaining weapons reduces the ability to acquire additional weapons. The grievance terms a3 and b3 can be positive or negative. At an equilibrium point of the dynamic process there is no change in weapon stocks, vil = 0 and vP2 = 0, yielding the reaction functions a2w - a3

a3

for country 1,

(3)

blW + b3 +b

for country 2.

(4)

W2 = -

al

W2b

b2

These reaction functions, illustrated in Figure 1, give the number of weapons each country holds as a (linear) function of the number held by the opponent, i.e., how each country reacts to weapons stockpiles of the other for an equilibrium to be attained. If the grievance terms a 3, b3 are positive and the stability condition a2b2 - albl >0 holds, then an equilibrium exists. This condition guarantees that the reaction functions

Ch. 6: Arms Races and Proliferation

117

will cross and that the resulting equilibrium will be stable. The equilibrium stockpile of weapons for country 1 is then given by a3b2 +alb3 a2b2 - albl

and a symmetric equation exists for vw' 2. If the stocks on both sides exceed the equilibrium then the force of the fatigue terms would offset the defense (and grievance) terms to reduce stocks on both sides to the equilibrium level. If, conversely, the stocks on both sides were below their equilibrium levels then the defense terms would offset the others to raise stocks on both sides. In the case where wl is "too large" and w2 is "too small" relative to the equilibrium, the force of the fatigue term for 1 and that of the defense term for 2 would reduce wl and increase w 2, restoring the equilibrium. While the Richardson model is the most influential model of an arms race, it is not the only such model. Another descriptive model is the stock adjustment model, such as those of Boulding (1962), Intriligator (1964), and McGuire (1977). Each country determines a desired stock of weapons, 1 and ¢2, and the rate of change of weapon stocks is assumed to be proportional to the discrepancy between desired and actual weapon stocks, as in the two differential equations: ¢

= a4 (1

Th2 = b4 (

-Wl),

(6)

- W2). ai2

(7)

Thus, each side acquires additional stockpiles in order to overcome a perceived deficiency in its stockpiles of weapons, the deficiency given by the gap between the desired and actual levels of weapons, for example, wl -w for country 1. In general, the desired stocks of weapons depend on the level of weapons in both countries. If, for example, each country desires to have a certain base level of weapons and to match increments of the other side's levels according to a fixed ratio then the desired stocks are linear functions of the levels of weapons held by the other side: ¢ = a5 + a6 w2,

(8)

W2 = b5 + b6 W1,

(9)

where a5 and b5 are the base levels and a6 and b6 are the fixed ratio of increments of one country's stocks relative to those of the opponent's. Inserting these in Equations (6) and (7) yields the Richardson model. Conversely, any Richardson model, as in Equations (1) and (2), can be interpreted as a stock adjustment model with the desired stocks linear functions of the weapons held by the opponent. 3. The dynamics of arms accumulation: the rational actor model Our goal in this section is to explain the arms race in terms of the behavior of decision makers using the postulates of rationality and maximizing behavior. In particular, we

D.L. Brito and M.D. Intriligator

118

derive the dynamics of arms accumulation in a model based on the axioms of rational choice, where it is assumed that each country can be modeled as a single rational actor. There are several schools of thought about the' stability of armament races. One is that arms races must have a stable equilibrium. This result comes either from a static model that treats an arms race as a game similar to "Prisoner's Dilemma" or from a dynamic model such as those of Richardson (1960), Saaty (1968), and Smoker (1965), which assume the equations of motion. A second school of thought is that arms races are unstable. This theory, often seen in the popular press, holds that, unless some agreement is reached, weapon stocks will increase in an ever-increasing spiral that must ultimately lead to bankruptcy or war. A third view is that a stable equilibrium may exist but that stability may only be a local property. Thus, a large disturbance of the system such as the introduction of a new weapons system may lead to an arms race. Here we will formulate a model that will permit us either to reconcile or to discriminate among these apparently conflicting viewpoints. Again consider a system with two nations. Each nation can use its production for either consumption or investment in its weapons stock. Each nation is trying to maximize a utility function over time whose arguments are consumption and defense. We address three questions: 1) Does there exist an equilibrium level of armaments? 2) If such an equilibrium exists, under what conditions will it be stable? 3) Under what conditions will a stable equilibrium occur at zero weapon stocks? We show, under very general assumptions given in Brito (1972), that there exists an equilibrium level of armaments. Further, we give some sufficient conditions for the equilibrium to be stable. Finally, we show that in a dynamic economic model total disarmament is unstable except under very special circumstances and that behavior very similar to that of the "Prisoner's Dilemma" game will be observed. We assume that: 1) Welfare in each country can be represented by a strictly concave twicedifferentiable concave utility function U(ci, D(wi, wj)) where ci is the consumption of the ith country and D(wi, wj) is a public good which will be called defense. We assume that at zero level of consumption the marginal utility of consumption is infinite. 2) The level of defense D(wi, wj) is given by the value function associated with a potential war or conflict that depends on the weapon stocks in each country, where wi is the weapons stock of the ith country. We assume that an increase in the weapons stocks of a country will increase its level of defense and that an increase in the weapons stocks of its rival will decrease its level of defense: (10)

0Di (Wi, wi) Owl

Ozi (i, wj)

o.

09w0 The exact dynamic game that defines the value function will not be specified for the moment, but it suffices to think of this function, associated with the outcome of a potential war, as a primitive construct, such as a production function.

Ch. 6: Arms Races and Proliferation

119

3) To reduce the dimensionality of the problem we assume that both countries are at an equilibrium capital-labor ratio and that they invest sufficiently to maintain their capital stocks. Further, we assume that the population is constant. The problem of the effect of armaments on growth and vice versa is important, but here we ignore it. Letting Y(t) be the net national product of the ith country and z i (t) be the expenditure on weapons at time t, the resource constraint is given by Y(t) = ci(t) + zi(t).

(12)

Let iwi be the resources necessary to maintain and operate weapon stock wi. Then W i j = Zi - 5iwi,

(13)

stating that net investment in weapons is gross expenditure less depreciation and operating expenses. 4) Finally, given the discount rate ri, we assume that at any point in time T, the ith country seeks to maximize welfare given by the present discounted value of utility, Ji(r) =

e- ri ( - r ) U(ci,Di (wi, wj)) dt

(14)

subject to the constraints given by Equations (12) and (13) and the jth country's reactions. Since this optimization depends only on the initial conditions, we can simplify notation by assuming that = 0. Assume initially that the parties to an arms race are Cournot-Nash players in that they are myopic with respect to the behavior of the other party, taking the behavior of the other party as given. We show that the dynamic equations for the arms race are given by general Richardson-type differential equations w i = R(wi, wj) ,

(15)

*j = Rj(wi, wj),

(16)

and that this dynamic system has a stable equilibrium at wii and rV. The CournotNash assumption of myopic behavior will be relaxed in subsequent sections, however it is useful to explore the implications of such simple strategic behavior. It provides a benchmark against which other strategic assumptions can be judged. Since the actual mechanism involved in the allocation of resources in the countries involved is a complex combination of political and bureaucratic behavior, there is some virtue in simplicity. The assumption of Cournot-Nash behavior implies that the level of the jth country's weapons stock wj is constant and parametric to the ith country's optimization. The current value Hamiltonian for the ith country's optimization is given by H = Ui (ci, Di (wi, wj)) + i (zi - 6ii) + i (Yi - i - ci),

(17)

where ~i is the costate variable associated with the ith country's weapon stocks and Xi is the Lagrange multiplier associated with the resource constraint. According to

120

D.L. Brito and M.D. Intriligator

the Pontryagin Maximum Principle the ith country would choose optimal paths c(t) and zi (t) (where * denotes the optimal solution) so as to maximize the Hamiltonian at each point in time for the ~*(t) which is the solution to the costate differential equation given below as well as the wi*(t) which is the solution for the state differential equation for the appropriate boundary conditions [Intriligator (1971)]. It follows from the Berge Maximum Theorem and the strict concavity of the Hamiltonian in ci that solution of the optimization c (i, Wi, Wj) and z (i, wi, Wj) are both continuous functions of ~i, wi, and wj. The assumption that the utility function is differentiable implies that A,'(i,wi,wj) is also continuous. For a given wj, the dynamics of the ith country's accumulation of arms is then given by the autonomous (time independent) differential equation system for the state and costate variables (18)

W,' Wij) - iwi,

Wi - Zi (i

(i +iUi)(c*, D i (w*, wj) = iOi

Di (w, wj)

(19)

Owi

Since the differential equation system is autonomous, Equations (18) and (19) can be combined to yield a differential equation in phase space, which, together with the boundary conditions, can be integrated with respect to wi to yield the optimal costate (20)

i*= O(wi*, wj). Equation (20) can then be substituted into Equation (18), resulting in Vi*

= Zi

(i (W, Wj) ,Wi,Wj) -

iWi,

(21)

and a similar equation can be derived for the jth country: Wi* = z

(

(Wi, Wj*) , W, Wj*) - (S6wj*.

(22)

Together, Equations (21) and (22) provide the rolling plan for both countries at time t. Implementation of the plan gives the dynamics of an arms race when the principals behave as Cournot-Nash players. This is a general nonlinear version of Richardson's equations as given by Equations (15) and (16). Letting -wi, be the solution of Ri (wi, wj) = 0, it is clear from Equation (20) that w =0 implies that Ii =0 and that from Equation (19), OUi (c*, Di ( ODi

wji) ) Di Di,(i, Wj) = (ri +- 5 ) O~i (iv, wj) OWi

(23)

This is a standard result in capital theory, stating that the marginal product of capital is equal to the cost of capital. Furthermore, solving the equations for wj as functions of wj yields i = g, (j),

i = 1,2 j=1,2,

(24)

representing a general non-linear version of the reaction curves of the Richardson model.

Ch. 6: Arms Races and Proliferation

121

Proposition 1. Given the assumptions we have made, if 5i > 0 then there exists an equilibrium level of armaments, where the equilibrium is characterizedby 4

R (wI,b2) = 0

(25)

=0.

(26)

R2(W I, W2)

We have demonstrated the existence of a resource-constrained equilibrium. The assumption that the utility function satisfies the condition that the marginal utility of zero consumption is infinite implies that this equilibrium is always in the interior of the feasible set. If we assume that 6i = 0, then we can still prove that an equilibrium exists if we assume decreasing returns to scale in arms as this would imply that Equation (24) is a contraction mapping, as discussed in Brito (1972).

4. The dynamics of nuclear war: strategy and outcomes A criticism that can be lodged against all the models introduced so far is that they treat an arms race from the outside, in terms of a mechanistic model, rather than from the inside, in terms of the decisions of defense planners, as discussed in Intriligator (1967). They all ignore strategic considerations, in particular, the roles of weapons in both deterring and conducting a war. Intriligator (1975) developed a model that took such factors into account, connecting the acquisition of weapons to their use in both peace (deterrence) and war (warfighting). This section connects this model that treats strategic considerations as perceived by defense planners to the arms race. In both country 1 and 2, at any time t, the political authorities will generally require the military authorities to justify their proposed budgets, in particular, any request for an increase in the level of weapons, here called "missiles". They may also question the existing levels of missiles in terms of their danger and/or expense. The military authorities would typically seek to justify both their budgetary requests for missile acquisitions zi (t) and their current inventories of missiles wi(t) in terms of national security considerations by showing their potential for deterrence or warfighting. Both the deterrence and the warfighting roles of weapons can be addressed via considering a hypothetical missile war, such as in war gaming or in computer simulation. The simulation of a missile war can be described by the time paths for missiles in both countries wl(r), W2 (T) and for casualties in both countries Xl(r), X2 (T), where T is an index of time during the simulated war which starts at the present time r = t. These time paths can be considered the solutions to the following coupled system of

4 Proof in Appendix.

D.L. Brito and M.D. Intriligator

122

differential equations and the associated boundary conditions at time t for the four state variables wl, w2, xl, and x2 , which provide a dynamic model representation of the hypothetical or simulated missile war 5. 1l = -VlrWI - e2w2rv2sW2,

(27)

-V2rW2 - elwvlrvlswl,

(28)

1w2

W2,

(29)

x2 = elxvr (1- Vls) WI.

(30)

= e2 V2r ( - V2

)

In Equation (27), the stock of country 's missiles starts at wl(t), its current level, and, assuming the hypothetical war begins at time t, the stock declines for two reasons. First, country 1 fires its missiles at the rate Vlr, SO -IrW represents the decline in country 1 missiles due to its own firing decisions. Second, country 1 missiles are destroyed due to country 2's missiles launched against them. Of the V2rW2 missiles launched at time t, as shown in Equation (28), the proportion v 2 is launched counterforce against country 1 missiles. Thus, v2rv2sW2 missiles are launched and targeted at country I missiles at time T. If each destroys, on average, e2w of country 's missiles, then e2wv2r v2sW2 of country 1 missiles are destroyed at time r, as shown in Equation (27). The remaining country 2 missiles launched at time r, given as 2r(1-2s)W2, are targeted countervalue, at country 's cities, and they inflict e2xv2r(l-2s)w2 casualties in country 1, as shown in Equation (29), where e2x is the average casualty inflicted per countervalue weapon of country 2. Equations (28) and (30) represent similar processes for country 2. The evolution of the simulated war described by Equations (27) to (30) is determined by the initial weapons stocks; the strategic decisions, over time, of rates of fire and targets; and the technical parameters determining the effectiveness of weapons against enemy weapons and their effectiveness against enemy cities. It should be emphasized that this is a planning model of a hypothetical war, e.g., in a war game or a computer simulation, which can be used by the military authorities to justify current budgetary requests and force levels. In particular, it can be used to justify the force levels and weapons acquisitions needed to deter an actual war. (It should also be emphasized that this differential equation model of a simulated war should not be confused with the earlier differential equation models of an arms race, such as the Richardson model.) Deterrence, in this context, refers primarily to the ability to keep the opponent from initiating a war, so consider what happens in the hypothetical war if the enemy initiates the war at the present time t. The value function of the associated optimization problem can be thought of as the Di(wi, Wj) term from the last section. If the planners

5 See Intriligator (1967, 1975), Saaty (1968), Brito and Intriligator (1972-1974, 1977), and Intriligator and Brito (1976a). For a related model see Kupperman and Smith (1972, 1976).

Ch. 6: Arms Races and Proliferation

123

in country 1 have the objective function J1 [Wl(T), w 2(T),xl(T),x 2(T)], defined on the terminal stage of a hypothetical war to be maximized, then D1 (Wl, w 2) = max J 1 [l(T), w 2(T),x (T),x 2 (T)], vlr 1 ,vs

(31)

where the choice variables in the maximization are strategic variables, here the rates of fire and the target selection during the war. If the objective of the military authorities in country 1 is to deter country 2 from striking by threatening to inflict upon it an unacceptable level of damage on a potential retaliatory second strike, then they should have enough missiles to inflict at least such a level of casualties. It is then possible to solve for the minimum level of missiles required for country 1 to deter country 2 by inflicting this level of casualties. The result is a linear reaction function, as in the Richardson model, but one which can be solved as an explicit function of certain underlying strategic, technical, timing, and social/psychological factors, such as in the example below. Thus, the Richardson model coefficients in Equations (1) and (2) can be interpreted as consequences of underlying strategic and related factors. In the case of each country acting so as to deter one another, a stable equilibrium for the resulting arms race exists if certain stability conditions are met. To give a numerical example, consider the symmetrical case, where the parameters are the same on both sides and where the maximum rate of fire is 10 percent per minute, it takes two weapons to destroy one enemy missile, one missile inflicts 250000 casualties, the first strike initiation interval is 15 minutes, the second strike retaliation interval is 10 minutes, and the minimum number of casualties required for deterring the other side is 40 million. Then the solution for the equilibrium level of weapons held by each side (where the linear reaction functions cross) is 414, that is, with 414 weapons each side has enough to deter the other side. At this level on both sides, either side can absorb the 15 minute first strike of the other and still have enough weapons left to inflict the required number of 40 million casualties on the other in the 10 minute retaliatory second strike 6. Geometrically, in the case of the two countries each trying to deter the other, these two reaction curves are shown in Figure 2 as the lines marked "country 1 deters", and the comparable line for country 2, marked "country 2 deters". At points to the right of the alai line country 1 has a sufficient number of missiles to deter country 2, while at points above the a2a2 line country 2 has sufficient missiles to deter country 1. The two reaction curves intersect at E, which is the equilibrium level of missiles, e.g., 414 on both sides in the above numerical example. The shaded upper cone of mutual deterrence represents the region of mutual deterrence, where each deters the other. As long as the levels of missiles remain in this cone, each side deters the other and the

For a discussion of the sensitivity of the equilibrium levels of missiles to changes in the strategic, technical, timing, and social/psychological parameters see Intriligator (1975). 6

D.L. Brito and M.D. Intriligator

124

Figure 2.

situation will be relatively stable against the outbreak of war. Arms control through arms limitations or reductions is feasible as long as the situation remains in the cone of mutual deterrence. Another goal of military authorities could be to avoid a preemptive strike by having enough missiles relative to those on the other side that it will not appear to the other side that an attack could be successfully implemented. If the other side believes that such an attack could be carried out then it might be forced to make its own preemptive strike in order to take advantage of being the first to move, using the element of surprise. If the number of casualties inflicted on country 2 in retaliation for its initiating the war were sufficiently large then country 1 would not fear an attack by country 2 and thus country 2 would avoid a preemptive strike. In Figure 2 the shaded cone of mutual deterrence is included in the area of the larger cone with vertex at E', for which both sides avoid preemption. At points to the right of the line blbl country 1 attacks, while at points to the left of the line b2 b2 country 2 attacks. Thus, in the cone with the vertex at E' both sides avoid preemption. This figure also shows the regions of initiation as the lower shaded sawtooth-shaped region in which one side, both, or neither avoids preemption nor deters the other side. For example, in region I1 country 1 can attack since it has enough missiles to attack country 2, but neither has enough missiles to deter the other. Thus country 1 will be forced to attack or country 2 will be forced to preempt, in either case leading to war. Region I2 is the obverse case in which either country 2 attacks or country 1 preempts. Region I12 near the disarmed state, is the most dangerous region of all since each side

Ch. 6: Arms Races and Proliferation

125

can successfully attack the other, neither can avoid preemption, and neither can deter the other. Each is forced to initiate the war in order to take advantage of striking first. Bilateral disarmament, interpreted geometrically as a movement to the origin in Figure 2, inevitably entails movement through the region of initiation. While the movement to the disarmed state need not necessarily lead to initiation, the movement from the disarmed state may be highly explosive. In such a situation, either country could acquire a small inventory of missiles which could enable it to attack the other with impunity. In fact, the danger of war may be greatest if one side detected that the other was starting to rearm. In such a situation the missile levels move through the regions of initiation in a crisis atmosphere. Figure 2 can also be used to study the effects of the arms race on the outbreak of war7 . Arms races are, in terms of Figure 2, movements in the weapons plane, representing changes in the weapons levels of both sides. An arms race in the usual sense of increasing levels of weapons on both sides would be represented by a movement up and to the right, while a disarming race would be represented by the reverse type of movement, that is, down and to the left. Other types of races can also be represented, for example, one in which one country is increasing while the other is decreasing its levels of weapons. The shaded areas indicate the likely effect of an arms race on the outbreak of war. For example, a movement from the disarmed state at the origin up into the region of initiation, such as to points in I12, would represent an arms race that is likely to lead to war. It results in an explosive situation in which neither country can deter and both can preempt. It is precisely this potential arms race that shows the danger of a completely disarmed situation, where rearming in a crisis atmosphere without the stabilizing influence of mutual deterrence could lead to the outbreak of war. Not all arms races, however, lead to the outbreak of war. Consider, for example, an arms race that starts in one of the regions of initiation, say in II. If both countries increase their levels of weapons via an arms race from a point in II to a point in the cone of mutual deterrence then the arms race has the effect of reducing the chance of war by the stabilizing influence of mutual deterrence. In fact, this type of increase may be one interpretation of the United States-Soviet Union nuclear arms race. The late 1950s and early 1960s were characterized by the instability of relatively low numbers of missiles, which were not sufficient to deter either side. There was, furthermore, asymmetry in the situation in that the United States, country I in the diagram, could attack, while the Soviet Union, country 2 in the diagram, could not. Thus, the situation was in region II, one of initiation. The fundamental instability in the situation was shown by repeated crises, culminating in the 1962 Cuban missile crisis, perhaps the time the world has come closest to nuclear war. In the mid- and late 1960s, however, there was a considerable buildup of Soviet weapons, substantially increasing w2, and

7 See Intriligator and Brito (1984-1986). See also Huntington (1958), Gray (1971, 1976), Lambelet (1975, 1985), Smith (1980), and Mayer (1986).

126

D.L. Brito and M.D. Intriligator

some buildup in US weapons, increasing wl. The result was that by the 1970s there was a situation of mutual deterrence, which represented a more stable situation. Thus, the arms race resulted in movement from regions of initiation to those in the cone of mutual deterrence. Such an arms race had the effect of reducing the chance of war. Once mutual deterrence was achieved there were no crises comparable to the Cuban missile crisis or to the other crises of the late 1950s and early 1960s. Both sides had become, by necessity, more restrained and less willing to "go to the brink" due to their mutual deterrence relationship.

5. Arms races and stability The existence of an equilibrium is important, but its implications depend crucially on whether the equilibrium is stable. In this section we will investigate sufficient conditions for an equilibrium to be stable. It is necessary to specify how each of the two countries will react when the weapon stocks are not in equilibrium. The simplest assumption we can make is that of myopic behavior, where each country behaves in a myopic manner, reacting only to the current level of the weapons stock of the opponent and disregarding information about the history of the levels of weapon stocks and the opponent's current rate of investment in weapons. This assumption was used by Richardson and others, and it is similar to the assumption used in the CournotNash solution of the duopoly problem. Under this assumption we have the following proposition: Proposition 2. An equilibrium point of an arms race is stable if each country behaves in a myopically optimal manner. Proposition 2 can be established by examining the phase diagram in Figure 38 This result, that the equilibrium of an arms race is stable if the countries involved act in a myopic manner, also depends on the assumption made in the model that the utility function is of the form U(ci, D(wi,wj)), where ci is the consumption of the ith country and D(wi, wj) is the public good, defense. Brito and Intriligator (1987) treat a more general problem using a utility function of the form u(ci, xi, zi) where xi is the level of defense and zi is the investment in weapons as before. If the preferences satisfy the condition that, on an indifference curve, > 0, then for a wide range of strategic behavior the resulting equilibrium is unstable. A necessary and sufficient condition for stability is that the elasticity of zi with respect to xi be less than 1. Thus, if expenditure on arms enters the utility function, then the problem becomes more complicated. The assumption that both countries behave in a myopic manner makes it very easy to analyze the stability of the equilibrium. This assumption, however, is very unrealistic.

8

The phase diagram is derived in the Appendix.

127

Ch. 6: Arms Races and Proliferation

'l =0

r 0

W1

Figure 3.

Clearly, military expenditures are based not only on the current arms level of the potential enemy, but also on its projected levels. A more realistic assumption is to assume that each country uses information about current arms levels and the rate of change in arms levels to predict future arms levels. This projection, which is revised periodically or continuously, is then used to plan current investment. Thus, the most realistic assumption to make about an arms race is that the participants monitor the behavior of the other parties and thereby learn about their characteristics. Such a formulation was attempted by Brito (1972) in an ad hoc fashion. Recent results on rational learning in games by Kalai and Lehrer (1993a,b, 1995) now enable us to address this problem in a very general fashion. The principal change in the model is that the choice set of the players must be discrete. In the context of the arms race, this assumption is, in fact, more realistic than the previous assumption that the choice set is the bounded interval which defines the budget set, given that warships, bombers, and missiles all come in discrete units. Letting p = 1/(1 + r) be the discount factor and recalling that, without loss of generality, we can choose the planning time as t=O, the objective function for country i in the discrete case can be written as the sum of discounted expected utility E (E

PtUi(citxit)

,

(32)

where cit is the level of consumption and xit is the level of defense in country i at time t, and the expectation is taken with respect to the probability distribution induced by the ith country's predictions of the other country's behavior. The objective function given by Equation (32) is maximized subject to the equation that gives the dynamics of weapons accumulation Wit+ = it +

- 6i) wit,

(33)

128

D.L. Brito and M.D. Intriligator

a discrete version of Equation (13), where zit is gross expenditures on arms zit in Zi, with Zi a finite set, and 6i is the rate of depreciation or cost of maintaining weapons. The budget set as in Equation (12), is given by Yit =

(34)

it + zit,

the level of defense associated with weapons accumulation is given as xit = Di (wit, Wjt) ,

(35)

and the ith country's forecast of the jth country's strategies for t = 1 .. co is given as (36)

jtQ+l = Qi (lit, wit, Wjt) .

where wjt is a random variable with a probability distribution fjt for t = 1, 2,.. , 0o, and it is the information available to the ith country at time t, which is revised by Bayesian updating. The optimal return function can be written Vi(wio, wjo, Ito)= max {E(

PtUi(Cit Xit))

}.

(37)

The solution to the problem is characterized by the choice of weapons stocks by country i at time 1, given as wi, and a sequence of contingent plans and forecasts Wit for t = 2, 3,..., oo and tj,for t = 1,2,... oo. The function Vi(wio,wjo,Iio) in Equation (37) is the value function of a dynamic programming problem. From Bellman's Principle of Optimality, Vi(wi0o, wjo, iio) = max {Ui (cio, xio) +pi Vi(wil, Wjl, Ii)},

(38)

which is the Bellman equation for this dynamic programming problem. The KalaiLehrer theorem states that if the players use Bayesian updating of their beliefs and certain technical restrictions are satisfied then the game will converge to an equilibrium. 6. A general model of an arms race Most of the early models of the arms race were heavily influenced by the Richardson model and, as a result, were formulated in continuous time. The goal of most of these models was often a derivation of the Richardson equations from more fundamental principles. Inasmuch as one of the objectives of these studies was an analysis of the stability of the arms race, continuous time models were convenient. The use of

Ch. 6: Arms Races and Proliferation

129

continuous time models was unfortunate, however, because continuous time makes it very difficult to model other essential features of an arms race, such as learning, strategic behavior, uncertainty, budgetary cycles, and lags in a continuous time model. All of these are important features of an arms race which need to be studied. A general model of an arms race which would encompass the many previous models can be formulated in discrete time. This general model has four components: a technology that describes the economic constraints faced by the country, a defense technology, an information or learning technology, and a choice function that characterizes the choices made by the country, given the technology and information available to them. Again assuming two countries, 1 and 2, at discrete points in time each country faces the economic constraints defined by Yit = F (cit, zit),

(39)

Wit+l = Zit + (1 - 6i) wit,

(40)

as in (12) and (13), where Yit are the resources available, cit is consumption, zit is expenditure on weapon stocks and wit are the weapon stocks for country i at time t. There is a process that produces a public good, which we will refer to as "defense" 9 , given by the function xit = Di (it, wjt) .

(41)

Equations (39), (40), and (41) define the feasible set for the ith country at time t, which we will denote as Bit. We will also assume that there is a stock of information lit which the ith country uses to predict the jth country's investment in weapons at time t + 1. This process is given by

wjt+l

= Qi (it,Wit, Wijt) ,

(42)

where v'jt+l is the prediction, which uses information as revised by the process it+ = i (it Wit, Wjt) ,

(43)

The assumption is that the ith country updates its information at time t + 1 based on what happened at time t. An example of such a process is Bayesian updating. The last element of the model is a choice function Ait that maps feasible sets and predictions into allocations. An example is utility maximization by a rational agent.

9 In the general context xi could be a vector variable that could be a combination of public goods and bads such as expected casualties, nuclear fallout and other such variables.

130

D.L. Brito and M.D. Intriligator

Formally, Ait : Bit x Qit - [cit,zit] -'+ wit+l,

(44)

where the vector variables are defined as Ct

2 ( c2t )

Wt

2t

Zt

(Wt )Z,

and the dynamics of an arms race are described by the mapping At: Bt x Qt -

[ct,zt] --+wt -

At+l: Bt+ x Qt+l.

(45)

The basic question of existence and stability can then be posed as whether there exists some vector i such that wt goes to iv as t -- oo for any given initial conditions.

This model is very general. However, if we make assumptions about the continuity of At and Qt and the compactness of Bt, then it is possible to prove: Proposition 3. IfAt and Qt are continuousfunctions of wt, the set Bt is compact, and there exists a compact set Q2 such that wt is in 2 for all t, then there exists at least one equilibriumfor the arms race o. This formulation is too general for specific analysis, but it does identify some of the essential elements of the arms race and it does suggest which specific assumptions are critical. The assumptions that the choice function At is continuous and that the feasible set Bt is compact are very standard economic assumptions. A fundamental problem, however, lies in the assumption that the learning technology Qt is continuous. It is in this process that strategic behavior is embedded. There is no general consensus in economics about what assumption can be made in this regard, although the recent results in game theory such as those of Kalai and Lehrer discussed above suggest that new developments in theory may lead to significant progress in addressing this critical element in the theory of arms races. 7. Power, bargaining, and war If arms races existed in a vacuum, without the possibility of war, they would at best be a form of conspicuous consumption that produced externalities in the form of research

'° The proof of this proposition rests on the fact that we have constructed a mapping from the compact set W to itself. The Berge Maximum Theorem implies that this mapping is upper semicontinuous, so it follows from the Kakutani fixed point theorem that an equilibrium exists.

Ch. 6: Arms Races and Proliferation

131

and other technological spillovers and at worst be a form of economic waste. They would be only of limited interest. Unfortunately, arms races can lead to wars and regime instability. The question then arises as to whether a theory that is premised on rational behavior can address a question such as war. Bueno de Mesquita's (1981) analysis of the determinants of war is persuasive in arguing that wars are consistent with rational decision making. Of the 58 wars that have been fought since the Congress of Vienna in 1815, the initiator won 42 of them. Bueno de Mesquita argued that if wars were nonrational events then there should not be any systematic relationship between the initiator and the victor, with such an event occurring only once every 2500 times. He then computed cost/benefit ratios and found that when wars occur it was usually in the interest of the initiating nation. He did not, however, explain why wars occur. Wars are usually Pareto inferior outcomes of a conflict, in that both parties would be better off if the expected loser compensated the expected victor by means of a transfer of resources without actually going to war. Thus, while the Bueno de Mesquita results demonstrate that wars are usually consistent with rational behavior on the part of the aggressors, his model does not explain why wars actually occur since it does not explain why the loser fought. Wallace (1979, 1981) studied the outbreak of war using data on 99 serious great power disputes occurring since 1815. He was particularly concerned with whether arms races lead to the outbreak of war, and he concluded that conflicts and disputes accompanied by arms races are much more likely to result in war. Of the 26 cases that led to war, 23 involved an arms race, while of the 73 cases in which there was no war, 68 did not involve an arms race. Conversely, of the 28 cases in which there was an arms race, 23 resulted in war, while of the 71 cases in which there was no arms race, 68 resulted in no war. In addition, Wallace found that the probability of escalating to war depends on whether the "revisionist" power during the period prior to the crisis wins the arms race. Of the 30 cases in which the revisionist power won, 12 resulted in war, while of the 65 cases in which the revisionist power did not win, 12 resulted in war. These results suggest that there is an asymmetric relationship between the winner of an arms race and the probability that a crisis will escalate to war. Siverson and Tennefoss (1983) studied the outbreak of war from 1816 to 1965. They divided conflict situations into three levels: threats, unreciprocated military action, and reciprocated military actions. They divided nations into four categories: allied major powers, unallied major powers, allied minor powers, and unallied minor powers. They found that major powers and allied minor powers seem to be involved in less hostility than unallied minor powers, while conflict between major powers is limited. While the major powers are stronger and hence less likely to be attacked, this explanation is not consistent with the observations that unallied minor powers initiated over 19 percent of the conflicts with major powers and that there is no case in which an allied minor power initiated a conflict. Smith (1980) attempted to identify arms races that are particularly war prone using statistical methods. She estimated the Richardson model and inferred stability from the estimated coefficients. She demonstrated that arms races do not always lead to war

132

D.L. Brito and M.D. Intriligator

and showed that the probability of war could be predicted by the coefficients of the Richardson model. She did not explain what factors led to this result, but she noted the importance of risk factors and uncertainties that are not treated in the Richardson model and in subsequent deterministic models. These factors are, however, treated in the model presented below. From this research, a pattern of the following empirical regularities appears to emerge: first, wars are usually consistent with rational decision making on the part of the initiator; second, an arms race is less likely to lead to war if the status quo power "loses" the arms race; third, conflict between major powers is limited and is less likely to escalate into war than conflict between major and minor powers; fourth, major powers and allied minor powers seem to be involved in less hostility than unallied minor powers; and, fifth, unallied minor powers initiated conflicts with major powers, but there is no case where an allied minor power initiated conflict. None of these results is yet definitive; however, they do form a reasonable starting point for a set of stylized facts that should be consistent with or explained by formal models of arms races and the outbreak of war. We now develop such a model. Consider a two-period model in which there are two countries in conflict for the rights to a flow of a single homogeneous good. For example, they may be in contention for a territory that yields a fungible resource, such as oil, or the right to a market that yields income. In the first period this good may be consumed or used to produce weapons, as in the usual guns or butter trade-off. This process is irreversible, so weapons cannot be converted back into goods during the second period. In the second period the distribution of the good can be altered either by war or negotiation. The outcome of war depends on the amount of fighting engaged in by either country, which, in turn, is limited by its weapon stocks, as determined by its choices during the first period. War produces an externality which is a public bad and which offsets the utility of consumption in the second period. We assume that the countries are rational and nonaltruistic and that they will choose to fight if the expected value of such a strategy is greater than that of not fighting. If they do not fight then they can negotiate a redistribution of resources in the second period. Each of the countries is informed as to all aspects of the problem other than possibly the way in which the war externality enters into the objective function of the other country. A basic assumption is that of rational agents maximizing utility. This assumption and the use of a two-period model probably bias the model in the direction of no outbreak of war. The assumption of rationality leads agents to adopt a cooperative solution. It is, in fact, very difficult to develop models in which rational agents carry out threats ex-post . The assumption of two time periods precludes early belligerent behavior to establish a reputation for toughness. The special assumptions of the model

" The models lack the property known as "subgame perfection". Games are subgame perfect if the optimal strategy in period 1 for period 2 play is in fact the optimal strategy in period 2 for period 2 play. See Fudenberg and Tirole (1991).

133

Ch. 6: Arms Races and Proliferation

that will be made, specifically that there exists a mechanism for reallocating resources and that war produces a negative externality that affects the countries in different ways, appear to be reasonable ones that are consistent with the reality of conflict between nations. The technical assumptions of differentiability, continuity, and compactness made earlier are standard, and most formal models in international relations make even stronger assumptions in which these properties are implicit. We assume for the two-period model that each country has a utility function of the form

ni

ni)·

(46)

where Cil is first-period consumption for the ith country, Ci2 is second-period consumption; x is a public bad common to both countries that is produced by war and that reduces utility, such as nuclear fallout or the loss of trade due to wartime disruption of commerce; and n i is a parameter that determines the disutility of war to the ith country. The utility function, u(-, ) is a strictly concave one such that au(cit,X)/Cit > 0 and Ou(O, O)/acit is finite 1213. Since there is no conflict in the first period and preferences are assumed to be the same in both periods, utility in the first period is given by u(cil,O). For simplicity we assume that the decision makers in each country believe that the parameter, ni, characterizing the other country's disutility of war takes one of two values: a with probabilityp or fi with probability I -p 14. Without loss of generality a < fi, so the disutility of war is greater for an a country than it is for an fi country as measured by the amount of utility of consumption needed to offset a particular level of the externality. In the first period each country faces a resource constraint of the usual form Yil = Cil + Wi,

(47)

as in Equations (12) or (34), where Yi, represents the total goods available to the ith country in the first period, and wi is the ith country's investment in weapons. We assume that total resources Y = Yit + Yjt are fixed; thus, Yi1 is sufficient to describe the initial allocation of wealth. This assumption is made so that the game is such that an agent has the option of not participating. The utility level of u(O, 0) can be thought of as the utility of staying home as opposed to engaging in adventures abroad in order to acquire a right to a share of total goods available Y, which, in turn, could be thought of as representing gains from international trade and diplomacy. 13 The assumption that x is public bad common to both countries is made to facilitate the graphical representation of the model. The formal mathematics do not change if we assume that xj is a public bad that is specific to the ith country. 12

14

This particular basis for uncertainty is not crucial to the properties of the model. For example, we

could have modeled the problem as one of uncertainty in the technology of conflict, where the jth agent is uncertain as to the ith agent's effectiveness. What is crucial is that there is uncertainty for the jth agent about the cost and/or outcome of conflict. Also we could have formulated the uncertainty in other ways, for example, a continuous distribution over the parameter n, reflecting the beliefs of agent j.

134

D.L. Brito and M.D. Intriligator

By the second period, both countries have chosen weapon stocks w i and wj, respectively, and there are then three possibilities. First, both countries could choose to behave passively and preserve the status quo. Second, one country could propose a redistribution agreed to by the second country. Third, one or both countries could propose a redistribution not agreed to by the other side that could lead to conflict and possibly to the outbreak of war. The first stage of the game involves the choice of the weapon stocks wi and wj. These weapon stocks in turn establish the parameters of the second stage of the game. Define as the war allocationthe distribution of resources and the externality that results from the maximization of second period utility by the two countries subject to the constraints x = D(vi, Vj),

(48)

Yi2 = Yil + 'Ifivi, vj),

(49)

Vi

Wi,

(50)

vj

w< W,

(51)

where vi is the amount of fighting by the ith country, as measured by the amount of weapons each of the countries commits to the war. We assume that committing more weapons shifts the redistribution, as measured in Equation (49) by Wi(vi, vj), in that country's favor, so that Olfi(vi, vj)/Ovi > 0 and 'fi(vi, j)/0vj < O0.The redistribution function i(vi, Vj) plays a role here similar to that of the threat point in the Nash bargaining problem. The amount of fighting the ith country is able to do is limited by its level of weapons, which was determined, in part, by the initial distribution of resources in the first period and the redistribution is from one to the other so 'i(vi, vj) = -Iyj(vj, vi). The war allocation defines the threat point of the second period bargaining game. The properties of this game can be illustrated in Figure 4. In this figure, the vertical axis represents second-period consumption for both countries and the horizontal axis is the level of the externality caused by war. The private consumption of country 1 is measured upwards from point 0, and the private consumption of country 2 is measured downwards from point YO. Country I has indifference curves that slope upward since x is the public bad; similarly, country 2 has indifference curves that slope downwards. If the initial allocation is at point ao, then, in the second period there are two possibilities: the war allocation could either be at a point like al or it could be at a point like a2. If the war allocation is al, both countries prefer the allocation at ao to the allocation at al. If, however, the war allocation is at a point like a2, then country 1 would prefer the allocation at a2 and thus has an incentive to initiate a war. However, both countries would prefer an allocation like a3 that is feasible and dominates a2. Rather than fight, country 2 should rationally negotiate and reach a settlement in the interval [a3, a4], representing the set of points that Pareto dominate a2 for both countries and do not

135

Ch. 6: Arms Races and Proliferation 1',Y

'22

C12

Figure 4.

involve conflict 15. If such voluntary transfers are possible and if each party knows the other's preferences, then war is not consistent with rationality since the set of points that Pareto dominate a2 is well defined. We assume that a threat cannot be rationally implemented by a party if it prefers the initial allocation to the threat allocation, and define an allocation to be stable if the initial allocation Pareto dominates the threat allocation. Classical bargaining theory assumes individual rationality, meaning that a rational individual would not agree to an allocation that is dominated by the threat point, and also assumes joint rationality, meaning that two rational individuals would not agree to an allocation if there exists an alternative feasible outcome that will make them both better off. Since any outcome in the interior of the feasible set is Pareto dominated, the final outcome must lie on the boundary of the feasible set and must consist of points that dominate the threat point. Defining N(a 3) = {Y 2, Yj2 Yi2 > a4 , Yo > Yj2 - as}, this is the set of points that Pareto dominate a3 and represents the negotiation set. There are several solution concepts that assign an allocation in the negotiation set as a function of the threat point 16. For the purpose of this section, however, it is only necessary to assume that the outcome of a negotiation will result in an allocation in the negotiation set. The cooperative allocation in the second period is described as follows: if the conflict allocation is not Pareto dominated by the initial allocation then the second period

i5 The standard Nash solution of the bargaining game leaves open the question as to why the threat

point is ever implemented given that it is Pareto dominated by the negotiation set. See Nalebuff (1982, 1984). 16

See Nash (1950), Kalai and Smorodinsky (1975), and Brito, Buoncristiani and Intriligator (1977).

136

D.L. Brito and M.D. Intriligator

allocation is given by the outcome of a cooperative game that allocates the surplus. However, if the conflict allocation is Pareto dominated by the initial allocation then the initial allocation is the outcome. With this formulation of the two-period model it is possible to show that there always exists a set of initial allocations of wealth that are stable and that do not lead to war since those initial allocations Pareto dominate any possible war allocation, as in Proposition 4. There exists a set of positive measure, So, the stable set, in the interval [0, Y] such that allocations in So will Pareto dominate any allocation that results from war 7 In the first period countries can choose two possible strategies for investing in weapons. First, they can adopt a war arms investment strategy in which they invest in weapons as if they planned to engage in fighting in the second period. Second, they can adopt a deterrence arms investment strategy in which they invest in a sufficient level of weapons so as to deter the other country from initiating a war in the second period. In the example illustrated by Figure 4, country 1 can choose to invest in weapons in the first period so that the resulting allocation is a5, making it better off since it would then have more first-period consumption and the allocation in the first period would still be ao. If the countries adopt a war arms investment strategy, the ith country chooses wi in the first period and, in the second period, maximizes, by choice of its level of fighting vi, its second period utility, given as: U Ci2 ,

)

(52)

subject to Equations (48)-(51), where it is assumed that Vj is chosen rationally by the jth country in response to the ith country's actions. An alternative to the war arms investment strategy is a deterrence arms investment strategy. Define T as a voluntary transfer from the ith country to the jth country in the second period. Another way to formulate the problem is for the ith country to maximize total utility of consumption over both periods: U (Cil, O) + U (Ci 2 , 0)

(53)

subject to the incentive constraintthat thejth country be no worse off after the transfer than the utility level it could have achieved by war, namely

u(Yj2 +Ti,

17

)

Proof in Appendix.

u(Yj2 +Ij(wi,Wj), D(wi, nj

)

(54)

Ch. 6: Arms Races and Proliferation

137

Here T is the transfer in the second period, given wj as well as the resource constraints for the ith country in both periods: Yi - cil - Wl = 0,

(55)

Yi2 - Ti - ci 2 = 0.

(56)

The incentive constraint in Equation (54) requires that the ith country define the feasible set of the jth country such that the jth country weakly prefers the nonwar allocation with the transfer to the war allocation 18. The resource constraint in Equation (55) states that the ith country's total resources in the first period are divided between weapons and consumption while that in Equation (56) states that total resources in the second period are divided between consumption and transfers to the jth country. The solution to this problem represents the maximum consumption that the ith country can achieve while still guaranteeing that it would not be optimal for the jth country to engage in war given that it has wj weapons. Proposition 5. If the transfer Tj > 0 then an equilibrium level of weapons exists and the allocation that results from the ith country adopting a deterrence arms investment strategy and the jth country adopting a war arms investment strategy is a Nash equilibrium. Note that if the ith country adopts a war investment strategy then the solution for the ith country strategy depends on its own parameter, ni, while if it is adopting a deterrence investment strategy, the solution depends on the other country's parameter, nj. This point is important in understanding why deterrence may lead to higher arms levels. If a country believes that the other country is tough, it may feel that it is necessary to invest in high weapon levels in order to deter. If, on the other hand, the ith country is adopting a war investment strategy, then it will choose a level of weapons, and thus a level of fighting, at which the marginal benefits from fighting are equated to the marginal costs of fighting. This apparent anomaly is due to the fact that in the deterrence arms investment strategy the level of arms acquired by the ith country is independent of n i but does depend on ni. Thus, if nj is much greater than ni then the deterrence arms investment strategy may have an equilibrium with a higher level of arms than the war arms investment strategy. Proposition 6. If the transfer T = 0 and if both countries adopt a deterrence arms investment strategy then the resulting arms investment strategy is not a pure Nash equilibrium. 1 The problem of determining how the surplus created by not fighting is distributed in a bargaining situation is not addressed. For the sake of simplicity, we are assigning the entire surplus to the party making the transfer. Clearly any rule from axiomatic bargaining theory could also be used. See the references in footnote 16.

138

D.L. Brito and M.D. Intriligator

It

cj 2

Ci2

Figure 5.

In Figure 5, point al is the allocation associated with the war arms investment strategy for both countries. This is Pareto dominated by ao so there will not be a war if both sides adopt investment strategies that result in al. Each country, however, can unilaterally adopt strategies that result in points like a2 and a3. These are the allocations that would result if one country adopts the war arms investment strategy and the other adopts a deterrence arms investment strategy. Again, these are Pareto dominated by ao so there will not be a war, but they are also Pareto dominated by a 4, the allocation that results if both countries adopt the deterrence arms investment strategy, each assuming that the other is using a war arms investment strategy. This point, however, is unstable, since it is Pareto dominated by as and a 6, which are the allocations that result if either country adopts a deterrence arms investment strategy, each assuming that the other is using a deterrence arms investment strategy. This process results in a sequence of weapon stocks that converges to the disarmed state, but the allocation associated with this state is not stable since either country would find it advantageous to switch to a war arms investment strategy. The resulting allocation will be ao which is unstable. A mixed strategy, however, will lead to a Nash equilibrium 19 Consider now the asymmetric information case in which only one country knows the utility function of the other. In this case, both informed and uninformed countries maximize expected utility. The informed country has the option of bluffing so as to increase the transfer made to it by the uninformed country. The uninformed country may call the bluff by offering the informed country a choice that involves a nonzero probability of war, a separating equilibrium. Alternatively, it can simply assume that the other country is not bluffing and not run the risk of war, a pooling equilibrium. Whatever equilibrium is chosen depends on the relative wealth of the two countries and the underlying distribution of types.

This follows from the standard result that all finite two-person games have an equilibrium in the space of mixed strategies [Nash (1950)]. 19

139

Ch. 6: Arms Races and Proliferation

T.

_ I

Figure 6.

Without loss of generality, we assume that thejth country is informed and knows the value of the war aversion parameter, ni, of the ith country. However, nj is not known to the ith country, which knows only that nj is a with probability p and /3 with probability 1 -p, where a [qVj(wi, wji, a) + (1 - q)u(Yj 2 + Ti, 0)].

(61)

If the jth country is of type fi it will choose the lottery, while if it is of type a it will optimally choose wji. This is possible because Vj(wi, W%~, 0) > Vj(w 1, wjf, a).

(62)

The expected outcome for the ith country is then given by the compound lottery:

Ai=

((l-p)[qu (Y + (w, w), D i (wi, wj) + (l I L ~ ni(63) / +p u(Yi2 - Ta,

q) u(Y

-

T,

0)

)}.

The ith country wants to maximize expected utility, that is, the utility of first-period consumption plus that of the lottery given by Equation (63). Let Cp(Yi, ni, a, ,p) be the value of the solution of the problem c (Yjl,ni, a, ,p) =

max

[u(cil,O)+ Ai],

q,rkr7,wi

(64)

subject to the incentive constraints given by Equations (60) and (61) and the resource constraints given by Equations (55) and (56). This problem can be solved recursively as a dynamic program, where the second-stage problem is given by

V = max A i + Xa (U(Y2 + Tia, 0)- (qVj(wi, wjfi, a) + (1 - q) u(Yj 2 + Ti, 0))) + ,f {u(Yj2 + Tp, 0) -

Vj(Wi, Wig,

)} .

(65) The first-order conditions for the problem are standard, and the process is illustrated in Figure 7. The point ao is the initial allocation, the point al is the conflict allocation, the point a 2 is the allocation that results from the transfer T, and the point a3 is the allocation that results from the transfer Ta,. The ith country chooses T and wi such that if the jth country is of type fi it is indifferent between the conflict allocation. Note that thejth country is indifferent between a and a2. Thus the ith country picks

141

Ch. 6: Arms Races and Proliferation

C2

Cl

r

Figure 7.

Tia and q such that if the jth country is of type a it is indifferent between a4 with probability q and a2 with probability (1 - q). Then- it is optimal for the jth country to reveal that it is of type a. An alternative equilibrium that has been treated in the literature on sequential games is a pooling equilibrium, where the ith country will choose a strategy that does not attempt to discriminate between the two types. This will happen if the optimal solution results in q = 0. We can characterize the configuration of possible equilibria. The important first-order condition is with respect to q. First, we can use the first-order condition with respect to Tia to solve for Xa in Equation (65), yielding: a (Y 2 - Ta,0)/Tia u (j2 - Tia, O) /Tia'

(66)

while the first-order condition with respect to q is

aq O)i-= (1 -p) [Mu ( Y i2 - Aa [11(Yj2 +

+ t t i (w i' w j / O'

j(Wi,

(

:)> D(wij))

ni

:

/

-U(i2-

'

ip, )](67)

(7

-U(yj2 + Tii, ]

The first term in Equation (67) is negative since the ith country prefers to make a transfer rather than have a war, while the second term in Equation (67) is positive since the jth country prefers a transfer to having a war if it is of type a, so the term in parenthesis is negative and the multiplier is positive ifp > 0. Ifp = 0 then the second term will be negative. In that case the ith country would choose q =0, and the equilibrium will be pooling for all Yil along the locus of points such that p = 0. Ifp = 1

142

D.L. Brito and M.D. Intriligator probability of war I I I I

e *

P'

~~~~P 1

Figure 8.

then the first-order condition becomes positive as the term that is multiplied by (1 -p) drops out and the optimal solution is q = 1, yielding a "separating equilibrium". The probability of a war for a given distribution of wealth is illustrated in Figure 8 where the probability increases linearly until it reaches some critical level p* and then drops discontinuously to zero. While this result is, to some extent, an artifact of the linearity of the model, wars occur in the model as a result of rational choice and they are associated with uncertainty, as suggested by the earlier cited empirical work on war initiation. Further results are available if we are willing to assume that the ith country is willing to make a transfer to an a country but not to a / country. This could be considered as the case in which a status quo power is challenged. It might be willing to make concessions to a strong challenger but not to a weak challenger. For most fixed levels of p

the terms in brackets are all positive, so I

pc(n) = 1

i=l

i=l

pi

(-pi(n)) exp (i=l

pi(n). i=

/

Since pi(n) > 0, a sufficient condition for Iin=, p~(n) to be an decreasing function of n is that the series _i=l pi(n) diverge as n - oo. The condition pi(n + l)-pi(n) I is not reflected in a correspondingly large achieved income ratio I1/I2. I shall be examining strong versus weak forms of the paradox: - POP (strong form): In mixed conflict-cooperation interactions, the contending parties end up with exactly identical incomes (11/12 = 1) regardless of the initial resource ratio R /R 1 2. - POP (weak form): In mixed conflict-cooperation interactions, the final distribution of income will have lesser dispersion than the initial distribution of resources. Thus, assuming contender 1 is initially the better-endowed side: 1 < I1/12 y

or I = 0,

and H =

-L-S-

I,

where the analysis assumes, for simplicity, that the nonnegativity constraint on H is not binding. [For this property to obtain in equilibrium, y must be sufficiently large.] Equation (2) says that time allocated to market production is negatively related to t. Also, the ruler and the potential insurgent leader face perfectly elastic supplies of soldiers and insurgents at expected rates of compensation equal to y. In this model consumable product per family, which equals La + yH, provides an index of current aggregate economic welfare. Time allocated to either soldering or insurgency can have private value, but it has no social value. 3.2. An insurrection? Next, consider the decisions of the potential insurgent leader. In this section, an insurrection would attempt to appropriate only the current income of the ruler's

198

H.I. Grossman

clientele. Accordingly, the objective of the potential insurgent leader is to maximize the expected net income of his clientele, denoted by G, where G = (1 -p)tLa - wiI.

(3)

In Equation (3), tLa is the gross income of the ruler's clientele and 1-p represents the fraction of tL a that the insurgent leader expects to capture. The realized value of this fraction can be either continuously or discretely distributed. For example, this setup allows as one possibility that the insurgent leader captures all of tL a with probability 1-p and captures none of tL a with probabilityp. This model also assumes that the only expected cost associated with organizing an insurrection is the expected compensation of insurgents. The expectations I-p and p depend on the relative numbers of insurgents and soldiers. Specifically, the technology of insurrection is such that p=

1

X and

I x=0-

0 >0.

(4)

In Equation (4), x measures the strength of an insurrection relative to the ruler's internal security measures. Equation (4) implies that 0 < p < 1 and that p is decreasing in x. The value of x in turn depends on the ratio of the number of insurgents to the number of soldiers and on the value of 0, which measures the effectiveness of insurgents relative to soldiers. The determination of 0 is discussed below. As stressed above, the theory developed in this essay focuses on the allocation of resources among productive and appropriate activities and, more specifically, on the factors that determine the allocation of resources to insurrection and to its deterrence and suppression. For these issues, the modelling of the effect of resource allocation decisions on the expected outcome of an insurrection, as exemplified by Equation (4), is critical. But, it is not necessary to be explicit about any noneconomic factors that help to determine the expected outcome of an insurrection. Also, assuming that agents maximize expected income or wealth, it is not necessary to model the random and idiosyncratic events that can cause the actual outcome of an insurrection to differ from the expected outcome. A complementary literature - see, for example, Kuran (1989) analyses the role of social psychology and random events in influencing the outcomes of specific insurrections. Another complementary literature - see, for example, Brito and Intriligator (1990) - models the dynamics of insurrectionary conflict and considers the role of other resources, such as arms and territorial control, as well as soldiers and insurgents in the technology of insurrections. The potential insurgent leader's problem is to select a nonnegative value for I to maximize G. In solving this problem, the potential insurgent leader takes as given tL a and S, which are determined by the ruler and the peasant or worker families, the expected compensation rate required by insurgents, which implies wi = , the technology of insurrection, given by Equation (4), and the value of 0. [Note that this

Ch. 8:

199

Insurrections

model precludes the possibility that the insurgent leader might induce some of the ruler's soldiers to desert.] The solution to this problem is either that G has an interior maximum at a positive value of I that satisfies dG =-OP tLa

=O

with

1>

(5.1)

and

I=0.

(5.2)

or that G is maximized with dG =

P tL a -y _

O

Conditions (5.1) and (5.2) say that, if the potential insurgent leader chooses a positive value for I - that is, if he decides to organize an insurrection - then I is such that the marginal contribution of insurgents to the expected net income of the insurgent leader's clientele equals the expected compensation rate required by insurgents, y. Alternatively, if the potential insurgent leader decides not to organize an insurrection, then at I equal to zero the marginal contribution of insurgents to the expected net income of the potential insurgent leader's clientele would be less than or equal to y. In addition, if I is positive, the compensation schedule for insurgents must be such that whatever the outcome of the insurrection the realized net income of the insurgent leader's clientele is nonnegative. Suppose that the insurgent leader captures all of tL a with probability 1 -p and captures none of tL a with probability p. Then, to satisfy this constraint, as well as the constraint wi = y, the leader would offer the insurgents compensation at rate y/(l -p) to be paid if and only if the insurrection succeeds. Substituting for dpl0I, calculated from Equation (4), conditions (5.1) and (5.2) imply that StL

S

IX__ _={ - ' 0

for S < , for S S,

where S= OtLaly. In Equation (6), S is the minimum number of soldiers - that is, the minimum allocation by the ruler to internal security measures - necessary to deter the potential insurgent leader from organizing an insurrection. If S < , then Condition (5.1) obtains, whereas, if S > S, then Condition (5.2) obtains. Moreover, if S is less than , then I depends positively on tLaly but is a hump-shaped function of SIO. 3.3. The deployment of soldiers Lastly, we analyze the decisions of the ruler. Given the transient goal of a potential insurrection, the objective of the ruler in this section is to maximize the expected net income of his clientele, denoted by R, where R = ptL' - w,S.

(7)

200

IHI.I Grossman

In Equation (7), p represents the fraction of their gross income that the ruler's clientele expects to retain, given the possibility of an insurrection. This model abstracts from the possibility that an insurrection might result in the destruction of some of the income of the ruler's clientele. [See Grossman and Kim (1995) for a more general analysis that allows predatory activity to be destructive.] The ruler's problem is to choose nonnegative values for t and S to maximize R. The ruler takes as given the expected compensation rate required by soldiers, which implies w, = y. In choosing t and S, the ruler takes into account how his policies will affect the decisions of the other agents. Consider the ruler's decision about the number of soldiers to deploy. In choosing S, the incumbent ruler takes into account how this choice will affect the choice of I by the potential insurgent leader. Equation (6) tells us that, if S is less than S, then I is positive, but that, if S is not less than S, then I equals zero. Moreover, with I= 0, p equals unity and Equation (7) implies that R is a decreasing function of S. Assume, for simplicity, that is small enough that, whether or not the ruler actually chooses S equal to , the ruler could choose S equal to S without causing the nonnegativity constraint on H to bind. This assumption requires that the ratio 0/y be not too large. It implies that the ruler could deploy enough soldiers to deter an insurrection without driving the expected compensation required by soldiers above y. [Note that this assumption is stronger than the assumption noted above that the nonnegativity constraint on H is not binding in equilibrium.] Given these assumptions, Equation (7) implies either that R has.an interior maximum at a value of S that satisfies OR

=

(Op OpId I ( + p d)

tLa

-

=0

with 0 < S
0

Conditions (8.1) and (8.2) say that, if the ruler chooses S less than S, then S is such that the marginal contribution of soldiers to the expected net income of the ruler's clientele, where the marginal contribution of soldiers includes both the direct effect of S on p and an indirect effect of S on p via the effect of S on I, equals the expected compensation rate required by soldiers, y. Alternatively, if the ruler chooses S equal to , then for all values of S less than the marginal contribution of soldiers to the expected net income of the ruler's clientele exceeds y. In addition, the compensation schedule for soldiers must be such that whatever the outcome of the insurrection the realized net income of the ruler's clientele is nonnegative. Suppose that the insurgent leader captures all of tL a with probability I -p and captures none of tL a with probability p. Then, to satisfy this constraint, as

Ch. 8:

201

Insurrections

well as the constraint ws = y, the ruler would offer his soldiers compensation at rate yip to be paid if and only if there is not a successful insurrection. Substituting for Op/OS and Opl0I, calculated from Equation (4), and for dUldS, calculated from Equation (6), Conditions (8.1) and (8.2) imply that

S=

tLa < S=a 40y

for 0 >

2(9) for 0 < .

Equation (9) says that, if and only if 0 is not larger than 1/2, then S equals S. In other words, if and only if insurgents would not be too effective against soldiers, then the ruler deploys sufficient soldiers to induce the potential insurgent leader not to organize an insurrection. Equation (9) also says that S is proportionate to tLaly. In addition, for S = , the ratio yS/tL is an increasing function of 0, whereas, for S < , yS/tL a is a decreasing function of 0. Thus, the maximum value of yS/tLa is 1/2, which occurs at 0=1/2. Substituting from Equation (9) into Equation (6), we determine the behavior of the potential insurgent leader taking account of the ruler's internal security measures, 1 ) tL 2 = 2~

for 0 > for

2 ~2

>

1 (10)

for 0 < 2

0

Equation (10) says that, if 0 > 1/2, then I is proportionate to tLaly. The maximum value of the ratio yI/tL a is 1/4, which obtains at 0 = 1. Substituting Equations (9) and (10) into Equation (4) gives the equilibrium value for p, p = min (

1

(11)

Equation (11) says that the fraction of the gross income of the ruler's clientele that the insurgent leader expects to capture, 1-p, depends only on 0, the measure of the effectiveness of insurgents relative to soldiers. If 0 is less than or equal to 1/2, then there is no insurrection and 1-p equals zero. Alternatively, if 0 exceeds 1/2, then the potential insurgent leader organizes an insurrection and 1 -p is positive and is an increasing function of 0. Note that, although an increase in y would increase the cost of recruiting insurgents, it also would increase the cost of recruiting soldiers. Moreover, because both S and I, as given by Equations (9) and (10), are proportionate to y, a change in y would not affect either the ratio IIS or the expected compensation bill for soldiers, yS. Similarly, although an increase in tL would increase the return to recruiting insurgents, it also would increase the return to recruiting soldiers and it also would not affect IIS.

H.I. Grossman

202

Accordingly, because p is a zero degree homogeneous function of I and S, with I and S chosen, respectively, to maximize the expected net incomes of the insurgent leader's clientele and the ruler's clientele, the equilibrium value of p is independent of both y and tLa. Importantly, the only effect of an change in y on the expected net income of the ruler's clientele would be through the negative effect of y on market production, L a , and thereby on ptLa. 3.4. The rulers tax policy To complete the analysis, consider the ruler's choice of t, the fraction of market production taken by the ruler's clientele. In choosing t, the ruler takes into account how this choice will affect the choice of L by the peasant or worker families. The ruler also takes into account the effect of his choice of t on the passive support that an insurrection would receive from noncombatants. The extent of passive support for an insurrection is reflected positively in the value of 0, the measure of the effectiveness of insurgents relative to soldiers. The key assumption is that the amount of passive support for an insurrection is inversely related to the ruler's popularity, which, in turn, is an increasing function of t. To make this story operational in a tractable way, the model assumes that 0 2exp[¢(t -z)] > 0~~~LY\L 1 - exp[(z - t)] < 2

for t 0 and z 0

for t >z for t z.

With the ruler choosing S to maximize R, and given this assumed dependence of 0 on t, how does R depend on t? Substituting Equations (9) and (11) into Equation (7) gives

R=

R=tL

40 (1- 0) tL a

for 0 >

(13) (13)

for 0 < '

Equation (13) says that, for all values of 0, R is positively related to tL a and negatively related to 0. For > 2, the negative relation between R and 0 obtains because the

Ch. 8:

203

Insurrections

negative effect of 0 on p outweighs the negative effect of 0 on S. For 0 < , the negative relation between R and 0 reflects the positive effect of 0 on S. Using Equation (2) to relate La to t, Equation (13) implies

La

a

dR40 dt

1-a )La

(-

t

t d\

lt (-

for 0 > lal-

I-O) a I -t

dO) 1 f- d

t 1-

for
h >-c and m > 0. It is in the terrorists' interest to attack provided that expected payoffs are positive, so that c 0. Additionally, the no-negotiation policy may be insufficient to deter an attack when the costs associated with failure are low or even negative. In fact, a fanatical group may perceive benefits from logistical failure, so that h > -c > 0. Clearly, the conventional wisdom hinges, in part, on a group being solely motivated by concessions. Next, we turn to the government, which expends D in deterrence regardless of the outcome. This expenditure determines, along with nature (chance), the terrorists' perceived probability of logistical failure. From Figure 4, we see that the government only incurs D if no attack occurs. If an attack occurs and fails, then the government incurs a ( 0) in addition to D; if, however, an attack succeeds then the government incurs an additional cost of h for capitulating and n for not capitulating. These additional costs involve crisis management and political fallout. At this juncture reputation costs are ignored. The government influences the terrorists' perceived failure rate through its expenditures on deterrence, in which 0 = K(D)

or

D = K-'(o) _ D(O),

(7)

where D(O) = 0, D' > 0, D" > O0. We further assume that the government does not know the resolve of the terrorists so that c > 0 is a random variable with probability density functionf(c). For the government, the likelihood of an attack is prob {c < c*}

='

=

j

C

f (c) dc.

(8)

By Equation (6), we have that c* and, hence, 2 depends on terrorists' beliefs concerning the likelihood of logistical success and government capitulation - i.e., 2= 2 (O,p). Partial differentiation of Equation (8) gives the following results: increased deterrence reduces the likelihood of an attack, and an increase in p (i.e., the perceived negotiation weakness of the government) encourages terrorist strikes. For simplicity, the two agents are assumed to have identical beliefs of the likelihood of attack.

Ch. 9:

Terrorism: Theory and Applications

227

When the decision to capitulate is made ex post, the government would concede whenever n > h, or capitulation is less costly. Following a hostage incident, ex post cost is min(n, h). From an ex ante perspective, the expected cost to the government is E[TC] = D(O) + 20a + Q2(1 - O)E[min(n, h)],

(9)

where n is a random variable with density function g(n), and E[.] is the expectations operator. To determine its optimal deterrence, the government chooses 0 to minimize Equation (9). We denote O* as the argmin E[TC] and TC* as minimized expected cost. Comparative statics applied to the resulting first-order conditions would show that O* increases as (i) the likelihood of attack goes up, (ii) the expected cost of successful attacks rises, and (iii) the ability to deter attacks increases (i.e., aQ/d0 rises). Desirability of precommitment never to negotiate can be examined by comparing the cost associated with such a policy with TC*. Let f2 denote the ex ante probability = 2(, 0). The government's Q2 of an attack when precommitment is credible, so that expected cost is then TC = D(O) + Q20a + •2(1 - O)E(n),

(10)

where E(n) is the expected cost of a successful attack. If a policy of precommitment eliminates all attacks (i.e., i2 = 0 when evaluated at O*), then precommitment dominates the ex post decision when Equation (9) is compared with (10). When, however, all attacks are not eliminated, precommitment may prove more costly ex post for those incidents where E(n) >E[min(n, h)] or capitulation is less costly. Credibility is not sufficient to make precommitment the optimizing policy, especially when the cost of holding firm cannot be known in advance. In essence, precommitment can lead to a time consistency problem if the game is extended to multiple periods, since a government may come to regret its pledge to hold firm once the actual cost of doing so is revealed. The analysis can be made more realistic by allowing multiple periods and reputation cost. The latter arises when concessions in one period make the terrorists raise their beliefs about future concessions. As p increases, more hostage taking will occur. Loss of reputation for the government raises the cost associated with capitulating and should curb, but not eliminate, such behavior [see Lapan and Sandler (1988) for further details]. Reputation cost arises because capitulation in any period of a multiperiod scenario augments the terrorists' perceived likelihood of future capitulations (i.e., Pt+l >pt if capitulation occurs in period t). An increase in p then augments the likelihood of an attack and, consequently, raises expected cost to the government. With reputation cost (R), the government will negotiate ex post only if n > h +R. Such cost decreases the likelihood of negotiation, but it does not necessarily eliminate the possibility unless h +R exceeds n for all realizations of n. The institution of a constitutional amendment that inhibits capitulation by imposing severe penalties may achieve this result.

228

W Enders and T Sandler

There are a number of aspects of the above representation that require future work. First, the analysis does not contain learning on the part of the two sets of agents. In the case of the terrorists, the model does not explicitly allow the terrorists to alter their beliefs with respect to p, based on the past behavior of the government. The same is true about the government's belief concerning p and other key variables. We return to this problem in Section 5. Second, the model permits an examination of a single incident and its ramification, but does not provide a theory of terrorists' campaigns, where the strategic choice for combining incidents over time is investigated. Third, the intertemporal aspects can be pursued to a greater extent. An optimal control framework, whereby terrorists are concerned about recruitment and resources over time, while the government is concerned about limiting these resources, may be required to study intertemporal aspects. This is an important direction for future work. The random nature of the government cost associated with not capitulating and the randomness of -c are important to the modeling. If all of the payoffs were deterministic, then one need only work backwards to ascertain with certainty whether the government will capitulate or whether the terrorists will attack [see, e.g., Lapan and Sandler (1993) and Selten (1988)]. The appearance of some randomness is crucial to making the problem interesting.

4. Game theoretic analysis 2: Terrorists' choice of targets In this section, we set up a simple one-period game in which two governments (targets) must determine deterrence expenditures when confronting the same terrorist group3. This exercise allows us to show the presence of a transnational transferable externality between the governments as they decide deterrence expense. Increased deterrence in one nation augments the probability that the attack will be in the other nation. Secondbest aspects arise when two or more choice variables are considered and cooperation extends to just one. We consider a case where a single terrorist group targets one of two nations, denoted by nation i, i= 1,2. When nation i increases its deterrence expenditure, D(Oi), the other target (nationj) experiences two "spillovers" or external effects: (i) By increasing the likelihood of terrorist failure, Oi, in attacks against nation i, nation i's actions increase the likelihood of attacks, j, against nationj. (ii) By increasing the likelihood of terrorist failure against nation i, nation i's actions reduce the expected cost to citizens (or property) of nationj visiting (or present) in nation i. The first externality is negative, while the second is positive. Clearly, the net outcome depends on the relative strengths of the opposing externalities. In a given period, we assume that the terrorist group can attack, at most, one country. Figure 5 depicts the corresponding game tree for a symmetric game in which each

3 This section draws, in part, from Sandler and Lapan (1988).

Ch. 9:

229

Terrorism: Theory and Applications - -.

-

- -- -

1

- ---

kovernmenTs Un oouuse

^1

--

--

U.Ierrenc1U

D(O,). D(E2) Terrorists decide whether to attack

ihe

xo'ao/ n

0

(

Zr~~~~~~~

-cQO

-C1 ,

D(@,)4 ( +

a + D(,)J

m,

ah+ D(E,)

~

h\+D()

()

0

,D(0)j

- -C~

.

- m

a +(D(,)

h + D(O,)1

(a + D()/

h + D(0,)j

Figure 5. Game tree for symmetric targets.

target experiences a cost, a, from a terrorist failure on its soil and a cost, h, from a terrorist success on its soil. In Figure 5, denotes the cost to country j of an unsuccessful attack against i, i •j, while h denotes the cost to countryj of a successful attack against i, i •j. Symmetry, invoked for convenience, ensures that the respective a and h values are also equal for each target. Reading from top to bottom, the three payoffs at the five end positions of the game indicate the net payoff to the terrorists, the cost to nation 1, and the cost to nation 2. Subsequently, we will assume full symmetry so that the terrorist group's perceived cost of failure, -ci, and its perceived payoff from success, mi, do not differ between targets. In Figure 5, the two governments simultaneously choose deterrence to thwart attacks, thereby determining the terrorists' perceived probabilities of logistical failure and success, Oi and I - Oi, in nation i. Based on their perception of these probabilities and the relevant payoffs, the terrorists then decide whether to attack and, if an attack is launched, which nation to strike. Let bi denote the terrorists' expected net payoff from attacking nation i. From the game tree, we have bi = -Oici +(l -

i)mi,

i = 1,2.

(11)

The decision rule is: bl < 0, b2
max[0, b2] b2 > max[bl,0]

no attack, attack 1,

-

attack 2.

If nation i is to increase the terrorist failure rate on its soil, the country must increase deterrence expenditures - i.e., D > 0, where primes denote derivatives. We also assume diminishing returns to effort so that Di > O0.By Equation (11), we have that increased

230

W Enders and T Sandler

defensive measures in country i reduce the expected benefits to terrorists from attacking nation i (i.e., Obil00 i < 0), but do not affect these benefits from attacking the other country (i.e., db1 /0Oi = O,i •j). Increased deterrence in country i raises the likelihood that either no attack will occur (o), or that country j (rj) rather than i, will be attacked. We assume that the government is less than perfectly informed concerning the terrorists' values and resources, so that there exists a continuous probability function, sr for attacks in nation i: ZY

= r(O1,

(12)

2),

where superscript u denotes uninformed. We assume that

a0i

0agr < ' < > , Oei

a, > 0, ' 0,

i,j= 1,2,

i X j.

(13)

Equation (13) indicates that an increase in the terrorists' perceived probability of failure in nation i decreases the likelihood of attacks there but increases this likelihood of attacks in the other country. Finally, an increase in Oi may increase the likelihood of no attacks, where ru = 1 - iur - r7u.We next denote the expected cost of an attack to government i as Ci(Oi, Oj), which includes deterrence cost and expected damages to country i's interest from attacks on its own soil and elsewhere. Assuming symmetric targets, we define l(o) = aOi +(I - Oi)h,

u(Oj) = aj

+ (1 - ej)h,

where: 1(0i) = expected costs to country i of an attack against i, v(Oj) = expected costs to country i of an attack againstj. Thus, the expected cost to government i is Ci({i,

9j) = D(oi) + Yril(Oi) + rijv(Oj),

i,j = 1,2,

i

j,

(14)

where the superscript u has been dropped from the srs. For each country, cost involves deterrence expenditures, the expected cost of a domestic attack, and the expected cost of a foreign attack on its interests. As is plausible in many cases, we assume a hostcountry disadvantage so that an attack in country i imposes greater costs on i's interests than onj's interests [i.e., l(0i) > v(Oi)]. An important exercise is to contrast the cooperative solution, where the governments work in unison to determine deterrence, with the Nash solution, where the governments

Ch. 9.'

231

Terrorism: Theory and Applications

work independently. If both countries were to cooperate, then they would choose their Ois to minimize aggregate cost: 2

CT(0 1, 02)

=

D(0

1 ) +D(0

2) + E

7i[l(Oi) + ()].

(15)

i=1

With full symmetry, so that the attack probability functions are also symmetric, it is apparent that O T = = O *, where a cooperative solution (O', 2*) satisfies CT(or, 0*)

(16)

CT(O 1 , 02)

for all nonnegative Ois. The optimizing OT satisfies

acT =D'(O*

ae

1

)[I(*'(*) '(*()] +

+

a(O*)] 2 =0

[0(*) +

a

D(*±1

=0

(17) An analogous equation holds for oCT/a0.2 A Nash noncooperative solution can be defined as (O1, O2) such that

C1(,

02)

C2(0,

0)2

C 1(0 1, 02) C 2 (0 ,

02)

01

V 02

0.

(18)

At a Nash equilibrium, each nation has chosen its deterrence to minimize its own cost, given the best-response level of 0 for its counterpart. The first-order condition for, say, ON is Oc 1

__

air

= D'(01) + r1 l'(0 1) + (01)+ o1 + v(0 2)

0t2

=

0,

(19)

where aC1/001 depends on (ON, oN). A similar expression holds for aC2 /00 2. A number of diverse scenarios can be depicted depending on our assumptions concerning collateral damage [i.e., the sizes of the v(.)s] or the relationship between v(O) and 1(0). First, we suppose that collateral damage is zero, so that v(O1) = V(02)=

(O1)

= v'(0 2 ) = 0.

Attacks in nation i are only against nation i's own interests. In this scenario, the only externality that arises is from inducing the terrorists to go elsewhere. By evaluating oCT/0Ol in Equation (17) at (ON , ON) that satisfies (19), we have

acT(oN, ON)

N

= )(ON)

a7r2

>

0

0.

(20)

Strict convexity of CT(-) ensures that Equation (20) implies overdeterrence, since 0* < ON. This follows because nations do not internalize the negative externality

W Enders and T Sandier

232

associated with the terrorist group going elsewhere. In consequence, each target expends too much effort on deterrence. If, however, an attack against country i imposes equal expected cost on both countries [i.e., (O)=1(0)], then the Nash equilibrium implies underdeterrence, because collateral protection and the general decrease in the likelihood of attacks are not taken into account [Sandler and Lapan (1988, p. 255)]. Whether noncooperative behavior leads to underdeterrence or overdeterrence hinges on: (i) the relative specificity of the costs of an attack against nation i; (ii) the differential in costs between successful and unsuccessful attacks; and (iii) the perceived responsiveness of the probability of an attack to changes in i. In the case of foreign attacks against American interests, underdeterrence is anticipated, because collateral damage on US interests is typically high and changes in Oi abroad have not induced terrorists until very recently to stage events on US soil. This last factor means that the negative externality is almost nil. Failure to account for the positive externality implies underdeterrence. We next demonstrate that parametric changes that would be beneficial in a world of cooperation may be counterproductive in a world of partial cooperation. Suppose that we extend our analysis of cost minimization to a case where each nation chooses two parameters: deterrence expenditure and the sharing of information. As before, deterrence expenditure determines the likelihood of logistical failure in country i. Policy parameter a denotes sharing information concerning the terrorists' preferences for attacking each country. We assume symmetry so that the expected cost to each government is i,j = 1, 2,

Ci = Ci(Oi, Oj, a) = F(Oi, Oj, a),

i

j,

(21)

where aFI00i >0, F/aOj > 0, and Fl/Oa = Fa < O0.The last partial indicates that a is welfare-improving in the sense of reducing costs. A cooperative solution for the O*s, given a and assuming full symmetry (i.e., O9 = * = 0*), satisfies

(22)

F 1(0*, 0*, a) + F2(0*, O*, a) = 0, where subscripts on F indicate partial derivatives (i.e., minimizing solution for the O*s, we have

F/00a

=F 1). At the cost(23)

c* = C* = F[O*(a), O(*(a), a]. A comparative static change with respect to a implies C = F + (F +F2) Oa

d

)

Fa

(24)

by the envelope theorem. Since F, < 0, full cooperation for both policy parameters is welfare-improving.

Ch. 9:

233

Terrorism: Theory and Applications

For partial cooperation, we examine the Nash equilibrium ON(a) that solves C~ = Fi[ON(a), ON(a), ] = 0, 0Oi

i = 1,2,

(25)

which implies Ci = F[ON(a), ON(a), a],

i = 1, 2.

(26)

Using Equation (26), we compute the change in expected cost owing to a change in a, which equals dC/F ( d - Fa+(F1+F2) da

dON

d =Fa+Fj-da )F

dON"\

d(7

(27)

by Equation (25). Implicit differentiation of Equation (25) gives dON

Fia

da

Fii +F O'

da Fj± i,j= 1,2, F' i

Xj,

(28)

which may be positive or negative depending on Fia. If, say, Fia is negative, then dON/da is positive since (Fii+Fi)>Oby the second-order conditions. A negative Fia is consistent with reducing a country's own marginal deterrence costs through the gain in information. In consequence, the sign of dCi/da in Equation (27) may be positive, thus indicating increased costs from sharing information when deterrence is decided independently. For this scenario, the increase in information allows a potential target to know better how to make its deterrence divert attacks, thereby leading to a larger negative externality and greater inefficiency. For terrorism, piecemeal policy coordination among governments may be worse than no coordination whatsoever. 5. Further game analyses In a recent paper, Lapan and Sandler (1993) extended the terrorism analysis to a setting of incomplete information, in which a signalling equilibrium 4 may allow the government to limit its expected cost from attacks, even though the likelihood of surrender may increase. These authors assumed asymmetric information: The government does not know the strength or the resources of the terrorists, but the terrorists know their own resources and the beliefs of the government 5. In this scenario,

4 Binmore (1992) contains an excellent discussion of signalling equilibrium. 5 Scott (1989, 1991) examined terrorism when the terrorists are uninformed about the type of government that they confront, but the government is informed. Scott did not consider a signalling equilibrium.

234

W Enders and T Sandler

first-period attacks of the terrorists may provide information to the government about the type of terrorist group that it confronts. Attacks serve as a signal that the government processes in order to update its posterior beliefs concerning the terrorists' resources. Based on these posterior beliefs, the government decides whether to capitulate or resist. The terrorist group needs to allocate resources between current and future attacks. Current attacks can signal strength and cause the government to concede, but these attacks limit resources for future attacks. Since it is the expected cost of future attacks that matters to the government, the terrorists face a resource allocation dilemma. In Lapan and Sandler (1993), action was sequenced as follows: Nature chooses the resources of the terrorist group, drawn from a distribution that is common knowledge. The terrorists, but not the government, observe the draw. Next, the terrorists decide first-period attacks, which impose costs on the government. Based on these first-period attacks, the government revises its beliefs and decides whether to surrender. If no surrender occurs, then the terrorists choose the level of period-2 attacks based on their remaining resources. These attacks impose further costs on the government. The game ends. The two-period structure is appropriate, since nothing further would be learned in subsequent periods. The authors derived a perfect Bayesian equilibrium that consists of a strategy combination and a set of beliefs such that, at each node, the strategies for the game's remainder are Nash (given beliefs) and the beliefs at each information set are rational, based on the facts revealed thus far in the game. The analysis indicated that regret may characterize the outcome when either a government capitulates to a terrorist group with insufficient resources, or else a government does not capitulate to a group with sufficient resources to induce capitulation under complete information. The government prefers the partial-pooling signalling equilibrium, despite probabilistic regret, over the never-surrender equilibrium, since its expected cost is reduced [see a similar analysis by Overgaard (1994)]. Much remains to be done in applying game theory to the study of terrorism. First, the rules of the game need to be changed to allow for learning over more than a single period. Without this important extension, the justification of a terrorist campaign over multiple periods cannot be established. To date, both the empirical6 and theoretical literature have focused on explaining single events and their repercussions. Second, incomplete information should be made two-sided so that both the terrorists and the government must update their priors over time. This extension can also justify the need for more than two periods in the game tree. Third, differential game techniques should be applied so that the accumulation of terrorists' resources over time can be examined under alternative antiterrorism policies. A host of alternative equilibrium concepts and learning paths could be introduced.

This empirical literature includes Atkinson, Sandler and Tschirhart (1987), and Enders, Sandler and Cauley (1990a, 1990b).

6

Ch. 9:

Terrorism: Theory and Applications

235

6. Choice-theoretic models of terrorism In the absence of a game-theoretic model, researchers have tended to use a choicetheoretic model to analyze various aspects of terrorism. Such models have typically focused on the terrorist who chooses among terrorist and nonterrorist activities subject to resource constraints 7 . Much of this literature has relied on the economics of crime methodology based on a state-preference approach. Landes' (1978) model of skyjackings is an apt example in which a potential hijacker is contemplating the forcible diversion of a commercial aircraft for political purposes. Such purposes could include publicity for a cause, economic freedoms, or political freedom, all of which adds to the individual's welfare in his/her desired destination. Of course, risks must be incurred in carrying out the act. To determine his/her course of action, a potential hijacker must first estimate the expected utility (EU) from hijacking an aircraft from country i to country j: EU = (1 - Pa) U(Wj) + PaP,U(Wi - S) +Pa(1- Pc) U(Wj- C),

(29)

where: Pa = hijacker's subjective estimate of probability of apprehension; = hijacker's estimate of conditional probability of imprisonment, given Pc apprehension; = hijacker's wealth from successful hijacking to country j; Wj Wi = hijacker's wealth in country i; S

= monetary equivalent of a sentence;

C = monetary cost associated with apprehension without imprisonment. The three right-hand utility expressions correspond to three possible states or outcomes that include success, apprehension with no conviction, and apprehension with conviction. The hijacker will engage in the incident provided that the expected utility from doing so exceeds the utility, U(Wi), from not hijacking the plane. That is, a hijacking takes place when EU > U(Wi). Based on this comparison, Landes (1978) specified an offense function, 0 = O(Pa, Pc, S, C, Z),

(30)

where overbars indicate average values, and Z denotes the average wealth differential between country j and country i. In essence, this function relates the number of hijackings to exogenous variables, some of which can be controlled by the authorities. 7 Sandler, Tschirhart and Cauley (1983), however, modeled both the choices of the terrorists and the government in an interactive framework, while Enders, Sandler and Parise (1992b) modeled the choice of a tourist who faces a risk of terrorism.

236

W Enders and T Sandler

Landes (1978) presented two regressions for US hijackings based upon US Federal Aviation Administration data on skyjackings for 1961-1976. The first regressed the quarterly total of hijackings on the probability of apprehension, the probability of conviction, sentencing, etc. The second regressed the time interval between hijackings on the same set of variables. Both analyses identified the probability of apprehension and the length of sentence (if convicted) as significant deterrents. For most regression runs, the conditional probability of conviction was almost significant. Landes also estimated that between 41 and 50 fewer hijackings took place in the USA from the start of 1973, following the installment of metal detectors in US airports through the third quarter of 1976. Other choice-theoretic models of terrorist behavior have employed a household production approach in which terrorists are depicted as rational actors who attempt to maximize a shared goal, subject to a resource constraint. This shared goal may be denoted as utility or expected utility derived from the consumption of basic commodities, produced from terrorist and nonterrorist activities. Basic commodities may include political instability, publicity, an atmosphere of fear, or extortion. Alternative terrorist attack modes are complementary or substitutable provided that they are used to produce the same basic commodities. Substitution possibilities are enhanced when attack modes are logistically similar and yield the same basic commodities in nearly identical proportions. Complementarity results when a combination of attack modes is required to produce one or more basic commodities and, moreover, such a combination enhances the marginal productivity of the constituent modes in producing the basic commodities. For example, threats when combined with actual bombings may create a greater state of fear than either alone. In any period, a resource constraint limits the terrorist group's expenditures among the choices not to exceed its monetary and nonmonetary resource endowments. The expenditures on any activity consist of the product of the activity's per-unit price and the level of the activity. Each terrorist and nonterrorist tactic has a per-unit price that includes the value of time and other resources needed to accomplish the act. Consider the class of terrorist acts. The prices faced by the terrorists for each tactic are determined, in large part, by the government's allocation of resources to thwart various acts of terrorism. If, for instance, the government were to secure its embassies or military bases, then attacks against such facilities would become more costly on a per-unit basis. If, moreover, the government were not at the same time to increase the security for embassy and military personnel when outside their facilities, then attacks directed at these individuals (e.g., assassinations) would become relatively cheaper. The installation of metal detectors in airports would make skyjackings relatively more costly than other kinds of hostage-taking missions such as kidnappings. If a choice-theoretic model of terrorist decisions were to be formulated, then some obvious comparative static results would follow. For example, government policies that increase the relative price of one type of terrorist tactic would cause a terrorist group to substitute out of the now-more-costly tactic into those terrorist and nonterrorist activities whose prices are now relatively less costly. Government interventions that

Ch. 9:

237

Terrorism: Theory and Applications

raised the price of all terrorist tactics or that reduced terrorists' resources would cause nonterrorist activities to increase relative to terrorist actions. Complementary tactics would respond in a similar fashion to relative price changes, while substitute tactics would respond in an opposing fashion. These propositions have been tested by some of the time-series papers to be described in Sections 7 and 8. To date, empirical applications have been to the choice-theoretical model rather than the game theory representations.

7. Intervention analysis A critical response to the rise in skyjackings occurred when the USA began to install metal detectors in all US airports in January 1973. Other international authorities followed shortly thereafter. To quantify the effects of installing metal detectors, we employ the basic framework used by Cauley and Im (1988) and Enders, Sandler and Cauley (1990a,b). Let yt = ao + alyt_l + cpt + Et,

al < 1,

(31)

where pt is the intervention (or policy) variable that takes on the value of zero prior to 1973:1 and unity beginning in 1973:1; yt is the number of skyjackings in quarter t; and Et is a white noise disturbance. For t < 1973:1, pt is zero so that the intercept term is a and the long-run mean of the series is ao/(1 -al). Beginning in 1973, the intercept jumps to ao + c (since P1973:1 jumps to unity). Thus, the initial effect of the metal detectors is given by the magnitude of c, whose statistical significance can be tested using a standard t-test. We would conclude that metal detectors reduced the number of skyjacking incidents if c is negative and statistically different from zero. The long-run effect of the intervention is given by c/(l - al), equal to the new longrun mean (ao +c)/(1 - al) minus the value of the original mean ao/(1 - al). If 0 < al < 1, the absolute values of the magnitude of the impacts are an increasing function of time. As we move further away from the date at which the policy is introduced, the greater is the absolute value of the magnitude of the policy response. There are several important extensions to the intervention example. For example, the model need not be a first-order autoregressive process. A more general autoregressive, moving average (ARMA) intervention model has the form Yt = ao +A(L)yt-_ + cpt + B(L) Et,

(32)

where A(L) and B(L) are polynomials in the lag operator L. Also, the intervention need not jump from zero to unity in 1973:1 and then remain at unity. With specific reference to skyjackings, there are other forms of the intervention function: (i) Pure Pulse: the values of Pt can be zero for all periods except for one particular period at which pt = 1. The pure pulse best characterizes a purely temporary intervention.

W Enders and T Sandler

238

Table 2 Metal detectors and skyjackings (Quarterly Time Series) Series

Pre-intervention

a,

Impact effect

mean

Transnational {TS,} US domestic {DS,} Other skyjackings {OS,}

3.032 (5.96) 6.70 (12.02) 6.80 (7.93)

Long-run effectb

0.276 (2.51)

0.237 (2.14)

-1.29 (-2.21) -5.62 (-8.73) -3.90 (-3.95)

-1.78 -5.62 -5.11

t-statistics are in parentheses. b The long-run effect is given by: c/(1-a,). a

(ii) Gradually changing function: an intervention may not reach its full force immediately. Often, the shape of the intervention function is clear from a priori reasoning. When there is an ambiguity, estimate the plausible alternatives and then use the standard Box-Jenkins model selection criteria to choose the most appropriate model. To illustrate the procedure, we consider the methodology used by Enders, Sandler and Cauley (1990b) to measure the effects of metal detectors on US domestic skyjackings {DSt}, transnational skyjackings {TSt} (including those involving the USA), and on all other skyjackings {OSt}. Since there are only 5 years of quarterly data (i.e., 20 observations) for the pre-intervention period, the best-fitting ARIMA model was estimated over the 1973:I-1988:IV period. Standard Box-Jenkins model selection criteria (including diagnostic checks of the residuals) resulted in an AR(1) model for the {TSt} and {OSt} sequences and a pure noise model (i.e., all autoregressive and moving average coefficients equal to zero) for the {DSt } sequence. The next step was to estimate the various models over the entire sample period including the effect of the intervention. The installation of metal detectors was tentatively viewed as an immediate and permanent intervention. As shown in Table 2, the installation of metal detectors reduced each of the three types of skyjacking incidents. The most pronounced effect was on US domestic skyjackings, which immediately fell by 5.62 incidents per quarter. All effects on {DSt} were immediate, since the estimate of al is zero. The situation is different for the {TSt} and {OSt} series, because the estimated autoregressive coefficients are different from zero. On impact, transnational skyjackings and other types of skyjacking incidents fell by 1.29 and 3.9 incidents per quarter. The long-run effects were estimated to be 1.78 and 5.11 fewer incidents per quarter, respectively. Diagnostic checking is particularly important since the observations from the preand post-intervention periods have been merged. A well-estimated intervention model will have three characteristics. First, the estimated coefficients should be of "high quality" in that all coefficients are statistically significant at conventional levels.

Ch. 9:

Terrorism: Theory and Applications

239

Second, the residuals should approximate white noise. If the residuals are serially correlated, forecasts from the estimated model cannot possibly be making use of all available information. Third, the tentative model should outperform plausible alternatives. In the skyjacking example, a plausible alternative was to model the intervention as a gradually increasing process. This is particularly true since the impact effect was immediate for US domestic flights and convergent for transnational and other domestic flights. The conjecture was that metal detectors were gradually installed in non-US airports and, even when installed, the enforcement was sporadic. As a check, we modeled the intervention as gradually increasing over the year 1973. Although the coefficients were nearly identical to those reported in Table 2, the Akaike Information Criterion (AIC) and Schwartz Bayesian Criterion (SBC) were slightly lower (indicating a better fit) using the gradually increasing process. Hence, it is reasonable to conclude that metal detector adoption was more gradual outside of the USA. An important question concerns whether terrorists, when confronted with more secure airports, substituted into other kinds of hostage-taking events. Now let y, denote the sum of barricade-and-hostage-taking incidents and kidnappings (BHKN). The estimated intervention model for the BHKN series showed an immediate increase of 6.663 incidents per quarter (i.e., c=6.663) as a result of the introduction of metal detectors in airports. The long-run effect was estimated to be an additional 10.59 incidents per quarter. Hence, the reduction in skyjackings cannot, by itself, serve as a true measure of success for the metal detector technology. 7.1. Other interventions: Increased embassy security In October 1976, US embassy security increased drastically. At the same time, improvements in other likely US targets subject to barricade and hostage seizure were undertaken. In this case, t represents the quarterly totals of "crimes against US diplomats" (CDA) over the 1971:II-1987:IV period and Pt = 0 for all observations prior to 1976:III. The most appropriate intervention model indicated that enhanced embassy security decreased CDA by 1.94 incidents per quarter; since all autoregressive coefficients equaled zero, the immediate and long-run impacts were identical. In addition, there was evidence of a mild substitution into attacks against non-US diplomats [Enders, Sandler and Cauley (1990b, pp. 13-14)]. 7.2. The Libyan raid and other retaliations On the morning of April 15, 1986, US warplanes attacked Libya. The stated reason for the attack was Libya's alleged involvement in the terrorist bombing of the La Belle Discotheque in West Berlin. Since 18 of the F-lll fighter-bombers were deployed from British bases at Lakenheath and Upper Heyford, England, the UK implicitly assisted in the raid. The remaining US planes were deployed from aircraft carriers in the Mediterranean Sea. Now let yt denote all transnational terrorist incidents directed

W Enders and T:Sandler

240

against the USA and the UK during month t. A plot of the {yt} sequence exhibits a large positive spike immediately after the bombing; the immediate effect seems to be a wave of anti-USA and anti-UK attacks to protest the retaliatory strike. One key issue concerns the permanence of the effects of the raid. We first let pt = 0 for all t < April 1986 and unity thereafter. Using standard Box-Jenkins model selection criteria, we obtain the following estimates (with t-statistics in parentheses): y, = 5.58 + 0 .3 3 6yt_1 +0 . 12 3 y,_5 +2 . 6 5pt,. (0.84) (3.26) (5.56)

(33)

Note that the coefficient of pt has a t-statistic of 0.84 that is not significant at the 0.05 level. Alternatively, when p, is allowed to be 1 only in the month of the attack, we obtain: y, = 3.79 + 0 .3 27 yt- +0.157yt-,5 + 38.9pt. (2.59) (6.09) (5.53)

(34)

In comparing the two estimates, it is clear that magnitudes of the autoregressive coefficients are similar. Although Ljung-Box Q-tests indicate that the residuals from both models approximate white noise, the second model is preferable. The coefficient on the pulse term is highly significant and the AIC and SBC both select the second specification. Our conclusion is that the Libyan bombing did not have the desired effect of reducing terrorist attacks against the USA and the UK; instead, the bombing caused an immediate quarterly increase of almost 39 attacks. Subsequently, the number of attacks declined; 0.327 of these attacks are estimated to persist for one period (0.327 x 38.9= 12.7). Since the autoregressive coefficients imply convergence, the long-run consequences of the raid are estimated to be zero. In a related study, Brophy-Baermann and Conybeare (1994) developed a rational expectations model of terrorist attacks. Optimizing terrorists will select a "natural rate" of attacks and only unanticipated retaliations will cause the actual rate to differ from the natural rate. The direct implication is that the retaliating country should follow a flexible, rather than a predictable, retaliation policy. Brophy-Baermann and Conybeare let the {y,} sequence represent transnational terrorist attacks against Israel over the 1968:I-1989:IV period. Six major Israeli retaliations were selected as the interventions. The authors argued that after the first major reprisal in 1972 (nine Israeli air raids against PLO bases in Syria following Black September's attack on Israeli athletes at the 1972 Olympic Games) the element of surprise became difficult to maintain. Hence, each intervention was modelled as a pure pulse function. Standard Box-Jenkins techniques indicated that the {yt} sequence is a white-noise process with a constant mean of 2.13 incidents per quarter. It was estimated that the 1972 retaliation induced an upward shift of 9.39 attacks on impact. The number of additional attacks fell to 0.5 attacks above the natural rate after three quarters. The other five Israeli retaliations were not statistically significant.

Ch. 9:

Terrorism: Theory and Applications

241

8. Vector autoregression models of terrorism An interesting development has been the application of vector autoregression (VAR) analysis to the study of terrorism. When it is not apparent whether a variable is exogenous, a natural extension of intervention analysis is to treat each variable symmetrically. Let the time path of each variable be affected by current and past realizations of itself and the other variables in the system. Consider the simple bivariate system Yt = blo - b12 zt + Y11Yt-1 + Y12Zt I + Eyt,

(35)

zt = b20 - b2 1Yt + Y21Yt-1 + Y22Zt l + Ezt,

(36)

where it is assumed that: (i) both Yt and zt are stationary; (ii) yt and Ezt are whitenoise disturbances with standard deviations of ay and oz, respectively; and (iii) {Eyt} and {Ezt} are uncorrelated white-noise disturbances. Equations (35)-(36) constitute a first-order VAR, since the longest lag length is unity. The structure of the system incorporates feedback because current realizations of Yt and zt are allowed to affect one another; however, the model is not in reduced form as each variable has a contemporaneous effect on the other. Fortunately, it is possible to transform the system of equations into the more usable form Yt = alo +allYt-i + a12 Zt-1 + elt,

(37)

zt = a20 + a2lYt-1 + a22t-1 + e2t,

(38)

where Eytl - 12Ezt 1-

12b21

Et - b2 1Eyt 1 - b12b21

More compactly, the system of equations can be written as Xt

= A 0 +Alxt

+ et,

(39)

where xt is the 2xl vector (yt,zt)'; Ao is the 2xl vector (al0 , a20)'; A is the 2x2 matrix with elements aij; and et is the 2xl vector (elt,e2t)'. Note that the right-hand side of Equation (39) contains only pre-determined variables and that the error terms are serially uncorrelated with constant variance. Hence, each equation in the system can be estimated using OLS. Moreover, OLS estimates are consistent and asymptotically efficient. Even though the errors are correlated across equations, seemingly unrelated regressions (SUR) do not add to the efficiency of the estimation procedure, because both regressions have identical right-hand-side variables. A test of causality is whether the lags of one variable enter into the equation for another variable. In terms of Equations (37)-(38), the {Yt} sequence is said to Grangercause the {zt} sequence if a21 is statistically different from zero. Similarly, the {zt} sequence Granger-causes the {Yt} sequence if a 2 is statistically different from zero.

W Enders and T Sandler

242

Impulse response analysis allows the researcher to trace out the time path of the various shocks on the variables contained in the VAR system. Write Equations (37) and (38) in matrix form to obtain:

I

[Yt LZ

a

[all

a20

a21

a1 2 [Yt-i] + [elt a22 J LZt-1 e2t

(40)

or, [yt1 Zt

[1 + Z

[all a12 i=

a21

]

a22 J

[elj]

(41)

Le 2 t-i

where y and Z are the mean values of {y } and {zt}, respectively. Equation (41) is called a vector moving average (VMA) model in that it expresses all variables in terms of the disturbances. Using the OLS estimate of Al, a researcher can shock elt and e2 by one standard deviation to trace out the effects of each disturbance on the time paths of the {yt} and {zt} sequences. The resulting time paths are called the impulse response functions. With this same methodology, we can decompose the variance of each sequence into the proportion due to each of the shocks. The main limitation of the VAR methodology is that it is not possible to recover (or identify) the pure Eyt and Ezt shocks without imposing additional restrictions on the parameters of the contemporaneous innovation matrix (i.e., on the values of bl 2 or b21). A typical decomposition strategy is to use the triangular Choleski decomposition so as to restrict either bl 2 or b21 to equal zero. As shown in Enders (1995), Granger-causality tests, impulse response analysis, and variance decompositions can be performed in norder multivariate generalizations of Equation (39). Enders and Sandler (1991) used the VAR methodology to estimate the impact of terrorism on tourism in Spain during the period from 1970 to 1988. Most transnational terrorist incidents in Spain during this period are attributed to the Basque Fatherland and Liberty (ETA) and its splinter groups. Because increases in tourism may generate terrorist acts, the VAR methodology was used to examine the reactions of tourists to terrorism and the reactions of terrorists to tourism. For a 24-month forecasting horizon, the variance decompositions were such that each time series explained the preponderance of its own past values. Terrorist incidents explained 8.7% of the forecast error variance of Spain's tourism, while tourism explained only 2.2% of the forecast error variance of terrorist incidents. More important, Granger-causality tests indicated that the effects of terrorism on tourism were significant at the 0.006 level, while the effects of tourism on terrorism were not significant at conventional levels. Thus, causality was unidirectional: terrorism affected tourism but not the reverse. The terrorism series appeared to be autonomous in the sense that neither series significantly explained the forecast error variance of current incidents. This result is consistent with the notion that terrorists randomize their incidents so that any one incident is not predictable on a month-to-month basis.

Ch. 9:

Terrorism: Theory and Applications

243

The impulse response functions were such that after a "typical" terrorist incident, tourism to Spain began to decline in month three. After the sixth month, tourism started to revert to its original level. There did appear to be a rebound in months eight and nine, followed by another drop in tourism in month nine that reached the maximum decline about one year after the original incident. Obviously, some of this pattern was due to the seasonality in the series. However, tourism slowly recovered and generally remained below its pre-incident level for a substantial period of time. Aggregating all 36 monthly impacts, we estimated that the combined effects of a typical transnational terrorist incident in Spain was to decrease the total number of foreign visits by 140 847 people. By comparison, a total of 5 392 000 tourists visited Spain in 1988 alone. Nelson and Scott (1992) used a VAR analysis to examine the interactions between media publicity and terrorist actions. The conventional wisdom is that media coverage of high-profile terrorist incidents induces other terrorist acts. This view would be plausible if the objective function of terrorists included notoriety. If, instead, terrorists are interested in political objectives, publicity alone should not encourage other terrorist acts. Specifically, Nelson and Scott examined the interrelationships between the number of terrorist incidents directed against Israel and the number of column inches devoted to terrorism in the New York Times. Their important finding was that the publicity variable does not Granger-cause terrorism. Enders and Sandler (1993) developed a methodology that incorporates intervention analysis into the VAR framework. Because terrorists must weigh the relative costs and benefits of various modes of attack when designing their campaigns, one should expect that the time series for one mode (or target) of attack may depend not only on the time series for that mode, but also on those of related modes. As such, univariate econometric models [see Landes (1978), Cauley and Im (1988)], which consider one attack mode at a time, may be misspecified. The interaction among the modes of attack can be written as a straightforward generalization of Equation (31): Yt = Ao +A(L)yt-_ + Cpt + Et,

(42)

where yt is the nx 1 vector (Ylt,Y2t,... ,Ynt)' with Yit being the number of type i ter-

rorist incidents occurring during time period t; A0 is an nx 1 matrix with elements consisting of the sum of a constant and three seasonal dummy variables; A(L) is an n x n matrix consisting of polynomials in the lag operator L; and Et is the n x vector (Elt, E2t,... , nt)' such that Eit is an independent identically distributed random variable and E[tE] = F is not necessarily diagonal. In Equation (42), matrices C andpt are of particular importance for assessing the success of the alternative interventions. Here, Pt is the K x 1 vector (Pt,... ,PKt)', where Pkt is an indicator of whether policy k is in effect at time t. For metal detectors, the policy intervention Pit is set equal to zero for all t < January 1973 and equal to unity for all t > January 1973. Each element Cik in the n xK coefficient matrix C measures the direct impact effect of policy intervention Pk on the time series of incident type Yi. Even if Cik is equal to zero, it is not possible

W. Enders and T Sandler

244

to conclude that incident type i is unaffected by policy intervention Pk. When the off-diagonal elements of A(L) are not zero, there are interactions among the various incident types. Thus, policy intervention pk may have an indirect effect on incident type i through its effects on a related incident type. This observation motivates the need to study terrorist-thwarting interventions in a VAR analysis. The direct effect of an intervention on a particular mode of attack may be zero (i.e., the relevant coefficient in the C matrix may be zero) even though the indirect effects, through the elements of A(L), may be nonzero. It is instructive to use one of the models estimated in Enders and Sandler (1993) as an example. In this model, we distinguished between nonresourceusing threats and hoaxes and resource-using events. Although costly for governments to police, the former are relatively inexpensive incidents for terrorists to instigate. In addition, the assassination (AS) series was included to ascertain whether the hardening of targets may have led to more killings as officials and others left more-secure grounds. The Yt vector consisted of a 5 x 1 vector of skyjackings (SKY), assassinations (AS), barricade-and-hostage-taking missions and kidnappings (BHKN), threats and hoaxes (THHX), and all other incidents (OT). The policy intervention matrix Pt was the 4 x 1 vector consisting of the indicators for the installation of metal detectors (METAL), US embassy fortifications in 1976 and 1985 (EMB76 and EMB85, respectively), and the Libya raid (LIBYA). An appropriate strategy to estimate the matrices in Equation (42) consists of the following steps: Step 1. Equation (42) was estimated with OLS for lag lengths of eight quarters, four quarters, and two quarters. As described by Enders (1995), lag lengths can be selected using the likelihood ratio test for the block restriction that the degrees of the polynomials in A(L) are all less than or equal to some specified value. Step 2. To ascertain the importance of the interrelationships among the various attack modes, the variance decompositions was obtained. With the use of the lag lengths from Step 1, the vector moving average (VMA) representation of Equation (42) was found and expressed each incident type i as being dependent on the current and past values of all the innovations and interventions. Because a VMA can not be estimated directly, we used the orthogonalized innovations obtained from a Choleski decomposition. When the innovations in the various series are correlated, the ordering of the variables used in the decomposition can be important. In such cases, it is standard to reverse the ordering of the variables used in the initial factorization. Table 3 reproduces a portion of the variance decompositions reported by Enders and Sandler (1993). To conserve space, Table 3 reports only the results for the ordering SKY

-

BHKN -+ AS

-

THHX -- OT

using a 24-quarter forecasting horizon. Although each incident type explains a large portion of its own forecast error variance, there is a complex set of interactions among series. THHX is associated with all of the other series: depending on the ordering, the other four series explain as much as 65% of the forecast error variance of THHX. The ordering is particularly important for the SKY and THHX series; the correlation coefficient between the residuals of the

Ch. 9:

245

Terrorism: Theory and Applications Table 3 Variance decomposition (24-Quarter Forecast Horizon) SKY

BHKN

AS

84.4 (0.726)a

5.05 (0.305)

2.62

2.37

(0.147)

(0.915)

(0.706)

BHKN

14.1 (0.279)

67.9 (0.554)

0.927 (0.692)

1.20 (0.259)

15.9 (0.242)

AS

14.1 (0.809)

6.87 (0.235)

59.8 (0.006)

0.761 (0.159)

18.4 (0.207)

THHX

35.8 (0.125)

8.03 (0.049)

11.3 (0.030)

34.7 (0.000)

10.2 (0.170)

OT

6.84 (0.823)

3.39 (0.731)

10.7 (0.804)

9.47 (0.277)

69.6 (0.211)

SKY

5.54

THHX

OT

The numbers in parentheses are the significance levels for the joint hypothesis that all lagged coefficients of the variable in question can be set equal to zero.

a

two series in the first ordering is 0.382 implying that SKY and THHX move strongly together. Step 3. The forecasts from an unrestricted VAR are known to suffer from "overparameterization". One way to eliminate the problem is to use the results of the Granger-causality tests as described by Enders and Sandler (1993). Reestimating the model after eliminating all insignificant groups of coefficients results in a parsimonious representation of the VAR process. Because the right-hand-side variables were no longer identical in each regression, each was estimated using Seemingly Unrelated Regressions (SUR). The SUR estimates of the coefficients (and their associated t-values) for the C matrix and for the long-run effects are reported in Table 4. Step 4. The SUR estimates of the C and A(L) matrices could be used to show the effectiveness of the various interventions. Each element in C shows only the impact effect of the intervention; the lagged impact and indirect effects are captured by the elements in the A(L) matrices. Because of the complexity of the interactions, the simplest way to demonstrate the short- and long-run policy effects is through the moving average representations [Enders and Sandler (1993)]. Diagnostic checking included the use of the Ljung-Box Q-statistic to determine serial correlation in the residuals. Signs of non-stationarity were investigated. The non-stationarity issue is difficult to address in a system of equations with multiple interventions; a variable may appear to be non-stationary owing to the effects of successive interventions. In addition to the standard Dickey-Fuller tests, the Perron (1989) methodology can be. used to test for a unit root in the presence of structural change.

w Enders and T2Sandler

246

Table 4 Coefficients and significance levels (t-statistics are in parentheses) Policies

SKY

Impact effects Metal -14.1"*

BHKN

AS

THHX

OT

11.6**

6.58*

1.75

11.3

EMB76

(-5.59) 2.51*

(3.74) -1.41

(2.30) 3.56'

(0.579) 8.67**

(1.15) -18.6*

EMB85

(1.83) 0.100

(-1.12) 3.54*

(1.89) -0.967

(4.39) -5.31*

(-1.89) -2.81

LIBYA

(0.041) -4.83

(2.21) -1.62

(-0.672) 1.57

(-2.11) 50.5*

(-0.408) 58.4*

(0.347)

(10.6)

(2.59)

4.1#

-9.5#

17.8 #

#

#

-21.1#

(-1.06) Long-run effects Metal -13.0

#

EMB76

0.98#

EMB85 LIBYAa

-0.52# NA

(-0.387)

5.3# #

-0.20 5.4' NA

8.2 -1.7 NA

11.8

4.1 # NA

-5.4 NA

·* Denotes a significant immediate impact at 0.05 level. * Denotes a significant immediate impact at 0.10 level. Denotes a significant long-run impact effect or that the impact effect was significant through its effects on an important explanatory variable in the VAR analysis. a All effects of LIBYA are temporary effects.

In Table 4, much of the increased terrorism generated by the Libyan raid augmented nonresource-using THHX. Nearly half of the terrorist reaction to the raid took the form of words and not deeds. Terrorists lashed back at the raid with relatively low-cost incidents (THHX and OT) in the hopes of imposing sizable costs on others. The impact and long-run effects of metal detectors on SKY are larger than in the intervention studies. In addition, we now find a larger substitution into BHKN and AS. The longrun effect on THHX is estimated to decline by over nine incidents per quarter. Since many THHX involved planes and embassies, which were made more secure by metal detectors, these threats may have declined as they became less credible. Terrorists may have substituted deeds for words owing to metal detectors. Both embassy fortifications had significant impact effects. EMB76 had the immediate impact of increasing THHX by almost nine incidents per quarter. There is only weak evidence that EMB76 affected the other series: at the 0.10 level for a one-tail test, EMB76 increased SKY and AS, while it decreased OT. The impact and long-run effects of EMB76 on AS are especially interesting, since they suggest that protected individuals were more prone to assassinations when leaving secured grounds. Insofar as THHX require few inputs, these events can increase without necessitating substitution out of other events. EMB85 is estimated to have increased BHKN at the expense of

Ch. 9:

Terrorism: Theory and Applications

247

THHX. After the late 1970s, BHKN consisted almost entirely of kidnappings; hence, it is not surprising that EMB85 led to substitution into BHKN. On impact, BHKN rose by 3.54 incidents per quarter. In the long run, BHKN gradually increased to 5.4 additional incidents per quarter.

9. Concluding remarks Economic methods and game theory present appropriate tools to analyze various aspects of terrorist behavior and the formulation of effective antiterrorism policies. Although economic studies do not provide a substitute for policy-relevant political science studies, economic analyses are complementary and introduce a degree of theorizing and empirical quantification that is needed. The recent focus in economics on asymmetric information, where parties are differentially informed, has provided tools (e.g., signalling games) that are especially appropriate to the study of the interface between a government and terrorists, where information is incomplete. Even the standard consumer-theory model yields insights when applied to terrorism - i.e., that antiterrorism policies may, through substitutions and complementarities, lead to unintended outcomes. These outcomes were quantified by the VAR analysis. In the future, dynamic methods can offer insights regarding the evolution of terrorist groups and their campaigns. More advanced VAR analyses can provide future empirical findings. Finally, empirical applications need to be developed for the game analysis.

References Alexander, Y., D. Carlton and P. Wilkinson, eds., 1979, Terrorism: Theory and practice (Westview, Boulder, CO). Atkinson, S.E., T. Sandler and J.T. Tschirhart, 1987, Terrorism in a bargaining framework, Journal of Law and Economics 30, 1-21. Binmore, K., 1992, Fun and games (D.C. Heath, Lexington, MA). Brophy-Baermann, B., and J.A.C. Conybeare, 1994, Retaliating against terrorism: Rational expectations and the optimality of rules versus discretion, American Journal of Political Science 38, 196-210. Cauley, J., and E.I. Im, 1988, Intervention policy analysis of skyjackings and other terrorist incidents, American Economic Review Papers and Proceedings 78, 27-31. Crenshaw, M., 1992, Current research on terrorism: The academic perspective, Studies in Conflict and Terrorism 15, 1-11. Dickey, D., and W. Fuller, 1979, Distribution of the estimator for autoregressive time series with a unit root, Journal of American Statistical Association 74, 427-431. Dickey, D., and W. Fuller, 1981, The likelihood ratio statistics for autoregressive time series with a unit root, Econometrica 49, 1057-1072. Enders, W., 1995, Applied Econometric Time-Series (Wiley, New York). Enders, W., and T. Sandler, 1991, Causality between transnational terrorism and tourism: The case of Spain, Terrorism 14, 49-58. Enders, W., and T. Sandler, 1993, The effectiveness of anti-terrorism policies: Vector-autoregressionintervention analysis, American Political Science Review 87, 829-844.

248

W Enders and T: Sandler

Enders, W., T. Sandler and J. Cauley, 1990a, U.N. conventions, technology and retaliation in the fight against terrorism: An econometric evaluation, Terrorism and Political Violence 2, 83-105. Enders, W., T. Sandler and J. Cauley, 1990b, Assessing the impact of terrorist-thwarting policies: An intervention time series approach, Defence Economics 2, 1-18. Enders, W., G.F. Parise and T. Sandler, 1992a, A time-series analysis of transnational terrorism: Trends and cycles, Defence Economics 3, 305-320. Enders, W., T. Sandler and G.F. Parise, 1992b, An econometric analysis of the impact of terrorism on tourism, Kyklos 45, 531-554. Gottman, J.M., 1981, Time-series analysis (Cambridge University Press, New York). Im, E.I., J. Cauley and T. Sandler, 1987, Cycles and substitutions in terrorist activities: A spectral approach, Kyklos 40, 238-255. Islam, M.Q., and W.N. Shahin, 1989, Economic methodology applied to political hostage-taking in light of the Iran-Contra affair, Southern Economic Journal 55, 1019-1024. Landes, W.M., 1978, An economic study of U.S. aircraft hijackings, 1961-1976, Journal of Law and Economics 21, 1-31. Lapan, H.E., and T. Sandler, 1988, To bargain or not to bargain: That is the question, American Economic Review Papers and Proceedings 78, 16-20. Lapan, H.E., and T. Sandler, 1993, Terrorism and signalling, European Journal of Political Economy 9, 383-397. Lee, D.R., 1988, Free riding and paid riding in the fight against terrorism, American Economic Review Papers and Proceedings 78, 22-26. Lee, D.R., and T. Sandler, 1989, On the optimal retaliation against terrorists: The paid-rider option, Public Choice 61, 141-152. Lichbach, M.I., 1987, Deterrence or escalation in repression and dissent, Journal of Conflict Resolution 31, 266-297. Mickolus, E.F., 1980, Transnational terrorism: A chronology of events 1968-1979 (Greenwood Press, Westport, CT). Mickolus, E.F., 1982, International terrorism: Attributes of terrorist events, 1968-1977 (ITERATE 2) (Inter-University Consortium for Political and Social Research, Ann Arbor, MI). Mickolus, E.F., T. Sandler and J.M. Murdock, 1989a, International terrorism in the 1980s: A chronology of events, Vol. 1 (1980-1983) and Vol. 2 (1984-1987) (Iowa State University, Ames, IA). Mickolus, E.F., T. Sandler, J.M. Murdock and P. Fleming, 1989b, International terrorism: Attributes of terrorist events 1978-1987 (ITERATE 3) (Vinyard Software, Inc., Falls Church, VA). Nelson, P.S., and J.L. Scott, 1992, Terrorism and the media: An empirical analysis, Defence Economics 3, 329-339. Overgaard, PB., 1994, Terrorist attacks as a signal of resources, Journal of Conflict Resolution 38, 452-478. Perron, P., 1989, The great crash, the oil price shock, and the unit root hypothesis, Econometrica 57, 1361-1401. Sandler, T., 1992, Collective action: Theory and applications (University of Michigan Press, Ann Arbor, MI). Sandler, T., and H.E. Lapan, 1988, The calculus of dissent: An analysis of terrorists' choice of targets, Synthese 76, 245-261. Sandler, T., J.T. Tschirhart and J. Cauley, 1983, A theoretical analysis of transnational terrorism, American Political Science Review 77, 36-54. Scott, J.L., 1989, The role of information in the war on terrorism, unpublished dissertation (University of South Carolina, Columbia, SC). Scott, J.L., 1991, Reputation building in hostage taking incidents, Defence Economics 2, 209-218. Selten, R., 1988, A simple game model of kidnappings, in: R. Selten, ed., Models of strategic rationality (Kluwer Academic Publishing, Boston, MA) 77-93.

Ch. 9:

Terrorism: Theory and Applications

249

Shahin, W.N., and M.Q. Islam, 1992, Combating political hostage-taking: An alternative approach, Defence Economics 3, 321--327. US Department of State, 1986, Patterns of global terrorism (US Department of State, Washington, DC). US Department of State, 1993, Patterns of global terrorism (US Department of State, Washington, DC). Wilkinson, P., and A.M. Stewart, eds., 1987, Contemporary research on terrorism (Aberdeen University Press, Aberdeen).

Chapter 10

DEFENSE EXPENDITURE AND ECONOMIC GROWTH * RATI RAM Illinois State University

Contents Abstract Keywords

1. Introduction 2. A methodological overview of the existing research 2.1. Benoit's work and a list of the variations adopted by other scholars 2.2. Some methodological issues 2.2.1. Theoretical foundations for the econometric models used 2.2.2. Exogeneity of the regressors and the possible SEM bias 2.2.3. Cross-section versus individual-country estimates

3. A summary of the evidence from existing studies 4. Some thoughts on future research, and a few notes of caution 5. Concluding remarks References

252 252 253 254 254 258 258 261 264 266 269 271 271

* This work owes much to the research reported by Sandler and Hartley (1995, ch. 8) and Macnair et al. (1995). Todd Sandler gave many helpful suggestions at all stages of the project. Keith Hartley provided insightful comments on an earlier version. Research assistance was rendered by Martin Vedder. However, the author alone is responsible for all errors and deficiencies. Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

252

R. Ram

Abstract This chapter reviews the empirical research on the relation between defense spending and economic growth. A broad flavor of the approaches adopted by different scholars is provided through a list of selected studies on the topic since 1973 and an outline of the variations over Benoit's original work that have been explored in the literature. A discussion of three major methodological issues is then undertaken. The following section summarizes the main substantive results reported in various studies, and another section contains some thoughts on the future research on the subject. The chapter ends with a few concluding and summarizing observations.

Keywords Defense expenditure, economic growth, military outlays, investment, factor productivity, external effects, Granger causality, regressor exogeneity, fixed-effect modelling, sensitivity analysis, regression diagnostics

Ch. 10:

Defense Expenditure and Economic Growth

253

1. Introduction An assessment of the impact of defense (military) outlays on economic performance is important. Despite the recent slowdown in the rate of increase of defense spending, the level of defense expenditures continues to be high in most parts of the world. For example, the estimate by the US Arms Control and Disarmament Agency [US ACDA (1994, p. 47)] indicates that the total world military expenditure exceeded one trillion dollars in 19911. In the developing world, military expenditure constituted about 4.5 percent of GNP and over 18 percent of all central government expenditure. In a similar vein, the United Nations Development Programme [UNDP (1994, pp. 170171)] points out that military expenditures in the developing countries, which stood at about 143 percent of their combined education and health expenditures in 1960, were still of the order of 60 percent of their total expenditures in those two categories in 1990-91. Thus military outlays continue to be huge while economic growth is of major concern in almost all countries. Reflecting on the massive use of resources for defense, Sivard (1993, p. 5) wrote, "In a world spending over $600billion a year on military programs, over 1billion people lack basic health care, one adult in four is unable to read and write, one-fifth of the world population goes hungry every day". Similarly, she noted, "At a cost of less than half their military expenditures, the developing countries could have a package of basic health services and clinical care that would save 10 million lives a year". It is, of course, evident that, although determination of an optimal defense structure is extremely difficult in a complex multicountry international context, military expenditure is usually incurred for external defense, which is an important intrinsic objective. National defense is expected to provide external security, and should thus enhance economic growth in the long run. However, its immediate objective is not directly related to economic growth. Therefore, even if defense spending did have an adverse effect on current economic growth, there is no necessary implication that such expenditures are harmful or too high. The importance of exploring the nexus between defense spending and economic growth arises from two kinds of considerations 2. First, even if defense outlays are optimal in some sense, it is useful to assess the cost of defense in terms of reduction in current consumption, investment or output growth. It is obvious that military expenditures include many items that necessarily divert scarce resources from current consumption, investment and growth. Second, there are several possible positive external effects of defense

These estimates are apparently derived from conversion of local-currency data into US dollars on the basis of market exchange rates. If Purchasing-Power-Parity (PPP) exchange rates were used, the amount for the less-developed countries (LDCs) might be significantly larger. 2 The many possible conceptual linkages between defense outlays and economic performance are discussed at great length in several studies, and it seems needless to repeat these here. Good discussions are provided by Benoit (1973, 1978), Chan (1985), Deger (1986b), Grobar and Porter (1989), Sandler and Hartley (1995, ch. 8), and several other scholars.

254

R. Ram

outlays on the rest of the economy, especially in regard to infrastructure, technological progress, and human-capital formation, and it should be useful to investigate how the positive external benefits compare with the cost of diverting resources to uses that cannot contribute directly to current consumption, investment or economic growth. While the question is important, its theoretical formulation and empirical implementation have been difficult and somewhat inconclusive after over 20 years of fairly sustained research. The main objective of this chapter is to provide a broad flavor of the models used, statistical procedures employed, and the results claimed by various researchers. The focus of the paper is empirical. Therefore, the effort is largely directed toward stating the statistical or econometric formats in which the question has been investigated, the conclusions claimed by different researchers, and the major methodological issues that have been raised and may need to be addressed in future research.

2. A methodological overview of the existing research Table 1 contains a list of the studies that have investigated the relationship between defense expenditure and economic growth since 1973. Although the list is certainly not exhaustive, it gives a reasonable overview of what has been done and includes most of the approaches that have been used. 2.1. Benoit s work and a list of the variations adopted by other scholars Benoit's (1973, 1978) work provided a strong impetus for most research on the subject during the last two decades. Although his theoretical model (1973, pp. 8-16) tended to emphasize the adverse effects of defense expenditures on growth, the econometric estimates based on a cross-section of 44 less-developed countries (LDCs) for the period 1950-1965 showed that "countries with a heavy defense burden generally had the most rapid rate of growth, and those with the lowest defense burdens tended to show the lowest growth rates" (1978, p. 271). Even though several possibilities of a spurious relation were explored, he noted persistent evidence of a positive effect of the defense burden on growth, and then tried to rationalize the findings in terms of (i) the possibility that the alternative uses of resources diverted to defense might not be very productive in many LDCs, (ii) total resource-use might increase with a heavier defense burden, and (iii) the military sector may impart several positive externalities to the rest of the economy and these benefits might outweigh the adverse effects. Benoit's empirical results were so unexpected that much of the subsequent research on the subject has been directed toward an assessment of the validity of his results. In addition to a general criticism of his work, many variations over his model, sample and estimation method have been explored. The following list provides a flavor of most of these variations and thus of the broad patterns in the research on the subject.

Ch. 10:

255

Defense Expenditure and Economic Growth

Table 1 List of selected studies of the defense-growth relationship (arranged in the order of authors' names)a

1. 2.

Model/sample/period

Adams, Behrman and Boldin (1991) Alexander (1990)

No effect of defense spending on growth Feder-type 3-sector model, LDC sample, 1974-1986 No effect of defense spending on growth Feder-type 4-sector model, 9 DCs, 1974-1985 Small positive and significant effect of Feder-type 2-sector model, defense spending on growth USA, 1949-1989 Traditional (ad hoc) model, Positive and significant effect of defense spending on growth 44 LDCs, 1950-1965 Positive and significant effect of defense Traditional and Feder-type 2-sector models, 74 LDCs, spending on growth 1981-1989 No significant effect of defense spending on Traditional and Feder-type 2-sector models, 58 LDCs, growth 1960-1970 and 1970-1977 No causality between ME/GDP and growth Granger-causality tests, in most countries 55 LDCs, time-series data Positive direct effect of defense spending on Traditional SEM growth, but negative indirect and total (3-equation) model, effects 50 LDCs, 1965-1973 Traditional SEM Positive direct effect of defense spending on growth, but negative indirect and total (3-equation) model, effects 50 LDCs, 1965-1973 Demand-side traditional A generally negative effect of defense model, 69 countries, spending on growth 1952-1970 Defense spending has a positive effect on Benoit's sample and model with breakup in subsamples growth in the group of 24 "resource-abundant" countries, but the effect is negative for the group of 9 "resource-constrained" countries No significant overall effect of defense Feder-type 3-sector model, spending on growth USA, 1952-1988 No significant externality or Feder-type 3-sector model, factor-productivity effect of defense USA, 1952-1988 spending on growth Causality runs from growth to defense Granger-causality tests, spending; and there is little evidence of 57 LDCs, 1962-1977, annual data causality from defense spending to growth Traditional models with Little effect of defense spending on growth many variables, 65 LDCs, 1960-1980

4.

Atesoglu and Mueller (1990) Benoit (1973, 1978)

5.

Biswas (1993)

6.

Biswas and Ram (1986)

7.

Chowdhury (1991)

8.

Deger (1986a,b)

9.

Deger and Smith (1983)

10.

Faini, Annez and Taylor (1984)

11.

Frederiksen and Looney (1983)

12.

Huang and Mintz (1990) Huang and Mintz (1991)

3.

13.

Main conclusions

Reference

14.

Joerding (1986)

15.

Landau (1986)

continued on next page

256

R. Ram

Table 1, continued Reference 16.

17.

18.

19.

20. 21.

22.

23.

Model/sample/period

Landau (1993)

Traditional (ad hoc) growth model, 71 LDCs, 1969-1989, defense variable (ME/GDP) as well as its square used Lebovic and Ishaq Traditional 3-equation model like Deger-Smith (1987) (1983), 20 Middle-Eastern LDCs, 1973-1982 Harrod-Domar growth Lim (1983) model, 54 LDCs, 1965-1973 Macnair et al. (1995) Extended Feder-type model, 10 NATO allies, 1951-1988, pooled annual data, spillins from allies included Flexible accelerator Mintz and Huang (1990) investment model, USA Mintz and Stevenson Feder-type 3-sector model, 103 countries, around (1995) 1950-1985 Mueller and Feder-type model with Atesoglu (1993) technical change, USA, 1948-1990 Traditional and Feder-type Ram (1994) 2-sector models, 71 LDCs, 1965-1973, 1973 1980 and 1980-1990, considers lowand middle-income groups and interperiod pooling with fixed-effect formats

24.

Rasler and Thompson (1988)

25.

Scheetz (1991)

26.

Smith (1980)

Demand-side investment models, 19th and 20th century "systemic leaders" Deger-type 3-equation model, Argentina, Chile, Paraguay, Peru, 1969-1987, time-series/pooled Keynesian model of investment demand, 14 OECD countries, 1954-73, time-series/pooled

Main conclusions Defense spending has an initial positive effect on growth, but the effect turns negative at higher levels of the defense burden. No significant effect in a subsample of 47 countries Defense spending has a negative effect on growth

Defense spending has a negative effect on growth Positive effect of defense spending on growth

Defense spending lowered investment and thus growth No significant relation between defense spending and growth in most of the individual-country estimates Small positive effect of defense spending on growth No significant effect of defense spending on growth in Feder-type models in any period, but the parametric structure differs across the periods. In traditional models, one can find positive as well as negative effects of defense spending by using different proxies for defense and in different periods and groups of countries Some evidence of an adverse effect of defense spending on investment Defense spending has a negative effect on investment

Defense spending has a negative effect on investment

continued on next page

Ch. 10:

257

Defense Expenditure and Economic Growth

Table 1, continued Reference

Model/sample/period

Main conclusions

27.

Stewart (1991)

Keynesian demand model, LDCs, simulation-based

28.

Ward and Davis (1992)

Feder-type 3-sector model, USA, 1948-1996

29.

Ward et al. (1991)

Feder-type 3-sector model, India, 1950 1987

Defense spending is conducive to growth, but non-defense spending is even more conducive Defense spending has a net negative effect on growth, even though the externality effect is positive Defense spending has a positive effect on growth

This listing is based largely on the information contained in Table 8.2 of Sandler and Hartley (1995). They provide more information on most studies. Landau (1993, pp. 43-45) gives a review and critique of six "major existing studies". Those six studies are included in this list. a

(1) (2) (3) (4)

(5)

(6)

(7)

(8)

Benoit's growth equation has been augmented by the addition of other variable(s). One frequent addition is the growth of labor force (population). Cross-country samples that are broader or narrower than his have been studied. While Benoit's conclusion was based almost entirely on the averaged data for 1950-1965, quite a few other periods have been studied. Instead of a single-equation framework, simultaneous-equations models (SEMs) have been estimated. The additional equations are typically for investment and defense outlays, which are treated as endogenous. In the SEM context, the "direct" as well as the "total" effect of defense spending on growth can be assessed. Instead of assuming exogeneity (or endogeneity) of the defense variable in the growth equation, direction of Granger causality across the defense and the growth variables has been investigated. While Benoit's growth equation was somewhat ad hoc, efforts have been made to provide a more explicit economic foundation for the growth model. In these cases, proxies other than Benoit's (ratio of defense spending to GNP) for the defense variable have been considered. One possibility is to use the rate of growth of defense spending as the proxy for the defense variable. An explicit consideration is given in some studies to the channels through which defense spending may influence growth. Such channels include the effect of defense outlays on investment, technical change, human-capital formation, public expenditures on health and education, productivity differences between defense and civilian sectors, and unspecified external effects of the defense sector on the rest of the economy. Besides considering the entire set of LDCs, structural variations across some subsets have been examined. In particular, it has been considered whether the structure for "resource-constrained" cases is different from that for "resource-

258

R. Ram

abundant" countries. Similarly, the position for low-income and middle-income LDCs has been compared. (9) Instead of the usual supply-side modelling in terms of a growth equation, demand-side effects have been estimated, especially in terms of the effect on investment. In some cases, the models include both demand-side and supply-side considerations. (10) Some scholars have focussed on investigating the relationship in individual countries instead of using the typical cross-section format. (11) While most studies are based on LDC samples, some scholars have extended the analysis to developed countries (DCs), and have introduced interesting refinements that are made possible by better data availability for the DCs. Although these studies do not directly shed light on the defense-growth nexus in the LDCs, they provide a useful comparative perspective. Of course, as Table 1 would also indicate, the foregoing list does not consist of mutually exclusive categories; some studies have included several variations over Benoit's work. 2.2. Some methodological issues While the foregoing list provides a quick view of the general patterns in the defensegrowth research, there are some methodological issues that need a more detailed discussion because these issues seem to have a significant bearing on the kind of empirical results one gets. Although other aspects might also be considered important by some scholars, the discussion here is limited to three major issues. 2.2.1. Theoreticalfoundationsfor the econometric models used As Table 1 indicates, much of the research in the 1970s and the early 1980s used growth models that were largely ad hoc and lacked a well-defined economic foundation. Although often plausible in a general sense, these models represented a somewhat subjective statistical formulation of the issue and thus were beset with uncertainty about the variables to be included and the proper proxies for these variables. For example, Benoit's (1973, 1978) format did not include a labor variable, but did include a foreign-aid term, and the proxy for the defense variable was the ratio of military expenditure to GNP. Similarly, Landau (1986) included a large number of variables, and well-defined criteria to determine the econometric formulation seemed lacking. Although models of economic growth derived from simple neoclassical production functions had been used in several types of studies, an explicit modelling of the channels through which the focus variable generated the effect on output or growth was lacking until Feder (1983) suggested an interesting two-sector framework for studying the effect of exports on growth. Adapting Feder's methodology, Biswas and Ram (1986) provided a simple framework that (i) postulated two mechanisms through which the

Ch. 10: Defense Expenditure and Economic Growth

259

defense (military) sector may affect aggregate output and growth, and (ii) starting with simple neoclassical production functions, led to a tractable growth equation whose parameters included the effect of defense outlays on growth. Although details of such models are now well known, it may be useful to recapitulate the basic features of their simple two-sector framework. Assuming that the economy consists of two distinct sectors, that labor (L) and capital (K) are the two conventional inputs, and that the size of the defense sector (M) exerts an "externality" effect on the rest of the economy (C), production functions for the two sectors may be written as M = M(LM,KM),

C = C(Lc, Kc, M),

(1,2)

where subscripts refer to the (exogenous) assignment of inputs to each sector. Treating total inputs of labor and capital as given, and writing total output (Y) as the sum of M and C L = LM +Lc,

K = KM + Kc,

Y = M + C.

(3a,b,c)

It is further postulated that input productivity in the two sectors is not necessarily identical, and a relative factor-productivity differential of the following type exists MK ML -- =1 + 6, (4) CK CL where the subscripts now denote the partial derivatives of M and C with respect to the subscripted input. Equation (4) suggests that labor (and capital) may be relatively more productive in the defense sector than in the civilian sector, and the difference is indicated by the (positive) value of 6. The model identifies two channels through which defense outlays might have a favorable effect on aggregate output and growth. One would be a positive CM (OCI/M), which is the marginal externality effect of defense on the rest of the economy. The other is a positive 6, which implies higher input productivity in the defense sector, and thus an increase in total output for given resources as inputs are shifted to the more productive defense sector. If data on sectoral inputs are available, estimates of CM and 6 can be obtained from growth-rate versions of Equations (1) and (2). However, since information on sectoral inputs is scarce for most LDCs, it is necessary to reformulate the model in terms of aggregate inputs. Following Feder (1983), Biswas and Ram (1986) indicated that taking the total derivative of Equation (3c) and using the information from Equations (1), (2), (3a), (3b) and (4), one can derive an equation for growth of aggregate output in the following form = L+

II

-

t

+

\.

A M

+ CM)

,

(5)

where a dot over the variable denotes its rate of growth (e.g., =dYIY), fi is an elasticity-type parameter and equals CL(L/Y), a is the marginal product of capital in

R. Ram

260

sector C, and I (dK) denotes aggregate investment. Addition of a constant term (which could reflect neutral technical change) and a stochastic disturbance in Equation (5) yields an econometric specification in which the coefficient of MI(MIY) indicates the sum of the externality and the factor-productivity effects of defense on growth. If one wishes to obtain separate estimates of the externality and the factor-productivity effects, assuming that the externality parameter is not CM, but is CM(M/C) and is denoted by 0, Biswas and Ram (1986) state the following reformulation of Equation (5):

Y = fIL+

Y

I \


0. The agent's income, I, will be given by the price he receives from the principal minus the production cost. Assume that the agent's utility over income, I, and effort, e, is given by the separable function 5 u(I) - O(e).

(2)

Assume that u(O)= 0, (0)=0, and that the agent's reservation level of utility is zero. This means that the agent is completely indifferent between accepting a costreimbursement contract and not participating at all 6 . Thus the principal would always have the option of hiring the agent under a cost-reimbursement contract, if he wished. This assumption is not necessary for the results but helps make the model intuitively clear. Assume that the agent evaluates income lotteries by calculating expected utilities and comparing them and that u' > 0 and u" < 0, so that the agent prefers more income to less, and is risk neutral or risk averse. Assume that ¢'(e) > 0 and 0"(e) > 0, so that the agent dislikes effort and the marginal disutility of effort is weakly increasing. Although it is not necessary for the analysis, it will be convenient to assume that the principal is risk neutral and places an infinite value on consuming the good. Thus the principal's goal is to procure the good at minimum expected cost. The relationship can be viewed as unfolding in three steps. At the first step, nature chooses and 0. At the second step, the principal offers a contract p(c) to the agent and the agent decides whether or not to accept it. At the third 'step, if the agent has accepted the contract, he chooses an effort level, e, that results in a production cost, c. Government receives the good and pays the firm p(c). With respect to actors' information, assume that both actors know the entire structure of the relationship given by the functions G, F, ¢, and u. Thus the only information asymmetry at the time of contracting is that the agent has observed 0 and the principal has not.

5

The separability assumption is basically for expositional convenience. The same basic qualitative

insights apply to the nonseparable case, only the analysis is more complicated because the agent's attitudes towards risk and preferences over income are potentially affected by his effort choice. 6 Under a cost reimbursement contract, p(c) =c, so that I always equals zero. The agent will choose the lowest possible effort, i.e., e=O, under this contract. Thus the agent's expected level of utility equals zero.

WP. Rogerson

322

3.3. Pure moral hazard: The model The pure moral hazard model is created by assuming that there is no asymmetric cost uncertainty. Formally, assume that 0 assumes some particular value with probability one. Since 0 does not vary, one can simplify notation by suppressing 0 entirely. Let G(e) and g(E) denote the distribution and density of E, and let F(e, E) denote the function that determines cost as a function of e and E. Suppose that the principal offers the agent a cost-reimbursement contract 7. From the agent's standpoint, there is no benefit from lowering production costs. Therefore, he will reduce his effort to a minimum. Since effort levels cannot be contractually specified, the only way to induce the agent to exert effort to lower costs is to give him a financial incentive to lower costs. That is, the agent must be told that if he succeeds in lowering costs by one dollar, the price that he is paid will not decline by a full dollar. Of course, the extreme case would be to offer the agent a fixed-price contract. In this case, the agent would receive all the benefits from exerting effort, so would have no incentive at all to shirk. However, a fixed-price contract creates another potential problem. Under a fixed-price contract, the agent bears all of the risk of cost overruns and underruns. If the agent is risk averse (recall that, by assumption, the principal is risk neutral), the optimal contract from an insurance perspective would be a cost-reimbursement contract, where the principal bore all the risk. This, then, is the fundamental contracting issue in the pure moral hazard model. From an incentive perspective, the ideal contract is a fixed-price contract. From an insurance perspective, however, this contract places all of the risk on the agent. If the agent is risk neutral, this poses no problem and the optimal contract is a fixed-price contract. However, if the agent is risk averse, the ideal contract from an insurance perspective is a cost-reimbursement contract. Therefore, we cannot simultaneously accomplish both goals of creating ideal incentives and ideal insurance. Not surprisingly, the optimal contract in general turns out to be a cost sharing contract, which tries to achieve an optimal trade-off between the two competing goals. The outlines of a formal analysis to this problem will now be sketched [see Grossman and Hart (1983), Holmstrom (1979), Rogerson (1985), and Shavell (1979) for more complete treatments]. It turns out that the analysis of this model is somewhat more convenient if we work directly with the distribution of cost induced by effort. Let F(c, e) denote the distribution function of c given that the agent exerts the effort level e. This is defined by F(c, e) = Pr{F(e, e) < c}.

(3)

Letf(c, e) denote the density function. The principal calculates the optimal contract to offer the agent by choosing a contract to minimize his expected payments subject to 7 The term cost-reimbursement contract will be used to describe the contract given by p(c)= c. The term fixed-price contract will be used to describe a contract of the form p(c) = k, where k is a constant.

Ch. 12: Incentive Models of the Defense Procurement Process

323

the constraint that the agent is willing to accept the contract. In reality, the principal chooses only the contract and predicts the agent's effort choice. However, as is typical in this sort of problem, the simplest way of formally describing the problem is to view the principal as choosing both the contract and effort subject to the extra constraint that the agent is actually willing to choose the effort that the principal specifies. The optimal contract is therefore the solution to the following program: Minimize p( ), e

J

f p(c)f(c, e) dc

u(p(c) - c)f(c, e) dc -

e C argmax

0(e)

u(p(c) - c)f(c,

subject to:

(4)

>0

(5)

e) dc - (

(6)

Equation (4) is the principal's payment to the agent. The principal chooses a contract p and effort level e subject to the constraints that the agent receives his reservation level of expected utility [Equation (5)], and that the agent will actually choose the level of effort specified by the principal [Equation (6)]. The standard method of solving this program is to replace constraint (6) by the first-order condition for the agent's effort choice problem,

I

u(p(c) - c)fe(c, e) dc - ¢'(e) = 0.

(7)

It is straightforward to use control theory to characterize the solution to the resulting program. Readers unfamiliar with control theory can derive the same result using standard Lagrangian techniques by assuming that c takes on n possible values, {C1 ,

,

n}.

In general the optimal contract is neither a fixed-price contract nor a costreimbursement contract. Rather, price depends on cost in a complex way. 3.4. Pure moral hazard: Discussion What light does this model shed on procurement problems and practices? At the level of logical foundations, it obviously makes a significant contribution. It clearly and explicitly describes the basic idea that incentive contracting may involve a trade-off between effort inducement and risk allocation and thus supplies a theory of incentive contracting. In order to be useful for normative or positive analysis, however, a theory of optimal contracts must describe how measurable features of a contract ought to vary with measurable features of the contracting environment. It turns out that this theory has been less successful on this front. Although the problem itself is very simple to state, formal derivation of optimal contracts is

324

W.P Rogerson

surprisingly complex and delicate. Furthermore, other than the statement that the optimal contract will, in general, be neither a fixed-price nor a cost-reimbursement contract, almost no general statements can be made. The nature of the optimal contract varies tremendously depending upon the precise functional forms of the utility function and distribution function for E. For normative purposes, the problem this creates is that the precise nature of the optimal contract is highly dependent on features of the contracting environment that government may be unsure about. For positive purposes, the problem is that the theory does not generate testable predictions. Therefore, it is probably fair to say that the major value of this model to date has been to help clarify the underlying incentive issues rather than to explain specific contracting phenomena. In reality, most contracts used by DoD are linear in cost. That is, they are of the form p = a + ic,

(8)

where a is a positive constant and fi is a constant between 0 and 1. If equals 0, it is a fixed-price contract. If fi equals 1, it is a cost-reimbursement contract. If 1 is between 0 and 1, the contract is a cost-sharing contract, and fi is usually spoken of as the share of the risk borne by DoD. Similarly, (1 - ,) is usually spoken of as the share of the risk borne by the firm. It is difficult to find simple examples where the optimal contract in the formal model is linear. This is unfortunate, but not a critical problem. In all likelihood, linear contracts are used because they are simple and information does not exist to support more finely-tuned calculations. By exogenously restricting oneself to linear contracts (or, by restricting oneself to a two-outcome case, where costs can be "low" or "high" 8), fairly well-behaved solutions can be calculated. In particular, it is possible to demonstrate a very intuitive result that will be referred to as testable result #1 (TR #1). Testable Result #1: Suppose that contracts are linear. Then under the optimal contract, the share of risk borne by government, , is always between zero and one. The share of risk borne by government grows larger as cost uncertainty increases. Government, does, of course, always choose 13 between zero and one. The interesting question is whether it can be shown that 13increases with cost uncertainty. Even attempting to relate this simple prediction to real behavior turns out to be problematic. It is a well-accepted stylized fact of the procurement process, that DoD bears successively lower fractions of risk as a program matures. At the early R&D phase, cost-reimbursement contracts are typically used. In development, the firm may accept a small share of the risk. In initial production, the firm will accept a large share of the risk and then, after initial production, fixed-price contracts will typically be used. A natural

8 In this case, the change in price divided by the change in cost as costs go from low to high can be

viewed as the slope, fl.

Ch. 12: Incentive Models of the Defense ProcurementProcess

325

hypothesis is that this behavior is explained by the simple moral hazard model. Namely, cost uncertainty declines over the life of the program and this is reflected by decreases in the value of 3. However, there may be another, more important, factor explaining this behavior. In the formal moral hazard model, the product is completely well-defined. Thus the government and the firm are able to sign a contract at the outset that completely specifies the nature of the product that will be delivered in an objectively verifiable fashion. The extent to which this assumption is satisfied varies enormously and systematically over the life of a program. In the R&D phase, government may be completely unable to specify many important aspects of what it wants. By the time of the third or fourth annual production lot, government may be able to describe quite precisely what it expects. Therefore, government's use of cost type contracts for R&D may be largely due to the fact that no product can be objectively defined at the date of contract signing. Fixed-price contracts would simply create an incentive to reduce the quality of the R&D supplied. In summary, in order to argue that the simple moral hazard model explains why ,3 varies across a group of contracts, it would be important to be able to control for the extent to which the product can be objectively described at the date of contract signing. This has not been done. 3.5. Pure self selection: The model Economic analyses of defense procurement dating back to Scherer (1964) have emphasized the fact that information about cost is asymmetric at the contracting stage and that this fact plays a key role in shaping the procurement process. That is, at the time of contract signing the firm is often much better informed than government about the likely cost of performing the contract. The pure self selection model focusses attention on this factor by assuming that the only cost uncertainty in the model is asymmetric cost uncertainty. That is, it is assumed that equals some fixed value with probability one. Since E is now constant, we can simplify notation by suppressing it entirely. Let G(O) and g(O) denote the distribution and density of 0 and let F(e, 0) denote the function determining cost as a function of e and 0. From the agent's perspective, there is no uncertainty in this model. Therefore without loss of generality, assume that u(I)=I. In the pure moral hazard model, the basic contracting issue was the trade-off between effort inducement and risk allocation. In the pure self selection model, the basic contracting issue is different. The principal is constrained to offer a contract that all types will accept 9 . One way of doing this is to offer a cost-reimbursement contract. By construction, every type is just willing to accept such a contract; however, the problem

If we relaxed this assumption, the same logic applies given whatever fraction of types the principal wishes to attract. 9

326

W

Rogerson

with this contract is that no effort is induced. Under a cost-reimbursement contract, all types will exert zero effort. The other extreme would be to offer a fixed-price contract. In this case, every type would choose the efficient level of effort. However, the problem with this contract is that, in order to guarantee that every type will accept the contract, the principal must offer a fixed price that is high enough that the highest type, max, (i.e., the highest-cost type), will accept it. Thus all other types are left with a positive surplus. Therefore, a cost-reimbursement contract enables the principal to extract all of the joint surplus generated by the relationship, but the level of joint surplus generated is low because no effort is induced. A fixed-price contract generates a high level of joint surplus because effort is induced, but the principal extracts only a small share of the joint surplus. Therefore, the optimal contract will in general be some form of cost sharing contract that trades off between these two factors. In the pure moral hazard model, the trade-off was between effort inducement (which is best accomplished by a fixed-price contract) and risk allocation (which is best accomplished by a cost-reimbursement contract.) The optimal contract is in general some sort of cost-sharing contract that seeks an optimal compromise between these two competing concerns. In the pure self selection model, rent extraction takes the place of risk allocation. Therefore in both cases, the basic intuition flowing from the models is quite similar. When cost uncertainty is high, government should choose contracts which have government bear more of the risk. In the pure moral hazard model, this is because extra symmetric uncertainty creates greater risk that the principal can more easily bear. In the pure self selection model, this is because extra asymmetric uncertainty makes it harder for the principal to capture rent through offering a fixed price, so cost sharing become more desirable. Although the basic "bottom line" qualitative insight is fairly similar, the method of formal analysis for the pure self selection model is actually quite different. This is because the principal does not know the agent's type. Therefore when the principal considers a contract, he must calculate whether each type will participate and what effort level each type will choose. Thus each type "self selects" into a price/cost pair from the contract. In general, each type will choose a different price/cost pair, and thus keeping track of and analyzing how different types self select creates an extra level of complexity in these models. It turns out that essentially the same mathematical structure applies to self selection models of a wide variety of phenomena, including models of price discrimination and quality choice by a monopolist, models of taxation and its incentive effects, principal agent models where the output of the agent is units of some physical commodity such as wheat, and principal agent models where the output of the agent is units of cost reduction. Of course this last interpretation is the one considered by this chapter. The basic mathematical structure of this model was developed by authors considering other interpretations [Guesnerie and Laffont (1984), Mirrlees (1985), Musa and Rosen (1978), Baron and Myerson (1982), Sappington (1983)]. Laffont and Tirole (1986) and McAfee and McMillan (1987a) were the first to apply this type of model to the case considered by this chapter.

Ch. 12:

Incentive Models of the Defense ProcurementProcess

327

The formal analysis of this model will now be outlined. Readers interested in further details should consult the references listed above or Laffont and Tirole (1993). It is convenient to reformulate the model slightly in order to analyze it. Because the agent faces no uncertainty, the agent can be viewed as choosing c directly instead of choosing effort. Let z(c, 0) denote the amount of effort needed to produce cost c when the agent's type is 0. (This is simply the inverse of F.) Then define 6(c, 0) to be the disutility of choosing c which is given by 6(c, 0) = ¢(z(c, 0)).

(9)

Rather than work directly with effort, it is more convenient to simply view the agent as choosing c at a disutility cost of 6(c, 0). Thus the effort variable is suppressed entirely0. Assumptions made previously that effort decreases cost and effort induces disutility (Fe < 0, ' > 0), and that higher values of 0 mean that costs are higher (Fo > 0) immediately imply that disutility both decreases in cost and increases in 0, i.e., 6c < 0

and

65 > 0.

(10)

The regularity assumption will be made that 6,0 < 0.

(11)

This means that the marginal disutility from lowering costs becomes greater as 0 increases. Thus increases in 0 mean that both the disutility from lowering costs and the marginal disutility from lowering costs increase. This is equivalent to the property that the efficient or first-best cost level increases in 0, which is also equivalent to the property that if the agent was given a fixed-price contract, the agent would choose a higher cost if his type was higher. This is of course a very intuitive requirement. We interpret higher values of 0 as meaning that the type is "worse". The assumption in Equation (11) states that as 0 becomes higher, we would observe the same contract resulting in higher costs. Therefore, Equation (11) simply formalizes what we intuitively mean by a worse-cost situation, and is a very natural assumption. This is an important point, because the main qualitative results depend on this assumption. For example, if 6 was additively separable in c and 0, this would imply that all types would choose the same cost. In this case, a fixed-price contract would

' All of the early analyses of the pure self selection problem followed this practice of suppressing the effort variable. This led to the inaccurate impression that the distinction between the pure moral hazard and pure self selection models is whether an effort variable exists. This is not true. In the pure self selection model, the agent makes a decision that affects his unobservable utility and this can always be thought of as an effort choice. The distinction between the two models is whether the uncertainty over cost is symmetric or asymmetric.

328

W.2P? Rogerson

extract all the rent from every type and this would be the optimal contract for the principal to offer. If 6 c0 < 0, then worse (i.e., higher) types choose higher costs. Thus, offering a price high enough to attract the high types leaves the low types earning rent. Recall from the above discussion of the intuition of the pure self selection problem, that this was the reason that a non-trivial contracting problem exists. One more regularity assumption will be made. For any type 0, the total disutility of the agent is given by the sum of production cost, c, and the disutility of effort, 6(c, 0), c + 6(c, 0).

(12)

It will be assumed that CSc > 0

(13)

and that for every order condition

there exists a unique value of c, cF(0), which satisfies the first-

I + c(CF(),0) = 0.

(14)

Therefore cF(0) is the unique first-best or efficient-cost choice for type 0, in the sense that it minimizes total disutility of c. Total differentiation of Equation (14) yields cFt(0)

=

o

c'

(15)

By Equations (11) and (13), this is positive, i.e., higher types should optimally produce at higher cost. Laffont and Tirole (1986) consider an example satisfying properties (10)-(14). The example is created by assuming that (e, 0) = 0- e

(16)

and that (e) is strictly increasing and strictly convex. The agent's type is therefore the value of cost if no effort is exerted. Exertion of effort results in cost reductions below 0. If F is defined by Eqation (16), then, 6(c, 0) = O(0 - c).

(17)

It is easy to verify that 6 satisfies the assumptions stated in Equations (10)-(14). The principal's contractual design problem can now be formally described. Government considers a contract, p(c). Given this contract, the principal predicts for each type of firm, 0, whether that type will participate, and what cost that type will choose to produce at, if it does participate. Among the contracts that cause all types

Ch. 12:

Incentive Models of the Defense Procurement Process

329

to participate , the principal picks the one that minimizes his expected cost. Let c(O) denote the cost chosen by type 0. Then, formally, the principal can be viewed as solving the following program: Minimize c( ), p( )

J p(c(O)) g(O) dO

c(O) G argmax{p(c) -

-

subject to

(e, 0)}

p(c(O)) - c(0)- 6(c(0), 0) > 0

for every 0,

for every 0.

(18) (19) (20)

This problem can be transformed into a well-behaved optimal control problem under various regularity assumptions. The main result is to showthat a unique solution c*(0) and p*(O) exists with the following three characteristics: (i) Higher types choose strictly higher costs, c*'(0) > 0.

(21)

(ii) The lowest type, 0i, chooses the first-best cost. All other types choose a cost strictly greater than the first best, C*(Omin) = CF( 0

,min)

c*(O) > cF(0)

for every 0

(22) (in,

Omax]-

(23)

(iii) The highest type, Omax, earns zero expected utility. All other types earn positive expected utility and expected utility is strictly decreasing in type: V*(0max) = 0,

V*'(0) < 0,

(24,25)

where *(0) is defined by v*(0) = p*(c*(O)) - c*(0) - 6(c*(0), 0).

(26)

None of these characteristics directly describe the nature of p*(c). However, based on these characteristics, it is straightforward to do so. When a type 0 chooses c, it solves the problem of maximizing p(c) - c - 6(c, 0).

(27)

The first-order condition for this problem is p'(c) - I - 8c(c, 0) = 0.

(28)

By comparing Equations (28) and (14), it is immediately apparent that the slope of p(c) at the c chosen by type 0 is sufficient to tell us whether the cost chosen by type " Recall that, to simplify the exposition, it is assumed that the principal places infinite value on the good and therefore definitely wants to consume it. If we relax this assumption, then in general the principal would rationally plan to make offers that would be rejected some of the time. However, nothing significant changes in the analysis.

W.P Rogerson

330 p*(c) P

I Cmin

C( 01)

C*(0

2)

C max

Figure 1.

0 is less than, equal to, or greater than the first best. In particular, c* (0) is greater than the first best iff p'(c*(O)) > 0 and equal to the first best iff p'(c*(O)) = O. That is, c*(O) < cF(0)

> p'(c*(O))

>

0.

(29)

Let [Cmii, cmax] be the interval of costs over which p*(c) is defined. The cost cmi is 0 chosen by type 0 min and the cost Cmax is chosen by max. Then it follows immediately that (30)

P'(Cmin) = 0 p'(c) > 0

for c

E (Cmin, Cmax].

(31)

That is, the slope of p(c) equals 0 at Cmin and is strictly positive elsewhere. This is the only property of the function p(c) that follows from the general analysis. A fixed-price contract is one where p'(c) is identically equal to zero. Therefore, the general result is that it is optimal for the principal to offer a cost-sharing contract, in the sense that prices increase with cost. Since p*'(c) equals zero at Cmin and is strictly positive for larger values of c, the function p*(c) cannot be globally concave. However, it is possible that it may be globally convex. This possibility is illustrated in Figure 1. The contract p*(c) is the convex function defined over [cmin,Cmax]. Each type of agent chooses a (p,c) combination off this curve to maximize his expected utility. Utility is increasing in price and decreasing in cost. Let 01 and 02 be two types with 01 < 02. The best indifference curve which can be achieved by each type is drawn in. The key geometric

Ch. 12: Incentive Models of the Defense Procurement Process

331

property illustrated by this graph is that higher types have flatter indifference curves 12 This follows from assumption (11) and is often referred to as the single crossing condition. This property is why higher types locate a tangency at a higher cost. An interesting result of Laffont and Tirole (1986) and McAfee and McMillan (1987a) is that if p*(c) is convex as drawn, it would be possible to replace the contract p*(c) with a menu of linear contracts. Consider the menu of linear contracts formed by all tangent lines to p*(c) over the interval [Cmin, Cmaxl] Suppose the principal initially offered the agent this menu. It is geometrically clear that each type 0 would choose the tangent line at c*(O) and would then choose the cost c*(O). Thus, the nonlinear contract p*(c) could be implemented by offering a menu of linear contracts. Furthermore, it is straightforward to show that one can add back some symmetric additive noise to the model, , and the menu of linear contracts continues to be optimal 13 This result is interesting for three reasons. First, it provides a formal theoretical model consistent with the observation of only linear contracts. Second, it extends the pure self selection model to cases including a symmetric noise term, . Third, it suggests the general idea that government should structure bargaining so that firms claiming to be lower cost select themselves into contracts where they bear a higher share of the risk. This qualitative insight is potentially quite applicable in a wide variety of situations. It turns out that it is difficult to identify broad classes of cases where p*(c) can be shown to be globally convex. In the case considered by Laffont and Tirole (1986), for example, one must assume that the third derivative of the disutility function is positive in order to guarantee that p*(c) is globally convex [Rogerson (1987)]. However, while this is unfortunate, it is not a "fatal flaw" of the model. In some over-all sense, p*(c) must be "generally convex" in the sense that its slope equals zero at Cmin and is positive elsewhere. Thus in many practical applications, the intuition that lower cost types should self select themselves into contracts where they bear a greater responsibility for cost overruns and underruns may turn out to be correct. More research is required on this point. Another interesting line of research explores how auditing should optimally be used in the pure self selection model [Baron and Besanko (1984)]. 3.6. Pure self selection: Discussion Just as for the pure moral hazard model, the pure self selection model obviously makes a significant contribution at the level of logical foundations. It clearly and 12 In Figure 1, 1, (2) is an indifference curve belonging to a low (high) type. The single crossing property means that 12 is flatter than I, at the point where they intersect. If this property holds for all indifference curves, the tangency for type 1 must occur to the left of the tangency for type 2. 13 See Rogerson (1987) for a thorough discussion of this geometric interpretation of the menu of linear contracts result.

332

W.P Rogerson

explicitly describes the basic idea that incentive contracting may involve a trade-off between effort inducement and rent extraction, and thus supplies a theory of incentive contracting. Unfortunately, the same basic criticism also applies to this model. Namely, the technical analysis is surprisingly delicate and complex, and the general nature of the optimal contract varies tremendously depending upon the precise functional forms one assumes. Just as for the pure moral hazard model, it is possible to derive various versions of TR #1. Once again, the problem is that it is difficult to actually test this result. Furthermore, since both theories predict the same result, it does not help distinguish between them. Five additional remarks should be noted about the pure self selection model. First, the basic idea underlying the pure self selection model is more complex and subtle than the basic idea underlying the moral hazard model. Thus, it is probably fair to say that the basic idea was not well understood by economists even on an intuitive basis prior to the creation of the formal models. Second, it may be that the pure self selection model is focussing on the more fundamental or important phenomenon. Asymmetric information at the time of contracting may turn out to be a more important explanation for various contracting practices than risk aversion. If all firms were risk neutral, it is not clear that contracting practices and institutions would necessarily be much different. However, if no firms had private information, there would be clearly be large differences. Third, the "linear menus" result for the pure self selection model possibly creates another testable result which will be called testable result number two (TR #2). Under the optimal menu, higher-type firms select a linear contract with a higher value of / and earn lower rents. The term "rent" refers to the agent's total utility including the unobservable part. Therefore, if we observed the same contracting problem played out n times, where nature drew an independent identically distributed (iid) value of 0 each time, we would observe a negative correlation between P/3and total utility. This is TR #2. Testable Result #2: Firms operating under linear contracts with higher values of /3 should earn lower rents. It is a well-accepted stylized fact that the measured ex-post rate of accounting profit earned by firms on fixed-price contracts is greater than the measure ex-post rate of accounting profit earned by firms on cost-reimbursement contracts. One might interpret this as supporting TR #2; however, this argument is problematic. The model predicts a correlation between unmeasurable total utility, p(c)-c-6(c, 0), and /3; it does not predict a relationship between measured profit, p(c)-c, and P/. Since the key feature of the model that generates all of the results is that there is unmeasurable utility, abstracting away from this for the purposes of creating testable predictions is clearly inappropriate. This stylized fact is most straightforwardly explained by some sort of risk/expected return relationship. Therefore, although testable results may flow from the linear menus analysis, TR #2 is not such a result.

Ch. 12:

Incentive Models of the Defense ProcurementProcess

333

Fourth, it may be that the linear menus analysis may help government to create better negotiation strategies with firms by explicitly offering them menu choices [see Reichelstein (1992) for a very interesting analysis along these lines]. Fifth, the SPP proceeds under the relatively artificial bargaining structure that government makes a single take-it-or-leave-it offer. In the moral hazard model, where information is symmetric at the time of contracting, this assumption is no great cause for concern; but, in the pure self selection model, where information is asymmetric, this assumption is much less palatable. Institutions that affect negotiations and the government's ability to commit may play a much more important role in determining contracting outcomes in this case. In support of this point, it is interesting to note that the pure self selection model has had its greatest success in explaining behavior, precisely in situations where commitment is not an issue. This is in models of a monopoly practicing price discrimination in a market with large number of anonymous consumers. A firm facing a large anonymous market of consumers truly does make a take-it-or-leave-it offer when it publishes a price schedule 14 3.7. The general model A general model would allow both types of cost uncertainty to exist simultaneously, and would allow the agent to be risk averse. This model would contain elements of both the pure moral hazard and pure self selection model, but has not yet been extensively analyzed [see Baron and Besanko (1987b, 1988) and McAfee and McMillan (1986)]. One way that some of these papers make progress is by exogenously restricting themselves to linear contracts and making other simplifying assumptions. This is a good strategy given the complexities of the general model. 4. The simple procurement problem with multiple agents A natural way to generalize the SPP of the preceding section would be to assume that there are initially n agents capable of producing the commodity, where n is greater than one. The principal only needs to hire one agent. If there is no asymmetric information, nothing fundamental changes. The principal would face n different pure moral hazard problems. He would calculate the optimal contract for each problem and the resulting expected payment. He would then simply choose the agent to minimize his expected wage payment. If there is asymmetric information, however, the introduction of multiple potential agents creates a fundamental change. In this case, the principal can conduct some type of auction in an attempt to extract more rent from the agents. Thus the question of how the principal should design auctions to produce the lowest expected payment arises. 14 The extra structure created by observable demand curves also make the models potentially more empirically relevant.

334

WP Rogerson

The auctions literature has concentrated on the case where it is assumed that the principal does not measure any production costs at all, so only fixed-price contracts are possible. Of course, if there is a single agent this problem is trivial from an analytic point of view 15. The pure self selection literature in a sense creates an interesting problem out of the single-agent case by assuming that some costs can be measured and that an effort variable exists which affects both measurable and unmeasurable cost. This creates a trade-off between rent extraction and effort inducement. When there are multiple agents, a non-trivial problem exists even when costs cannot be measured. There are many different ways to run a fixed-price auction for producing a commodity. Which will minimize the principal's expected procurement cost and why? The auctions literature has been one of the most successful literatures in information economics. A large number of sharp and interesting results have been derived. This literature is too voluminous to attempt to survey here [see the excellent survey by McAfee and McMillan (1987b)]. This literature is mentioned simply to place it in the context of the models described in this chapter. The auctions literature simplifies the SPP by assuming that costs are unobservable so that only fixed-price contracts can be signed. Thus the issue of effort becomes irrelevant and we can simply assume that each agent has an unobservable cost of production drawn from some distribution. Anton and Yao (1989, 1992) have analyzed a number of models of "share auctions" where two firms bid prices for various shares of the total buy. This is a particularly interesting type of auction from the perspective of defense procurement because it resembles the DoD practice of dual sourcing [Anton and Yao (1990)]. One of the main insights from these models is that, in a one-period model, it would generally be highly undesirable to conduct a share auction. The principal would almost always be better off by conducting a winner-take-all auction, because it induces more competitive behavior. Therefore, the value of dual sourcing must be that it preserves competition for the future. A number of papers have considered two-agent self selection models where extra structure is added by distinguishing between a low-cost incumbent and a higher-cost potential entrant [Anton and Yao (1987), Demski, Sappington and Spiller (1987), Stole (1994)]. The general flavor of the results is that it may be desirable for government to commit ex-ante to use the potential entrant in circumstances where it would be cheaper to use the incumbent, ex-post. This threat can force the incumbent to bid lower. Recently, a number of papers have begun to generalize the general n-person auction model to consider cases where there is unobservable effort, measurable cost, unmeasurable disutility, and effort affects both measurable cost and unmeasurable disutility [Laffont and Tirole (1987), McAfee and McMillan (1987a), Riordan and 15 In

the formal model of the previous section, it was assumed that the principal places infinite value

on consuming the good. Therefore the principal would offer a fixed price equal to the lowest price the agent would accept with certainty. In a more realistic case where the principal places a finite value on consuming the good, the optimal price would generally be one where the agent refuses the contract with positive probability. Calculation of the optimal price is analytically straightforward.

Ch. 12: Incentive Models of the Defense Procurement Process

335

Sappington (1987)]. These papers generalize the pure self selection model to the case where there are n agents. The general flavor of the result is that, in cases where it would have been optimal for the principal to offer a menu of linear contracts with a single agent, it is now optimal for the principal to announce a menu of linear contracts and a preference ordering over them, and then to have agents bid a linear contract from the menu. The agent bidding the most preferred linear contract wins the auction. An interesting technical feature of these models is that there is often a sort of "separability" between the effort-inducement problem and the rent-extraction problem. Adding more agents does not change the relationship between type and effort for the winning bidder. It only changes the amount of rent extracted. Another interesting line of research has investigated how government should run auctions where firms bid both price and quality, in situations where quality can vary and some aspects of it can be objectively measured [Che (1993)]. In such cases government announces a "scoring rule" that it uses to compare bids with different prices and qualities.

5. Research and development As discussed in Section 2, innovation is a major product of the procurement process. Producing innovation quite clearly requires unobservable "effort" on the part of defense firms. Therefore government faces an incentive problem. A natural first approach to modeling this would be to attempt to use the models developed in Sections 3 and 4. In the SPP of Sections 3 and 4, the agent exerted unobservable effort to produce units of "output". We chose to interpret units of output as units of "cost reduction", but the same mathematics would still apply if we chose another interpretation for output. In fact, most of the incentives literature does choose other interpretations. Why then, don't the models of Sections 3 and 4 describe the government's problem of inducing firms to perform R&D? To be more concrete, suppose that the value of consuming a weapons program to government is given by v and the cost of producing the program is given by c. Both of these are affected by R&D work at the design stage. Finding a better design increases v and finding a cheaper way to produce the same design decreases c. The net value or surplus from producing the program is v- c, which I will denote by s. Why can't the models of Sections 3 and 4 be applied by assuming that the agent exerts effort to produce units of s? The answer is that the models of Sections 3 and 4 assume that units of output can be objectively measured for purposes of contracting upon. For the R&D case, this would amount to assuming that government can describe all possible designs in an objectively verifiable fashion, and specify the reward that the firm will be paid depending upon what design it comes up with. Generally speaking this is not possible. There are too many possible designs and no way of objectively measuring their surplus to government. Of course the ex-post cost of production can be measured

336

W.P Rogerson

to some extent, and this is one of the factors that determines s. However, so long as the performance of the weapon cannot be completely specified at the beginning of the design phase, attempting to write contracts which rewarded the firm if the expost production cost turned out to be low would create serious incentive problems. Namely, a firm could always reduce production costs by settling on a design which offered less performance. Therefore, from a modelling standpoint, the distinguishing feature of the R&D phase is that it is a principal agent problem where the agent's output cannot be objectively measured for purposes of contracting upon. Therefore the solution of signing explicit incentive contracts as modelled in Sections 3 and 4 cannot be followed. In purely commercial markets where there are generally fairly large numbers of consumers for a product, incentives for innovation are created by the patent system. A firm which invents a new idea is allowed to be the owner of it for a number of years and charge "whatever the market will bear". Thus the prospect of earning monopoly profit at the production stage gives firms the incentive to innovate. In the case of defense procurement, the policy of telling firms to invest their own money in R&D and then to charge government "whatever the market will bear" would create severe problems. Firms would generally be unwilling to invest in any R&D because there would be no guarantee of recovering any investment at the production stage, due to the hold-up problem. At a minimum, allowing prices at the production stage to be determined by "free and unfettered" bargaining would create enormous uncertainties and bargaining costs. As discussed in Section 2, one of government's main responses to this has been to directly fund a large share of R&D itself. That is, unlike most purely commercial markets, government directly buys the intermediate product R&D, as well as buying the products that incorporate this R&D. However, this does not create any incentives for firms to exert effort at the R&D stage. How does government create incentives for firms to exert effort at the design phase? The method that government appears to use is a "prize" system. Suppose that multiple firms are creating different designs at the R&D phase. Although there is no objectively verifiable signal of the value of s that each firm comes up with, there is an objectively verifiable signal of which firm comes up with the highest s. That is, government will choose the design with the highest value of s to enter production. Therefore, government could create incentives for innovation by committing to a set of regulatory institutions that guaranteed that firms which generate ideas good enough to be adopted by government would receive prizes in the form of economic profit on the production phase of the system. Furthermore, if profit was awarded approximately as a percentage of cost, this might also tend to award larger prizes to more important innovations, at least in a rough sense. It appears that government has followed this strategy [see Rogerson (1989, 1994) for a more extensive discussion]. The basic idea that offering a prize to the winner of a contest will create incentives for contestants to work hard, is simple and clear enough that perhaps no formal modelling is required to elucidate it. However, three more complex and

Ch. 12:

Incentive Models of the Defense ProcurementProcess

337

subtle issues related to this do require modelling, and have been considered in the incentives literature. First, R&D is not simply a one-shot process where a number of firms simultaneously spend money and then each receives a design. R&D is a sequential process, where one can begin with many possible design approaches and narrow the field gradually as intermediate results become known. Taylor (1995) explicitly analyzes how a prize system should be organized in light of the fact that R&D is sequential. Second, the main focus of the incentives literature has been on the role of asymmetric information at the sole source selection phase in allowing government to make credible commitments to give prizes for innovation. The issue is "How does government credibly commit to award prizes?" Suppose, as was done in Sections 3 and 4, that the principal has all the bargaining power at the source selection phase, in the sense that the principal can make a take-it-or-leave-it offer. If there is no asymmetric information at the source selection phase then, in the absence of any other commitments on the part of the principal, the agents know that they will earn no surplus at the source selection phase regardless of how good their design is. Working backwards, they will therefore have no incentive to perform R&D at the design phase. However, now suppose that there is asymmetric information at the source selection phase. Then agents will expect to earn positive rents at the source selection phase, and in many plausible situations agents will perceive that their expected rents are increasing in R&D effort, so this creates an incentive for R&D. Therefore asymmetric information at the source selection phase may be desirable because it creates a credible commitment on the part of government not to expropriate all the benefits of firms' R&D. This then suggests that government might be able to influence the amount and quality of R&D that firms perform by committing to institutional procedures that affect its ability to extract rent at the source selection phase. In particular, by committing to procedures that reduce competition or reduce the amount of information that government has available at the source selection phase, it may be that firms' incentives to perform R&D would increase [see Laffont and Tirole (1988b), Piccione and Tan (1993), Riordan and Sappington (1989), Tan (1992), Tirole (1986a), and Sappington (1986) for models that explore aspects of this issue]. The third subject investigated by formal models is the "franchise bidding" aspect of the problem that is created by the fact that multiple firms typically perform R&D and then a sole source is selected for production. For a moment, ignore the effects of source selection procedures and pricing procedures in the sole source phase on firms' incentives to perform R&D - this was described above. Even if firms' R&D efforts can be directly controlled by government, the fact that multiple firms bid for the sole source franchise creates some interesting questions. In particular, it is reasonable to expect that much of the profit earned at the sole source production stage will be bid away at the source selection phase. If this is true, there is no reason to necessarily insist on using regulatory institutions at the sole source production stage that extract all the rent at that stage. In particular, one could allow rents to be earned at this stage if this would create some other benefit. Two possible benefits that have

338

W.P Rogerson

been discussed are creating better incentives for the sole source to lower production costs [Riordan (1993), Rogerson (1994)] and creating incentives for more competitive bidding at the source selection phase [Bower and Osband (1991)]. Bower and Osband's argument is particularly clever. They build a model where allowing higher profit rates on sole source production contracts actually reduces over-all expected contractor profits and increases government's welfare. The intuition for their result is that the differential subsidization created by profit policy at the sole source phase (highercost producers receive higher markups) encourages more aggressive competition at the source selection phase. This paper very strikingly illustrates the general idea that the procurement stages are interrelated and that policy instruments applied at one stage may have quite surprising and unintended consequences for behavior at other stages. Many more insights remain to be gleaned from procurement models that explicitly model procurement as occurring over three stages - design, source selection, and production - and consider the interlinkages between stages.

6. Multiple periods of production, regulatory lag, and the ratchet effect An important facet of many real procurement situations that the SPP abstracts away from is the fact that DoD often purchases the same product or closely related versions of the same product over a number of years. Because the design continues to evolve in difficult-to-anticipate ways, and because of uncertainty regarding future demand, longterm contracts cannot be signed. In such situations, even if the firm is initially better informed than government, we might expect much of the firm's private information to be revealed over time as costs of production for each year are observed. One focus of the theoretical literature has been to highlight the crucial role of commitment ability in such situations. Consider the pure self selection model with a single agent from Section 3. Now, however, assume that the procurement will occur twice and that the agent's type remains the same over both procurements. Such a model has been considered by Laffont and Tirole (1988a). Let p*(c) denote the optimal contract to offer in the one-period version. It is straightforward to show that, if the principal can make a long-term commitment, the best policy for the principal is to commit to offerp*(c) both times. However, now consider what happens if the principal cannot make a binding commitment to the period-2 contact. Suppose, for the sake of argument, that the principal initially promises to offer p* (c) both periods and the agent initially believes him. Then, in period 1, the agent will choose a cost according to his type as described in Section 3. In particular, the principal will be able to precisely infer the agent's type, since higher types choose strictly higher costs. This means that, at the start of period 2, the principal knows the agent's type. Since the principal now knows the agent's type, he will of course no longer want to offer p*(c). Realizing this at stage 1, the agent will alter his selection of c to take into account the effects of this on the principal's period-2 choice. Laffont and Tirole (1988a) solve for equilibrium behavior in this model. The general flavor of the results is that the inability to commit

Ch. 12: Incentive Models of the Defense Procurement Process

339

may make the principal significantly worse off, and that the formal analysis becomes much more complicated than for the one-period case, because the incentive constraints are not as well-behaved. Laffont and Tirole (1990) and Baron and Besanko (1987a) consider the effect of allowing government certain types of commitment ability. In reality, the most overwhelming stylized fact related to the repeat nature of procurement is that a type of simple regulatory lag behavior seems to occur [Rogerson (1994)]. Each annual procurement occurs under a fixed-price procurement and this year's price is determined largely by last year's cost. This creates a distinct incentive for firms to find ways to lower costs. Namely, if they find a way to lower cost, they can "keep" the savings on the current contract. It is not completely clear how to square this observation with the theoretical models described above. Should regulatory lag be modelled as the result of government commitment or is it what we would observe when government can make no commitments? The topic of using formal incentive models to normatively and positively analyze the observed behavior of regulatory lag is an interesting subject for future research.

7. Incentives within government 16 7.1. Introduction To focus attention on the incentive problem between government and defense firms, the discussion to this point has implicitly viewed government as a single rational actor. But DoD's budget is the size of a small country's GNP and decisionmaking must necessarily be decentralized among thousands of individuals with potentially conflicting objectives. The way that government organizes its decisionmaking procedures, and the incentives faced by individual acting in their own selfinterest, will affect how these decisions are made. Thus, government faces a massive planning, organizational, and incentive problem of its own, independent of any problems it faces with defense firms [Baron (1993), Leitzel (1993)]. Three underlying factors characterize the defense decision-making process. First, since final decision-making authority rests with Congress and the President, there is no single rational actor with well-defined preferences in control of defense procurement. However, this chapter will set aside political aspects of the procurement process [Mayer (1990)], and instead describe the organizational problem that remains even if a single rational principal - call it "Congress" - pursued the goal of adequate defense at minimum cost. Second, the defense decision making process is enormous and complex. This means that Congress must delegate substantial decision-making authority to lower levels. Within DoD, this delegation has taken the form of individual military services

16

This section draws from Rogerson (1994).

340

W.P Rogerson

essentially choosing their own weapons, while the Office of the Secretary of Defense and Congress play a supervisory role. Third, many aspects of DoD's performance like "preparedness", and whether it was accomplished at minimum cost - are very difficult to measure objectively. In the absence of convenient bottom-line measures of defense performance, it becomes harder to delegate, and more necessary for Congress to control and manage the process. Thus, while the technical complexity and sheer magnitude of the procurement process make delegation of decision-making authority necessary and desirable, the lack of objective overall performance measures limits its value. An analogy to the organizational problem of a profit-maximizing firm is illuminating. When a board of directors of a large firm delegates authority to management, it can monitor the results by using (admittedly imperfect) objective performance measures like profits and stock-market value. But there is no analog to profit or stock-market value in the Congress/DoD relationship, which is what makes the delegation problem more difficult. 7.2. Two-leuel models There are two different types of principal agent models that can be used to capture different aspects of the problem. One type of model is the normal two-level principal agent model where the principal is interpreted as Congress and the agents are interpreted to be the military services or other agencies within DoD. This type of model therefore completely abstracts away from the incentive problem between government and defense firms. One interesting conclusion that has emerged from this type of model is that the military services may derive some of their power to affect policy through strategically controlling program decisions made early in a program's life. A theme of the literature on government decision-making dating back to Niskanen (1971) is that a goal of government bureaucrats may be to maximize the size of their own budget. It is well-accepted that military services act as though this is one of their primary goals [Fox (1988), McNaugher (1989), Stubbing (1986)]. In Niskanen's (1971) theory, bureaucrats are assumed to be able to maximize their budget by exercising a sort of monopoly power. He assumes that an individual government bureau is the only possible supplier of a particular product and is able to make Congress a take-it-or-leave-it offer. Rather than offer Congress the first-best quantity (where marginal cost equals marginal benefit), the bureau offers Congress the largest quantity that Congress would prefer to having nothing (where total cost equals total benefit). Faced with the offered quantity or nothing, Congress chooses the offered quantity. The bureau then receives a budget sufficient to produce this quantity. A problem with this theory is explaining why bureaucrats have monopoly power. Why can't Congress simply choose the quantity it wants to? In the case of defense procurement, a different, though somewhat related, theory which does not exhibit this problem can be created to explain how bureaucrats are able to increase their budget. The key fact which this theory is based on is that defense programs are executed over

Ch. 12: Incentive Models of the Defense ProcurementProcess

341

many years. Congress exercises budget authority and thus decides how many units to purchase each year. However, many relatively complex and technical decisions made early in the program's life affect the marginal benefits and marginal costs that Congress will face when it makes annual quantity decisions. In many cases, these early decisions are delegated to the military services and Congress has a difficult time evaluating the technical merits or consequences of these decisions. This means the military services can strategically manipulate Congress's future decisions through their decisions made early in the program's life which affect future marginal benefits and marginal costs. In this theory, the source of the bureau's power to influence its budget is its informational advantage. Congress can still be viewed as "moving first" or as designing the over-all mechanism that is played. At the outset, Congress has two choices. If it delegates certain decisions to the military, better decisions will be made in a variety of technical dimensions because of the military's greater technical expertise. However, the military may also purposely distort its decisions to alter future marginal benefits and marginal costs of the program and thus manipulate Congress's future decisions over quantities to procure. Congress must weigh the benefits and costs of delegation and then determine how much authority to delegate. Two examples of such decisions will now be described. The first example concerns the trade-off between quality and quantity. Many institutional analyses of defense procurement have argued that the same expenditures would produce a more effective defense if larger numbers of less elaborate and less technically sophisticated weapons were purchased [Gansler (1980, pp. 15-21), Peck and Scherer (1962, ch. 13), Stubbing (1986, ch. 8)]. This outcome can also be seen as the result of an agency problem between Congress and the military. Rogerson (1990) describes a simple framework where the military chooses the quality of weapons, and then Congress chooses quantity. In this model, the military's goal is to maximize the benefits of military preparedness, while Congress's goal is to maximize the benefits of military preparedness minus the costs. The major result is that when quality and quantity are not good substitutes, that the military will purposely choose a quality higher than the efficient level. The intuition behind this result is that the military can increase military preparedness by purposely increasing quality above the efficient level because this induces a relatively small decrease in Congress's quantity choice. An interesting feature of this model is that the military is assumed to be as good an agent as one could realistically hope for. The military agrees with Congress's definition of military preparedness and makes a good-faith effort to maximize this. If given a fixed budget, the military in this model would always choose quality and quantity to maximize military preparedness. The distortion arises because the military is able to manipulate Congress into increasing the budget by increasing quality. In this model, budget-maximizing behavior on the part of defense bureaucrats is caused by the relatively idealistic goal of maximizing the bureau's production of social benefits. A second example concerns the decision of what scale of production facility to build. In a model related to the above quality vs. quantity model, Rogerson (1991b) shows that the military can induce Congress to increase the quantity purchased of a weapon

342

W.P Rogerson

by purposely selecting a production technology of too high a scale. The underlying insight is that higher-scale technologies exhibit higher fixed costs, but lower marginal costs. Thus, so long as Congress does not cancel the program, it will buy more when faced with a higher scale technology. This model explains the well-accepted stylized fact that weapons production systematically occurs in production facilities designed to produce at much higher rates than they are actually operated at [CBO (1987), Gansler (1989), Rogerson (1991a)]. 7.3. Three-level models The other type of model to build is a three-level model with a principal, a supervisor, and an agent. In this model, the principal is interpreted as Congress, the supervisor as the military service, and the agent as the defense contractor. The focus of this type of model is to explicitly analyze the hierarchical nature of control within defense procurement. The study of such models is in its infancy, and a large literature on such models addressing the structure of control and authority in organizations is likely to develop over the next few years. One of the most interesting qualitative insights explored by this literature is the idea that control problems within government may affect the type of incentive arrangements that government is able to offer defense firms. In particular, Congress may limit the type of contractual arrangements that the military services have with defense contractors, because Congress is worried about being strategically manipulated by the military services. Kelman (1990) argues that this explains why Congress delegates so little discretion to DoD officials. Laffont and Tirole (1991) and Marshall, Meurer and Richard (1994a) have built formal models that exhibit this phenomenon. Marshall, Meurer and Richard (1991, 1994b) have analyzed models where the bid protest process is modelled as a device to help Congress monitor and control defense bureaucrats. An interesting issue that arises in this type of model, is modelling the possibility of collusion between the supervisor and the agent. See Tirole (1986b) and Campbell (1994) for some approaches to this. Tirole (1986b) calculates the optimal "collusionproof" contract in a model where collusion occurs through legally binding contracts between the supervisor and employee. Campbell (1994) expands upon this line of research by modelling collusion as occurring through non-binding agreements.

8. Conclusion Formal models of incentive contracting have contributed to our understanding of defense procurement by forcing us to think more explicitly (and thus more clearly) about the nature of the incentive and information problems that affect defense procurement. Much of the modelling thus far has been devoted to the necessary first step of considering fairly simple stylized principal agent problems. Research that builds upon these efforts, to develop models with more institutional and empirical content and

Ch. 12: Incentive Models of the Defense Procurement Process

343

models which capture aspects of the multi-stage nature of defense procurement, will surely yield many new insights over the next decade.

References Anton, J., and D.A. Yao, 1987, Second sourcing and the experience curve: Price competition in defense procurement, Rand Journal of Economics 18, 57-76. Anton, J.J., and D.A. Yao, 1989, Split awards, procurement and innovation, Rand Journal of Economics 20, 538-552. Anton, J.J., and D.A. Yao, 1990, Measuring the effectiveness of competition in defense procurement: a survey of the empirical literature, Journal of Policy Analysis and Management 9, 60-79. Anton, J.J., and D.A. Yao, 1992, Coordination in split award auctions, Quarterly Journal of Economics 107, 681-707. Baron, D.P, 1989, Design of regulatory mechanisms and institutions, in: R. Schmalensee and R.D Willig, eds., Handbook of industrial organization, Volume II (North-Holland, Amsterdam) 1347-1447. Baron, D.P., 1993, Defense procurement: politics, management, and incentives, in: J. Leitzel and J. Tirole, eds., Incentives and procurement contracting (Westview Press, Boulder, CO) 7-24. Baron, D.P., and D.E. Besanko, 1984, Regulation, asymmetric information, and auditing, Rand Journal of Economics 15, 447-470. Baron, D.P, and D.E. Besanko, 1987a, Commitment and fairness in a dynamic regulatory relationship, Review of Economic Studies 54, 413-436. Baron, D.P., and DE. Besanko, 1987b, Monitoring, moral hazard, asymmetric information, and risk sharing in procurement contracting, Rand Journal of Economics 18, 509-532. Baron, D.P., and D.E. Besanko, 1988, Monitoring of performance in organizational contracting: The case of defense procurement, Scandinavian Journal of Economics 90, 329-356. Baron, D.P., and R.B. Myerson, 1982, Regulating a monopolist with unknown costs, Econometrica 50, 911-930. Besanko, DE., and D.E.M. Sappington, 1987, Designing regulatory policy with limited information (Harwood Academic Press, London). Bower, A.G., and K. Osband, 1991, When more is less: Defense profit policy in a competitive environment, Rand Journal of Economics 22, 107-119. Burnett, WB., and EM. Scherer, 1990, The weapons industry, in: W. Adams, ed., The structure of American industry (Macmillan, New York) 289-317. Caillaud, B., R. Guesnerie, P. Rey and J. Tirole, 1988, Government intervention in production and incentives theory: A review of recent contributions, Rand Journal of Economics 19, 1-26. Campbell, Colin, 1994, Collusion within hierarchies, mimeograph (Northwestern University, Evanston, IL). CBO (US Congressional Budget Office), 1987, Effects of weapons procurement stretchouts on costs and schedules (Congressional Budget Office, Washington, DC). Che, Y.-K., 1993, Design competition through multidimensional auctions, Rand Journal of Economics 24, 668-680. Crocker, K.J., and K.J. Reynolds, 1993, The efficiency of incomplete contracts: An empirical analysis of Air Force engine procurement, Rand Journal of Economics 24, 126-146. Demski, J.S., D.E.M. Sappington and P.T. Spiller, 1987, Managing supplier switching, Rand Journal of Economics 18, 77-97. Fox, J.R., 1988, The Defense management challenge (Harvard Business School Press, Boston, MA). Gansler, J.S., 1980, The defense industry (MIT Press, Cambridge). Gansler, J.S., 1989, Affording defense (MIT Press, Cambridge). Goldberg, VP., 1976, Regulation and administered contracts, Bell Journal of Economics 7, 426-448.

344

We Rogerson

Grossman, S.J., and O.D. Hart, 1983, An analysis of the principal-agent problem, Econometrica 51, 7-46. Guesnerie, R., and J.-J. Laffont, 1984, A complete solution to a class of principal-agent problems with an application to the control of a self-managed firm, Journal or Public Economics 25, 329-369. Holmstrom, B., 1979, Moral hazard and observability, Bell Journal of Economics 19, 74-91. Kelman, S., 1990, Procurement and public management (AEI Press, Washington, DC). Kovacic, W.E., 1991, Commitment in regulation: Defense contracting and extensions to price caps, Journal of Regulatory Economics 3, 219-240. Laffont, J.-J., and J. Tirole, 1986, Using cost observation to regulate firms, Journal of Political Economy 94, 614-641. Laffont, J.-J., and J. Tirole, 1987, Auctioning incentive contracts, Journal of Political Economy 95, 921-937. Laffont, J.-J., and J. Tirole, 1988a, The dynamics of incentive contracts, Econometrica 56, 1153-1175. Laffont, J.-J., and J. Tirole, 1988b, Repeated auctions of incentive contracts, investment and bidding parity, Rand Journal of Economics 19, 516-537. Laffont, J.-J., and J. Tirole, 1990, Adverse selection and renegotiation in procurement, Review of Economic Studies 75, 597-626. Laffont, J.-J., and J. Tirole, 1991, The politics of government decisionmaking: A theory of regulatory capture, Quarterly Journal of Economics 106, 1089-1127. Laffont, J.-J., and J. Tirole, 1993, A theory of incentives in procurement and regulation (MIT Press, Cambridge, MA). Leitzel, J., 1993, The choice of what to procure, in: J. Leitzel and J. Tirole, eds., Incentives in procurement contracting (Westview Press, Boulder, CO) 91-99. Lichtenberg, F.R., 1988, The private R&D investment response to federal design and technical competitions, American Economic Review 78, 550-559. Marshall, R.C., M.J. Meurer and J.-F. Richard, 1991, The private attorney general meets public contract law: Procurement oversight by protest, Hofstra Law Review 20, 1-71. Marshall, R.C., M.J. Meurer and J.-F. Richard, 1994a, Unwarranted bidder exclusions and biased bid evaluations in federal procurements, mimeograph (Duke University). Marshall, R.C., M.J. Meurer and J.-E Richard, 1994b, Curbing agency problems in the procurement process by protest oversight, Rand Journal of Economics 25, 297-318. Mayer, K.R., 1990, Patterns of congressional influence in defense spending, in: R. Higgs, ed., Arms politics and the economy (Holmes and Meier, New York) 202-235. McAfee, R.P., and J. McMillan, 1986, Bidding for contracts: A principal agent analysis, Rand Journal of Economics 17, 326 338. McAfee, R.P., and J. McMillan, 1987a, Competition for agency contracts, Rand Journal of Economics 18, 296-307. McAfee, R.P., and J. McMillan, 1987b, Auctions and bidding, Journal of Economic Literature 25, 699-738. McNaugher, T.L., 1989, New weapons, old politics: America's military procurement muddle (Brookings Institution, Washington, DC). Mirrlees, J., 1985, Optimal taxation, in: K. Arrow and M. Intriligator, eds., Handbook of Mathematical Economics, vol. 3 (North-Holland, Amsterdam). Monteverde, K., and D.J. Teece, 1982, Supplier switching costs and vertical integration in the automobile industry, Bell Journal of Economics 13, 206-213. Musa, M., and S. Rosen, 1978, Monopoly and product quality, Journal of Economic Theory 18, 301317. Niskanen, W.A., 1971, Bureaucracy and representative government (Adeline-Atherton, Chicago, IL). Peck, M.J., and F.M. Scherer, 1962, The weapons acquisition process: An economic analysis (Graduate School of Business, Harvard University, Cambridge, MA).

Ch. 12: Incentive Models of the Defense Procurement Process

345

Piccione, M., and G. Tan, 1993, Cost-reducing investment, incentives, and procurement contracts, mimeograph (University of British Columbia). Reichelstein, S., 1992, Constructing incentive schemes for government contracts: An application of agency theory, The Accounting Review 67, 712-731. Riordan, M.H., 1993, Incentives for cost reduction in defense procurement, in: J. Leitzel and J. Tirole, eds., Incentives in Procurement Contracting (Westview Press, Boulder, CO) 135-146. Riordan, M.H., and D.E.M. Sappington, 1987, Awarding monopoly franchises, American Economic Review 77, 375-387. Riordan, M.H., and D.E.M. Sappington, 1989, Second sourcing, Rand Journal of Economics 20, 41-58. Rogerson, W.P., 1985, The first-order approach to principal agent problems, Econometrica 53, 13571367. Rogerson, W.P., 1987, On the optimality of menus of linear contracts, Discussion Paper 714R (Northwestern University Center for Mathematical Studies in Economics and Management Sciences). Rogerson, W.P., 1989, Profit regulation of defense contractors and prizes for innovation, Journal of Political Economy 97, 1284-1305. Rogerson, W.P., 1990, Quality vs. quantity in military procurement, American Economic Review 80, 83-92. Rogerson, W.P., 1991a, Excess capacity in weapons production: An empirical analysis, Defence Economics 2, 235-250. Rogerson, W.P., 1991b, Incentives, the budgetary process, and inefficiently low production rates in defense procurement, Defence Economics 3, 1-18. Rogerson, W.P., 1991c, Profit regulation of defense contractors and prizes for innovation, R-3635-PA&E (RAND Corporation, Santa Monica, CA). Rogerson, W.P., 1992a, Overhead allocation and incentives for cost minimization in defense procurement, R-4013-PA&E (RAND Corporation, Santa Monica, CA). Rogerson, W.P., 1992b, Overhead allocation and incentives for cost minimization in defense procurement, The Accounting Review 67, 671-690. Rogerson, W.P., 1992c, An economics framework for analyzing DoD profit policy, R-3860-PA&E (RAND Corporation, Santa Monica, CA). Rogerson, W.P., 1994, Economic incentives and the defense procurement process, Journal of Economic Perspectives 8, 65-90. Sandler, T., and K. Hartley, 1995, The economics of defense (Cambridge University Press, Cambridge). Sappington, D.E.M., 1983, Limited liability contracts between pricipal and agent, Journal of Economic Theory 29, 1-21. Sappington, D.E.M., 1986, Commitment to regulatory bureaucracy, Information Economics and Policy 2, 243-258. Sappington, D.E.M., 1991, Incentives in principal-agent relationships, Journal of Economic Perspectives 5, 45-66. Scherer, EM., 1964, The weapons acquisition process: Economic incentives (Graduate School of Business, Harvard University, Cambridge, MA). Shavell, S., 1979, Risk sharing and incentives in the principal agent relationship, Bell Journal of Economics 10, 55-73. Stole, L., 1994, Information expropriation and moral hazard in optimal second source auctions, Journal of Public Economics 56, 463-484. Stubbing, R.A., 1986, The defense game (Harper and Row, New York). Tan, G., 1992, Entry and R&D costs in procurement contracting, Journal of Economic Theory 58, 41-60. Taylor, C.R., 1995, Digging for golden carrots: An analysis of research tournaments, American Economic Review 85, forthcoming. Tirole, J., 1986a, Procurement and renegotiation, Journal of Political Economy 94, 235-259.

346

WI. Rogerson

Tirole, J., 1986b, Hierarchies and bureaucracies, Journal of Law Economics and Organization 2, 181214. Williamson, O.E., 1985, The economic institutions of capitalism (The Free Press, New York).

Chapter 13

THE ECONOMICS OF MILITARY MANPOWER* JOHN T. WARNER Clemson University BETH J. ASCH RAND, Santa Monica, CA

Contents Abstract Keywords 1. Introduction 1.1. Some definitions 1.2. Summary statistics

2. The supply of military manpower 2.1. Initial enlistment supply 2.1.1. Theoretical model of the enlistment process 2.1.2. Empirical models 2.1.3. Empirical estimates 2.2. Retention 2.2.1. Theoretical models of retention 2.2.2. Empirical studies

3. Demand for military manpower 3.1. Framework 3.2. Studies of personnel productivity 3.3. Force mix issues

4. Global procurement issue: to draft or not to draft? 4.1. Economic theory of the draft 4.2. Other issues

5. The structure of pay 5.1. Stylized facts about military compensation 5.2. Theory

6. Force management issues 6.1. Women in the military

348 348 349 349 350 352 353 354 355 357 360 360 363 367 367 368 371 373 373 379 380 380 381 386 386

* We would like to thank Carl Dahlman, Judy Fernandez, Curt Gilroy, Glenn Gotz, Keith Hartley, Jim Hosek, Bernie Rostker, and especially Todd Sandler, for comments on previous drafts.

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

348

JT Warner and B.J Asch

6.2. Reserve force management issues

387

7. Civilian returns to military service 8. Summary References

390 393 394

Abstract The USA and other countries spend a significant portion of their defense budgets on personnel. Effective management of military forces requires an understanding of the economics of military manpower. Over the past three decades economists have produced a substantial body of research about the subject. This chapter distills this literature for a general audience. Topics surveyed include the supply of personnel, personnel productivity and the demand for personnel, procurement by conscription versus voluntary means, the structure of pay, the use of women and reservists, and the civilian return to military training and experience. It also points to directions for future research.

Keywords military, manpower, enlistment supply, retention, draft, conscription, volunteer force, military compensation, reserves, military women, military productivity, cost effectiveness and manpower

Ch. 13:

The Economics of Military Manpower

349

1. Introduction The USA and other countries spend a significant portion of their defense budgets on military manpower and employ large numbers of personnel. In 1991, for example, the USA spent $84 billion on 1.9 million active duty military personnel and 1.1 million in ready reserve components. It is important that policymakers have an understanding of the economics of military manpower. Effective management of military personnel requires the understanding of a number of questions including the responsiveness of personnel supply to pay and other policy tools, the optimal amount of training, the optimal experience and quality mixes, and the proper mix of pay and other incentives. Because these are all questions in applied labor economics, the study of military manpower should prove interesting to a wider audience than just those responsible for policymaking. Economists have produced a substantial literature about military manpower, but as Sherwin Rosen noted in the 1986 Handbook of Labor Economics, much of their work is unavailable in the formal literature. The purpose of this chapter is to distill this literature for a general audience. The emphasis is on the USA, but studies based on other countries are selectively cited where appropriate. This survey proceeds as follows. The remainder of this introduction makes some preliminary definitions and presents some summary statistics about the size and composition of military forces in the USA and elsewhere. Section 2 examines the supply of military manpower. Section 3 examines the demand for military manpower by reviewing studies relating the characteristics of the force to measures of productivity. Implications for the optimal experience and quality mixes of the force are considered. Section 4 then reexamines the global manpower procurement issue: should military manpower be procured by a draft or by voluntary means? Section 5 addresses questions about the structure of compensation that are now only beginning to be studied. Section 6 looks at two contemporary force management issues: women in the military, and the management of the reserve forces and their relationship to the active forces. Section 7 examines the civilian return to military training and experience. Finally, Section 8 concludes the chapter. 1.1. Some definitions The military personnel systems of the USA and its major allies share many common attributes. All countries distinguish between officers and enlisted personnel. Officers are usually college graduates and have leadership and command responsibilities. Enlisted personnel usually have less education and assume the responsibilities of executing the orders of the officer corps. Within each of these designations there is a fixed rank or paygrade structure consisting of between 7 and 10 ranks. Rank structures are hierarchical, with large numbers of personnel in low ranks and declining numbers in, and lower rates of promotion to, the upper ranks. In the USA, for example, only about 10 percent of the enlisted forces occupy the top three enlisted ranks and less than 15 percent of the officer forces occupy ranks of Lieutenant Colonel and above. There

350

JT Warner and B.J Asch

is a heavy reliance on up-or-out rules to control the rank and experience distribution of the force (especially in the USA) as well as a general lack of lateral entry. Because of the lateral entry constraint, senior personnel must be "grown" from the ranks of junior personnel. In a closed military personnel system, the steady-state distribution of the force (F) and the requirement for new enlistments (E) can be derived as follows. Let ct be the continuation rate at year of service (YOS) t. Then the survival from the initial entry point to year t is st = cj. The total steady-state force is _j=

F =E + slE +

+STE=

1+

t) E.

The steady-state enlistment (or accession) requirement is therefore E=-

F

1+ T=1 St

Manyears per accession (y) are I + ETI= st. Obviously, the factors that raise continuation rates raise y and reduce E. It is important to distinguish between the total force (F) and the ready force (M). Some fraction of force is located in the training establishment and therefore not available for immediate deployment. To illustrate this idea, assume that the entire first year is spent in training and that the trainers have more than one year of experience. Let k denote the required ratio of trainers to trainees. Then the ready force is M=F-k

(-,

T

E - E =( -k)

st

t=1

)

stE. t=1

M rises with continuation rates and with the number of enlistments, but falls with the required trainer/trainee ratio. The expression for M is easily modified to account for variations in the length of training. 1.2. Summary statistics Table I shows the size of the active forces and the reserve force in the USA and its major allies. The USA, Germany, and the UK maintain the largest standing military forces, with the USA by far the largest. Germany and the UK have a somewhat greater reliance on reserve forces than does the US. All countries must enlist large numbers of personnel each year to maintain the force. In Fiscal Year (FY) 1991, the US active forces accessed 200 000 new enlisted personnel and 20 000 officers. Although with the end of the Cold War force levels are declining in the USA and elsewhere, the USA

Ch. 13:

351

The Economics of Military Manpower Table 1 Force levels for USA and major allies'

Country USA Germany UK Canada Australia a

Ground forces 884 287 216 20 29

Naval force

Air force

558 31 59 12 15

Active total

502 90 7 21 19

1944 408 275 78 63

Reserve force 1806 697 349 37 29

Numbers given in thousands. Table 2 YOS distribution of US forces, FY 1990'

YOS

Army

Navy

Air Force

Marine Corps

Enlisted 0-4 5-10 11-20 21-30

50.2 26.3 21.2 2.3

49.9 27.2 20.3 2.7

35.8 31.3 27.6 5.3

59.3 23.4 15.3 1.8

Officers 0-5

35.6

45.7

30.4

37.5

6-11

29.5

26.7

34.4

26.5

12-20 21-30

25.1 9.8

19.5 7.3

25.3 9.2

28.2 7.5

a

YOS, years of service; all values given in percent.

will still have an active force of over 1.4 million when its drawdown is completed in FY 1997. Table 2 shows the year of service (YOS) distribution of US enlisted and officer forces at the end of FY 1990. The table reveals that the bulk of US personnel are found in the low YOS, with the Marine Corps having the largest percentage of inexperienced personnel and the Air Force the smallest. Less than 10 percent of US forces have more than 20 years of experience. It is evident from the table that retention and the average experience level are higher among officers than enlisted personnel. Not revealed in the data is the substantial increase in the experience of the enlisted forces since the elimination of the US draft in 1973. Army enlisted personnel with more than 4 YOS (hereafter denoted "careerists") made up 32.6 percent of the Army enlisted force in FY 1974. By 1990 the percentage had grown to 49.8 percent. The Navy careerist percentage rose from 40.4 percent in 1974 to 50.1 percent in 1990. In the Air Force the percentage increased from 51.3 percent to 74.2. The Marine Corps careerist growth has been smaller, 32.9 percent to 40.7 percent. The change

352

JT: Warner and B.J Asch

in the officer experience distribution since the end of the draft has been much less dramatic. A major theme of the research cited below has been whether these increases in experience have led to a more productive force and a more efficient force.

2. The supply of military manpower Studies abound of both the decision to join the military (the enlistment decision) and the decision to remain after the expiration of the initial or subsequent terms of service (the reenlistment decision). The decision to enlist or reenlist is conveniently described using standard occupational choice theory [Rosen (1986)]. Suppose that there are two sectors of the economy - the military sector and the civilian sector. Individuals deciding whether to join the military must compare the pay and non-pecuniary benefits available in each sector. Military service is often arduous and involves exposure to risk and loss of life. But military service offers many non-pecuniary advantages over non-military employment - pride of service to one's country, the opportunity for travel, and more stable employment. Assume that individuals are able to weigh the many non-pecuniary aspects of employment in each sector and place an overall value on the non-pecuniaries associated with employment in each sector, rM and rC, respectively. Let WM denote the military wage and WC denote the civilian wage. Then the utility of joining the military is UM = WM + rM while the utility of remaining in the civilian sector is UC = WC + TC. Individuals join the military only if UM > UC, which implies that WM - WC > T - M. Simply stated, individuals join only if the pay differential (WM - WC) exceeds their net preference for civilian life, r = TC - TM The distribution of t over the relevant population determines the level of the supply curve for military service and its elasticity with respect to pay. Suppose that r M and r c follow a bivariate normal distribution over the eligible population with mean = btC i _ PM and variance 2 = a2 + 2 - 2 p7M oc. A positive value of Mindicates that, on average, the eligible population values the non-pecuniary aspects of civilian life more than the non-pecuniary aspects of military life. The shape of the supply curve is determined by the variance of the net preference factor , ( 2 ). Suppose that a 2 = 0. Such will be the case if (a) everyone has identical preferences for the two sectors or (b) p = 1 and M= c. If 2 = 0, then everyone has the net preference r equal to , in which case no-one will join if WM - WC < or WM < WC + . In this latter formulation, WC +,u is the individual's opportunity cost of serving in the military, i.e., the wage foregone plus the difference value he or she places on the non-pecuniary aspects of life in the two sectors. The military wage must exceed the civilian wage by the net preference factor before anyone will join. But if WM > WC+ then everyone will want to join, so that the supply curve is perfectly elastic at WM = WC + t. In the case of homogeneous preferences the parameter is the compensating wage differential required to make all individuals indifferent between military and civilian service. It follows that the more heterogeneous preferences are (the larger is 2 ), the less elastic the supply curve will be. Figure 1 sketches two different supply curves on

Ch. 13:

353

The Economics of Military Manpower c2

0

.5N

Number of Enlistments

Figure 1. Enlistment supply curves based on small and large variations in preferences for military service.

the assumption that the net preference factor is normally distributed. N is the eligible population. When preferences are heterogeneous, only individuals for whom WM - WC > r are paid in excess of that required to induce them to join (or stay) and are said to earn economic rents (payments in excess of opportunity cost). The intercept of each supply curve is WC + r., where Tmin is the net preference of the person who is the least averse to military service. If tastes are normally distributed, then the cumulative density function of tastes, and hence the supply curve, will be S-shaped, as illustrated in Figure 1. When tastes are normal, enlistments are less responsive to pay when pay is very low or very high than when pay is in the middle range of possible values. If, instead, is uniformly distributed across the population with pdff(T) = /(m,,,ax in), then the supply curve will be linear over the range WC + Tmin to WC + max and a given pay change will have the same effect on enlistments at every point on the supply curve. 2.1. Initial enlistment supply The occupational choice framework provides a starting point for thinking about initial enlistment supply, but it needs to be expanded to understand more fully enlistment behavior. The first factor to consider is the role of human capital development in the initial enlistment decision. Individuals may join because they want to acquire skills that will be useful to them later. Individuals will be more willing to join if the skills are transferable than otherwise. Thus, the supply of potential enlistees depends upon skill transferability such that the supply curve of potential recruits to military skills that provide transferable training will lie to the right of the supply curve to the skills that provide military-specific training. The military will have to offer a higher wage in the latter skills to attract the same number of enlistments. Another human capital factor

354

J5T: Warner and B.J Asch

is the availability of educational benefits. In the USA since World War II, individuals who successfully serve for some period of time qualify for post-service educational benefits. A second consideration is that individuals do not make their enlistment decision in a vacuum. The decision to enlist is shaped by environmental influences such as the advice of family and friends and societal attitudes towards military service. Orvis and Gahart (1989) analyzed the Youth Attitude Tracking Survey, a DoD-sponsored survey of US youth, and showed the importance of these "influencers" on youth's intentions to enlist and their subsequent enlistment rates. The third consideration is that the recruiting establishment also affects enlistment outcomes. The military services make a wide range of decisions regarding how to manage recruiting resources, including selecting recruiters, training them, and allocating them to recruiting stations throughout the country and selecting the level and allocation of advertising resources across media type. The services manage recruiters by assigning them quotas for the quantity of enlistments they make and for various enlistment categories (e.g., male versus female). They also generally use incentive plans that reward recruiters for various aspects of their productivity, such as certificates, badges, and improved promotion chances. The US services emphasize the recruitment of "high quality" youth, defined as highschool degree graduates who score in the upper half of the Armed Forces Qualification Test (AFQT) score distribution. During the 1980s high-quality enlistments averaged about half of total Army, Navy, and Marine Corps enlistments and 71% of Air Force enlistments. There have been significant swings in recruiting of high quality personnel since the beginning of the all-volunteer force (AVF) in 1973. Recruiting of high-quality males fell sharply in the late 1970s but rebounded during the 1980s. Factors that explain these swings are now considered. 2.1.1. Theoretical model of the enlistment process Since high-quality youth have better civilian opportunities, they are more difficult to enlist and so are thought to be supply constrained; consequently, recruiters are given more points and higher quotas for enlisting them rather than the lesser qualified. Because of these incentive and quota systems, recruiters do not passively process enlistments but may respond to enlistments by varying their level of effort and the allocation across enlistment categories. Following Dertouzos (1985), the enlistment process and the role of recruiter effort can be illustrated as in Figure 2. The recruiting production possibility curve (PPC) is given by AA'. This curve shows the feasible combinations of high- (H) and low- (L) quality enlistments that a recruiter can achieve for a given set of economic conditions, recruiting resources, and net tastes for service in the population. The mix of enlistments that the recruiter chooses depends on the shape of the PPC and the incentives he or she faces. Point Q in the diagram is the recruiter's quota for high- and low-quality recruits. If an enlistment determinant such as a recruiting resource is increased, the range of feasible enlistment outcomes increases, and if recruiters continue to supply the

Ch. 13:

355

The Economics of Military Manpower

M

0

A'

C'

B'

L

Figure 2. Recruiting production possibilities curves.

same level of effort, the PPC shifts out to BB'. The increase in potential high-quality enlistments (the desired category of enlistments from the service's perspective) holding the number of low-quality enlistments constant is given by the movement from point Q to point M. However, the movement to M assumes that recruiters maintain effort levels. As Dertouzos (1985) shows, recruiters do not have a strong incentive to overproduce because doing so can result in a higher future quota. If so, recruiters may reduce effort so that the shift in the PPC is smaller, as shown by CC'. If recruiters also have incentives to attain low-quality enlistments, their optimal outcome may therefore be a point like D, representing fewer high-quality enlistments than M. Therefore, studies that fail to account for the role of recruiter incentives will tend to underestimate the actual supply effect, the movement from Q to M. 2.1.2. Empirical models There have been two generations of research on enlistment supply during the allvolunteer force (AVF). The first-generation models were reduced-form models that ignored the potentially important role of recruiter behavior [see, e.g., Goldberg (1982), Ash, Udis and McNown (1983), Dale and Gilroy (1985), and Brown (1985) and the summary by Nelson (1986)]. The second-generation models recognized that recruiters respond to their incentive plans by varying the intensity and direction of their effort and held recruiter effort constant in their analyses. These studies include Dertouzos (1985), Daula and Smith (1985), Polich, Dertouzos and Press (1986), and Berner and Daula (1993). First-generation studies typically estimated a model of the form lnH =

lnX,

(1)

356

JT Warner and B.J Asch

where H denotes high-quality enlistments and X is a vector of determinants. Since variables were entered logarithmically, parameter estimates were interpreted as supply elasticities. Following Polich, Dertouzos and Press (1986), the second-generation studies recognized that recruiters maximize their utility where their welfare is assumed to depend on the number of enlistments, quotas (Q), and recruiter effort (E). Formally, U = U(E, H, L, QH, QL),

(2)

where QH and QL are the quota for high- and low-quality recruits, respectively. We expect: UE < 0, UQ,, < O, UQL > 0, but UH > 0 and UL > O0. Recruiters are constrained in their maximization of U by the variables included in X, the factors determining the PPC in Figure 2. Accounting for recruiter utility maximization and following past literature in specifying functional form, the (structural) enlistment supply equation is specified as InH = lnL + lnX + InE,

(3)

where A is the tradeoff between H and L on the PPC. Although E is unobserved, it is posited that the level of effort depends on how well the recruiter is performing relative to quotas, or InE

=

ylln(()+

721n(

)

(4)

Substituting Equation (4) into Equation (3) gives InH = allnL + a21lnX + a3 ln QH + a4ln QL.

(5)

Since L and H are jointly determined, Equation (5) can be estimated by a two-step procedure using the following equation for low-quality recruits: InL = 0 + rllnX +

zt21n

QH + r31n QL.

(6)

Simultaneous estimation of Equations (5) and (6) gives coefficient estimates for Equation (4) which allows identification of the underlying structural parameters in Equation (3). Daula and Smith (1985) take a somewhat different approach in incorporating recruiter incentives. They estimate a switching regression model that recognizes that high-quality enlistments are demand constrained if goals are set too low (e.g., point Q is demand constrained when BB' is the PPC) and are supply constrained when Q is infeasible. When resources are increased, recruiters will move from Q to a point like D in Figure 2 in a demand-constrained environment, but to a point like M in a supply-constrained environment. Thus, the estimated effects of an increase in resources will be smaller than when enlistments are demand constrained. Daula and

Ch. 13:

The Economics of Military Manpower

357

Smith's disequilibrium model yields qualitatively similar results to other studies that incorporate recruiter behavior, as shown below. 2.1.3. Empirical estimates Table 3 presents elasticities estimated by studies of US Army enlistments that have controlled for recruiter behavior. [Warner (1990) provides Army estimates similar to those in Table 3 and makes estimates for the other services.] Some of the estimates in Table 3 come from data generated during controlled national experiments. For example, between 1980 and 1981, the Educational Assistance Test Program (EATP) was conducted whereby the USA was geographically divided into a control and three test cells for the purposes of estimating the effects of varying the structure of educational benefits on high-quality enlistments [Fernandez (1982)].1 Between 1982 and 1984, the Enlistment Bonus Test (EBT) was conducted for the Army whereby the level and the distribution of bonuses across terms of service were varied across the control and test cells [Polich, Dertouzos and Press (1986)]. Relative military pay and the civilian unemployment rate are consistently found to influence high-quality enlistments, with relative pay elasticities ranging from about 0.15 to 1.89, with a central tendency of about 0.5 to 1.0, and unemployment elasticities ranging from 0.49 to 1.36. Table 3 also presents estimated elasticities for various recruiting resources. Of these resources, high-quality enlistments are the most sensitive to the number of recruiters - elasticity estimates are on the order of 0.5. On the other hand, studies find much smaller elasticity estimates for national advertising, with a central tendency of around 0.05 to 0.10. Polich, Dertouzos and Press (1986) find that advertising has a depreciation rate of less than 100 percent, i.e., the positive effects of more advertising persist for some time even if advertising expenditures fall to their previous level. They estimate that the long-run effect of a change in advertising is about 1.4 times its initial effect. Studies also find that educational benefits have a greater effect on high-quality enlistments than do enlistment bonuses. The estimates from the EATP indicate that introducing the Army College Fund in 1982 increased high-quality enlistments by about 9 percent whereas estimates from the EBT indicate that expanding enlistment bonuses increased them by about 5 percent (which translate into elasticity estimates of 0.17 and 0.7, respectively). Both the Army College Fund and the enlistment bonus programs are targeted toward specific occupations and both programs thus have the potential to channel recruits into hard-to-fill occupations. Although educational

The GI Bill was eliminated in 1977 and replaced with a much less generous program called Veteran's Educational Assistance Plan (VEAP). Poor recruiting in the 1978-1979 period brought about the EATP experiment. The Army College Fund and the Montgomery GI Bill, an enhancement to the VEAP, were outgrowths of this experiment. Unlike the GI Bill, under these programs new recruits must participate in a contribution plan to be eligible for future benefits.

358

J.8T Warner and B.J Asch Table 3 High-quality enlistment supply elasticity estimatesa

M/C

0.49 -0. 5 5 g 1.20 0.15-0.62 0.48 1.89 0.82

UnEm

Recr

0.56 0.94 0.59 0.57-0.65 0.49 1.36 0.99

0.59 0.60 0.15 0.48-1.15 0.27 1.11 0.83

Ads

References (1) Daula and Smith (1985) (2) Polich, Dertouzos and Press (1986) (3) Goldberg (1991)

Bonus

0.09b b c 0.06 b 0.05 0.14 -0.29 0.43 0.72 0.16-0.17 0.21 -0.04 0.46 0.13b c e 0.07 k C e

Table based on Berner and Daula (1993) Abbreviations: M/C, Relative military to civilian pay; UnEm, Unemployment rate; Recr, Recruiters; Ads, National advertising; Edu, Educational benefits; Bonus, Enlistment bonus; L-Q E, Low-quality enlistments; H-Q G, High-quality goal. b Impressions, not expenditures. Study has dummy variables for the availability of various educational benefit programs, but does not estimate an elasticity of supply with respect to educational benefit levels. d Study disaggregates low-quality enlistments into two groups: AFQT category 1 3A, Non-high school graduates, and Other Enlistments (AFQT categories 3B-4). The first estimate shown is for Other Enlistments. e Pooled sample. a

Edu

L-Q E

H-Q Goal

-0.02to 11" -0.31

0.41 0.22 0.33

-0.20 to 0.38 d -0.08to 0.17

0.19

Ref.

1e,f

2h 3i 4j 5k 1m, 1n'f

f Panel data, recruiting battalion by month,

10/80-6/84. The results shown do not account for battalion effects. They do not report their results for the pooled sample. g Civilian pay only, not relative military to civilian pay. h Panel data, Military Entrance Processing Stations (MEPS) by month, 7/81-6/84. Data are expressed as difference from corresponding month in base period, 7/81-6/82. i Panel data, recruiting battalion by month, 10/80-9/88. J Panel data, recruiting battalion by quarter, 1981-1990. k Berner and Daula (1993). Panel data, recruiting battalion by month, 10/80-1/90. Results control for endogeneity of goals. e Study has dummy variables for the availability of various benefit programs, but does not estimate an elasticity of supply with respect to bonus benefit levels. m Supply constrained. m Demand constrained. (4) Kearl, Home and Gilroy (1990) (5) Berner and Daula (1993)

benefits have a larger market expansion effect, evidence suggests that enlistment bonuses are more effective at skill-channeling. 2

2 Estimates show that enlistments rose by 30-40% in the skills eligible for the enlistment bonus, holding the change in the total enlistments constant [Polich et al. (1986)]. The skill-channeling effect for educational benefits was estimated to be 17% [Fernandez (1982)].

359

Ch. 13: The Economics of Military Manpower

Table 4 Estimates of the marginal cost of recruiting resources Resource Entry basic pay Enlistment bonus National advertising Recruiters Educational benefits a

Estimated marginal cost (1990 $) 34 800 18 600 8100 7300 6900

Estimates based on Polich et al. (1986), Asch and Dertouzos (1994), and Asch et al. (1992).

The coefficient estimates for the high-quality quota and the high-quality/low-quality tradeoff parameter in the table show the importance of controlling for recruiter behavior. For example, the estimate of A from Polich, Dertouzos and Press (1986) of -0.31 implies that high-quality recruits are about 4 times as hard to recruit as lowquality ones. Daula and Smith's 1985 analysis implies a trade-off of about 8-to-1. Thus, if recruiters face insufficient rewards for achieving high-quality recruits (e.g., the point trade-off in the incentive plan is at worst less than 4-to-1), then recruiters will allocate their effort towards achieving low-quality ones. The size and significance of the high-quality quota coefficient estimates also suggest that quotas are one of the primary determinants of high-quality enlistments. Furthermore, the importance of accounting for recruiter behavior is also seen by comparing the supply elasticity estimates when enlistments are demand-constrained versus supply-constrained (the final two rows in Table 3). These estimates are smaller in the demand-constrained environment, as predicted by theory. The interaction between recruiter incentive systems and enlistment quotas is further studied in Asch (1990), Asch and Karoly (1993), and Berner and Daula (1993). Asch (1990) finds evidence that the structure of the Navy recruiters' incentive plan affected the timing and quality mix of enlistments, while Asch and Karoly (1993) show that the structure of the incentive plan for job counselors affected the number of highquality enlistments and the fill rates of various occupations. Berner and Daula (1993) indicate evidence that recruiting goals are endogenous (e.g., their size is based on past production and supply) so that recruiters who overproduce are penalized with higher quotas. The elasticity estimates in Table 3 provide the means to compute the marginal enlistment cost of various recruiting resources. Estimates from various studies are shown in Table 4. These estimates show the importance of considering both the resource effect (e.g., the elasticities in Table 3) and the policy cost in determining the optimal mix of recruiting resources. Although Table 3 indicates that (relative) military pay has a larger elasticity than advertising, Table 4 indicates that increasing pay is a relatively costly recruiting policy while increasing advertising is not. Raising pay is costly because to attract an additional high-quality recruit DoD must also raise the pay

360

JT Warner and B.J Asch

of everyone who would have enlisted anyway (i.e., pay rents). Enlistment bonuses are also a relatively costly recruiting resource. Table 4 suggests that recruiters, advertising, and educational benefits are the most cost-effective resources. However, effectiveness is measured in terms of high-quality enlistments. If other criteria for effectiveness were used, such as skill-channeling, a different ordering might result. Withers (1978) estimated enlistment supply models for the UK, Canada, Australia, and the USA for the period 1967-1973. Although the shortness of the time period and multicollinearity plagued the analysis for several of the countries studied, he obtained robust estimates for enlistments into the UK Army, with a pay elasticity estimate of 1.46 and an unemployment elasticity of 0.90. 2.2. Retention 2.2.1. Theoretical models of retention The analysis of retention poses a theoretical problem that is generalizable to any labor market setting. Consider an individual at time period t who is considering whether to stay or leave. Suppose that the individual can stay to t + 1 and then leave, t + 2 and then leave, etc. The theoretical problem is deciding the time horizon that is relevant to retention decision-making. The problem is particularly important in the US military because of the lumpiness of the military income stream - individuals are vested in a sizeable pension only upon completion of 20 years of service. The simpler solution was provided by the Annualized Cost of Leaving (ACOL) model [Enns, Nelson and Warner (1984), Warner and Goldberg (1984)]. To illustrate the ACOL approach, suppose that for an individual at year t: (1) Wj is expected military pay in each future yearj, (2) WCt is civilian earnings in future yearj if the individual leaves at t, (3) WCn is civilian earnings in future year j if the individual separates after future year n, (4) Rn is the expected present value at future year n of retired pay and other separation benefits if the individual separates after year n, (5) Rt is the present value at year t of retired pay and other separation benefits if the person leaves now, (6) Tm and r c are the preference factors previously defined, and (7) is the individual's subjective discount rate on future income. Then the present value of the future benefit from staying from period t to period n is: I (1

S,, =

j=t+l

M

R -I + (1 +p) + +

+Y

Tm

- 1 (1+ (1 +p)-

j=t+l

W

t

Y1

(lp

+ TC

(7)

j=n+

The value of leaving immediately is: wet v +

L =

c

(l +p)I + R,. ,=1+1

(8)

Ch. 13:

361

The Economics of Military Manpower

The cost of leaving is Ct,n =St,n -Lt, which can be written as: E

WnY-WC

/

Rn

(1+p)t + (1 + p)n-t Rt ~~~j=t+~~

-

E

Wj -WJj

E

( + ) t-

-T(

+(9)

j=t+1

j=n+l

The first term accounts for the military-civilian wage differential over the interval from t to n, the second term measures the change in the value of retirement benefits, the third term accounts for the change in civilian opportunities brought about by service from t to n (see Section 7), and the last is the present value of the net preference for civilian life. The individual is presumed to stay if there exists at least one future time horizon over which Ct,n is positive. Of course, it could be positive over multiple periods. How to separate stayers from leavers? Saying that the individual stays if there exists at least one future horizon over which Ct,n is positive is equivalent to saying that the individual stays if the following is true over at least one future time horizon: n fint r -

m

R )WWE)n+l

w.-w.wC

M

C

Rt-R

_W~+ ( (ll~-

W

C

(l+p)J


O0.Then at each point in the career an individual with ability parameter a must decide whether to remain in service and how much effort to supply. Likewise, the military must decide whether to retain the individual and whether to promote him or her. The military decides whom to promote by administering an evaluation (Ei,t) to everyone in rank i at time t, rank-ordering the evaluations, and promoting some fraction ri*,. An individual's promotion probability (ri,t) depends on his or her ability and effort and the abilities and efforts of others (a0 and e, respectively). The military may also use the evaluation to separate those whose evaluation Ei,t falls short of some minimum E'. The model is a generalization of the Gotz-McCall dynamic programming model. It has the following structure: To begin with, the expected value of future utility at the end of period t is Vi*t = cti,E(S,t G*, + Ei,, > 0) + (1 - i,t)Li,t,

(17)

where: Pi,t = pr(E' < Ei,t) pr(G Gi,t =

i+l,t+l [T'

t

+ Ei,t > 0)

+ 6 i+ + Wil, t+l +

+ (1 - ni+l it 1t+l)[Tm + i + Wm

1

= p

, )2

Vil,t+l - Z(eit+l)] +P

+l - Z(eit+l)] - Li,,

(18)

Li,t = Cit, + Ri,t + Ft,

i+l,t+ = T(a, ei,t, a, t, eitl, 7ri, t+l)

In words, expected utility Vi*t is a weighted average of the expected return to staying and the return to leaving immediately, where the weight i,t is the product of the (independent) probabilities that the military wants to retain the individual (,t) and the individual wants to stay (p,2t). The expected gain to staying Gi*t is a weighted average of the payoff to achieving next rank in period t + 1 and the return to remaining in the current rank. These returns depend on the pecuniary reward associated with the ranks (WM) and the value of rank-specific non-pecuniaries (6). The expected gain to staying G*, is the expected return to staying minus the return to leaving, which equals the present value of civilian earnings (Ci,t), vested retired or separation pay (Ri,t), and the value of non-pecuniaries in the civilian sector (Ft). Individuals desire to stay only if Gi,t + Ei,t > O, where Ei,t is the random shock to the retention decision. Individuals with higher permanent tastes for military service (Tm) are more likely to stay. If the individual is at a mandatory separation point in rank, then ]i, t = 0 and Vi t =Li,t.

Ch. 13:

The Economics of Military Manpower

383

The individual's expected future utility at the start of period t is: rm + WM +fi V t-Z(eit ).

(19)

The individual supplies effort in t to maximize Equation (19). Raising effort can raise expected future utility because (1) effort raises the probability of promotion, which conveys higher monetary and non-monetary rewards, and (2) effort reduces the probability of involuntary separation arising from a failure to meet minimum performance standards. Promotion also pushes further into the future the date at which the individual might be subject to a high-year-of-tenure rule. The first-order condition for utility maximization is: 'Pt[Ti (Wi+lt+ - Wit+l + 6 i+1 - 6i + (Vi+,t+l - Vit+l))] + p0,t [Git',t

(20)

+ (E0J2,] K - Z'(ei,t) = 0.

The first two expressions in (20) represent the discounted marginal benefit of effort. The last term, Z'(ei,t), is the marginal cost of effort. Individuals supply effort to the point at which marginal benefit equals marginal cost. The first marginal benefit term measures the direct pecuniary and non-pecuniary payoff from effort. It is apparent that the direct benefit of effort increases as the period t+ I interrank pay spread (W ,,t+ -W t+l) increases and as the difference in the value the individual places on the non-pecunaries associated with the next rank and the current rank (6i+l - 6i). Furthermore, the direct benefit of effort of all differentials beyond period t+ I are summarized in the term (V+l,t+ - Vi*t+). The direct benefit of effort is weighted by the probability of staying, such that individuals who are more likely to stay, and thus realize the reward to effort in the current period, have a higher expected reward to effort and will thus work harder. The term grt is the marginal effect of current period effort on the probability of promotion in the next period (i.e., Jr' = Ori+l,t+l/0ei,t).The direct benefit of effort is larger the more marginal effort raises the probability of promotion. Asch and Warner (1994a) show that J' is largest when the expected probability of promotion is 0.5. Marginal effort has little value when the probability of promotion is either very high or very low. They show that JT' declines the more important are the idiosyncratic determinants of promotion (i.e., "luck"). Finally, they demonstrate that y' increases with the scale of the contest. The more participants there are in the contest, the larger is the marginal value of effort. Scale matters because the more participants there are, the greater is the chance that the individual can surpass other contestants by supplying more effort. The second term of Equation (20) measures the value that effort has in avoiding separation due to failure to meet minimum performance standards 9. Since all terms 9 The term ¢0, shows the effect of marginal effort on the probability of being allowed to stay; ¢0, is

the effect of a small change in G-, on the probability of wanting to stay, and K is the effect of marginal effort on the evaluation. See Asch and Warner (1994a) for details.

384

JT Warner and B.J Asch

in this expression are unambiguously positive except for G*, extra effort has value in terms of reducing the threat of involuntary separation except for those who have a very large negative expected gain to staying. For most personnel, minimum performance standards and up-or-out rules are a stimulant to effort. A main implication of the model is that, ceteris paribus, interrank pay differentials should rise as individuals progress through the ranks (i.e., the system should be skewed). Without skewness individuals will reduce effort as they reach the higher ranks because of the effect of declining promotion probabilities on the return to effort. But several factors might explain the observed lack of skewness in the US and foreign systems, including (1) the greater probability of staying as YOS and rank increase, (2) the value of rank-specific non-pecuniaries increases with rank, (3) the sheer size of the promotion contests. Additionally, an oft-cited factor that reduces the degree of skewness is that military "output" is team-oriented. Team production requires cooperation and not competition, which excessive skewness might promote [Rosen (1992)]. The lateral entry constraint also serves to reduce skewness. Since about two-thirds of enlistees remain for only one enlistment, their enlistment decisions will be based mostly on entry level pay. The lack of lateral entry means that the military must access enough personnel at the entry level to fill lower-level positions today and higherlevel positions in the future. Since ability has an increasing effect on performance as individuals progress through the ranks, there must be a sufficient number of highability personnel in the entry cohort to fill the upper-level positions in the future. But the military cannot just selectively recruit sufficient numbers of high-ability personnel because true ability is unobservable at entry. However, when entry pay is increased, the ability mix improves because higher entry pay attracts more applicants who have observable characteristics that are correlated with ability (education level and AFQT test scores) and the military can and does in fact screen on these characteristics. An implication of this discussion is that many junior personnel are overpaid due to the lack of lateral entry (i.e., they earn economic rents). The analysis highlights several potential problems with the US military compensation system. First, for the most part, intragrade pay raises are not performance based, but are provided in a lock-step fashion based on time-in-service. Some use of performance-based intragrade pay should be explored. Second, because longevity increases are based on time in service, early promotions convey only a temporary gain. In fact, slow promotees often earn more upon promotion that those who have held the rank several years. A system based on time in service blunts the advantage to working harder and achieving an early promotion and it is likely to produce a significant amount of adverse selection. A system of longevity increases based on time in grade has been recommended a number of times and the pay tables of a number of other countries, including Canada, the United Kingdom, France, and Australia, are based on time in grade. There should be further investigation in the USA of such a system. The US military retirement system has often been criticized for its high cost, its unfairness to those who separate without benefits, and its inflexibility in force

Ch. 13:

The Economics of Military Manpower

385

management. But is it a coincidence that the US and its allies all have retirement systems that postpone retirement to a late date but provide relatively generous benefits to retirees? Asch and Warner argue not. The lateral entry constraint places the military in a much different situation from civilian employers. It must access and train large numbers of entrants before identifying for advancement those who have the talent to perform the higher-level tasks in the organization. It therefore wants to provide incentives for the most talented to stay and seek advancement and for others to leave after they discover that they are unsuitable for the upperlevel positions. That is, it must provide the proper incentives for personnel to selfsort. Delayed vesting of retired pay induces self-sorting because only those who think that they can achieve the requisite rank and longevity will stay early on while others will leave. Deferred retired pay can also motivate work effort, especially when combined with minimum performance standards for retention and up-or-out rules that prevent low-ranking personnel from staying long enough to collect retirement benefits. This discussion begs the question of when vesting should occur. But notice that there is a trade-off between the vesting date and the military's ability to pay new entrants. If the military is to meet a fixed budget constraint, earlier vesting will dissipate its capacity to raise entry pay and attract a higher-quality entry cohort. Contrary to critics of delayed vesting, it is not necessarily unfair to the bulk of entrants who never qualify for retirement benefits because they are generally overpaid as a result of the lateral entry constraint. The question now arises why retirement benefits should be part of the self-sorting mechanism. After all, why not just pay a bonus to all who reach the requisite rank and years of service? The answer has to do with retired pay's role as a separation incentive. At some point the military wants everyone, including the best personnel, to separate, even when they may still be individually very productive. The longer individuals remain in the top positions the slower will be the promotion rates for younger (and potentially equally able) personnel. Unless offset by changes in the structure of pay, reduced promotion opportunities in the junior ranks will discourage work effort in those ranks and will cause those junior personnel with the best external opportunities (i.e., the more able) to leave. Without the proper inducement, the senior personnel may not want to leave voluntarily if their military pay exceeds their best private sector alternatives. Such is especially likely to be the case for those trained in the militaryspecific skills. There is, of course, no reason why the separations required to maintain personnel flows could not be accomplished with other policy tools such as up-or-out rules. However, excessive reliance on involuntary separation to control the experience structure of the force can be bad for morale, impacting on recruiting, retention, and work effort. These adverse effects might require the payment of a "regret premium" to compensate for the prospect of involuntary separation. A relatively generous retirement system for senior personnel may be the only way to quell these problems. Asch and Warner (1994b) evaluate a number of past proposals to overhaul the US military

386

JT Warner and B.J Asch

retirement system. Considered in the light of the model developed in this section, most would do more harm than good. Finally, note an important implication of Equation (20). Draftees cannot be induced by compensation and personnel policies to supply effort because draftees know that their probability of staying and receiving contingent compensation () is low. This result implies that draft armies will have to devote more resources to direct monitoring and disciplining of draftees, a factor which further reduces the availability of ready personnel. Furthermore, draftees will have to be motivated by penalties for poor performance (e.g., bad conduct discharges).

6. Force management issues 6.1. Women in the military With the exception of nursing positions, the military has traditionally been a male organization. As women's role in society at large has been transformed over the last few decades, the question arises of whether the role of women in the military should more closely mirror the greater role of women in the civilian sector, and specifically whether all military positions, including combat ones, should be open to women. Binkin and Bach (1977) discuss the economics of sex integration into the armed forces. The question they pose is: what mix of military men and women can meet US national security needs at the lowest cost? Adressing this question requires information on the differential costs and effects associated with different mixes of men and women and of filling a given position with a man versus a women. Under the cost category, information is needed on the differential recruitment, training, retention, and separation (including retirement) costs. There may also be adjustment costs, such as those associated with changing facilities to accommodate women. The costs associated with military benefits, including medical costs, housing costs, and the cost of dependents may also differ depending on the mix of men and women. Are men and women equally productive? Those who oppose women serving in some military positions argue that women lack men's upper body strength and therefore could not perform equally as men in some jobs. Some also argue that because of pregnancy, women are less likely to be deployable and thus, units which have a greater mix of women are less ready. Still others argue that women would adversely affect unit cohesion, and that they are more likely to be emotionally strained by the rigors of combat. Studies of the relative costs and effectiveness of women versus men in the military generally focus on variables that are amenable to analysis, such as enlistment, attrition, and retention behavior. On these important dimensions, past studies indicate that the behaviors of women and men are often similar and respond to many of the same factors. Hosek and Peterson (1990) use a choice-based sample to estimate a logit model of the individual enlistment decisions of young men and of women and find

Ch. 13:

The Economics of Military Manpower

387

that for most variables, the coefficient estimates are statistically equivalent for men and women. They use individual rather than aggregate level data since in the aggregate, women are demand constrained - more want to enlist in a given year than the services demand at the prevailing level of compensation. While they find that female and male enlisted supply generally depend on similar factors in similar ways, they do find that the intercept is smaller for women - reflecting the demand constraint - and that immediate marriage plans are a greater deterrent to women's enlistment. Overall first-term attrition rates are higher for women than for men [Buddin (1988)], but there are differences among the services. Attrition rates are higher for Navy women than Navy men during the early months of terms of service, but the rates converge over time [Fletcher et al. (1994)]. Attrition rates for men and women in the Army also converge over time, if attrition due to pregnancy is excluded [Martin (1994)]. But when attrition due to pregnancy is included, the rates do not converge. Retention behavior is similar for men and women [Quester (1988)]. When retention differs, women in fact have somewhat higher first-term retention and slightly higher long-term retention rates than men [Quester (1988), Shiells and McMahon (1993)]. The evidence on sex differences in productivity is spotty. The services presented evidence before the 1992 President's Commission on the Assignment of Women in the Armed Forces that women are less effective on two counts. First, service data suggested that women have lower deployment rates than men and that lower deployment rates detract from unit readiness. Second, evidence was presented purporting to show that women are at a disadvantage in performing military jobs requiring physical strength. Yet, the greater reliance on technical skills and the reduced reliance on physical strength over the last few decades among many military jobs, especially in the Air Force and Marine Corps, weakens the physical strength argument against the employment of women in the military. Furthermore, as noted by Fletcher et al. (1994), other physical attributes such as compact body type that are important for military jobs that are limited by space (such as in ships, submarines, and aircraft) favor women over men. For the many reasons discussed earlier, the services stress the recruitment of highquality personnel. Because of demand limitations the fraction of women who are high quality exceeds the male high-quality fraction. During the 1980s, about half of male enlistees were in the high-quality category whereas over 60 percent of female enlistees were. The relevant comparisons are not between high-quality female enlistees and highquality male enlistees, but between high-quality female and low-quality male enlistees. Along many dimensions, the women may be the more productive group. 6.2. Reserve force management issues Several characteristics of reserve service distinguish it from active service. First, only a small fraction of reservists work full-time; during peacetime most reservists are civilians. Consequently, reserve units must draw personnel from local labor markets. Second, unlike the active components where most individuals who enlist have no

388

J. T Warner and B.J Asch

prior military experience, about half of all reserve enlistees do [Marquis and Kirby (1989a)]. Prior service personnel are generally recruited to fill the mid-career and senior positions whereas non-prior service personnel generally fill junior positions initially. Third, since reservists are primarily civilians and cannot be relocated to meet service demands in any location, skill matching is important; reserve personnel of the appropriate skill type must be recruited and retained to meet the occupation mix of the reserve units in the local area. Finally, there is no clear-cut reenlistment point in the reserves. Although reservists sign enlistment contracts, the timing of separations from the reserves bears little relation to the formal contract expiration dates. Thus, there is little practical distinction between continuation and reenlistment in the reserves. Several studies have estimated enlistment supply of non-prior service and of prior service personnel to the reserve components. Generally these studies find that the same factors affect reserve as active duty enlistment supply and in very similar ways. For non-prior service personnel, Tan (1991) finds pay elasticities between 0.43 and 0.67, unemployment rate elasticities between 0.25 and 0.45, and recruiter elasticities between 0.4 and 1.0. He also finds evidence of a negative tradeoff between nonprior service and prior service recruits; as recruiters allocate their effort towards priorservice recruits, non-prior service enlistments fall. Studying prior service enlistments into the Naval Reserve, Kostiuk and Grogan (1987) estimate a pay elasticity of 0.82 and a recruiter elasticity of 0.50. Marquis and Kirby (1989a) find a pay elasticity of 1.17 for prior service personnel affiliating with the Army Reserve and the Army National Guard. A unique set of factors that are hypothesized to affect reserve but not active duty supply are employer attitudes and policies towards reservists. Evidence from Grissmer, Kirby and Sze (1992b) suggests that reserve service imposes a cost on reservists in terms of conflicts with their primary employers, especially among those working in small firms. However, family/spousal attitudes toward reserve service had a larger effect on reservists' reenlistment decisions than did perceived civilian supervisor attitudes. It has also been argued that reserve duty differs from other secondary jobs because military service is national service and involves the risk of mobilization and the possibility of family hardships, financial loss, and injury. Mehay (1991) finds evidence indicating that several of the determinants of secondary job participation differ for reservists and civilians but that individuals do appear to choose to moonlight (regardless of whether it is in the reserves or in the civilian sector) based on relative compensation and local labor market conditions. Asch (1993) expands the moonlighting model to explicitly incorporate the risk of mobilization for reservists, and Grissmer, Kirby, Sze and Adamson (1992a) examine the options for offering insurance protection against losses incurred during mobilization. The model of the decision to stay in the reserves is fundamentally similar to the stay-leave decision in the active components. Evidence on reserve attrition indicates that the factors that are important in explaining active attrition also tend to be important in explaining reserve attrition. For both prior service and non-

Ch. 13:

The Economics of Military Manpower

389

prior service personnel, higher reserve pay and higher education levels are usually associated with lower attrition, as is greater aptitude [Kostiuk and Follmann (1988), Marquis and Kirby (1989b), Grissmer and Kirby (1988)]. These studies provide mixed results on attrition of women and nonwhites compared with males and whites. One issue of concern has been high attrition among non-prior service reserve personnel. Grissmer and Kirby (1988) find attrition rates of 35 to 40 percent two years after entry. High attrition rates are costly because of the substantial training investment made by the military, and the relative short time non-prior service individuals have been in service relative to prior service who entered the reserves at the same time. But Kirby and Grissmer (1993) show that reserve attrition is not excessive because a significant fraction of those who leave the reserves later rejoin or, alternatively, join the active force. Although reserve attrition is high, only one-third to one-half of the losses are to civilian life. The rest join the active force (about 25 percent of reserve separations) or later rejoin a reserve component. When attrition is defined as separation to civilian life, Kirby and Grissmer find that reserve attrition rates are often lower than they are for comparable non-prior service personnel in the active force. Another issue of concern is the skill qualification levels of reservists. Individuals who are not skill qualified are not deployable which, in turns, adversely affects reserve unit readiness. Some evidence indicates that personnel in the reserve components are often less skill qualified than their active duty counterparts [Grissmer et al. (1994)]. Among enlisted personnel, the skill qualification level among Naval Reserve personnel is 13 percentage points lower than that of active Navy enlisted personnel. This figure is 10-13 percentage points for the two Army reserve components. One reason for the lower skill qualification level in the reserves is that most prior service accesssions require retraining; they do not enter the skills that were trained in while they were on active duty. Less than one half of prior service personnel match their active and reserve jobs when they affiliate with the reserves. The match rate is higher for the Naval Reserves (70 percent) than for the Army National Guard, where the match rate is less than one-third (Grissmer et al. 1994). These figures imply that the training cost saving associated with employing prior service over non-prior service personnel is less than is generally believed. Another reason for lower skill qualification levels in the reserves is the significant amount of personnel turnover in reserve units. The Air reserve components have the greatest stability - about 70 percent of the positions are filled by the same enlisted personnel over an 18-month period. But the Army reserve components have stability rates of only 45 to 50 percent and the Naval Reserve has the lowest rate, one-third. Eighty percent of this turnover is due to switching units within local areas rather than geographic relocation [Buddin and Grissmer (1994)]. The cause of this turbulence appears to be due to individuals' desires to change jobs or to seek promotion opportunities. The aggregate costs or benefits of this turnover remains an open question.

390

J T Warner and B.J Asch

The overriding issue is the structure and the appropriate mix of active and reserve forces 10. The US Congress has generally favored a greater reliance on the reserves than the services have. At issue is the substitutability of reserve and active forces and their relative cost. The ease of substitution depends on the speed with which reserve forces can be mobilized and on the relative unit readiness of the reserve and active forces. Some reserve force components are virtually perfect substitutes for their active force counterparts and have the advantage of lower peacetime costs. Examples are Air National Guard and medical units. But anecdotal evidence indicates that in direct combat units the reserve forces require a long lead time for mobilization and are not as combat-ready as the active forces II . One of the key force mix questions is whether a given active/reserve mix that meets national security goals is sustainable. Analyses of alternative mixes assume that reserve readiness can be maintained regardless of the mix. However, a constraint on the degree to which the active force can be reduced and the reserves increased is the supply of active veterans to the reserves. As the active force gets smaller, the flow of veterans to the reserve forces falls. This constraint is important because of the contribution of prior service personnel to reserve readiness. While direct evidence on the relative contributions of prior service versus non-prior service personnel is scarce, there is an important difference between them that affects readiness, and that is cumulative military experience. Grissmer et al. (1994) shows that among personnel in similar paygrades, non-prior service personnel have about half the military experience of prior service personnel. Debate over the constitution of the active and reserve forces will no doubt continue. More research is needed to understand better the productivity differences between prior service and non-prior service personnel in each of the reserve components.

7. Civilian returns to military service The military has long been viewed as a good training ground for the nation's youth, and the military can have a significant impact on the general society given the thousands of individuals who enter and leave service each year. From an academic perspective, it is interesting to know whether the training and experience acquired in service have a payoff once individuals separate. It is also important from a policy perspective, since the design of an effective compensation system depends on how military service impacts future civilian alternatives.

'o A recent Congressionally mandated study addresses this issue in great detail. The discussion in the text draws heavily from the study's final report [National Defense Research Institute (1992)] as well as from Grissmer et al. (1994). " Some Army Reserve armor units activated early in the Gulf crisis were not ready for overseas deployment even after months of training at the National Training Center in California.

Ch. 13:

The Economics of Military Manpower

391

Underlying the returns-to-service literature are the human capital hypothesis and the screening hypothesis. According to the human capital hypothesis, once individuals leave military service their earnings relative to nonveterans will depend on the amount of military training and experience received and their transferability to the civilian sector. For the same amount of training and experience, veterans' civilian earnings will be less than the earnings of comparable non-veterans if all or part of that training and experience is not transferable, because those who directly entered the civilian labor market accumulated civilian human capital while those who entered the military did not. Some types of military training and experience are more transferable than others. According to the screening hypothesis, even if military training does not provide skills that can be used in the civilian sector, military service may still give a positive (negative) return if civilian employers view military service as evidence of desirable (undesirable) work qualities that are inherent to those who serve. Thus, military experience may serve as a signal of the innate characteristics of those who enter the military [DeTray (1982)]. Evidence of a positive or negative return to service is thus consistent with both the human capital hypothesis and the screening hypothesis. However, from the standpoint of addressing the question of what roles the military might play in, say, developing youth, knowing which hypothesis is the correct one is crucial because they yield different policy prescriptions. The general approach to estimating whether there is a veteran's premium in the civilian labor market is to estimate an equation of the form In W = 6V+Xfi + ,

(21)

where W is the individual's civilian wage, V represents veteran's status, X is a vector of other individual characteristics, E is a random error term, and 6 and i are parameters to be estimated. The parameter represents the veteran's premium (if it is positive) or penalty (if it is negative). Studies that apply ordinary least squares (OLS) to Equation (21) generally find that World War II veterans earn a premium [Fredland and Little (1980), Rosen and Taubman (1982)] but that Vietnam-era veterans suffer a penalty [Rosen and Taubman (1982), Berger and Hirsch (1983)]. Furthermore, there is some evidence that the veteran premium is larger for non-whites and for the less educated 12 A problem for all of these studies is the failure to control for sample selection bias. Sample selection bias occurs when the unobservable factors that determine earnings () are correlated with veteran's status (V). Such will be the case if individuals are not randomly assigned to the veteran and non-veteran groups. Non-random assignment can occur two ways: (1) those who enter military service are not a random sample of Xie (1992), for example, tracks earnings of synthetic cohorts of veterans in the March Current Population Survey of each year from 1964-1984. Among white veterans, he finds negligible earnings differences between veterans and non-veterans once a host of other factors are controlled for. But black veterans earn 6% more than non-veterans. 12

392

JT Warner and B.J Asch

all qualified youth and (2) those who leave military service are not a random sample of all enlistees. In fact, considerable evidence indicates that the decision to enter or separate from service is not random. The correlation between E and V can be positive or negative, depending on the circumstances. Given that the military screens potential entrants and tends to reject those with lower aptitude and physical exam scores, it could be that V and E are positively correlated since those who score low are likely to be low earners as well. In this case, OLS estimates of 6 will be biased upward. But the correlation could be negative if the more able find enlistment unattractive (as during the early AVF period) or if they are successful in avoiding the draft (as may have been the case during the Vietnam era). Sample selection bias can also arise if the decision to leave the military is not random. The bias could be negative if up-or-out rules force the separation of less able personnel (while the more able stay), but it could also be positive if those with the best civilian alternatives are the ones who actually separate. Only five studies have attempted to correct for selectivity bias: Angrist and Krueger (1994), Angrist (1989, 1990), Gilroy et al. (1992), and Bryant et al. (1993). The latter two employ Heckman's two step method of correcting for sample selection bias [Heckman (1979)]. Angrist's various studies use a simpler instrumental variables technique: find a variable correlated with veteran's status (V) but not the unobservable determinants of earnings (E) and use it as an instrument for V. He takes advantage of the lottery nature of the draft, which randomly assigned youth into draft status based on birth date to form an instrument for V. Analysis of results based on simple OLS with those based on instrumental variables techniques find that estimates -based on OLS are inconsistent whereas those based on the instrumental variables method are not. Angrist (1989) finds a penalty of 28 to 35 percent for white Vietnam-era veterans and a premium (that is only marginally significant) of between 20 to 40 percent for black veterans. Using alternative data sources, Angrist (1990) arrives at qualitatively similar results. Angrist and Krueger (1994) restudy the earnings of World War II veterans. Using an instrument for veteran status based on birth year and month, they estimate a wage penalty of 6.4 percent using the instrumental variable method but a 7.8 percent premium using OLS. Specification tests indicate that the negative estimates based on the instrumental variable technique are consistent while the OLS estimates are not. The two studies of AVF veterans find mixed results. Gilroy et al. (1992) find that white veterans initially earn less upon leaving service but then earn more relative to a comparable group of nonveterans who entered the civilian market upon leaving high school. Over an 8-year time horizon, the premium for non-Hispanic whites is estimated to be about 5 percent. But they find a zero returns for blacks and for Hispanics. Bryant et al. (1993) also focus on AVF veterans. Contrary to Gilroy et al. and more consistent with the various Angrist studies, they estimate a veteran wage penalty of between 1 and 8 percent. The wage penalty is estimated to be greater for whites than for nonwhites. The bulk of the veterans in these studies served for relatively short periods. Thus studies that contrast veteran and non-veteran earnings tell little about how longer periods of service affect future civilian earnings capacity compared with civilian

Ch. 13:

The Economics of Military Manpower

393

experience. And since none of these studies controlled for military occupation, they tell nothing about whether post-service earnings depend upon the type of training. One such study that addresses these questions is Borjas and Welch (1986), who studied the earnings of military retirees who were surveyed in the 1977 Defense Retiree Survey along with a matched sample of veterans from the March 1977 Current Population Survey (and who presumably served for relatively short periods) 13. Over the span of the second-career, officer retirees who work year-round earn 14 percent less than their veteran counterparts and enlisted retirees earn 20 percent less. Enlisted personnel in the Combat Arms skills lost an additional 13 percent. Goldberg and Warner (1987) found similar patterns studying the Social Security earnings over the period 1972-1977 of a large sample of enlisted personnel that separated in FY 1971. For three occupation categories that appear a priori to provide transferable training (e.g., Electronics Equipment Repair), the authors could not reject the hypothesis that additional military experience adds as much as civilian experience to future civilian earnings. In these skills, military and civilian experience appear to be perfect substitutes. But other skills do not appear a priori to provide as much transferable training (e.g., Combat Arms), as a year of military experience was found to add less to subsequent civilian earnings than a year of civilian experience. Over a 20-year career, the earnings loss is about 20 percent in the non-transferable skills. To summarize, the bulk of evidence fails to support a finding that military service conveys a post-service return that is in excess of the return from civilian experience. Collectively, the studies suggest that for short periods of service the return to military service is roughly comparable to the return to civilian experience. In fact, that those who enter for one term of military service do not suffer a large loss should be considered a positive result, for it suggests that military service is a legitimate avenue for the acquisition of human capital. But those individuals who serve for longer periods in non-transferable skills do suffer a loss of earning power, a fact that must be considered in the design of the military compensation system.

8. Summary It is evident from this survey that over the past two decades economists have contributed a substantial body of literature on the economics of military manpower. For the USA at least, supply relationships and the relationships between experience and productivity are now reasonably well understood. Future research should focus on the relationship between productivity and incentives and how effectively military personnel systems sort personnel into their most suitable positions. The bulk of research has focused on the USA. But pay and personnel practices do vary across

13 Comparing retirees with (short-service) veterans should eliminate some of the entry-level selection bias since all entrants met initial screens.

394

. T Warner and B.J Asch

countries. Whether differences in practices generate differences in outcomes would be a particularly fruitful avenue of future work. As militaries in the US and elsewhere accumulate more data about the effectiveness of women and reservists, economists will no doubt play a role in the analysis and interpretation of those data. Finally, this survey has said little about the military downsizings now underway in the USA and elsewhere. It is too early to tell what effects they will have on various measures of effectiveness and how military organizational structures might be modified to minimize their impact. But the downsizings, which generate exogenous changes in force size and force structure, will create numerous opportunities to better understand the economics of military manpower.

References Albrecht, M., 1979, Labor substitution in the military environment: implications for enlisted force management, R-2330-MRAL (RAND, Santa Monica, CA). Altman, S., and A. Fechter, 1967, The supply of military personnel in the absence of a draft, American Economic Review 57, 19-31. Angrist, J., 1989, Using the draft lottery to measure the effects of military service on civilian labor market outcomes, in: R. Ehrenberg, ed., Research in labor economics (JAI Press, Greenwich, CT) 265-310. Angrist, J., 1990, Lifetime earnings and the Vietnam era draft lottery: evidence from social security administration records, American Economic Review 80, 313-336. Angrist, J., and A. Krueger, 1994, Why do World War II veterans earn more than nonveterans?, Journal of Labor Economics 12, 74-97. Asch, B., 1990, Navy recruiter productivity and the Freeman plan, R-3713-FMP (RAND, Santa Monica, CA). Asch, B., 1993, Reserve supply in the post-Desert Storm recruiting environment, MR-224-FMP (RAND, Santa Monica, CA). Asch, B., and J. Dertouzos, 1994, Educational benefits versus enlistment bonuses: a comparison of recruiting options, MR-302-OSD (RAND, Santa Monica, CA). Asch, B., and L. Karoly, 1993, The role of the job counselor in the military enlistment process, MR-314-P&R (RAND, Santa Monica, CA). Asch, B., and J. Warner, 1994a, A theory of military compensation and personnel policy, MR-439-OSD (RAND, Santa Monica, CA). Asch, B., and J. Warner, 1994b, A policy analysis of alternative military retirement systems, MR-465-OSD (RAND, Santa Monica, CA). Asch, B., J. Dertouzos, J. Hosek and B. Rostker, 1992, Restructuring DoD's accession plans in the coming decade, White paper (RAND, Santa Monica, CA). Ash, C., B. Udis and RF. McNown, 1983, Enlistments in the all-volunteer force: A military personnel supply model and its forecasts, American Economic Review 73, 144 155. Beland, R., and A. Quester, 1991, The effects of manning and crew stability on the material condition of ships, Interfaces 21, 111 120. Berger, M., and B. Hirsch, 1983, The civilian earnings experience of Vietnam-era veterans, Journal of Human Resources 18, 455-79. Berner, K., and T. Daula, 1993, Recruiting goals, regime shifts, and the supply of labor to the army, Defence Economics 4(4), 315-328. Binkin, M., and S. Bach, 1977, Women and the military (The Brookings Institution, Washington, DC).

Ch. 13: The Economics of Military Manpower

395

Black, M., R. Moffitt and J. Warner, 1990, The dynamics of job separation: the case of federal employees, Journal of Applied Econometrics 5, 245-262. Borjas, G., and F. Welch, 1986, The post-service earnings of military retirees, in: C. Gilroy, ed., Army manpower economics (Westview Press, Boulder, CO) 295 313. Brown, C., 1985, Military enlistments: what can we learn from demographic variation?, American Economic Review 75, 228-234. Browning, E., 1987, On the marginal welfare cost of taxation, American Economic Review 77, 11-23. Bryant, R., V. Samaranayake and A. Wilhite, 1993, The effect of military service on the subsequent civilian wage of the post-Vietnam veteran, Quarterly Review of Economics and Finance 33, 15-31. Buddin, R., 1988, Trends in attrition of high quality military recruits, R-3539-FMP (RAND, Santa Monica, CA). Buddin, R., and D. Grissmer, 1994, Skill qualification and turbulence in the army national guard and army reserve, MR-289-RA (RAND, Santa Monica, CA). Buddin, R., D. Levy, J. Hanley and D. Waldman, 1992, Promotion tempo and enlisted retention, R-4135FMP, The (RAND, Santa Monica, CA). Clark, R., 1978, Capital-labor ratios in a military service: a putty-clay application, in: R. Cooper, ed., Defense manpower policy (RAND, Santa Monica, CA) 11-23. Cooke, T., and A. Quester, 1992, What characterizes successful enlistees in the all-volunteer force: a study of male recruits in the U.S. Navy, Social Science Quarterly 73, 238 252. Cooke, T., A. Marcus and A. Quester, 1992, Personnel tempo of operations and navy enlisted retention, CRM 91-150 (Center for Naval Analyses). Cooper, R., 1975, The social cost of maintaining a military labor force, R-1758-ARPA (RAND, Santa Monica, CA). Dale, C., and C. Gilroy, 1985, Estimates in the volunteer force, American Economic Review 75, 547-441. Daula, T., and R. Moffitt, 1991, Estimating a dynamic programming model of army reenlistment behavior, in: C. Gilroy, D. Home and D. Smith, eds., Military compensation and personnel retention (US Army Research Institute, Alexandria) 181-201. Daula, T., and R. Moffitt, 1992, Estimating dynamic models of quit behavior: the case of military reenlistment (US Military Academy). Daula, T., and D. Smith, 1985, Estimating enlistment supply models for the U.S. army, in: R. Ehrenberg, ed., Research in labor economics (JAI Press, Greenwich, CT) 7, 261-309. Daula, T., and D. Smith, 1992, Are high quality personnel cost-effective? the role of equipment costs, Social Science Quarterly 73, 266-275. Dertouzos, J., 1985, Recruiter incentives and enlistment supply, R-3065-MIL (RAND, Santa Monica, CA). DeTray, D., 1982, Veteran status as a screening device, American Economic Review 72, 133-142. Enns, J., G. Nelson and J. Warner, 1984, Retention and retirement: the case of the U.S. military, Policy Sciences 17, 101-121. Fernandez, J., 1992, Soldier quality and job performance in team tasks, Social Science Quarterly 73, 253-265. Fernandez, R., 1982, Enlistment effects and policy implications of the educational assistance test program, R-2935-MRAL (RAND, Santa Monica, CA). Fisher, A., 1969, The cost of the draft and the cost of ending the draft, American Economic Review, 59, 239-254. Fletcher, J., J. McMahon and A. Quester, 1994, Women in the navy: the past, the present, and the future, Occasional paper (Center for Naval Analyses). Fredland, J., and R. Little, 1980, Long-term returns to vocational training: evidence from military sources, Journal of Human Resources 15, 49-66. Friedman, M., 1972, An economist's protest (Thom, New Jersey).

396

JT Warner and B.J Asch

Gilroy, C., T. Daymont, P. Andrisani and R. Phillips, 1992, The economic returns to military service: race-ethnic differences, Social Science Quarterly 73, 340-359. Goldberg, L., 1982, Enlistment supply: past, present, future, CNS 1168 (Center for Naval Analyses). Goldberg, L., 1991, Recent estimates of enlistment supply models (Economic Research Laboratory, Reston, VA). Goldberg, M., and J. Warner, 1982, Determinants of navy reenlistment and extension rates, CRC 476 (Center for Naval Analyses). Goldberg, M., and J. Warner, 1987, Military experience, civilian experience, and the earnings of veterans, Journal of Human Resources 22, 62-81. Gotz, G., 1990, Comment on the dynamics of job separation: the case of federal employees, Journal of Applied Econometrics 5, 263-268. Gotz, G., and J. McCall, 1980, Estimation in sequential decisionmaking models: a methodological note, Economics Letters 6, 131-136. Gotz, G., and J. McCall, 1984, A dynamic retention model for air force officers: theory and estimates, R-3028-AF (RAND, Santa Monica, CA). Gotz, G., and R. Roll, 1979, Defense resource management study supporting papers: first-term career mix of enlisted military personnel (RAND, Santa Monica, CA). Gotz, G., and R. Stanton, 1986, Modeling the contribution of maintenance manpower to readiness and sustainability, R-3200-FMP (RAND, Santa Monica, CA). Grissmer, D., and S. Kirby, 1988, Changing patterns of nonprior service attrition in the army national guard and army reserve, R-3626-RA (RAND, Santa Monica, CA). Grissmer, D., S. Kirby, M. Sze and D. Adamson, 1992a, Insuring mobilized reservists against economic losses, unpublished manuscript (RAND, Santa Monica, CA). Grissmer, D., S. Kirby and M. Sze, 1992b, Factors affecting reenlistment of reservists: spousal and employer attitudes and perceived unit environment, R-401 I-RA (RAND, Santa Monica, CA). Grissmer, D., S. Kirby, R. Buddin, J. Kawata, J. Sollinger and S. Williamson, 1994, Prior service personnel: a potential constraint on reserve forces, MR-362-OSD (RAND, Santa Monica, CA). Hammond, C., and S. Horowitz, 1990, Flying hours and crew performance, P-2379 (Institute for Defense Analyses). Hammond, C., and S. Horowitz, 1992, Relating flying hours to aircrew performance: evidence for attack and transport missions, P-2609 (Institute for Defense Analyses). Hansen, L., and B. Weisbrod, 1967, Economics of the military draft, Quarterly Journal of Economics 82, 395-421. Heckman, J., 1979, Sample selection bias as a specification error, Econometrica 47, 153 162. Hogan, P., D. Smith and S. Sylwester, 1991, The army college fund: effects on attrition, reenlistment, and cost, in: C. Gilroy, D. Horne and D. Smith, eds., Military compensation and personnel retention (US Army Research Institute, Alexandria) 317-354. Horowitz, S., and A. Sherman, 1980, A direct measure of the relationship between human capital and productivity, Journal of Human Resources 15, 67-76. Hosek, J., and C. Peterson, 1985, Reenlistment bonuses and retention behavior, R-3199-MIL (RAND, Santa Monica, CA). Hosek, J., and C. Peterson, 1990, Serving her country: an analysis of women's enlistments, R-3853-FMP (RAND, Santa Monica, CA). Kearl, E., D. Horne and C. Gilroy, 1990, Army recruiting in a changing environment, Contemporary Policy Issues 8, 68-78. Kester, J., 1986, The Reasons to Draft, in: W. Bowman, R. Little and T. Sicilia, eds., The all-volunteer force after a decade (Pergamon-Brassey's, New York) 286 315. Kirby, S., and D. Grissmer, 1993, Reassessing enlisted reserve attrition: a total force perspective, N3521-RA (RAND, Santa Monica, CA). Kleinman, S., and W. Shughart, 1974, The effects of reenlistment bonuses (Center for Naval Analyses).

Ch. 13: The Economics of Military Manpower

397

Kostiuk, P., and D. Follmann, 1988, Retention of navy veterans in the selected reserve, CRM 88-72 (Center for Naval Analyses). Kostiuk, P., and D. Follmann, 1989, Learning curves, personal characteristics, and job performance, Journal of Labor Economics 7, 129-146. Kostiuk, P., and J. Grogan, 1987, Enlistment supply into the naval reserve, CRM 87-239 (Center for Naval Analyses). Lee, D.R., and R. McKenzie, 1992, Reexamination of the relative efficiency of the draft and the all-volunteer army, Southern Economic Journal 59, 644-654. Mackin, P., P. Hogan and L. Mairs, 1993, A multiperiod model of U.S. army officer retention decisions (SAG Corporation, Falls Church, VA). Marcus, A., 1982, Personnel substitution and naval aviation readiness, Professional Paper 363 (Center for Naval Analyses). Marquis, S., and S. Kirby, 1989a, Reserve accessions among individuals with prior military service: supply and skill match, R-3892-RA (RAND, Santa Monica, CA). Marquis, S., and S. Kirby, 1989b, Economic factors in reserve attrition: prior service individuals in the army national guard and army reserve, R-3686-1-RA (RAND, Santa Monica, CA). Martin, T., 1994, Who stays, who leaves? an analysis of first-term army attrition, PhD thesis (RAND Graduate School, Santa Monica, CA). Mehay, S.L., 1991, Reserve participation versus moonlighting: are they the same?, Defence Economics 2, 325-338. Miller, J., ed., 1968, Why the draft? The case for the volunteer army (Penguin, Baltimore, MD). National Defense Research Institute, 1992, Assessing the structure and mix of future active and reserve forces: final report to the Secretary of Defense, MR-140-1-OSD (RAND, Santa Monica, CA). Nelson, G., 1986, The supply and quality of first-term enlistees under the all-volunteer force, in: W. Bowman, R. Little and T. Sicilia, eds., The all-volunteer force after a decade (Pergamon-Brassey's, New York) 23-51. Oi, W., 1967, The economic cost of the draft, American Economic Review 57, 39-62. Orvis, B., and M. Gahart, 1989, Quality-based analysis capability for national youth surveys, R-3675-FMP (RAND, Santa Monica, CA). Orvis, B., M. Childress and M. Polich, 1992, Effect of personnel quality on the performance of patriot air defense system operators, R-3901-A (RAND, Santa Monica, CA). Payne, D., J. Warner and R. Little, 1992, Tied migration and returns to human capital: the case of military wives, Social Science Quarterly 73, 324-339. Polich, M., J. Dertouzos and J. Press, 1986, The enlistment bonus experiment, R-3353-FMP (RAND, Santa Monica, CA). Quester, A., 1988, Non-prior-service male and female recruits: historical comparisons of continuation, promotion/demotion, and desertion rates, CRM 89-314 (Center for Naval Analyses). Quester, A., and A. Adedeji, 1991, Reenlisting in the marine corps: the impact of bonuses, grade, and dependency status, CRM 91-64 (Center for Naval Analyses). Quester, A., and M. Nakada, 1983, The military's monopsony power, Eastern Economic Journal 4, 295-308. Report of the President's Commission on an All-Volunteer Armed Force, 1970, (Collier-Macmillan, London). Ridge, M., and R. Smith, 1991, UK military manpower and substitutability, Defence Economics 2, 283-294. Rosen, S., 1986, The theory of equalizing differences, in: O. Ashenfelter and R. Layard, eds., Handbook of labor economics (North-Holland, Amsterdam) 641-692. Rosen, S., 1992, The military as an internal labor market: some allocation, productivity, and incentive problems, Social Science Quarterly 73, 227-237. Rosen, S., and P. Taubman, 1982, Changes in life-cycle earnings: what do social security data show?, Journal of Human Resources 17, 321-338.

398

JT Warner and B.J Asch

Scribner, B., D. Smith, R. Baldwin and R. Phillips, 1986, Are smarter tankers better? AFQT and military productivity, Armed Forces and Society 12, 194-206. Shiells, M., and J. McMahon, 1993, Effects of sea duty and advancement on first-term retention, CRM 92-205 (Center for Naval Analyses). Smith, D., S. Sylwester and C. Villa, 1991, Army Reenlistment Models, in: C. Gilroy, D. Horne and D. Smith, eds., Military compensation and personnel retention (US Army Research Institute, Alexandria) 43-173. Tan, H., 1991, Non-prior service reserve enlistments: supply estimates and forecasts, R-3786-FMP/RA (RAND, Santa Monica, CA). Tollison, R., 1968, Racial balance and the volunteer army, in: J. Miller III, ed., Why the draft? The case for a volunteer army (Penguin Books, Baltimore, MD) 149-158. Warner, J., 1990, Military recruiting programs during the 1980s: their success and policy issues, Contemporary Policy Issues 8, 47 57. Warner, J., and M. Goldberg, 1984, The influence of non-pecuniary factors on labor supply: the case of navy enlisted personnel, Review of Economics and Statistics 66, 26-35. Warner, J., and G. Solon, 1991, First-term attrition and reenlistment in the U.S. army, in: C. Gilroy, D. Home and D. Smith, eds., Military compensation and personnel retention (US Army Research Institute, Alexandria) 243-280. Withers, G., 1978, International comparisons of manpower supply, in: R. Cooper, ed., Defense manpower policy (RAND, Santa Monica, CA) 116-136. Xie, Y., 1992, The socioeconomic status of young male veterans, 1964-1984, Social Science Quarterly 73, 379-396.

Chapter 14

THE DEFENSE INDUSTRIAL BASE J. PAUL DUNNE University of Leeds

Contents Abstract

400

Keywords

400

1. Introduction 2. Defining the defense industrial base 3. The defense equipment market 4. The military industrial complex 5. The structure and evolution of the DIB 6. Efficiency and economic effects

401 402 406 409 411 422

7. Developments in the DIB

424

8. Conclusions

426

References

427

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T. Sandler © 1995 Elsevier Science B. V All rights reserved

400

JP Dunne

Abstract This chapter provides a survey of research on the defense industrial base (DIB) focusing on the advanced industrial capitalist economies. It starts by looking at the problems of definition and measurement and how these have been dealt with in practice. This is followed by an overview of the nature of the defense equipment market and the concept of the military industrial complex. An analysis of the evolution and the structure of the DIB is then presented and the evidence on the efficiency of the DIB and its economic effects are reviewed. Finally, the restructuring of the DIB that has taken place since the end of the Cold War and the likely future developments are discussed.

Keywords Defense industrial base (DIB), military industrial complex, defense equipment, efficiency, economic effects, restructuring

Ch. 14:

The Defense Industrial Base

401

1. Introduction This chapter will consider the Defense Industrial Base (DIB), a concept which has common usage in debates over the economics of military spending, but, as shall be seen, a rather ephemeral character. It is generally accepted that many of the major economies have a national DIB in the sense that they have a sector or groups of industries that are dependent to some degree on defense spending and upon which the state is dependent for some degree of self sufficiency in the production of the means of defense and war. The major companies in the DIB are often easily identifiable, but trying to analyse the full extent of the DIB leads to considerable conceptual and practical problems. This is evident in the variety of theoretical and empirical analyses of the DIB and the way they differ in their assessment of its scope, role, and importance. While many studies focus on the purely functional role of the DIB, its efficiency in producing military related products and the importance and affordability of self sufficiency, others stress more general aspects of the DIB. They see the DIB companies not simply as passive suppliers of weapons systems but as active participants in the determination of the level of defense spending, as an important part of a set of vested interests which make up the " Military Industrial Complex" (MIC). In this way the DIB can have influence and effect well beyond that of providing the means for defense of the realm. Both the MIC and the DIB have their roots in the mechanization of weaponry and war towards the end of the last century, but an important spur to their development has been the Cold War and the unprecedentedly high peacetime military spending associated with it. With the end of the Cold War the military industries are facing declining demands for their products in increasingly competitive markets. This has led to a profound restructuring and downsizing of the companies and changes in the relations between the companies and the government. The trends towards collaboration and internationalization of defense companies may mean that the idea of a national defense industrial base may soon be irrelevant (see Chapter 16 of this Handbook). Understanding the nature of the DIB and the transformations it is undergoing is an important task. The DIB is an important part of many advanced economies, and if not handled properly downsizing can cause economic problems, which can fuel opposition to further disarmament. Companies can also respond to reduced domestic demand by exporting arms to volatile areas and the internationalization of the DIB reduces the control national governments have over it. Such developments present clear dangers to world peace and could prevent the move to a post-Cold War environment with reduced military burdens. The rest of this chapter will attempt to provide an understanding of the nature and extent of the DIB and the developments taking place in it. To this end Section 2 considers the not inconsiderable problems of trying to define the DIB. This is followed by a discussion of the problems of the nature of the defense equipment market in which the DIB operates. When the DIB is not simply treated as a passive actor in

J.P? Dunne

402

the dynamics of defense policy making, the interaction with the rest of society and the proactive activities of vested interests have to be considered: the DIB becomes the Military Industrial Complex. This is discussed in Section 5, followed by an analysis of the evolution and structure of the DIB in Section 6. Section 7 then considers the issues surrounding the efficiency and economic impact of the DIB, and Section 8 outlines the developments that have been taking place with the restructuring at the end of the Cold War. Finally Section 9 presents some conclusions. 2. Defining the defense industrial base At a superficial level the idea of a Defense Industrial Base is fairly straightforward. It constitutes those companies which provide defense and defense related equipment to the defense ministry. But if we try to operationalise this definition there are many problems. Defining the defense industry itself is not straightforward, as the range of products involved can be very wide. Weapons differ greatly, from large technologically advanced and expensive weapons systems to inexpensive small arms, and there are numerous other more general commodities consumed by the military. One useful way of classifying the DIB products is to consider their relation to military action or warfare. There are: (i) Lethal large or small weapons systems. (ii) Non-lethal but strategic products (e.g. vehicles and fuel). (iii) Other products consumed by the military (e.g. food and clothing). It is important to recognize that this hierarchy does not necessarily reflect importance to the military or to the production of security. Most weapons systems could not operate without the strategic products, fuel and transport, and soldiers could not survive without food. The companies that make up the DIB will also differ, both in the degree of their dependence on military production and in their importance to the DIB. There is no reason why these two should be positively related. Some large diversified companies may only consider their DIB activities as marginal but may be vital producers of particular weapons systems, while some smaller companies may be wholly dependent on military related orders but not important suppliers. We could classify companies by the following taxonomy: Dependence on DIB Importance to DIB

High

... ·

Low

Low The closer a company is to the top left hand comer the stronger its claim to DIB membership. Companies would range from the high dependence high importance ones

Ch. 14:

The Defense Industrial Base

403

(e.g. VSEL in the UK, Northrop in the USA, GIAT in France), through low dependence high importance (e.g. GEC in the UK, McDonnell Douglas in the USA, Daimler Benz in Germany), through the high dependence but relatively low importance (subcontractors who make very specific but relatively easily sourced components for the main contractors), through to the companies which make relatively low importance products, such as paper towels, and have little dependence on the military. Of course the importance of a product to the military may not necessarily be reflected in its monetary value. Some researchers have focused upon the technological aspects of the DIB. Walker et al. (1988) and Schofield (1993) suggest that it is important to recognize that the prime defense contractors are systems manufacturers integrating a variety of subsystems into a complete final product. There is then a hierarchy of products from systems to low level components which have different generic and specific characteristics and a decreasing differentiation between military and civil products. Indeed, many of the low level technologies are "dual use". The hierarchy they suggest is:

(i) (ii)

Military strategies and concepts (high level planning) Integrated weapon and information systems (e.g. national early warning systems) (iii) Major weapon platforms and communication systems (e.g. aircraft, battleships, etc.) (iv) Complete weapon and communications component parts (e.g. torpedoes) (v) Sub-systems (e.g. gyroscopes) (vi) Sub-assemblies (e.g. sights, fuses) (vii) Components (e.g. integrated circuits) (viii) Materials (e.g. semi-conductors). These categories range from the high unit cost, complex systemic integration products, with long product life, at the top to low unit cost mass produced products at the bottom. At any level the characteristics of products will also differ depending on whether they are custom made or standard. This taxonomy is useful in highlighting the difference in civil and military products and so suggesting what constitutes the core of the DIB, but as with the other taxonomies, it does not say where to draw the line in defining the DIB. Even if we were clear on what constituted the DIB there are problems involved in measuring it. A direct measure of the DIB companies will tend to understate its importance as there are many sub-contractors who are dependent to different degrees on military-related orders but through the intermediary of other contractors. Indeed, some firms will not know they are part of the DIB if they are making intermediate products or components (e.g. ball bearings). There are also important distributional issues as many defense producers provide the main industrial presence in outlying regions. The definition of what constitutes military production is not always straightforward and can differ across countries. The production and export of weapon components (e.g.

404

1J Dunne

the UK-Iraqi supergun affair), anti-insurrection and riot control equipment, razor wire and torture and other such equipment will not necessarily be classified as military. Such problems are being exacerbated by the tendency for governments to put considerable emphasis on the development of dual use technologies, which blur the military-civil distinction. It is also not uncommon for defense related work to be found under civil headings. Walker (1988) suggested that the UK dependence on military production tends to be considerably underestimated, particularly in the electronics industry. Given all of these problems, how the DIB is defined in any particular study is likely to depend on the questions being asked, how the information is to be used, and the resources available. If the concern is the capability of the UK to produce the range of military equipment necessary for super power status then the major defense producers need to be considered. If concern is with both the production of weapons and the protection of strategically important aspects of the economy we would include many more. Indeed, in the interest of maintaining a credible defense of its neutrality Sweden had a policy of maintaining a strong domestic production capability in all strategic industries: this even included the clothing industry [Thorsson (1984)]. When the definition of the DIB is this wide its maintenance can look remarkably like a form of industrial planning. So far we have focused on the domestic industry, but it is also possible to see foreign suppliers as part of the DIB. To illustrate this Hartley (1991) suggests two extreme definitions of the DIB at the opposite ends of the policy spectrum. The "free market" definition of the DIB, which would include only those UK companies which have a comparative advantage, and a "fortress Britain" approach in which all aspects of military capability are included. In the free market case the government would simply take the best deal and if UK companies happened to provide it all well and good. In between these extremes lie the other definitions such as all firms receiving MoD contracts, or a minimum core of national assets, with some level of self sufficiency. There has been considerable debate over the importance of self sufficiency in arms production. The security advantages are presented as the ability to produce in house all advanced means of warfare and to be able to increase production quickly in times of crisis and war. For example, Gansler (1980) argued that the problems of the DIB in the USA were the result of a failure to recognize the need for proactive policies by government. He made the point that the defense market operated far from a competitive ideal and attempts to use free market policies piecemeal in individual cases and sectors had led to "second best" problems, making the DIB inefficient in the production of defense materiel and unresponsive in terms of production speed up to meet an emergency. With the increasing sophistication and costs of advanced weaponry, however, maintaining a national capability across the whole range of weapons does not come cheap. It is also not clear that it is an advantage or a realistic option. As Hartley et al. (1987) argue self sufficiency may not be important for the UK given the degree of collaboration already underway and the likely nature of future conflicts. Also, a complete national DIB in any country is still likely to require imports of components.

Ch. 14:

The Defense Industrial Base

405

Even the USA is now dependent on Japanese components for the production of advanced weapons systems. This is underlined by Ratner and Thomas (1990) who argue that rather than dichotomizing the imports/self sufficiency arguments planners should look at the vulnerability of different products to disruption (e.g. some are made by safe countries, as Canada is for the USA, some are easily substituted). It should be possible to specify scenarios, work out the risks involved, and design an optimal procurement policy which minimizes supply disruptions and does not simply reject the possibility of imports. A failure to recognize this has tended to lead governments to see "defense industry deficiencies" and to be biased towards subsidizing relatively inefficient domestic companies when safe alternatives were available. It is also important to recognize that the DIB has tended to fulfil a role beyond providing the means of warfare. During the Cold War the main justification used for maintaining an increasingly expensive DIB was the security from the Soviet threat. However, arguments that the DIB provided skilled employment and technological progress through spin-off were also given considerable weight. The arguments that the DIB simply crowded-out private sector investment and employment were generally ignored. Certainly the DIB had an important role in government industrial and technology policy (especially in the USA, France and the UK) whether explicit or implicit. These factors mean that any evaluation of the role and nature of the DIB has to be more wide ranging than merely a consideration of self sufficiency. In the post-Cold War period the emphasis has in fact shifted from arguments that it is necessary to maintain complete production facilities to the possibility of maintaining strategic autonomy through maintaining the technological capability to produce advanced weapons systems. This can be through encouraging the development of "dual use" technologies, which are intended to allow the defense contractors to diversify into civil markets and make them less dependent on defense procurement, but still maintain weapons production capabilities. But dual use technologies may simply saddle companies with restrictions which make them less competitive in the civil market and increase their problems of adjusting to cuts in procurement. In addition, there are alternatives. Technological capability can be maintained through licensing, collaboration and joint ventures and it is to these last two options that many major defense contractors have been turning [see Coopey et al. (1994)]. The problems of defining the DIB inevitably lead to problems of measurement. These are compounded by the secrecy surrounding the information on defense companies in many countries. This secrecy may be the result of government sensitivity or firm sensitivity. In companies that are not solely defense producers it is often difficult to get a breakdown of the defense and civil work and its contribution to profits, etc. This can be because of the difficulty of accounting joint costs, but even in companies where defense divisions are kept separate any cross subsidization of civil production by lucrative defense contracts is likely to be hidden. It is also difficult to obtain data on the employment and skill composition of the labor force, especially on the employment of qualified scientists and engineers because of the secrecy surrounding research [Buck

406

JP Dunne

et al. (1993)]. In the USA the private defense firms are legally required to provide information and the government has made relatively explicit use of the defense sector as a tool of industrial policy, meaning that more information is available than in other countries [Cypher (1987)]. In contrast, in the UK public ownership (pre-1977) and generally secrecy has limited the information available. However, privatization, takeover bids, and recent changes in contractual arrangements have made the defense sector more transparent [Dunne and Smith (1992)]. To measure the full extent of the DIB in terms of employment would require identifying all firms involved in defense production, including the supply chain of subcontractors. This is difficult because it can be difficult to identify the proportion of the workforce involved in defense work, even in the prime contractors, and the problem becomes worse when firms are involved in the production of dual use technologies. In addition, defense production will have an induced multiplier effect on local economies through consumption of the workers and this is difficult to identify [Paukert and Richards (1991) provides contributions which cover these issues; see also Chapter 17 of this Handbook and Dunne (1986)]. Overall, it would appear that the DIB is not a simple well defined and easily measured concept. In practice the definition of the DIB used in a particular study is likely to be dependent on the questions being asked and the likely use of the information, while data and resource availability are likely to determine the scope of measurement. With any definition of the DIB, however, there are characteristics of the market for defense equipment which make it different from most civil markets. The next section considers these characteristics.

3. The defense equipment market The determining feature of the defense market is the monopsonistic role of the government. In such a situation the government is the prime moving force in determining both the demand side of the market and much of the supply side, though the final form of the DIB is also historically contingent. The government can influence the size of the industry, its structure, entry and exit, prices and profits, efficiency, ownership and level of technology. Its decisions on what to buy, where to buy it, how to buy it, and the importance of wider policy objectives will be paramount, while its policies of protection, support, financial assistance, infrastructural and capital investment, education, and ideology will also be important. Of course, in principle the DIB companies although tied to government procurement also have the option of exporting but in practice this is normally regulated by the government [Smith (1990)]. As we shall see it was the high military expenditures post-World War II which encouraged corporate involvement in lucrative defense orders, while the high R&D expenditures influenced the structure and performance of the companies. The high R&D costs in turn influenced the trend in costs, making them higher than civil costs

Ch. 14:

The Defense IndustrialBase

407

and the nature of production, small batch, technologically advanced and concerned with performance rather than cost minimization, limited the extent of economies of scale and learning that could be achieved. All of these factors influenced the size and structure of the DIB and how it developed over the post-1945 period [Baack and Ray (1985)]. While other large companies are similar in structure the products they produce and the sub-systems they integrate will have a different technological forms and requirements embodied in them. The further one moves up the hierarchy the more different the civil and military products and production processes will be, meaning that the nature of capital equipment, labor skills, and the organization of production become increasingly specific to the sector [Hooper and Buck (1991)]. This still leaves the problem that many important defense companies have large civil operations but evidence suggests that the different requirements of the defense market and the particular social relations of production they engender have led companies to keep their defense divisions very separate from their civil divisions [see Maddock (1983), Southwood (1991)]. The general characteristics of defense production which result from the monopsonistic structure of the market are: (i) An emphasis on performance of high technology weaponry rather than on cost [Kaldor (1991)]. (ii) Risk borne by government who often finance R&D and in some cases provide investment in capital and infrastructure. (iii) Elaborate rules and regulations on contracts. This is necessary in the absence of any form of competitive market and to assure public accountability [Peck and Scherer (1962)]. (iv) Close relations between contractors, procurement executive and military. The "revolving door" is common, which sees military and civil servants moving to defense contractors they had dealings with and defense contractors moving into the bureaucracy. (v) Outside of the USA many companies will be national monopolies or close to it, meaning that any introduction of competition will need to be in the form of foreign firms. This can be implicit rather than explicit, by creating contestable markets with potential competition from abroad, although there may be problems of proving this a credible threat [Baumol et al. (1982)]. These characteristics will tend to favor those firms who specialise in defense work. They will know their way around the red tape and will have the contacts. They will become experts at getting money out of government, rather than being successful in commercial markets. The companies seek involvement in the development programs for technologically advanced weapons systems as the best means of obtaining the subsequent production contracts. This can lead to 'buy ins', where firms understate risk or cost to win initial contracts, making up the losses later. In addition, past programs have seen 'gold plating' where the military continually ask for extras or continuous technological improvements over the contract period. This allows

408

.R?Dunne

renegotiation of contracts or additional payments, usually to the advantage of the contractor [Laffont and Tirole (1993) provide a theoretical overview of many of these issues]. As a result of the structure of the market there are both barriers to entry and barriers to exit, which have led, until recently, to the DIB showing remarkably stability in terms of its composition of main contractors [see Gansler (1980)]. These barriers can be categorized as market, technological and procedural. The barriers to entry are: (i) Market. The marketing of military products is different to commercial products as personal contacts and networking are likely to be more important than general advertising. Market demand for arms is also limited by the government and is likely to be inelastic. This means that entrants cannot rely on an expansion of the market to accommodate them, as prices are reduced, but are likely to have to fight against and replace incumbents. There is also likely to be considerable brand loyalty given the nature of the products. Customers may require compatibility with previously purchased weapons systems, or may provide follow-on orders from previous contracts. (ii) Technological. The nature of defense production promotes a concern for technical characteristics and so proven capability is important to the buyer and difficult for potential entrants to prove. Many projects will need to be taken through the development stage to production and this will require a highly skilled workforce and expensive specialized capital equipment. Management styles will also need to be different to civil production. (iii) Procedural. Political considerations are likely to influence the winning of large contracts, and personal contacts that are built up over time are likely to be important. It is likely that mastering the detailed procedural regulations and accounting practices that have been developed will be difficult for the newcomer. In addition, larger contractors may not play by the rules (e.g. they may "buy in", putting in very low bids at the beginning of the contract in the expectation of either being able to make money later when the inevitable specification changes are requested or by winning future contracts). Also new firms need security clearance and this can be expensive and time consuming to obtain. These barriers will effect both defense sub-contractors trying to move to the top league and other large civil contractors trying to break in. There have been some attempts to break these barriers in the USA and the UK by changing contractual arrangements, paying competing firms to produce prototypes before deciding on the winner, and by allowing foreign competition. In fact the threat of outside competition is usually intended to create contestable markets and make domestic suppliers more efficient, rather than to replace them. This does require that the threats are credible to prevent complacency. This policy was seen clearly in the UK where the government did open up the defense markets as part of the policy of reform. Foreign suppliers replaced UK contractors in, for example, the cancellation of Nimrod and the purchase of US AWACS.

Ch. 14:

The Defense Industrial Base

409

The barriers to exit are: (1) Market. The marketing of civil products and the need to compete on price make the civil market place very different to the world of defense companies. Defense contracts can be safe and profitable and often long term commitments. The market is cyclical and even in lean spells it may be worth staying in the market in the expectation of better times, particularly as the government is still likely to bail out major contractors in trouble. In addition, even when there are cuts in domestic sales governments are happy to provide assistance for foreign sales (including providing contacts, VIP visitors, export credits and assistance in the country), and these can be lucrative. (2) Technological. Government defense R&D sponsorship is valuable to companies and not something they would want to lose, with some firms using governmentowned plant and equipment. In addition, the specialist skills and capital equipment are difficult (or impossible) to convert to civil production. The management techniques required in civil production are also very different and the emphasis on quality regardless of cost rather than quantity and low cost may be difficult to alter. (3) Procedural. Major defense contractors have become successful at handling the complex procedures and getting money out of government. Having developed these skills they will not want to give them up. In addition, some firms may also fight to stay in a declining defense market because they have a sense of duty to the nation, in producing the means of defense, rather than for commercial motives. These factors mean that not only has it been difficult for companies to enter into the defense sector to produce weapons systems, or to upgrade from sub-contractor status, but also that it is difficult for the defense companies to leave the industry [Gansler (1980, 1989)]. 4. The military industrial complex So far we have discussed the DIB as a passive component of national security policy, molded by government policy. In doing so we are implicitly arguing that national governments make decisions about the need for defensive and offensive capabilities, decide the best way to achieve these in terms of force structures and weapons procurement, and then decide on the form of DIB best suited to achieving these ends. This can be seen in the literature both on input budgeting and program or output budgeting, where the problem is seen as finding the most efficient means of producing particular outcomes, taking a national security perspective [see Sandler and Hartley (1995), Rogerson (1994), and Chapter 12 of this Handbook on procurement for a discussion of these issues]. In many ways this reflects the neo-classical approach to military expenditure which is based upon the notion of a state with a well defined social welfare function, reflecting some form of social democratic consensus, recognizing some well defined national interest, and threatened by some real or apparent potential enemy. In this approach the

410

J.P.F Dunne

DIB would simply be determined as the most efficient way of producing the optimal level of security. The problem with this perspective is that the size and importance of the DIB within many countries, particularly the USA, has inevitably led to it linking into other parts of society and the economy. Once we move beyond seeing the DIB as a passive capability to provide weapons systems and recognize the fact that it may have proactive tendencies (rent seeking and trying to capture the regulators) these linkages become important. More recent neo-classical literature has addressed these problems and attempted to integrate political factors, such as bargaining and interest groups, that determine sources of weapons and levels of protection [e.g. Grossman and Helpman (1994)]. But this still represents a partial analysis focusing on particular aspects of the problem. It does not address the complex dialectical interaction between the demand side and the supply side in which both will influence each other and set the parameters for decision making and the outcome will be a complex dynamic process, which can be both contradictory and conflictual. More general analyses locate the DIB firmly within the context of the wider military industrial complex (MIC). This term was introduced by Dwight Eisenhower, an exmilitary Republican President of the USA. He was concerned that coalitions of vested interests within the state and industry could lead to decisions being made which were in the interest of the coalition members and not necessarily in the interests of national security. These coalitions could include some members of the armed services, of the civilian defense bureaucracy, of the legislature, of the arms manufacturers and of their workers. From the point of view of someone like Eisenhower this was a case where the interests of capitalism [discussed in Smith (1977)] might diverge from the interests of capitalist defense firms. Indeed, Schwarz (1990) in a detailed historical analysis of the development of the MIC uses the term 'military industrial congressional complex' to emphasize the importance of political linkages. In addition, Schwarz points out that the development of the MIC in the USA was linked to the New Deal economists and their attempts to introduce economic planning. Following Smith (1977) we can characterise this approach as the liberal or institutional approach. It hinges on the nature of a 'military industrial complex' composed of conflicting interest groups and institutional linkages. The MIC becomes a self generating structure (agency) which embodies the interests of various groups in society. The strength of the vested interests and their competition for resources, leads to internal pressures for military spending, with external threats providing the justification. For this approach there is a national interest but it is distorted by vested interests. The MIC imposes a burden on the rest of society and has adverse effects on the civilian sector. It crowds out civilian resources, and the companies involved develop a culture which leads to inefficiency and waste and an increasing reliance on defense contracts as they become less able to compete in the civilian market [see Melman (1970, 1971, 1985) and Dumas (1986)]. The theoretical underpinnings of this work were originally based on C. Wright Mills' 1956 analysis of the power elite, but there are also variants which follow a Weberian

Ch. 14: The Defense Industrial Base

411

focus on the role of bureaucracy and the work of Galbraith on coalitions [see Slater and Nardin (1973) and Moskos (1972)]. There is also a considerable amount of work undertaken from a Marxist perspective. While this approach is often typified as a focus on ruling class concept of the MIC [see Brunton (1988)] it is much richer and more varied than this suggests. The role of military expenditure in the development of capitalism is seen as much wider and more pervasive than in the institutional approach, but with the military industrial complex constrained by the laws of motion of the capitalist system. Within the Marxist approach there are a number of strands which tend to differ in their treatment of crisis and in the extent to which they see military expenditure as necessary for capital accumulation [see Dunne (1990)]. The underconsumptionist approach developed from the work of Baran and Sweezy (1966) sees military spending as important in preventing realization crises as it allows the absorption of surplus without increasing wages, unlike other forms of government spending, and so maintains profits. In this way the MIC provides a valuable service to maintaining capitalism. A similar perspective, but one which focuses on the tendency for capitalist economies to overproduce, is the permanent arms economy [Kidron (1968)]. In this theory military expenditure is wasteful and pushing resources into it prevents overheating. Thus the inefficiencies of the MIC and the DIB can be seen as positive factors and the development of the MIC and the DIB play a positive role in capitalist development [see Howard and King (1992)]. Empirical work, starting with Smith (1977) has, however, failed to find support for the underconsumptionist approach and its prediction of a positive economic effect of military spending. Pivetti (1992) and Smith and Dunne (1994) provide recent contributions to the debate. There is no clear conceptualization of the MIC. Indeed, the concept appears to be most of value as a descriptive rather than an analytical concept [see Fine (1993)]. This has led some researchers to emphasize the changing nature of the MIC and the need to focus on its dynamics at an empirical level. Smith and Smith (1983) argue that the MIC should be seen as a coalition of interests and that the focus should be on the structural pairings that have developed between particular sections of private industry and particular parts of the military which have inevitably led to mutual interests. In contrast, Brunton (1988) argues that the MIC should be seen as an evolving system of institutions rather than focusing on individual components. There clearly is a set of vested interests which influence policy on military spending - a MIC - in the same way as there are vested interests that influence policy decisions in health and education and other such areas. There is no conceptual difference between the MIC and other industrial complexes, but there are important differences in detail, in particular in the way the DIB operates, and it is to these we now turn. 5. The structure and evolution of the DIB The DIB has been in existence at least since the end of the 19th century. It was the result of the rise of capitalist industry which revolutionized weapons technology, with

412

JP Dunne

mass production and mechanization. At this time, the arms industry was markedly international in orientation and was dominated by a handful of large companies all of whom did well in the rearmament for the First World War. At the end of the war the companies formed cartels to carve up the international market in the face of slack demand. This continued until the 1930s when governments in the major powers became more inward looking and made conscious efforts to build up indigenous arms production capabilities. In the UK and the USA this took the form of direct government involvement in a set of core industries, with the institutions of the New Deal playing a major role in US war mobilization [Schwarz (1990)1. The Second World War then saw the capitalist economies militarized to an unprecedented degree and the DIB in its "national" form was established [Lovering (1990, 1993)]. With the end of World War II there was a massive reconversion of plants back to civil production, but the new technologies developed during the war suggested the need for new specialized industries. In particular, the importance of air power in the war had spurred technological developments which, in the post-1945 period, suggested the need for both civilian and military developments in the aircraft industry. There was a growing civil market and a large scale development of jet propelled fighters and it became necessary for the state to involve itself to maintain and develop capabilities [Edgerton (1991)]. The devastation of World War II (after the war, of the advanced economies only the USA, the UK and Sweden had their defense industries largely intact) and the onset of the East-West confrontation underlined the need for the development of national capabilities in arms research, development and production. This led to considerable turbulence in the defense industries, especially in the USA [see Reppy (1993), Gansler (1980)] with the entry of new firms ready to develop and produce the new weapons of the missile age. In the 1960s things settled down, although the speed of technological development continued and there was a shift towards computer-related technologies and a growth in the ability of prime contractors to internalise technological change. Rather than new entrants, large contractors internalized technological developments and developed ever closer links with the military. R&D complexes, funded by the Department of Defense (DoD) were formed in the USA. The barriers to entry increased, with increasingly complex procurement regulations and accounting procedures, while the barriers to exit were increased by the same factors. As a result the list of major defense contractors varied surprisingly little since the 1960s until the recent restructuring. There were certainly some mergers and exits but much less than among non-defense contractors. There was less turbulence and more government involvement in other countries, particularly the UK, mainly because of the smaller size of their markets, but a similar pattern over time was apparent. This changed in the UK in the 1980s with the privatization of the main defense contractors and then the introduction of competition policy in the mid-1980s. The thawing out and end of the Cold War also saw a final end to the stability of the defense industry. Cuts in procurement expenditures and changes in government policy have led to a marked restructuring which we will consider in more detail later [Wulf (1993c)].

Ch. 14:

413

The Defense Industrial Base Table 1 NATO's defense budgets

Country

Belgium Canada Denmark France Germany Greece Italy Luxembourg Netherlands Norway Portugal Spain Turkey UK USA NATO Europe NATO total a

b

Defense as percentage of GDP Average, 1992 1985-1989 2.9 2.1 2.1 3.8 3.0 6.2 2.3 1.1 2.9 3.2 3.2 2.2 3.3 4.6 6.3 3.3 4.7

Source: NATO. Equipment expenditures include R&D.

1.9 1.9 2.0 3.4 2.2 5.6 2.0 1.2 2.5 3.4 3.0 1.6 3.9 3.8 5.2 2.7 3.7

% equipment expenditures b in defense Average, 1992 1985-1989 12.1 19.7 14.0 n.a. '

8.2 18.6 17.8 n.a.

19.6 18.2 19.7 3.5 19.8 21.7 7.6 n.a. 18.2 24.8 25.6

13.3 23.4 15.0 4.6 14.2 24.4 2.2 10.9 24.8 18.1 22.9

' n.a.: not available.

On the demand side the USA is clearly dominant, both in terms of the size of its domestic market and the level of its military expenditure. As Table 1 shows the USA has by far the highest defense expenditure as a share of GDP in NATO, with the European military economy dominated by France, Germany and the UK, who spend almost 70 percent of the total for all European NATO countries. As far as the DIB is concerned the major concern is the trends in procurement expenditure, and again the USA is dominant: indeed it spends more than twice as much on the procurement of major weapons systems as all European NATO countries put together, including France. Also, while the real military expenditures in France and Germany were of similar magnitudes to the UK, the UK had much the highest procurement expenditure. The pattern over time shows a general, more pronounced, increase in the early 1980s and a general decline in expenditure in the late 1980s. This shrinkage in demand was combined with a declining and increasingly competitive international arms market (see Chapter 18 of this Handbook). An important aspect of military spending is Government funded defense R&D, and as Table 2 shows this is rather irregularly distributed among the six largest R&D spenders in the OECD. In 1990 the USA was, and continues to be, by far the largest

J.P Dunne

414 Table 2 Research and Developmenta Military R&D in 1990 Country

R&D expenditure (millions of £)

Percentage of total government R&D 62.6 5.4 40.0 13.5 6.1 44.0

24294 324 3145 991 255 2183

USA Japan France Germany Italy UK R&D as percentage of GDP Country

1967

1975

1985

1988

1990

Defense R&D USA Japan France Germany Italy UK

1.01 0.01 0.54 0.21 0.02 0.60

0.63 0.01 0.35 0.14 0.01 0.64

0.85 0.02 0.46 0.14 0.08 0.67

0.83 n.a. 0.52 0.13 0.08 0.51

0.74 0.02 0.57 0.14 0.05 0.40

Industry financed R&D 1.01 USA 0.83 Japan 0.61 France 0.94 Germany 0.35 Italy 1.00 UK

1.01 1.12 0.69 1.12 0.47 0.80

1.35 1.84 0.94 1.58 0.58 0.96

1.38 1.95 0.96 1.78 0.54 1.06

a

Source: OECD and UK Government Statistics.

spender on military R&D. It allocates some 13 percent of its defense budget to R&D, a figure which reflects a growth of 80 percent between 1980 and 1990, mainly due to the Strategic Defense Initiative (SDI). Despite some cuts in the late 1980s the US defense R&D still remains high, four times as high as the total for the other countries in 1990. At the other extreme, Japan has by far the smallest allocation of R&D resources to defense, which is not surprising given its commitment to limiting its military spending to less than 1 percent of GDP. In 1990 just over 5 percent of government R&D spending was military related in Japan, representing only 0.02 percent of GDP. However, Japan and to a lesser extent Germany do pay for R&D via licensed production of US equipment (e.g. the F15 fighter).

Ch. 14:

The Defense Industrial Base

415

The UK, France, and Germany all spent about the same amount on defense in absolute terms, but the expenditure patterns for R&D are very different. The share of GDP devoted to R&D was considerably higher in Germany, but the share of defense R&D was considerably lower than in both France and the UK. Only 13.5 percent of government R&D was defense related in Germany, while the figures for the UK and France were 44 percent and 40 percent, respectively (there are suggestions that some defense R&D work is hidden under other categories in Germany). In addition, in the UK and France nuclear R&D is an important element of the high military R&D expenditure. Between 1967 and 1988 there is some evidence of a slight increase in industry financed R&D as a share of GDP in most countries. In the late 1980s there does seem to have been a decline in defense R&D as a share of GDP allocated to R&D, but rather less than expected given the cuts in spending. There is also some evidence that, while the Cold War arms race may be over, the dynamic of weapons development may be continuing [see PRODEM (1993)]. In 1988 SIPRI started to collect information on the world's 100 largest firms. Until then there were no consistent data sources on the international DIB, only collections of case studies of particular countries [e.g. Ball and Leitenberg (1983)]. While these were valuable, the creation of a consistent data source allows a better investigation of the changing structure of the DIB. Using this source, Table 3 shows the top 100 defense companies in 1991, by country, and gives their value of arms sales, arms sales as a share of total sales, and employment in 1991. Some care must be taken in interpreting the data as there are problems of reliability: some companies report defense related work while for others it has had to be estimated. One striking feature of this list is the degree to which the large defense firms have maintained their dominance. The major contractors are generally the same companies who would have been in the top twenty 10 or 20 years ago. It is also noticeable that the European national champions are quite small relative to their American competitors: in 1988 the combined arms sales of the 100 top European companies was about the same as the those of the top 10 US companies. In world terms in 1991 the largest European contractor, BAe, ranked 7th. There have been a number of changes since 1991, through merger and takeover in the USA and Europe, but the general picture remains the same. The general expectation is that the market will soon comprise a few large players and a periphery of specialist niche producers. As Table 4 shows, the US and Western European companies dominate the top 100 list, with the US companies producing 61% of the total arms sales and the European ones 33%. After the USA, the UK has the largest number of contractors in the top 100, but France has the highest arms sales; together they make up 61% of non-US OECD sales. The increasing involvement of Japanese companies in arms production is evident, with them accounting for over 3% of OECD arms sales, the same as Italy. The DIB contractors tend to be concentrated in the aerospace, electronics, ordnance, and shipbuilding industries. Table 5 shows that in the European countries with the largest defense sectors, the largest shares in military procurement expenditure tend to

416

J.P Dunne Table 3 100 Largest arms producers in 1991: by country

Rankb

USA 1 2 4 5 5s 6 7 8 9 11 12 13 18 20 21 23 24 25 25s 25s 26 28 29 31 33 37 40 41 43 46 48 49 49s 59 62 66 67

Name

McDonnell Douglas General Dynamics General Motors Lockheed Hughes Elec (GM) General Electric Northrop Raytheon Boeing Martin Marietta Rockwell Intemat United Technol Litton Industries Grumman TRW Loral Westinghouse E Tenneco Newport News (Tenneco) Pratt & Whitney (UT) Texas Instruments LTV Textron Unysis E-Systems IBM GTE ITT FMC Gencorp Alliant Tech Systems Allied Signal Aerojet (Gencorp) Harris Bath Iron Works Thiokol Science Applicat Intl

Arms sales (million US$)

10200 7620 7500 6900 6600 6120 5100 5100 5100 4560 4000 4000 3150 2900 2800 2600 2300 2220 2220 2100 1950 1800 1800 1750 1550 1300 1200 1200 1170 1110 1100 1100 1090 760 720 690 680

Arms sales as percentage of total

55 80 6 70 57 10 90 55 17 75 34 19 60 72 35 90 18 16 100 29 29 30 23 20 78 2 6 6 30 56 93 9 95 25 96 54 100

Employees

109 123 80 600 756000 71300 93 000 284 000 36200 71 600 159100 60 500 87 000 185100 52 300 23 600 71300 22 000 113664 89 000 28 100 41 000 62 939 34 600 52 000 60 300 18622 373 815 175 000 110000 23 150 14500 6700 98 300 n.a. 30 700 10 000 11500 13 100 continued on next page

Ch. 14:

417

The Defense Industrial Base

Table 3, continued Arms sales as percentage of total

Employees

650 620 610 600 600 530 500 450 440 400 400 390 390 380

1 29 32 20 5 68 16 n.a. 91 6 24 n.a. n.a. 17

317 100 26500 15700 17324 102 000 8200 29400 n.a. n.a. 58 182 12100 n.a. n.a. 14400

British Aerospace GEC Rolls Royce VSEL Consortium Smiths Industries Lucas Industries Westland group Thorn EMI Racal Electronics Devonport Manage ment Hawker Siddley Hunting Dowty Group

7550 3960 1680 920 600 570 530 470 440 430 420 400 390

40 24 27 100 52 14 64 7 15 100 11 30 32

123 200 104 995 57100 13 028 12100 54 900 9060 53 757 35 384 11 460 40500 7302 13 000

Thomson SA Thomson CSF DCN Aerospatiale Dassault Aviation CEA Industrie SNECMA Groupe GIAT Industries

4800 4800 3710 3450 1870 1750 1320 1220

38 77 100 40 74 25 31 61

105000 44500 30 000 43 287 11914 37300 27236 17000

Rankb

Name

68 69 70 72 73 77 78 83s 86 92 93 96 97 98

AT&T Computer Sciences Sequa Hercules Motorola Avondale Industrie s Teledale AVCO (Textron) Esco Electronics Honeywell Penn Central Dyncorp Mitre Olin

UK 3 14 32 52 71 75 76 82 85 87 89 91 95 France 10 l0s 16 17 27 30 36 39

Arms sales (million US$)

continued on next page

418

J.P Dunne

Table 3, continued Rank

b

Name

Arms sales (million US$)

Arms sales as percentage of total

Employees

47 50 53s 54s 61s 74 90s

Alcatel-Alsthom Matra Groupe Matra Defense SNECMA Eurocopter(Aerospatiale) SAGEM Groupe SAGEM

1100 1050 890 850 720 590 400

4 26 100 33 42 28 43

213 100 21 334 2500 13 816 7525 15076 6006

Germany 15 16s 32s 53 54s 55 56 57 58 64s 86s 89s 90 90s 94

Daimler Benz DASA (Daimler Benz) MBB (DASA) Siemens Telefunken S (DASA) Diehl Bremer Vulkan Rheinmetall Thyssen MTU (DASA) Systemtechnik N (BV) Blohm & Voss (Thyssen) Mannesmann Krauss-Maffei (Mann ... ) Lurssen

3920 3620 1540 900 810 800 780 770 770 690 430 400 400 400 390

7 49 51 2 82 44 39 37 3 32 75 49 3 47 n.a.

379 252 56465 20 730 402 000 8846 15 529 15021 13 661 148 557 17052 2441 5758 125 188 5004 1000

Italy 19 20s 38 44 81s 84s 89s

IRI Alenia (IRI) EFIM FIAT Oto Melara (EFIM) Fincantieri (IRI) Agusta (EFIM)

3270 2140 1270 1140 480 440 410

5 55 29 3 99 20 71

368267 30099 35 489 287 957 2149 19750 6998

Japan 22 51 60 63 88

Mitsubishi HI Kawasaki HI Ishikawajima-Harima Mitsubishi Electric Toshiba

2630 1050 740 710 420

14 15 9 4 1

n.a. n.a. n.a. n.a. 168 000 continued on next page

Ch. 14:

419

The Defense Industrial Base

Table 3, continued Arms sales as percentage of total

Employees

1410

88

17 100

490 450

94 98

8500 5100

INI (Spain) CASA (INI)

1330 670

7 75

142 295 9338

Ordnance Factories Hindustan Aeronautics

1120 370

96 98

173 000 35 000

870 740

47 99

14 508 6274

520

11

29 329

Canada 61 CAE Industries 84 Bombardier

730 440

80 16

10000 26 692

South Africa 64 Armscor

710

70

20 000

1170 690

46 95

19 138 4495

450

94

4265

Rankb

Name

Israel 34

Israel Aircraft Ind.

81 83

Israel Military Ind. Rafael

Spain 35 67s India 45 99 Sweden 54 59s 78

Celsius Swedish Ordnance-Bofors (Celsius/Nobel) Saab-Scania

Switzerland 42 Oerlikon-Buhrle 65 Eidgen. Riistung. Switz Netherlands 83s Hollandse Signaal (Thomson-CSF) (Netherlands) a

Arms sales (million US$)

Source: SIPRI Yearbook (1993).

b s means fully owned subsidiaries. These are included in the figure for the parent company but the rank

indicates where the company would have been if it was independent.

go to the aerospace and electronics sectors. As the second part of the table shows the aerospace industry is, apart from the small and highly specialized munitions sector, the most dependent on arms production, while the electronics sector has a much larger customer base. A similar pattern is observed in the USA. There have been numerous case studies of defense intensive sectors, which have provided a range of information

420

JP Dunne Table 4 Regional/national shares for 100 largest arms producers 1991a

#

Region/Country (billion US$)

47 40 7 6 100

USA Western European OECD Other OECD Developing countries Total

108.9 58.6 6.7 4.6 178.8

Arms sales Percentage of total 60.9 32.8 3.8 2.5 100.0

Breakdown of OECD #

Country (billion US$)

13 11 8 5 3 2 2 2 1 aSource:

UK France FRG Japan Italy Switzerland Sweden Canada Spain

18.4 21.4 8.7 5.5 5.6 1.9 1.4 1.2 1.3

Arms sales Percentage of total 10.3 11.9 4.9 3.1 3.1 1.0 0.8 0.7 0.8

SIPRI Yearbook (1993).

on the character and development of the DIB. The studies by Wulf (1993b) and Brzoska and Lock (1992) are recent examples. There are still problems in interpreting the figures in Table 5, however, as they tell us little about the importance of defense production to the individual firms. In addition,

it is extremely difficult to get information on the sub-contractors who supply the main contractors even in the USA. As a result, most studies simply state how important they are and then move on. One exception is Hartley and Hooper (1990) which, in a study of Vickers, provides evidence on the supply chain for the UK's Challenger 2 tank. Countries with large defense sectors differ in a number of ways, both in the structure of the DIB and its links with the state and the military. A major factor is the ownership of the contractors. In the USA all of the companies have remained in private hands; in the UK they have recently (early 1980s) been privatized and in France they are mainly state-owned. However, what this implies is not clear. Ownership does matter, in that it confers specific property rights, but the mere fact that one of the companies was nationalized and one privatized, does not distinguish these two companies. It is not state or private ownership itself that is important, but the character of the ownership, the constraints and the policies followed. The monopsonistic character of the defense market and the inherent strategic and technological uncertainty about needs, costs and

421

Ch. 14: The Defense IndustrialBase Table 5 Percentage shares of military production and procurement Country

Aerospace

Shipbuilding

Machinery Electronics Munitions, Transportation weapons

A. Share of industrial sectors in military procurement expenditure 25.0 6.9 17.5 Germany (1987-1990) n.a. 7.0 31.0 UK (1987-1988) 8.0 10.0 34.0 France (1986-1988) n.a. 4.0 10.0 Belgium (1986-1987)

25.2 23.0 27.3 16.0

B. Share of military production in total industrialproduction 1986-1990 n.a. n.a. 20 45 Germany 39 n.a. 50 50 UK 57 n.a. 50 69 France 10 n.a. 25 50 Italy 4 3 10 35 Belgium 5 n.a. 30 10 Netherlands 8 n.a. 70 50 Spain a

Source: Wulf (1993a).

b

n.a.b 8.0 n.a. 50.0

7.5 n.a. n.a. 20.0

n.a. 100 60

n.a. 3 n.a.

n.a. 80 100

2 2 3

20

6

n.a.: not available.

performance raise moral hazard problems, as we saw in the discussion of the MIC. The actors in the DIB can use this fundamental uncertainty to capture and exploit government purchasing in their own interest. The question is whether such exploitation is minimized by having the companies private at arms length, where the government has no responsibility for their profitability, production and employment, and thus is less open to lobbying and leverage; or whether such exploitation is minimized by having the companies owned by the state where the Government can directly control them to stop such exploitation. If one believes that the exploitation is a direct result of a capitalist firm's drive for profits, then public ownership appears more attractive; however, if one allows for a more general concept of surplus or rents which can be extracted by state employees, this is not the case. Nationalized arms firms tend to be as expansionist and acquisitive as private ones. Within a state the transfers of surplus tend to be less obvious: it can be difficult to identify the true cost of military activity. Within a market the transfers are somewhat more transparent [see Dunne and Smith (1992)]. Irrespective of the form of ownership the government can determine the forms of contractual arrangements with the DIB. During the Cold War the structure of these relations across the major producers were remarkably similar, but recent developments have seen changes taking place, particularly in the USA and UK, with the introduction of competition and new forms of contracts [see Chapters 12 and 16 of this Handbook]. While the role of the government will clearly structure the form of the DIB in any particular country, the final outcome will depend upon other social, economic and political factors. These will include the nature of the firms, the nature of management

422

JP Dunne

and workers, the nature and strength of government, the role of the industry in government policy, the relative success of the DIB, attitudes to foreign competition, the existence or not of nuclear capability, and the nature and extent of R&D and its diffusion. The studies by Wulf (1993a) and Brzoska and Lock (1992) provide some country comparisons. In our discussion of the DIB we have focused on the NATO economies and have ignored what was the largest DIB in the world, that of the former Soviet Union. This is partly because of the very different nature of the Soviet DIB, the paucity of high quality and believable studies, the problems of interpreting the information available, and the profound restructuring that has taken place since the end of the Cold War. Cooper (1991, 1993) provides a comprehensive analysis. In addition, we have not really dealt with the fact that many developing countries have developed the capability of producing weapons. These 'third tier' producers [Krause (1992)] show some similarities to the DIBs in the advanced economies but also differences [see Brauer (1991) and the collection in the volume edited by Katz (1986)]. Having provided an overview of the evolution and structure of the international DIB, focusing on the NATO and OECD economies, we can see that, while retaining some common features, DIBs can differ in form, content and behavior across countries. We now move on to consider the debate over the efficiency of DIBs and the effects of maintaining a DIB on the rest of the economy.

6. Efficiency and economic effects There has been considerable debate on the effect of the DIB on economic performance, usually as part of the debate on the economic effects of the MIC. As we have seen, most of the theoretical approaches tend to see the MIC as generating pressures for an increased military burden independent of any threat. The characteristics of the DIB that have developed over the post-1945 period have provided a client industry for the procurement executive but the DIB is also an active participant in this process. The economic effects of this will depend upon whether or not the MIC and the DIB are a force for economic growth or economic decline. It is worth pointing out that there are different issues involved in evaluating the DIB. The aims of the government are likely to be to provide weapons and other material to deter aggression and provide security and to maintain the capability. Any attempt to achieve more value for money may be at the expense of this capability. In addition, there may be a trade-off between the benefits of security and any negative economic effects on the rest of the economy. Military spending may be necessary but economically costly. The contradictions are clear: if the military causes economic decline it is also likely to undermine the ability to recreate the security. The issue of hegemony and the potentially contradictory role of the military has been the subject of considerable debate [Dunne (1990), Kennedy (1987)].

Ch. 14:

The Defense IndustrialBase

423

As we saw in Section 2 there is also considerable debate over the benefits of a DIB in terms of self sufficiency and foreign sourcing. There is some evidence from EU countries that opening up defense markets to foreign competition could lead to cost savings of about 25% on defense equipment purchases [Hartley and Cox (1995)]. The arguments that the DIB has a positive effect on economic growth point to the spin-off from technology, the provision of jobs, in particular highly skilled ones, and the externality effects through linkages with the rest of the economy. The DIB is also seen to have provided a form of industrial planning in some countries, particularly the USA, which has benefited the whole of industry [Cypher (1987)]. In addition, the underconsumptionist and permanent arms economy approaches, discussed above, see military production as allowing underconsumption or overaccumulation crises to be averted. In contrast, the arguments that the DIB has a negative effect point to the increasing paucity of spin-off and the increase of spin-in. The DIB is seen as an industrial structure protected from the ravages of competition that has developed to be expert at getting money and making profits from government and not at producing low cost products for a competitive market. As a result it has developed into an inefficient producer and has had a damaging impact on the rest of industry. It has crowdedout resources, both investment and human capital, has reduced civil technological development, and has had externality effects on other companies. It has reduced industrial efficiency and international competitiveness. Melman (1985) argues that taking up lucrative military contracts weakens a firm's ability to behave as a proper capitalist firm. They do not have to reduce costs and strive instead to increase subsidies and maximise costs to push up the money they get from the government. Inefficiency thus becomes profitable and endemic, and bad habits spread from the defense divisions to the rest of companies and to the rest of the economy. The development of an MIC reinforces the creation of state capitalist system which leads to reduced international competitiveness. Much of the empirical support for spin-off relies on anecdotal evidence, and this has generally been refuted by recent studies which have shown more concern for the problems of civil R&D in the UK [ACOST (1989)]. Attempts at more formal empirical analysis have failed to find supporting evidence [e.g., see Buck et al. (1993) for the UK and Chakrabarti et al. (1992) for the USA]. Another possible source of spin-off is the mobility of scientists and engineers from military to civilian industry [Lerner (1992) finds evidence of spillover for the USA]. It is most likely that the role of R&D is historically specific. During wars it is likely to provide innovations, and certainly post-World War II there were a number of important civilian innovations that came from defense, such as transistors, semiconductors and computing (though one can never say what the opportunity cost was). In general, however, it has been the countries with limited military R&D that have advanced faster, economically and technologically. Indeed, it would appear that civilian technology is increasingly leading military technology, with "spin-in" becoming increasingly important. Many advanced weapons systems are relying on

424

JP Dunne

civil production for advanced components. There are some exceptions, such as stealth technology, but the dominant trend is towards a dependence on civil innovation, or at least a limiting of freedom to pursue technologies where there is no civil connection [see Markusen and Yudken (1992)]. In addition to the descriptive, historical and qualitative analyses of the DIB there have been some more theoretical and quantitative economic studies in the industrial economics literature mainly based within the neo-classical paradigm. The structureconduct-performance paradigm provided an understanding of industrial performance depending on structural factors reflected in the size distribution of firms and entry conditions. Industries could be evaluated on the basis of their distance from the competitive ideal through the analysis of these factors. A classic study is by Peck and Scherer (1962). More recent developments moved away from the competitionmonopoly dichotomy and focused on the oligopolistic nature of most markets, using game theoretic and principal agent models to analyse industries and introducing concepts of contestable markets, with the focus on threat rather than actual competition determining market structure [Baumol et al. (1982)] and transaction cost models of firms [Laffont and Tirole (1993)]. Empirical studies of industrial structure and performance have focused on areas where adequate data are available. Employment equations have been estimated for defense firms, normally derived from neo-classical production functions and compared with results from civil firms. These studies provide some evidence of labor hoarding by defense companies and inelastic responses to factor prices and cuts in demand [see Hartley and Lynk (1983) and Lynk and Hartley (1985) for the UK]. In a similar manner the relative profitability of defense firms has been analysed, either by market based analysis or the estimation of profit functions (with all the attendant measurement problems). Studies have suggested that competition reduces profitability and that wars can lead to higher profits [Dunne (1993), Hartley and Watt (1981)]. Evidence also suggests that the financial performance of defense companies has been better than that of comparable large corporations. According to Trevino and Higgs (1992) the top US defense firms outperformed the market by a huge margin 1970-1989, though the post-Cold War period has seen much harder times [see also Dunne (1993), Stigler and Friedland (1971), Bohi (1973), and Kaun (1990)]. Overall, the evidence is mixed but it certainly does not support the view that an emphasis on developing and maintaining a DIB provides economic benefits. Instead, it suggests that there are likely to be clear economic costs, as well as the direct resource costs. With the end of the Cold War even the security benefits of maintaining a comprehensive DIB are being brought into question and the resulting cuts in military expenditure have already precipitated marked changes in the DIB. 7. Developments in the DIB With the end of the Cold War the international DIB is undergoing a profound restructuring. In addition to the cuts in procurement expenditure, the drive of

Ch. 14:

The Defense Industrial Base

425

technology towards more complex and expensive systems has challenged the ability of national economies to support major players. While the changing relation between military and civilian technology has led to new competition from civil producers, there is limited regulation of the arms producers internationally and the trend is towards a reduced role for the state in many countries. Although arms production is not covered by the Treaty of Rome, these developments are underpinned in the EC by the Independent European Policy Group (IEPG) Action Plan announced in 1986, which is intended to remove the protectionist policies in national arms markets. It is not intended to create a free market, but to allow "balanced" access to markets. This has spurred the restructuring of the industry in Europe [see Brzoska and Lock (1992)]. More recent developments have seen the IEPG being absorbed into the Western European Union (WEU) as the institutional architecture of Europe continues to evolve and it attempts to develop an alternative security agenda. The developments in the international arms market have led to increasing concentration and a marked restructuring. In Europe there has recently been a growth of crossborder mergers and international takeovers in response to the increasing uncertainties of the market and partly in response to the Single European Market. The international arms market has become increasingly competitive with an increase in intermediate producers and some switch in patterns of demand away from high technological equipment. In addition, as former Warsaw Pact countries have tried to sell weapons for hard currency, often at bargain prices, they have squeezed out less sophisticated weapons. This excess supply and 'cascadence' of CFE Treaty limited items has increased pressures in the arms market in particular for marginal and intermediate producers [Dunne (1994), Krause (1992)]. A further development has been the growth of cross share ownership and the creation of new companies by the defense giants to integrate production and research and development. There are clear trends towards the formation of transnational defense companies, as Far East Asian companies become involved [see Walker and Willett (1993)]. Internationalization in these forms and in terms of licensed production, joint ventures, collaboration and mergers, is increasing in all military industrial sectors. It is most advanced in the electronics and aerospace sectors, for both military and civil production, with other sectors, military vehicle and warship production, being more mixed but nevertheless increasingly moving towards internationalization in the face of overcapacity [see Chapter 16 of this Handbook, and Sk6ns (1993)]. This globalization of the arms producers could mark a significant change from the Cold War DIB, based mainly upon large national defense producers owned or controlled by the nation states and dependent upon them for much of their sales. The changes taking place seem to be a move back to the pre-Second World War industrial structure, where the large private arms producers had considerable autonomy and little national allegiance [Lovering (1993)]. The changes should not be overstated: there still remains considerable national regulation of defense firms, indeed there has even been the prevention of takeovers.

426

J.P Dunne

But there are changes taking place and they clearly have important implications for the regulation and control of the arms market. As regards future prospects, the differing economic situations and the differing forms of defense industrial base within the main producers mean that the effects of the cuts and restructuring are likely to differ across countries. There is also considerable uncertainty about the future. The process of continued cuts in military spending by the major powers is looking to have ended, with pressures in the USA from the Republican Congress to maintain or increase it. Nevertheless, the changed international environment makes a return to Cold War expenditure unlikely to say the least, and the economic situation of potential arms importers suggests a relatively stagnant market for exports. The World Bank and IMF are pushing for reduced military burdens in debtor countries, and many countries have reduced their orders of equipment from the industrialized countries. It is likely that these trends will continue. Indeed, it is also likely that the market will become increasingly competitive, as former WTO countries fall back on arms exports in desperation, and intermediate producers develop their facilities. This means that there is probably still some way to go in the restructuring of the DIB. The size of the US market and the resulting cost advantages may cushion many of the blows to the defense contractors, but there is still likely to be further concentration. Within Europe, Germany is likely to suffer least, given its strong industrial and technological capacity, while France is likely to retain a high level of government involvement to maintain its industry [see Walker and Willett (1993)]. The UK is the most vulnerable DIB, with a weak technological and industrial base and government policies biased towards introducing competition [see Dunne and Smith (1992)]. Also, British companies tend to be poorly placed for change relative to their main competitors. For example, DASA is very strong in civil markets compared to BAe, and Thomson CSF is stronger than GEC in civil electronics. [Renner (1992), Wulf (1993a) and Brzoska and Lock (1992) provide detailed country studies].

8. Conclusions This chapter has endeavored to provide an overview of the issues and research on the defense industrial base. It has shown the real problems of actually defining the defense industry and the DIB. Problems that should be taken seriously when undertaking applied work. The definition of the DIB needs to be chosen to suit the empirical analysis being undertaken and the practical problems of data availability. The defense industry is seen to be very different from the civil sector, mainly because of the monopsonistic role of the government, the emphasis on developing a national DIB, and the Cold War. While there are clear policy issue to deal with, such as determining the costs and benefits of the DIB and how to get best value for money, it is important to see the DIB within the context of a wider and more pervasive MIC. The scale and scope of the MIC and the influence the social, economic and political structures borne

Ch. 14:

The Defense Industrial Base

427

of the Cold War arms race have to be recognized, as does the complex interaction of the demand and supply sides of the defense sector over time and their feedback on one another. The only way to understand this process is through careful empirical work on the development of the MIC, a task that is clearly of importance given the restructuring of the DIB taking place in the mid-1990s. While there are signs that the path of military spending reductions is no longer declining as rapidly it is still likely that there will be further cuts and most certainly the defense industry will become increasingly competitive. There is bound to be further restructuring with budget pressures to collaborate and to buy off-the-shelf weapons. It seems sensible for countries to consider what the nature and structure of the international DIB should be and not to leave it to the firms to restructure away from national and international control. What is needed is more detailed empirical research on the impact of disarmament on the DIB, taking advantage of the considerable cross country experience since the end of the Cold War. This will allow governments to plan further cuts to overcome economic costs and to gain from them. The evidence certainly suggests that military spending is an economic burden and that in the long run there can be benefits, but there are likely to be problems of structural adjustment. Policies are needed to aid structural adjustment, reallocate resources, retrain skilled workers and reorient science and technology away from the military. Disarmament needs to be treated as an investment process and policies for conversion developed to achieve the illusory 'peace dividend' [Chapter 19 of this Handbook, Hartley (1993), Dunne and Willett (1992)]. Central to this process will be a thorough understanding of the DIB and the changes it is undergoing.

References ACOST, 1989, Defence research and development: A national resource (Advisory Council on Science and Technology, HMSO, London). Baack, B., and E. Ray, 1985, The political economy of the origins of the military industrial economy in the United States, Journal of Economic History 45, 369-75. Ball, N., and M. Leitenberg, 1983, The structure of the defense industry: An international survey (Croom Helm, London). Baran, P., and P.M. Sweezy, 1966, Monopoly Capital (Monthly Review Press, London). Baumol, W., J. Panzar and R. Willig, 1982, Contestable markets and the theory of industry structure (Harcourt Brace Jovanovich, San Diego, CA). Bohi, D.R., 1973, Profit performance in the defense industry, Journal of Political Economy 81, 721-728. Brauer, J., 1991, Arms production in developing nations: The relation to industrial structure, industrial diversification and human capital formation, Defence Economics 2, 165-175. Brunton, B.G., 1988, Institutional origins of the military industrial complex, Journal of Economic Issues 22, 599-606. Brzoska, M., and P. Lock, eds., 1992, Restructuring of arms production in Western Europe (Oxford University Press, Oxford). Buck, D., K. Hartley and N. Hooper, 1993, Defence research and development, crowding-out and the peace dividend. Defence Economics 4, 183-212.

428

JP Dunne

Chakrabarti, A.K., H.M. Glismann and E.-J. Horn, 1992, Defence and space expenditures in the US: An interfirm analysis, Defence Economics 3, 169-90. Cooper, J., 1991, The Soviet defence industry: Conversion and reform (Pinter/RIIA, London). Cooper, J., 1993, The Soviet Union and the successor republics: Defence industries coming to terms with disunion, in: H. Wulf, ed., Arms industry limited (SIPRI, Oxford University Press, Oxford). Coopey, R., M. Uttley and G. Spinardi, 1994, Defence science and technology: Adjusting to Change (Harwood, Reading). Cypher, J., 1987, Military spending, technical change, and economic growth: A disguised form of industrial policy, Journal of Economic Issues 21, 33-50. Dumas, L.J., 1986, The overburdened economy (University of California Press, California). Dunne, J.P., 1986, The employment consequences of military expenditure: A comparative assessment (ILO Disarmament and Employment Programme, working paper no. 6, International Labour Office, Geneva). Dunne, J.P., 1990, The political economy of military expenditure: An introduction, Cambridge Journal of Economics 14, 395-404. Dunne, J.P., 1993, The changing military industrial complex in the UK, Defence Economics 4, 91-112. Dunne, J.P., 1994, Conversion in Europe: Challenges and experiences, Discussion Paper (School of Business and Economic Studies, University of Leeds). Dunne, J.P., and R. Smith, 1992, Thatcherism and the UK defence industry, in: J. Michie, ed., 1979-92, The economic legacy (Academic Press, London) 91-111. Dunne, J.P., and S. Willett, 1992, Disarming the UK: The economics of conversion, Discussion Paper No. 2 (Centre for Industrial Policy and Performance, University of Leeds). Edgerton, D., 1991, England and the Aeroplane: An essay on a militant and technological nation (Macmillan, London). Fine, B., 1993, The military industrial complex: An analytical assessment, Cyprus Journal of Economics 6, 26-51. Gansler, J.S., 1980, The defence industry (MIT Press, Cambridge, MA). Gansler, J.S., 1989, Affording defence (MIT Press, Cambridge, MA). Grossman, G.M., and E. Helpman, 1994, Protection for sale, American Economic Review 84, 833 850. Hartley, K., 1991, The economics of defence policy (Brasseys, London). Hartley, K., 1993, Economic aspects of disarmament: Disarmament as an investment process (UNIDIR, Geneva). Hartley, K., and A. Cox, 1995, The costs of non-Europe in defence procurement (European Commission, Brussels). Hartley, K., and N. Hooper, 1990, The economic consequences of the UK government decision on a Chieftain tank replacement, Research Monograph Series, Vol. 1 (Centre for Defence Economics, University of York). Hartley, K., and E. Lynk, 1983, Labour demand and allocation in the UK engineering industry: Disaggregation, structural change, and defence reviews, Scottish Journal of Political Economy 30, 42-53. Hartley, K., and P. Watt, 1981, Profits, regulation and the UK aerospace industry, Journal of Industrial Economics 26, 413-428. Hartley, K., F. Hussain and R. Smith, 1987, The UK defence industrial base, Political Quarterly 58, 62-72. Hooper, N., and D. Buck, 1991, Defence industries and equipment procurement options, in: N. Hooper and S. Kirby, The cost of peace: Assessing Europe's security options (Harwood, Reading). Howard, M.C., and J. King, 1992, A history of marxian economics: Volume 2, 1929-90 (Macmillan, London). Kaldor, M., 1991, The baroque arsenal (Abacus, London). Katz, J.E., ed., 1986, The implications of Third World military industrialization (Lexington Books, Lexington, MA).

Ch. 14:

The Defense Industrial Base

429

Kaun, D.E., 1990, War and Wall Street: The impact of military conflict on investor attitudes, Cambridge Journal of Economics 14, 439-452. Kennedy, P., 1987, The rise and fall of the great powers (Unwin, London). Kidron, M., 1968, Western Capitalism since the war (Weidenfeld and Nicolson, London). Krause, K., 1992, Arms and the state: Patterns of military production and trade (Cambridge University Press, Cambridge). Laffont, J.-J., and J. Tirole, 1993, A theory of incentives in procurement and regulation (MIT Press, Cambridge, MA). Lerner, J., 1992, The mobility of corporate scientists and engineers between civil and defence activities: Implications for economic competitiveness in the post-Cold War era, Defence Economics 3, 229-42. Lovering, J., 1990, Military expenditure and the restructuring of capitalism: The military industry in Britain, Cambridge Journal of Economics 14, 453-468. Lovering, J., 1993, After the Cold War: The defence industry and the new Europe, in: P. Brown and R. Crompton, 1993, The new Europe? Economic integration and social exclusion (University of London Press, London). Lynk, E., and K. Hartley, 1985, Input demands and elasticities in UK defence industries, International Journal of Industrial Organization 3, 71 83. Maddock, I., 1983, Civil exploitation of defence technology (National Economic Development Organisation, London). Markusen, A., and J. Yudken, 1992, Dismantling the cold war economy (Basic Books, New York). Melman, S., 1970, Pentagon capitalism (McGraw-Hill, London). Melman, S., ed., 1971, The war economy of the United States (St. Martins Press, New York). Melman, S., 1985, The permanent war economy (Simon and Schuster, New York). Mills, C.W., 1956, The power elite (Oxford University Press, Oxford). Moskos Jr, C.C., 1972, The military industrial complex: Theoretical antecedents and conceptual contradictions, in: S.C. Sarkesian, ed., 1972, The military industrial complex: A reassessment (Sage, London) 3-24. Paukert, L., and P.J. Richards, eds., 1991, Defence expenditure, industrial conversion, and local employment (International Labour Office, Geneva). Peck, M.J., and F.M. Scherer, 1962, The weapons acquisitions process: An economic analysis (Harvard University Press, Cambridge, MA). Pivetti, M., 1992, Military spending as a burden on growth: An underconsumptionist critique, Cambridge Journal of Economics 16, 373-384. PRODEM, 1993, Militarism versus disarmament - Challenging the new arms race, The new militarism briefings, No. 2 (Project on Demilitarisation). Ratner, J., and C. Thomas, 1990, The defence industrial base and foreign supply of defence goods, Defence Economics 2, 57-68. Renner, M., 1992, Economic adjustment after the Cold War: Strategies for conversion (UNIDIR, Dartmouth Publishing Co., Dartmouth). Reppy, J., 1993, The United States: Unmanaged change in the defence industry, in: H. Wulf, ed., 1993, Arms industry limited (SIPRI, Oxford University Press, Oxford). Rogerson, W.P., 1994, Economic incentives and the defence procurement process, Journal of Economic Perspectives 8, 65-90. Sandler, T., and K. Hartley, 1995, The economics of defense (Cambridge University Press, Cambridge). Schofield, S., 1993, Defence technology, industrial structure and arms conversion, in: R. Coopey, M. Uttley and G. Sporardi, eds., Defence science and technology: Adjusting to change (Harwood, Reading). Schwarz, J.A., 1990, Baruch, the New Deal and the origins of the military industrial complex, in: R. Higgs, ed., Arms, politics and the economy: Historical and contemporary perspectives (Holmes and Meier, New York). SIPRI Yearbook, annual, SIPRI Yearbook (Blackwell, Oxford, for Stockholm International Peace Research Institute).

430

JP Dunne

Skons, E., 1993, Western Europe: internationalisation of the arms industry, in: H. Wulf, ed., Arms industry limited (SIPRI/Oxford University Press, Oxford). Slater, J., and T. Nardin, 1973, The concept of the military industrial complex, in: S. Rosen, ed., 1973, Testing the theory of the military industrial complex (Lexington Books, Lexington, MA) 27-60. Smith, D., and R. Smith, 1983, The economics of militarism (Pluto Press, London). Smith, R., 1977, Military expenditure and capitalism, Cambridge Journal of Economics 1, 61-76. Smith, R., 1990, Defence procurement and industrial structure in the UK, International Journal of Industrial Organisation 8, 185-205. Smith, R., and J.P. Dunne, 1994, Is military spending a burden? A marxo-marginalist response to Pivetti, Cambridge Journal of Economics 18, 515-521. Southwood, P., 1991, Disarming military industries (Macmillan, London). Stigler, G.J., and C. Friedland, 1971, Profits of defense contractors, American Economic Review 61, 692-94. Thorsson, I., 1984, In pursuit of disarmament: Conversion from military to civilian production in Sweden (Liber, Stockholm). Trevino, R., and R. Higgs, 1992, Profits of US defense contractors, Defence Economics 3, 211-218. Walker, W., 1988, UK defence electronics: A review of Government statistics (PICT Policy Research Paper, ESRC, London). Walker, W., and S. Willett, 1993, Restructuring the European defence industrial base, Defence Economics 4, 141-160. Walker, W., M. Graham and B. Harbor, 1988, From components to integrated systems: Technological diversity and interactions between military and civilian sectors, in:P. Gummett and J. Reppy, eds., The relation between military and civilian technologies (Kluwer Academic Publishers, Dordrecht, Netherlands) 17-37. Wulf, H., ed., 1993a, Arms industry limited (SIPRI/Oxford University Press, Oxford). Wulf, H., 1993b, Arms industry limited: The turning point, in: H. Wulf, ed., Arms industry limited (SIPRI/Oxford University Press, Oxford). Wulf, H., 1993c, Western Europe: Facing over-capacities, in: H. Wulf, ed., Arms industry limited (SIPRI/ Oxford University Press, Oxford).

Chapter 15

ECONOMICS OF DEFENSE R&D FRANK R. LICHTENBERG Columbia University Graduate School of Business and National Bureau of Economic Research

Contents Abstract Keywords 1. Introduction 2. Design competitions 2.1. How much private investment in defense R&D?

3. 4. 5. 6. 7.

432 432 433 434 436 437 441 444 448

Independent R&D subsidies The profitability of defense (R&D and other) contracts The impact of defense and other R&D on productivity growth The effect of defense R&D on nondefense R&D investment How do Government decision-makers respond to cost information yielded by defense R&D? 450 8. Are defense R&D programs dynamically optimal? 453 9. Conclusions 455 References 456

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

432

FR. Lichtenberg

Abstract This chapter examines a number of aspects of defense R&D, including: mechanisms design competitions, and independent R&D subsidies - used by the US government to encourage firms to invest their own funds in defense R&D; theory and evidence concerning both the private and social benefits of, or returns to, R&D conducted by defense contractors; the effect of defense R&D on nondefense R&D investment; the response of government decision-makers to cost information yielded by defense R&D; and the dynamic optimality of R&D projects.

Keywords research and development, procurement, design competitions, innovation, productivity, profitability, government expenditure, R&D subsidies, weapons acquisition, defense policy

Ch. 15:

433

Economics of Defense R&D Table 1 b Defense R&D expenditures in 1991 of G-7 countries

Country

Total R&D

33

123

26.8

1 2 4 3 0 0

61 30 21 16 11 6

1.6 6.7 19.0 18.8 0.0 0.0

US Japan Germany France U.K. Italy Canada a

Defense R&D as % of total R&D

Defense R&D

Defense R&D as % of GNP 0.7 0.0 0.1 0.5 0.4 0.1 0.0

Billions of 1987 dollars.

b Source: Science and Engineering Indicators [National Science Board (1993)], pp. 375-376.

1. Introduction It is the government's role to promote investment in research and development (R&D) that yields innovations in the production of public goods, such as armaments for national defense and equipment for space exploration. As Table I indicates, it is estimated that in 1991, the G-7 nations collectively spent $43 billion (in 1987 dollars) on defense R&D. (This figure may significantly understate the true amount because, as we argue below, only R&D that is directly funded by the government is officially classified as defense R&D, whereas a substantial amount of "privately-funded" R&D may also be defense-related.) About three-fourths of total estimated defense R&D was conducted by the United States; for this reason, and also because more is known about defense R&D conducted in the US, we will devote most of our discussion to US military R&D. This chapter will examine a number of aspects of defense R&D. Sections 2 and 3 describe two mechanisms - design competitions, and independent R&D subsidies used by the US government to encourage firms to invest their own funds in defense R&D. The next two sections discuss theory and evidence concerning both the private and social benefits of, or returns to, R&D conducted by defense contractors. In Section 4 we argue that the private returns, or profitability, of defense R&D may depend on contractors' ability to shift costs from commercial to government R&D. A production function/productivity growth framework for assessing the social returns to defense R&D is presented in Section 5. The effect of defense R&D on nondefense R&D investment is examined in Section 6. The question, how do government decision-makers respond to cost information yielded by defense R&D?, is pursued in Section 7. Section 8 investigates whether independent R&D projects conducted by defense contractors are dynamically optimal, in the sense defined by Grossman and Shapiro (1986). The chapter's conclusions are summarized in Section 9.

434

ER. Lichtenberg

2. Design competitions In addition to the two obvious institutional arrangements the US government 1 uses to call forth defense R&D investment - performing R&D in government laboratories, and contracting with private firms and nonprofit organizations (such as universities) to perform R&D - there is a third and quantitatively important method by which the government promotes R&D investment relevant to the provision of social goods: awarding major contracts by a method of acquisition known as "procurement by design and technical competition". In essence, this method consists in the government's simply revealing its demand for certain types of technological innovations, and encouraging private firms to sponsor the necessary R&D, the costs of which the sponsor will recover from profits on the sale of the product. Edwin Mansfield (1971) has observed that before World War II, the government tended to issue relatively few R&D contracts to industry or universities. If the government wanted private firms to perform defenserelated R&D, it would encourage them to finance it on their own. Since the beginning of World War II, the real value of government R&D contracts has increased very rapidly, and contracting has apparently become the most important of the three methods used by the government to promote R&D investment related to social goods. But the government continues to induce a considerable amount of privately financed, federal mission-oriented R&D expenditure by sponsoring design competitions. A design and technical competition (henceforth abbreviated "design competition") begins "officially" when a federal agency (such as the Department of Defense) issues a formal Request for Proposals, which is often 1100 to 2500 pages long. Three or four firms typically submit proposals (which may range from 23 000 to 38 000 pages in length) in response to requests issued by the Defense Department, which then begins an elaborate review process. The firm that submits the proposal receiving the highest "score" 2 is generally selected as the contractor, and is essentially guaranteed (unless Congress decides to cancel the project) to be awarded a sequence of contracts for R&D, production, spare parts, maintenance, training, and so forth, over a number of years. (The extent of competition beyond the design stage - e.g. prototype competition, competition for production and maintenance - appears to be quite limited.) The contracts that are initially awarded to the successful firm are officially designated as "competitive" contracts. But most of the revenue that the firm will receive by virtue of having won the competition will come from subsequent, "follow-on" contracts, which are officially designated as "noncompetitive" contracts. In fiscal year 1984, for example, the value of noncompetitive follow-on contracts after design competition

Other countries with major defense industrial bases follow a similar system, but they often lack alternative domestic suppliers; hence foreign bids (e.g. in the aerospace industry) are necessary to avoid a domestic monopoly. 2 Projected price (or life-cycle cost) is only one of many factors affecting the total score, and has a small impact on the award of the initial contract. As discussed in Section 7, however, revisions in cost estimates appear to affect significantly the government's ultimate demand.

Ch. 15: Economics of Defense R&D

435

was 2.72 times as large as the value of competitive contracts associated with these competitions. Because the winner of the design competition is virtually assured of eventually being awarded the relatively large follow-on contracts, it is often suggested that contractors are willing to incur losses (by "buying in", or submitting bids below anticipated costs) on the initial competitive contracts. I think that it is natural to pose the question of why the government does promote mission-oriented R&D by sponsoring design competitions, in addition to doing so by directly contracting with firms to perform R&D. Perhaps the most compelling possible explanations are based on imperfect information considerations. A design competition appears to be an almost perfect example of a contest, in the sense defined by Barry Nalebuff and Joseph Stiglitz (1983) and others. In a contest, or competitive compensation scheme, the individual's reward or compensation (for example, whether he is awarded a contract) is determined only by his rank vis-a-vis his competitors, rather than by his "individualistic" output (or marginal product), as is the case in the classical model of pure price competition. These authors have shown that when the principal (in our case, the government) cannot directly and costlessly observe the level of input (effort) of the agents (contractors), rewards based on relative output are superior to payments based on individualistic output. They argue that competitive compensation schemes have "... greaterflexibility and greater adaptability to change in the environment than do other forms of compensation" (p. 41), so that contest may be preferred when the risk associated with common environmental variables (for example, the difficulty of achieving technical progress in a given area) is large. Moreover, "the use of a contest as an incentive device can induce agents to abandon their natural risk aversion and adopt "riskier" and more profitable production techniques" (p. 23). Inability of the principal to monitor the (relative) ability, or productivity, of various agents, is a second type of imperfect information which may render competitive compensation schemes optimal. Suppose that both government contracts and potential contractors are heterogeneous, in the sense that one firm is more qualified to perform a given contract that other firms, but the government is uncertain about the identity of the most qualified supplier. A number of theoretical models of markets characterized by this kind of imperfect information show that it is equilibrium behavior for sellers to invest in acquiring, and for buyers to rely on, signals of quality and ability. It may be useful to interpret design competitions as signaling phenomena, in which the signal of contractor ability that the government relies on is the score on the technical proposal. Factors other than imperfect information about contractor effort and/or ability may account for the existence of design competitions. For example, judging from periodic congressional hearings and reports on the subject [see, for example, US Congress (1969)], there is strong congressional demand for competition in procurement, perhaps because Congress believes that competitive procurement promotes economic efficiency or fairness. The passage of the Competition in Contracting Act also reflects this demand.

436

FER. Lichtenberg

2.1. How much private investment in defense R&D?

There are two types of evidence (aside from case studies and anecdotal evidence concerning company proposal efforts related to specific design competitions) concerning the amount of private investment in defense R&D. The first is provided by EM. Scherer's (1984) analysis of "linked" R&D and patent data of the largest R&D performing companies. Scherer attempted to classify each of about 15 000 US patents (obtained by 443 companies between June 1976 and March 1977) by "industry of use", that is, to identify the sector(s) of the economy in which (most intensive) use of the invention was anticipated. Two of the industries of use defined by Scherer were "defense and space operations" and "government, except postal and defense". He estimated the value of company-sponsored R&D "used" by these sectors to be $1206.3 million and $378.7 million, 11.3 and 3.6 percent, respectively, of the total amount of company-funded R&D ($10.64 billion) attributed to these companies. Thus, according to Scherer's methodology, the federal government is the primary beneficiary of about 15 percent of company-sponsored industrial R&D expenditure. The second is Lichtenberg's (1988) econometric study of the private R&D investment response to government procurement in general, and design competitions in particular. The basic strategy was to estimate, using longitudinal firm-level data, regressions of private R&D expenditure on three variables: the value of the firm's competitive contracts, of its noncompetitive contracts, and of its nongovernmental sales. (The sum of these three variables is total sales.) The coefficient of the first variable was the primary concern; the other two variables were included mainly to "control" for their influence and to provide benchmarks against which to measure the effect of competitive procurement. Variants of the model were estimated on annual 1979-1984 panel data for 169 industrial firms. A major defense buildup occurred during this period: the value of federal contracts more than doubled, whereas total sales increased by only about 35 percent. The estimates implied that a $1 increase in government sales tends to induce a 9.3 cent increase in private R&D while a $1 increase in nongovernment sales induces only a 1.7cent private R&D increase. These estimates enabled calculation of the fraction ' of the total induced increase in private R&D induced by the increase in government procurement during the period. The point estimate (standard error) of v was 0.528 (0.050). The point estimate implies that slightly over half of the total induced increase in private R&D between 1979 and 1984 was induced by the increase in government sales; the limits of the 0.95 confidence interval on this share are 0.430 and 0.626. The estimates also suggested that a $1 increase in competitive procurement induces a 54 cent increase in private R&D expenditure. As noted earlier, the prospect of substantial future noncompetitive follow-on contracts awarded to the winner of the design competition makes firms willing to make R&D investments which are large, relative to the value of the initial competitive contracts. The coefficient on

Ch. 15: Economics of Defense R&D

437

noncompetitive contracts was negative but insignificant, suggesting that the entire stimulus to private R&D from government procurement comes from competitive acquisition. The remainder of the empirical analysis was devoted to extending and amplifying the major finding that a considerable quantity (and share) of private R&D investment is induced by competitive procurement. The estimates indicated that most of the longrun response of private R&D to competitive procurement is a response to future procurement. They also suggested that competitive contracts (whether or not for R&D) have a large positive effect on private R&D, noncompetitive R&D contracts have an even larger negative ("crowding-out") effect, and other noncompetitive procurement has essentially no effect. The award of noncompetitive R&D contracts signals the end of the design and technical competition. At this stage of the procurement cycle, there are incentives for firms to reduce private R&D. Losers of the competition reduce spending because the prize is no longer at stake; the winner reduces spending because the government is now willing to directly sponsor the R&D via contracting. A $1 increase in noncompetitive R&D procurement tends to reduce private R&D by more than $2. Noncompetitive non-R&D has essentially no effect on private R&D investment. In contrast to previous studies of the effect of government R&D on private R&D, which did not control for non-R&D procurement and which did not distinguish competitive from noncompetitive procurement, Lichtenberg found that the net effect of R&D procurement on private R&D is negative. But non-R&D procurement, which is about five times as large, has a stimulatory effect, so the net effect of procurement in general is positive and quantitatively important. The government therefore appears to play a larger role in determining the allocation of the nation's scientific and technical resources, hence the rate and direction of technical progress, than is perhaps generally recognized.

3. Independent R&D subsidies The Department of Defense (DoD) encourages private military R&D not only by awarding prizes in design competitions, but also by subsidizing expenditures dedicated towards winning the prizes. In other words, DoD promotes this contractor activity both by creating returns to it and by reducing the (private marginal) costs of it. The DoD policy that provides a subsidy to private military R&D is its policy regarding so-called "independent" R&D. Independent R&D (IR&D) is contractorinitiated and directed technical effort that is not sponsored by, or required in performance of a contract or grant. The contractor selects the projects that comprise its IR&D program. The DoD and its contractors consider independent, or non-contract, R&D to be "company-funded", and it is reported as such in financial statements and official government R&D statistics. But under the Defense Procurement Regulations some of the costs of IR&D are "allowable", i.e., they can be included as indirect

ER. Lichtenberg

438

costs (overhead) in contractors' DoD contracts. Each year, ceilings on the amount of allowable IR&D costs are negotiated in advance by the DoD and each of its major contractors. The existence of a subsidy to private military R&D is due to the way in which these ceilings are negotiated or determined. To develop a model of DoD's policy of allowable-cost determination, we adopt the following notation: C= ceiling on allowable costs, X= total costs incurred, R= costs recovered from DoD,

S = total sales of the contractor, D= DoD sales of the contractor. R, the amount of costs recovered from DoD, is determined by the following formula D R = min (X, C).

(1)

Cost recovered is the fraction of firm sales accounted for by sales to DoD times either costs incurred or the ceiling, whichever is lower. In practice, X > C in the case of all firms. The maximum value of the ratio CIX = 1; the mean and median values are 0.823 and 0.872, respectively. Hence min(X, C)=C. Let us define O=DIS as the DoD share of sales; we shall treat this fraction as a parameter. Then Equation (1) reduces to R = OC.

(2)

If the firm has pure cost-plus contracts with the government, then its private cost P of conducting a level of investment S is P=X-R=X- OC,

(3)

and the marginal rate of subsidy (MRS) to contractor IR&D expenditure is dR dC MRS= dX = O-, dX dX'

(4)

The marginal private cost of investment, which is generally hypothesized to determine the equilibrium rate of investment, is I - MRS. The marginal rate of subsidy, hence marginal private cost, depends on the derivative dC/dX. The value of this derivative is an empirical matter. The ceiling C is set in an agreement negotiated by the firm and DoD prior to the investment of funds. One might, therefore, regard C as predetermined, i.e. independent of the realized expenditure X. In this case, dC/dX = 0, and IR&D reimbursement has

439

Ch. 15: Economics of Defense R&D

no effect on the marginal private cost of investment; it reduces only total costs, like a lump-sum payment. But the proposition that C does not depend on X, if it is true at all, is probably true only in the short run. Winston (1985, p. 22) maintained that "the accepted ceilings are set at a fraction of the contractor's anticipated IR&D expenditures that are deemed to meet the PMR [potential military relevance] criterion and to be of value to DoD". This hypothesis can be represented as follows: Cit = aXi,A

(5)

where Cit denotes the ceiling negotiated for firm i in period t, XA denotes anticipated expenditure during the period, and 0 < a < 1. The contractor announces his anticipated expenditure in the course of his negotiations with the government, but unfortunately we do not observe the announced value. Therefore, in order to make the model operational we need to specify a relationship between the unobservable XA and actual (current and/or lagged) expenditure X. Lichtenberg (1990) pursued two alternative approaches, one based on "rational", the other on "adaptive", expectations (or anticipations); we describe here only the second approach, in which XA is assumed to be forecast by a distributed lag function of past actual expenditures, with geometrically declining lag coefficients: XA = r (Xi,t_ + HXi,t 2

H2Xi,_3 +

).

(6)

The following equation can be derived from Equations (5) and (6): Cit = /Xi,t-1 + HCi,t-1,

(7)

where - a F. In this model there is a distinction between the short-run and (long-run) steady-state derivatives of C with respect to X; these are /f and f/i(1-H), respectively. The steady state is reached when Cit = Ci,t-l. It is interesting to note that an alternative "structural" model for determining C can generate a similar (although restricted) reduced-form equation. Suppose that the change in a firm's negotiated ceiling is proportional to the degree of "excess demand" for IR&D funds, as measured by the difference between lagged costs incurred and the lagged ceiling: Cit - Ci,t_l =

(Xi,t_1 - Ci,tl) .

(8)

Such a ceiling-adjustment process would be in keeping with that old saw about the politics of the budgetary process, that the more a department (or organization) exceeds its budget the more its budget will be increased (and that a department risks having its budget cut by not spending up to the limit). Adding Ci,t,_ to both sides, Cit =

Xi, t-_1 + (1 - fi) Ci,t_

(9)

In Equation (9) the coefficients on Xi,t_1 and Ci,t_1 sum to 1, whereas, this is not the case in Equation (7). This restriction can be tested by estimating both equations.

440

FR. Lichtenberg

Lichtenberg (1990) estimated these equations using cross-sectional data for about 275 IR&D sponsoring organizations. To attenuate heteroskedasticity, and because negotiations may focus on percentage rather than absolute changes, he estimated logarithmic versions of the equations, i.e. C and X were defined as logarithms of the respective variables. In this case /i and PI(1-H) are, respectively, the short-run and (long-run) steady-state elasticities of C with respect to S. The (long-run) steady-state derivative of C with respect to X is

1 -HX

X)LR

()

and the (long-run) steady-state marginal rate of subsidy (from Equation (4)) is MRSLR

D

C

SI- HX

(11)

This expression was evaluated at the sample aggregate values of D, S, C and X to obtain an "average" estimate of the (long-run) steady-state marginal rate of subsidy. The DCAA reports data on expenditures, ceilings, and reimbursements pertaining to another type of technical effort related to IR&D: the preparation of bids and proposals (B&P). The reasoning that led to the specification of Equations (7) and (9) applies as well to B&P as it does to IR&D expenditure. The disturbances of the IR&D and B&P equations are likely to be correlated, in part because there is some fungibility of expenditures between the two categories. More efficient parameter estimates could therefore be obtained by estimating the IR&D and B&P equations jointly, using Zellner's seemingly unrelated regressions (joint generalized least-squares) technique. This procedure also enabled testing the hypothesis that the parameters of Equation (7), hence MRSLR, are the same for B&P as they are for IR&D. Joint generalized least-squares estimates of the parameter f (the coefficient on Xt-1) in both equations were positive and significantly different from zero. Thus, conditional on the lagged ceiling, the higher was lagged expenditure, the higher the current ceiling. The hypothesis that /3+H=1, which is implied by the model (8), was decisively rejected in the case of IR&D but was not rejected (at the 0.1 level) in the case of B&P. The cross-model residual correlation was -0.093, suggesting that there is some substitution between IR&D and B&P ceilings and/or expenditure. The estimates implied that the government pays 41.3% of the marginal cost of IR&D and B&P expense. This is slightly (although significantly) less than the average subsidy rate of 47.4%. The marginal rates of subsidy to IR&D and B&P appear to be quite high. The R&D Tax Credit created by the 1981 Economic Recovery Tax Act provides a benchmark against which we can compare this subsidy. The provisions of the credit allowed firms to deduct from their tax liability an amount equal to 25% of their R&D spending above a certain base-period amount. But due to certain technical features of the credit particularly the way in which the base-period amount was calculated - the effective

Ch. 15: Economics of Defense R&D

441

rate of subsidy it provided was much lower than 25%; Baily and Chakrabarti (1988, p. 119) estimate that the effective rate of subsidy was 7-8%. Moreover, the credit was only temporary - it was originally due to expire at the end of 1985 after four and a half years. In contrast, DoD's IR&D policy has remained essentially unchanged since 1970, and predecessor policies can be traced back to the 1930's. The above estimates were derived on the basis of particular assumptions about (1) the relationship between the ceiling Cit and anticipated expenditures XA, and (2) the relationship between XiA and actual expenditure, but further analysis and calculations suggested that the estimates were not very sensitive to minor changes in model specification. While these estimates revealed the magnitude of the subsidy to private military R&D, they do not reveal how much (additional) R&D is undertaken due to the existence of the subsidy. To determine this with any precision, we would need to know the price elasticity of supply of R&D, a parameter about which little is known. But the existence of the subsidy is no doubt partly responsible for the fact that the marginal private R&D intensity of government (primarily defense) sales is much higher than the R&D intensity of non-government sales - 9.3% as opposed to 1.7%. Why does the government provide a subsidy for private military R&D, in addition to establishing prizes for innovation? Presumably if the subsidy were abolished the government could continue to induce the same amount of private R&D investment by increasing the value of the prizes. The answer may be that the cost to the government of promoting a given level of investment may be lower with a combination of prizes and subsidies than it would be with prizes alone. By providing a subsidy the government in effect shares with the contractor the risk of investment; the agency theory literature suggests that such risk sharing is often optimal from the principal's (government's) point of view. Given that the government wants to provide a subsidy, why does it not do so explicitly by announcing "we'll pay 40% of your independent R&D expense, with no ceiling", rather than by imposing ceilings that are in reality influenced by past expenditures? It is not possible to offer a definitive answer to this question, but two factors may account for the institutional arrangements we have described. First, while the major objective of IR&D policy is to strengthen the defense technology base by promoting private military R&D investment, for political reasons the DoD may need at least to appear to be "controlling costs" by imposing ceilings. Secondly, the government is interested not only in encouraging private innovation but also in transferring technologies and knowledge developed in the course of independent R&D to government officials. The negotiation of ceilings provides the government with the opportunity to monitor contractor activity, thereby facilitating technology transfer. 4. The profitability of defense (R&D and other) contracts Having described some of the mechanisms by which the government induces firms to invest in defense R&D, we turn to a discussion of the private and social returns to

442

P.R. Lichtenberg

this investment. First we consider the issue of the private returns, or profitability, of defense contracts 3 . The Department of Defense (DoD) has periodically conducted comprehensive studies of defense contractor profitability. The most recent of these studies was the Defense Financial and Investment Review (DFAIR) completed in June 1985 [DoD (1985)]. With regard to profitability, the primary goal of DFAIR was to compare "profits realized on negotiated defense contracts with the profits realized on comparable commercial work" (V-26). To accomplish this, DoD asked its contractors to "isolate the profits achieved on negotiated defense contracts [by] segregat[ing] and reporting] operating results of various categories of business conducted within individual business segments" (V-26) 4. DFAIR's overall conclusion was that the "profitability of defense contracts has not been unreasonable" (IX-3). DFAIR also performed a much more cursory comparison "of profitability for DoD business to profitability for commercial business within the same business segments" during the period 1977-1983 (V-43). The estimate of this difference for all product groups combined was about 8 percentage points. It was argued, however, that this difference in average profitability was entirely attributable to a single product group aircraft - for which the differential was 14 percentage points. But research by Rogerson (1992), Thomas and Tung (1992), and Lichtenberg (1992a), imply that such comparisons of the profitability of defense contracts to the profitability of commercial work may not tell us much about the true profitability of defense contracting. This is because most (about 3/4, on average) of the sales of government contractors are to commercial (non-government) customers, and government contractors appear to be able to shift some of the overhead and pension costs of their commercial operations to the government. Rogerson (1992) argues that the methods used by defense firms for calculating the cost of products - in particular, the allocation of overhead in proportion to directly charged labor use - enables firms to shift overhead from commercial to defense business. Under this overhead allocation scheme, by incurring an additional dollar of direct government costs, the firm's government revenue will increase by more than $1 - perhaps by as much as $1.20-$1.40 - since the share of overhead reimbursed by the government increases. Rogerson argues that the overhead allocation rule provides firms with the incentive to "under-capitalize production of products with cost-sensitive revenues [e.g., defense items] and over-capitalize production of products with costinsensitive revenues [e.g., commercial items]".

3 We will discuss theory and evidence concerning the profitability of defense contracts in general, not just defense R&D contracts. We are not aware of any empirical studies of defense R&D in particular; moreover, distinguishing between the profitabilty of defense R&D and non-R&D contracts is probably even difficult in principle. 4 Since a significant fraction of costs and assets are joint between government and commercial work, the validity of these "segregations" is questionable.

Ch. 15:

Economics of Defense R&D

443

Thomas and Tung (1992) argue that government contractors are able to reduce their cost of doing nongovernment business by overfunding pension plans when employees work on defense contracts (these contributions are reimbursed by the government) and withdrawing excess pension assets when employees work on commercial business. Evidence suggests that DoD faces considerable difficulty in reclaiming its share of any overfunding. Although, in principle, regulatory authorities such as the Defense Contract Audit Agency could prevent this cost-shifting, they may focus their limited resources on more obvious abuses, such as the inclusion of nonexistent costs or costs that clearly belong to commercial activities. One could tell several (we believe reasonable) stories why the government might allow defense contractors to earn a higher rate of return on their assets than other firms earn. First, the government may wish to attract a "queue" of potential suppliers 5. Second, potential suppliers may be heterogeneous with respect to the quality of goods and services they provide, and quality is not observable ex ante by the government. A number of models [see, for example, Stiglitz and Weiss (1981)] imply that under these conditions the equilibrium price will be above the market-clearing price: the government will find it optimal to pay a price above contractors' reservation price (and allow firms to earn higher returns than they could elsewhere). When evaluating the relative profitability of government contractors we need to distinguish between three different profitability concepts: r = profitability of government business done by government contractors; :r2 = profitability of commercial business done by government contractors; and 7r3 = profitability of commercial business done by non-government contractors. By shifting costs, contractors may earn above-normal returns on commercial work, and normal returns on government business. In this case, (rl-3r 2) is clearly a downward-biased estimate of the "effect of government contracting on profitability". We need to know, not only whether arl > Jr2 (and rt > Jr3), but also whether st2 > :r 3. Unfortunately, this question was not directly examined by DFAIR or by previous DoD studies. Lichtenberg (1992a) tried to make inferences about the difference between the profitability of commercial work performed by contractors and that performed by other firms, using Compustat Industry Segment on over 9000 industry segments, 8% of which were government contractors. He was able to measure only the overall profitability of the firms in his sample, but since about 80% of contractor assets are employed in commercial operations, overall contractor profitability is predominantly determined by the profitability of their commercial business. The measure of profitability he used was return on assets: the ratio of operating profit (loss) to identifiable assets. He estimated variants of the following equation: PROFITit = ai + 6t + I3GOVPOSit + uit, 5 This situation is analagous to the classic two-sector labor market model of a developing economy

formulated by Todaro (1969). There, urban employers offer an above-market-clearing wage and attract a queue of (unemployed) rural workers.

444

R. Lichtenberg

where PROFITi, denotes the profitability of segment i in year t (t = 1984, ... , 1989); GOVPOS = 1 if the segment had positive government sales, and zero otherwise; and ai and 6 t are fixed firm- and year-effects, respectively. In this model, f3 may be interpreted as the (population) mean profitability of defense contractors minus the mean profitability of other segments, and an estimate of fi can be used to assess the magnitude and significance of the profitability differential. All of the estimates of fi were positive and extremely significant, providing strong support for the hypothesis that the profitability of government contractors is substantially higher than the profitability of other segments. The estimates indicated that the mean profitability of segments that do any business with the government is 3.0 to 3.6 percentage points higher than that of segments with no government sales, whose mean return on assets is about 4.4%. Hence the profit rate of contractors is 68 to 82% higher than that of other segments. Other estimates implied that a 50 percentage-point increase in the ratio of government sales to total sales (GOVSHR) will increase return on assets by 4.5 percentage points - from 4.4% to 8.9%. These findings presumably did not surprise most DoD contract officers: survey evidence reported in DFAIR indicates that twice as many of these officers agreed with the statement, "the profits earned by defense contractors are too high" as disagreed with it. Lichtenberg (1992a) further showed that although, in theory, asset mismeasurement could bias upward his estimates of the government/nongovernment profitability differential, a limited attempt to correct for such mismeasurement produced no evidence that this was the case. He also presented evidence that highly governmentoriented segments are significantly less capital-intensive than less government-oriented segments and noncontractors (controlling for industry). A speculative interpretation of the findings of this and the previous sections is that DoD wants to provide generous subsidies and profits to contractors, but does not wish to appear to be doing so to members of Congress and the public. 5. The impact of defense and other R&D on productivity growth This section sketches a theoretical framework for assessing the rate of return to investment in R&D in general, and defense R&D in particular, and surveys some of the empirical evidence that has been developed using this framework. The basic model that has been most widely used to assess econometrically the economic impact of R&D investment is a production function generalized to include the stock of "knowledge capital" as a factor of production: Y(t) = F (K(t), L(t), Z(t)),

(12)

where Y, real output; K, physical capital service flow (usually assumed to be proportional to the capital stock); L, real labor input (e.g. hours worked6); and Z,

6

For simplicity, we ignore issues related to "labor quality" or human capital.

445

Ch. 15: Economics of Defense R&D

knowledge capital. Just as fixed capital is a distributed lag function of investment in plant and equipment (INV), knowledge capital is a distributed lag function of R&D investment (RD). Assuming geometric depreciation, K(t) =

i (1 - 6K) i INV(t - i),

Z(t) = i (1 - 6z)' RD(t - i).

(12a)

Let us assume that the production function (12) is Cobb-Douglas, and that there are constant returns to scale with respect to the conventional inputs K and L: Y = KaLI-aZ.

(13)

Knowledge capital is assumed to be a pure public good, whereas physical capital is a "congestible" good. Therefore to double output we need double only the quantities of the conventional inputs. /3 is the elasticity of output with respect to the stock of knowledge capital, and is the key parameter for measuring the impact of R&D or knowledge capital. Taking logarithms and differentiating with respect to time, Y' = aK' + (1 - a)L' + PZ',

(14)

where a prime after a variable denotes its growth rate, e.g. Y' (dln Y)/dt. From Equation (14), the rate of labor productivity growth (Y'-L') is determined by the rate of physical capital deepening (K'-L') and by the growth rate of the knowledge capital stock (Z'): Y'- L' = a (K'- L') + Z'

(15)

Z' also influences the growth rate of total factor productivity (TFP). TFP is defined as the ratio of output to an index of conventional inputs: Y TFP= KaLl-a Hence TFP' = Y' - [aK' + (1 - a)L'] = iZ'.

(16)

Equation (16) implies that the growth rate of total factor productivity is equal to the growth rate of the knowledge capital stock times the elasticity of output with respect to knowledge capital. The objective is to estimate fl. Suppose that we have time-series data on Y, L, K, and RD. (As we will discuss below, there may be serious problems with accurately measuring some of these variables, particularly Y.) To calculate TFP', we require an estimate of a, and to calculate Z', we require an estimate of 6 z. Most analysts have been willing to assume that the conventional factors are paid the value of their marginal products, and therefore that a may be set equal to capital's share

446

FR. Lichtenberg

in total production cost (or national income), sK7. Then TFP' can be estimated by TFP' = Y' - (sKK' + (1 - SK)L'). Getting a reliable estimate of 6z is more difficult. One feasible approach is (i) to calculate the Z series using the accumulation Equation (12a) under alternative assumed values of this parameter; (ii) to estimate Equation (16) using these different series; and (iii) to select the 6 z value which provides the best fit of the TFP growth Equation (16). Griliches and Lichtenberg (1984) took this approach using industry level data for US manufacturing, and found that 6 z = 0 provided the best fit. If we are willing to assume that 6z=0, i.e., that knowledge capital does not depreciate, then Equation (16) can be re-expressed in an even simpler form. dY Z dZ dY dZ RD TFP'= - = 2(17) dZY Z dZ Y Y where -2 (dY/dZ)= the marginal product of knowledge capital. Since the rate of depreciation is zero, the net change in the knowledge capital stock (dZ) is equal to gross R&D investment (RD). Equation (17) says that the rate of TFP growth (output growth controlling for the growth in conventional inputs) equals the ratio of R&D investment to output times the marginal product of knowledge capital. Under our assumptions, Q2 may also be interpreted as the rate of return to investment in R&D. Suppose that a firm spends an extra dollar on R&D this year. Since knowledge capital does not depreciate, its stock will be $1 higher in every future year. If the marginal product of knowledge capital is Q2, its output (revenues) will be $r in every future year. Hence, 2 is the rate of return on R&D investment. Because the labor and capital engaged in R&D are usually already included in the conventional input measures L and K - that is, they are "double-counted" - Q2 represents the excess rate of return to R&D investment - the additional return received for employing these factors in R&D, rather than in ordinary production. Equation (17) embodies the implicit assumption that the rates of return to different types of R&D (e.g., defense and non-defense) are equal. To estimate the rate of return to defense R&D performed in industry, and test the null hypothesis that it is equal to the rate of return to non-defense R&D, we need to disaggregate R&D and generalize the model as follows: TFP'= 12RD, + 2RD (18) or TFP' = 21lrdl + Q2 0 rdo,

(19)

where RD 1 denotes defense R&D; RDo (-- RD-RD 1) denotes non-defense R&D; and rdi (RDi/Y), i = 0,1. Equation (19) has been estimated on data at a number of levels 7 Paul Romer (1986) has argued that the elasticity of aggregate output with respect to physical capital

substantially exceeds capital's share innational income (about 30%), and that there are increasing returns with respect to conventional inputs. However Mankiw, David Romer and Weil (1992) have argued that the apparently high capital elasticity disappears when human capital is accounted for.

Ch. 15: Economics of Defense R&D

447

of aggregation - national, industry, firm and line of business - for a large variety of sectors and samples. Due to data limitations, in many cases government-funded R&D is used as a proxy for defense R&D, and privately-funded R&D is used as a proxy for non-defense R&D. Lichtenberg and Siegel (1991) obtained the following estimates of a firm-level regression of productivity growth on federally-funded (FRD/Y) and company-funded R&D intensity (CRD/Y): TFP*' = const. + 0.03(FRD/Y) + 0.35(CRD/Y), (0.8) (13.1)

(20)

where the figures in parentheses are t-statistics. The estimated rate of return to company-funded R&D was positive and highly significant, but they were unable to reject the null hypothesis that the return to federally-funded R&D performed in industry is zero. Further analysis suggested that the impact of federally-funded R&D on productivity growth was highest among small firms, although even for them the impact was not statistically significant. Lichtenberg (1992b) estimated a somewhat different but related model using crosssectional country-level data. His estimates also implied a decisive rejection of the hypothesis that the rate of return to government-funded R&D is equal to the return to privately-funded R&D, and inability to reject the hypothesis that the return to government R&D is zero. The apparently low social rate of return to government-funded R&D should be interpreted with caution. It does not necessarily imply that government R&D does not contribute to social welfare. A substantial fraction of government-sponsored R&D is devoted to the production of intangible goods, such as national defense and health, or the reduction of "bads", such as destruction of the environment, whose value is reflected imperfectly, at best, in national accounts data. Suppose that social welfare depends on two things - national security, and everything else (or guns and butter) and that the usual static tradeoff between the two (along the production possibilities frontier) applies. Suppose further that the production of butter, but not of guns, is included in GNP. The greater the fraction of R&D and other investment a society devotes to gun rather than butter production, the lower its measured growth rate will be. Under this view, the negative coefficient on government R&D-intensity merely reflects the existence of the static tradeoff between measured output (butter) and unmeasured output (guns). Even if the negative sign of this coefficient is not surprising, however, its magnitude is of interest, since it indicates the opportunity cost (in terms of conventionally-measured output) of government-funded research. It is sometimes hypothesized that this opportunity cost is substantially reduced by pervasive spillovers from government research to the private sector. Since these

8

In contrast, the returns to company-funded R&D were statistically significantly higher for large firms.

448

FR. Lichtenberg

estimates of the coefficient on government R&D should be interpreted as the opportunity cost of government R&D net of spillovers, it appears that, even net of these spillovers, the opportunity cost to the private sector of government research is substantial. The difference between the estimated returns to company and federal R&D may be overestimated due to (more severe) problems associated with measuring, and adjusting price indexes for, quality change in military industries. However there is evidence from case studies that the return to US-government-sponsored civilian R&D programs has also been very low. According to Cohen and Noll (1991, p. 365), only one of six major federal R&D commercialization programs (evaluated using retrospective benefit-cost analysis) achieved its objectives and can be regarded as worth the effort, and that program was killed.

6. The effect of defense R&D on nondefense R&D investment The evidence surveyed in the previous section suggests that, holding nondefense R&D constant, government-funded defense R&D has essentially no effect on productivity growth. But, in principle, defense R&D could have a nonnegligible (positive or negative) indirect impact on productivity growth by affecting the quantity of nondefense R&D investment. Suppose nondefense R&D responds to the level of defense R&D according to the formula rdo = 6rdl + u.

Then the total derivative of TFP' with respect to rdl is 21 + 6t0, which depends on the returns to nondefense R&D and on the response of rdo to rdl, as well as on the returns to defense R&D. If, as the econometric evidence suggests, Q1 = 0, defense R&D could exert a negative effect on the growth rate of productivity if it tends to depress or "crowd out" nondefense R&D expenditure. There have been a number of empirical studies of the effect of government (primarily defense) R&D expenditure on private R&D. Edwin Mansfield and Lorne Switzer (1984) simply asked a sample of industrial R&D officials how company expenditures would respond to specified, hypothetical changes in federal R&D support. On the basis of their survey, these authors concluded that changes in federal support tend to induce changes in the same direction in company spending, although some of their findings seemed to suggest that crowding out may in fact occur. Most other studies have been econometric analyses of the statistical relationship between company and federal R&D expenditure in either the aggregate time-series, industry-level cross sectional, or firm-level cross sectional dimension. With the exceptions of Jeffrey Carmichael (1981) and Neil Kay (1979), most early analysts obtained estimates of 6 that were positive and significantly different from zero. In particular, using linear specifications of the relationship between company R&D

Ch. 15: Economics of Defense R&D

449

expenditure and contract R&D expenditure, David Levy and Nestor Terleckyj (1983) obtained estimates around 0.27 from aggregate times-series regressions, Richard Levin (1980) obtained a similar estimate from industry cross-sectional data, and John Scott (1984) reported estimates around 0.07 from the Federal Trade Commission Line of Business ("below" firm-level) data. These investigators seemed to interpret their results as offering at least tentative support for the hypothesis that federal support of R&D performed by industry stimulates company R&D activity. But Lichtenberg (1984) argued that errors in the specification of the company-federal R&D relationship, and/or errors in the measurement of real R&D activity by source of financing, which these studies may have committed probably resulted in estimates of 6 that were substantially upward biased. He identified two potential sources of bias - failure to allow for "fixed (industry- or firm-) effects", and erroneous "deflation" of both types of R&D expenditure - and proposed and implemented appropriate, albeit perhaps partial, remedies. Using firm-level panel data on R&D expenditure, and industry-level panel data on both R&D expenditure and employment, he was able to investigate the issues of lags, fixed effects, and deflation bias in the analysis of the government/private R&D relationship. Lichtenberg (1984) began by estimating industry-level regressions of the change in company-funded R&D expenditure on the change in federal R&D expenditure, both deflated by the GNP deflator. The expenditure regressions were estimated on annual data for twelve manufacturing industries over the period 1963-1979. The coefficient on federal R&D was positive but insignificant in equations that included industry and/or year effects. Since one might reasonably hypothesize that private R&D responds with a lag (of greater than a year) to changes in government R&D expenditures, he reestimated the models with two lagged values of the federal R&D variable included; the (sum of) the government R&D coefficient(s) was again positive but insignificant. Indeed, the hypothesis that contract R&D expenditure stimulates company-funded research appeared to be undermined rather than strengthened by allowing for lagged effects. Lichtenberg also estimated similarly specified models using data on employment of R&D scientists and engineers, rather than R&D expenditure, by source of funding; due to the difficulty of properly deflating R&D expenditures, these employment data may serve as a better index of real R&D input than inaccurately deflated cost series. These estimates implied a strong, negative contemporaneous effect of federally-funded on company-supported R&D employment. When lagged values of federal R&D were included, some estimates implied essentially no, or perhaps a weakly positive, net effect of federal R&D, but other (less restrictive) models indicated a negative long-run impact: a federally funded increase of 100 R&D scientists and engineers this year will result in a reduction of company-sponsored employment of 39 within the year, essentially no change next year, and an increase of 7 in two years. The third and last type of data Lichtenberg analyzed was longitudinal, firm-level data on R&D expenditures as a fraction of sales. He performed regressions of the ratio of company R&D to sales on the ratio of federal R&D to sales, for the years 1967, 1972, and 1977, and corresponding regressions involving changes in these variables

450

FR. Lichtenberg

during the periods spanned by these years. Estimating models in both level and firstdifference form reveals the importance of allowing for fixed effects in the analysis of R&D performance. In the 1967 and 1972 cross-sectional regressions, the coefficient on federal R&D/sales was positive and significant, indicating that firms which do more contract R&D tend to perform more own-financed R&D. It is rather surprising that the coefficient in the 1977 equation was negative, as well as larger in magnitude and more significant than the 1967 and 1972 coefficients. Only Carmichael had found negative and significant cross-sectional coefficients on contract R&D, and his estimates were smaller (of the order -0.08). But coefficient estimates in the first-difference form of the model for 1967-1972, 1972-1977, and 1967-1977 were all negative and highly significant. The point estimate of -0.26 for the ten-year interval as a whole is similar in magnitude to the coefficients in the industry-level R&D employment regressions reported above. Thus improved measurement of, and specification of the relationship between, company and federal R&D yielded evidence that was quite inconsistent with the hypothesis that defense R&D investment tends to stimulate nondefense or civilian R&D. But even these tests are not entirely conclusive, because the distinction between government and private R&D is not the same as the distinction between defense and civilian R&D, and because the defense/civilian R&D relationship may be characterized by very long lags.

7. How do Government decision-makers respond to cost information yielded by defense R&D? A "major thesis" of Peck and Scherer's (1962) seminal monograph on defense procurement was that "there is uniqueness in both the magnitude and the diverse sources of uncertainty in weapons acquisition" (p. 17). They defined two broad classes of uncertainties: internal and external. The extent of internal (and possible also of external) uncertainty about a weapon system is greatest at the beginning of its "life cycle". As resources are devoted to the system's research and development, information about the true cost of acquiring the system is generated, and the degree of technological uncertainty is reduced. How do defense decision-makers - the people in the Pentagon and Congress who make decisions about the allocation of defense resources - respond to the arrival of new information concerning the cost of weapons acquisition? Because in most economic settings, it is inefficient not to change behavior in response to new information, this question relates to the degree of efficiency of defense procurement, an issue of considerable concern to policy-makers and the public. In other words, how elastic is the government's demand for individual weapon systems? When the government's estimate of the cost of acquiring a given weapon changes as a consequence of data generated in the course of R&D, how much (if

Ch. 15: Economics of Defense R&D

451

at all) does the desired "buy" (quantity) change? An important determinant of the elasticity of demand for a specific weapon is the degree of substitutability between it and other weapons actually or potentially being acquired. Scherer (1964, pp. 54-53) suggests that even systems that have no obvious technical or operational substitutes are "threatened" by rival systems in the bureaucratic competition for budgetary support. In the early 1960's, for example, (offensive) Polaris missiles and the Nike Zeus ballistic missile defense programs were regarded by top Defense Department (DoD) officials as substitutes, in effect. Early in a program's life cycle, there is often intense competition among a few potential suppliers to develop and produce the weapon. But once the design and technical competition is over, the system is likely to be produced on a sole-source basis. The winning contractor then enjoys a monopoly with respect to the supply of the system. The extent of the contractor's market power, and his ability to earn monopoly profits, are inversely related to the elasticity of the government's demand. The higher the demand elasticity, the lower the price the contractor will seek to set and the lower his profit. It may also be the case that the higher the demand elasticity, the lower the optimal level of government expenditure to monitor and regulate the costs and profits of defense contractors. One research strategy for determining the demand elasticity is to examine empirically the relationship between revisions in cost estimates and revisions in quantity estimates across the population of major weapon systems. The revisions are from original or "baseline" estimates (made around the start of full-scale development) to "current" estimates (made at a subsequent date). Lichtenberg (1989a) formulated a simple model of revisions in weapons system quantity and cost estimates. He postulated that at any given time in the life cycle of a weapons system, the Pentagon has estimates of (the slopes and intercepts of) both the marginal cost schedule (the supply curve) and the marginal benefit schedule (the demand curve) of the system 9 . In particular, the Pentagon has such estimates at two times: the date at which full-scale development begins (time 0), and at a later date (time t). (Estimates made at time 0 are referred to as "baseline estimates".) Assume that the supply and demand schedules are log-linear. The baseline schedules may be written: InMC = 0 - aln Q,

(21)

InMB = 00o-

(22)

-1 In Q,

where MC denotes marginal cost, MB denotes marginal benefit, Q denotes quantity, and is the elasticity of demand". For simplicity, he also assumed that the baseline 9 Marginal cost consists primarily of production costs, since development costs are essentially fixed indeed "sunk" costs: they are independent of quantity purchased. '° Because the cost of developing and producing high-technology equipment is subject to considerable

uncertainty, a disturbance term should be added to equation (21); it is suppressed here for simplicity.

452

FR. Lichtenberg

quantity chosen by the government is the one satisfying the equality between InMC and InMB: InMC =lnMB4 lnQo=

-

(23)

In order for Q0 to be an equilibrium quantity, it must be the case that ,fi- > a: the demand curve must be more negatively sloped than the supply curve. Because, as we shall see below, weapons systems typically exhibit decreasing marginal costs, the condition is not a trivial one. As time passes following the start of full-scale development, decision-makers will revise their estimates of the supply and/or demand schedules. Information generated during the course of development about the cost or difficulty of acquiring the system would result in supply-curve revisions. Changes in the actual or perceived nature of the "threat" from enemy forces, and revisions in supply-curve estimates of other (complementary or substitute) systems under development would result in demandcurve revisions. We represent the Pentagon's estimates of the supply and demand curves at time t (t > 0) as follows: In MC = t - a In Q,

(24)

InMB = Ot -

(25)

-1 In Q.

Lichtenberg assumed that only the intercepts, and not the slopes, of the supply and demand curves are subject to revision; data limitations would not allow him to identify changes in the slopes. Equilibrium quantity at time t therefore satisfies: '

In Qt = _

(26)

The revision in equilibrium quantity between time 0 and time t can be calculated by subtracting Equation (23) from Equation (26): ln()Qt

Q

-5 - +

-00

' (27)

The log-change in quantity is due to both supply- and demand-curve revisions, each divided by the difference between the slopes of the curves. Equation (27), along with the baseline supply curve (21), can under certain assumptions provide a basis for estimating the parameters a and /f. Lichtenberg had data from the Selected Acquisition Reports, for 84 major weapons systems, on the quantity- and supply-shift variables ln(Qt/Qo) and (6t-60). Unfortunately, he lacked data on the demand shift (t-e 0 ). But suppose, as seems reasonable, that demand shifts are uncorrelated with supply

Ch. 15:

Economics of Defense R&D

453

shifts across weapons systems. Moreover, assume that a and fi do not vary across weapons systems. Then the regression equation, In(Qt ) = - (-l

- a)' (6t- 60)i + Ei,

(28)

where the i subscript denotes weapon system i and E is a disturbance term, will yield a consistent estimate of the nonlinear function of the parameters -(3-'-a)-l. Of course, neither a nor /f can be separately identified from this equation alone, but the available data permitted him to estimate another equation which identified a. By simultaneously estimating the system of two equations, he was able to identify both parameters. His empirical results may be summarized as follows. When the data were standardized by program base year - in effect comparing a program only with those other programs entering full-scale development at about the same time - there was a significant negative correlation between quantity and cost changes. The estimated elasticity of demand was 0.55, and was significantly different from both zero and unity. This suggests that the government's demand for specific weapons is inelastic, but not perfectly inelastic. The estimates also imply that weapons acquisition is characterized by increasing returns: the mean and median values of the elasticity of total cost with respect to quantity were 0.78 and 0.72, respectively. Further analysis revealed that the negative correlation between quantity and cost revisions - hence the nonzero demand elasticity - was entirely attributable to one component of cost revisions: those associated with changes in the acquisition schedule" . The elasticity of quantity with respect to schedule-related cost increases is about twice as great as the elasticity with respect to cost increases generally. In principle, it is possible that schedule-related cost increases are due to demand-induced stretch-outs of programs rather than supply-related, or technological, shocks. But it is not clear on theoretical grounds that unobserved demand shocks could account for the correlations he observed, and the demand-shock interpretation was also not supported by one econometric attempt to correct for it.

8. Are defense R&D programs dynamically optimal? Grossman and Shapiro (1986) studied the optimal pattern of outlays for a single firm pursuing an R&D program over time. Treating dynamic R&D investment as an optimal control problem facing a single firm, they characterized the profile of R&D expenditure as a single R&D project progresses. In this section, we briefly review the assumptions and implications of their model and report an analysis of data contained in the Defense

" As Hartley and Tisdell (1981, p. 355) have observed, system modifications are another important source of cost escalation.

454

FER. Lichtenberg

Technical Information Center's IR&D data bank to determine whether they are broadly consistent with the theory. Grossman and Shapiro assume that a firm seeks a prize of size W and that to obtain this prize it must "travel" a distance L. The instantaneous rate of advance is determined by the rate of R&D expenditure. There are decreasing returns to effort at any point in time, given that some progress is being made, but there may be a fixed start-up cost at any moment. The firm's problem is to choose expenditures at every point in time up to some terminal date to maximize the present discounted value of net profits, subject to the constraint that total progress attained at the termination date be sufficient to complete the project. The major implication of the model is that it generally is not optimal for a firm to devote a constant level of resources to its research program, even if the relationship between effort and progress is unchanging. Rather, the firm should vary its R&D expenditure directly with the current expected value of the project. In many circumstances, this value will increase as the firm achieves progress. "Circumstances" refers to whether or not there is uncertainty about the "difficulty" of the project (i.e., the distance to be traveled) or about the relationship between effort (expenditure) and progress. They show that if neither type of uncertainty is present (the deterministic case), it is optimal to increase effort over time as the project nears completion, in part because discounted R&D costs can be decreased for any given duration of the project by shifting expenditures from early to later stages. If there is only uncertainty concerning the relationship between effort and progress, a monotonously increasing effort profile remains optimal. If there is uncertainty about the difficulty of the project, the optimal pattern of investment depends on the hazard rate function. If this function is everywhere nondecreasing - i.e., if whenever success is not realized, researchers become more optimistic that a breakthrough is imminent - the optimal program again involves rising R&D outlays over time. If, alternatively, the firm learns that the project is more difficult than was originally believed, it may be optimal to reduce the scale of the R&D program, or even to abandon it entirely. Grossman and Shapiro argue that the deterministic case applies more closely to development projects than to pure research programs. Hence, their model implies that development projects should exhibit increasing effort profiles, whereas projects subject to greater uncertainty as to their difficulty (such as basic research projects) would not necessarily do so, and might be expected to exhibit flatter (or even negatively sloped) profiles. Some of the assumptions of the model are highly unrealistic, and one might therefore not expect actual data to be consistent with the theory. First, the model omits all R&D rivalry, i.e., strategic interaction among firms. Also, the assumption that the progress function is stationary over time may be more reasonable in some contexts than in others. It may be easier to make progress once some initial groundwork has been laid, though the groundwork itself cannot be rushed due to diminishing returns to more effort on this "subproject". This case would tend to reinforce the GrossmanShapiro results. But it may also be easy to make progress early on, due to a long list

Ch. 15:

Economics of Defense R&D

455

of "easy ideas to try". Then progress may slow once the difficult stage of the program is reached. Despite the possible lack of realism of the assumptions, it is of interest to assess the degree of consistency of the theory with the data. Lichtenberg (1989b) used data on over 9000 Independent R&D projects to test the hypothesis that the rate of investment in a project tends to increase as the project approaches completion. He computed, for each continuing project, the ratio (denoted R) of expected investment in the next year to average investment to date. To guard against the influence of a relatively small number of observations with very large values of R (some of which may have been outliers), he excluded from his sample observations with values of R greater than 4; his hypothesis tests are therefore likely to be conservative. The mean value of R for all projects was 1.30 and highly significantly different from 1; the median was also greater than 1 (1.14), albeit smaller than the mean. This is consistent with the general implication that the rate of investment increases as the project approaches completion. But the data were not consistent with the implication that R should be lower in the case of projects involving greater uncertainty about the difficulty of completion. Mean and median values of R were essentially the same for basic research projects as they were for applied research and development; Lichtenberg (1989b) presented evidence that the latter two categories are subject to less uncertainty than the former. He also used the IR&D data bank to develop some stylized facts about R&D investment at the project level. These are: (1) Research projects (both basic and applied) are longer and less intense than development projects; (2) the elasticity of cumulative investment with respect to project duration is greater than 1 for research projects and less than one for development projects; (3) the distributions of duration, average investment, and cumulative investment are highly skewed; (4) the shape of the duration distribution is close to lognormal, indicating that the conditional probability of project completion initially rises and then declines; (5) the degree of uncertainty about the project completion date is greater for basic research and concept formulation projects than it is for applied research and development projects.

9. Conclusions The major conclusions of this chapter may be summarized as follows: (1) Direct R&D contracting is not the only way in which the government induces private firms to invest in defense R&D; it also does so by sponsoring design competitions and providing subsidies to "independent" R&D. (2) A considerable quantity (and share) of private R&D investment is induced by competitive defense procurement. (3) The effective rate of subsidy to independent R&D exceeds 40% - much higher than the apparent (nominal) subsidy or the subsidy to other R&D. (4) The profitability of government contractors is 68 to 82% higher than the profitability of other producers; this is consistent with the hypothesis that the former are able to shift costs from their

456

FER. Lichtenberg

commercial operations to the government. (5) Both micro and aggregate estimates of the (social) "rate of return" to investment in government-funded (largely defenserelated) R&D - or its impact on productivity growth - are insignificantly different from zero, and are much smaller than estimates of the return to privately-funded R&D. (6) The best available evidence does not support the hypothesis that defense R&D tends to stimulate civilian R&D, and thereby has an indirect positive effect on productivity growth. (7) In two respects, the conduct of defense R&D and procurement appears to be "efficient" (or at least fails to be grossly inefficient). First, defense decisionmakers do seem to respond to the arrival of new information about the cost of weapons acquisition yielded by R&D: the elasticity of demand for weapons is significantly greater than zero (it is about 0.55). (8) Second, most independent R&D projects appear to be "dynamically optimal": the rate of investment increases as the project approaches completion.

References Baily, M., and A.K. Chakrabarti, 1988, Innovation and the productivity crisis (The Brookings Institution, Washington, DC). Carmichael, J., 1981, The effects of mission-oriented public R&D spending on private industry, Journal of Finance 36, 617-627. Cohen, L., and R. Noll, 1991, The technology pork barrel (The Brookings Institution, Washington, DC). DoD (US Department of Defense), 1985, Defense financial and investment review (Department of Defense, Washington, DC). Griliches, Z., and FR. Lichtenberg, 1984, R&D and productivity at the industry level: Is there still a relationship?, in: Z. Griliches, ed., R&D, patents, and productivity (University of Chicago Press, Chicago, IL) 465-496. Grossman, G.M., and C. Shapiro, 1986, Optimal dynamic R&D programs, Rand Journal of Economics 17, 581 593. Hartley, K., and C. Tisdell, 1981, Micro-economic policy (Wiley, New York). Kay, N., 1979, The innovative firm: A behavioral theory of corporate R&D (St. Martin's Press, New York). Levin, R., 1980, Toward an empirical model of Schumpeterian competition, unpublished paper (Yale University, May). Levy, D., and N. Terleckyj, 1983, Effects of government R&D on private R&D and productivity: A macroeconomic analysis, Bell Journal of Economics 14, 551 561. Lichtenberg, ER., 1984, The relationship between federal contract R&D and company R&D, American Economic Review 74, 73-78. Lichtenberg, F.R., 1988, The private R&D investment response to federal design and technical competitions, American Economic Review 78, 550-559. Lichtenberg, FR., 1989a, How elastic is the government's demand for weapons?, Journal of Public Economics 40, 57-78. Lichtenberg, ER., 1989b, IR&D project data and theories of R&D investment, Journal of Economic Dynamics and Control 13, 271-282. Lichtenberg, F.R., 1990, U.S. government subsidies to private military R&D: The Defense Department's independent R&D policy, Defense Economics 1, 149-158.

Ch. 15: Economics of Defense R&D

457

Lichtenberg, ER., 1992a, A perspective on accounting for defense contracts, The Accounting Review 67, 741-752. Lichtenberg, F.R., 1992b, R&D investment and international productivity differences, in: H. Siebert, ed., Economic growth in the world economy (J.C.B. Mohr, Tfibingen, Germany) 89-110. Lichtenberg, ER., and D. Siegel, 1991, The impact of R&D investment on productivity: New evidence using linked R&D-LRD data, Economic Inquiry 29, 203-228. Mankiw, N.G., D. Romer and D. Weil, 1992, A contribution to the empirics of economic growth, Quarterly Journal of Economics 107, 407-437. Mansfield, E., 1971, Technological change (W.W. Norton, New York). Mansfield, E., and L. Switzer, 1984, Effects of federal support on company-financed R&D: The case of energy, Management Science 30, 562-571. Nalebuff, B.J., and J. Stiglitz, 1983, Prizes and incentives: Towards a general theory of compensation and competition, Bell Journal of Economics 14, 21-43. National Science Board, 1993, Science and engineering indicators - 1993 (US Government Printing Office, Washington, DC). Peck, M.J., and F.M. Scherer, 1962, The weapons acquisition process: An economic analysis (Harvard Business School, Boston, MA). Rogerson, W.P., 1992, Overhead allocation and incentives for cost minimization in defense procurement, The Accounting Review 67, 671-690. Romer, P.M., 1986, Crazy explanations for the productivity slowdown, NBER Macroeconomics Annual 2, 163-202. Scherer, EM., 1964, The weapons acquisition process: Economic incentives (Harvard Business School, Boston, MA). Scherer, EM., 1984, Using linked patent and R&D data to measure interindustry technology flows, in: Z. Griliches, ed., R&D, patents, and productivity (University of Chicago Press, Chicago, IL) 417-461. Scott, J., 1984, Firm versus industry variability in R&D intensity, in: Z. Griliches, ed., R&D, patents, and productivity (University of Chicago Press, Chicago, IL) 233-245. Stiglitz, J., and A. Weiss, 1981, Credit rationing in markets with imperfect information, American Economic Review 71, 393-409. Thomas, J., and S. Tung, 1992, Incentives under cost-reimbursement: Pension costs for defense contractors, The Accounting Review 67, 691-711. Todaro, M., 1969, A model of labor migration and urban unemployment in the less developed countries, American Economic Review 59, 138-48. US Congress, 1969, Senate, Committee on the Judiciary: Competition in defense procurement, hearings before the subcommittee on antitrust and monopoly (US Government Printing Office, Washington, DC). Winston, J., 1985, Defense-related research and development in industry, Congressional Research Service Report No. 85-205 S, 18 October 1985.

Chapter 16

INDUSTRIAL POLICIES IN THE DEFENSE SECTOR KEITH HARTLEY University of York

Contents Abstract Keywords 1. Introduction 2. The stylized facts and the policy problem 3. The economics of standardization 4. Taxonomy: project appraisal and procurement options 5. International collaboration 5.1. Collaboration: a public choice analysis 5.2. Empirical results

6. 7. 8. 9.

Licensed and co-production Offsets The military production function: internal efficiency Conversion 9.1. Direct conversion 9.2. Re-allocating resources and public policy

10. Conclusion: a research agenda References

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

460 460 461 461 465 471 473 475 477

479 481 483 484 485 486 487 487

460

K. Hartley

Abstract Voluntary military alliances, such as NATO, are often criticized for failing to exploit the opportunities for equipment standardization and free trade. However, nationalism means that governments adopt a variety of industrial policies for purchasing defense equipment, leading to departures from the competitive free trade model. Different procurement policies offer varying combinations of military and national economic benefits. Work sharing can be achieved through international collaboration, licensed production and offsets. Policies designed to improve efficiency in equipment procurement also affect the military production function. Job losses associated with the closure of military bases and defense plants raise issues of conversion.

Keywords Collaboration, conversion, cost-benefit analysis, licensed and co-production, market failure, military production functions, NATO free trade area, offsets, procurement options, public choice, specialization, standardization, strategic trade theory

Ch. 16: Industrial Policies in the Defense Sector

461

1. Introduction The supply side of military alliances is an under-researched field (see Chapter 5 of this Handbook). Critics of voluntary military alliances, such as NATO, stress their inefficiencies in the provision of both defense equipment and armed forces. They point to the wasteful duplication of costly research and development (R&D) programs for defense equipment and small national orders so that there is a failure to exploit the available economies of long production runs. Similarly, amongst armed forces there seems to be wasteful duplication as each member state of the alliance provides its own defense ministry, army, navy and air force, their associated training and support facilities and the necessary infrastructure of military bases and communications networks. On this basis, it seems that voluntary military alliances "fail" to exploit the available opportunities for international specialization based on comparative advantage and the associated potential for exploiting the resource savings from decreasing cost industries. These issues will become even more important in an era of falling defense budgets. Whilst many of the examples in this Chapter are based on NATO experience, the general supply-side principles apply to any military alliance. The Chapter starts by outlining the policy problem and considers the economics of standardization; it then presents a taxonomy of equipment procurement options, each of which is analyzed using a cost-benefit framework. Consideration is also given to internal efficiency issues in the armed forces and their implications for the military production function. Inevitably, policies aimed at improving efficiency involve winners and losers. Some groups will be made worse-off as reflected in the closure of defense plants and military bases. A final section focuses on conversion issues.

2. The stylized facts and the policy problem Military alliances often focus on burden sharing as reflected in each member state's share of GNP devoted to defense. Such ratios are misleading since they measure inputs and not the output of defense, so that they provide no indication of the efficiency of military effort. At the same time, the rising costs of weapons and, at the margin, a greater preference for civil and social welfare expenditures (e.g., roads, hospitals, schools) has meant that in a period of comparative peace members of a military alliance will be under continued domestic electoral pressures to reduce defense budgets. In these circumstances, a military alliance is likely to place much greater emphasis on methods of increasing the efficiency of existing (and falling) levels of defense spending. A whole new jargon has emerged with objectives expressed in such terms as standardization, inter-operability, rationalization, managed competition, two-way streets, collaboration, co-production and offsets. For NATO, the policy problem was seen to be a failure of the allies to agree on common tactics, common training, common weapons and international specialization resulting in a waste of resources and adverse effects on military effectiveness (e.g.,

462

K. Hartley

an excessive fragmentation of NATO defense procurement markets). As a result, NATO was believed to be an inefficient organization in both its weapons markets and its armed forces. Equipment is not standardized nor inter-operable, whilst weapons production is "unco-ordinated" with "wasteful duplication" of costly programs. Standardization is advocated as the preferred solution with this policy objective embracing "commonality" in tactics, weapons, training and logistics. Attractive though it might seem, such "commonality" or standardization has major policy implications for each member of a voluntary club. To replace "wasteful duplication" in R&D and production requires nations with different preferences (and their governments seeking re-election) to agree on which weapons to buy, how and from whom. Inevitably, efforts to create a NATO free trade area in weapons will involve gainers and losers with no guarantees that the losers will be compensated. Similarly, nations in a voluntary military alliance will be reluctant to sacrifice the independence and guaranteed protection offered by their own national army, navy and air force for the dependence and interdependence associated with the international specialization of armed forces. Such international specialization based on comparative advantage might require, say, Germany to specialize in providing armored forces for the collective defense of the alliance, with France providing air defense, Turkey supplying infantry forces, the UK providing naval forces and the USA offering communications, early warning and the strategic nuclear umbrella. The gains from sacrificing nationalism and independence are believed to be substantial. One of the pioneering efforts in the field estimated annual waste in NATO at some $11 billion [1975 prices: Callaghan (1975)]. This estimate rather cavalierly assumed that waste comprised all European defense R&D expenditure together with 10% of US procurement, 25% of European procurement, 10% of direct American annual NATO cost and 15% of European general purpose force expenditures. The preferred solution was a North Atlantic Common Defense Market characterized by international specialization based on comparative advantage, with Europe and the USA each developing, producing, supporting and providing the other with the equipment it is best able to make. The result was expected to be the end of "wasteful duplication" and the creation of long production runs leading to lower unit costs of standardized and inter-operable equipment, with a "two-way street" in military trade between Europe and the USA. In addition to the economic benefits, standardization was expected to increase NATO military effectiveness from 30-50% for most units to as much as 300% for certain tactical air units [Callaghan (1975, p. 111)]. The case for a NATO Common Defense Market or Free Trade Area appears impressive. The performance of defense industries in any military alliance is increasingly important in an era of expensive equipment, rising weapons costs and falling defense budgets. Indeed, cuts in military spending will force procurement agencies and the armed forces to change their traditional purchasing policies in order to seek cheaper methods of acquiring weapons (e.g., by buying off-the-shelf). Table 1 gives examples of the high costs of modern defense equipment. Not only is equipment expensive but cost trends are upwards reflecting improvements in quality

463

Ch. 16: Industrial Policies in the Defense Sector Table 1 Unit procurement costs US equipment

Average unit cost (million $, 1995 prices)

Army Comanche helicopter Advanced field artillery system

19 7

Navyb

5675 1700 825 90

Nuclear aircraft carrier New attack submarine Aegis destroyer or successor F/A- 8E/F (fighter aircraft) Air Forceb F22 (fighter aircraft) C17 (transport aircraft) F16 replacement (strike aircraft)

105 260 45

Rates of cost escalation: UK and US procurement Equipment Fighter aircraft Guided missiles Frigates Helicopters Destroyers Submarines a

Cost escalation' (% pa) 11 11 10.5 9.5 9 9

Sources: CBO (1994), Pugh (1993).

b Average unit cost for Navy and Air Force equipment is mid-point of low and high cost figures.

Cost escalation refers to unit production costs adjusted for inflation and differences in production quantities.

and performance. New military equipment costs in real terms (unit for unit) far more than the old items which it replaces: typically such cost increases average about 10% per annum resulting in a doubling in cost every 7.25 years [Pugh (1993)]. The changing ratio of unit costs to defense budgets means that such cost increases affect the numerical strengths of armed forces, with consequent effects upon suppliers (i.e., economics as the ultimate arms controller). It also means that policies to improve the efficiency of procurement only buy a brief respite once for all time. With cost increases of 10% per annum, cost reductions of some 20% (e.g., as have resulted from greater competition) achieve relief from cost escalation for about 2 years [Pugh (1993)]. Such cost trends have led some commentators to make long-run forecasts of a one-ship navy and a one-aircraft air force [Starship Enterprise: Kirkpatrick and Pugh (1983)].

K. Hartley

464 Table 2a Duplication Number of types

Equipment 1 2 3 4 5 6 a

Tanks Armored combat vehicles Fighter-bomber aircraft Ground-attack-trainer aircraft Anti-tank helicopter Frigates

Europe a

USA

4 12 7 6 7 11

1 3 5 1 3

Europe = EC plus EFTA. Table 2b Output of combat aircraft

Country

Orders for national forces

Total orders

France Mirage 2000 Rafale

325 312

542 312

UK Hawk Sea Harrier

176 75

746 98

Sweden Gripen

300

300

Internationalcollaboration AMX (Italy-Brazil) Alpha Jet (France-Germany) Tornado (UK-Germany-Italy) EF 2000 (UK-Germany-Italy-Spain) Harrier AV8B (UK-USA)

192 351 872 602 405

192 503 992 602 439

1103 2222 >979 442

1251 3989 >1410 442

USA F15 F16 F18 F22 a

Source: Jane's (1994).

Inefficiency in NATO weapons markets is reflected in the "wasteful duplication" of costly R&D programs and relatively short production runs. The stylized facts are given in Tables 2a,b which show two distinctive features. First, the proliferation of different

465

Ch. 16: Industrial Policies in the Defense Sector

types of defense equipment in NATO and especially within Europe (Table 2a). For example, NATOs Rapid Reaction Corps is expected to fight together with six kinds of helicopter, ten different makes of vehicles, seven kinds of anti-tank weapons and four different communications systems [Beard (1993)]. Second, the major differences in the scale of output between the USA and individual European nations (Table 2b). Whilst the examples of scale differences in Table 2b are for aircraft, the same point applies for a whole variety of equipment, including main battle tanks, armored combat vehicles, missiles, helicopters and warships. For example, between 1979 and 1991, the US forces purchased almost 7800 M-1 tanks compared with the UK where the Army bought 426 Challenger I tanks.

3. The economics of standardization The simple economic case for standardization and free trade in defense equipment markets is shown in Figure 1. Consider two nations, with nation A operating initially on the long-run average cost curve LAC 1 purchasing Q2 at a unit cost of Cl; whilst nation B on a lower long-run average cost curve purchases Q1 at a unit cost of C 2. In each nation, decreasing costs show that a greater output leads to cost savings. However, if both nations combined their orders (Ql + Q2 = Q3) and purchased common equipment from the lowest cost supplier (nation B), then the total output of Q3 would be provided at a unit cost of Co. As a result of both standardization and free trade, nation A saves C -Co and nation B saves C 2 - Co. In addition to reductions in unit productions costs, there will be further savings from reduced duplication in R&D (e.g., one R&D program rather than two). If standardization is as beneficial as claimed, why has it not occurred? Are there some obvious market failures preventing worthwhile international transactions or are the existing arrangements optimal once it is recognized that all policies and exchange Average Cost

Cl

------

L

LAC 1 (A) Co01~~~~!

LACO (B)

0 Q 3 Figure 1. Standardization2 Figure 1. Standardization and free trade.

Output

466

K. Hartley

involve costs? In fact, governments themselves are often a major barrier to free trade in defense equipment and hence the source of market failure. Governments can ensure that their national defense markets are characterized by entry barriers, preferential purchasing and restrictions on the transfer of technology so that anticompetitive behavior results from the demand side of the market. Such policies reflect a traditional desire for independence and security of supply which necessitates protection of a nation's defense industrial base (DIB). Also, nations might dislike the expected outcome of free trade within a military alliance. In NATO, for example, it is predicted that a free trade area will result in the USA specializing in high technology defense equipment with Europe confined to "metal bashing" and having to accept a reduction in the size of its "key" technology industries such as aerospace. National governments can always justify support for their DIB by pointing to its apparent industrial and economic benefits in the form of protecting jobs, supporting high technology and spin-off and contributing to the balance of payments (see Chapter 14 of this Handbook). These apparent benefits are attractive to votemaximizing governments, but they should be seen as claims made by governments rather than economic statements about the social benefits of import replacement. Governments are entitled to believe that imports are "bad" and that exports are "good". However, if economists accept that the international division of labor and international trade are welfare enhancing, it follows that both exports and imports are welfare enhancing: imports create jobs by freeing resources for more productive alternative uses elsewhere in the economy. Indeed, in defense markets governments are often influenced by claims about technology and as a result ignore the opportunity cost question: would the resources currently allocated to the DIB make a greater contribution to technology, jobs, the balance of payments and ultimately GDP (economic welfare) if they were used elsewhere in the economy? There are circumstances in which state intervention in defense markets can be justified. One set of circumstances arises from market failures in the form of public goods, external economies and social benefits, including failures in R&D markets [Arrow (1962), Demsetz (1969)]. Strategic trade policy provides a further case for state intervention. It arises where international trade is characterized by strategic rivalry between a small number of giant firms or "national champions" of different countries, or indirectly between governments acting on their behalf. In such markets, an established firm might try to create strategic entry barriers to maintain and enhance its market power. However, by using subsidies, governments have the potential to influence exit and entry decisions in oligopoly industries, thereby shifting rents between countries. The aerospace industry is often suggested as strategic in a theoretical sense in that it generates "excess rents" and subsidies can be used for "rent-snatching" (e.g., Airbus representing Europe's efforts to achieve a share of those rents [Pomfret (1991), Tyson (1992)]). A simple example is presented in Figure 2 which displays two game matrices in normal form showing super-normal profits and losses. Assume that Lockheed is established in the world military transport aircraft market with its Hercules aircraft.

467

Ch. 16: Industrial Policies in the Defense Sector (a) No Subsidy

(b) With Subsidies to FLA

European FLA

European FLA

Produce Don't Produce 10,-10

Lockheed Hercules

Don't

Produce

25,

Produce Don't Produce Lockheed

Produce

Hercules

Don't

-10,40

125,0

1

Produce

Figure 2. Successful intervention. (a) No subsidy; (b) With subsidies to FLA. There is no incentive for the Europeans to enter the market with their rival Future Large Aircraft (FLA) without a subsidy, because entry would involve losses (-10). A commitment by the European governments to pay the subsidy changes the payoffs and makes production a dominant strategy. With a subsidy (50), the Europeans are better off producing than not producing, whatever Lockheed does, so that they will enter the market. As a result, Lockheed will lose money and will exit if it has no access to a subsidy. At the same time, the Europeans achieve rents of 175 for a subsidy of 50. However, intervention through subsidies will not always pay off and there are examples of costly failures. Indeed, public choice analysis predicts government failure, whilst Austrian economics stresses the role of uncertainty and the impossibility of making accurate predictions about the unknown and unknowable future [Mueller (1989)]. The potential losers from a free trade area will seek to oppose change and lobby for compensatory policies. As a result, there is often an emphasis on managed (fair) competition, balanced trade and transitional measures. These policies involve departures from the principles of a competitive free trade area as national defense ministries seek to "manage competition" so as to minimize the costs imposed on the losers. For example, efforts will be made to ensure that the opening-up of national defense markets occurs slowly over a number of years allowing time for adjustment. Similarly, an emphasis on "balanced" international trade in defense equipment can be used to protect national defense industries. However, such balance is likely to be achieved at a price of departing from principles of specialization by comparative advantage and seeking "balance" in a specific product group, namely, defense equipment as distinct from an overall balance across all international transactions. A free trade area is one possible solution for weapons procurement in a military alliance. However, this is not the only possibility: there are at least three alternative scenarios each of which needs to be specified more clearly. Indeed, rarely are efforts made to operationalize any of these scenarios. First, a competitive free trade area could be created whereby each member state would open up its national defense market, allowing firms from other member states to bid for its national defense contracts. In NATO, such a free trade area could be created for the whole military alliance; or, it could apply to a sub-set of member states, namely, European countries and the Single European Market. Furthermore, such a free trade area could apply to all defense

468

K. Hartley Table 3 Alternative alliance scenarios

Scenario 1 2 3

National procurement by national defense ministries Centralized procurement agency replacing national defense ministries Twin track (a) competition (b) collaboration

Liberalized competitive market: Extent of market Alliance-wide World-wide ...

...

...

...... ......

equipment or there could be limited liberalization under which certain equipments would be excluded (e.g., nuclear systems). Second, the alliance could create a single centralized procurement agency which would replace national defense ministries and would purchase common, standardized equipment for all alliance forces. Third, there is the twin track scenario comprising a mixture of competition and collaboration. Competition would apply to small and medium-sized equipment (e.g., ammunition, small arms, small-medium sized missiles) where purchasing would be undertaken by national defense ministries. Large equipment projects would be undertaken on a collaborative basis with work allocated by either juste retour or by competition (e.g., combat aircraft, large missiles, nuclear systems [Moravcsik (1990), Sandler and Hartley (1995)]). For each scenario, competition could be restricted to firms from member states of the alliance or their markets could be opened-up to firms from the rest of the world (i.e., outside the alliance), as shown in Table 3. Of the various options, the creation of a centralized procurement agency purchasing common equipment offers the greatest potential cost savings but, politically, it is the most difficult to implement. Whichever scenario is preferred, it has to be recognized that purchasing defense equipment is a complex process involving a variety of choices under uncertainty. Decisions are required on what to buy, whether to buy on a national or a collaborative basis, at which stage in the project's life cycle the equipment is to be selected, who to buy from and how to buy. In other words, choices are required on the type of equipment to be purchased, the role of competition and the extent of the market in the selection process, the choice of contractor and the type of contract. Table 4 presents a taxonomy which identifies and classifies the variety of arrangements for both purchasing and supplying defense equipment of which alliance defense industrial policies form a subset. Broadly, there are two extremes, namely, national independence whereby a nation acting alone might buy all its defense equipment from its domestic DIB; alternatively, it could import all its defense equipment by buying off-the-shelf from overseas suppliers. Between these extreme policies, there are a variety of intermediate policies such as international collaboration, the licensed or co-production of a foreign design or imports

469

Ch. 16: Industrial Policies in the Defense Sector Table 4 Procurement options Choices

Choice of buyer b (B) Joint Procurement (A) National procurement'

1. What to buy: performance features

1

2

II. When to select in project life cycle (i) Design stage (ii) Development prototype stage (iii) Production stage

3 5 7

4 6 8

9

10

11

12

13 15

14 16

17 19

18 20

IV Choice of contract (i) Fixed price (ii) Target cost - incentive type (iii) Cost-plus

21 23 25

22 24 26

V Supply-side: industrialorganization (i) Single producer (ii) Licensed coproduction (iii) Offsets (work sharing) (iv) Collaboration-consortia

27 29 31 33

28 30 32 34

35

36

37 39

38 40

III. Role of competition and extent of market (i) Competitive procurement: open versus selective (ii) Non competitive: direct negotiation (iii) Market open to a. National firms only b. Suppliers in alliance member states c. Firms in other regions d. The world

VI. Location of supplier (i) National (ii) Overseas a. Alliance b. Rest of world

One alliance nation. b Two or more alliance nations including centralized procurement. a

of foreign equipment with some form of offset or work-sharing arrangement for the domestic industry of the importing nations. Table 4 can be used to classify the three alliance defense industrial policies (as shown in Table 3). National procurement by national defense ministries in a liberalized

470

K. Hartley

competitive market would be represented by A, I 1, II 3 or 5 or 7, III 9, 15 or 19, IV 21, V 27 or 33, VI 35 or 37 or 39. This scenario involves one alliance nation buying its defense equipment on a competitive fixed price contract with competition restricted to firms in alliance member states or open to the world. Supply could be from a single producer or a consortium located within the nation or overseas. In contrast, a centralized procurement agency buying equipment for all alliance members on a competitive fixed price basis would be represented: B I 2, II 4 or 6 or 8, III 10, 16 or 20, IV 22, V 28 or 34, VI 36 or 38 or 40. Whilst the various alliance defense industrial policies appear to offer substantial cost savings, their assumptions need to be specified. Only then is it possible to assess their implications and their likelihood of success (and the reasons for departing from "first best" solutions for such regimes as managed competition). For each scenario it is assumed that: (i) The national or centralized procurement agency acts as a competitive buyer with no work sharing or juste retour constraints and the centralized procurement agency can make and enforce its equipment choices. For example, the agency would need an agreed set of rules for making procurement choices (e.g., unanimity or majority voting). (ii) Firms are awarded competitively-determined fixed price contracts for development and production work. (iii) There are clearly specified requirements for defense equipment applicable to all bidders. (iv) Suppliers receive no preferential treatment from their national governments (e.g., subsidies; indirect support such as for civil R&D with possible military applications). In other words, there needs to be non-discriminatory purchasing and a level playing field. The problems of creating a competitive free trade area in a military alliance (e.g. NATO or the Single European Market) cannot be exaggerated. Within a liberalized market with procurement by national defense ministries, problems arise in enforcing the procedures for market openness and in preventing anti-competitive behavior by national governments. Defense equipment purchases can always be rationalized on grounds of 'national security'. And whichever the scenario, there will always be pressures from producers to restrict the market to firms from alliance member states, with the potential for cartelization and lack of contestability. Indeed, the history of trade policy shows how hard it is to defeat the power of producer lobbies. There will also be pressure from the military-industrial complex favoring 'fair and managed' competition with cost based contracts awarded on a juste retour basis. Inevitably, the long-run trend towards a smaller number of larger defense contractors will lead to departures from the competitive model and the creation of oligopolies and monopolies where there are opportunities for cartels and collusive tendering. Scenarios which create and protect alliance monopolies so as to obtain economies of scale and avoid the wasteful duplication of costly R&D will allow firms to charge monopoly prices as well as pursue a quiet life leading to dynamic costs rather

Ch. 16: Industrial Policies in the Defense Sector

471

than benefits - all of which will reduce the estimated savings from achieving less duplication in R&D! Here, contestability of alliance defense markets is crucial to firm behavior and performance. Contestability can be achieved in a variety of ways through, for example, opening-up the alliance market to firms from the rest of the world; or by offering sub-contract work to the losing firm in a major competition; or by recognizing that for some equipments, it is possible that the rivals who lose a major contract might be able to re-enter the market at some future date. Or, with a centralized procurement agency, competing development teams might be maintained by covering the cost of developing cheap competing prototypes. However, where there is an alliance monopoly supplier, the centralized procurement agency might have to maintain competition by funding an overseas rival development program (i.e. in a non-member state), although this might not be politically acceptable. Moreover, unless the centralized procurement agency is prepared to award the contract to an overseas firm from outside the alliance, the alliance monopolist will not take the threat of rivalry seriously; and if the contract is awarded overseas, the alliance monopolist might exit from the industry! In such a situation, the alliance monopolist might have to be subject to regulatory requirements (e.g. profit controls) and will behave like any firm subject to regulation.

4. Taxonomy: project appraisal and procurement options Purchasing defense equipment requires nations to make difficult choices in a world of uncertainty (e.g. over time horizons of up to 40 years). Where a country is a member of an alliance, such choices might be made on an individual national basis or collectively whereby all members agree on the type of equipment to be purchased. Cost-benefit analysis provides a framework for the economic evaluation of alternative equipment. In principle, this requires an analysis of the costs of the various rival equipments, their performance in relation to the national or alliance defense requirement and any wider industrial or economic benefits. For each alternative defense equipment (e.g. rival tanks or ships or combat aircraft), a cost-benefit analysis requires the following: (i) The estimated life-cycle costs of the alternatives. Acquisition prices represent only part of life-cycle costs: these are the true total costs of a project which need to be expressed in terms of present values. Cost estimates for equipment under development and not yet in-service can also be unreliable and for such equipment, this unreliability is even greater for cost estimates for in-service support, operations and maintenance. Nor should cost figures be compared on a per unit basis only: for example, combat aircraft differ in their performance so that the relevant basis for any comparison is the total cost of thefleet required to meet an operational requirement. (ii) The militaryfeatures and strategic implications of the alternatives. This requires an analysis of the performance, numbers required to meet the operational requirement, and delivery schedules. The alternatives might differ in their contributions to alliance

472

K. Hartley

standardization and inter-operability, to international collaboration and to maintaining the national defense industrial base. iii. National economic benefits. Equipment procurement choices involve wider economic benefits in the form of their impact on jobs, technology and the balance of payments. Employment impacts involve numbers of jobs and their skill content; technology includes any spin-offs to the rest of the economy; and the balance of payments embraces exports and import-saving. In some cases, efforts are made to include Treasury or Exchequer impacts as reflected in tax receipts and any expenditure on unemployment and social welfare payments (e.g. following the cancellation of a project). However, the inclusion of Exchequer impacts is controversial. It raises issues of transfer payments versus the resource costs of the unemployed in periods of largescale unemployment; the role, if any, of shadow pricing; and the extent to which local labor markets act as efficient clearing mechanisms (see Chapter 17 of this Handbook). Not all costs and benefits will be quantifiable: there will be some intangibles which need to be brought to the attention of decision-makers (e.g. the security offered by a national defense industrial base). Even where variables can be quantified, they might be subject to risk and uncertainty, and government as the final decision-maker might attach different valuations to those recommended by economists. For example, a Defense Ministry might focus on defense criteria only, namely, costs, performance, delivery and the risks attached to the various competing proposals, so ignoring any wider industrial and economic benefits. On the other hand, when governments have to approve an equipment purchase, their final decision might be determined by wider economic benefits, such as jobs impacts in marginal constituencies and, ultimately, the desire for re-election. Whether or not it is a member of an alliance, a nation has a variety of industrial policy options for purchasing defense equipment. Each option offers the purchasing nation a different degree of industrial involvement or work sharing in the development and/or production of the system being acquired. The options also have different implications for costs, program risks, control over equipment specifications, as well as wider industrial and economic benefits for the purchasing nation [Sandler and Hartley (1995, Chapter 9)]. Typically, governments prefer the indigenous development and production of defense equipment. Domestic sourcing enables governments to claim that their policies generate new technologies (which might spill-over into civil industry), provide jobs, improve the balance of payment and the security of supply. However, the costs of an independent DIB and the lack of a suitable technological capability often restrict purchasers of defense equipment to sharing the production work associated with equipment they wish to buy, rather than undertaking indigenous development or a collaborative project, where both development and production work are shared. Whichever procurement policy is adopted, care is needed to distinguish between the economic welfare implications of international trade and the beliefs, myths and special pleading which national producer groups will use to influence equipment choices and which governments will use to 'rationalize' their decisions. A taxonomy of various work-sharing arrangements is shown in Table 5. For example, importing involves

473

Ch. 16: Industrial Policies in the Defense Sector Table 5 Work sharing Policy

National-domestic procurement (DIB) International collaboration Licensed production; coproduction Offsets direct indirect Importing

Sharing arrangements a Selected defense equipment b Other defense equipmentb D P S D P S (100%)

(100%)

(100%)

/ (S%)

V/ (S%)

I/ (S%)

Civil goods and services

V V

(S%)

(%)

x

x

/(S%) x

/(S%)

V (S%)

V

,, work share; x, no work share. h D, development; P, production; S, support work. a

c S%, sharing rate between partner nations, where S < 100%.

no work sharing on the contract, whereas purchasing from the national DIB offers 100% work on development, production and support for the program. Offsets can involve direct work sharing on the selected equipment or indirect involvement through other defense projects or civil goods and services (e.g. tourism).

5. International collaboration International collaboration or joint ventures can be viewed as clubs involving two or more nations sharing the total development and production costs and work on defense equipment projects. Europe has considerable experience of such collaboration, especially in aerospace projects (aircraft, helicopters, engines and missiles). Examples include the British-French Jaguar strike aircraft; the UK-Germany-Italy Tornado combat aircraft, and the 4-nation Eurofighter 2000 air superiority fighter aircraft [Draper (1990), Hartley (1983), Matthews (1992)]. The stylized facts of international collaboration in aerospace can be summarized as follows: (i) During the 1960s, international collaboration in aerospace was largely intraEuropean. (ii) During the 1970s, there was an increase in the proportion of collaborative projects with a non-European partner. (iii) Both the number and percentage of all projects which were collaborative was higher in the 1980s than in the 1960s. In other words, there has been a movement away from national projects towards collaboration.

474

K. Hartley Table 6 Cost savings from collaboration' R&D costs (£m)

Production order

Unit production cost (m)

Total production cost (£m)

Total program cost (Em)

Independence Nation A Nation B

1000 1000

200 200

20 20

4000 4000

5000 5000

Internationalcollaboration Nations A&B: equal sharing Savings from collaboration

1000 1000

400 -

18 2

7200 800

8200 1800

a For collaborative production a 90% unit production cost curve is used, showing a 10% reduction in

unit costs for a doubling in cumulative output.

(iv)

Between the 1960s and 1980s, there was an upward trend in the mean number of partner nations per collaborative project. (v) The larger European economies, namely, France, Germany and the UK, have been most active in collaborative projects, usually collaborating with each other. (vi) The limited population of collaborative projects has involved different partner nations, various organizational arrangements and a heterogeneous set of projects (e.g. combat and trainer aircraft, helicopters, missile and aero-engines). (vii) Collaboration is not restricted to aerospace: a wide variety of military equipment has been developed and produced collaboratively [e.g. electronics based projects such as airborne and naval radar: Hartley and Martin (1993a)]. European collaboration is believed to offer a variety of economic, industrial, military and political benefits. In theory, there are cost savings in both R&D and production. Members of the collaborative club can share the costs and risks of costly R&D programs, and by combining their national orders they can achieve the economies of scale and learning from a longer production run. In this ideal case, the partner nations obtain cheaper equipment and avoid the 'wasteful duplication' of costly R&D work associated with competing national programs. An example is shown in Table 6 where two-nation collaboration offers cost savings of 18%. There are also possible industrial benefits, especially in costly high technology industries such as aerospace where collaboration creates larger European groups able to compete with the USA. Further benefits arise from greater equipment standardization between NATO members as well as demonstrating a European commitment to the cohesion and credibility of the NATO alliance [HCP 247 (1991), Pugh (1986)]. Collaboration, however, has its costs leading to departures from the 'ideal model'. Governments are involved in a set of choices about the type of equipment to be purchased, with which partner nations, with which firms, how to organize the joint venture, and the criteria for sharing the total work on the program. These

Ch. 16: Industrial Policies in the Defense Sector

475

choices will be made in the political market place of governments, voters, political parties, bureaucracies and producer groups within the partner nations: such market arrangements provide opportunities for inefficiency in collaborative ventures. 5. 1. Collaboration:a public choice analysis The typical election arrangements mean that voters are unable to express their preferences between alternative procurement policies (e.g. nationalism, buy abroad, joint projects or licensed production); nor are they provided with information on the costs and benefits of different policies because of the alleged need to 'protect national security'. As a result, a government has discretion in interpreting the 'national interest' and it will be influenced by the views of specialists in the form of bureaucracies and interest groups of contractors and scientists. A defense ministry and the armed forces aiming to maximize their budget and the opportunities for power, prestige and personal satisfaction, will stress and exaggerate the apparently attractive cost savings from international collaboration and its major social benefits through supporting jobs, the balance of payments and promoting high technology. Bureaucrats will welcome the opportunities for regular international travel and the prestige involved in negotiating an international treaty, protecting the national interest, allocating contracts and 'policing' the program. They have an incentive to negotiate international contracts which make it costly for any partner nation to withdraw unilaterally, and to devise a set of monitoring arrangements which maximize their involvement in project decision-making (e.g. under the guise of public accountability and protecting the national taxpayer). Some of these activities represent bureaucratic behavior as 'on-the-job' leisure disguised as 'output' [Peacock (1992)]. The case for collaboration will be supported by producer groups of management and unions seeking new contracts, and by the scientific lobby with its desire to expand the frontiers of knowledge - all at the taxpayers expense. As a result, the armed forces, bureaucracies, contractors and scientists within each partner nation will insist upon imposing their requirements, ideas and technical aspirations. Bargaining is inevitable. At the start of the program, each partner nation's armed forces will insist upon their operational requirements; firms will compete for project leadership; and each country's scientists will demand to be involved in the most exciting technical advances. International rivalry will dominate the early stages of negotiations, as each national interest group submits its claims, uses its bargaining skills, exaggerates its orders and threatens to withdraw (it resembles collective bargaining between employers and trade unions). Inevitably, any international agreement will be a compromise between the maximum bids of each partner and the minimum terms required to persuade it to join and remain a member of the club. Such compromises are likely to reflect bargaining behavior rather than efficiency criteria: behavior within committees will determine outcomes rather than commercial considerations of efficiency and profitability. Each partner will demand a 'fair' share of both development and production work, including an involvement in all aspects of high technology. For example, on aircraft development,

476

K. Hartley

each partner will demand a share of high technology work on the airframe, engine, avionics and equipment. In its desire to obtain a preferred part of the project, a nation might be willing to sacrifice to its partners a major share in a less preferred part of the program, or even offer compensation in other unrelated policy areas (e.g. support in a vital vote on EU agricultural or regional policy). As a result, work is allocated on the basis of political and equity (juste retour) rather than efficiency criteria. Collaborative projects once started are difficult to stop. International collusion replaces rivalry, and national governments will be confronted by a combined interest group of scientists, producers and bureaucrats in the partner nations. Such international groups of experts are likely to be influential in persuading vote-conscious governments of the technological, military, diplomatic and other social benefits of continuing with a collaborative venture. Their views on technology and estimates of development costs and time-scales are likely to be decisive although optimistic; and the costs of optimism are borne by each nation's numerous tax-payers. As a result, budget-sensitive bureaucracies in partner nations, supported by scientists and defense contractors, will tend to exaggerate the social benefits and cost savings of joint programs in order to persuade vote-conscious governments to undertake and continue the work. European collaborative ventures will also be attractive to governments seeking reelection. Partner governments can be seen to be promoting European unity, avoiding dependence on US technology and weapons, and obtaining wider economic benefits (the military-industrial-political complex). A public choice analysis predicts that joint European ventures will depart from the ideal model and will be characterized by inefficiency. The political market place and the associated bargaining process creates constraints on the choice of the most efficient organizational arrangements and contractors. On collaborative aerospace ventures, inefficiency is reflected in: (a) The administrative and organizational costs resulting from excessive government bureaucracy and duplicate organizations for procurement and supply. For example, the most efficient organization for managing a project might require a prime contractor-supplier relationship, but such an arrangement might be unacceptable to two equal partner nations and their prime contractors. Governments are also likely to select the major contractors on political, rather than commercial, criteria (e.g. awarding work to marginal constituencies and to their national champions). (b) Work sharing. Typically, some high technology development work has to be allocated to a partner country lacking the necessary technical knowledge and expertise. For example, a nation might join the club as a means of entering a new field of high technology. (c) The duplication of R&D and production work with each partner requiring a flight testing centre and a final assembly line. (d) Each nation requiring modifications on its military projects. Compared with the ideal model, this will raise R&D costs and possibly reduce the scale economies from a long run of one type, as well as reducing standardization.

Ch. 16: Industrial Policies in the Defense Sector

477

(e) Delays due to design by committee using unanimity rules and pursuing 'overelaborate' specifications reflecting the technical aspirations of the military and scientists in each partner nation. Delays also result from frequent meetings and paperwork, the need to consult all the partner nations before a decision can be taken, and the slow decision-making process by 'over-involved' governments. Further problems arise from differences in language, culture, national pride, and managerial philosophy and practices. As a result, there are transaction costs of creating new international organizations and doing business with strangers selected on political criteria. (f) The substantialcosts of controllingjoint programs. They require a complex set of international transactions and contracts between governments and firms, embracing high technology projects which cannot be clearly-defined ex ante. The resulting contracts are costly to monitor and cancel. For example, on the Concorde project it was stated that "... the inherent difficulty of the multiplicity of contracts, the number of interested parties and the participation of two sovereign states ... meant that the project had acquired a life of its own and was out of control" [HCP 265 (1981)]. Some of the early experience with collaboration provided lessons which were applied to subsequent joint ventures. 5.2. Empirical results The various benefits, costs and inefficiencies of collaboration can be formulated into a set of testable hypotheses about the impact of collaboration on costs and development times. Evidence from an international survey shows that compared with national ventures, there was considerable support for the hypotheses that joint European programs involved [Hartley (1995)]: (i) Higher total R&D costs, varying between an extra 5% to an additional 50% with a median cost penalty of 10-25%. (ii) Lower unit production costs, with cost savings varying between 10% and 25%, with a median saving of 20%. This result was completely contrary to UK experience which suggests that on collaborative production work, the expected scale and learning economies are not achieved: unit production costs are either unchanged by collaboration or might even be higher than for a national equivalent [Hartley and Martin (1993b), HCP 247 (1991)]. (iii) Longer development times, ranging from an extra 10% to 100% and a median of an additional 20% (representing an extra 2 to 9 years in development times). The international questionnaire survey found a variety of reasons for collaboration inefficiencies. They included compromises, overlap, cross-checking and duplication, the time needed to reach agreements, work-sharing arrangements, bureaucracy, cultural differences, national rivalry, frequent specification changes, the number of partners, and the sheer complexity of some joint programs. Indeed, one respondent carefully explained that longer development periods often reflected the complexity of the program and not its joint nature. In principle, some of these hypotheses can be tested

478

K. Hartley

statistically. For example, a model of development time of the following form can be estimated:

r

= T(C, Z, N),

(1)

where T = total development time, C = project development cost, Z = a vector of characteristics of the defense equipment (e.g. aircraft speed, weight, range, etc) N = number of partner nations. Inevitably, empirical work on development time scales and other hypotheses about collaboration is limited by the lack of data. For defense equipment, there are problems in obtaining reliable data on development costs both in total and on an annual expenditure basis. Similarly, data on cost penalties are not publicly available: hence the need for interview-questionnaire surveys [Hartley (1983, 1995)]. Empirical work is further complicated because of a limited and heterogeneous population of collaborative defense projects involving different partner nations and different organizational arrangements. Finally, there are problems with the counter-factual: what would have happened in the absence of a collaborative venture? Estimation of Equation (1) provided only limited support for the hypothesis that collaborative projects take longer to develop than national ventures. For a sample of military aircraft, collaboration appeared to add an extra seven months from the first flight of the prototype to delivery for every partner nation involved in the program, with this period increasing to fourteen months for combat aircraft. However, for the total development period from start to delivery for both military aircraft and helicopters, collaboration generally had no statistically significant effect. As an exception, one equation suggested an extra thirteen months for a two-partner program [compare the questionnaire survey results: Hartley and Martin (1993b)]. How might collaboration be improved to avoid its worst features and inefficiencies? There are at least three guidelines for government policy towards improving the efficiency of international collaboration: * Rule I: allocate both development and production work on the basis of competition, with nations specializing according to their comparative advantage. * Rule II: there needs to be a prime contractor subject to contractual incentives and penalties, which place it at risk (i.e. via competitively-determined fixed price contracts). * Rule III: the principle of compensation. Adequate arrangements are needed to compensate the losers from policies designed to improve efficiency. Such compensation need not be organized within the program: it could involve offsets from other defense projects or more general assistance in the form of regional aid and manpower policies (e.g. retraining, mobility).

Ch. 16: Industrial Policies in the Defense Sector

479

6. Licensed and co-production When buying foreign defense equipment, the importing nation often demands some form of compensatory or 'offsetting' economic activity for its domestic industry. One solution is for the importing nation to undertake the domestic production of the foreign-designed equipment (see Table 5). Production sharing can be achieved through licensed or co-production. Licensed production occurs where the purchaser obtains a share of the production work for its own order and, sometimes, for any exports to third parties. The license may cover the manufacture of the entire item, or specified parts, or only final assembly. Both the UK (Westland) and Italy (Agusta) have frequently purchased licenses to produce American-designed helicopters. Japan has licensed produced a range of US combat aircraft (e.g. F4, F15) and is co-developing a version of the US F16 aircraft [Chinworth (1992)]. A similar arrangement is co-production, where the nation buying foreign equipment is given a share in the manufacturing work for its own order, the supplying nation's order and orders from third parties. For example, the General Dynamics F16 European co-production contract in 1975 was based originally on sharing the manufacture of a 998 aircraft program. The European consortium (Belgium, Denmark, the Netherlands and Norway), which ordered 348 aircraft, was originally allocated work on ten percent of the initial US order (650 aircraft), forty percent of their own order and fifteen percent of export sales to other countries. Exports were estimated at 500 units and the achievement of this figure gave the Europeans manufacturing business to the value of 80 percent of their total order, namely, 279 aircraft [Hartley (1983)]. Licensed and co-production offer a variety of military, budgetary, industrial and wider economic benefits. Purchasing existing equipment avoids the risks of failure; it provides operational equipment in an acceptable time scale; and it contributes to military standardization. Manufacturing under license also saves substantial R&D resources which would be required for an independent national venture. Furthermore, licensed and co-production provide support for a nation's DIB; they offer some technology transfer (e.g. in management and production); and contribute to employment and balance of payments objectives (via import savings). There is however, little quantitative evidence on the magnitude of these benefits: a deficiency which suggests the need for good economic case studies. For example, a study of Japanese experience with licensed production of US aircraft estimated that Japanese work content varied from 85% of the F104 to 90% of the F4 and 70% of the F15 [Chinworth (1992)]. The claimed benefits of licensed and co-production are not free gifts: there are usually cost penalties compared with buying directly off-the-shelf from the original manufacturer. These cost penalties have been estimated at up to 50%, with a typical cost penalty of 10-15% [Chinworth (1992), Hartley (1995)]. On the AmericanEuropean co-production of the F16 it was estimated that the Europeans incurred a 34% cost penalty compared with buying the aircraft directly from General Dynamics

480

K. Hartley Unit Cost C2 C1

Co I __- _

----------

ACO

I u

Q1

Q2

Q3

Output

Figure 3. Economics of licensed production.

[Rich et al. (1981)]. Such cost penalties reflect entry or start-up costs for the licensed producer (e.g. production investment in tooling), the costs of transferring technology (e.g. drawings, etc.), relatively short production runs and work starting at the top of the learning curve. An example is shown in Figure 3, where a nation undertakes the licensed production of OQl, which is equal to Q2 - Q3. Assuming that the licensed producer and the original manufacturer operate on the same unit cost curve (ACo) then licensed production of OQI results in a unit cost of C 1 compared with a direct buy which increases output of the main manufacturer from Q2 to Q3 and results in unit costs of C0 . More likely, the licensed producer will operate on a higher unit cost curve (ACI) so that the unit cost of OQI is even higher, at C2. In an interesting exception, a Rand study of Japanese experience with the licensed production of the US F104 aircraft estimated that the unit costs of the Japanese aircraft were about 10% less than the price it would have paid for US produced aircraft: a result which reflected the transfer of learning technology from Lockheed and lower Japanese labor costs [Hall and Johnson (1967)]. There are further costs of licensed production in the form of time scales and restricted access to key equipment technologies. Japanese experience with the licensed production of the US F15 aircraft provides useful case study evidence [Chinworth (1992)]. On the F15, the time scale between signing the agreement and the first delivery of Japanese-built F15 aircraft was 5.5 years. However, this time scale needs to be compared with the likely delivery schedule if the aircraft had been purchased directly from the USA. In addition, US companies used 'black boxes' to protect their key technologies and design capabilities. As a result, equipment failures involving 'black box' technologies meant that the equipment had to be returned to the USA for repair by US firms. Inevitably, repairs and servicing in the USA were time-consuming, so reducing the operational effectiveness of the Japanese-built F15s and showing the limitations of licensed production in contributing to an independent DIB. Indeed, Japanese experience with F 15 repairs and servicing, parts supplies, technology transfer

Ch. 16:

Industrial Policies in the Defense Sector

481

and pricing further contributed towards the country's desire to create an independent aircraft industry [Chinworth (1992)].

7. Offsets Governments are often unwilling to pay the (sometimes considerable) premium for licensed production. Nevertheless, governments and interest groups believe that industrial benefits are desirable and this has led to the development of offsets as a further variant within the work-sharing taxonomy. Typically, an offset occurs when the foreign supplier places work to an agreed value with firms in the buying country, over and above what it would have bought in the absence of the offset. Defining offsets is difficult because different countries have different criteria for assessing whether a particular transaction is eligible towards an offset obligation. One distinction which is sometimes made is between direct and indirect offset. Direct offsets involve goods and services for the equipment which the purchaser is buying. For example, if the UK buys an American aircraft and the landing gear for those aircraft came from a UK supplier then this would be a direct offset. Indirect offsets involve goods and services unrelated to the purchase of the specific foreign defense equipment. For example, the sale of American F18 aircraft to Spain allowed aid to Spanish tourism to count as part of the offset agreement. Indirect offsets can include foreign investment and countertrade transactions such as barter, counter-purchase and buy-back [Marin and Schnitzer (1994), Martin and Hartley (1995), Udis and Maskus (1991); see also Table 5]. In addition to the distinction between direct and indirect offsets, there are cases where the offset may be unrelated to the purchase of goods. If the industrial base in the country purchasing the equipment is relatively small and underdeveloped, there may be few goods that can be purchased and thus inward investment might constitute the major component of any offset [e.g. as in the Al Yamamah sale of military aircraft by British Aerospace to Saudi Arabia: Martin (1995)]. Offsets are a growing feature of the international trade in defense equipment, particularly aerospace equipment, and this is a field which is relatively unexplored by economists. Offsets impose conditions on the foreign seller of defense equipment enabling the purchasing government to recover or offset some or all of its purchase price [Udis and Maskus (1991)]. Offset schemes are usually designed to achieve a relocation of economic activity from the country of the equipment supplier to the purchasing nation. Such relocation resembles trade diversion and has been criticized by economists as welfare reducing. For the seller, offsets reflect the desire of profitseeking firms for 'doing business' with governments: they can be regarded as part of the sales package and as an alternative to price discounts. For the supplying nation, there are concerns about the impact of offsets on its defense industries, especially suppliers, on employment and on technology transfer to potential rivals. For the buying nation, offsets appear to provide industrial benefits in the form of jobs, technology transfer, support for the defense industrial base and foreign currency savings [Hall and

482

K. Hartley

Markowski (1994), OMB (1987)]. But appearances can be deceptive. Many of these alleged benefits are attractive to vote-sensitive governments and such claims need to be subject to careful and critical evaluation. There might be alternative and cheaper methods of achieving these objectives. Offsets need not necessarily be inefficient and welfare-reducing. In some circumstances, offsets might contribute to efficiency improvements if they remove non-tariff barriers and lead prime contractors to search for, and to discover, more efficient subcontractors located overseas (i.e. by extending market information and knowledge, and removing entry barriers to the use of foreign sub-contractors). Alternatively, defense offsets can be viewed as a sub-set of the price-quality-quantity trade-offs which characterize negotiations surrounding complex transactions. In a world of imperfect markets, oligopoly rents, complex transactions and asymmetrical information, offsets might enhance the welfare of the purchaser (e.g. a single contract with an offset might economize on transaction costs compared with a contract for defense equipment and a variety of separate contracts for the 'benefits' supplied under the offset). On the other hand, some forms of mandatory offset obligations may inhibit the buyer's flexibility in negotiating advantageous deals and so result in inefficient procurement [Hall and Markowski (1994)]. Clearly, in competitive bidding for foreign defense contracts, overseas firms have every incentive to offer an attractive offset package as part of their bid: maximizing offsets becomes part of the competitive process, with firms seeking new and ingenious methods of satisfying their contractual obligations. They might, for example, use specialist agencies, such as banks, as well as their suppliers to achieve their offset targets; and they will try to claim as much business as possible as offset. At the same time, vote-sensitive governments have every inducement to claim the maximum size of offset deal so that they can justify the import of defense equipment in terms of protecting the national defense industry, jobs and technology. There are, however, worries about the real benefits of offsets. Reservations have been expressed about the following issues [Martin and Hartley (1995)]: (i) New work. There are concerns about the extent to which offset business represents genuinely new work which would not otherwise have been obtained without the offset agreement. For example, on the UK purchase of American Boeing AWACS aircraft, the UK negotiated an impressive 130% offset agreement (i.e. offsets to 130% of the value of the UK order). However, critics have suggested that, typically, genuinely new business might be 25-50% of the total offset. (ii) Civil work. On the UK purchase of AWACS, Boeing was allowed to count its purchases of Rolls-Royce civil aero-engines for its commercial aircraft as part of the offset obligation. This was criticized for allowing civil aerospace work to count against a defense offset commitment; and because the arrangement was generous to Boeing, representing more than 50% of Boeing's total offset commitment. In fact, Rolls-Royce aero-engines would have been purchased irrespective of the offset agreement: hence they were not new work resulting from the offset obligation [HCP 286 (1989)].

Ch. 16: Industrial Policies in the Defense Sector

483

(iii) High technology work. Whilst offsets can appear to offer substantial work shares, it is unlikely that they will provide much high technology work. This is because offsets are for the purchase of existing defense equipment where all the major development work has been completed. Moreover, the original manufacturer will wish to maintain its property rights in valuable technologies.

8. The military production function: internal efficiency Equipment forms a major input into the military production function. The result of combining equipment with military personnel and other inputs (e.g. land) is the production of military outputs in the form of national protection, security and peace. The policies used to purchase defense equipment have major implications for its unit costs and hence affordability from a fixed defense budget. Purchasing from a high cost national DIB means less equipment for the nation's armed forces raising questions about whether policy is aiming to protect the DIB or the nation's citizens. Once equipment is acquired, military commanders have to determine its most efficient use. In theory, this is a simple optimization problem which in the private sector would be 'solved' by profit conscious firms in competitive markets seeking to minimize costs. Such firms would have an incentive to substitute cheaper for more expensive factor inputs which would be reflected in substitutions between different equipments and between equipment and personnel (e.g. quality versus quantity of defense equipment; air systems replacing land forces). However, incentives to make such substitutions are usually absent in military 'firms': the lack of competition and the nature of employment contracts for military personnel means that there are no strong pressures and incentives for individuals and groups to minimize costs [Sandler and Hartley (1995, p. 157)]. On the contrary, each service will seek to protect its budget and maintain its traditional monopoly property rights (i.e. over air, land and sea), so reducing the opportunities for efficient substitution between the armed forces. Efficient substitution is also unlikely within each of the armed forces. Commanders of army, navy and air forces bases will be concerned with promotion opportunities which are usually related to military rather than economic performance. Indeed, military commanders seeking power and prestige will aim to maximize the size of their bases. As a result, they have every incentive to 'hoard' equipment and personnel, as well as accommodation and real estate for training purposes. Bases located in valuable city center sites or close to national capitals are especially attractive. The continued retention of such bases will be rationalized and justified on military criteria, such as the need for valuable training areas or proximity to the national defense ministry: in reality, bases in attractive locations confer non-military benefits (e.g. proximity to cultural and leisure facilities in a nation's capital; the opportunities for using training areas for leisure pursuits). It is feasible to construct institutional and budgetary arrangements which would provide military commanders and their units with efficiency incentives. Within the

484

K. Hartley

army, navy and air force, various military bases could be required to bid competitively for the provision of contractually agreed services for a fixed budget. For example, air force bases could compete for providing an agreed number of combat aircraft patrols per annum; and army units could compete to offer combat ready infantry units available for rapid deployment anywhere in the world. Military commanders and their units would have every incentive to economize on factor inputs. Once they have been allocated defense equipment (e.g. combat aircraft; tanks, etc.), they would have incentives to operate the equipment efficiently. They might substitute civilians for military personnel to maintain, service and repair the equipment: for example, repair and maintenance work could be 'contractorized' and undertaken by private firms (e.g. civilian garages repairing military vehicles). Similarly, there would be incentives to economize on spares acquisition and to substitute training simulators where these are cheaper than 'real' training which uses expensive equipment and large acres of ground. The aim of the scheme would be to provide military commanders with opportunities and incentives for entrepreneurship.

9. Conversion Policies to improve the efficiency of procurement and the internal efficiency of the armed forces will mean a reduced demand for resources in the military-industrial sector. Some defense firms will lose from these changes with the inevitable job losses and plant closures. Military establishments will also be affected as the armed forces declare personnel redundant and close their bases. Where defense plants and military bases are major employers in a town or region, their closure has significant adverse impacts on the prosperity of local economies [see Chapter 17 of this Handbook and EC (1992)]. Faced with plant and base closures and associated job losses, all societies and different types of economies have to consider how best to transfer resources from the military to the civilian sector. Two broad solutions are possible. First, efforts can be made to utilize existing defense establishments and their workforce to produce civil goods and services rather than defense equipment. This is the classic direct conversion of swords to ploughshares, tanks to tractors, aircraft carriers to oil tankers and bomber aircraft to civil airliners. Second, the closure of defense plants and bases might be accepted, with the focus on re-allocating the resources released to other civil firms, industries and regions in the economy. Both solutions raise questions about the appropriate role and extent of state intervention in the adjustment process. For direct conversion, state intervention might operate through a state conversion or diversification agency. For assisting resource re-allocation, state intervention might take the form of general manpower policies (e.g. retraining; mobility; information and careers advice).

Ch. 16: Industrial Policies in the Defense Sector

485

9.1. Direct conversion Direct conversion appears attractive. If successful, it utilizes existing and valuable human and physical capital so appearing to avoid the costs, upheaval and 'wastes' associated with plant closure and redundancies. However, such direct conversion often fails to occur simply because of the costs required to exit defense markets and to enter new civilian markets. Some of these costs reflect the demand- and supplyside characteristics of defense contractors. On the demand side, some defense firms are highly dependent on military sales to a single government customer, operating in a protected market with support for 'national champions' resulting in a 'cosy relationship' between contractors and the Defense Ministry [a culture of dependency: Melman (1971)]. On the supply side, a defense firm's assets might be highly specific to defense, with little if any alternative use value. Examples include some defense technologies such as stealth and armor and the specialist defense plants needed to construct nuclear-powered submarines and inter-continental ballistic missiles [Sandler and Hartley (1995, p. 293), Dumas (1977, 1986)]. Further understanding of the problems of direct conversion can be obtained by using a taxonomy which focuses on the following economic characteristics for such conversion: (i) The type of economy, namely, market, centrally-planned or transitional economy with a further distinction between developed and developing nations [UN (1993)1. (ii) The ownership of defense plants, either publicly or privately owned [Ott and Hartley (1991)]. (iii) The distinction between defense plants and military bases. In market economies, military bases usually have to change ownership to enter civilian markets (e.g. military air bases can be converted to civil airports, although there are examples of the joint uses of such bases). (iv) The extent to which a defense company is involved in one or all of the following activities: R&D, production, repair and maintenance. Direct conversion at each of these stages could involve different problems reflecting differences in demand requirements and asset specificity. For example, military R&D is characterized by secrecy and highly specialized military requirements with an emphasis on performance rather than costs: such features are less marketable in the civil sector [Nadal (1994), UN (1993)]. In market economies, the direct conversion of specialist defense plants is not costless and instantaneous. There will be costs involved in re-equipping defense plants for civil production and in retraining managers, scientists and workers. It will also take time to change the culture of an organization from defense dependency to an enterprise culture: typically, an adjustment period of up to five years might be needed. Further costs will arise in searching for, and entering, new civil markets; and crucially, there is a need to discover civil markets which are expected to be profitable after bearing the costs of conversion and entry. Not surprisingly, faced with such adjustment costs,

486

K Hartley

some specialist defense firms have decided to seek new military markets at home and overseas, to lay-off labor or close plants, or to diversify by acquiring new civil business. Adjustment will differ between the short and long run. In the short run, a defense firm's adjustment will be constrained by fixities and contractual commitments, so that variable labor without long-term contracts is most likely to bear any immediate adjustment costs. In the long run, all factors can be varied and the most efficient adjustments can be made [Hooper and Hartley (1993)]. Conversion in some of the transitional economies such as Russia is occurring at the same time as the economy is changing from a centrally planned to a market economy. One of these tasks is difficult enough: to undertake both simultaneously represents a major challenge [UN (1993), Wiseman (1991)]. Conversion in Russia and the Ukraine has some further distinctive features. First, the magnitude of the adjustment problem. In 1992, employment in the military-industrial complex in the Russian Federation totalled some 5.4 million people equivalent to 7.4% of total employment, with examples of regions where over 40% of employment was defenserelated and some cities which were 'closed communities' wholly dependent on defense work [e.g. Zhukovsky, a city of 100000 people, was the center of Russian aviation R&D: OECD (1995)]. The Ukraine was confronted with a similar massive conversion challenge. In the past, 700 industrial enterprises accounting for 18% of industrial production and 1.2 million workers were dependent on defense orders; by 1994, there were no orders! Second, Russian experience distinguishes between bureaucratic and technocratic conversion. The bureaucratic type involved top officials determining the output of civil goods using the old management systems and directives to produce TV sets, fridges, radios, cameras and washing machines, regardless of production costs. Technocratic conversion involved restructuring from below with the focus on purely technological criteria rather than the needs of the economy or on market research. Technocratic conversion was more successful than the bureaucratic type but it postponed 'greater conversion'. At the same time, the public was deceived as its political leaders, in attempting to preserve the existing structure, passed off small changes as major steps towards demilitarization. Conversion decisions were made by engineers rather than businessmen and economic planners and as a result many large defense production plants made poor choices and took on unrealistic or wasteful programs [Bougrov (1994)]. 9.2. Re-allocating resources and public policy Re-allocating resources from the military-industrial complex to the civil sector involves adjustment costs and takes time [UN (1993)]. Much depends on the transferability of human and physical resources and how well and how quickly economies can adjust to change. For example, some of the skills of redundant military personnel and redundant defense workers might be highly specific with value only to the defense sector (e.g. missile operatives; nuclear weapons scientists). Other military skills might be general with value to large numbers of civil firms (e.g. air

Ch. 16: Industrial Policies in the Defense Sector

487

traffic controllers; computer skills; drivers; electronic and electrical repair specialists). Ultimately, though, the operation of the labor market is a major determinant of the efficiency of the adjustment process in any economy. How well and how quickly do labor markets work and clear? Similar questions need to be asked about the operation of other factor markets (capital, land, raw materials, energy). State intervention might then be rationalized on grounds of market failure, although public-choice analysis recognizes the possibility of government failure! One further issue has to be addressed, namely, is defense 'different' so that it qualifies for 'special' public policies favoring defense firms and redundant defense workers? Those who claim that defense is 'different and special' point to the importance, role and responsibility of government in determining the size and location of the military-industrial complex; the need to compensate the potential losers from change so as to minimize the barriers to change; and the fact that disarmament contributes to peace. Critics of this view claim that defense is not different and should be treated like any other declining industry in the civil sector; and that defense workers have been rewarded for the risks and uncertainties of employment in the industry. This is not to suggest that general economy-wide compensation policies should not be available for assisting industrial, employment and regional adjustment (e.g. via regional and manpower policies and income-deficiency payments). However, the question of whether defense is 'different and special' is likely to be the focus of myths, emotion and special pleading by interest groups likely to lose from policies aimed at improving efficiency in the military-industrial complex. Economists have to subject such special pleading to economic analysis, empirical evidence and critical evaluation. They need to identify the economic logic of any policy proposals for favorable assistance to defense industries, distinguishing between efficiency and equity issues. 10. Conclusion: a research agenda There is no shortage of research questions. There are extensive opportunities for applying public choice models (e.g. to analyze collaboration and the military production function). Much more reliable quantitative evidence is needed on the costs and benefits of different procurement policies. What are the cost and time penalties of collaboration, licensed and co-production; how valuable is equipment standardization and inter-operability; should wider industrial and economic benefits enter into equipment procurement choices; and what is the 'best' way of organizing and managing collaborative projects? What are the lessons from successful and failed conversions of defense plants and military bases in both market and transitional economies? Contrary to the prevailing fashion in economics, answers to many of these questions require the use of good quality case studies. References Arrow, K.J., 1962, Economic welfare and the allocation of resources for invention, in: D.Lamberton, ed., 1971, The economics of information and knowledge (Penguin, London).

488

K. Hartley

Beard, R., 1993, NATO armaments co-operation in the 1990s, NATO Review 41(2) (NATO, Brussels) 23-28. Bougrov, E., 1994, Conversion in transitional economies: The case of the former USSR and Russia, Defence and Peace Economics 5, 153-166. Callaghan Jr, T.A., 1975, US-European economic co-operation in military and civil technology (Center for Strategic and International Studies, Georgetown University, Washington, DC, September). CBO (US Congressional Budget Office), 1994, The cost of the Administrations plans for the Army, Navy and Air Force through the year 2010 (Congressional Budget Office, US Congress, Washington, DC, November). Chinworth, M., 1992, Inside Japan's defense (Brassey's, Washington, DC). Demsetz, H., 1969, Information and efficiency: another viewpoint, Journal of Law and Economics 12, 1-22. Draper, A., 1990, European defence equipment collaboration (RUSI/Macmillan, London). Dumas, L.J., 1977, Economic conversion, productive efficiency and social welfare, Journal of Sociology and Social Welfare 4, 567-596. Dumas, L.J., 1986, The overburdened economy (University of California Press, London). EC, 1992, The economic and social impact of reductions in defense spending and military forces on the regions of the Community (EC DGXVI, Brussels). Hall, G.R., and R.E. Johnson, 1967, Aircraft production and procurement strategy (Rand, Santa Monica, CA). Hall, P., and S. Markowski, 1994, On the normality and abnormality of offset obligations, Defence and Peace Economics 5, 173 188. Hartley, K., 1983, NATO arms co-operation (Allen & Unwin, London). Hartley, K., 1995, The analysis of the costs of non-Europe, in: K. Hartley and A. Cox, eds., The costs of non-Europe in defence procurement (Main Report, Part B: European Commission, Brussels). Hartley, K., and S. Martin, 1993a, The political economy of international collaboration, in: R. Coopey, M. Uttley and G. Spinardi, eds., Defence science and technology (Harwood, Reading). Hartley, K., and S. Martin, 1993b, Evaluating collaborative programmes, Defence Economics 4, 195211. HCP 247, 1991, Ministry of Defence: Collaborative projects (National Audit Office, HMSO, London). HCP 265, 1981, Concorde (Second report from the Industry and Trade Committee, HMSO, London). HCP 286, 1989, The working of the AWACS offset agreement (Defence Committee, House of Commons, HMSO, London). Hooper, N., and K. Hartley, 1993, UK defence contractors adjusting to change, centre for defence economics research monograph series, Vol. 3 (University of York, England). Jane's, 1994, Jane's all the world's aircraft 1994-1995 (Jane's Information Group, London). Kirkpatrick, D.L.I., and P. Pugh, 1983, Towards the Starship Enterprise: Are the current trends in defence unit costs inexorable? Aerospace, May. Marin, D., and M. Schnitzer, 1994, Tying trade flows: A theory of countertrade, Discussion Paper 946 (Centre for Economic Policy Research, London). Martin, S., 1995, The economics of offsets: Issues, policies and prospects (Harwood, Reading). Martin, S., and K. Hartley, 1995, UK firms experience and perceptions of defence offsets: survey results, Defence and Peace Economics 6, forthcoming. Matthews, R., 1992, European armament collaboration (Harwood, Reading). Melman, S., 1971, ed., The war economy of the United States: Readings on military industry and economy (St. Martins Press, New York). Moravcsik, A., 1990, Defence co-operation: The European armaments industry at the crossroads, Survival XXXII, 65-85. Mueller, D.C., 1989, Public choice II (Cambridge University Press, Cambridge). Nadal, A.E., 1994, Military R&D: The economic implications of disarmament and conversion, Defence and Peace Economics 5, 131 151.

Ch. 16: Industrial Policies in the Defense Sector

489

OECD, 1995, The OECD observer 192, February, March (OECD, Paris) 39-43. OMB (US Office of Management and Budget), 1987, Impact of offsets in defense-related exports (Office of Management and Budget, Washington, DC). Ott, A., and K. Hartley, eds., 1991, Privatization and economic efficiency (Elgar, Aldershot). Peacock, A.T., 1992, Public choice analysis in historical perspective (Cambridge University Press, Cambridge). Pomfret, R., 1991, The new trade theories, rent-snatching and jet aircraft, The World Economy 14, 269-277. Pugh, P., 1986, The cost of sea power (Conway, London). Pugh, P., 1993, The procurement nexus, Defence Economics 4, 179-194. Rich, M., W. Stanley, J. Birkler and M.I. Hesse, 1981, Multi-national co-production of military aerospace systems (Rand, Santa Monica, CA). Sandler, T., and K. Hartley, 1995, The economics of defense (Cambridge University Press, Cambridge). Tyson, L., 1992, Industrial policy and trade management in the commercial aircraft industry, in: Who's Bashing Whom? Trade Conflict in High Technology Industries (Institute for International Economics, Washington, DC). Udis, B., and K.E. Maskus, 1991, Offsets as industrial policy: lessons from aerospace, Defence Economics 2, 151-164. UN (United Nations), 1993, The economic aspects of disarmament: Disarmament as an investment process (United Nations, New York). Wiseman, J., 1991, Privatization in the command economy, in: A. Ott and K. Hartley, eds., Privatization and economic efficiency (Elgar, Aldershot).

Chapter 17

THE REGIONAL IMPACT OF DEFENSE EXPENDITURE DEREK BRADDON University of the West of England

Contents Abstract Keywords 1. Introduction 2. Data problems 3. Theoretical approaches for evaluating regional impacts 3.1. Economic base models and regional multipliers 3.2. Regional defense expenditure and local labor markets 3.3. Input-output analysis 3.4. Regional econometric models

4. The regional distribution of defense expenditure 5. Selected case studies 6. Policy implications 7. Conclusions References

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T Sandler © 1995 Elsevier Science B. V All rights reserved

492 492 493 494 495 495 498 499 502

506 513 515 518 519

492

D. Braddon

Abstract This chapter examines the current state of knowledge regarding the economic impact of defense expenditure at the level of the regional economy from a global perspective. Appropriate theoretical and analytical approaches are identified and selected case studies reviewed. The regional concentration of defense expenditure is explored using a variety of indicators and perceived differences in the economic impact of defense industry decline and military base contraction and closure are evaluated. The chapter considers relevant labor market issues and examines the different impacts to be expected in developing countries. Policy implications are explored in the context of the post-Cold War environment confronting the defense sector.

Keywords regional multiplier, regional models, supplier network, military bases, labor market, adaptive capacity, regional policy

Ch. 17: Regional Impact of Defense Expenditure

493

1. Introduction For many regions of the world, particularly in the advanced industrial nations, economic prosperity has depended historically upon defense-related industrial production and/or military base-related activities, especially in periods when the adverse effects of national and international economic recession would have otherwise eroded the buoyancy of the local economy. In the 1990's, however, the vulnerability of many defense-dependent regional economies to deep and sustained reductions in defense expenditure has become a prominent economic and political issue. Sharply declining national defense budgets, impacting upon almost all significant military goods producers and military infrastructures, have posed a serious threat to short-term regional prosperity and, in some cases, even to long-term regional and sub-regional economic viability. While similar or more extreme reductions in defense budgets have been experienced over the last 50 years (e.g. after World War II, the Korean War and Vietnam), on those occasions, the continual presence and rapid escalation of the Cold War super-power arms race made a cyclical resurgence in defense budgets inevitable [Grawe (1991)]. Few analysts now anticipate such a cyclical recovery in defense expenditure in the future. Consequently, the effects of permanently lower levels of defense spending on the regional industrial and technological base and on the local communities, companies and workers that it supports is likely to be profound. Furthermore, the revision of military strategies following the end of the Cold War has initiated a new phase of military base closures and realignments by most of the major powers with significant regional consequences. In particular, base closures by both the USA [US General Accounting Office (1991)] and the former Soviet Union will have the greatest impact within Europe [Sharp (1990)]. The nations within the European Union most vulnerable to US military withdrawal are Germany, the UK and Italy. The case of Germany best illustrates the regional economic significance for a nation acting as host to large numbers of foreign troops stationed within its territory. During the 1980's, some 400000 NATO troops and their families were stationed in West Germany, generating significant expenditure of DM 21 billion (mainly in specific regions) with direct and indirect employment effects of some 230 000 jobs. Similarly, in 1988, some 425 000 Soviet troops were stationed in East Germany. However, the regional and national income and employment effects in this case were negligible as these troops were strictly isolated from local communities and almost all civilian posts were filled by Soviet citizens. In total, estimates suggest that the impact on demand levels from the Soviet forces amounted to no more than 4.5 billion marks in 1989 or about 1% of industrial output [Maneval (1993)]. The scale of regional impact associated with the withdrawal of such foreign troops can be extremely significant, depending upon the specific economic condition of the host region at the time of withdrawal. Current estimates suggest that, together with planned manpower reductions in the Bundeswehr, NATO and Soviet troop withdrawals have combined to reduce the number of soldiers and their families stationed within

494

D. Braddon

Germany by some 850 000 over a period of just four years. Such rapid manpower reductions and associated multiplier consequences can generate serious regional consequences as, for example, in the Mainz region where, in the 1980's, some 5700 people were employed in a US Army repair facility. By 1994, over half of these jobs had been lost and recent developments suggest that the remainder may be at risk [Lock and Vob (1994)]. Taking an international perspective, this chapter first identifies and evaluates relevant analytical techniques employed to assess the regional impact of defense expenditure, reviewing evidence from a number of important studies, and then explores the regional distribution of that expenditure globally. Similarities and differences between the regional economic and labor market impacts associated with the defense industry and its supply chain and those associated with military bases and establishments are considered, along with particular issues confronting developing countries. The chapter concludes with an assessment of policy implications at the regional level.

2. Data problems The precise geographical incidence of defense expenditure (defined to include spending on military personnel and bases; defense equipment supplied by the entire defense industrial base; and other relevant spending: e.g. works, maintenance, etc.) is extremely difficult to estimate and compare empirically for several reasons. First, the quantity, quality and level of disaggregation of both national and regional defense expenditure varies greatly in availability, accuracy and comparability. Secondly, government statistics relating to the distribution of defense procurement expenditure usually indicate only the region (or city) in which recipient firms have headquarters or principal locations, neglecting the regional impact of second and third stage procurement through sub-contracting and supply chains. This problem is exacerbated by the fact that major defense companies tend to operate in more than one region and in more than one industry, while frequently themselves being just one division of a national or international conglomerate engaged in many other non-defense activities. Data problems in this area of analysis are complicated further by the national security requirements surrounding defense activities (particularly constraining information concerning industrial enterprise and military base location) and by differing views as to precisely which industries are contained within the "defense" sector. Furthermore, both final demand and intermediate products in the defense sector tend to be highly specific and do not always fit neatly into standard industrial categories. Accurate evaluation of the industrial and regional impact of defense expenditure, then, requires a particularly detailed level of data disaggregation to capture fully the real effects. A further problem relates to the selection of appropriate indicators to best reflect the economic significance of defense activity at the regional level. These could include actual or estimated numbers in defense employment within a region; the share of total regional employment emanating from the regional defense sector; the value of regional

Ch. 17: Regional Impact of Defense Expenditure

495

output or income generated by regional defense activities; the share of total regional output or income emanating from regional defense business; the value of defense prime contracts awarded to a region, adjusted for supply chain leakages and so on. No single indicator, however, is likely to be suitable for the international comparison of regional defense-dependence. Even within a single nation, different indicators may be relevant since the regional economic significance of a particular kind of defense-dependence will be a function of that region's unique dimensions and characteristics as well as the economic condition and policy approach of the nation of which it is a part.

3. Theoretical approaches for evaluating regional impacts The assessment of the regional impact of defense expenditure has conventionally been undertaken through the application of economic base models, regional multiplier estimates, the regional application of input-output analysis and through the development of more complex econometric models. Defense expenditure received by one specific region is essentially income which will, in turn, encourage other economic activities and generate additional income flows in the local economy. Regional impacts will comprise the direct effects of prime contracts awarded to companies in the region (or direct expenditure on military establishments located there); the indirect effects from resulting supply and sub-contracting business generated; and the induced effects from consequent changes in regional consumption and related investment activity. In practice, the actual impact of variations in defense expenditure within a region (and between regions) will depend upon what form the expenditure change takes and the economic structure and industrial inter-relationships of the region concerned. 3.1. Economic base models and regional multipliers These models seek to identify what proportion of regional output or employment is dependent upon external and internal markets. That proportion of regional economic activity driven by external markets is seen as the basic and essential requirement for regional prosperity. Non-basic economic activity with an entirely local focus is seen as being dependent upon this external stimulus. Consequently, the capacity of a specific region to maintain and expand prosperity and employment will depend upon the basic/non-basic activity ratio which can be utilized, in turn, to derive a simple regional multiplier. Separating the regional economy into its two components: Et = Eb+E,

(1)

where Et is the total regional income or employment; En, non-basic sector income or employment within region; Eb, basic sector income or employment within region.

D. Braddon

496 With z as the ratio of non-basic to total employment, this implies that:

(2)

En = zE,.

If demand now increases for basic goods such as defense production in a region, the resulting change in basic employment will lead to a change in non-basic and total employment. Substituting Equation (1) into (2) gives: En = z(Eb +En)

and :

En=

Eb,

(3,4)

from which the multiplier can be derived as l/(1-z). The value of economic base multipliers, however, in evaluating the regional impact of exogenous economic change is constrained by two important problems. First, localized (non-basic) economic activity may be affected by factors other than total regional income, implying that the proportionality assumption incorporated above cannot be sustained. However, assuming time-series or cross-section data exists for total regional income or employment (Et) and for income or employment generated by the basic sector (Eb), regional multiplier estimates can be obtained by regressing Et on Eb. Second, the basic sector of a regional economy is likely to comprise a set of individual industries with significantly different import propensities and labor force socioeconomic characteristics. As a result, regional economic consequences emanating from changes in export demand will depend upon the specific industry or industries affected, a problem which can be overcome by calculating a set of basic industry-specific regional multipliers. The regional economic impact of military base contraction and closure will depend on several factors. These include the nature and size of a base scheduled for contraction or closure and the relevant closure timetable; the density of bases and the proportion closing within a specific region; base proximity to rural and urban communities and their capacity for economic regeneration; and the potential for retaining part of the operational functions of a base after closure in the form of a "contractorized enclave". It will also include employment of local civilians on the base; demand for housing for service personnel and their dependents; demand by military and civilian staff for local services and utilities; demand for local businesses to service the base and the business rates and taxes payable to local authorities [Braddon et al. (1991)]. In terms of local economic impact, the incomes and expenditure of military base personnel and their families should be particularly significant since salaries in the armed forces tend to be on a par with or considerably higher than for equivalent employees elsewhere in the public sector. However, military bases tend to be relatively closed economies with many goods and services provided and consumed internally, thereby diminishing the economic stimulus on the local or regional community by generating multiplier consequences significantly lower than for a comparable loss of civilian jobs. The regional multiplier effects of military base operations can also be affected by the age profile of employees. Where a base contains a large number of

497

Ch. 17: Regional Impact of Defense Expenditure Table 1 Multiplier estimates for defense-related regional activities Study

Year

Location

Activity

Weiss and Gooding

1968

USA

Demilitarized

1991

IFO ADCC

1991 1992

UK Germany Netherlands Germany UK UK

Aztec

1992

naval shipyard air base manufacturing defense industry defense industry defense industry defense industry shipyard defense electronics USAF air base aircraft factory

UK UK

Employment multiplier 1.55 1.35 1.78 0.83 1.0-2.0 1.33 2.1 1.96 1.27 1.15 1.5

younger service personnel, the marginal propensity to consume locally (and therefore

the multiplier effect) is likely to be above average. Regional multipliers in defense-dependent regions can also be derived by modifying the Kahn/Keynes employment-income multiplier process. At the regional level, withdrawals from the flow of income are likely to be significant through high import propensities, depending on the degree to which the regional defense industry supply chain extends beyond regional borders. Table 1 shows the range of regional multiplier estimates for defense expenditure generated by a number of studies. The differences in multiplier estimates observed in Table 1 can be attributed to the unique characteristics of each region, its location and the precise form defense expenditure takes. Where a region is closely inter-linked with other regions, the multiplier effect is likely to be reduced. Where defense companies or military bases have a high degree of self-sufficiency, the multiplier effect will similarly diminish. Multipliers estimated for defense activities in rural areas will differ from those in urban areas where the effects of more advanced infrastructure, high-technology industry, sophisticated services provision and a more flexible labor market will be likely to exert a significant effect. However, recent research focusing upon the supply chain of a major defense prime contractor in the highly defense-dependent South West region of the UK [Braddon et al. (1992)] suggested that conventional approaches to the evaluation of multiplier impacts may fail to capture adequately the fundamental changes taking place within the defense industry supply chain as it adjusts to the new business conditions of the 1990's. The research confirmed that the defense industry regional supply matrix includes a much wider range of business activity than is commonly supposed, implying significantly

498

D. Braddon

greater regional economic impact from defense expenditure variations than might normally have been expected. Again, the extent of regional inter-dependencies in the defense supply industry appears much greater than previously imagined. Companies were identified in different tiers of the network supplying not only one or more defense prime contractors directly but also other companies located in different tiers of the network (including their own) and, through them, indirectly supplying other prime contractors. Examples were found of companies within the region acting as first tier suppliers for one customer; second tier suppliers for another; and, simultaneously, as suppliers to several other sub-contractors and other suppliers in the same or separate product groups. In many cases, the companies themselves were unaware of the final destination of their output, exacerbating the problem of identifying and measuring both the content and extent of the regional defense industrial base and the regional impact associated with variations in the level and content of defense procurement expenditure. At the same time, however, the intensification of competitive pressures has fractured established regional buyer-seller relationships in the defense industry with prime contractors seeking to maintain their competitive edge in world markets by retaining only the most efficient and flexible suppliers, drawn from an increasingly wide geographic area. In general, most studies of conventional regional multipliers in the defense sector have derived estimates for defense employment of between one and two, indicating that for each direct job in defense, up to one additional job will be indirectly supported by related economic activity elsewhere in the regional economy, excluding induced multiplier effects. Conflicting evidence questions the employment-creating capacity of defense expenditure compared with other forms of government expenditure [Hartley and Hooper (1993), Smith (1991)]. 3.2. Regional defense expenditure and local labor markets At a time when, almost universally, defense expenditure is declining in arms-producing nations, the capacity of a particular region to absorb adverse impacts upon employment and income will depend largely on the flexibility and responsiveness of the local and national labor markets and the potential for new injections of alternative economic activity at the regional level. Initial labor market impacts may include job-shedding as part of the planned rationalization of prime defense contractors and of small and medium sized enterprises in the defense sector; together with unplanned and sudden job-shedding by companies located in the lower tiers and levels of the defense industry supply chain, unaware of the degree to which their business is defense-dependent. Further job-shedding in nondefense sectors of the economy will then occur as the regional multiplier consequences of reduced demand impact upon local business. Where defense industrial companies or military bases actually close, the labor market impact is unlikely to be restricted to a neat set of geographical boundaries but will extend across local government areas with variable severity, depending

Ch. 17: Regional Impact of Defense Expenditure

499

on the geographic distribution of defense suppliers, military bases and associated establishments and their suppliers. Some regions which have high concentrations of both defense supply companies and military bases are likely to experience regional "clustering" effects and, in the absence of compensating employment gains in other sectors following a sustained reduction in defense expenditure, their economic base could be severely eroded. The capacity of the regional labor market to respond to reductions in defense spending will be affected significantly by the mobility of displaced labor and the evolving skill profile required in any regeneration of the regional economy. There is likely to be a significant difference here between the regional impact of defense expenditure reductions which affect employment in defense industry and that in military bases and establishments. While the extent and duration of both forms of unemployment will depend upon the absorptive capacity of the local economy and the flexibility of the local labor market, redundant personnel in the armed forces are likely to be more geographically mobile, tending to move away from their place of discharge (perhaps returning to their home town, seeking new employment). With the exception of some highly-qualified defense scientists and technicians, however, unemployed defense industry workers are likely to remain in the same location, seeking alternative employment while effectively swamping the local labor market with similar skill and experience profiles. The issue of labor mobility, therefore, adds a further dimension to the variation in regional multiplier estimates in the defense sector. 3.3. Input-output analysis Another prominent type of empirical model employed to estimate the regional economic impact of a change in defense expenditure has been input-output analysis [Richardson (1985), Miller and Blair (1985)]. Devised initially by Leontieff (1951) with important developments introduced by Isard (1960) and Isard and Langford (1971), input-output analysis is particularly appropriate to the study of the effects of demand-driven changes in the regional economy since it focuses specifically upon the inter-dependencies that bind together separate sectors of the regional economy. Such inter-relationships have been categorized as backward and forward linkages with the former relating to input provision and the latter to output utilization [Hirschman (1958)]. Backward linkages in one particular sector are induced by additional final demand for the output of that sector and, while the effects may spread widely throughout the economy, the original demand increase is said to emanate from the sector of origin. On the other hand, linkage effects could be calculated with respect to the sector of destination. Key sectors of the economy will exhibit varying degrees of linkage with, by definition, primary production having virtually no backward linkage but significant forward linkage, while manufacturing offers the possibility of significant backward linkages. In general, the nearer a sector is to production for final demand, the greater the scope for backward linkage. Within a region's industrial and consumer sector, these linkages depend upon the structure and scope of the input and component

D. Braddon

500

chain stretching from regional suppliers to their prime contractor customers and other consumers. Three variants of the input-output approach are appropriate for economic impact analysis and, suitably adapted, may be employed for regional evaluation and estimation. The open comparative static input-output model employs standard input-output tables, supplemented with average labor coefficients for relevant industries, to determine the employment effects of defense expenditure. Closed static models can be used to estimate the multiplier consequences of defense expenditure by incorporating a private consumption sector in the model. Finally, dynamic models supplement inputoutput analysis with econometric equations, designed to estimate demand, both factor and final, with the additional attribute that they can be extended to incorporate the important effects of changing technology. Overall, closed static models will generate higher values than open static models, through the incorporation of full multiplier consequences. The results estimated by static and by dynamic models will differ since the former employ average coefficients while the latter employ marginal coefficients and incorporate more extensive feedback effects. The input-output approach begins with the assumption that the output of any good or service in an economy is either utilized for final consumption by households, government or exports or is employed as an input in the production of goods and services within that economy (including itself). For any regional economy, then, each output can be represented by a production function equation where output equals final consumption plus the sum of all its uses as inputs in producing the output activity of that region. To avoid excessive complexity, input-output models are usually constructed on the basis of three key assumptions. Each industry is assumed to produce a single homogenous product or two or more fixed-proportion joint products. Furthermore, each industry is assumed to employ a fixed combination of relevant factors of production or inputs in producing its output. Constant returns to scale are assumed to apply throughout the model. To produce each unit of the jth product, then, will take alj amount of the first required input, a2j of the second input and, in an n industry economy, a of the nth input. The term a denotes the input coefficient. For this n industry economy, input coefficients are conventionally arranged in a matrix A = aij as indicated below: I

II

III

...

N

I

all

a12

a 13

...

al

II III

a21 a3

a2 2 a3 2

a2 3 a3 3

... ...

a2n a3n

N

anl

an2

an3

...

ann

In this matrix, the amount required from each input industry to produce a unit of output of another industry (or of itself) is indicated by column data. In addition to an n industry formulation, an input-output model may contain an open sector (often

501

Ch. 17: Regional Impact of Defense Expenditure

the household sector) which generates a final demand exogenously and provides labor inputs not supplied by the n industries. This element of non-input demand means that the sum of elements in each column of the input matrix A must now be less than one in order to satisfy this additional demand. Thus:

Z

ai

v-

Cd

)

0

i

)

-u

>lc

a

-

E .

)C 4-1 0

0_ 0.

',

Ia

.o

0~ Ca

~0 0

v

C)

0

.0

U4

0. 0. c V0

C

C

(d Ob~ a e0l Q

o C) v a 0 0

0

C

-

a

a Ca Co c

50 0

. 0

3_

. 00

-pp - -

o

Co 0

E

C)

s

C

B

C0 Ca2

c

-Co 0

00u

Ca

b

_ P.

~

+0

M'9!U

00 D0. CW

.0o

0

C o)

0 0

Mr- L Cs

C)

~a

a

8 00B

a

m F0

o

*0 -C)

.0

u

CD

-$4

o

o

V)

0 .

."q

C-

o\ OC

0

3 _

I

0 ~o 03

CD u 0 =S

0 cn

Ch. 18: Economics of Arms Trade

527

trade. This is unfortunate because economists have much to offer in understanding the causes and consequences of arms transfers. A large body of international trade theory remains untapped. Even during the Cold War, non-economists cited the importance, albeit secondary, of economic motivations for arms transfers [e.g., Stanley and Pearton (1972, ch. 6); Pierre (1982, p. 24); Ferrari et al. (1987, pp. 63-66)]. With the end of the Cold War there has been an increase in the relative importance of economic causes and consequences of arms transfers. This chapter is about the economics of arms trade. Our aim is to review the half dozen or so major topics in arms trade research, but with a hefty bias toward economic aspects of the trade. Despite our emphasis on the economics of arms trade, our intention is not to discount geopolitics; it is crucial for understanding arms transfers. The relative research gaps in the arms trade literature, however, lie more in economics than geopolitics. In Section 2 we describe the major trends in the international arms market over the past twenty years. The primary economic impacts of the end of the Cold War on the arms market are woven throughout our data analysis in Section 2. In Section 3 we explore the causes of arms trade, primarily in the context of international trade theory. In Sections 4 and 5 we move to empirical and theoretical studies of selected geopolitical topics. The relationship between arms trade and conflict is explored in Section 4. In Section 5 we cover the relationship between arms trade and foreign policy leverage. Concluding comments are offered in Section 6.

2. The structure of the international arms market The decline of the Cold War led to a precipitous drop in the world arms trade. Figure 1 shows the trend in arms deliveries in constant 1993 dollars over the 1983-1993 period. The real volume of weapons trade reached a peak in 1984, fell for two years, rose in 1987, then declined by over two-thirds up to 1993. World arms sales agreements also declined substantially from a peak of $69.8 billion in 1985 to $38.0 billion in 1993 in nominal terms [US ACDA (1993/94)]. From 1992 to 1993, however, arms sales agreements rose in nominal terms from $28.7 billion to $38.0 billion [US ACDA (1993/94)]. Since arms sales agreements reflect probable future arms sales deliveries, the recent increases may signify that the volume of world arms trade has bottomed out and may start increasing. The major source of the increase in arms export agreements from 1992 to 1993 was the United States, whose nominal export agreements more than doubled from $13.5 billion in 1992 to $28.3 billion in 1993. 2.1. Arms imports The decline in the volume of world arms trade over the last ten years is reflected in regional breakdowns. Arms imports fell in both the developed and developing countries. Real developing country arms imports fell from a peak of $60.8 billion

528

C.H. Anderton

80 70' 60 50 billions of 199340 dollars 30 20 10 0

\r,

Cam cX

'r co

LO co

(D co

rc

CO co

0 c

0 0)

N 0)

0

M 0)

Year Figure 1. World arms trade, 1983-1993 [US ACDA (1993/94)].

35 30 25

-

Middle East

}---- Africa

Billions of 199320 dollars 15 10 5 0 CM

t

LO

cD

0

0

0

0

r

0

C

0

0

_

0

0

0

0Y

0 0)

-

East Asia

-

South Asia

Cl o 0 0

Year 12 10

-

-

--

Billions of 1993 6 dollars 2

-

o CM ' U (CD rC CD CD CD CD CD X 0 0 0 0 0 0)

c CD 0)

0 0) 0)

l

-

0) 0)

0 0

co 0 0

A

Eastern Europe North America

--

4

Western Europe

Central America & Carribien South America

Year Figure 2. Arms imports by selected region, 1983 1993 [US ACDA (1993/94)].

in 1984 to $17.2 billion in 1993 while real developed country imports fell from a peak of $16.4 billion in 1986 to $4.8 billion in 1993 [US ACDA (1993/94)]. Figure 2 shows real arms imports by selected region over the 1983-1993 period. From 1990 to 1993, dramatic reductions occurred in the weapons imports of all regions.

Ch. 18: Economics of Arms Trade

529

Table 2 Arms import share by selected regional, 1973, 1983, 1993 (in percent) Region

1973

1983

1993

Middle East Europe, All Western Europe Eastern Europe East Asia South Asia Africa North-America South-America Central America & Caribbean

28.4 26.0 31.1 3.4 3.6 -

40.4 20.0 9.7 10.2 9.6 4.7 14.2 3.0 3.7 3.4

42.6 20.8 15.9 4.9 12.4 2.1 1.3 7.4 1.0 0.7

a

-

Source: US ACDA (1984, 1993/94).

In addition to dramatic declines in the arms imports of regions, there have been significant changes in the distribution of arms imports. Table 2 shows regional arms import shares, comparing 1973, 1983, and 1993. The Middle East has dramatically increased its import share from 28.4% in 1973 to 42.6% in 1993. Europe is the second largest region in import share even though it has declined from 26.0% in 1973 to 20.8% in 1993. East Asia had a decline in import share from 31.1% in 1973 to 9.6% in 1983 (due, primarily, to the end of Vietnam hostilities). Since 1983, East Asia's import share has risen to 12.4%. Africa has significantly decreased its share of weapons imports since 1983, from 14.2% to 1.3% in 1993. In 1971, South Vietnam, West Germany, Egypt, Iran and North Vietnam had the highest arms import shares, primarily reflecting the geopolitical rivalry between the USA and USSR. By 1981, the top four importers were from the Middle East: Iraq, Libya, Saudi Arabia and Syria. By 1991, Saudi Arabia held the dominant share of the world's arms imports, followed by Afghanistan, the USA and Iran. It is likely that Iraq would have been the second largest arms importer in 1991, but for the Gulf War embargo. 2.2. Arms exports The dramatic reductions in arms imports, of course, are reflected in the arms export data. Real developed country arms exports fell from a peak of $66.7 billion in 1987 to $20.2 billion in 1993 while real developing country arms exports fell from a peak of $10.5 billion in 1984 to $1.8 billion in 1993 [US ACDA (1993/94)]. Developed countries continued to be the major exporters of weapons, providing more than 90% of world arms exports in 1983 and 1993.

C

CH. Anderton

530 0

30 25

=

Eastern Europe

Billions of 1993 20 dollars 15

[]

Western Europe

10

'

North America

5 0

-----CO' o) o -

T Co C) -

n Co C -

( W a

rco ) -

Co Co a) -

a) CO C

0 ) o

-

_ C) a> -

cJ 0) 0)

c')

*,

East Asia - Middle East

0

Year Figure 3. Arms exports by selected region, 1983-1993 [US ACDA (1993/94)]. Table 3 Arms export share by selected regional, 1973, 1983, 1993 (in percent) Region

1973

1983

1993

Middle East Europe, All Western Europe Eastern Europe East Asia South Asia Africa North America South America

0.2 59.6

0.7 70.5 23.5 46.9 5.2 0.6 0.2 22.6 0.4

1.0 45.1 32.2 12.8 5.0 0.0 0.6 47.9 0.3

a

1.9 0.0 0.0

Source: US ACDA (1984, 1993/94).

Figure 3 shows real arms exports by selected region over the 1983-1993 period. Beginning in the late 1980s, significant reductions took place in the arms exports of all regions. Arms export market share by major region is shown in Table 3. Europe and North America dominated the arms export market in 1983 and 1993, accounting for greater than 90% of export market share. The market share of Eastern Europe declined dramatically between 1983 and 1993, due primarily to the breakup of the Soviet bloc. The market shares of the four major arms exporters from 1983-1993 are shown in Figure 4. The decline in market share in the USSR/CIS is dramatic. Many believe that the CIS's loss of market share is due to the relatively poor performance of Soviet weapons in the Gulf War and buyer perceptions that the CIS cannot provide adequate after-sales services for weapons given its recent political turmoil. In 1993, two developing countries were among the top ten weapons exporters: China and Israel [US ACDA (1993/94)]. With roughly $950 million in arms exports in 1993 (4.3% of market share), China has emerged as one of the top five or six weapons

531

Ch. 18: Economics of Arms Trade I30

g 30 ..

United States

-

40

Soviet Union/CIS

---

20

-

10

0---

lO 0) o a)

'I 03 a)

03 a)

(D 03 a)

N. co a)

o co am

0

, a

o)

a)

CO 0) 01

a() au

United Kingdom France

0C 0 a

Year Figure 4. Arms export market share, 1983-1993 [US ACDA (1993/94)].

exporters in the world. Although Israel is in the top ten of weapons exporters, its market share was less than one percent. Another perspective on the arms trade is to consider the types of weapons transferred. Table 4 shows the supplies of major weapons to developing countries broken down by major weapons category and selected suppliers over the 1991-1993 period. The Soviet Union/Russia and the United States dominated the delivery of major weapons systems to developing countries over this period. France had a substantial presence in land armaments and missile exports to developing countries. China was a major deliverer of land armaments, especially artillery, and supersonic aircraft. 2.3. Structural change: end of the Cold War We can gain insight into the structural changes emerging in the international arms market by considering the four major arms trade submarkets identified by Catrina (1988, p. 42): industrial supplier and industrial recipient (submarket 1), industrial supplier and developing recipient (submarket 2), developing supplier and developing recipient (submarket 3) and developing supplier and industrial recipient (submarket 4). According to Catrina, the highest volume of arms trade occurs in submarket 2, followed by submarkets 1, 3 and 4. The highest level of arms trade sophistication (i.e., technology) occurs in submarket 1 followed by submarkets 2, 4 and 3. Catrina's volume of trade rankings are borne out by SIPRI data for the 19881992 period [see SIPRI Yearbook (1993, p. 444)]. His descriptive analysis of the levels of weapons technology occurring in the four submarkets is persuasive. Submarket I is dominated by highly sophisticated arms transfers among industrialized countries which are often allies. The second submarket generally varies widely in technological sophistication, but some highly sophisticated weapons pull up the average, e.g., F-14s, F-15s, F-16s, MIG-23s, MIG-25s, MIG-27s, MIG-29s, Mirage 2000s, Tornados and AWACs [Catrina (1988, p. 43)]. The fourth submarket (ranked third in sophistication) is a recent phenomenon in the arms market. Catrina (1988, p. 44) provides two examples of somewhat sophisticated weapons transfers from

CH. Anderton

532

Table 4 Number of arms delivered, cumulative 1991-1993, by selected supplier, all developing country recipients, and major weapon typea China

USSR/ Russia

USA

France

UK

920 610 1200

470 175 713

60 380 30

70 80

380 1300 20

Naval craft Major surface vessels Other surface vessels Submarines Missile attack boats

2 6 4 -

10 -

2 12 -

4 12 -

4 12

Aircraft Combat - supersonic

60

208

10

20

130

-

51

-

20

-

Helicopters Other aircraft

110 40

90 39

60

10 30

60

Missiles Surface-to-air

880

986

1310

330

100

Weapon type

Land armaments Tanks Artillery Armored personnel carriers and armored cars

Combat - subsonic

a

2

Source: US ACDA (1993/94).

developing to industrial countries: Brazil's T-27 trainer aircraft sale to the United Kingdom and Israel's lease of Kfir aircraft to the USA. The third submarket (ranked fourth in sophistication) has long been characterized by retransferred weapons of relatively low sophistication. More recently, indigenous production and export of fairly sophisticated arms from developing countries to other developing countries has emerged as an important, though hardly dominant feature of the overall arms market. For example, Brazil exported no weapons until 1975 [US ACDA (1971-1980, p. 83)]. Brazil has emerged as one of the top ten exporters of major conventional weapons to developing countries. Over the 1988-1992 period, Brazil was the ninth leading major conventional weapons exporter to the developing world [SIPRI Yearbook (1993, p. 444)]. Developing nations received 99% of Brazil's major conventional exports over that period. The end of the Cold War implies far more than a relative shift in the weights attached to economic and political motivations for arms transfers. The nature of the international arms market has fundamentally changed along a number of dimensions that will reshape the arms trade in the four submarkets. The dramatic reduction of superpower competition in various regions has led to the virtual elimination of arms aid by the

Ch. 18: Economics of Arms Trade

533

USA and Russia [Johnson (1994, p. 111)]. Recipients must now pay full price for their weapons systems and they have become choosier consumers. Developing countries cannot afford the most expensive weapons systems, but instead have increased their demand for upgrades. Industrial nations are demanding the best that money can buy -not only technologically sophisticated weaponry and upgrades but also guarantees of effective after-sales service, training, and access to reliable spare parts. It is likely that the average sophistication in each of the four arms trade submarkets will rise as arms producers and governments compete to bolster sales in a leaner world market. The reduction of superpower competition has also led to lower military budgets in many regions of the world. Cuts in military procurement tend to increase the unit costs of weapons production. Many nations look to weapons exports to increase production runs, lower unit costs and maintain defense industrial bases. But all nations cannot be successful in increasing weapons exports in a declining market. Competitive pressures will force some nations to cancel indigenous weapons programs and rely more on imports. Some nations will be forced to specialize in a narrower range of weapons exports (e.g., Israel in aircraft and missile defense, Germany and Sweden in naval systems). A few nations may establish niches in high technology systems for customers that most Western nations would not supply (e.g., Russia, China, North Korea to customers like Iran, Syria, Libya, Iraq). The competitive international arms market will hurt the ability of France, the United Kingdom and the United States to sell a broad range of weapons systems. It is likely that only the United States and possibly a revitalized Russia will be able to maintain a defense industrial base over the full range of advanced weapons systems. One response of defense firms facing more stringent competitive pressures is to merge or form alliances with other defense firms, domestic and foreign [see SIPRI Yearbook (1993, p. 434 and p. 441) for US arms industry mergers and joint ventures in 1992 and Sk6ns and Wulf (1994, p. 51) for international projects of the Russian military aerospace industry in 1992 and 1993]. Global partnerships among defense firms allows them to share the expense of research and development of new weapons systems. For sophisticated weapons, the research and development costs are often too much for one firm to bear alone. Mergers and alliances also enhance the ability of the defense firms to market weapons systems in their partners' countries. The post-Cold War international environment is also testing old alliance patterns between nations, leading to the emergence of new alliances, formal and informal. The changing world order is characterized by a greater presence of multilateral peacekeeping and humanitarian aid operations. The use of coalition forces in places like Iraq, Somalia and former Yugoslavia imply further development of inter-operable forces between nations. A partnership between defense firms in various countries can be better positioned to meet this new demand than a single defense firm. New security-sharing arrangements based on multilateral approaches, when combined with declining defense budgets and political pressure from defense workers and firms to maintain production, leads to further demands for co-developed and co-produced weapons systems and

534

C.H. Anderton

various payments restrictions (e.g., offsets, barter, debt reduction) rather than outright imports. The new arms trade environment emerging from the remnants of the Cold War also has important implications for arms trade restraint. During the Cold War, the Coordinating Committee for Multilateral Export Controls (COCOM) was formed by the United Sates and other Western nations to keep defense-related products and computer technology from the Eastern Bloc nations. Today, Western nations are encouraging economic development of their former adversaries and COCOM has been dissolved. This will lead to sophisticated technology transfers from West to East (and some from East to West) which could ultimately improve the defense industrial bases and sophistication of weapons exports from Eastern nations. Russia may be relatively uninterested in arms export restraint in light of its dwindling market share and its suffering defense industrial base. Although the end of the Cold War led to an initial optimism regarding arms export restraint, there are potent economic and political forces creating skepticism about future prospects for restraining the arms trade [Moodie (1994)]. While the new market will be characterized by a lower quantity of arms sales, the quality of the arms transfers will continue an upward trend which was empirically observed by Baugh and Squires (1983a). Carus (1994) considers the possibility that revolutionary changes in the technology of war may lead to a fundamental transformation of the international arms trade, perhaps as significant (or more so), than the structural shock associated with the end of the Cold War. The new technology of war could be more information-intensive than previously imagined, relying on advanced sensors, communications, data processing, long-range precision guided weapons and advanced simulation techniques. If this view is correct, dual-use equipment could become central to military capability. While specialized defense hardware would remain an important component of military capability, dual-use equipment could become a dominant feature of the arms trade. As a result, it would become more difficult to track and control the trade in militaryrelated equipment. Even the traditional arms trade data generated by US ACDA and SIPRI might have to be re-thought in order to more fully incorporate the transfer of information systems and technology [Carus (1994, p. 174)].

3. Causes of arms trade Bajusz and Louscher (1988, pp. 13-15) identify the following causes of arms trade: maintenance of indigenous defense industries, economies of scale, R&D cost savings, defense industry employment, increased trade balance, access to hard currency, foreign policy influence over recipients, maintenance of local power balances, promotion of internal security of recipients, stronger allies, and inter-operability of weapons between allies. While it is fairly easy to list causes of the arms trade, it is much more difficult to develop theoretical and empirical frameworks which discern the relative importance of causal variables in an economic and political context. There has been relatively little

535

Ch. 18: Economics of Arms Trade

systematic cumulative theoretical and empirical work on the causes and consequences of arms trade. Indeed, some scholars believe that such work may not even be possible: "The subject [of arms transfers] is quite probably of a complexity well beyond the level where one might contemplate tight causal models, and explicitly set forth formal relationships between sets of dependent and independent variables." [Neuman and Harkavy (1979, p. 315)]. While the Neuman and Harkavy position might be true for some segments of the arms trade, it is unlikely to be true generally. Our failure to date to develop explicit sets of formal relationships between dependent and independent variables in the arms trade is due primarily to a lack of trying. For example, when considering the economic causes of the arms trade, a large body of international trade theory remains untapped. This is unfortunate given the variety of models applicable to the arms trade and the insights which can be gleaned from applications of international trade theory. 3.1. Economic causes Traditional international trade theory through the late 1970s was a "powerful monolithic structure in which all issues were analyzed using variants of a single model." [Krugman (1989, p. 1214)]. The new literature of the last fifteen years has broken the grip of a single international trade model by emphasizing economies of scale and imperfect competition. Increasingly, international trade is a field where many models are taught and research is an eclectic mix of theories. 3.1.1. Supply and demand model Assume that conditions are met for a perfectly competitive international arms market between two countries: Rongovia and Mendac [see Alexander et al. (1981, p. 23)]. The market in linear form is: Rongovia: Mendac:

Wr = ao - alPr, Wm = co - clP ,

Ws = bo + blPr, W m = do + dlP m ,

(1)

where Wr, W, WTm, W m are the domestic demand and supply of weapons with respective superscripts for Rongovia and Mendac, pr and pm are the domestic prices of weapons in Rongovia and Mendac, and the other terms are parameters where ao, co, bl,dl >0, al,cl > 0, and bo , do 0. Assume that Rongovia's autarky price is lower than Mendac's. The export supply and import demand schedules are respectively: ES = W - W = (bo - ao) + (b + al)P,

(2)

ID = W - W

(3)

= (co - do) - (ci + dl)P,

536

C.H. Anderton

where P is the world price of weapons. The equilibrium free trade world price, P*, and weapons trade volume, W*, are: (co - do - bo + ao) (ai +bl +cl +dl)'

(4)

W* = (bo - ao) +(bl +al)P*.

(5)

p*

The arms trade volume is determined by the variables which define the domestic demands and supplies of the two countries. Domestic demand for a country is defined by: (1) the strength and stability of its perceived external and internal threats; (2) plans for military action; (3) national income; (4) supply of foreign exchange (if importing); and (5) supply of substitutes for countering perceived threats or carrying out military action, e.g., the strength and reliability of allies [Alexander et al. (1981, pp. 7-8)]. Domestic supply for a country is defined by: (1) price of resources, (2) technology, and (3) price of alternative goods which could be produced. Even the simple supply and demand model challenges conventional wisdom in the arms trade literature. For example, some arms trade scholars take a condemnatory "merchants of death" view of the arms trade [e.g., Engelbrecht and Hanighen (1934)]. These scholars implicitly or explicitly assume a positive relationship between the volume of arms trade and the volume of arms in the international system. Since a nation's international arms transactions and indigenous arms production are jointly determined economic activities, a positive relationship between the volume of arms trade and volume of arms in the international system does not necessarily hold. For example, an increase in Rongovia's domestic demand reduces the volume of arms trade but increases indigenous weapons production in each country leading to more weapons in the system. Alexander et al. (1981, pp. 12-13) find empirical evidence of this for the United States during the primary Vietnam buildup years 1966-1968: as US domestic demand for weapons rose, US weapons production increased while its weapons exports decreased. Obversely, a fall in Rongovia's domestic demand increases the volume of arms trade but lowers indigenous weapons production in each country, leading to less weapons in the system. 3.1.2. Neoclassical trade model Figure 5 illustrates a free trade equilibrium between Rongovia and Mendac where (Pw/Pf), terms-of-trade price of weapons relative to food; (Pw/Pf)R, (Pw/Pf)M, autarky relative prices in Rongovia and Mendac; R*, M*, autarky points of Rongovia and Mendac; Rp, Mp, production points of Rongovia and Mendac with trade; Rc, Mc, consumption points of Rongovia and Mendac with trade. Rongovia has a comparative advantage in weapons and Mendac in food. Rongovia exports A-B weapons and imports C-B food; Mendac is the other side of the trade. The volume of weapons trade and the relative price of weapons are determined by the three key variables of the neoclassical model: tastes, production technology,

Ch. 18: Economics of Arms Trade

537 -

o...Ioft rlr I

t t

Food

Weapons Rongovia

Weapons Mendac

Figure 5. Neoclassical trade model.

and resource endowments. Tastes depend on perceived external and internal threats, reliability and strength of allies, and plans for military action. In the Heckscher-Ohlin model, production technology and tastes are assumed to be identical between nations. Differences in factor endowments between nations explains the volume and terms of trade. For example, if weapons production is capital-intensive and Rongovia is capital abundant, the Heckscher-Ohlin theorem predicts that Rongovia will export weapons. The Stolper-Samuelson theorem relates international trade and real factor rewards. According to the Stolper-Samuelson theorem, the factor used intensively in the rising price industry has an increase in real rewards. The opening of trade between Rongovia and Mendac causes the relative price of weapons to rise in Rongovia and fall in Mendac. Assuming that weapons are capital-intensive, real returns to capital owners rise in Rongovia and fall in Mendac. A short-run variation of the neoclassical model, the specific factors model, assumes that one or more factors are specific to an industry while one or more other factors are mobile between industries within each country. For example, Rongovia might have capital specific to weapons production, capital specific to food production, and labor which is mobile between food and weapons. Starting from a free trade equilibrium in Rongovia, a shift in tastes away from weapons would cause the real rewards of weapons capital to decline and the real rewards of food capital to increase. There is an ambiguous effect on the real returns to labor. Figure 5 holds constant the other countries that influence the tastes of Rongovia and Mendac (e.g., enemies and allies). These countries could be built into an extended general equilibrium framework where arms race and alliance impacts are endogenous. For example, Jones (1988) examines the impact of alliance formation on the arms trade. In Figure 6 assume that Rongovia and Mendac are identical in every respect, have homothetic preferences, and are "small", i.e., price takers on the world market. (Pw/Pf)0 is the world relative price of weapons and PPFR and PPFM are the production

538

CH.Anderton Food

weapons

Weapons Mendac

Rongovia

Figure 6. Trade creation impact of alliance formation.

possibility frontiers of Rongovia and Mendac, respectively. Prior to alliance formation the countries produce and consume at R*, M* under free trade. Since the world price is the same as the autarky prices of both countries in this particular example, there is no trade. This special case allows Jones to isolate the pure impacts of alliance formation on trade in the small country case. When Rongovia and Mendac form an alliance the spill-over effects shift the consumption possibilities for both countries out to CPFR and CPFM. Since world price is given, production remains at R* and M*, but excess supplies of defense goods emerge in each country. Rongovia exports R*A of defense goods and imports AB in food from the rest of the world (i.e., nonallies). A similar opportunity exists for Mendac (not shown). Rongovia's consumption at Rc is made up of OFR of food, FRB of domestically produced weapons which were not exported, and BRc of implicit provision of weapons spilling over from the alliance with Mendac. The alliance has created weapons exports as well as trade in food. If Rongovia and Mendac were "large", either individually or jointly, the new weapons exports resulting from their alliance would lower the relative world price of weapons. World production of weapons would fall with the lower price. Many other variations of the Jones model are possible [see Jones (1988, pp. 133-138)]. The neoclassical model highlights the general equilibrium nature of arms trade. The arms trade is co-determined with weapons and non-weapons production and trade and factor markets in both countries. The interdependence of these markets are significantly influenced by alliances and arms races. The neoclassical model predicts that a country will trade with other countries that are different from itself in factor endowments, production technology or tastes. For example, from 1981-1991 the United States exported $3.6 billion of arms to Latin America while Latin America exported essentially zero weapons to the United States [US ACDA (1991/92, p. 138)]. Focusing on purely economic motivations, the Heckscher-Ohlin theory offers a possible explanation of the "North-South" trade in weapons between the USA and Latin America: the capital abundant US exports

Ch. 18: Economics of Arms Trade

539

capital-intensive weapons to labor abundant Latin America in return for labor-intensive civilian goods. The Heckscher-Ohlin prediction of US-Latin weapons trade seems plausible, but even if it received empirical support it would not be general because much of the weapons trade occurs between countries which are similar in relative factor endowments, production technologies and tastes. For example, from 1987-1991 the United States exported $2.8 billion in weapons to West Germany and imported $1.1 billion in weapons from West Germany [US ACDA (1991/92, p. 132, 134)]. This is intra-industry trade, i.e., two-way trade of differentiated products within a broad product group. The Intra-Industry Trade Index (IIT) can be used to gauge country i's intra-industry trade in arms: aims- I

(6

IExportsans - Importsrs Exports/ + mportrm s

.

If IIT = 0, then intra-industry trade is not present. If IIT = 1, then a nation's exports and imports of weapons are equal implying the highest degree of intraindustry trade. Some intra-industry trade indexes for selected countries based on 19871991 US ACDA data are as follows: United States=0.30, United Kingdom=0.25, Brazil = 0.66, Columbia = 0, Greece = 0.05, Switzerland= 0.76. The indices imply that intra-industry weapons trade is prevalent in the world economy. 3.1.3. Trade models with economies of scale and learning economies Arms trade scholars believe that some weapons production processes are characterized by economies of scale. Given fixed input prices, economies of scale lead to lower unit costs when output levels are increased. This is shown by the average cost (AC) curve in Figure 7. Figure 7 shows why economies of scale can be such a powerful motivation for weapons exports. Let Dd represent a perfectly inelastic domestic demand curve for $ unit costs L

d

D

ac1= P1 pC

i

ac 2

_., -

-'

AC I-

f 1

f2

fighters

Figure 7. Economics of scale motivation for weapons exports.

540

C.H. Anderton

fighter aircraft. Pi is the "controlled price" set by the government buyer of fighters. The weapons producer is capturing revenues (PI f l) just equal to total costs (total costs include normal profits). If the weapons producer increases its production to f2 it will reap lower unit costs due to economies of scale (acl to ac 2 ). This will improve the cost competitiveness of the weapons producer, enhancing its ability to export weapons. In the figure, weapons exports equal f2 -fl. If PI is the price of fighters on the world market and the government lowers the price it pays for domestic needs to P, both the government and the weapons producer gain from weapons exports. The government's "defense burden" for fighters falls from P flI to PI fl. The weapons producer receives "above normal" profits equal to (PI f2) - (a' 'f2). Using a cross-sectional pooled times series analysis, Snider (1987) found that arms exports significantly lowered the defense burdens of Great Britain, France and West Germany. The average yearly procurement savings from arms exports for each country ranged from $844 million to about $1 billion. See also Kolodziej (1987) for a discussion of the perceived procurement benefits to industrialized and developing countries from arms exports. If economies of scale were fully exploited in the global economy, each country would specialize in producing a limited range of weapons systems. For example, Germany might specialize in naval systems and Israel in aircraft and missile defense. If each country produces only a subset of the weapons systems, then weapons are produced more efficiently than if each country tried to produce everything for itself. The result would be greater production of each weapon system in the world economy. The specialized economies would then trade with each other to consume the full range of weapons. Hartley (1987, pp. 297-298) estimated that NATO weapons standardization policies, which led to economies of scale and gains from trade, likely reduced unit costs by at least 20%. Further savings were available through the elimination of duplicate R&D and from economies in life-cycle costs [Hartley (1987, p. 298)]. Of course, weapons are not the same as civilian goods. For security or domestic political reasons, most nations restrict weapons exports or produce a wider subset of weapons than suggested by a free trade economies of scale model. For example, the USA accepted scale economy sacrifices to maintain production sources for its Sidewinder, Maverick, and Sparrow missiles and for Los Angeles and Trident submarines [Scherer and Ross (1990, p. 529)]. This allowed the USA to have a substantial and diverse industrial base in these weapons systems as well as a capacity to maintain competitive bidding over the weapons. Despite many examples of unspecialization in weapons production throughout the world, economies of scale provides a strong motivation for specialization and trade in weapons. Learning economies (also known as dynamic increasing returns or dynamic scale economies) are present when unit costs fall with cumulative production over time, rather than with the current rate of production [Krugman and Obstfeld (1994, p. 144)]. If a nation has extensive experience in an industry, it may have lower unit costs than an inexperienced nation even if the inexperienced country's learning curve is

Ch. 18:

Economics of Arms Trade

541

lower because of lower input prices. Dynamic scale economies can thus "lock in" an initial advantage or head start in an industry [Krugman and Obstfeld (1994, p. 144)]. Weapons exports can be an effective mechanism for a nation to accumulate experience in weapons production, garnering lower unit costs through learning effects. The lower unit costs, as with economies of scale, imply the two-fold benefit to the weapons exporter: (1) lower defense burden for the domestic country's weapons acquisition and (2) lower unit costs which further enhances weapons export competitiveness. 3.1.4. Imperfect competition and trade Economies of scale, learning economies and intra-industry trade of differentiated weapons generally imply imperfectly competitive market structures: monopolistic competition, oligopoly or monopoly. In imperfectly competitive markets for differentiated weapons systems, a nation that produces a wide variety of weapons will not reap the scale advantages that could be gained from larger production runs over a narrower range of weapons. On the other hand, a nation that pursues large production runs over a narrow range of weapons gains the scale advantages but has a limited variety of weapons. Weapons trading improves the scale/variety tradeoff for nations. Krugman and Obstfeld (1994) develop a monopolistic competition model of trade that is applicable to the international arms market. Assume that there are n symmetric weapons firms, i.e., the demand and cost functions are identical for all firms, even though they are producing and selling somewhat differentiated products. The average cost of each firm depends on the size of the market and the number of firms in the industry: AC= (n. F) +c,

(7)

where AC, average cost of weapons production; F, fixed cost; S, total weapons sales of the industry (in physical units); c, firm's marginal cost. Equation 7 implies that average costs are higher the more firms there are in the industry (since with more firms, each produces less). The upward-sloping relationship between n and the average cost is the CC schedule in Figure 8. Krugman and Obstfeld also derive a relationship between the number of firms in the industry and the price charged by each firm: P'c+

6nl

(8)

where P, price charged by each firm; b, parameter measuring the sensitivity of each firm's market share to the price it charges. Equation 8 implies that the more firms there are in the industry, the lower the price each firm will charge. The downward-sloping relationship between P and n in Figure 8 is the PP curve.

C.H. Anderton

542 Price, P

_,_.

-

firms, n Figure 8. Monopolistic competition model of weapons trade.

Figure 8 is a monopolistically competitive model of production, consumption and trade. The equilibrium number of firms is n* and the equilibrium price charged by each firm is P*. P*=AC* reflects the long-run zero-profit equilibrium. Trades in differentiated weapons (not shown) are the differences between each firm's total weapons sales and its weapons sales to its home country. The integrated market of Figure 8 supports more firms, each producing at a larger scale and selling at a lower price, than national markets alone [Krugman and Obstfeld (1994, p. 128)]. The end of the Cold War led to a decline in the quantity of weapons demanded in the world market. This led to decreases in total weapons sales S (see Figure 1) that shifts the CC curve to the left in Figure 8. The "shakeout" in the world defense industry causes n to fall from n*to n2 . The surviving firms each produce smaller weapons quantities at higher prices, even though overall demands for weapons have fallen. The new long-run equilibrium of P=AC occurs at higher price P2 because lower production runs raise average costs. In the high technology weapons markets (e.g., fighters, missiles, submarines), smallgroup oligopoly is more applicable than large-group monopolistic competition. With oligopoly comes the potential for collusion and strategic behavior. Firms can raise profits through explicit or tacit collusion. Co-production or co-development weapons agreements are often driven by political considerations, but they also may be motivated by gains from collusion. Strategic behavior, i.e., actions that affect the behavior of competitors in desirable ways, might lead firms to accept lower profits to deter entry of a rival or governments to subsidize its weapons manufacturers to shift profits and unit cost competitiveness from foreign to domestic firms. Figure 9 provides a typology of the arms trade in terms of weapons type, trade theory, submarkets, and number of suppliers.

Ch. 18: Economics of Arms Trade

543 A

Few Suppliers

I Submarkets 1 23

Oligopoly Trade Theories

I I

/ S

/

PGMs, aircraft carriers, sophisticated submarines, I --.Zenth

.\

helicopters

~~~~~~~~~~~~~~~~~~~~~I I

Monopolistic Competition Trade Theories

/ Slow fighter aircraft, transport aircraft, detection

technologies, military communications systems, large destoyers, submarines, frigates, transport helicopters, tanks, howitzers, night vision technology

All submarkets

I Many suppliers

/

Mortars, rifles, machine guns, grenades, jeeps, trucks,

F

Competitive Market Trade Theories . supply & demand neoclassical model

ammunition, minelayers and minesweepers, patrol boats, gun boats, light helicopters, simple communications and detection systems, simple aircraft

Figure 9. Typology of the weapons trade.

3.2. Domestic political economy

3.2.1. Defense industrial base Many governments view the maintenance of a defense industrial base (DIB) as essential to national security. In the event of war or in negotiations where war is a possible outcome, a nation with a strong DIB can deter a rival or improve a negotiated settlement. The DIB argument can enhance or contract the arms trade depending on its impact on trade policy and the nation's position in the arms market. For example, Rongovia might subsidize its machine gun production and move from an importer to an exporter of machine guns. Alternatively, Rongovia might subsidize its machine gun production, import less, but remain a machine gun importer. If economies of scale are widespread throughout weapons industries of the world, then government efforts to maintain defense industrial bases over many weapons systems might actually shrink the weapons trade and reduce the number of weapons in the international system. Indigenous procurement and subsidization of production lines for many weapons systems may not reap the gains from economies of scale that subsidization or even laissez faire of a few lines would. The result is production of many weapons systems in many countries, missing out on large production gains from specialization and economies of scale. This is shown in Figure 10. Assume average costs are the same in Germany and Israel for each weapons system. Assume Germany specializes in ships, spends P .S 1 on ships, and then exports half of its ships for half of Israel's missiles. Israel produces Ml missiles, exporting half

C.H. Anderton

544 ships

missiles

price

price

p

P. S2

.5S1

S1 quantity

M2

.5M1

M 1

quantity

Figure 10. Costs of maintaining defense industrial base.

to Germany. Now, suppose Germany subsidizes its missile production and Israel its ship production to have a defense industrial base in each system. Assume that trade ceases and each country produces S 2, M 2 for its own needs. Germany and Israel spend (P2.S 2 ) + (P2.M 2 ) for weapons which is higher than what they used to spend, (P .SI1) and (Pn.*MI), respectively (we assume that PS 2 = =Ps S 1 and Pm M 2 =P .M 1). The difference between the old and the new weapons expenditure levels represents the subsidy to maintain defense industrial capacity in both goods. In spite of greater spending for weapons by both countries, the arms trade has ceased and the volume of weapons in the system has declined (2S 2 <S1 ; 2M 2 <M 1). Of course, it is ultimately an empirical question whether the subsidy effect, which tends to increase weapons production, is outweighed by the unspecialization effect, which tends to diminish weapons production. 3.2.2. Employment Associated with a decline in demand for defense goods is a decline in employment in the defense industry. Weapons exports are often cited as a mechanism to maintain employment in the defense industrial sector in the face of declining domestic demand for weapons. Table 5 presents selected empirical studies of economic impacts of arms trade. While the three studies highlighted in the table consider a broad range of economic variables affected by the arms trade (e.g., jobs, inflation, exchange rates, etc.), employment is a focal point of the studies. In each case, significant job displacement occurs from arms export restraints. Job displacement occurs not only directly in defense manufacturing, but indirectly in other sectors via multiplier effects [as in CBO (1976) and Bajusz and Louscher (1988)] and inter-sectoral linkages [as in Grobar et al. (1990)]. One of the weaknesses of the multiplier analyses of CBO (1976) and Bajusz and Louscher (1988) is that they are partial approaches. The employment displacements that occur in these frameworks ignore the employment replacements that can occur in other sectors of the economy through the release of resources from defense production. The advantage of the Grobar et al. (1990) model is that it considers how other

Ch. 18:

Economics of Arms Trade

545

.q)

.9

u

o

o

00

ON O

0

o .0 C

o

ON C

U) -

ON% O

N O 0 U)a~

C. o 0

6

UO

.5 0.0 00~

Ut 0E 00

VC

00 0)

0)

S

On

ON

0025

.-a~

0

.5-

-"I

D U) 0 O

00

0

'O0 y0 c9

ES

0 .o 0

o

O

F-0. O

O

":O

D

0

ON

0)0)

0 .)

S 0

(d

'0

Y:

I

0

0

Sa .

ONs

o

aj u) 4. .v

Co

bQ

S

o~

D.._

o00

° ON .

0

.CU)

,0

-)

Cs

b

Z °E! N

PC)

O

. ,

000

.0

0E

E c

-

:

0)

b^ O

cd

3 ·E

6

r4

tON

r

. *0 04

S

U) U),

- E E

5n

U) 0. U) ;) 4.C .0)

Z o .

5

00

0 I z v ._

0 ".0

S~

.00 I

.r

o

i

B N 00000

9 c2

0 U) B

.0

o 00O

R 0 0 DS I >.

0 u

e

0 E

M . 0-

-O

"

SO

6o'0

U)

000

5i 0. ' 0

w

'b

>0 U)

S 0

t 9 O U)

'5

oa

H

0 0)

5-E

oo~a.5o

P 0

5B 0

)Cx

0a o

5 5~~

ON

CU 0. 0)0 0 0.

.-

egt QO r A

bD

00

U

0)

00

00)00 00.o~D

n

,

o

oO

00

0 w

-n

-

U)

C.H. Anderton

546

sectors of the economy can expand with a decline in weapons production and exports. Grobar et al. found that a unilateral US embargo on exports and imports of military goods led to significant net employment declines in the US transport equipment and electrical machinery sectors. Roughly two dozen other sectors of the US economy, however, experienced net employment increases from the unilateral US embargo. Broadly similar results were obtained for the USA under a multilateral arms export restraint system. 3.3. Economic and political synthesis Levine, Sen and Smith (1994) have developed a model of the international arms market that synthesizes two major motivations of arms transfers: (1) economic benefits and (2) security repercussions of recipient behavior. The model describes an arms market consisting of a set of similar competing suppliers, who have both economic and security objectives, selling to a single generic recipient. The total amount of weapons supplied to the recipient, Qt, and the end-of-period aggregate stock of weapons held by the recipient, St+l, are defined by: Qt = Qit +qit,

St+l = (1 - d)St + Qt,

(9,10)

where: Qit, amount of weapons supplied by all countries other than i; qit, amount of weapons supplied by country i; d, depreciation rate. The recipient's demand schedule for weapons is: Pt(Qt) = a - bQt = a - b(Qit + qit),

(11)

where P is the price of weapons. Supplier i's single-period objective function, Uit, is made up of profits from the arms sales, rit, and security repercussions, V(St), which is a function of the stock of arms held by the recipient: Uit = Tit + V(St),

(12)

where sit = P(Qt)qit - cqit, V(St) = gSt + eSt2, (13,14) with c being the supplier's average and marginal cost and Equation 14 being a quadratic security function. The parameters g and e in Equation 14 represent various security aspects of arms transfers. Two are the straightforward cases of sales to allies (g >0, e > 0) and adversaries (g < 0, e < 0). Sales to allies increase a supplier's security while sales to adversaries reduce security. In two other cases, the short-run and long-run effects of arms transfers differ. In the "legitimate security limits" case (g > 0, e < 0), arms transfers increase security up to a point, but beyond that point reduce security. In the "potentially cooptive" case (g < 0, e > 0), transfers reduce security in the short run, but in the longer run the supplier can coopt the recipient thus increasing security. In a dynamic version of the model, the authors derive the following propositions (among others):

Ch. 18: Economics of Arms Trade

547

(1) If the recipient is an ally or is sufficiently capable of being coopted, the total volume of arms sales exceeds the pure profit maximizing level. (2) If the recipient is an adversary or represents a "legitimate security limit", the total volume of arms sales is below the pure profit maximizing level. (3) Greater competition in the arms market (i.e., greater number of suppliers) increases the total volume of arms sales. (4) If the recipient is an ally or is sufficiently capable of being coopted and the suppliers can credibly pre-commit to future deliveries of weapons and spare parts, then the effect of forward-looking behavior by the recipient is to import a greater volume of weapons. If the suppliers lack a reputation for pre-commitment, then the effect of forward-looking behavior by the recipient is to import less weapons. The Levine, Sen and Smith model not only treats the commercial and security motives for arms exports, it also highlights the likely lobbying strategies of arms manufacturers. Weapons manufacturers have a profit incentive to counteract arms export restraints. If the recipient is questionable from a security point of view (as in the "legitimate security limit" case), arms manufacturers would tend to discount security considerations and highlight the economic benefits of arms exports. On the other hand, when security is promoted by weapons sales (as in the allies case), arms manufacturers would highlight the importance of security considerations in arms sales. An extension of the Levine et al. model could be the development of an analytical framework for how the supplier's objective function is a product of the domestic political process.

4. Arms trade and conflict SIPRI (1971, p. 73) states that "perhaps the most important question about arms supplies [is] ... what effect they have on the development of wars - on the likelihood of wars breaking out, on the course of wars and on their general severity". The "destabilizing school" generally believes that arms transfers promote conflict by exacerbating regional or international tensions, strengthening the clout of the military, and providing the tools to carry out a greater level of conflict. The "stabilizing school" generally believes that arms transfers can restrain conflict by restoring a balance of power in unstable regions. Connections between arms trade and conflict, however, are not easy to make. Arms transfers can increase or decrease the likelihood of inter-state war or internal conflict depending upon a variety of intervening variables: conflict setting (i.e., intensity of rivalry, issues at stake, history, geography), types and quantities of weapons transfers, existing weapons stocks, goals and perceptions of political leaders, commitments of allies, and prospects of foreign intervention. In many cases there will be too many intervening variables to establish causality between arms trade and conflict. But, if the word "causal" is used in the weaker sense of Dessler (1991, p. 347) to mean "an insufficient but necessary part of an unnecessary but sufficient condition", then what we would really mean when we use the word causal is whether arms transfers promote or restrain conflict [Kinsella (1994, p. 20)].

548

C.H. Anderton

The relationship between arms transfers and the outbreak and longevity of wars is similar to the question of the relationship between arms rivalry and war. Since arms transfers are part of the material of arms racing (along with the rivals' indigenous weapons production), understanding the impact of arms rivalry on the probability and severity of war would help us understand connections between arms transfers and war. The literature, however, is ambiguous on the relationship between arms races and war (see chapter 6 of the Handbook). The necessary research directions on the relationship between arms trade and conflict are theoretical and empirical. Can arms trade scholars sort out the set of variables and conditions under which arms transfers promote and restrain conflict? Relatively little systematic theoretical and empirical work has been directed to this question, although useful case study research and anecdotal evidence abounds. 4.1. Theoretical models of arms trade and conflict Figure 11 shows attack and defend regions in a rivalry between J and L. The weapons stocks of the two nations, Wj and WL, are measured on the axes. If the weapons vector is in region 1, mutual security exists, i.e., neither party can successfully initiate attack against the other. If the weapons vector is in region 2J or 2L, then respectively, J and L could attack the other and win the war. The slopes and positions of the attack/defend lines can be derived from the Lanchester war model [Anderton (1990)]. A variety of variables affecting the course of war - quality of weapons, terrain, defense production capacity, perceptions of decision makers, resource constraints - can be embodied in the attack/defend lines depending on the variant of the Lanchester model chosen [Anderton (1992, 1993)]. Even the simple model of Figure 11 illustrates the ambiguity in the relationship between arms trade and conflict. Suppose the weapons vector is at point A. Based on the technical specifications of the Lanchester model, L could attack and defeat J. Some policy makers might conclude that the regional imbalance in weapons between J and L might lead to aggression by L. A third nation, say K, could move the weapons vector from A to B by exporting weapons to J under the pretext that a regional weapons balance would reduce the possibility of war. This stabilizing view of weapons exports underlies the arguments of some US officials for allowing US weapons exports to Bosnia in its conflict with the Serbs. It could be argued, however, that the weapons export illustrated in Figure 11 increases the likelihood or intensity of war. Party L might perceive the arms exports to J as eliminating L's advantage or eventually moving to a point where J could attack L. L might initiate war (i.e., preempt) or prosecute an existing war more fervently before it loses its dominance or becomes vulnerable. In this case, the weapons exports to J serve as a catalyst for war. The preemption motive for war based on arms transfers is cited as a possibility in Pakistan's 1965 attack on India [Gerner (1983, pp. 20-21)] and Israel's 1956 attack on Egypt [Becker (1977, p. 4)]. Critics of US arms export restraints on Bosnia claim that months of speculation about whether the restraints

Ch. 18: Economics of Arms Trade

549

can attack

W Figure 11. War creation and war diminishing aspects of arms trade.

would be lifted gives the Serbs an incentive to be extra aggressive to gain territory before their weapons advantages are diminished. The Wittman (1979) model can be used to evaluate theoretically the promotion or restraining influence of arms transfers on war initiation and termination. The variables in a model of possible war between J and L are (for i =J,L): Pi, country i's subjective probability of winning the war; U, country i's utility function; s, settlement outcome (e.g., amount of territory gained or lost); i, unconditional surrender by i; U(w), i's expected utility of initiating or continuing a war; U'(s), i's expected utility of settlement or not initiating a war. Figure 12 shows two cases: (1) the feasibility of peace between J and L and (2) necessary conditions for war initiation or continuation. If J receives expected utility U,(w) =A ° from initiating or continuing a war, J will only accept a settlement that gives it at least as much utility as A°. Settlements to the left of A' give J more utility than U}(w). L will only agree to a settlement if it is to the right of B,. The area of overlap between points to he left of At and points to the right of B' indicate the feasibility of a negotiated settlement. On the other hand, if U5(w)=AJ is J's expected utility from initiating or continuing a war, the gap between At" and Bt implies that a peaceful settlement is not feasible. In this case, necessary conditions are met for war initiation or war continuation. In Figure 12, suppose U(w)=A° and peaceful settlement holds. If L receives new weapons imports and J does not, it could create the necessary conditions for war initiation. Since L has additional weapons and J does not, L's expected utility from war would rise while J's would fall. If L's expected utility rises sufficiently more than J's expected utility falls, the settlement region could vanish, possibly giving way to war initiation. Alternatively, the settlement region might not vanish but it would move in L's favor as L's new source of weapons gives it greater relative negotiating power. In Figure 12, suppose UJ(w) =Al and J and L are at war. Suppose L receives new weapons imports. If L's expected utility rises by a sufficiently smaller amount than J's expected utility falls, a settlement region could emerge. US exports of Stinger missiles

550

C.H. Anderton

settlement region

settlement region

settlements favorable to L settlements favorable to J

Figure 12. Peace and war in the Wittman model.

in the 1980s to the Mujahadeen may have dramatically reduced the Soviet Union's expected utility from continuing the war, thus contributing to their withdrawal. The Wittman model seems inherently ambiguous on the relationship between arms trade and conflict initiation or termination. Only if the weapons trade has a sufficiently asymmetric impact on expected utilities of war for the two parties will a peaceful settlement shift to war or a war settlement shift to peace. Where the Wittman model is unambiguous is on the effects of weapons trades on the relative negotiating powers of the conflicting parties. 4.2. Empiricalstudies Table 6 summarizes selected empirical studies of the relationships between arms trade and conflict. Relative to the arms race modeling literature where over one hundred empirical studies can be found, the arms trade/conflict empirical literature is quite sparse. The majority of the studies cited in Table 6 indicate that arms trading promotes conflict, but there are enough exceptions to suggest that the relationships between arms trade and conflict are not amenable to general statements. The results vary quite widely depending on the nations, data sets, empirical models, and time frames of the studies. Some of the ambiguity of the arms trade/conflict relationship, which is borne out in the theoretical models, is prevalent in the empirical literature as well. The necessary empirical research directions on the arms trade/conflict relationship are two-fold. First, some of the new empirical work should draw upon existing studies and data sets in order to develop a cumulative effort. Second and perhaps more importantly, empirical research should, if possible, identify the conditions under which arms transfers seem to promote conflict in some situations but not in others.

Ch. 18: Economics of Arms Trade

O 44)

551

ll6

N

o

aC

so

O

0r oN

-r

PI

Id

C

'o 'N O

ON

ON

"I

ON -

O0

O ONi

0 0

4)0

I "0 Q o

U

E .3

Id

U

e)

0

m

.

-u

o 0

U

4

O>

m

0, 4 2

-_

N S

0 i

-o2

= Cl E ON

^m

v

0 4 .0

.2 o

4)~~

24)500ZoS·

2

=

004

t

0I

0~6 Ots

N0

0. .2-C_ wt .0q .,

z 0

i

>

0

m 2

8 0 -. . .n

253

02

0'

0

22 .> D

.2

B' 4 0 2~ E4)' · .2 2

Q-

0"·f 0. . E

44Nr

4)O

0

4)0i

F z;;

0

20 o 2

0

~

9

, 4)

0 0 0

.2

3

0

og

Q

:

0

.5

,,,0 B

ON O

2

I o=

U

o

5.22

O

a- 4 2 2

o

.2

e;

2

540 ON

0o--~q 3Cg

-

"0

2

a

04 ON

Co~a d -.0

U

., tU

O

2

)

.4

4)

~

~..

424 o

o

*aZ o

N-g:

g

00

.0

o

o ) .0

mb

~

a


va

ON

8·

t

0 0 En

4)

bn

..& O .. -

ON

3004

U .

:5

o

~

ONg 4)4)

a

4H 0

U

o "0

0.

i,;

0 ON . -5

Du

0

c D

0 0.~

D' 9

.

N

5 5 'E

0~

'0

0 4) "0

0

M

ON

0,

z

;

OI

~

ON

02

Q

4) 0 0

3

0

0

= .E

0

4

t$" Ž0 2 S0J

~-

"

0

(C

"0

2 o 2,

2 4 Ž) 00 . 0)

-

g.
4 ->.

2.

~

u

o

4)4

u)

)w

._

c u

4)

a0

4

4)

aO a0 . n ,,, ,

0

EX

a

22

24)2N)

)

a;~~~~

2 .

0

.

4)

22

4) e40

Q0

2'c-

4)o

ao

ON~ V)

222

2f" Xs~~

0 U 00 "·fisg~~~~4

c 44)

<m

·

. 4.-

"

Ute

i

O

>. 4.

>

o

d~~S as~~4~4)-

._s

c

w

.2

4)4)

c2g 2~2

4).

2! S 4)

.e

11-

.

z

40: 4)

4

r~

._

04

.0 4 4)

a

-

~~ ~ ~

0

) 4

0

4.

Ch. 18: Economics of Arms Trade

553

4.3. Case studies The arms trade literature is replete with useful qualitative studies on the relationship between arms trade and conflict. An early example is Kemp's (1970) study of the correlations between arms trade and Third World conflict between 1945 and 1970, with an emphasis on the 1960s. During this period, many Third World conflicts took place and "nearly all the weapons used in these conflicts ... originated from the factories of the industrial powers." [Kemp (1970, p. 5)]. Nevertheless, Kemp maintains that "it cannot be argued from this evidence alone that arms cause or contribute to conflict." [Kemp (1970, p. 9)]. The relationships between arms supplies, conflict and Third World stability are exceedingly complex. Kemp's paper is designed to cull out general principles about arms transfers, military requirements, and conflict in the Third World and then apply these principles on a regional case-by-case basis. A summary of Kemp's principles is that in evaluating the impacts of arms transfers on Third World conflict, it is important to estimate the effectiveness of the weapons transferred. Effectiveness is a multi-dimensional variable depending on military environments, technical specifications of weapons and political factors. "Only if it is possible to apply qualitative criteria based on some measure of a weapon's "effectiveness" can one argue that the transfer of some types of weapons would introduce more destabilizing factors in the Third World security environments than the transfer of others." [Kemp (1970, p. 22)]. Kemp recognizes that it is difficult if not impossible to label weapons as "offensive" or "defensive" based on technical specifications alone, but this does not mean that no judgment can be made. "When making a qualitative distinction between various classes of weapons, it is essential to be more precise about the various types of military, geographical, and technical environments that exist, or are believed to exist, in various areas of the Third World." [Kemp (1970, p. 26)]. Kemp undertakes a region-by-region analysis of the impacts of arms trade and arms trade control on Third World environments in terms of peace and security in those areas. Kemp (1970, p. 70) reaches the following conclusions: ". .. it has been suggested that the arms trade to the Third World is neither "good" nor "bad", that certain types of arms transfers can both serve and harm the interests of suppliers and recipients and, more particularly, that there is no a prioritruth that predetermines that arms transfers inevitably contribute to the outbreak of conflict." Figure 11 above illustrates how the arms trade might restrain or promote conflict. The direction of causality is from arms trade to conflict. Neuman (1986) considers how conflict, by fueling the demand for weapons, can cause more arms trade. She investigates the impacts of war on arms transfers for eight less developed country wars - five conventional inter-state wars and three insurgencies. Neuman finds that for most warring states, the dollar value of arms imports escalated during conflict, but the contributions of the USSR and, particularly, the USA declined, with other suppliers moving in to meet the demand. Neuman (1986, p. 92) concludes that both superpowers showed concern about escalating Third World conflicts with arms exports,

554

C.H. Anderton

and exercised restraint, particularly during the early stages of war or until a political or strategic advantage emerged. Harkavy (1987) also considers the interrelationships between arms trade and ongoing conflict. He lays out the myriad dimensions of the subject and creates a preliminary classification of cases of arms resupply during conflict. Harkavy believes that there has been so little systematic treatment of the subject that it is premature to venture generalizations or describe causal interrelationships. Pearson et al. (1992) analyzed the effects of arms transfers on ten inter-state wars, reaching the following general conclusions: (1) Arms transfers are a factor in decisions to go to war because of considerations of military superiority, perceptions of changes in the balance of power, and interest in establishing links with supporting states. (2) Supplying states have little leverage in conditioning or even determining the outcome of hostilities. (3) Arms transfers generally prolonged and escalated wars, resulting in more suffering and destruction. (4) Arms embargoes, whether partial or total, in and of themselves had little chance of compelling warring parties to stop wars or come to the negotiating table. They did serve to contain the actual fighting. (5) The two superpowers were more prone to use arms transfers as a means to influence warring parties than were other suppliers. During wars the superpowers' share in deliveries generally declined. Consistent with the theme of ambiguity in some of the theoretical and empirical literature described above, Pearson et al. (1992, p. 399) conclude: Arms transfers have both precipitant and deterrent effects on armed conflicts; they are also a factor in the conduct and the cessation of wars. Often heard categorical statements either that arms transfers are inherently conflict enhancing or that, properly managed, they are a trustworthy instrument for stability - ignore the complex record. Despite the obvious relevance of these factors, the literature contains little systematic evidence of the actual effects of arms transfers on conflicts.

5. Arms trade and foreign policy influence Arms suppliers often use their weapons exports to create foreign policy influence over arms recipients. For example, Snider (1978) found empirical evidence that, under certain conditions, some weapons exporters gained influence over several Middle Eastern countries through arms transfers. Roeder (1985) found that Soviet economic and military aid to Third World countries from 1960-1980 increased the recipient's trade dependence on the Soviet Union, which in turn induced political compliance. Case studies of the US-Israeli relationship, e.g., Nachmias (1988), Pollock (1982), and Wheelock (1978) found that the USA sometimes used its weapons exports

555

Ch. 18: Economics of Arms Trade Table 7 Frequency of US arms influence attempts by administration', 1950-1992 Administration Truman (1950-1952) Eisenhower (1953-1960) Kennedy (1961-1963) Johnson (1964-1968) Nixon (1969-1974) Ford (1975 1976) Carter (1977 1980) Reagan (1981-1988) Bush (1989 1992) Total (1950-1992) a

Average attempts per year 3.0 2.0 3.0 6.4 3.8 8.5 10.5 3.5 3.8 4.5

Source: Sislin (1993, p. 159).

to influence Israel. Table 7 shows the frequency of 191 arms influence attempts by US Presidential Administration. Although arms influence attempts are widespread, they are successful only part of the time. For example, Snider (1978, p. 262) says that his relationships do not always hold and Gelb (1976/77) shows examples of both US successes and failures in arms influence. Although these studies and others identify both successes and failures in arms influence attempts, they fail to solve the key puzzle: why are such practices successful only part of the time? Answering this question requires that the context of arms trade influence attempts be thoroughly examined. "The key question then becomes: What environmental, contextual, or intervening variables involved in the influence attempt are important in determining success?" [Sislin (1993, p. 4)]. 5.1. Sislin 's theory of arms influence success Sislin (1993) makes a major contribution toward identifying the determinants of arms influence successes by offering a theory of arms as influence. He considers two decision-makers: recipient and supplier that are assumed to behave as unitary actors [for caveats see Sislin (1993, pp. 47-48)]. Stage 1 has the recipient engaged in some behavior which the supplier does not approve of. In Stage 2, the supplier decides to manipulate arms exports to alter the behavior of the recipient. Finally, in Stage 3, the recipient decides how to respond to the supplier's influence attempt.

Although Sislin did not present his three-stage theory mathematically, it can be done as follows: Stage 1 Vs(c) > VS(nc), where VS(c) is the value to the supplier of compliance by the recipient; V(nc) is the value to the supplier of non-compliance by the recipient.

556

C.H. Anderton

Condition 15 is necessary for an arms influence attempt, but it is not sufficient. "It is possible, and frequently the case, that the costs of the influence technique, ... , would be so high as to lower the benefit the supplier receives from compliance to the point where more value could be obtained by simply doing nothing." [Sislin (1993, p. 47)]. Stage 2 (Supplier Decision Calculus) E [VS] = Pc V(c) +(1 - Pc) V(nc) - Cs(i) > VS(nc) where E[Vs] is the expected value to the supplier of an arms influence attempt; P,, probability of recipient compliance as perceived by the supplier; Cs(i), cost of influence attempt as perceived by the supplier. Ignoring risk preference, condition 16 is sufficient for an arms influence attempt. A variety of arms influence techniques might satisfy condition 16 [see Sislin (1993, pp. 51-54)]. Assuming that only Pc and Cs are functions of the arms influence attempt, the supplier chooses an arms influence instrument to maximize E[VS]. Stage 3 (Recipient's Decision Calculus) Vr(c) > Vr(nc) - PiCr(i),

(17)

where Vr(c) is the value to the recipient of compliance; Vr(nc), value to the recipient of non-compliance; Cr(i), costs to the recipient of arms influence instrument; Pi, probability that costs of arms instrument will actually be incurred. Ignoring risk preference, condition (17) is sufficient for the recipient to comply with the arms influence attempt. 5.2. Determinants of arms influence success What are the key determinants under which Equations (16) and (17) would hold, giving rise to arms influence success? Sislin (1993) addresses this question for the United States based on his data set of 80 successful influences out of 191 attempts over the 1950-1992 period. Table 8 lists the variables that Sislin identifies as possible determinants of arms influence success. Using a winnowing procedure on multivariate logit analysis, Sislin identifies the core determinants of arms influence success. He finds six variables that seem to play a significant role in the probability of successful influence: sanction type, policy, regime, hegemony, supplier arms, and alternate supplier. Sislin's (1993, p. 187) general conclusion is: "Influence attempts are more likely to succeed when (1) sanctions are positive (reward or promise) rather than negative; (2) the focus of the attempt is to alter the recipient's foreign rather than domestic policy; (3) the recipient is a civilian regime, not a military one; (4) the United States was more powerful in the world system; (5) the recipient receives more of its arms from the United States; and (6) the recipient has less alternative suppliers available to it." Sislin's research implies that arms influence success is not determined by one or two key variables, but by a set of variables which defines the context for a high probability

Ch. 18:

Economics of Arms Trade

557

Table 8 Possible determinants of arms influence successa I. Characteristics of influence attempt 1 Sanction type - supplier's sanction is positive or negative 2 Goal - supplier's goal is deterrence or compellence 3 Policy - supplier is attempting to alter recipient's foreign policy or domestic policy II. Characteristics of recipient A. Political 4 Regime - recipient state ruled by military elites or civilian elites 5 Recipient cohesion - recipient elites are cohesive or fragmented B. Military 6 Conflict - recipient engaged in international conflict or not 7 Security - recipient feels secure or insecure 8 Civil strife - recipient leaders face domestic conflict or domestic tranquility C. Economic 9 Resource - recipient's resource base 10 Indigenous arms - recipient's level of indigenous arms production III. Characteristics of interaction between recipient and supplier 11 Supplier arms - degree of dependence of recipient on supplier's arms 12 Trade - degree of economic interaction between supplier and recipient 13 Precedent - outcome of previous influence attempt (if any) IV. Characteristics of supplier 14 Supplier cohesion - supplier elites are cohesive or fragmented 15 Presidential style - active-positive US presidents versus other types V. System characteristics 16 Hegemony - trend of US hegemony 17 Cooperation - degree of friendliness or hostility between the US and USSR 18 Alternate supplier - number of alternative arms suppliers available to the recipient a

Source: Sislin (1993, pp. 108-109).

of arms influence success. Altering any one of the variables in the set has only a small impact on the probability of arms influence success. Sislin's results challenge some of the conventional wisdom in the arms influence literature. For example, the recipient's level of indigenous arms production is sometimes cited as a variable which diminishes a recipient's vulnerability to arms influence. Sislin finds that indigenous arms production is not a significant determinant of arms influence success or failure. Another example is the belief by some analysts that a nation involved in conflict is more likely to be influenced through weapons restraints. Sislin, however, finds that this variable is generally insignificant.

558

C.H. Anderton

5.3. Recipient influence Sislin (1993, pp. 196-200) identifies directions for future research: (1) test other suppliers, (2) consider other forms of influence attempts, (3) test the robustness of his work, and (4) comparative analyses of arms influence with other forms of influence. One direction of research that Sislin did not mention is recipient influence. The literature on arms influence generally presumes that the supplier has the leverage to initiate an arms influence attempt. Although the presence of alternative suppliers might diminish the recipient's vulnerability to the arms influence attempt, the action is still from the supplier to the recipient. It is possible, however, that the consumer of weapons could exercise foreign policy influence over the supplier. The economic perspective reminds us that a market is made up of two sides. Under conditions of pure competition or bilateral monopoly (or bilateral oligopoly, etc.), each side of the market might lack a clear path for creating arms influence. Other market structures (e.g., monopsony, oligopsony) might allow the buyer to levy influence on the seller. The arms trade literature to date, theoretically and empirically, has not investigated arms trade influence from the buyer to the seller. This part of the spectrum could be relevant in the immediate post-Cold War era which is often characterized as a "buyer's market". Nykvist (1995) has undertaken theoretical and case study research on arms trade foreign policy leverage from the buyer's side.

6. Concluding comments The international arms market experienced a dramatic structural change with the end of the Cold War. Economics has risen in importance, relative to geopolitics, in understanding the motivations and consequences of the arms trade. Nevertheless, there is surprisingly little theoretical and empirical development of the economics of arms trade. Almost the whole body of international trade theory remains untapped. It seems reasonable to conclude that the economics of arms trade is essentially an unexplored sub-field of defense economics, ripe for foundational theoretical and empirical contributions. Fields other than economics have contributed much to our understanding of arms transfers. A myriad of useful case studies exist, offering important insights and data on the geopolitics of the arms trade. Empirical sub-literatures on foreign policy leverage and the relationship between arms trade and conflict are now in place beckoning researchers to further develop these sub-fields. The international systems model of Harkavy (1975), Laurance (1992) and others is a useful theoretical and applied approach for understanding the geopolitics of arms transfers. Catrina (1994) identifies many arms trade research directions, including some on the economics of arms trade. At this point, the central problem of the arms trade literature, for both economists and geopolitical scientists, is how to go beyond the proliferation of policy position articles and descriptive studies (which are useful and important in their own right), to

Ch. 18: Economics of Arms Trade

559

a more solid base of theoretical and empirical models. A new theoretical synthesis of economics and geopolitics would represent a major breakthrough. The development of the economics of arms trade must move beyond its incipient stages for this synthesis to occur, but it seems that Levine, Sen and Smith (1994) have already taken steps in

this direction. Ultimately, it would be desirable to see theoretical and empirical work on the arms trade narrow the range of outcomes which are regarded as plausible.

References Alexander, A.J., W.P. Butz and M. Mihalka, 1981, Modelling the production and international trade of arms: An economic framework for analyzing policy alternatives (The Rand Corporation, Santa Monica, CA). Anderton, C.H., 1990, The inherent propensity toward peace or war embodied in weaponry, Defence Economics 1, 197-218. Anderton, C.H., 1992, Toward a mathematical theory of the offensive/defensive balance, International Studies Quarterly 36, 75-100. Anderton, C.H., 1993, Arms race modeling and economic growth, in: J.E. Payne and A.P. Sahu, eds., Defense spending and economic growth (Westview Press, Boulder, CO) 55-82. Bajusz, WD., and D.J. Louscher, 1988, Arms sales and the U.S. economy: The impact of restricting military exports (Westview Press, Boulder, CO). Banks, A., 1971, Cross-polity time-series data (MIT Press, Cambridge, MA). Baugh, W, and M. Squires, 1983a, Arms transfers and the onset of war part I: Scalogram analysis of transfer patterns, International Interactions 10, 39-63. Baugh, W., and M. Squires, 1983b, Arms transfers and the onset of war part II: Wars in Third World states, International Interactions 10, 129-141. Becker, A.S., 1977, Arms transfers, great power intervention, and settlement of the Arab-Israeli conflict (The Rand Corporation, Santa Monica, CA). Bobrow, D.B., P.T. Hopmann, R.W. Benjamin and D.A. Sylvan, 1975, The impact of foreign assistance on national development and international conflict, Journal of Peace Science 1, 39-60. Carus, W.S., 1994, Military technology and the arms trade: Changes and their impact, The Annals of the American Academy of Political and Social Science 535, 163-174. Catrina, C., 1988, Arms transfers and dependence (Taylor & Francis, New York). Catrina, C., 1994, Main directions of research in the arms trade, Annals of the American Academy of Political and Social Science 535, 190-205. CBO (US Congressional Budget Office), 1976, The effect of foreign military sales on the U.S. economy (Congressional Budget Office, Washington, DC). Dessler, D., 1991, Beyond correlations: Toward a causal theory of war, International Studies Quarterly 35, 337-355. Engelbrecht, H.C., and FC. Hanighen, 1934, Merchants of death: A study of the international armament industry (Dodd, Mead, New York). Ferejohn, J., 1976, On the effects of aid to nations in an arms race, in: D.A. Zinnes and J.V Gillespie, eds., Mathematical models in international relations research (Praeger, New York) 218-251. Ferrari, P.L., J.W. Knopf and R.L. Madrid, 1987, U.S. arms exports: Policies and contractors (Investor Responsibility Research Center, Washington, DC). Frey, B.S., 1975, Weapon exports and aid to developing countries, Journal of Peace Science 1, 117-126. Gelb, L., 1976/77, Arms sales, Foreign policy 25, 3-23. Gerner, D., 1983, Arms transfers to the Third World: Research on patterns, causes and effects, International Interactions 10, 5-73.

560

C.H. Anderton

Grobar, L.M., E.M. Stern and A.V. Deardorff, 1990, The economic effects of international trade in armaments in the major western industrialized and developing countries, Defence Economics 1, 97-120. Harkavy, R.E., 1975, The arms trade and international systems (Ballinger, Cambridge, MA). Harkavy, R.E., 1987, Arms resupply during conflict: A framework for analysis, in: C. Schmidt, ed., The economics of military expenditures (MacMillan, London) 141-175. Harkavy, R.E., 1994, The changing international system and the arms trade, The Annals of the American Academy of Political and Social Science 535, 11-28. Hartley, K., 1987, Efficiency, industry and alternative weapons procurement policies, in: C. Schmidt, ed., The economics of military expenditures (MacMillan, London) 283-300. Johnson, J., 1994, Financing the arms trade, The Annals of the American Academy of Political and Social Science 535, 110-121. Jones, P.R., 1988, Defense alliances and international trade, Journal of Conflict Resolution 32, 123-40. Kemp, G., 1970, Arms traffic and Third World conflicts, International Conciliation 577, 1-77. Kiefer, D., 1988, Interstate wars in the Third World: A Markov approach, Conflict Management and Peace Science 10, 21-36. Kinsella, D., 1994, The impact of superpower arms transfers on conflict in the Middle East, Defence and Peace Economics 5, 19-36. Kolodziej, E.A., 1987, Re-evaluating economic and technological variables to explain global arms production and sales, in: C. Schmidt, ed., The economics of military expenditures (MacMillan, London) 304-335. Krugman, P., 1989, Industrial organization and trade, in: R. Schmalensee and R. Willig, eds., Handbook of industrial organization, Vol. 2 (North-Holland, Amsterdam) 1179-1223. Krugman, P., and M. Obstfeld, 1994, International economics: Theory and policy, 3rd edition (HarperCollins, New York). Laurance, E.J., 1992, The international arms trade (Lexington Books, Lexington, MA). Leiss, A.C., 1970, Changing patterns of arms transfers: Implications for arms transfer policies (Center for International Studies, MIT, Cambridge, MA). Leiss, A.C., G. Kemp, J.H. Hoagland, J.S. Refson and H.E. Fischer, 1970, Arms transfers to less developed countries (Center for International Studies, MIT, Cambridge, MA). Levine, P., S. Sen and R. Smith, 1994, A model of the international arms market, Defence and Peace Economics 5, 1-18. Majeski, S., and D. Jones, 1981, Arms race modeling: Causality analysis and model specification, Journal of Conflict Resolution 25, 259-288. Maniruzzaman, T., 1992, Arms transfers, military coups, and military rule in developing states, Journal of Conflict Resolution 36, 733-755. McGuire, M.C., 1987, US foreign assistance, Israeli resource allocation and the arms race in the Middle East: An analysis of three interdependent resource allocation processes, in: C. Schmidt, ed., The economics of military expenditures (MacMillan, London) 197-238. Milstein, J.S., 1972, American and Soviet influence, balance of power, and Arab-Israeli violence, in: B.M. Russett, ed., Peace, war, and numbers (Sage, Beverly Hills, CA) 139-166. Moodie, M., 1994, Constraining conventional arms transfers, The Annals of the American Academy of Political and Social Science 535, 131-145. Muller, J., 1967, The foreign aid programs of the Soviet bloc and Communist China (Walker, New York). Nachmias, N., 1988, Transfer of arms, leverage, and peace in the Middle East (Greenwood Press, New York). Neuman, S.G., 1986, The arms trade in recent wars: The role of the superpowers, Journal of International Affairs 40, 77-100. Neuman, S.G., and R.E. Harkavy, 1979, Arms transfers in the modern world (Praeger, New York).

Ch. 18: Economics of Arms Trade

561

Nykvist, K., 1995, A nontraditional approach to arms trade foreign policy leverage, Department of Economics Honors Thesis (Holy Cross College, Worcester, MA). Pearson, ES., M. Brzoska and C. Crantz, 1992, The effects of arms transfers on wars and peace negotiations, in: SIPRI yearbook 1992: World armaments and disarmament (Oxford University Press, New York) 399-415. Pierre, A.J., 1982, The global politics of arms sales (Princeton University Press, Princeton, NJ). Pollock, D., 1982, The politics of pressure (Greenwood Press, Westport, CT). Roeder, P., 1985, The ties that bind: Aid, trade, and political compliance in Soviet-Third World relations, International Studies Quarterly 29, 191-216. Sandler, T., and K. Hartley, 1995, The economics of defense (Cambridge University Press, Cambridge). Scherer, F.M., and D. Ross, 1990, Industrial market structure and economic performance, third edition (Houghton Mifflin, Boston, MA). Schrodt, P., 1983, Arms transfers and international behavior in the Arabian Sea area, International Interactions 10, 101-127. Sherwin, R., 1983, Controlling instability and conflict through arms transfers: Testing a policy assumption, International Interactions 10, 65-99. SIPRI (Stockholm International Peace Research Institute), 1971, The arms trade with the Third World (Almqvist & Wiksell, Stockholm). SIPRI Yearbook, annual, SIPRI yearbook: World armaments and disarmament (Oxford University Press, New York). Sislin, J., 1993, Arms as influence: The elusive link between military assistance and political compliance, Ph.D. Thesis (Department of Political Science, Indiana University). Sivard, R.L., 1987, World military and social expenditures, 1987-1988 (World Priorities, Washington, DC). Sivard, R.L., 1993, World military and social expenditures, 1993 (World Priorities, Washington, DC). Skbns, E., and H. Wulf, 1994, The internationalization of the arms industry, The Annals of the American Academy of Political and Social Science 535, 43-57. Smaldone, J., and R. Anderton, 1995, Arms transfers and regional conflicts in Africa, mimeograph. Small, M., and J.D. Singer, 1982, Resort to arms: International and civil wars, 1816-1980 (Sage, Beverly Hills, CA). Smith, R., A. Humm and J. Fontanel, 1985, The economics of exporting arms, Journal of Peace Research 22, 239-247. Snider, L., 1978, Arms transfer and recipient cooperation with supplier policy references: The case of the Middle East, International Interactions 5, 241-266. Snider, L., 1987, Do arms exports contribute to savings in defense spending?: A cross-sectional pooled times series analysis, in: D. Louscher and M. Salomone, eds., Marketing security assistance (Lexington Books, Lexington, MA). Stanley, J., and M. Pearton, 1972, The international trade in arms (Praeger, New York). Starr, H., and B.A. Most, 1980, Diffusion, reinforcement, geopolitics and the spread of war, American Political Science Review 74, 609-636. US ACDA (US Arms Control and Disarmament Agency), annual, World military expenditures and arms transfers (US Government Printing Office, Washington, DC). Wheelock, T., 1978, Arms for Israel: The limit of leverage, International Security 3, 123-137. Wittman, D., 1979, How a war ends: A rational model approach, Journal of Conflict Resolution 23, 743-763.

Chapter 19

THE ECONOMICS OF DISARMAMENT JACQUES FONTANEL Universit PierreMendes France de Grenoble

Contents Abstract Keywords 1. Introduction 2. Economic definitions of disarmament 2.1. Disarmament via physical reductions and a ban or limitation on forces 2.2. Military expenditure reduction 2.3. Thoroughgoing disarmament

3. Theoretical economic analysis of disarmament 3.1. 3.2. 3.3. 3.4.

The peace dividend approach Disarmament as an investment Development as the main factor of disarmament Disarmament/armament as elements of a national strategy

4. Empirical analysis of disarmament

564 564 565 565 566 566 567 569 570 571

572 572

575

4.1. The United States 4.2. France

576

4.3. United Kingdom 4.4. Simulations for developing countries

578

5. Disarmament for development

578 579 580

5.1. The nature of the transfers

580

5.2. The economic effect of a disarmament for development process

581

6. The role of public policy

583

7. Conclusion References

585 585

Handbook of Defense Economics, Volume 1, Edited by K. Hartley and T: Sandler O 1995 Elsevier Science B. V All rights reserved

564

J Fontanel

Abstract The economics of disarmament is a new discipline. It involves analyses on the economic causes of the arms race, the definitions of disarmament and the economic determinants and military expenditure. Simultaneously, disarmament is considered as an obstacle to economic development, a peace dividend or an investment. The construction of economic models of disarmament produces some controversial results. Questions arise concerning the economic effects of a reduction in military expenditure on growth, employment, inflation, budget deficits, the costs of conversion or the dual use of military products, capital capacity and R&D. Long term disarmament needs a development process and an adequate public policy.

Keywords disarmament, military expenditure (ME), conversion, arms industry, peace dividend, disarmament for development, national security, investment, growth, econometric models

Ch. 19: Economics of Disarmament

565

1. Introduction Economists have not derived any real inspiration from disarmament for four main reasons. First, they have attempted to establish scientific regularities, they are involved with optimization problems in a peaceful paradigm and usually they are not interested in the multidisciplinary analysis of the real constraints of power, risks and conflicts. Second, if the reduction of military expenditure (ME) is seen as an instrumental variable that may modify the conditions of economic growth, economic factors condition the disarmament process. Third, public interest in security questions is variable reflecting the course of conflicts. Fourth, the defense budget is often analyzed as an unproductive expenditure and it seems to be economically uninteresting; moreover, the theory of public expenditures has been neglected for much of the history of economic thought [Goodwin (1991)]. However, states have to protect their wealth. The basis of conflict and war exists in the contexts of the struggle for power and of religious and ideological convictions. It can reflect a struggle for resources or economic competition. In peacetime, governments have to take into account the balance that must be maintained between impoverishing military expenditure and a necessary level of national security. There is a choice to be made between today's and tomorrow's security and the state must settle on the "razor edge" of spending, choosing just the amount needed to give itself military protection in the short term without threatening the increase in wealth needed to maintain the defense effort. There are three main reasons for disarmament. First, the reduction of "excess" armaments is based on the definition of excessive stocks of weapons [Fontanel (1991)]. Second, disarmament can be imposed by disastrous national economic conditions (the "guns or butter" dilemma). Third, the goal of disarmament is to reduce the risk of armed conflict, that is to increase security. In this context, the economic aspects are of secondary importance. Disarmament may develop the idea of establishing a lasting peace able to overcome the ideological, political and religious differences that may lead to wars. This chapter deals with (i) the economic definitions of disarmament; (ii) the theoretical economic analysis of disarmament; (iii) the empirical analysis of disarmament; (iv) disarmament for development and (v) the role of public policy.

2. Economic definitions of disarmament A disarmament process leads to a modification of a nation's military strategies. There are many forms of disarmament, such as reduction of military expenditure, reduction or destruction of stocks of specific weapons, a ban or limitation on the production of some types of military equipment, reduction in the numbers of military personnel in their geographical location, limitations on arms transfers, control of defense R&D, and the monitoring and verification process [Sandler and Hartley (1995)]. There are

566

J Fontanel

three main forms of disarmament, namely destruction and reduction of military forces, reduction of military expenditure, and "thoroughgoing" disarmament. 2.1. Disarmament via physical reductions and a ban or limitation on forces The superpowers might find an acceptable level of weaponry which they believe ensures their security. Disarmament is then a way of reducing the stockpile of destructive power. Defense spending in an arms race can be analysed by the Prisoner's Dilemma problem of game theory (but it is often difficult to know who are the players, what is the mutual information and what are the nation's preferences?). Classical arms control is useful to decide a lower level of defense spending for each nation, with the same national security, with a strong verification regime [Schelling (1966)]. The destruction of military stocks is a costly business and its initial effect is an increase of the economic burden of defense [Fontanel and Ward (1993)]. Bischak and Oden (1989), attempting to compare the reduction of military purchases, operational and maintenance costs and personnel costs with the financial commitments occasioned by the destruction of arms and the verification of the INF Treaty, estimated that the overall reduction of ME for the USA were equal to $1.5 billion for 1988-2000. Unclassified assessments by the Russian military estimate that implementing START would cost Russia 90 to 95 billion rubles, equivalent to $6 million (1992 prices). Ukraine has requested $2.8 billion to ship nuclear arms to Russia for dismantling [Renner (1994)]. Annual funding for the US Army's Chemical Material Destruction Agency has risen from $200 million in the late 1980s to $500 million in 1994. The US Army demilitarized 300000 tons of ammunition during 1990-1995, at a cost of $300 million, when Congressional appropriations to procure new ammunition add up to $6.6 billion. In Europe, conversion funding by government is marginal ($320 million for Italy, $160 million for France, but German expenditures are greater: $7.6 billion of grants and credits available to facilitate the pullout of ex-Soviet troops). Without reductions of military outlays, quantitative limits to armaments must focus on currently developed weapon systems, and may release resources to develop new and destabilizing weapon systems that are not covered by any treaty and which involve a redesign of strategic doctrine, at greater costs. It is certainly useful to control R&D and to establish reductions of most weapons systems. However, it should be recognized that below certain thresholds physical weapons stockpiles have little utility, either in battle or in preventing it. Under the Ricardian theory of international trade, the government normally seeks to keep those arms of which it has the greatest comparative advantage or opts for the agreement that is most costly in economic terms to its potential adversary. 2.2. Military expenditure reduction For a long time, Soviet leaders felt that attaining a "superpower" status required a large military and were convinced that the free world would not be able to preserve

Ch. 19:

Economics of Disarmament

567

its standard of living and its military advantage. At the same time, the US government felt that pressure on the Soviet Union's military would bring about a final blow to the Soviet economy. These considerations brought about a "rational" arms race only when security has been defined as a zero-sum game. The dynamics of military expenditure progressively undermined the economic basis of the great powers. Military spending provides evidence of the price that is collectively attached to security, but the economic factors are not likely to be the best source of reliable indicators of security. This arises because some types of delivery vehicles (e.g. nuclear missiles) are clearly more effective than conventional weapons, and at a relatively low cost in terms of the strategic, military and political advantages. Moreover, relations of strength are often expressed in terms of thresholds and an equal reduction of ME for each state does not necessarily result in the former balance of forces being maintained. Disarmament by reduction of defense budgets implies that the structure of expenditure has to be taken into consideration. Cuts in defense spending reduce the demand for manpower and capital in the military sector (armed forces and defense industries). Costs of conversion can be very important and last a long time for groups, regions and industries [UNIDIR (1993)1. A disarmament predicated on the notion of cost-savings will also have as a primary goal the increased efficiency of reduced expenditures. Military-industrial institutions will try to find ways, by giving greater attention to waste, fraud, and mismanagement, to produce the same level of weapon stocks with a reduced budget and to exploit economies of scale in order to reduce unit costs or to attempt to produce "better" weaponry. In responding to cost-saving measures, governments will tend to specialize their defenses in areas in which they may have the greatest economic advantage. Reduction of military expenditure must be analyzed in opportunity cost terms. A disarmament predicated on responding to an economic crisis, may well lead towards a less stable and more dangerous global environment. Simply constraining the flow of resources for military purposes will be insufficient to ensure a disarmament that increases global security. 2.3. Thoroughgoing disarmament Thoroughgoing disarmament appears when military expenditure becomes more of an endogenous variable than a true instrumental variable of the government. It may have several causes - economic, political, strategic and even ideological. Economic competition involves cost reductions, better information on the basic rules of defense, the search for international co-operation, some opening up of the market to foreign companies, privatization of state undertakings, limited industrial policy by the state, and the search for economies of scale. The pursuit of a new international co-operation over armaments is almost invariably dictated by the reduction of unit costs of weapons which have become very expensive. When it works properly, which is rarely the case, this international co-operation leads a more equitable calculation of the expenditure to be collectively undertaken [Fontanel and Smith (1990)]. Under these conditions the

568

J Fontanel

Table I Some econometric results Partial economic endogenous variables

References

Civil budget of state Bureaucratic procedures in the allocation of military resources Availability of public funds Per capita income Material well-being of citizens and social spending Economic cyclical movement

Fontanel (1980), Gyimah-Brempong (1992) Fontanel and Smith (1990), Looney (1988, 1989), Byers and Peel (1989) Hewitt (1991) Lotz (1970), Fontanel and Smith (1990) Hewitt (1991) Griffin, Wallace and Devine (1982), Smith and Smith (1983) Correa and Ji-Won Kim (1992) Deger and Smith (1983), Looney (1989) Galbraith (1972), Melman (1974) Looney (1989)

Inertial economic effect Petro-dollars Military industrial complex National arms production Other partialendogenous variables Militarization of the society Power struggle or imagined threats of the ruling class National security, wars and arms race

Urbanization Arms race strategies Ideological priorities of government Political situation (monarchy, dictatorship, ... ) Geographical variables

Maizels and Nissanke (1987), Thee (1982), Whynes (1979) Looney (1989), Griffin, Wallace, Devine (1982) Adams and Gold (1987), Byers and Peel (1989), Hewitt (1991), Brito and Intriligator (1987) Lotz (1970) Correa and Ji-Won Kim (1992), Fontanel and Smith (1990), Rajmaira and Ward (1991) Thee (1982), Hewitt (1991) Hewitt (1991), Nincic and Cusack (1979), Grindle (1986), Kende (1980) Hewitt (1991)

defense sector loses its priorities and is then regarded as a burden. It has been stated that national security can no longer be analysed in purely military terms: economic security must also be taken into account. There is considerable econometric research into the analysis of the economic foundations of ME (Table 1). The results obtained have not always been either coherent or generalizable. However that may be, current economic analysis is still failing to produce decisive results in this field. Military expenditure, an economic variable which is partly endogenous and partly exogenous, is limited by the available resources, although national security implies costs that are often determined by other than economic considerations (e.g. threat or probability of conflicts). Econometric methods established several economic, political and geographical determinants of

569

Ch. 19: Economics of Disarmament

military expenditure, which reduce the exogenous character of a disarmament process. Disarmament is not only a public decision and, without reference to an internal and external situation, it depends on economic and social variables which modify its content and its importance [Fontanel (1995)]. 3. Theoretical economic analysis of disarmament The history of economic thought on defense and development is very controversial. In a world of sovereign states the security of nations is both a priority and a constraint that economists have not always viewed in the same way. Disarmament has been analyzed as (i) an obstacle to economic and power development; (ii) a positive economic effect; (iii) a process which has some contradictions with economic systems such as socialism and capitalism and (iv) a complex process producing various economic results, depending on geographical, political and economic situations (Table 2). The history of economic thought is instructive in this context [Fontanel (1994, 1995)]. There are four main economic analyses of disarmament: the peace dividend approach, disarmament as an investment, development as the main factor of disarmament, and armament/disarmament as elements of a national strategy. Table 2 History of economic thought and disarmament Main analyses Disarmament as an obstacle to development Good relationship between armament and development. Armament is a stimulus for production, innovation and employment in capitalist countries. Economic development makes international antagonism inevitable. Laissez-faire leads to the domination of the strongest and military forces ensure the sovereignty of states threatened by less developed countries. Underdevelopment is a product of development of the great powers and disarmament is an element of game theory in international relations. The power of the Prince is the main economic objective. War is a solution to excessive population growth. The threat of war is essential in keeping anti-social trends under control. Disarmament as a positive economic effect Military expenditures are not productive. War and its preparation are explicit variables of the monopoly spirit. Laissez-faire produces development and development produces disarmament. Military expenditures reduce the competitiveness of a nation. Disarmament produces "peace dividends".

Authors

Plato Rae (1834), Mandel (1975), Attali (1978) Seligman (1915), Emmanuel (1969), Freyssinet (1966), List (1840), Veblen (1917), Brewer and Shubik (1979), J.S. Mill (1824) Mercantilism Malthus (1826), Bouthoul (1961) J.S. Mill (1824), Galbraith (1972) A. Smith (1776), J.B. Say (1803), Ricardo (1817) Melman (1974), Nef (1949) Continued on next page

570

J Fontanel Table 2, continued

Main analyses

Authors

Disarmament as an investment process. Military force is a means to deter agression, but it increases the likelihood of wars. A negotiated disarmament is needed to avoid wars.

UNIDIR (1993) Richardson (1960), Brito and Intriligator (1987)

Disarmament and economic systems Wars are conditioned by antagonist social relations. Disarmament is desirable only with socialism. Military investment is a catalyst of primitive accumulation and an instrument of colonial domination. The capitalist exploitation of the world and imperialism need an armament process. Military expenditure exerts a positive influence on profits, capitalist technology and demand for labor. It serves to absorb the economic surplus of monopoly capitalism. Disarmament is impossible with colonialism. Disarmament is impossible with imperialism.

Marx (1867), Engels (1878) Luxemburg (1913) Lenin (1917), Hilferding (1910) Baran and Sweezy (1966), Kidron (1970) G. Frank (1972) Boukharine (1915)

The complex relations between disarmament and development Military preparedness is necessary as a national insurance policy. Costs of defense are tolerable so long as combat does not disrupt the civilian sector.

Pigou (1921), Giffen (1872)

Military expenditure is a public good, which is not the best solution for a deficit spending policy. War is tending to disappear with the industrial class and the elimination of pauperism. Disarmament is an effect of development.

Keynes (1940), Samuelson (1964) Proudhon (1861), Fourier (1848), Saint-Simon (1819)

3.1. The peace dividend approach Disarmament is likely to contribute to peace and the reduction of military expenditure makes a contribution to the peace dividend. Resources released from defense become available for alternative use, such as social, health, education, environment or other economic uses [UN (1983), Klein (1990), Eisner (1993)]. The "peace dividend" must be a simple reallocation of public expenditure from defense to other budget headings, or a rather good substitutability between defense and other economic activities [UNIDIR (1984)]. With disarmament, there is the release and reallocation of resources from military to civilian uses, which is likely to have favorable effects on growth and international trade [UN (1989)]. There are two main myths surrounding the peace dividend [UNIDIR (1993)]: it is reputed to be large and immediate and to be able to solve the main social and economic

Ch. 19: Economics of Disarmament

571

problems of a nation. According to Kennedy (1988), military superpowers devote a large share of the nation's economic power to unproductive military expenditure and they take the risk of eroding their national economic competitiveness. The economic effects of disarmament are transmitted through four main channels, namely the possible increase of investment [Smith (1980)]; substitution effects for the work force [Richards (1991)]; reduction in the application of a military technology that has less respect for the economic rules of profitability [UN (1989)]; and limitation of public effective demand [Melman (1974)]. The positive effects of disarmament have often been high-lighted but it does not necessarily produce the cumulative effects that are a feature of economic development. Besides, it is paradoxical to note that economists have sometimes explained the growth of Germany and Japan between 1930 and 1940 by investment in armaments while explaining the economic miracle of the same two countries since 1945 by the weakness of their military expenditure. It has to be conceded that there are far more essential arguments to account for these two periods. According to Deger (1986), a percentage point reduction in military expenditure increases economic growth by a third of a percentage point and completely eliminates the external capital requirements of the least advanced countries. The competitiveness of the economy is increased and the deficit on the trade balance is reduced. There is evidence that some exporting activities are impoverishing, notably when payment conditions become difficult or even impossible, and when the largely subsidized arms industry is persuaded to sell at a loss to the community [Smith, Humm and Fontanel (1985), Chesnais (1990)]. But expectations of a peace dividend must be challenged, because econometric analyses seem relatively incapable of providing indisputable information on the existence of the negative or positive relationship under consideration. Moreover, the peace dividend alone is not able to solve all the world's problems, and major reallocations of resources involve costs and take time. 3.2. Disarmament as an investment Peace dividends are likely to be small in the short-term; they are mainly affected by the adjustment costs associated with the reallocation of resources from military to civilian uses in order to obtain long-term benefits. They have to solve the problem of unemployment and all investments are not successful. In some cases, society's rate of return from disarmament will be low or even negative [Intriligator (1994)]. In the opinion of Alexander (1990), the shift into non-defense products will generally be unsuccessful because of mismatches between defense and civilian experiences, skills and technologies, but primarily because the culture and management of the defense industry militate against civilian success. In the end, the task is not converting the defense industry from military to commercial markets but changing the industry. Diversification reduces the company's dependence on defense by acquisition or organic growth of non-defense operations. There are various kinds of diversifications, mainly in production (dual technologies or military specialization

572

J Fontanel

for market needs), in markets (exports and new buyers), in regions (relocation of installations and markets), in portfolios (financial operations involving the purchase and sale of companies) and in the negotiation of stable and lasting agreements with other companies. Innovative uses of military resources do not assure success in the new market. Complete adjustment may require the mobility of both labor and capital. Public policies are needed such as low-interest loans to industry or aid to R&D, exports, unemployment, retraining and regional development policies. Institutional culture is highly resistant to change. Good management of disarmament has positive long-term effects on civilian R&D, on the real productivity of national economies, and on confidence in international trading relations that cannot be quantified by econometric studies. According to the experts of UNIDIR (United Nations Institute for Disarmament Research) [UNIDIR (1993)], disarmament has major economic consequences involving costs as well as benefits, and its economic dividends are likely to be small in the short-term. But, in the long-term, disarmament can produce significant benefits through the production of civil goods and services as resources are reallocated to the civilian sector (the peace dividend). Disarmament is like an investment process involving shortterm costs and long-run benefits. Disarmament can be an worthwhile investment for economic development, ceteris paribus,if the countries feel that their national security is not threatened by the process [Hartley (1993)]. 3.3. Development as the main factor of disarmament Advocates of the New Economic Order see disarmament more as a consequence of development. For Russett (1982), the long term burden of US military spending helps to explain the decline of US productivity relative to Japan. The Cold War left a legacy of significant debt, both internal and external, as well as an undercapitalized and declining infrastructure, each due in part to the costs of maintaining the liberal structure of the international system [Goldstein (1988)]. The provision of public goods requires devoting considerable resources to military and foreign affairs resulting in inadequate investment [Kennedy (1988)]. The World Bank has adopted the position that high levels of ME are detrimental to industrialization, at least at the level of opportunity costs and now low levels of military outlays are required by the recipients of development aid. It established that for Japan, the economic miracle has been created, in large part, by avoiding high levels of ME/GDP and concentrating on national economic policies, but Japan obtained free US protection. Ward and Davis (1990) found evidence of a significant negative trade-off between ME and economic growth. The long-term impact of such significant trade-offs between defense effort and investment is thought to be reduced economic growth and national well-being. 3.4. Disarmament/armamentas elements of a nationalstrategy Colonial wars are historically seen as basic, although not exclusive causes of underdevelopment [Fontanel (1990)]. They promoted export products to the detriment of the

Ch. 19: Economics of Disarmament

573

crops that provided food for the population, and they had the effect of stimulating the mining industry and destroying local craft industries, the economy being developed in a disjointed way, entirely geared to the economy of the colonial power. These relationships remain unchanged. Major international economic negotiations provide reminders of the existence of relations of conflict and the importance of the status of a nuclear power. The threat of armed force is always a decisive factor in the spread of the dominant modes of production and in the maintenance of existing powers. Moreover, the major economic powers use military and non-military aid to maintain their political, economic, ideological and even moral influence. Western technology is strongly influenced by the militarization of the advanced economies, which is still dominant in contemporary economic development despite the examples of Germany and Japan, where the contrary holds. Under these conditions, disarmament definitely does modify relationships of strength, and this is also reflected in the economic sphere. Economic war fulfills all the classical functions of war. The economy has become an instrument of power often applied to relations of conflict between states [Fontanel (1994)]. War must be understood as the projection of national power through means that are political and economic as well as military [Wolfson and Farrell (1987)]. Insecurity is dependent on the arms race, on inequality, international domination, and even social exploitation. Indirect strategies of dissuasion, economic forms of retaliation, embargoes and boycotts are all powerful weapons with economic and political effects dependent on defense measures, international expressions of solidarity and the potential for substitution [Baldwin (1985)]. All the dimensions of social life have to be taken into consideration in international strategy. Direct (military) strategy being difficult by the strength of nuclear forces, states are led to follow a strategy of "indirect manoeuvreing". The idea is to sell only those products that would not improve the economic growth of the country receiving them more than that of the country selling them. In other words, unequal exchange is something to be demanded and sought. There are many differences of opinion as to the usefulness of these measures. Some authors are of the opinion that the grain embargo did not adversely affect the Soviet economy, but American governments have calculated differently. The defense of the industrial base undertaken by the Pentagon has proved to be extremely costly and has gradually eroded the competitiveness of the American economy. This policy was initially based on three essential principles: that the USA is an economic leader and can therefore control advanced technology; that the exports concerned are not overly important for the American economy and the cost of such controls would be relatively slight; lastly, that dual-purpose technologies are relatively few in number and can readily be isolated. There are many indirect strategies which can be pursued by nations: (a) Impoverishment resulting from the strain of preparing for war is often presented as one of the basic reasons for the collapse of the Soviet economy. An arms race reduces the development potential of states, but the poorest are the most rapidly impoverished, with the result that the might of the richest countries is increased [Brito and Intriligator (1987)].

574

.1Fontanel

(b) The strategy for effecting the break tries to create economic problems in the rival country so as to increase its political and social difficulties. An embargo is an instrument of deterrence capable of changing the behavior of a potential enemy, but it is a weapon that is dangerous to the user, because prolonged interruption of trade flows may lead to a permanent loss of outlets and illegal trading (black markets). The economic weapon enables a country to dominate another country through the exercise of the power conferred by the exercise of a monopoly over the supply of goods and services vital to the country's survival. The USA often uses food as a weapon against developing countries [Strange (1985)]. (c) The aim of a containment strategy is to develop ties of economic interdependence capable of ensuring peace; commercial ties had a moderating role on ideologies. The impact of this strategy cannot really be measured [Tinbergen (1993)]. For Kennedy (1988), the history of the rise and later fall of the leading countries in the Great Power system shows a very significant correlation over the longer term between production and revenue-raising capacities on the one hand and military strength on the other. Ward, Davis and Lofdahl (1994) found that after 1945, Japan's defense policy was shaped with a variety of non-military goals in mind. They demonstrate that Japanese defense spending was driven by macroeconomic goals as well as the desire for political and economic co-operation with the USA, and they identified two main effects: (i) Size effects convey the contribution of the overall spending levels to national product. In this way, government purchases may contribute to national productivity by increasing the volume of activity. (ii) Externalities through spill-over contributions. Government spending (an intermediate, not final demand) on education may contribute to the efficiency and productivity of non-governmental sectors, by an indirect effect upon the technical skills of the labor force. The defense-growth trade-offs in the USA and Japan change in character and dimension, sometimes rather quickly. Ward, Davis and Lofdahl (1994) built an economic model of the defense-growth trade-off. The estimating equation is: DY

I

Y_

Y_

bDL

(

dM

L d(l +M) DM DN + qM_ + qN- + e, S_,

)DM Y1

dN

) DN

+d

Y_-

S_1

where Y is national income, I is national investment, L is labor, M is military spending, N is non-military state spending, q is the externality and d is the factor productivity differential. Size effects convey the contribution of the spending levels to national product and the externalities are the spin-off contributions. The USA emerges as a society in which there has been a recent growing linkage between non-military spending and economic productivity. When the central government purchases defense goods, the

Ch. 19: Economics of Disarmament

575

volume effect appears to be negative, while the spin-offs are positive but declining. Government spending for non-defense programs does not contribute to economic growth. Disarmament is unlikely to bring about an aggregate downturn in the economy though local dislocations are already evident in some communities. This result suggests that disarmament will generate greater positive short-run size-effects on the aggregate economy for the USA, and not for Japan, but there is no evidence that higher ME is likely to be deleterious to the Japanese economy. For Japan, government spending does not contribute positively to the volume aspects of economic output in the way it does in USA and this is true for defense as well as non-defense government spending, but spin-off effects of ME are strong and growing for the Japanese economy. This finding contradicts Kennedy's hypotheses. The size impacts were negative over the entire period and the externalities have been positive and growing for about a century.

4. Empirical analysis of disarmament All the methods in general use in economic analysis may be applied to the economics of disarmament. No economic situation is ever repeated in exactly the same form and the experimental approach is practically impossible in economics. Under these conditions the methods of analysis most employed are deductive analysis based on postulates (e.g. see Table 2), empirical simulation models, evaluations of reduced forms, input-output models, computational general equilibrium models and historical studies (which can provide useful knowledge about past disarmament such as World Wars I and II, the Korean and the Vietnam Wars) and monographic studies (such as regional or sectoral case studies which are very useful for the understanding of firm behavior). The information from statistical methods [Atesoglu and Mueller(1990)] must be interpreted and there is often a confusion of simple correlations with causation. Econometric analysis in which chronological series of data are used lay the emphasis on short-term effects, since all the information considered relates to the recent past. Input-output analyses are mainly interested in the technological structures and interindustry relationships in the economy [Leontieff and Duchin (1983)]. They are very useful for the measure of the direct or indirect changes in output, employment or capital utilization induced by a change in the structure of the demand of final goods. These analyses require detailed data on the actual technological structures of military and civilian productions. The validity of a model cannot be restricted to the truth of the hypotheses, but must also be dependent on the capacity of the system as set out to predict the behavior of the real system, even if, in some instances, the cause and effect relationships or simultaneous relationships misrepresent or simplify the reality. Models have to be evaluated in terms of how well they communicate and the quality of the information created in relation to the objectives laid down at the time of construction. The economist has therefore to consider which variables should be accepted, the level of their aggregation, the relationships being tested and the basic aims of the study.

J Fontanel

576

Nevertheless, this methodology is not without its shortcomings, prominent among which are the dubious significance of the results obtained, equation misspecifications [Kirby and Hooper (1991)], the excessive generalization of the results [H6bert (1993)], the obvious risk that the relationships will be unstable and the low degree of relevance as a tool in economic decision-making [Dunne and Smith (1990a,b)]. Whereas a number of economic models have been constructed to analyze the impact of increased ME, there are few formal analyses of the economic effects of disarmament. But, in econometric modelling it is assumed that the response to an increase or decrease is symmetric; this may not be appropriate when considering a large reduction of ME. It is interesting to analyze some studies of the American, French and British economies. 4.1. The United States Paul Kennedy (1988) formulated the hypothesis that the decline of the American economy could be due to the high level of ME. Under those conditions, peace dividends from disarmament could appear as much in favor of social programmes as of national productivity. The reduction of defense expenditure did further the competitiveness of enterprises, since the technological spin-off from the military sector was considered to be inefficient and costly [Melman (1974)]. With the available scientific tools, econometric analysis does not provide irrefutable proof of a negative relationship between capital formation and ME, and the productivity problems of American industry cannot be ascribed to the defense effort alone [Gold and Adams (1990)]. Kinsella (1990) indicated by the use of an autoregression vector that no substantial causal relationship existed between ME and prices, unemployment and the interest rate. A direct positive relationship (without a time lag and causal relationship) did exist between ME and national output, but only for monthly or quarterly information, notably when the economy was in crisis and thus was under the effect of the Keynesian multiplier. Huang and Mintz (1990) confirmed this analysis. Atesoglu and Mueller (1990) showed that there was a positive relationship between ME and economic growth but that the multiplier was very small. They used for each regression equation the Cochran-Orcutt iterative procedure assuming first order autocorrelation of the disturbance terms since the ordinary least squares estimates indicated the possibility of autocorrelation. With the growth in real input GQR, the investment ratio (of real gross fixed investment to the real GNP) IQ, the growth in employment LRG, the growth in real defense spending DGR, then: QGR=-0.078+0.551IQ +0.965LGR+0.062.DGR (1.59) (1.75) (3.35) (2.47)

with

u=0.399 (2.47)

(2)

period: 1949-1989, R 2 =0.720, SE=0.015, DW=2.058. A change in defense spending has positive and significant effects on the growth rate of the economy and both defense effects (the change in the rate of defense spending and the relative size of the defense sector) are individually significant. A study by

Ch. 19: Economics of Disarmament

577

Ward and Davis (1990) revealed the existence of contradictory effects, the result of which was dependent both on the magnitude of defense expenditure in the US economy and on the importance of the military effort in the US budget. The relative weight of ME in the national economy is a burden for the economy, whereas public non-ME tends to have a positive effect. It must be pointed out that the state often has to be responsible for activities (justice, freedom, democracy) that are not very productive but are necessary for society as a whole, and that cannot be handed over to the private sector for economic considerations of profitability or the principle of public service. Ward and Davis concluded that the global effect of American ME during the period 1952-1988 had been very slightly positive, but that its contribution to national productivity had declined, in contrast to that of governmental civilian expenditure. Furthermore, Aschauer (1989) was of the opinion that whereas expenditure on the state infrastructure had very positive effects on growth, the same could not be said of military capital expenditure. Klein (1990, 1993) thought that the effect of disarmament on the American economy was bound to be positive. Because there is a crowding-out effect between ME and investment, the production of civilian capital goods will be very useful when it takes the place of military production, because the former creates new income flows, whereas the latter is used to destroy without any economic return. Thus, a motorway will still have utility far beyond the expenditure period and will further the expression of other utilities. With disarmament, the government ought simultaneously to pursue a policy of compensatory public expenditure or a reduction of the budget deficit. In the medium term, lower interest rates will lead to more investment (for the LINK model a 3 percent reduction in ME, accompanied by a more flexible monetary policy would produce a reduction of 2 percent in interest rates), and that situation will favor both international trade and a reduction of the public deficit. For Olszewski, Payne and Ross (1993), there is a cointegrating relationship between real MEs, real output and the real dollar-mark exchange rate. ME cannot significantly affect the real dollar exchange rate value in the long run. For Eisner (1993), ME exerts undoubted inflationary pressures, because the economy provides salaries without a flow of goods being produced in exchange. The result of disarmament should therefore be an increased propensity for investment. By using two time periods and two macroeconomic and industrial models, Thomas, Stekler and Glass (1991) concluded that reducing the level of US defense expenditure will reduce real output, price level, and employment, with an attenuation of the effects after five years. With a constant reduction, changing the composition of the cuts had a small but perceptible effect. For Hormats (1992), the so-called "peace dividend", which is not real money, will be insufficient by itself to cure the structural ills afflicting the US economy (budget and commercial deficits, insufficient investment, inadequate national savings and deficient training methods). It is preferable to create incentives for savings, investment and education, and to invest a few thousand million dollars more to help former socialist nations to consolidate democratic and market reforms.

578

J Fontanel

4.2. France A very simple model focusing on chronological series (1950-1980) showed that progressive disarmament had a weak positive effect on the GDP of France [Fontanel (1980)]. This impact was due mainly to a reduction of the inflationary tensions to which the armament effort gives rise and to the appreciable increase in national investment. Above all, this analysis took into consideration the dual nature of ME: as public expenditure it exerts important effects on global demand, capable of exerting Keynesian multiplier effects; as ME in the strict sense it must be analyzed in comparison with other forms of public and private expenditure. Under these conditions ME has quite high opportunity costs, notably because it reduces investment. Disarmament improves economic growth very quickly, because it reduces inflation and competitiveness. With a model of the input-output type, Aben (1981) analyzed the effects of a 40 percent reduction of ME on the main sectors of the economy and on employment. It concluded that a reduction without a substitution of civilian expenditure for ME led to an appreciable decline in national production (34 percent for the armaments sector, shipbuilding and aeronautics) and a loss of 400 000 jobs; when military expenditure is substituted for public civilian expenditure, none of the possibilities for the conversion of defense activity made it possible to avoid a worsening of the employment situation in the short term (a loss of some 140 000 to 213 000 jobs). However, very few economic activities subjected to such a reduction of their outputs would emerge with such a satisfactory result. More recently, Aben and Daures (1993) used an input-output model to describe the economic consequences of a transfer of the budget from the defense sector to education. Based on a Parliamentary hypothesis of growth, their simulations confirmed, on strict economic grounds, that it is better to have chalk rather than guns, except if chalk really means education equipment. 4.3. United Kingdom Dunne and Smith (1984) are of the opinion that disarmament is more of an opportunity than an economic problem. Using the Cambridge Growth Project model, they made several simulations of the economic impact of disarmament in Great Britain. With compensation, disarmament produced a slight improvement in the balance of payments, a limitation of inflationary pressures, the creation of new jobs connected with the labor-intensive nature of the defense expenditure relative to the civilian governmental expenditure, and additional growth. Without compensation, there are reductions in GDP, a fall of the price index, and a worsening of the jobs situation. This situation was brought about more by the deflationary policy than by the actual reduction of ME. The structural changes were, however, mainly of a microeconomic and sectoral order and highly heterogeneous. These results are in line with the historical experience of the United Kingdom [Smith and Smith (1983)] and with studies that stress the effects of substitution of military R&D to the detriment of civilian products

Ch. 19: Economics of Disarmament

579

Table 3 Cuts in UK defense spending: simulation results with and without compensation Variables

1993 with

Consumers' expenditure Government consumption Investment Exports (G&S) Imports GDP at factor cost Balance of payments Unemployment

0.22 -0.21 0.25 -0.02 0.09 0.10 -0.04 -0.05

1996 without -0.31 -1.87 -0.3 -0.01 -0.50 -0.45 0.16 0.06

with 1.73 -1.31 1.58 -0.25 0.99 0.63 -0.40 -0.22

2000 without -1.97 -6.99 -1.77 0.06 -2.34 -2.03 0.93 0.25

with 5.39 -2.79 4.27 -0.79 3.21 1.84 -1.36 -0.52

without -4.41 -12.13 -3.64 0.69 -4.39 -3.64 2.17 0.46

Percentage difference from base. Source: Barker, Dunne and Smith (1991).

a

[Hartley (1993)], poor use of military technology [Maddock (1983)], and the crowdingout effects of investment [Smith (1980)]. Barker, Dunne and Smith (1991) used a multi-sectoral macro-economic model to evaluate the impact of cutting UK defense spending. The scenario under consideration was one of a reduction in ME, with real cuts of 8.5% a year (starting in 1993), with UK ME reduced to around 2% of GDP by the year 2000. It assumes that the cuts fall broadly proportionately on the major components and that the current balance between personnel, equipment and other spending is maintained. Without compensation, disarmament produces adverse economic results at the beginning; but a reduction in ME to one-half of its 1992 level by the year 2000, with the released "peace dividend" allocated to other government current and capital expenditures, would lead to a reduction in unemployment of 0.52 million and an increase in GDP factor costs above base by 1.84% (Table 3). The transmission mechanism is thus likely to depend on the macro-economic policy adopted. Questions arise as to how UK companies are responding to present reductions in ME and how they are likely to respond to future changes. There is a new, more commercial procurement regime, with increased competition, the use of fixed price contracts and a greater willingness to import. The consequence has been massive restructuring in the UK defense industries and a prompt diversification. Such problems of transition may not be as difficult as in the past, as the restructuring of the UK armaments industry is already taking place because of the present reduction in defense spending. 4.4. Simulations for developing countries For Linden (1992), a two-sector growth model is derived to indicate the effects of the military burden and government expenditure on the growth of GNP in selected Middle Eastern countries (1974-1985). A four equations simultaneous model (growth rate,

580

J Fontanel

defense burden, skilled labor rate and investment rate) of Gyimah-Brempong (1989) is used to investigate the relations between defense outlays and economic growth and their channels in less-developed countries (LDCs). Defense burdens affect economic growth directly through providing an increased supply of skilled labor and decreased investment. For African countries, there is a trade-off between a high defense burden and economic growth. With a simple neoclassical formulation of economic growth to account for the role of military exports in economic productivity, Ward (1990) showed that for India, only the military aspects of government spending appear to generate short-term economic productivity gains. Since these are a small part of the government budget, the overall impact of government spending is likely to be slightly negative. There are negative externalities for military spending when examined separately from its marginal productivity for civilian production and separated from size effects of the overall governmental budget. The relative marginal productivity of military and nonmilitary spending in both Brazil and India is also negative. There is strong crosssectional evidence that increased defense expenditures actually serve to retard the rate of resource creation, compounding the negative effect of defense expenditures on growth and development [Deger and Smith (1983)]. The potential benefits of security expenditure that lie in the possibility of modernization are indisputable, but the negative impact on savings and investment that seem to result from higher levels of defense spending are such that an alternative means of acquiring the modernizing results could be more beneficial to the economy. The conclusion is that military spending has negative consequences for economic growth when compared to alternative potential allocations. The application of simulations based on the reduction of ME involves the hypothesis that the effects on the national economy of an increase or a decrease in the economic effort of defense will be inverse but similar. Now, it is probable that threshold effects will emerge to render such a generalization questionable.

5. Disarmament for development Disarmament and development are two essential objectives of our time and the problem is to know whether it is desirable that they should be linked. French governments have frequently called for the establishment of an international fund of disarmament for development that would enable the amounts saved on arms to be used to develop productive activities that would speed up economic development and aid arrangements to the developing countries [Fontanel (1986), UNIDIR (1987)]. 5.1. The nature of the transfers International aid effects are interesting to analyze. The question is differently analyzed when the military effort is accompanied by assistance from a developed country. Aid

Ch. 19:

Economics of Disarmament

581

linked to disarmament should normally have the same effects as existing traditional forms of aid, except perhaps at the level of the size of transfer, which may exceed the threshold above which the solidarity accorded has significant economic effects. It should, however, be noted that the confiscation of the savings by a social grouping that decides to devote it to unproductive uses such as exports of capital or imports of luxury goods, is not favorable to economic development. A transfer makes economic sense only if it finds expression in a highly productive activity and not by aggravating inflationary trends owing to the appearance of demonstration effects. A resource transfer may also be the occasion of new markets for the developed countries and new dependence for the poor countries [Lebovic (1988)]. The developing countries must therefore avoid the transfers being impoverishing, when it reintroduces unequal trade flows or transfers in kind which competes directly with the national activities of the recipient and which bears the stamp of a culture and the dominant values of the society in which it was created. Aid from developed countries may lead to distorted development. Some forms of transfers prove ultimately to be costly, notably in operating costs, and political constraints are rarely absent from such an undertaking by rich countries. Aid to developing countries may be devoted to prestige expenditure and to the increasing of social inequalities. The main aim of disarmament for development must be to narrow the disparities in growth. 5.2. The economic effect of a disarmamentfor development process Reduction of ME has been simulated in several models of the world economy built for international organizations [UN, UNITAD (UN Institute for Trade and Development), OECD and IMF) or for OECD countries with the three-equation model of Cappelen, Gleditsch and Bjerkholt (1984). For the UN model of Leontieff and Duchin (1983), the transfer of resources to poor countries furthers their economic development. Ongoing disarmament would have a positive effect for all the regions of the world and the transfers of resources would appreciably increase consumption and the per capita GDP of the arid countries of Africa, and the low-income countries of Asia and tropical Africa. However spectacular these results may seem to be for the sparsely populated underdeveloped countries, they are not very significant statistically. Thus, an annual reduction of ME by 1.2 percent between 1980 and the year 2000 would further the economic growth of almost all countries: 1 percent for Japan, 1.5 percent for North America, 3 percent for Europe, 10 percent for the planned-economy countries of Asia, 20 percent for the low-income Asian countries and tropical Africa, over a period of 20 years. It may be noted that aid for development appreciably improves the situation of the poorest countries. The econometric results obtained do not seem to be very significant. They give only a very global idea of the economic impact of disarmament. The model of Leontieff and Duchin (1983) is not very adequate as an indicator of the ways in which the groups of countries under consideration were growing in 1990. For example, the excellent results foreseen for the planned-economy countries are very wide of the mark, because the formalized system

582

J Fontanel

had to use optimistic figures (based on bureaucratic planned prices) supplied by the socialist governments, and it was scarcely possible to analyze the latent political and economic crisis of the Eastern countries on the basis of the raw official data. The UNITAD world model simultaneously examines an initial disarmament scenario which is not accompanied by an international redistribution of the amounts saved, while a second scenario studies a policy based on the satisfaction of basic needs. Whereas the reduction of ME depresses demand and causes an increase in unemployment in the short term, its spin-off is expressed in the longer term in the growth of the developing countries by an increase in demand for the most developed regions of LDCs, mainly when the policy is based on basic needs. These effects will be strengthened by measures for the easing of trade and currency restrictions, but the gains may rapidly become losses if there is an increase either in interest rates or in the amount of the annual repayments of the debt burden of LDCs. Disarmament may relax some economic constraints but it cannot resolve all the problems of the developing countries. Even on the most satisfactory hypothesis, the real per capita consumption in Africa south of the Sahara will continue to worsen because the population is growing at too fast a rate, and because of the endemic weakness of agricultural production. The LINK model [Klein (1990)] shows that a 10 percent reduction of ME accompanied by an improvement in international assistance (0.7 percent of the GNP of the developed countries) is capable, in the best case, of a transfer relating exclusively to capital goods, of leading to a 1.7 percent growth of the GNP in the developing countries and an 0.2 percent growth in the developed countries. For this transfer to be positive for all parties (donors and recipients), 60 percent of the aid must be spent on capital goods [Klein (1990)]. The MULTIMOD model of the IMF (1993) considered the long term benefits of a 20% homogenous reduction of ME, international arms transfers and military aid. It calculates these benefits, for the long run, to be more than 10 000 billions of constant (1992) US dollars. For the first year, there is a decline of 6 billion US dollars of the GDP of developed countries, but during the second year, with the reductions of interest rates and public 'outlays and the increases of private consumption and investment, the effects will become positive. For the eleventh year, there will be 60 billion US dollars of benefits for developed countries and 11.5 billion US dollars for LDCs. The indebted countries will obtain the best results with the interest rate reduction and the increased exports. This model is interesting and positive, but it does not introduce economic policies, the effects of disarmament on regions, sectors and enterprises or the conversion problems. The model of Cappelen, Gleditsch and Bjerkholt (1984) developed a theoretical model, consisting of three equations, which was tested empirically for seventeen OECD countries in the period 1960-1981. This study suggests a negative relationship between economic growth and defense spending and does several simulations of the world model. Without transfers, the developed countries are the main beneficiaries of ongoing disarmament. With significant international aid the gap between rich and poor countries tends to narrow. Maneval, Rautsola, Wiegert (1991) tested a slightly modified

Ch. 19: Economics of Disarmament

583

model with a different kind of estimation for the four main neutral European nations, which appeared to be rather homogenous in regard of this study (populations, standard of living, annual per capita incomes, politically stable). They used the OLS-estimation for structural equations (unmodified single equations and additionally the best-fitting lag-versions). If GDP is the growth rate of gross domestic product; MANU, the growth rate of manufacturing output; MIL%, defense spending relative to GDP; INV%, investments relative to GDP; u, the stochastic terms in the individual equations; X%, exports relative to GDP; and DGC%, defense spending relative to government consumption, then: GDPt = ao + alMANUt + a 2INV%t + a3MIL%t + ult,

(3)

MANUt = bo + blINV%t +b2 X%t +b3 MIL%t + U2t,

(4)

INV% = c0 + cl GDPt + c2 DGC%t + c3 MIL%t + U3t.

(5)

The main finding was that defense spending had a positive effect on manufacturing output, but a negative effect on investment; also that the net effect on economic growth was, when significant, negative for the whole sample, but it varied from country to country. Disarmament cannot be confined to quantitative choices on the limitation of ME. It calls for a range of decisions on political priorities, the international economic order, the nature of development, the rate and direction of technological progress that could be developed in a less militarized society, the management of the natural environment, and the reallocation and redistribution of economic resources. Disarmament that failed to take into account the insecurity constituted by economic arms would not be very durable. Furthermore, foreign aid is often part of a package with military aid. Disarmament must take account of economic weapons.

6. The role of public policy Conversion is defined broadly as a new allocation of resources from the military sector to civilian uses. It is the process of deploying the resources released by military activities in favor of civilian sectors, in a continuous macro-economic process depending on the proper functioning of labor and capital markets. Technological conversion could take place as an acceleration of the spin-off mechanisms, but a British report concluded that less than 20 percent of the UK military R&D had civil spin-offs [Hartley (1993)]. Defense firms have difficulty in transition because their "core competencies" are derived from defense production and are unsuited to civilian activities [Cronberg and Hansen (1992)]. The military-industrial complex requires a dichotomy between civil and military production (length of the product cycles, importance of weapons performance, radical innovations and planned obsolescence). The economic policy of the government needs to help the enterprises to solve these problems.

584

J Fontanel

Defense managers have a lack of skills in the civil sector (e.g. the importance of effective advertising campaigns, the public acceptance of the new product line and the prices for the penetration of a new market), because in the military market they have mainly to know the administrative rules, to develop good working relationships with key procurement government personnel, and to lobby politicians [Dumas (1982)]. Military enterprises respond to disarmament in six main ways: (i) by developing weapons exports, (ii) by the restructuring of companies, implying a reduction in production volume by laying-off workers and selling or closing plants, (iii) by developing dual technologies, (iv) by specialization, narrowing the range of products manufactured and identifying niches with high profit potential, (v) by diversification, with the production of civil and military products, (vi) by complete conversion, based on a transfer of activity away from the military to the civilian sector. There are a variety of policy options, which are summarized in Table 4. Table 4 Policy options Type of policy

Examples

Manpower policy

Training Retraining of managers and workers Job information Labor mobility Early retirement Retooling old plants and equipments Investing in new plants and equipments Producing new consumer goods Use of scientists and engineers New civil research and development programmes, e.g. energy; environment; space exploration Location of industry policy Building airports, roads and expanding telecommunications Subsidies to civil research and development Subsidies to labor and/or capital Government contracts for civil goods Aiming to assist the conversion of defense plants to civil markets Using government expenditure to avoid recessions Support for exports and import saving Aimed at compensating the losers from disarmament: e.g., unemployment pay and redundancy pay (social safety net) Role for international agencies in disseminating information and experience on adjustment

Capital policy

Science and technology policy

National regional policy Social infrastructure policy Industry policy

State conversion agency Aggregate demand policy International trade policy Income deficiency payments International action

a

Source: UNIDIR (1993).

Ch. 19: Economics of Disarmament

585

There are some economic principles for disarmament which are important to follow [UNIDIR (1993), Fontanel (1993)]. These principles are: (i) reductions of ME should be gradual and predictable, allowing for smooth economic and social adjustments; (ii) a distinction must be made between stock conversion and expenditure flow diversion; (iii) public policies must assist change and resource allocation in order to minimize the costs of disarmament; (iv) joint conversion project should be an important aspect of international economic co-operation; the management of conversion is difficult and sometimes it is better simply to abandon specialist defense plants; (v) disarmament requires control of military technology, especially in military research and development, and control of arms transfers; (vi) industrialized countries might use some of the benefits of disarmament to assist developing countries. 7. Conclusion The economics of disarmament is a new discipline. Much work, involving multidisciplinary analysis, is needed to explain the causes of war, the arms race and disarmament models (arms limitations, disarmament, conversion); the definition and the intertemporal and international comparisons of military expenditure; the analysis of the determinants of military expenditure; the economic effects of the arms trade and of international aid; the economical and political role of military alliances (international public goods); the micro-economics of demand and supply (defense markets, public procurement, defense industries and regions, research and development, military manpower, the employment in defense industries); and the cost of civil defense. Questions arise concerning the costs and policies of conversion, the dual use of military products, the substitutability of military and civilian research and development, the effects of disarmament on capital capacity, the systemic and military transitions for the former Soviet Union, the relevance of the various models for the economic analysis of disarmament, the costs of wars, the economic effects of arms exports on growth, employment or technology. This agenda is not exhaustive of the variety of research questions on the economic aspects of disarmament. However, economists have to remember that the main dividend of disarmament is peace itself and that disarmament without development is not synonymous to durable peace [Fontanel and Ward (1993)]. References Aben, J., 1981, Desarmement, activity et emploi, Defence Nationale (Mai). Aben, J., and N. Daures, 1993, Chalks vs guns: some economic consequences of an announced French scenario, Defence Economics 4, 353-364.

586

J Fontanel

Adams, G., and D. Gold, 1987, The economics of military spending. Is the military dollar really different? in: C. Schmidt and F. Blackaby, eds., Peace, defence and economic analysis (Macmillan Press, London) 266-300. Alexander, W.R.J., 1990, The impact of defence spending on economic growth: A multi-sectoral approach to defence spending and economic growth with evidence from developed countries, Defence Economics 2, 39-55. Aschauer, D., 1989, Is public expenditure productive?, Journal of Monetary Economics 23, 177-200. Atesoglu, H.S., and M.J. Mueller, 1990, Defence spending and economic growth, Defence Economics 2, 19-27. Attali, J., 1978, La nouvelle conomie franqaise (Flammarion, Paris). Baldwin, D.A., 1985, Economic statescraft (Princeton University, Princeton, NJ). Baran, P., and P.M. Sweezy, 1966, Monopoly capital (Penguin Books, London). Barker, T., J.P. Dunne and R. Smith, 1991, Measuring the peace dividend in the United Kingdom, Journal of Peace Research 28, 345-358. Bischak, G., and M. Oden, 1989, The INF Treaty and the United States' experience: The industrial, economic and employment impacts, disarmament and employment programme, Working Paper no. 11 (ILO, Geneva). Boukharine, N., 1915, Economic mondiale et imp6rialisme (Moscou, Anthropos, Paris, 1971). Bouthoul, G., 1961, Sauver la guerre (Grasset, Paris). Brewer, G., and M. Shubik, 1979, The war game: A critique of military problem solving (Harvard University Press, Harvard, MA). Brito, D.L., and M.D. Intriligator, 1987, Arms race and the outbreak of war: application of principal-agent relationships and asymmetric information, in: C. Schmidt and E Blackaby, eds., Peace, defence and economic analysis (Macmillan Press, London) 104-120. Byers, J.D., and D.A. Peel, 1989, The determinants of arms expenditures of NATO and the Warsaw Pact: Some further evidence, Journal of Peace Research 26, 69-77. Cappelen, A., N.P. Gleditsch and O. Bjerkholt, 1984, Military spending and economic growth in the OECD countries, Journal of Peace Research 4, 361 373. Chesnais, F., 1990, Comp6titivit6 internationale et d6penses militaires (CPE Economica, Paris). Correa, H., and Ji-Won Kim, 1992, A causal analysis of the defense expenditures of the USA and the USSR, Journal of Peace Research 29, 161-174. Cronberg, T., and K.H. Hansen, 1992, From military to civil production (Tekster om Technologivurdering, Lyngby). Deger, S., 1986, Military Expenditure in Third World countries. The economic effects (International Library of Economics, Routledge and Kegan Paul, London). Deger, S., and R. Smith, 1983, Military expenditure and growth in less developed countries, Journal of Conflict Resolution 27, 335-354. Dumas, L.J., 1982, The political economy of arms reduction (Westview, Boulder, CO). Dunne, J.P., and R. Smith, 1984, The economic consequences of reduced UK military expenditure, Cambridge Journal of Economics 8, 297-310. Dunne, J.P., and R. Smith, 1990a, The peace dividend and the UK economy, Cambridge Econometrics, Report 1 (Spring). Dunne, J.P., and R. Smith, 1990b, Military expenditure and unemployment in the OECD, Defence Economics 1, 57-73. Eisner, R., 1993, Quelques considerations macro6conomiques sur le d6sarmement, Cahiers de l'Espace Europe de Grenoble 4, 53-71. Emmanuel, A., 1969, L'6change in6gal (Maspero, Paris). Engels, E, 1878, Anti-Diirhing (International Publishers, New York, 1971). Fontanel, J., 1980, Military expenditure and economic growth (France, Morocco), Rapport pour le Groupe d'experts des Nations Unies sur le D6sarmement pour le D6veloppement (CEDSI, Grenoble).

Ch. 19: Economics of Disarmament

587

Fontanel, J., 1986, The international disarmament fund for development, Disarmament 9 (United Nations, New York) 21-28. Fontanel, J., 1990, The economic effects of military expenditure in third-world countries, Journal of Peace Research 27, 461-466. Fontanel, J., 1991, Effects for developing countries of the East-West disarmament process, in: Serge Sur, ed., Disarmament agreements and negotiations: the economic dimension (UNIDIR, Aldershot, Dartmouth). Fontanel, J., ed., 1993, Economistes de la paix (Presses Universitaires de Grenoble, ECO+, Grenoble). Fontanel, J., 1994, The economics of disarmament: A survey, Defence and Peace Economics 5, 87-120. Fontanel, J., 1995, Les dpenses militaires et le dsarmement (Publisud, Paris). Fontanel, J., and R. Smith, 1990, The impact of strategy and measurement on models of French military expenditure, Defence Economics 4, 261-273. Fontanel, J., and M.D. Ward, 1993, Military expenditures, armament and disarmament, Defence Economics 4, 63-78. Fourier, C., 1848, Le nouveau monde industriel et soci6taire, Les Oeuvres completes (Librairies socidtaires, 1929, Paris). Frank, G., 1972, Le dveloppement du sous-ddveloppement (Maspero, Paris). Freyssinet, J., 1966, Le concept de sous-ddveloppement (Editions Mouton, Paris). Galbraith, J.K., 1972, The new industrial state (Hamish Hamilton, London). Giffen, 1872, The cost of the Franco-German war, reprinted in Economic Inquiries and Studies (Bell, London). Gold, D., and G. Adams, 1990, Defence spending and the American economy, Defence Economics 1, 275-293. Goldstein, J., 1988, Long cycles prosperity and war in the modem age (Yale University Press, New Haven, CN). Goodwin, C.D., 1991, Economics and national security, history of political economy 23, Annual supplement (Duke University Press, Durham). Griffin, L.J., M. Wallace and J. Devine, 1982, The political economy of military spending: Evidence from the United States, Cambridge Journal of Economics 6. Grindle, M.S., 1986, The politics of military budgets in Latin America (International colloquium on Defence, Security and Development, Birkbeck College, London). Gyimah-Brempong, K., 1989, Defense spending and economic growth in Sub-saharan Africa: An econometric investigation, Journal of Peace Research 26, 79-90. Gyimah-Brempong, K., 1992, Do African governments favor defense in budgeting?, Journal of Peace Research 29, 191-206. Hartley, K., 1993, Aspects conomiques du dsarmement, in: J. Fontanel, ed., Economistes de la Paix (ECO+, Presses Universitaires de Grenoble, Grenoble) 119-132. Hebert, J.P., 1993, Conversion of military R&D: the French case, in: Pugwash project on conversion of military R&D (Oxford Press, London). Hewitt, D.P., 1991, Military expenditure. Econometric testing of economic and political influences (Fiscal Affairs Department, International Monetary Fund, May). Hilferding, R., 1910, Finance capital (Routledge and Keegan Paul, London, 1971). Hormats, R.D., 1992, Priority uses for the peace dividend (Washington Economics Reports, United States Information Agency, March 4). Huang, C., and A. Mintz, 1990, Ridge regression analysis of the defence-growth trade-off in the US, Defence Economics 2, 29-37. IMF (International Monetary Fund), 1993, World economic outlook (International Monetary Fund, Washington DC, Annex, September). Intriligator, M.D., 1994, Economic aspects of disarmament: Arms race and arms control issues, Defence and Peace Economics 5, 121-129.

588

J Fontanel

Kende, I., 1980, Local wars 1945-76, in: E. Eide and M. Thee, eds., Problems of contemporary militarism (Croom Helm, London). Kennedy, P., 1988, The rise and fall of the Great Powers (Random House, New York). Keynes, J.M., 1940, How to pay for the war? (Harcourt Brace Jovanovich, New York). Kidron, M., 1970, Western capitalism since the war (Penguin Press, London). Kinsella, D., 1990, Defence spending and economic performance in the United States: A causal analysis, Defence Economics 1, 295-310. Kirby, S., and N. Hooper, eds., 1991, The cost of peace: Assessing Europe's security options (Harwood Academic Publishers, Chur). Klein, L.R., 1990, The economics of turning swords into plowshares, Challenge, March April, 1-5. Klein, L.R., 1993, D6veloppement et d6sarmement. La signification in: J. Fontanel, ed., Economistes de la paix (ECO+, Presses Universitaires de Grenoble, Grenoble) 81-88. Lebovic, J.H., 1988, National interests and United States foreign aid, Journal of Peace Research 25, 115-136. Lenin, 1917, Imperialism, the highest stage of capitalism, reprinted in 1977, Lenin, Selected Works (Progess Publishers, Moscow). Leontieff, W., and F. Duchin, 1983, Military spending: Facts and figures, worldwide implications and future outlook (Oxford University Press, London). Linden, M., 1992, Military expenditure, government size and economic growth in the Middle East in the period 1973-85, Journal of Peace Research 29, 265-270. List, F., 1840, Le systeme national d'6conomie politique (Guillaumin, 2nd edition, Paris 1857). Looney, R.E., 1988, Third-world military expenditure and arms production (Macmillan Press, London). Looney, R.E., 1989, Internal and external factors in effecting Third World military expenditures, Journal of Peace Research 26, 33-46. Lotz, J.R., 1970, Patterns of government spending in developing countries, The Manchester School 36, no. 2. Luxemburg, R., 1913, The accumulation of capital (Routledge and Keegan Paul, London, 1951). Maddock, I., 1983, Civil exploitation of defence technology (Ministry of Defence, London). Maizels, A., and M.K. Nissanke, 1987, The causes of military expenditures in developing countries, in: S. Deger and R. West, eds., Defence, security and development (Frances Pinter, London) 129-139. Malthus, R., 1826, An essay on the principle of population, in: The works of Robert Malthus, Vol.2 (William Pickering, London, 1986). Mandel, E., 1975, Late capitalism (New left book). Maneval, H., P. Rautsola and R. Wiegert, 1991, Military spending and economic growth: A comment on Cappelen, Gleditsch and Bjerkholt, Journal of Peace Research 28, 425-430. Marx, K., 1867, Capital (Lawrence and Wishart, eds., London, 1970). Melman, S., 1974, The permanent war economy: American capitalism in decline (Simon and Schuster, New York). Mill, J.S., 1824, War expenditure (Westminster Review 2, London). Nef, U., 1949, La route vers la guerre totale (Armand Colin, Paris). Nincic, M., and TR. Cusack, 1979, The political economy of the U.S. military spending, Journal of Peace Research 16, 101 115. Olszewski, E.A., J.E. Payne and K.L. Ross, 1993, U.S. military expenditure and the dollar: Another look, Defence Economics 4, 249-258. Pigou, A.C., 1921, The political economy of war (Macmillan, London). Proudhon, J.J., 1861, La guerre et la paix (Correspondance, Dalloz, Paris, 1953). Rae, J., 1834, Statement of some new principles on the subject of political economy (R. Warren James, ed., University of Toronto Press, 1965). Rajmaira, S., and M.D. Ward, 1991, Reciprocity and evolving norms in the Reagan and Gorbachev eras (Presented at the 1990 Annual Meeting of the American Political Science Association, San Francisco, California).

Ch. 19:

Economics of Disarmament

589

Renner, M., 1994, Budgeting for disarmament. The costs of war and peace, Worldwatch Paper 122, 1-70. Ricardo, D., 1817, On the principles of political economy and taxation (edited in The Works and Correspondence of David Ricardo, Cambridge University Press, Cambridge, 1951). Richards, P.J., 1991, Disarmament and employment, Defence Economics 2, 295-312. Richardson, L.F., 1960, Arms and insecurity (The Boxwood Press, Pittsburgh, PA). Russett, B.M., 1982, Defence expenditures and national well-being, American Political Science Review 76, 767-777. Saint-Simon, 1819, Essai sur la politique (Oeuvres completes, Anthropos, Paris 1966). Samuelson, P., 1964, Economics: An introductory analysis (McGraw-Hill, New York). Sandler, T., and K. Hartley, 1995, Economics of defense, Surveys of Economic Literature, Cambridge University Press, Cambridge, 260-283. Say, J.B., 1803, Trait6 d'6conomie politique (Calmann Levy, Paris, 1972). Schelling, T.C., 1966, Arms and influence (Yale University Press, New Haven, CT). Seligman, E.R., 1915, An economic interpretation of the war (Reprinted in Essays in Economics, Macmillan, New York, 1925). Smith, A., 1776, An inquiry into the nature and causes of the wealth of nations (Reedited in 1976 by R.H. Campbell and A.S. Skinner, Clarendon Press, Oxford). Smith, D., and R. Smith, 1983, The economics of militarism (Pluto Press, London). Smith, R., 1980, Military expenditure and investment in OECD countries 1954-1973, Journal of Comparative Economics 4, 19-32. Smith, R., A. Humm and J. Fontanel, 1985, The economics of exporting arms, Journal of Peace Research 22, 239-247. Strange, S., 1985, Protectionism and the world politics, International Organization 39, 233-259. Thee, M., 1982, Third world armaments: Structure and dynamics, Bulletin of Peace Proposals 13, 2. Thomas, R.W., H.O. Stekler and W. Glass, 1991, The economic effects of reducing US defence spending, Defence Economics 2, 183-198. Tinbergen, J., 1993, Paix et 6conomie, in: J. Fontanel, ed., Economistes de la paix (Presses Universitaires de Grenoble, Grenoble) 15-19. UN (United Nations), 1983, Economic and social consequences of the arms race and military expenditures (A/37/386, United Nations Publications, E.83.IX.2, New York). UN (United Nations), 1989, Study on the economic and social consequences of the arms race and military expenditures (A/43/368, United Nations publications, E.89.IX.2., New York). UNIDIR (UN Institute for Disarmament Research), 1984, Establishment of an international disarmament fund for development (UNIDIR, Geneva). UNIDIR (UN Institute for Disarmament Research), 1987, Dsarmement-developpement: au-delA des impasses, quelques suggestions pratiques (UNIDIR/86/13, Geneva). UNIDIR (UN Institute for Disarmament Research), 1993, Economic aspects of disarmament: Disarmament as an investment process (UNIDIR/92/94, New York). Veblen, T., 1917, An inquiry into the nature of peace and the terms of its perpetuation (Macmillan, New York). Ward, M.D., 1990, Technologies militaires et d6veloppement economique, une comparaison de 1'Inde et du Bresil, Ares, Economie de la Defense, 12, 99-122. Ward, M.D., and D.R. Davis, 1990, Economic growth and military spending in the United States, 1952-1988 (National Science Foundation, WPO-90-05, Boulder, CO, and Grenoble). Ward, M.D., D.R. Davis and C.L. Lofdahl, 1994, A century of tradeoffs between defense and growth: The case of Japan and the United States (Working Paper, Boulder, CO, July 8). Whynes, D.K., 1979, The economics of Third World military expenditure (University of Texas Press, Austin, TX).

590

J Fontanel

Wolfson, M., and J.P. Farrell, 1987, Economic warfare between the superpower, in: C. Schmidt and E Blackaby, eds., Peace, defence and economic analysis (International Economic Association, Macmillan Press, London) 155-181.

AUTHOR INDEX Abdallah, W.K. 58 Aben, J. 578 Aben, J., see Schmidt, C. 83 ACOST 423 Adams, F.G. 255, 260 Adams, G. 568 Adams, G., see Gold, D. 576 Adamson, D., see Grissmer, D. 388 ADCC 497 Adedeji, A., see Quester, A. 364 Albrecht, M. 368, 371 Albrecht, U. 53, 300 Alesina, A. 303 Alexander, A.J. 535, 536 Alexander, W.R.J. 255, 260, 261, 266, 571 Alexander, Y. 215 Alogoskoufis, G. 78 Altman, S. 373 Anderton, C.H. 73, 115, 548 Anderton, C.H., see Isard, W. 115 Anderton, R., see Smaldone, J. 551 Andreoni, J. 94 Andrisani, P., see Gilroy, C. 392 Angrist, J. 392 Annez, P., see Faini, R. 27, 255, 284, 291, 292 Anton, J.J. 334 Arora, MB. 60, 63 Arrow, K.J. 466 Asch, B. 359, 382, 383, 385, 388 Aschauer, D. 577 Ash, C. 3, 355 Aten, B., see Heston, A. 55, 56 Atesoglu, H.S. 255, 575, 576 Atesoglu, H.S., see Mueller, M.J. 256 Atkinson, S.E. 234 Attali, J. 569 Augustine, N. 53 Aztec Training and Enterprise Council 497 Baack, B. 407 Bach, S., see Binkin, M. 386 Baik, K.H. 169 Bailey, M.J. 20, 38 Baily, M. 441 Bajusz, W.D. 525, 534, 544, 545 Baldwin, D.A. 573 Baldwin, R., see Scribner, B. 370 Ball, D.N.F. 503 Ball, N. 49, 51, 52, 56, 58, 60, 65, 280, 415

Banks, A. 552 Baran, P. 411, 570 Barker, T. 579 Baron, D.P. 311, 326, 331, 333, 339 Barro, R. 294, 303, 304 Battistelli, .E 514 Baugh, W. 534, 551 Baumol, W. 407, 424 Bayoumi, T., see Arora, VB. 60, 63 Beard, R. 465 Becker, A.S. 548 Becker, G.S. 182 Behrman, J.R., see Adams, EG. 255, 260 Beland, R. 368, 370 Belsley, D.A. 270 Ben-Habib, J. 303 Benjamin, R.W., see Bobrow, D.B. 552 Benoit, E. 5, 9, 27, 253-255, 258, 266, 280 Berg, A. 301 Berg, E., see Berg, A. 301 Berger, M. 391 Bergstrom, T.C. 94, 97 Berner, K. 355, 358, 359 Berthelemy, J.-C. 295 Besanko, D.E. 311 Besanko, D.E., see Baron, D.P. 331, 333, 339 Betzig, L. 184 Binkin, M. 386 Binmore, K. 233 Birkler, J., see Rich, M. 480 Bischak, G. 514, 566 Bishop, P. 515 Biswas, B. 9, 255, 258-261, 263, 266, 267, 292 Bitzinger, R.A., see Kosiak, S. 509 Bjerkholt, O., see Cappelen, A. 581, 582 Black, M. 362 Blackaby, F.E 61 Blair, R., see Miller, R. 499 Blume, L., see Bergstrom, T.C. 94, 97 Bobrow, D.B. 552 Bohi, D.R. 424 Boldin, M., see Adams, F.G. 255, 260 Borjas, G. 393 Bougrov, E. 486 Boukharine, N. 570 Boulding, K.E. 19, 20, 24, 30, 37, 117, 171 Bouthoul, G. 569 Bower, A.G. 338

592 Braddon, D. 496, 497 Brauer, J. 422, 511 Brewer, G. 569 Brito, D.L. 5, 19, 20, 111, 113 115, 118, 121, 122, 126, 127, 135, 147, 198, 568, 570, 573 Brito, D.L., see Intriligator, M.D. 32, 113, 115, 122, 125, 155 Brophy-Baermann, B. 240 Brown, C. 355 Browning, E. 375 Brunton, B.G. 411 Bryant, R. 392 Brzoska, M. 52, 57, 61, 63, 64, 420, 422, 425, 426 Brzoska, M., see Pearson, F.S. 554 Buck, D. 405, 423 Buck, D., see Hooper, N. 407 Buddin, R. 364, 366, 369, 387, 389 Buddin, R., see Grissmer, D. 389, 390 Bueno de Mesquita, B. 131, 151 Buoncristiani, A.M., see Brito, D.L. 135 Burnett, W.B. 313 Burton, R. 504 Butz, WP., see Alexander, A.J. 535, 536 Byers, J.D. 568 Caillaud, B. 311 Callaghan Jr, T.A. 462 Campbell, Colin 342 Campbell, J.M., see Gregory, P.R. 262 Cappelen, A. 581, 582 Carlton, D., see Alexander, Y. 215 Carmichael, J. 448 Cars, H.C. 55 Carus, WS. 534 Catrina, C. 525, 531, 558 Cauley, J. 216, 237, 243 Cauley, J., see Enders, W 234, 237-239 Cauley, J., see Im, E.I. 220 Cauley, J., see Sandler, T. 18, 152, 215, 217, 235 CBO (US Congressional Budget Office) 342, 463, 544, 545 Center for Studies in Defense Resources Management 504 Chakrabarti, A.K. 423 Chakrabarti, A.K., see Baily, M. 441 Chan, MW.L. 80 Chan, S. 253, 281 Che, Y.-K. 335 Cheng, B., see Gregory, P.R. 262 Cheshire, P.C. 515 Chesnais, F. 571 Chiang, A.C. 502 Childress, M., see Orvis, B. 370 Chinworth, M. 479481

Author Index Chowdhury, A.R. 80, 255, 263 Clark, R. 370 Cochran, M., see Ward, M.D. 257, 292, 293 Cohen, L. 448 Cole, S. 506 Commission of the European Communities 510, 511 Conybeare, J.A.C. 26, 72, 98, 99, 106 Conybeare, J.A.C., see Brophy-Baermann, B. 240 Cooke, T. 364, 369 Cooper, J. 422, 510 Cooper, R. 35, 379 Coopey, R. 405 Cooter, R. 167 Comes, R. 94, 95, 97 Correa, H. 568 Cox, A., see Hartley, K. 423 Cox, J.C., see Warf, B. 514 Crantz, C., see Pearson, F.S. 554 Crenshaw, M. 215 Crocker, K.J. 315 Cronberg, T. 583 Crow, R.T. 504 Cruz, J., see Simaan, M. 115 Cummins, J.M. 3 Cusack, T.R. 76 Cusack, T.R., see Nincic, M. 568 Cypher, J. 76, 406, 423 Dacy, D.C. 34 Dale, C. 355 Daly, M. 174 Daula, T. 355, 356, 358, 359, 364, 372 Daula, T., see Berner, K. 355, 358, 359 Daures, N., see Aben, J. 578 Davidson, R. 270 Davidson, R., see MacKinnon, J.G. 106 Davis, D.R., see Ward, M.D. 257, 292, 293, 572, 574, 577 Davis, Z.S. 112 Daymont, T., see Gilroy, C. 392 Deardorff, A.V, see Grobar, L.M. 525, 544, 545 Deaton, A. 73, 77 Defense Budget Project 516 Deger, S. 3, 27, 84, 253, 255, 256, 262, 266, 268, 279-281, 284, 289-292, 294, 298-300, 305, 512, 568, 571, 580 Demilitarised 497 Demsetz, H. 466 Demski, J.S. 334 Denoon, D.B.H. 25, 27 Dertouzos, J. 354, 355 Dertouzos, J., see Asch, B. 359 Dertouzos, J., see Polich, M. 355-359 Dessler, D. 547

Author Index DeTray, D. 391 Deutsch, K.W 151 Devine, J., see Griffin, L.J. 75, 568 Diamond, J., see Heller, P.S. 299 Dickey, D. 220 Diehl, P.E, see Goertz, G. 47, 64 Dixit, A. 169, 176 DoD (US Department of Defense) 53, 442, 504 Dowdall, P., see Braddon, D. 496, 497 Draper, A. 473 Duchin, F., see Leontieff, W. 575, 581 Dudley, L. 25, 34, 75, 79, 93, 101, 105 Dumas, L.J. 410, 485, 584 Dunne, J.P. 73, 77, 80, 406, 411, 421, 422, 424-427, 504, 505, 576, 578 Dunne, J.P., see Barker, T. 579 Dunne, J.P., see Smith, R. 76, 411 Dyckman, J., see Burton, R. 504 Easterly, W. 304 EC 484 Edgerton, D. 412 Eisner, R. 570, 577 Elliott, M.V, see Flegg, A.T. 505 Ellsberg, D. 17 Elrod, M.A., see Oneal, J.R. 94, 97, 100, 102 Emmanuel, A. 569 Enders, W. 219, 220, 234, 235, 237-239, 242-245 Engelbrecht, H.C. 536 Engels, F. 570 Engineer, M., see Usher, D. 193 Enns, J. 360 Evron, Y. 148 Faini, R. 27, 255, 284, 291, 292 Farrell, J.P., see Wolfson, M. 27, 573 Fechter, A., see Altman, S. 373 Feder, G. 9, 258 260 Fei, E.T. 60 Ferejohn, J. 525 Fernandez, J. 370 Fernandez, R. 357, 358 Ferrari, P.L. 527 Filip-Kohn, R. 504 Fine, B. 411 Fischer, H.E., see Leiss, A.C. 525 Fisher, A. 28, 373 Flegg, A.T. 505 Fleming, P., see Mickolus, EF. 216 Fletcher, J. 387 Follmann, D., see Kostiuk, P. 370, 389 Fontanel, J. 19, 35, 53, 55, 79, 565-569, 572, 573, 578, 580, 585 Fontanel, J., see Cars, H.C. 55 Fontanel, J., see Smith, R. 34, 78, 525, 571

593 Forbes, J.E, see Sandler, T. 100, 103, 104, 152 Fourier, C. 570 Fox, J.R. 340 Frank, G. 570 Frankel, B. 148 Frankel, B., see Davis, Z.S. 112 Frederiksen, P.C. 255, 267 Fredland, J. 391 Frey, B.S. 525 Freyssinet, J. 569 Friedland, C., see Stigler, G.J. 424 Friedman, D. 20, 30, 37 Friedman, J.W. 171 Friedman, M. 373 Fritz-ABmus, D. 84 Fudenberg, D. 132 Fuller, W., see Dickey, D. 22:0 Gahart, M., see Orvis, B. 354 Galbraith, J.K. 568, 569 Gallois, P. 151 Gansler, J.S. 341, 342, 404, 408, 409, 412 Garfinkel, M. 38 Gelb, L. 555 Georgiou, G. 80 Gerner, D. 525, 526, 548 Giffen 570 Gilroy, C. 392 Gilroy, C., see Dale, C. 355 Gilroy, C., see Kearl, E. 358 Glass, W., see Thomas, R.W. 577 Gleditsch, N.P. 76, 111 Gleditsch, N.P., see Cappelen, A. 581, 582 Glickman, N.J. 502, 505 Glickman, N.J., see Klein, L.R. 505 Glismann, H.M., see Chakrabarti, A.K. 423 Goertz, G. 47, 64 Gold, D. 576 Gold, D., see Adams, G. 568 Goldberg, L. 355, 358 Goldberg, M. 364, 366, 393 Goldberg, M., see Warner, J. 360, 361, 364 Goldberg, VP. 315 Golden, J., see Olvey, L.D. 28 Goldstein, J. 572 Gonchar, K. 57 Gooding, E., see Weiss, S. 497 Goodwin, C.D. 565 Gottman, J.M. 219 Gotz, G. 362-364, 366, 371, 372 Graham, M., see Walker, W. 403 Granger, C.W.J. 80 Grawe, O.R. 493 Gray, C. 125 Greenwood, D. 80 Gregory, P.R. 262 Griffin, L.J. 75, 568

594 Griliches, Z. 446 Grimmett, R. 64 Grindle, M.S. 301, 568 Grissmer, D. 388 390 Grissmer, D., see Buddin, R. 389 Grissmer, D., see Kirby, S. 389 Grobar, L.M. 253, 290, 525, 544, 545 Groenwold, N. 506 Grogan, J., see Kostiuk, P. 388 Grossman, G.M. 410, 433, 453 Grossman, H.I. 33, 38, 179, 193-195, 200 Grossman, S.J. 322 Groth Jr, C.H., see McGuire, M.C. 25, 99, 106 Guesnerie, R. 326 Guesnerie, R., see Caillaud, B. 311 Gupta, D., see Venieris, Y. 303 Gyimah-Brempong, K. 568, 580 Haavelmo, T. 193 Hagger, A.J., see Groenwold, N. 506 Hall, G.R. 480 Hall, P. 481, 482 Hamilton, W.D. 174 Hammond, C. 369 Hanighen, F.C., see Engelbrecht, H.C. 536 Hanley, J., see Buddin, R. 364, 366, 369 Hansen, K.H., see Cronberg, T. 583 Hansen, L. 104, 373 Hansen, L., see Murdoch, J.C. 25, 75, 93, 105 Happe, N. 61-63 Harbor, B., see Walker, W. 403 Harkavy, R.E. 525, 554, 558 Harkavy, R.E., see Neuman, S.G. 535 Hart, O.D., see Grossman, S.J. 322 Hartley, K. 3, 83, 404, 420, 423, 424, 427, 453, 473, 474, 477-479, 498, 516, 540, 572, 579, 583 Hartley, K., see Buck, D. 405, 423 Hartley, K., see Hooper, N. 486 Hartley, K., see Lynk, E. 424 Hartley, K., see Martin, S. 481, 482 Hartley, K., see Ott, A. 485 Hartley, K., see Sandler, T. 75, 98, 111, 251, 253, 257, 261, 264, 265, 292, 311, 312, 409, 468, 472, 483, 485, 525, 565 Hausman, J.A. 262, 263, 269, 305 HCP 247 474, 477 HCP 265 477 HCP 286 482 Hebert, J.P. 576 Heckman, J. 392 Heller, P.S. 299 Heller, P.S., see Tait, A.A. 79, 299 Helpman, E., see Grossman, G.M. 410 Hendry, J.B. 34 Herrera, R. 49, 52, 61 Herrera, R., see Berthelemy, J.-C. 295

Author Index Hersh, S.M. 149 Hesse, M.I., see Rich, M. 480 Heston, A. 55, 56 Heston, A., see Summers, R. 55 Hewings, G.J.D., see Jensen, R.C. 506 Hewitt, D.P. 61, 63, 568 Hibbs, D. 303 Higgs, R., see Trevino, R. 424 Hildebrandt, G.E. 47 Hilferding, R. 570 Hilton, B. 83 Hirsch, B., see Berger, M. 391 Hirschman, A.O. 23, 499 Hirshleifer, J. 20, 30, 98, 172, 174-176, 179, 188, 193 Hitch, C.J. 4, 15, 17 Hitiris, T. 516 Hoagland, J.H., see Leiss, A.C. 525 Hoffenberg, M., see Leontieff, W. 24 Hogan, P. 366 Hogan, P., see Mackin, P. 365, 366 Holland, D., see Hughes, D. 514 Holmstrom, B. 322 Holzman, F.D. 56 Hooper, N. 407, 486 Hooper, N., see Buck, D. 405, 423 Hooper, N., see Hartley, K. 420, 498, 516 Hooper, N., see Kirby, S. 576 Hopmann, P.T., see Bobrow, D.B. 552 Hormats, R.D. 577 Horn, E.-J., see Chakrabarti, A.K. 423 Horne, D., see Kearl, E. 358 Horowitz, S. 368, 370 Horowitz, S., see Hammond, C. 369 Hosek, J. 364, 386 Hosek, J., see Asch, B. 359 Howard, M.C. 411 Hsiao, C. 506 Hsiao, C., see Chan, M.W.L. 80 Huang, C. 255, 576 Huang, C., see Mintz, A. 256 Huck, B.J. 514 Hufbauer, G. 28 Hughes, D. 514 Huisken, R. 61 Humm, A., see Smith, R. 34, 78, 525, 571 Huntington, S. 125 Hussain, F., see Hartley, K. 404 IFO 497 Ihori, T. 19, 36 IISS (International Institute for Strategic Studies) 62 Im, E.I. 220 Im, E.I., see Cauley, J. 216, 237, 243 IMF (International Monetary Fund) 58, 582 IMF (International Monetary Fund) GFSY 62

Author Index Intriligator, M.D. 5, 16, 19, 32, 33, 111-113, 115, 117, 120-123, 125, 155, 171, 186, 571 Intriligator, M.D., see Brito, D.L. 20, 113, 114, 122, 126, 135, 147, 198, 568, 570, 573 Isard, W. 115, 499 Ishaq, A., see Lebovic, J.H. 256 Islam, M.Q. 224 Islam, M.Q., see Shahin, W.N. 216, 224 Iyigun, M., see Grossman, H.I. 193 Jack, B. 26 Jacobs, R.L. 263 Jacobsen, C.G. 57 Jane's 464 Jensen, R.C. 506 Joerding, W. 255, 262, 263 Johnson, J. 533 Johnson, R.E., see Hall, G.R. 480 Jones, D., see Majeski, S. 551 Jones, E.L. 204 Jones, P.R. 537, 538 Kahn, H. 151, 152, 170 Kalai, E. 127, 135 Kaldor, M. 407 Kamlet, M.S. 76 Kan, S. 65 Kaplan, M.A. 151 Karoly, L., see Asch, B. 359 Katz, J.E. 422 Kaun, D.E. 424 Kawata, J., see Grissmer, D. 389, 390 Kay, N. 448 Kaysen, C. 35 Kearl, E. 358 Kelly, R., see Olvey, L.D. 28 Kelman, S. 317, 342 Kemp, G. 553 Kemp, G., see Leiss, A.C. 525 Kende, I. 568 Kendry, A., see Braddon, D. 496, 497 Keng, C.WK., see Chan, M.W.L. 80 Kennedy, P. 422, 571, 572, 574, 576 Kent, G.A. 18, 20, 24, 33 Kester, J. 379 Keynes, J.M. 570 Kidron, M. 411, 570 Kiefer, D. 551 Kim, Ji-Won, see Correa, H. 568 Kim, M., see Grossman, H.I. 179, 193, 200 King, J., see Howard, M.C. 411 Kinsella, D. 547, 551, 576 Kirby, S. 389, 576 Kirby, S., see Grissmer, D. 388 390 Kirby, S., see Marquis, S. 388, 389 Kirkpatrick, D.L.I. 463

595 Klein, L.R. 24, 505, 570, 577, 582 Kleinman, S. 364 Knopf, J.W., see Ferrari, P.L. 527 Knorr, K. 23, 95 Kolodziej, E.A. 540 Kosiak, S. 509 Kostiuk, P. 370, 388, 389 Kovacic, W.E. 314 Krause, K. 422, 425 Kravis, I.B. 54, 55 Kreps, D.M. 171 Krueger, A., see Angrist, J. 392 Krugman, P. 535, 540-542 Kuh, E., see Belsley, D.A. 270 Kupperman, R. 122 Kuran, T. 198 Laffont, J.-J. 3, 311, 326-328, 331, 334, 337-339, 342, 408, 424 Laffont, J.-J., see Guesnerie, R. 326 Lambelet, J.C. 125 Lanchester, F.W. 30 Landau, D. 255 258, 263, 264, 266, 268, 269, 294 Landes, W.M. 215, 235, 236, 243 Langford, TW., see Isard, W. 499 Lapan, H.E. 224, 227, 228, 233, 234 Lapan, H.E., see Sandler, T. 228, 232 Laurance, E.J. 64, 525, 558 Leamer, E.E., see Jacobs, R.L. 263 Lebovic, J.H. 256, 581 Lee, D.R. 223, 373 Lehrer, E., see Kalai, E. 127 Leiss, A.C. 525 Leitenberg, M., see Ball, N. 415 Leitzel, J. 339 Lenin 570 Leontieff, W. 24, 499, 575, 581 Lerner, J. 423 Levin, R. 449 Levine, P. 34, 78, 525, 546, 559 Levine, R. 270 Levy, D. 449 Levy, D., see Buddin, R. 364, 366, 369 Lichbach, M.I. 217 Lichtenberg, ER. 318, 436, 439, 440, 442-444, 447, 449, 451, 455 Lichtenberg, F.R., see Griliches, Z. 446 Lim, D. 256 Lincoln, G.A. 18, 21 Linden, M. 4, 579 List, F. 569 Little, R., see Fredland, J. 391 Little, R., see Payne, D. 366 Lock, P. 57, 494 Lock, P., see Brzoska, M. 420, 422, 425, 426 Lofdahl, C.L., see Ward, M.D. 574

596 Looney, R.E. 83, 301, 568 Looney, R.E., see Frederiksen, P.C. 255, 267 Lotz, J.R. 298, 299, 568 Louscher, D.J., see Bajusz, W.D. 525, 534, 544, 545 Lovering, J. 412, 425, 506 Lutkepohl, H. 263 Luxemburg, R. 570 Lynk, E. 424 Lynk, E., see Hartley, K. 424 Mackin, P. 365, 366 MacKinnon, J.G. 106 MacKinnon, J.G., see Davidson, R. 270 Macnair, E.S. 6, 251, 256, 260, 261, 264-266, 269, 293 Maddala, G.S. 264, 270 Madden, J.R., see Groenwold, N. 506 Maddock, I. 407, 579 Madrid, R.L., see Ferrari, P.L. 527 Mairs, L., see Mackin, P. 365, 366 Maizels, A. 79, 298, 299, 568 Majeski, S. 551 Malthus, R. 569 Mandel, E. 569 Maneval, H. 493, 582 Maniruzzaman, T. 551 Mankiw, N.G. 446 Mansfield, E. 434, 448 Marcus, A. 369 Marcus, A., see Cooke, T. 364 Marin, D. 481 Markowski, S., see Hall, P. 481, 482 Marks, S., see Cooter, R. 167 Markusen, A. 424, 506 Marquis, S. 388, 389 Marra, R.E, see Ostrom Jr, C.W. 79 Marshall, R.C. 342 Martin, S. 481, 482 Martin, S., see Hartley, K. 474, 477, 478 Martin, T. 387 Marx, K. 570 Maskus, K.E., see Udis, B. 481 Matthews, R. 473 Mayer, K.R. 339 Mayer, T. 125 Mayer, W. 19 McAfee, R.P. 3, 326, 331, 333, 334 McCall, J., see Gotz, G. 362-364, 366 McGinnis, M.D., see Williams, J.T. 76, 80 McGregor, P.G. 505 McGuire, M.C. 4, 19, 20, 24, 25, 33, 34, 36, 77, 98, 99, 106, 111, 117, 525 McGuire, M.C., see Andreoni, J. 94 McKean, R.N., see Hitch, C.J. 4, 15, 17 McKenzie, R., see Lee, D.R. 373

Author Index McMahon, J., see Fletcher, J. 387 McMahon, J., see Shiells, M. 387 McMillan, J., see McAfee, R.P. 3, 326, 331, 333, 334 McNaugher, T.L. 340 McNicoll, I.H., see McGregor, P.G. 505 McNown, R.F., see Ash, C. 3, 355 Mehay, S.L. 388 Mehay, S.L., see Looney, R.E. 83 Melman, S. 410, 423, 485, 568, 569, 571, 576 Meurer, M.J., see Marshall, R.C. 342 Mickolus, E.E 216, 217 Mihalka, M., see Alexander, A.J. 535, 536 Milgrom, P., see Kreps, D.M. 171 Mill, J.S. 569 Miller, J. 373 Miller, R. 499 Mills, C.W. 410 Milstein, J.S. 525, 552 Ministere de la Defense 504 Mintz, A. 75, 256 Mintz, A., see Chan, S. 281 Mintz, A., see Huang, C. 255, 576 Mirrlees, J. 326 Mnookin, R., see Cooter, R. 167 Moffitt, R., see Black, M. 362 Moffitt, R., see Daula, T. 364 Monteverde, K. 318 Montmarquette, C., see Dudley, L. 75, 79, 93, 101, 105 Moodie, M. 534 Moravcsik, A. 468 Morgenstern, 0. 111 Morgenthau, H.J. 151 Mori, K., see Klein, L.R. 24 Moskos Jr, C.C. 411 Most, B.A., see Starr, H. 551 Mountain, D.C., see Hsiao, C. 506 Mowery, D.C., see Kamlet, M.S. 76 Muellbauer, J., see Deaton, A. 73, 77 Mueller, D.C. 467 Mueller, M.J. 256 Mueller, M.J., see Atesoglu, H.S. 255, 575, 576 Muller, J. 552 Murdoch, J.C. 19, 25, 75, 77, 82, 84, 93, 95-97, 102-105 Murdoch, J.C., see Conybeare, J.A.C. 26, 72, 98, 99, 106 Murdoch, J.C., see Hansen, L. 104 Murdoch, J.C., see Macnair, E.S. 6, 251, 256, 260, 261, 264-266, 269, 293 Murdoch, J.C., see Sandler, T. 25, 77, 97, 99, 104, 105 Murdock, J.M., see Mickolus, E.E 216, 217 Musa, M. 326 Myerson, R.B., see Baron, D.P. 326

Author Index Nachmias, N. 554 Nadal, A.E. 485 Nakada, M., see Quester, A. 380 Nalebuff, B.J. 135, 435 Nardin, T., see Slater, J. 411 Nash, J.F. 135, 138 National Defense Research Institute 390 National Science Board 433 Nef, U. 569 Nelson, G. 355 Nelson, G., see Enns, J. 360 Nelson, P.S. 243 Neuman, S.G. 535, 553 Nickell, S. 78 Nincic, M. 568 Niskanen, W.A. 340 Nissanke, M.K., see Maizels, A. 79, 298, 299, 568 Njolstad, O., see Gleditsch, N.P. 76, 111 Noh, S.J., see Grossman, H.I. 194, 195 Noll, R., see Cohen, L. 448 Nykvist, K. 558 Obstfeld, M., see Krugman, P. 540-542 Oden, M., see Bischak, G. 514, 566 OECD 486 Office of Economic Adjustment 515 Ohlson, T., see Blackaby, E 61 Oi, W. 28, 373 Okamura, M. 19, 77, 102 Olson, M. 4, 18, 25, 27, 91, 94, 95, 100 Olszewski, E.A. 577 Olvey, L.D. 28 OMB (US Office of Management and Budget) 482 Oneal, J.R. 94, 97, 100, 102, 103 Orvis, B. 354, 370 Osband, K., see Bower, A.G. 338 Ostrom Jr, C.W. 76, 79 Ott, A. 485 Overgaard, P.B. 234 Ozler, S., see Alesina, A. 303 Palda, E 33 Palmer, G. 91 Panzar, J., see Baumol, W. 407, 424 Parise, G.F., see Enders, W. 219, 220, 235 Pashardes, P., see Dunne, J.P. 73, 77 Paukert, L. 406, 508 Payne, D. 366 Payne, J.E. 80 Payne, J.E., see Olszewski, E.A. 577 Peacock, A.T. 475 Pearson, ES. 554 Pearton, M., see Stanley, J. 525, 527 Peck, M.J. 4, 28, 312, 341, 407, 424, 450 Peel, D.A., see Byers, J.D. 568

597 Penubarti, M., see Ward, M.D. 257, 292, 293 Perotti, R., see Alesina, A. 303 Perron, P. 245 Peterson, C., see Hosek, J. 364, 386 Phillips, R., see Gilroy, C. 392 Phillips, R., see Scribner, B. 370 Pi, C.-R., see Macnair, E.S. 6, 251, 256, 260, 261, 264-266, 269, 293 Piccione, M. 337 Pierre, A.J. 525, 527 Pigou, A.C. 570 Pilandon, L., see Schmidt, C. 83 Pivetti, M. 76, 411 Plosser, C.I. 263, 269 Polachek, S.W. 19, 38 Polich, M. 355-359 Polich, M., see Orvis, B. 370 Pollock, D. 554 Pomfret, R. 466 Popkin, S.L. 195 Porter, R.C., see Grobar, L.M. 253, 290 Powell, R. 179, 209 Press, J., see Polich, M. 355-359 PRODEM 415 Proudhon, J.J. 570 Pugh, P. 463, 474 Pugh, P., see Kirkpatrick, D.L.I. 463 Quester, Quester, Quester, Quester,

A. A., A., A.,

364, 380, 387 see Beland, R. 368, 370 see Cooke, T. 364, 369 see Fletcher, J. 387

Rae, J. 569 Rajmaira, S. 568 Rajmaira, S., see Ward, M.D. 257, 292, 293 Ram, R. 9, 256, 260, 266-268 Ram, R., see Biswas, B. 9, 255, 258-261, 263, 267, 292 Ramsey, J.B. 263, 264 Rapoport, A. 115 Rasler, K. 256 Ratner, J. 405 Rattinger, H. 76 Rautsola, P., see Maneval, H. 582 Ravenhill, J. 301 Ray, E., see Baack, B. 407 Reay, S., see Braddon, D. 497 Rebelo, S., see Easterly, W. 304 Reed, L.W., see Kaysen, C. 35 Refson, J.S., see Leiss, A.C. 525 Reichelstein, S. 333 Renelt, D., see Levine, R. 270 Renner, M. 426, 566 Report of the President's Commission on an All-Volunteer Armed Force 373 Reppy, J. 32, 412

598 Research Seminar in Quantitative Economics 504 Rey, P., see Caillaud, B. 311 Reynolds, K.J., see Crocker, K.J. 315 Ricardo, D. 569 Rich, M. 480 Richard, J.-F., see Marshall, R.C. 342 Richards, P.J. 571 Richards, P.J., see Paukert, L. 406, 508 Richardson, H.W. 499 Richardson, L.F. 4, 8, 19, 30, 111, 115, 118, 570 Ridge, M. 370 Riker, W., see Bueno de Mesquita, B. 151 Riordan, M.H. 334, 337, 338 Roberts, J., see Kreps, D.M. 171 Roeder, P. 554 Roemer, J.E. 193, 195 Rogerson, W.P. 311, 312, 314, 315, 318, 319, 322, 331, 336, 338, 339, 341, 342, 409, 442 Roll, R., see Gotz, G. 371, 372 Romer, D., see Mankiw, N.G. 446 Romer, P.M. 304, 446 Rosecrance, R.N. 151 Rosen, S. 349, 352, 381, 384, 391 Rosen, S., see Musa, M. 326 Ross, D., see Scherer, EM. 540 Ross, K.L., see Olszewski, E.A. 577 Ross, K.L., see Payne, J.E. 80 Rostker, B., see Asch, B. 359 Rothfels, H. 178 Rothstein, R.L. 297, 302 Roubini, N., see Alesina, A. 303 Rowen, H. 27 Russett, B.M. 95, 96, 100, 153, 572 Saaty, T. 118, 122 Saint-Simon 570 Samaranayake, V., see Bryant, R. 392 Samuelson, P. 570 Sandler, T. 5, 18, 25, 31, 75, 77, 92, 94, 95, 97-100, 103-105, 111, 152, 170, 215, 217, 223, 228, 232, 235, 251, 253, 257, 261, 264, 265, 292, 311, 312, 409, 468, 472, 483, 485, 525, 565 Sandler, T., see Atkinson, S.E. 234 Sandler, T., see Conybeare, J.A.C. 26, 72, 98, 99, 106 Sandler, T., see Comes, R. 94, 95, 97 Sandler, T., see Enders, W. 219, 220, 234, 235, 237-239, 242-245 Sandler, T., see Hansen, L. 104 Sandler, T., see Hartley, K. 83 Sandler, T., see Im, E.I. 220 Sandler, T., see Lapan, H.E. 224, 227, 228, 233, 234

Author Index Sandler, T., see Lee, D.R. 223 Sandler, T., see Macnair, E.S. 6, 251, 256, 260, 261, 264-266, 269, 293 Sandler, T., see Mickolus, EF. 216, 217 Sandler, T., see Murdoch, J.C. 19, 25, 75, 77, 82, 84, 93, 95-97, 102-105 Sappington, D.E.M. 311, 326, 337 Sappington, D.E.M., see Besanko, D.E. 311 Sappington, D.E.M., see Demski, J.S. 334 Sappington, D.E.M., see Riordan, M.H. 334, 337 Say, J.B. 569 Scheetz, T. 256, 292, 300, 301 Schelling, T.C. 3, 4, 17, 20, 24, 27, 32, 111, 153, 566 Scherer, EM. 4, 311, 312, 325, 436, 451, 540 Scherer, F.M., see Burnett, W.B. 313 Scherer, F.M., see Peck, M.J. 4, 28, 312, 341, 407, 424, 450 Schmidt, C. 83 Schmidt, P., see Ramsey, J.B. 264 Schnitzer, M., see Marin, D. 481 Schofield, S. 403 Schott, J., see Hufbauer, G. 28 Schrodt, P. 551 Schwarz, J.A. 410, 412 Schwert, G.W., see Plosser, C.I. 263, 269 Scott, J. 449 Scott, J.L. 233 Scott, J.L., see Nelson, P.S. 243 Scribner, B. 370 Seiglie, C. 35 Seligman, E.R. 569 Selten, R. 216, 228 Sen, A. 281 Sen, S. 52, 55, 61 Sen, S., see Berthelemy, J.-C. 295 Sen, S., see Deger, S. 84, 281, 284, 290-292, 300, 305, 512 Sen, S., see Levine, P. 34, 525, 546, 559 Shahin, W.N. 216, 224 Shahin, W.N., see Islam, M.Q. 224 Shambaugh, D. 52 Shapiro, C., see Grossman, G.M. 433, 453 Sharp, J. 493 Shavell, S. 322 Shell, K. 286 Sherman, A., see Horowitz, S. 368, 370 Sherwin, R. 552 Shibata, H., see McGuire, M.C. 19, 36 Shiells, M. 387 Shogren, J.E, see Baik, K.H. 169 Shubik, M. 19 Shubik, M., see Brewer, G. 569 Shughart, W., see Kleinman, S. 364 Siegel, D., see Lichtenberg, F.R. 447 Simaan, M. 115

Author Index Singer, J.D., see Deutsch, K.W. 151 Singer, J.D., see Small, M. 551 Singh, B. 511 SIPRI (Stockholm International Peace Research Institute) 56, 62, 525, 547 SIPRI Yearbook 419, 420, 531-533, 551 Sislin, J. 555-558 Sivard, R.L. 62, 253, 551 Siverson, R. 131 Skaperdas, S. 38, 176, 193 Skogh, G. 179 Sk6ns, E. 52, 53, 425, 533 Slater, J. 411 Smaldone, J. 551 Small, M. 551 Smith, A. 569 Smith, D. 364, 366, 369, 411, 568, 578 Smith, D., see Daula, T. 355, 356, 358, 359, 372 Smith, D., see Hogan, P. 366 Smith, D., see Scribner, B. 370 Smith, H., see Kupperman, R. 122 Smith, R. 27, 34, 53, 75-78, 80, 81, 83, 256, 406, 410, 411, 498, 525, 571, 579 Smith, R., see Alogoskoufis, G. 78 Smith, R., see Barker, T. 579 Smith, R., see Deger, S. 27, 255, 256, 262, 266, 290, 568, 580 Smith, R., see Dunne, J.P. 73, 77, 80, 406, 421, 426, 504, 576, 578 Smith, R., see Fontanel, J. 79, 567, 568 Smith, R., see Georgiou, G. 80 Smith, R., see Hartley, K. 404 Smith, R., see Levine, P. 34, 78, 525, 546, 559 Smith, R., see Ridge, M. 370 Smith, R., see Smith, D. 411, 568, 578 Smith, T.C. 125, 131 Smoker, P. 118 Smorodinsky, M., see Kalai, E. 135 Snider, L. 540, 554, 555 Sollinger, J., see Grissmer, D. 389, 390 Solon, G., see Warner, J. 366, 369 Southwood, P. 407 Spiegel, M., see Ben-Habib, J. 303 Spiller, P.T., see Demski, J.S. 334 Spinardi, G., see Coopey, R. 405 Squires, M., see Baugh, W. 534, 551 Srinivasan, T.N. 19 Stanley, J. 525, 527 Stanley, W., see Rich, M. 480 Stanton, R., see Gotz, G. 372 Starr, H. 551 Stekler, H.O., see Thomas, R.W. 577 Stern, E.M., see Grobar, L.M. 525, 544, 545 Stevenson, R.T., see Mintz, A. 256 Stewart, A.M., see Wilkinson, P. 215 Stewart, D.B. 257, 291

599 Stigler, G.J. 424 Stiglitz, J. 443 Stiglitz, J., see Nalebuff, B.J. 435 Stole, L. 334 Stoll, R.J. 76 Strange, S. 574 Stuart, C., see Skogh, G. 179 Stubbing, R.A. 340, 341 Summers, R. 55 Swagel, P., see Alesina, A. 303 Sweezy, P.M., see Baran, P. 411, 570 Switzer, L., see Mansfield, E. 448 Sylvan, D.A., see Bobrow, D.B. 552 Sylwester, S., see Hogan, P. 366 Sylwester, S., see Smith, D. 364, 366, 369 Syropoulos, C., see Skaperdas, S. 38 Sze, M., see Grissmer, D. 388 Tait, A.A. 79, 299 Tan, G. 337 Tan, G., see Piccione, M. 337 Tan, H. 388 Taubman, P., see Rosen, S. 391 Taylor, C.R. 337 Taylor, L., see Faini, R. 27, 255, 284, 291, 292 Teece, D.J., see Monteverde, K. 318 Telser, L.G. 169 Tennefoss, M., see Siverson, R. 131 Terleckyj, N., see Levy, D. 449 Thaler, D.E., see Kent, G.A. 18, 33 Thee, M. 568 Thomas, C., see Ratner, J. 405 Thomas, J. 442, 443 Thomas, R.G. 298 Thomas, R.W. 577 Thompson, E.A. 20, 30, 39 Thompson, W.R., see Rasler, K. 256 Thompson, W.R., see Zuk, G. 301 Thorsson, I. 404 Tinbergen, J. 574 Tirole, J. 3, 337, 342 Tirole, J., see Caillaud, B. 311 Tirole, J., see Fudenberg, D. 132 Tirole, J., see Laffont, J.-J. 3, 311, 326-328, 331, 334, 337-339, 342, 408, 424 Tisdell, C., see Hartley, K. 453 Todaro, M. 443 Tolley, G. 19 Tollison, R. 380 Trevino, R. 424 Tschirhart, J.T., see Atkinson, S.E. 234 Tschirhart, J.T., see Sandler, T. 215, 217, 235 Tsipis, K. 144 Tullock, G. 30, 151, 195 Tung, S., see Thomas, J. 442, 443 Tyson, L. 466

600 Udis, B. 481 Udis, B., see Ash, C. 3, 355 UN (United Nations) 49, 52, 53, 55, 56, 485, 486, 570, 571 UNDP (United Nations Development Programme) 59, 253 UNIDIR (UN Institute for Disarmament Research) 567, 570, 572, 580, 584, 585 US ACDA (US Arms Control and Disarmament Agency) 52, 59, 62, 253, 278, 527-532, 538, 539 US Congress 435 US Department of Commerce 53 US Department of State 216 US General Accounting Office (GAO) 53, 493 US Strategic Bombing Survey 17, 21 Usher, D. 193 Uttley, M., see Coopey, R. 405 van Ypersele de Strihou, J. 95 Varian, H., see Bergstrom, T.C. 94, 97 Veblen, T. 569 Vehrencamp, S.L. 184 Venieris, Y. 303 Villa, C., see Smith, D. 364, 366, 369 Vob, W., see Lock, P. 494 Vu, A., see Hilton, B. 83 Wakeman-Linn, J., see Happe, N. 61-63 Waldman, D., see Buddin, R. 364, 366, 369 Walker, W. 403, 404, 425, 426 Wallace, M., see Griffin, L.J. 75, 568 Wallace, M.D. 131 Waltz, K.N. 113, 151 Wan, H.Y. 23, 28 Wandschneider, P., see Hughes, D. 514 Ward, M.D. 257, 292, 293, 572, 574, 577, 580 Ward, M.D., see Cusack, T.R. 76 Ward, M.D., see Fontanel, J. 566, 585 Ward, M.D., see Rajmaira, S. 568 Ward, M.P., see Jacobs, R.L. 263 Warf, B. 514 Warner, J. 357, 360, 361, 364, 366, 369 Warner, J., see Asch, B. 382, 383, 385 Warner, J., see Black, M. 362 Warner, J., see Enns, J. 360 Warner, J., see Goldberg, M. 364, 366, 393 Warner, J., see Payne, D. 366 Warr, P.G. 91, 94 Watt, P., see Hartley, K. 424 Webber, C.D., see Flegg, A.T. 505 Weidenbaum, M. 24, 27

Author Index Weil, D., see Mankiw, N.G. 446 Weisbrod, B., see Hansen, L. 373 Weiss, A., see Stiglitz, J. 443 Weiss, S. 497 Welch, F., see Borjas, G. 393 Welsch, R.E., see Belsley, D.A. 270 Wentz, WB. 151 West, G.R., see Jensen, R.C. 506 West, R. 55 Wezeman, S., see Laurance, E.J. 64 Wheelock, T. 554 White, H. 270 White, H., see MacKinnon, J.G. 106 White, H., see Plosser, C.I. 263, 269 Whynes, D.K. 568 Wiberg, H. 47 Wiegert, R., see Maneval, H. 582 Wilhite, A., see Bryant, R. 392 Wilkinson,P. 215 Wilkinson, P., see Alexander, Y. 215 Willett, S., see Dunne, J.P. 427 Willett, S., see Walker, W. 425, 426 Williams, J.T. 76, 80 Williamson, O.E. 28, 314 Williamson, S., see Grissmer, D. 389, 390 Willig, R., see Baumol, W. 407, 424 Wilman, J., see Tolley, G. 19 Wilson, E.O. 174 Wilson, M., see Daly, M. 174 Wilson, R., see Kreps, D.M. 171 Winston, J. 439 Winter Jr, S. 18, 24 Wiseman, J. 486 Withers, G. 360 Wittman, D. 20, 30, 175, 549 Wolf Jr, C., see Rowen, H. 27 Wolfson, M. 27, 33, 171, 573 Wong, K. 25, 36 World Bank 51, 270, 509 Wulf, H. 412, 420-422, 426, 511 Wulf, H., see Laurance, E.J. 64 Wulf, H., see Sk6ns, E. 533 Xie, Y. 391 Yao, D.A., see Anton, J.J. 334 Yudken, J., see Markusen, A. 424, 506 Zeckhauser, R., see Olson, M. 4, 18, 25, 91, 95, 100 Zimmermann, K., see Fritz-ABmus, D. 84 Zou, H.F. 305 Zuk, G. 301

SUBJECT INDEX accidental nuclear war 113, 154, 155, 159 active forces 350, 351 adaptive capacity 515, 517, 518 adjustment costs 486, 571 aerospace 481 Africa 582 agency problem 341 aircraft industry 412 all-volunteer force, see AVF alliances 18, 22, 25, 26, 36, 89-106, 461, 462, 466-472, 474, 533, 538 burden sharing 3, 5, 10, 26, 91, 94-100, 106 empirical studies 100 oligarchy model of 105 technology of supply 98, 106 American procurement system 311 antiterrorism policies 234, 247 embassy security 236, 239 Libyan retaliatory raid of 1986 239, 244, 246 metal detectors 216, 217, 236-239, 243, 244, 246 appropriative activities 193 195 arms control 566 treaties 156 arms exports 529-531 restraints 534, 544-548 arms imports 527-529 arms industry 571, 579 arms races 3 5, 8, 15, 19, 22, 25, 109-159, 292 alliance models of 93 and bankruptcy 118 and bargaining 111, 114, 130, 134, 135, 137, 143, 146, 147, 149 and equilibrium 112-114, 116-119, 121, 123, 126, 128, 130, 137, 138, 141, 143 and optimal paths 120 and rational learning in games 127 and stability 112, 113, 115, 116, 118, 123, 126-128, 130, 131, 145, 150-152, 155, 157, 159 and war 118 general model of 128-130 rational choice models 111, 118, 142, 143 reaction curves 120, 123 arms trade 3, 5, 8, 10, 22, 34, 523-559 and conflict, empirical studies of 551, 552

and foreign policy 554-558 and imperfect competition 535 causes of 534-547 economic causes 535 data on 63, 64 empirical studies of 545 historical attributes of 526 neoclassical trade model of 536, 537 typology of 543 arms transfers 525, 527, 531, 532, 534, 535, 546-551, 553, 554, 558 economic causes of 527, 535 asymmetric information 111, 114, 138, 143, 337 auctions 333-335 AVF (all-volunteer force) 354, 355, 363, 367, 371, 373, 379, 380, 392 bargaining, see under arms races best shot 98, 106 Bosnia 548 budget-maximizing behavior 341 budgeting 10, 51 burden sharing, see under alliances bureaucrats 340 case studies 292, 298, 487, 513-515 co-production 542 coalitions 36, 37 COCOM (Coordinating Committee for Multilateral Export Controls) 534 Cold War 18, 20, 24, 26-34, 47, 64, 157, 270, 278, 350, 401, 402, 405, 412, 415, 421, 422, 424-427, 493, 525-527, 531-534, 542, 558, 572 collaboration 38, 461,464, 468, 469, 472-475, 477, 478, 487 collective action 89-106 collusion 342 commitment ability 338 comparative advantage 461, 462, 467, 478, 566 compensation 478 competition 167, 170, 171, 174, 179, 188, 313, 314, 316, 317, 334, 337, 338, 421, 426, 434, 435, 468, 471, 579 imperfect 535, 541 Competition in Contracting Act 435 cone of mutual deterrence 124, 125

602 conflict 165-188, 527, 547, 548, 551, 552, 557, 558 and alliances 170 and binding agreements 169 and equality 179 and escalation 170 and limited stakes 169 and plays 168 and settlement 169, 170 models of 171, 179 reaction curves 185 sources of 172-175 technology of 175, 179 conscription 28, 50, 55, 379, 380 see also draft Contest Success Function, see CSF contestability 471 contests 178 contracting 28 contractorization 484 contracts 311, 313, 314, 316, 318, 320, 323, 324, 326, 328, 331, 336, 338, 342 contract design 3, 5 cost-plus 438 cost-reimbursement 314, 316, 321-326, 332 cost-sharing 324 fixed-price 313, 314, 316, 322, 324, 326, 327, 330, 332, 334, 339 linear 324, 331, 333, 335 conversion 3, 5, 7, 10, 304, 427, 461, 484, 485, 487, 517, 566, 567, 578, 582-585 Coordinating Committee for Multilateral Export Controls, see COCOM cost-effectiveness analysis 23 cost effectiveness and manpower 360, 367, 372 cost escalation 463 cost of defense 253 cost overruns 331 cost-plus contracts, see under contracts cost-reimbursement contracts, see under contracts cost-benefit analysis 23, 461, 471 costs of wars 51 country data 264-266, 269, 290 country studies 300, 305 crisis stability 113 cross-section data 264-266, 269, 290 crowding-out 280, 283, 284, 296, 577, 579 CSF (Contest Success Function) 176-178 culture of dependency 485 data 505, 506, 518, 575, 582 international comparisons 47, 50, 51, 54, 57, 65 limitations 47, 48, 50-52, 55, 57, 58, 63, 64 problems 494, 495

Subject Index sources 47, 49, 51, 53, 54, 58, 59, 63 deadweight loss 375-377 defense budget 17, 39 equipment 401, 402, 404, 406, 413, 423, 426 industry 534, 542, 544 policy 437, 438 procurement, see under procurement defense economics 6, 7, 15 defense expenditure, see ME defense industrial base, see DIB defense R&D, see under R&D deflators 52-54, 57, 61 delays 477 demand for defense 69-85 demand functions Cobb-Douglas 77, 94 Constant Elasticity 77 Linear Expenditure System 77 Murdoch-Sandler 77 Stone-Geary 77 Department of Defense, see DoD design competitions 433-437, 451, 455 determinants of arms influence success 556, 557 deterrence 15, 17-19, 23, 24, 32, 33, 114, 121-123, 125, 126, 136-138, 143, 145, 152, 156, 157, 159, 171, 186 development times 477, 478 DIB (defense industrial base) 399-427, 472, 479, 480, 533, 534, 540, 543, 544 and government 406 benefits of 423 definitions of 402-406 entry barriers 408 exit barriers 409 in terms of employment 406, 534, 544, 545 disarmament 3, 5, 7, 10, 15, 22, 23, 563-585 and history of economic thought 565, 569, 570 and role of public policy 565, 583 585 definitions of 565-569 econometric results 568 empirical analysis of 565, 575-580 for development 565, 580-583 theoretical analysis of 565, 569 575 disarming race 125 discretionary budget 3 diversification 571 DoD (US Department of Defense) 313318, 324, 334, 338-340, 342, 437-439, 441-444, 451 downsizing 7, 8, 401 draft 349, 351, 352, 373-377, 392 see also conscription dual-use equipment 534

Subject Index dual-use technologies 405, 406 dummy variables 265 EC regions 510 econometrics 3, 6 methods 71, 80, 81, 84 models 258, 568, 571, 575, 576 results 568 economic development 27, 33, 34 military outlays 253, 257, 259, 261, 262, 264, 266-269, 271 relation with growth 281 relation with military spending 298 relation with security 294, 296 economic effects 422 economic growth 251-271, 565, 570-583, 585 externality model 284, 285 models of 258 relation with development 281 relation with military expenditure 289 economic principles 16 economic regeneration 517, 518 economic sanctions 23, 28, 36 economic warfare 15, 21, 22, 27 economics of security 21 economics of transition 34, 35 economies of learning, see under learning economies of scale 8, 313, 314, 316, 317, 534, 535, 539-541, 543, 567 economies of scope 8, 319 efficiency 15, 17, 401, 402, 406, 422, 423 effort 320, 323, 335, 336, 381, 382 Egypt 548 elasticity 453, 456 electronics industry 404 embargo 574 enlistment decision 352 supply 353, 355, 356, 358, 360, 377,388 entry 384 entry barriers, see under DIB equilibrium criterion 39 escalation rate 53 ethnic conflict 34 European Union 516 exchange rates 54 56, 61 exit barriers, see under DIB exogenous and endogenous growth models 282 exploitation hypothesis 94 external effects 253, 257, 261, 264 externalities 6, 8, 292-294, 304, 574, 580 transboundary 228 factor productivity 255, 259-261, 264, 266 fixed-effect modelling 256, 265, 270, 271 fixed-price contracts, see under contracts force mix 371-373

603 foreign aid 15, 25, 33, 34 foreign policy influence 525, 527, 534, 554 France 578 free rider 31, 34, 36, 104 free trade 536-538, 540 free trade area 467, 468 game theory 3 5, 9, 17, 38, 216 dominant strategy 223 Nash equilibrium 5, 93, 94, 96, 97, 103, 106, 112, 186, 223, 231-233 Nash reaction paths 93 Nash-Cournot behavior 25 Nash-Cournot equilibrium 26, 186, 187 perfect Bayesian equilibrium 234 Prisoner's Dilemma 221, 223, 224, 566 repeated games 111 signalling game 233, 234, 247 geopolitics 525, 527, 558, 559 Germany 493, 571, 573 gold plating 407 governance 277 government expenditure 448, 449, 451 Granger causality, see under time-series analysis Gulf War 8 high technology 483 hijacking, see skyjackings under terrorist attack modes history of defense economics 4 history of economic thought, see under disarmament hold-up problem 314, 315, 336 human capital 280, 281, 284, 291, 294-297, 304, 353, 391, 393 IEPG (Independent European Policy Group) 425 IMF (International Monetary Fund) 51, 426, 582 see also author index imperfect competition, see under competition imperfect information 435 impulse response functions 242 incentives 311, 312, 314, 315, 317-319, 322-326, 332, 335-339, 342, 484 income distribution 193, 195, 203, 209 Independent European Policy Group, see IEPG India 548 indigenous arms production 412 indirect costs 50 industrial policies 3, 5, 6, 9, 10 inflation 53, 54, 61 information systems 534 innovation 312, 318, 335-338, 433, 434, 441 input-output analysis 495, 499-502, 575, 578

604

Subject Index

insurrections 3, 5, 8, 9, 191-211 and free-rider problem 195 and internal security measures 194 and resource allocation 198 and ruler's tax policy 202 economic theory of 209 general equilibrium theory of 191-211 technology of 194, 198, 202, 204, 205, 210 international aid 580-582 international conflict 20, 29, 30 International Monetary Fund, see IMF international terrorism, definition of 216 intra-industry trade 539, 541 investment 253, 256, 257, 260-262, 264, 266-269, 271, 569-572, 574, 576-580, 582, 583 Israel 548 Japan 571-575 job shedding 498 joint-product model 91, 92, 95-100, 102-106 juste retour 468, 476 knowledge capital KONVER 517

444-446

labor market 487, 494, 497-499, 518 100 largest firms 415-420 learning 480 economies of 474, 539 541 legitimate governments 302 Lenin 195, 196, 204 licensed and co-production 468, 473, 475, 479, 480, 487 life cycle 316, 450, 451 Lindahl equilibrium 99, 105, 106 LINK 582 local economic impacts 496 Machiavelli's Theorem 188 manpower elasticity estimates 358, 366, 388 marginal cost 451 market failure 465, 466, 487 ME (military expenditure) 45-65, 74-76, 7880, 83, 84, 251-271, 565-572, 575-582, 585 alternative units 57 and development 279 and economic growth 279 as share of GDP 58, 59 as share of government expenditures 58, 59 as share of human development expenditures 58 CIA estimates of 56, 79 comparison 56 data 78-80 definitions of 47, 48, 51, 52 employment impacts of 506

estimates of 55, 56, 61, 79 examples of 83 HMS measure of 57 IISS estimates of 61 IMF data 60 in developing countries 275-305 determinants of 299 median-voter models of 75 military burden measure of 57 neoclassical model of 71 optimizing models of 71, 76, 80, 85 political science models of 76 PPP estimates of 55, 56, 79 SIPRI estimates of 61 UN data 60 US ACDA data 60 median-voter model 93, 101, 105 militarization 300-302 military aid 48-50, 52, 61, 62 military alliances, see alliances military bases 493-499, 510, 514, 517, 518 military compensation 380, 381, 384 military expenditure, see ME military forces, stocks and flows of 73 military governments 301 military industrial complex 401, 402, 409-411 military manpower 3, 6, 9, 347-394 military production function 461, 483, 487 military productivity 349, 367, 368 military rotation policy 370 military strategy 17, 27 military strength 47, 57, 58 military women 349, 386 monopolistic competition 542 monopoly 451, 471, 483, 541 monopsony 380 moral hazard 322 333 self-selection models 325-333 multipliers 292, 578 Nash equilibrium, see under game theory Nash reaction paths, see under game theory Nash-Cournot behavior, see under game theory Nash-Cournot equilibrium, see under game theory national accounts 53, 60 national economic benefits 472 national monopolies 407 national security 17, 22-24, 31, 35, 565, 566, 568, 572 NATO 19, 51, 94, 95, 97, 99-106, 156, 266, 277, 278, 413, 461, 462, 464-467, 470, 474, 540 flexible response 97 NATO free trade area 462 New Economic Order 572 non-pecuniary aspects 352

Subject Index nuclear proliferation 15, 22, 32, 111-113, 149-151, 154-156, 158, 159 nuclear war model 113, 121, 146 nuclear weapons 144-149 number of arms delivered 532 occupational choice 353 offense versus defense 179 offsets 461, 469, 473, 478, 481 oligopoly 541, 542 oligopoly models 24 operations research 21-23 opportunity costs 30, 47, 50, 53-55, 373-377, 447, 567, 578 outbreak of war 111-114, 124, 125, 131, 132, 134, 143, 151, 156, 159 ownership 420, 421 Pakistan 548 paradox of power, see POP peace 8, 15, 19, 32, 549 peace dividend 427, 569-572, 576, 577, 579 peace science 10 pensions 360 policy options 584 politics 167, 171, 178, 187 pooling equilibrium 141, 143 POP (paradox of power) 182, 183 Potential Settlement Region, see PSR PPP (Purchasing Power Parities) 55, 56 precommitment 114, 143, 147, 149 preemption 548 principal agent models 319, 320, 340 Prisoner's Dilemma, see under game theory privatization 406, 412, 420 prizes for innovation 336, 337, 441 probability of war 113, 114, 132, 142, 143, 151-154 due to imperfect information 113 due to redistribution of resources 114 due to risk taking 114 procurement 3, 5, 6, 8, 9, 434-437, 450, 455, 456 defense procurement 309-343 options 461, 468, 469, 471 productivity 47, 53, 433, 435, 444, 445, 447, 448, 456 profitability 433, 441-444, 455 property rights 20, 37, 193, 196, 204, 209, 210 PSR (Potential Settlement Region) 173, 175, 187 public choice 3, 7, 467, 475, 476, 487, 525 public economics 9 public goods 5, 6, 8, 17, 18, 23, 25, 281, 294 296, 298, 433 pure 91, 92, 94 104, 106

605 public policy 10, 572, 583 public utility regulation problem 317 Purchasing Power Parities, see PPP rational-actor model 216, 236 R&D (research and development) 311, 312, 314 319, 324, 325, 335 338, 406, 407, 409, 412-415, 422, 423, 431-456, 566, 572, 578, 583 defense 10 demand elasticity 451, 453, 456 private and social benefits of 433 subsidies 433, 437, 455 recruiter incentives 356 recruiting 354, 355, 359 redistribution 167, 178, 187 redundant personnel 499 reenlistment 364, 365 decision 352 regional econometric models 502 506 regional impact of defense spending 10 regional models 502, 504, 505, 518 regional multiplier 495-499, 505, 514 regional policies 494, 515, 518 regression diagnostics 271 regressor exogeneity 261 regulation 471 rent seeking 9, 20, 30 rents 332, 337, 353 reputation 226, 227 research and development, see R&D research opportunities 10 reserve forces 349-351, 381, 388-390 resource allocation 17 restructuring 401, 402, 412, 422, 424-427 retention 351, 360-363, 366, 372, 377, 382, 385 387 retirement benefits 361, 385 systems 381 revolutions 193, 195, 204, 210 Richardson model 111, 112, 114-117, 120, 122, 123, 128, 131 see also arms races defense term 115 fatigue term 116 grievance term 116 sanctions 556 screening hypothesis 391 secrecy 47, 65 security definition of 35 equilibrium 39 selectivity bias 392 self-selection models, see under moral hazard SEM (simultaneous-equations models) 255, 262, 269, 290, 291

606 sensitivity analysis 261, 270, 271 settlement 167, 168, 171-173, 175, 184 simultaneous-equations models, see SEM Single European Market 425, 470 SIPRI (Stockholm International Peace Research Institute) 51 see also author index social benefits 48, 51 Socialist System 24 sociopolitical instability index, see SPI sovereign power 193, 204 Soviet bloc 277, 278 Soviet Union 8, 19, 29, 516 military spending 79 Spanish tourism and terrorism 242, 243 specialization 461, 462, 467 SPI (sociopolitical instability index) 303, 304 spin-offs 280, 287, 294, 296, 304, 423, 583 stability, see under arms races standardization 461, 462, 465, 472, 474, 476, 479, 487 Star Wars 20 stock adjustment model 115, 117 Stockholm International Peace Research Institute, see SIPRI strategic behavior 311, 341, 342, 542 strategic competition 28 strategic doctrine 97, 102, 104 strategic interactions 19 strategic trade theory 466 structure-conduct-performance paradigm 424 subsidy 437-441, 467, 544 substitution 367-370, 390 supplier network 498, 502, 518 supply chain 497 systems analysis 17 tax policy 194, 195, 202 team production 384 technical progress 288 technology of struggle 167, 188 terrorism and paid riding 223 cycles 217-220 data sources 216 definition of 215 ITERATE (International Terrorism: Attributes of Terrorist Events) 216-218 state-sponsored 216, 222 trends 216 terrorist attack modes 215, 217, 218 assassinations 217, 244-246 bombings 217 hostage taking 217, 218, 244-246 skyjackings 217, 218, 235-239, 244-246 threats and hoaxes 217, 244-246 terrorists' choice between legal and illegal activities 217

Subject Index terrorists' net payoff 229 threats 169, 171 time-series analysis 216-218 ARMA (autoregressive moving average) 237 Granger causality 80, 241-243, 245, 255, 257, 263, 269 intervention analysis 216, 218, 237 spectral analysis 219 stationarity 221, 245 timing of crises 123 trade theory 527, 535, 542, 558 training 349, 350, 353, 369, 377, 386, 389-391, 393 costs 375-377, 389 transaction costs 482 transboundary externalities, see under externalities transitional economies 486 typology of arms trade 543 UK 578, 579 UN 47, 51 UN Institute for Trade and Development, see UNITAD uncertainty 312, 313, 450, 454, 455, 467 unequal exchange 573 UNIDIR, see author index unit costs 533 procurement 463 production 477 UNITAD (UN Institute for Trade and Development) 582 United States Arms Control and Disarmament Agency, see US ACDA in author index US ACDA, see author index USA 576, 577 VAR (vector autoregression) 6, 80, 216, 218, 241-244 vector autoregression, see VAR Vietnam War 28 volunteer force 373-380 war

15, 18-24, 29, 32, 33, 35, 38, 122, 125, 167-169, 171, 185-187 causes of 130, 131, 172 war arms investment strategy 136-138 warfighting 121 Warsaw Pact 19 weakest link 98, 106 weapon stock 119 weapons acquisition 28, 450, 453, 456 work sharing 472, 473, 476 world arms trade 527-529 World Bank 426, 572 see also author index zero-sum game

567