Brookings Trade Forum 2005 (Brookings Trade Forum)

Brookings Trade2005Forum Offshoring White-Collar Work Susan M. Collins and Lael Brainard editors brookings institution...

Author: Susan M. Collins | Lael Brainard

110 downloads 1624 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Brookings Trade2005Forum Offshoring White-Collar Work Susan M. Collins and Lael Brainard editors

brookings institution press Washington, D.C.

Copyright © 2006 THE BROOKINGS INSTITUTION 1775 Massachusetts Avenue, N.W., Washington, D.C. 20036 ISSN 1520-5479 ISBN-13: 978-0-8157-1284-8 ISBN-10: 0-8157-1284-7

For information on subscriptions, standing orders, and individual copies, contact Brookings Institution Press, P.O. Box 465, Hanover, PA 17331-0465. Or call 866-698-0010. E-mail [email protected]. Visit Brookings online at http://bookstore.brookings.edu. Brookings periodicals are available online through both Online Computer Library Center (contact OCLC subscriptions department at 800-848-5878, ext. 6251) and Project Muse (http://muse.jhu.edu). Authorization to photocopy items for internal or personal use or the internal or personal use of specific clients is granted by the Brookings Institution for libraries and other users registered with the Copyright Clearance Center Transactional Reporting Service, provided that the basic fee is paid to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923. For more information, please contact CCC at 508/750-8400. This authorization does not extend to other kinds of copying, such as copying for general distribution, or for creating new collective works, or for sale. Specific written permission for such copying must be obtained from the Permissions Department, Brookings Institution Press, 1775 Massachusetts Avenue, N.W., Washington, DC 20036; fax, 202/536-3623; e-mail, [email protected].

Brookings Trade2005Forum Offshoring White-Collar Work Editors’ Summary

ix

What Do We Learn from Trade Theory?

james r. markusen Modeling the Offshoring of White-Collar Services: From Comparative Advantage to the New Theories of Trade and Foreign Direct Investment

1

Comments by Alan V. Deardorff and Douglas A. Irwin 24 Discussion 30

daniel trefler Service Offshoring: Threats and Opportunities

35

Comments by Dani Rodrik and Pol Antràs 61 Discussion 66

Exploring the Empirics

j. bradford jensen and lori g. kletzer Tradable Services: Understanding the Scope and Impact of Services Offshoring

75

Comments by Jared Bernstein and Robert C. Feenstra 117 Discussion 127

maria borga Trends in Employment at U.S. Multinational Companies: Evidence from Firm-Level Data

135

desirée van welsum and xavier reif Potential Offshoring: Evidence from Selected OECD Countries

165

Comments by Robert Z. Lawrence and Catherine L. Mann 195 Discussion 201

Offshoring—India’s Role

t. n. srinivasan Information-Technology-Enabled Services and India’s Growth Prospects

203

Comment by Anne Krueger 232 Discussion 236

rafiq dossani Globalization and the Offshoring of Services: The Case of India

241

Comment by Arvind Panagariya 268 Discussion 274

Lessons from Industry Studies

clair brown and greg linden Offshoring in the Semiconductor Industry: A Historical Perspective

279

Comment by Jeffrey T. Macher 323 Discussion 329

rosemary batt, virginia doellgast, and hyunji kwon Service Management and Employment Systems in U.S. and Indian Call Centers

335

Comment by Vivek Agrawal 361 Discussion 366

ravi aron and ying liu Determinants of Operational Risk in Global Sourcing of Financial Services: Evidence from Field Research

373

ashish arora The Emerging Offshore Software Industries and the U.S. Economy

399

frank levy and ari goelman Offshoring and Radiology

411

Discussion 424

What Role for Policy?

lael brainard, robert e. litan, and nicholas warren A Fairer Deal for America’s Workers in a New Era of Offshoring 427 Comment by Lawrence Mishel 448 Discussion 452

kimberly a. clausing The Role of U.S. Tax Policy in Offshoring Comment by Kevin A. Hassett 483 Discussion 486

457

the brookings institution The Brookings Institution is a private nonprofit organization devoted to research, education, and publication on important issues of domestic and foreign policy. Its principal purpose is to bring the highest quality independent research and analysis to bear on current and emerging policy problems. The Institution was founded on December 8, 1927, to merge the activities of the Institute for Government Research, founded in 1916, the Institute of Economics, founded in 1922, and the Robert Brookings Graduate School of Economics and Government, founded in 1924. Interpretations or conclusions in Brookings publications should be understood to be solely those of the authors. Board of Trustees John L. Thornton Chair Strobe Talbott President Zoë Baird Alan R. Batkin Richard C. Blum Geoffrey T. Boisi James W. Cicconi Arthur B. Culvahouse Jr. Kenneth W. Dam Vishakha N. Desai Thomas E. Donilon Mario Draghi Kenneth M. Duberstein

Lawrence K. Fish Cyrus F. Freidheim Jr. Bart Friedman David Friend Ann M. Fudge Jeffrey W. Greenberg Brian L. Greenspun William A. Haseltine Teresa Heinz Samuel Hellman Glenn H. Hutchins Joel Z. Hyatt Shirley Ann Jackson Kenneth Jacobs Suzanne Nora Johnson

Michael H. Jordan Harold Hongju Koh William A. Owens Frank H. Pearl John Edward Porter Steven Rattner Haim Saban Leonard D. Schaeffer Lawrence H. Summers David F. Swensen Larry D. Thompson Laura D’Andrea Tyson Antoine W. van Agtmael Beatrice W. Welters Daniel Yergin

Honorary Trustees Leonard Abramson Elizabeth E. Bailey Rex J. Bates Louis W. Cabot A. W. Clausen William T. Coleman Jr. Alan M. Dachs D. Ronald Daniel Robert A. Day Bruce B. Dayton Charles W. Duncan Jr. Walter Y. Elisha Robert F. Erburu Henry Louis Gates Jr. Robert D. Haas Lee H. Hamilton F. Warren Hellman

Robert A. Helman Roy M. Huffington James A. Johnson Ann Dibble Jordan Vernon E. Jordan Jr. Breene M. Kerr Marie L. Knowles James T. Lynn Jessica Tuchman Mathews David O. Maxwell Donald F. McHenry Robert S. McNamara Mary Patterson McPherson Arjay Miller Mario M. Morino Maconda Brown O’Connor Samuel Pisar

J. Woodward Redmond Charles W. Robinson James D. Robinson III Judith Rodin Warren B. Rudman B. Francis Saul II Ralph S. Saul Henry B. Schacht Michael P. Schulhof Joan E. Spero Vincent J. Trosino John C. Whitehead Stephen M. Wolf James D. Wolfensohn Ezra K. Zilkha

Brookings Trade Forum is a series of annual volumes that provide authoritative and indepth analysis on current and emerging issues in international economics. The series aims to explore questions on international trade and macroeconomics in an interdisciplinary fashion with both practitioners and academics and seeks to gather in one place papers that provide a thorough look at a particular topic affecting international economic policy. Leading experts in the field will contribute to each volume. This eighth issue contains edited versions of the papers and comments presented at a conference held at the Brookings Institution, May 12–13, 2005. This year’s forum focused on the offshoring of high-skilled work. Contributors use trade theory, empirics, industry studies, and policy analyses to illuminate the issues and implications of this complex phenomenon. The conference and journal were made possible by a generous grant from the Alfred P. Sloan Foundation. Additional support of the 2005 Trade Forum came from the William and Flora Hewlett Foundation and the Tokyo Club Foundation for Global Studies.

Coeditors

Staff

Advisers

Contributors

Susan M. Collins, Brookings Institution and Georgetown University Lael Brainard, Brookings Institution Lindsey Wilson, production assistant Janet Mowery, editorial associate Gabriel Chodorow-Reich, research assistant and verifier Eric Haven, research verifier Stephen Robblee, research verifier Nicholas Warren, senior research analyst Jagdish Bhagwati, Columbia University Richard N. Cooper, Harvard University Avinash Dixit, Princeton University Geza Feketekuty, Monterey Institute of International Studies Jeffrey A. Frankel, Harvard University Gene Grossman, Princeton University Gary Horlick, Wilmer, Cutler & Pickering Gary Hufbauer, Institute for International Economics John H. Jackson, Georgetown Law School Paul R. Krugman, Princeton University Maurice Obstfeld, University of California–Berkeley Sylvia Ostry, University of Toronto Kenneth Rogoff, International Monetary Fund Laura Tyson, London Business School Paula Stern, The Stern Group Alan Wolff, Dewey Balantine Vivek Agrawal, McKinsey and Company Pol Antràs, Harvard University Ravi Aron, University of Pennsylvania Ashish Arora, Carnegie Mellon University Rosemary Batt, Cornell University Jared Bernstein, Economic Policy Institute Maria Borga, U.S. Bureau of Economic Analysis

Lael Brainard, Brookings Institution Clair Brown, University of California–Berkeley Kimberly A. Clausing, Reed College Alan V. Deardorff, University of Michigan Virginia Doellgast, Cornell University Rafiq Dossani, Stanford University Robert C. Feenstra, University of California–Davis Ari Goelman, Massachusetts Institute of Technology Kevin A. Hassett, American Enterprise Institute Douglas A. Irwin, Dartmouth College J. Bradford Jensen, Institute for International Economics Lori G. Kletzer, University of California–Santa Cruz Anne Krueger, International Monetary Fund Hyunji Kwon, Cornell University Robert Z. Lawrence, Harvard University Frank Levy, Massachusetts Institute of Technology Greg Linden, University of California–Berkeley Robert E. Litan, Brookings Institution and Kauffman Foundation Ying Liu, University of Pennsylvania Jeffrey T. Macher, Georgetown University Catherine L. Mann, Institute of International Economics James R. Markusen, University of Colorado–Boulder Lawrence Mishel, Economic Policy Institute Arvind Panagariya, Columbia University Xavier Reif, Organization for Economic Cooperation and Development Dani Rodrik, Harvard University T. N. Srinivasan, Yale University Daniel Trefler, University of Toronto Desirée van Welsum, Organization for Economic Cooperation and Development Nicholas Warren, Brookings Institution Conference participants

Susan Aaronson, University of North Carolina Starynee Adams, Brookings Institution Mary Amiti, International Monetary Fund Claude Barfield, American Enterprise Institute Joshua Bivens, Economic Policy Institute Ronald Blackwell, AFL-CIO Mary Jane Bolle, Library of Congress Barry Bosworth, Brookings Institution Chad P. Bown, Brookings Institution and Brandeis University Sydney F. Collins, University of Miami Rishi Daga, Reliance Infocomm Annie Davis, Brookings Institution I. M. Destler, University of Maryland and Institute for International Economics Rebecca Dillender, Bureau of Internal Labor Affairs Howard Dobson, United States Department of Labor Kristin Forbes, Council of Economic Advisers Barbara Fraumeni, Bureau of Economic Analysis Richard Freeman, Harvard University

Conference participants (continued)

Carol Graham, Brookings Institution Gene Grossman, Princeton University Jane T. Haltmaier, Federal Reserve Board Ronil Hira, Rochester Institute of Technology Ned Howenstine, Bureau of Economic Analysis Kent Hughes, Woodrow Wilson International Center for Scholars Graham Ingham, International Monetary Fund Jane Irhig, Federal Reserve Board Karen Johnson, Federal Reserve Board Robert Johnson, Cargill Douglas Kaden, Oak Hill Capital Aslihan Kes, International Center for Research on Women Judy Knepper, United States Government Accountability Office Ralph Kozlow, Bureau of Economic Analysis Martha Laboissiere, McKinsey and Company Frank Langfitt, National Public Radio Thea Lee, AFL-CIO Philip I. Levy, Council of Economic Advisers Paul Magnusson, Business Week Magazine Dalia Marin, University of Munich Richard Mataloni, Bureau of Economic Analysis Lawrence McNeil, Bureau of Economic Analysis Emily McWithey, Brookings Institution Theodore Moran, Georgetown University Janet Norwood, New York Conference Board Gail Pesyna, The Sloan Foundation Anbihn Phan, Princeton University David Ratner, Economic Policy Institute J. David Richardson, Syracuse University and Institute for International Economics Fernando Robles, George Washington University Howard Rosen, TAA Coalition Jeffrey Russell, Duke University Kenneth Ryder, National Academy of Public Administration Isabelle Sawhill, Brookings Institution Gary Saxonhouse, University of Michigan Phillip Swagel, American Enterprise Institute Strobe Talbott, Brookings Institution Daniel Tartullo, Georgetown University Law Center Michael Teitelbaum, The Sloan Foundation Edwin M. Truman, Institute for International Economics Robert Vilhauer, Boeing Company David Walters, Office of the United States Trade Representative Timothy Wedding, United States Government Accountability Office Shang-Jin Wei, International Monetary Fund Obie Whichard, Bureau of Economic Advisers Beth Anne Wilson, Federal Reserve Board Karen Wilson, Boeing Company Loren Yager, United States Government Accountability Office William J. Zeile, Bureau of Economic Analysis

LAEL BRAINARD SUSAN M. COLLINS

Offshoring White-Collar Work: Editors’ Summary

O

ffshoring of services burst into America’s public consciousness in 2003, raising concerns about the nation’s future competitiveness. For the first time, highly skilled white-collar workers in the United States perceived themselves to be in direct competition with lower-paid foreign workers. Viewed from across the Indian Ocean, services offshoring appeared to promise accelerated development— where poor countries could compete successfully with much richer ones by focusing their talent pool on a select group of high-value activities. For the first time in many decades, mainstream economists launched a serious debate over the possibility that trade could be zero sum (albeit only in special circumstances). With the media stoking the popular imagination through dramatic, though often anecdotal or inconsistent “statistics,” the Brookings Institution embarked on an effort to analyze the offshoring phenomenon. First, seeking to separate fact from speculation, we convened a full-day conference in 2004 that focused narrowly on the available data, their implications, and their limitations. We then launched a broader research project aimed at illuminating offshoring from a variety of complementary views: those of theory and empirics, industries and labor markets, and developed and developing countries. Participants with diverse perspectives presented their analyses at the Brookings Trade Forum Conference in May 2005. This volume contains the fourteen revised papers, as well as invited commentary by fourteen additional experts and summaries of the conference discussions. Here we highlight some of the key themes and conclusions that emerged from the project, and then briefly summarize the papers. The term offshoring has itself been the source of some confusion. In this volume, we distinguish between questions of location (whether an activity is undertaken in the home market or offshore) and questions of ownership (whether an activity is undertaken within an enterprise or is outsourced to an arm’s-length ix

x

Brookings Trade Forum: 2005

provider. Thus, by offshoring we mean the assignment of part or all of the value chain to an offshore (foreign) location, where an activity could be done either within the firm (in-house) or by a third party (at arm’s length). In contrast, we use the term outsourcing to mean the assignment of services to a third party, whether domestic or foreign. The research focused on the new dimensions of offshoring, which reflect both sector and occupation: offshoring of services, as well as offshoring of white-collar work. The contributors to this volume demonstrate that existing economic theory can go a long way toward capturing key dimensions of the services offshoring phenomenon. In particular, trade theory can explain why production processes that rely on skilled workers may migrate to locations in developing countries where such workers are relatively scarce; lack of complementary factors such as know-how may make them relatively inexpensive in these markets. In many respects, the issues raised by the new wave of offshoring parallel those that arose from the globalization of manufacturing that began some decades ago. But recent developments affect workers who tend to be more educated and to earn higher salaries. They often involve only a fragment of the value chain. And because of the nature of the relevant processes, recent theory suggests that levels of institutional development might constrain the extent to which developing countries can participate. Under quite sensible assumptions, standard trade models do not necessarily predict that high-wage economies such as the United States will gain from greater offshoring of services. One contribution of the project is to clarify the channels through which gains and losses might occur. Since existing evidence consistently shows that relatively few service and white-collar jobs have been offshored to date, concerns focus on what might happen in the future. How much of the labor force in the United States and elsewhere is really vulnerable? As the 2004 workshop made clear, the answer is difficult to discern from existing data. Thus a contribution of this project is to present and discuss alternative approaches to constructing such indicators—one based on quantitative analysis across U.S. industries and occupations and the other based on a more subjective assessment for a large number of OECD countries. The results suggest that more of the industrial country workers exposed to international trade are employed in services than in manufacturing, and show that those workers suffer substantial losses when they are displaced. They also highlight that trade goes in both directions and that foreign firms locating domestically provide job opportunities for workers in tradable service sectors. The spread of offshoring to services raises questions that are at least as important and consequential for developing countries as for the advanced ones. In this context, India emerges as a focus of both public and academic attention. Its econ-

Lael Brainard and Susan M. Collins

xi

omy is a study in contrasts—home to both a world-class information technology (IT) services sector and to one-quarter of the world’s poorest. Contributors to the project highlight both lessons and debates from recent research. There is broad agreement that India’s IT success is founded on a pool of highly educated English speakers and that IT services have thrived by working around poor infrastructure and taking advantage of episodic openings in the restrictive policy regime. Interestingly, among the potential constraints to future growth is a limited supply of suitably skilled workers, not concerns about India’s institutional weaknesses (such as corruption and patchy intellectual property enforcement). Significant disagreement emerged over the importance of India’s diaspora community to the IT success; whether the experience in IT will drive broader economic liberalization; and whether India could have reaped even greater dividends by pursuing a broader growth strategy premised on access to high-quality primary education, public investment in infrastructure, and sweeping policy reform. Analysis of individual service sectors yields rich insights into the offshoring phenomenon. Software services have a long history of offshoring, particularly in India, followed by call centers, semiconductor design, and business process outsourcing (BPO). Radiology is an example of a relatively new focus of public attention. Services differ profoundly from manufacturing in that quality inheres in the underlying processes rather than in a final physical output. As a result, the decision to offshore depends on a firm’s ability to specify and monitor processes and on its choice of internal or arm’s-length organizational forms. In some sectors—notably call centers—a cookbook approach to offshoring appears to prevent foreign providers from fully utilizing their skilled workers or achieving maximum efficiency. Unexploited opportunities for arbitrage also are present in the BPO sector, where field surveys suggest the striking finding that advanced economy buyers and developing economy suppliers perceive complexity very differently. For instance, while processes requiring intensive algorithmic computation tend to be rated as complex by managers in the United States and the United Kingdom, managers in India and Singapore give high complexity ratings to processes requiring judgmentdriven communication. Concerns about the loss of high-value services appear to be overblown in software services, call centers, and semiconductor design: the vertical decomposition of these services preserves the highest-value activities in the home market, while shedding lower-value processes to overseas providers. And professional credential requirements interact with other factors to make extensive offshoring of radiology services unlikely to come to pass. What is the policy agenda that emerges from the spread of offshoring into services? The papers in this volume focus on two areas: social insurance and corporate taxation. The United States lags behind other wealthy economies on

xii


social insurance. With offshoring and technological change accelerating the rate at which workers’ job specific-skills depreciate, there is a strong case to be made for wage loss insurance, which would encourage workers to broaden their employment search and go back to work more quickly, while defraying the cost to employers of hiring and providing on-the-job training to new employees from different sectors. The policy agenda should also address economy-wide cost disadvantages. In this context, a case can also be made for simplifying the U.S. system of taxing multinationals’ worldwide income and for lowering the corporate tax rate while broadening the base.

Part I: What Do We Learn from Trade Theory? In the first paper of the volume, James Markusen examines offshoring of white-collar work from the perspective of a trade theorist. His approach is to construct a set of alternative conceptual frameworks, instead of focusing on a single detailed model based on one particular framework. The paper provides new and useful benchmarks for understanding the effects of offshoring on welfare and factor incomes among industrialized economies (which he calls the North) and emerging economies (the South). Markusen begins by identifying the real-world features of offshoring that he believes models should consider. First, he argues that offshoring of white-collar services is largely about technical and institutional innovations that allow new things to be traded, and not about marginal liberalizations that allow more and less costly trade in existing traded goods and services. Further, the newly traded services tend not to be final goods, but instead parts of a production chain that is decomposed or “vertically fragmented” geographically. These offshored services can be either upstream (such as software design) or downstream (such as call centers that offer after-sales services). And the offshored service activities in his model can be fragmented from the production of a final manufactured good as easily as from the production of a final service. Second, borrowing from mainstream trade theory, Markusen argues that it is important to incorporate differences in both technology across productive activities and factor supplies across countries. Fragments of the production process often differ in their factor intensities (that is, their relative use of unskilled labor, skilled labor, physical capital), while economies often differ in their factor endowments (the relative abundance of skilled and unskilled labor and physical capital). This may suggest, for example, that if production of a certain service


xiii

requires mid-level skills, its production may be best suited to locations where such labor is cheap. However, the simplest off-the-shelf trade model in which factor intensities interact with factor endowments (the two-factor Heckscher-Ohlin model) implies that factors are always expensive where they are scarce. This is in stark contrast to the empirical fact that skilled labor is relatively cheap in some countries where it is also relatively scarce, such as India or China. Thus Markusen suggests expanding the basic model to incorporate an additional factor, which he labels “know-how.” The know-how could be physical capital, such as computers, networks, and satellite links, or knowledge capital, such as managerial, organizational, or marketing expertise. He makes the key assumption that know-how is complementary to skilled labor, thereby creating a potential missing input for countries otherwise well suited to the production of skill-intensive fragments. This feature provides an explanation for the “puzzle” that scarce factors can be relatively cheap. Markusen presents a series of five simple models. The first is a standard twogood, two-factor, two-region Heckscher-Ohlin model with perfect competition. His innovation is to permit fragmentation of the initially more skill-intensive good, leading to a shift of the service activity from North to South—the offshoring of services. The second model incorporates “know-how,” the third factor assumed to be located primarily in the North and a necessary complement to the skilled labor used in services. Markusen’s third model omits know-how but introduces multinational firms that produce with increasing returns to scale under conditions of imperfect competition. The fourth model combines the missing input model with the multinational production structure. Markusen also briefly discusses a fifth, somewhat more speculative, framework that considers some issues related to outsourcing—or whether offshoring is conducted within or outside the ownership boundaries of the multinational firm. Markusen’s preferred framework is the missing input model, particularly with its extension to multinationals. His models illustrate how the ability to offshore white-collar services can exploit this gap between skills and know-how, bringing modern corporate knowledge to these skilled workers. Although there is clearly a benefit to the world economy overall, the model also highlights that there are winners and losers, and in particular, that some northern groups may be vulnerable. The implications of offshoring that these models suggest to be unambiguous are welfare gains for the South overall and welfare gains for skilled labor in the South. The implications for unskilled labor in both regions, for skilled labor in the North, and for overall welfare in the North are all mixed.

xiv


The second paper, by Daniel Trefler, elaborates on some of the issues Markusen considers in his fifth model. A central point of his analysis is that institutional development in potential host economies is likely to constrain the extent to which activities can be offshored, which he defines as the use of workers located abroad to provide sophisticated services to U.S. customers. Trefler sees two key features of offshoring that distinguish it from other types of international trade. He identifies each feature with a set of concerns from the North. First, service offshoring uses some of the most dynamic information and communication technologies (ICT). It thus has implications for the corporate innovation strategies that lie at the heart of U.S. competitiveness policies. He argues that this feature raises the concern that U.S. firms may be crowded out of the most innovative lines of business, and that there is already some evidence of this among Indian multinationals such as Satyam, HCL, and Tata. Second, Trefler notes that the employment of highly skilled white-collar workers in low-cost countries such as India raises the concern that offshoring may displace “good” U.S. jobs and depress salaries of high-paid workers, both of which would reduce the incentives of Americans to invest in their own human capital. A related concern is that the disruption from increased service offshoring may make it less worthwhile for firms to make long-term investments in their best workers. Although existing evidence suggests that service offshoring is currently small, Trefler stresses that many worry that, over the next ten to twenty years or more, imports from China and India will devastate the United States. He believes this concern is misplaced for two reasons. First, it ignores the law of comparative advantage, which states that no country can export all goods. He notes that similar concerns were expressed about Japan, where wages in 1959 were 10 percent of U.S. levels. However, the law of comparative advantage does not rule out the possibility that China and India will export high-tech goods and services, leaving the relatively low-skilled activities for Americans. This raises Trefler’s third and perhaps central theme: the crucial role of institutions. Current thinking about innovation-based long-term growth focuses on (1) institutions that protect property rights from preying politicians and bureaucrats, (2) institutions that provide a fully functioning legal framework for arm’slength transactions, and (3) institutions that balance the needs of innovators inside the corporation against the needs of investors outside the corporation. Trefler observes that these institutions are only beginning to take shape in China, India, and many of the other emerging markets that are potential offshoring hosts, and argues that they are unlikely to evolve quickly, even over a quartercentury horizon. Thus, Trefler concludes that China and India are a long way from being the world’s innovation giants.


xv

Finally, Trefler believes that most of the sensible policies aimed at fostering American competitiveness in the service offshoring market are investmentpromoting framework policies. These encourage American workers, firms, and governments to invest in building productive assets such as human capital and new technologies. While acknowledging that such framework policies address a whole host of domestic competitiveness issues and are not unique to offshoring, Trefler argues that they should still be seen as appropriate ways to address concerns raised by services offshoring. He advocates providing more assistance to workers who are displaced, both skilled white-collar workers and unskilled workers displaced by low-end manufacturing imports. Many of these issues are addressed in the paper on wage insurance by Lael Brainard, Robert Litan, and Nicholas Warren.

Part II: Exploring the Empirics The next three papers in the volume address empirical issues. Traditionally, economists have treated services as nontraded activities. However, as globalized production expands beyond manufacturing to include a variety of services, this perspective is obsolescent, and the need for better measures of tradability is clear. Existing data on international flows of services provide at best an incomplete picture of the magnitude and dimensions of this growing phenomenon. J. Bradford Jensen and Lori Kletzer present a new method for identifying which service activities are vulnerable to international trade. Their approach distinguishes occupations as well as industries and enables them to examine the implications of globalizing the production of services for domestic employment and job loss. Specifically, Jensen and Kletzer use the geographic concentration of service activities in the United States to identify which service activities are traded domestically. They then classify activities that are traded domestically as potentially tradable internationally. Using the identified industries and occupations, they develop estimates of the number of workers who are in tradable activities in all sectors of the economy. This enables them to compare the demographic characteristics of workers in tradable and nontradable activities and the employment growth in traded and nontraded service activities. Their approach to measuring tradability of services relies on the economic intuition that production of services will be geographically concentrated only if the services are traded. This builds from the observation that production of traded manufactured goods tends to be geographically concentrated (to capitalize on

xvi


increasing returns to scale, access to inputs like natural resources, etc.), while production of those that are not traded tends to be more widely distributed. Indeed, they show that the geographic concentration of some service activities within the United States is nearly as great as in manufacturing. Their paper highlights three main findings. First, they conclude that the number of workers potentially exposed to international trade in services is actually larger than the number of exposed workers in manufacturing. Second, workers in tradable sectors have higher skills and significantly higher earnings. The higher earnings do not appear to be solely a result of higher skill levels—in regressions controlling for observable characteristics, workers in selected tradable service activities have 16–17 percent higher earnings than similar workers in nontradable activities in the same sector. Third, employment in tradable activities grew more slowly in the period 1998 to 2002, but Jensen and Kletzer show that this was due primarily to employment losses in manufacturing. Within services, employment in tradable and nontradable activities grew at similar rates except at the lowest end of the skill distribution. Average employment growth was negative in low-skill tradable industries and occupations but positive (though low) in nontraded low-skill services. Jensen and Kletzer use the 2004 Displaced Worker Survey to examine the scope and cost of involuntary job loss. Briefly, they find some evidence that displacement rates are higher from tradable than from nontradable service industries. The difference is most notable in the Information sector; however, the authors note that this could simply represent displacements associated with the tech/telecom bubble and may not reflect offshoring. They also find somewhat higher displacement rates from tradable than from nontradable white-collar occupations. The same gap is not evident for blue-collar workers. Consistent with the employment characteristics, they find that workers displaced from tradable service activities are more educated, and have higher earnings, than workers displaced from nontradable activities. They are also more likely to be reemployed. Finally, they document that job loss from tradable and nontradable service activities is costly to workers (they typically endure a period of unemployment and are unlikely to earn as much in new jobs as they did in their former jobs). Generalizing from what is known about manufacturing worker job loss, the authors speculate that lower levels of job tenure and higher levels of educational attainment may be advantages in seeking reemployment. They would favor a less porous safety net, provided, for example, by extending Trade Adjustment Assistance (TAA) to services workers and extending wage insurance beyond TAA.


xvii

The next two papers, by Maria Borga and by Desirée van Welsum and Xavier Reif, address evidence of existing and potential offshoring by U.S. and selected OECD multinational companies. Borga uses firm-level data from the Bureau of Economic Analysis (BEA) to examine the evidence on the extent of offshoring by U.S. parent companies to their foreign affiliates, and then to determine if, and how, offshoring is associated with changes in U.S. parent employment. Using survey data from 1,117 U.S. parent companies, she studied their offshoring behavior by examining the shares of imports of goods from affiliated and unaffiliated parties and of imports of services from affiliated parties in their total purchased inputs. Between 1994 and 2002, average employment growth was 39 percent for U.S. parent companies, a result of both mergers and acquisitions and the expansion of existing operations. Examining parents’ offshoring behavior, she finds distinctions between goods and services as purchased inputs, as well as between purchases from affiliates versus nonaffiliates. Overall, she concludes that the vast majority of U.S. parents’ purchased inputs are acquired from domestic sources, not imports in both 1994 and 2002. The average share of foreign affiliates’ sales to the local market increased, reaching 78 percent in 2002, while shares of sales to both third countries and the United States declined. The high share of sales to local markets demonstrates the importance of serving local customers in the parent’s decision to invest abroad. This suggests that serving local markets was an increasingly important motive for U.S. MNCs’ overseas expansion. Borga also compares the data for parent companies with employment gains and those with employment losses. The average increase in employment (82 percent) is roughly the same as the average increase in sales (84 percent) for parents with employment gains. But for parents that lost employment, the average decrease in employment was 30 percent, well above the 10 percent average decrease in sales. Both sets of parents increased their reliance on purchased inputs. However, those parents whose employment declined strengthened their ties to their affiliates by increasing their reliance on imported goods and, to a lesser extent, on imported services from them. On the other hand, parents that gained employees had a smaller increase in reliance on affiliated imports of goods and reduced their reliance on affiliated imports of services. The paper then turns to a decomposition of changes in parent employment among three factors: the change in output, the change in labor productivity, and the change in the use of purchased inputs in production. This decomposition shows that the increase in labor productivity was the most important factor for those parents that lost employment, followed by the loss of output and by the

xviii


increased use of purchased inputs. For those parents that gained employment, increased output more than accounted for the change in employment. Because most of the inputs purchased by parents are supplied by domestic firms, rather than imported, the loss of jobs due to greater imports accounted for less than one-third of the total change in employment attributed to the greater use of purchased inputs. In the final section of the paper, Borga considers a variety of correlations, including those between employment and output, employment and labor productivity, employment and reliance on purchased inputs, and imports by parents and parents’ employment. In their paper, Desirée van Welsum and Xavier Reif also focus on the growing tradability of services and the implications for white-collar jobs previously shielded from international competition. Like Jensen and Kletzer, their work is motivated by the absence of any official data measuring the extent of the offshoring of services activities. Instead of a statistical approach, they consider a checklist of attributes and use their own judgment to identify tradable occupations. The authors use data on trade in services and occupational employment to estimate the current extent of globalization of services in OECD economies and its potential growth. They find that in many countries often mentioned in the offshoring debate both exports and imports of business and computer and information services have grown rapidly. OECD countries still account for over 75 percent of exports of these services, but their share is declining. Drawing on their earlier detailed analysis of occupational data for selected OECD countries, Van Welsum and Reif seek to determine the share of total employment that could potentially be affected by the international sourcing of IT- and ICT-enabled services. Their measure suggests that close to 20 percent of total employment could potentially be affected by offshoring, particularly in business services (for example, accounting, consulting), financial services, and research and development. The authors provide a simple descriptive regression analysis of the relationship between the share of employment potentially affected by offshoring and other economic and structural developments for OECD economies between 1996 and 2003. They do not find any systematic or significant evidence that either net outward investment or imports of business services are associated with declines in the share of employment potentially affected by offshoring—at least at the aggregate level. However, they do find that exports of business services are positively associated with the share of employment potentially affected by offshoring, suggesting that increases in demand and production have also raised demand for these types of ICT-using occupations. Other key factors found to be


xix

positively associated with the share of employment potentially affected by offshoring are the comparative size of the service sector, the growing share of ICT investment in total fixed investment, and human capital.

Part III: Offshoring—India’s Role The next two papers, by T. N. Srinivasan and Rafiq Dossani, consider the economic implications of services offshoring for India. T. N. Srinivasan finds the contribution of information technology services to Indian growth to be significant. He traces the spectacular development of India’s IT sector to several sources, especially public investment in higher education and the creation of elite engineering schools. The policy regime was also influential: first as a restrictive force until the mid-1980s, subsequently as an enabling factor before the reforms of 1991, and thereafter as a proactive supportive force. Telecommunications reforms, government incentives for software technology parks (STPs), foreign investment, and venture capital all played prominent roles. Srinivasan argues that the Indian IT diaspora concentrated in Silicon Valley was a significant factor in the rise of India’s IT sector, although Rafiq Dossani finds evidence for the opposite view. Srinivasan notes that a significant number of Indian engineers who held senior positions in U.S. companies in the late 1990s subsequently helped persuade their senior management to establish operations in India. Moreover, the diaspora’s influence is spreading beyond the narrow confines of the Indian IT industry to the broader contours of India’s economic development and growth. Srinivasan includes a brief analytical discussion of the influence of IT in the growth process and as a source of dynamic comparative advantage. IT services are in effect universal intermediates, which are essential to any production activity and possibly most, if not all, consumption activities. Any technical progress in the IT sector is reflected first in productivity gains (or cost reductions) in the IT sector. Other sectors begin to experience productivity gains as they invest in equipment and processes to take advantage of the new lower-cost information technology, which diffuses the total factor productivity (TFP) gains gradually over time. Srinivasan next examines the prospects for and constraints on India’s fulfilling the high expectations for IT-led growth. The spectacular growth of software exports and information-technology-enabled services and business process outsourcing (ITES-BPO) has been underpinned by the relative abundance of relatively low-wage English-speaking skilled workers. Srinivasan notes that current

xx


trends could lead to excess demand for ITES/IT workers, which would put upward pressure on compensation. If the share of Indian graduates employed in the industry does not decline, the implied annual growth rate of 11.3 percent is far below the 35 percent growth in the value of India’s ITES/IT output projected by India’s National Association of Software and Service Companies (NASSCOM) for 2003–12. Any increase in the compensation of ITES/IT workers in India with no change in their productivity would cut into India’s cost competitiveness. While India is likely to remain competitive for the foreseeable future at the low end of the skill spectrum, India’s competitive edge is likely to erode at the higher end of the spectrum, as Indian wages rise relative to U.S. wages. Already, worker turnover is high, particularly in smaller lower-value-added firms, which should put upward pressure on compensation. Potential future competitors include Bangladesh, Ireland, Pakistan, Sri Lanka, and the Philippines, all of which have pools of English-speaking workers suitable for ITES/IT employment, although only Bangladesh currently has wages below India’s. Competition from China also cannot be ruled out, since English-speaking ability can be acquired. Srinivasan believes the IT sector can make a significant contribution to India’s future growth, but cautions that accelerating the rate of GDP growth to 8 percent or more per year and sustaining it for several decades is a necessary though not sufficient condition for achieving the overarching objective of eradicating poverty. He also believes that the success of the IT sector can help to broaden and strengthen political support for reforms in other sectors—if three things happen. First, politicians must recognize the link between reforms and the success of the IT sector. Second, a much greater proportion of the population must experience the benefits from efficient and inexpensive IT services. Third, political differences over the pace and sequencing of further reforms will have to be resolved. In the next paper, Rafiq Dossani examines India’s experience with providing offshored services to developed nations, where certain components of services exporting have been well established for decades. India has catapulted itself in recent years into a leadership position among developing economies on services exports related to information technology. Dossani’s analysis offers central insights into how the services component of international trade has expanded and how a developing country like India could succeed in exporting services to developed countries. Dossani begins with a discussion of how technology has helped to overcome the intrinsically greater obstacles to trade in services relative to manufactured goods. First, digitization allowed the conversion of service flows into stocks of


xxi

information, making it feasible to separate and store a service. The expanded ability to subdivide tasks reduced costs by offering the possibility of preparing the standardized components with lower-cost labor and, possibly, at another location. Second, digitization permitted noninformation service flows to be converted into information service flows—for instance, replacing the need for inperson sampling in a showroom with virtual sampling of goods delivered over the Internet. Third, costs associated with transmitting digitized material dropped significantly. Services such as writing software programs, which were offshored to India in the early 1970s, were enabled by digitized storage and, in the 1980s, by the standardization of programming languages. Still later, as digital transmission costs fell in the 1990s, even nonstorable services, such as customer care, could be offshored. Advances in information technology made the recent growth in offshoring possible by parsing the provision of certain services into components requiring different levels of skill and interactivity. Dossani finds that the initial impact in India of services offshoring was to generate high and growing levels of employment. However, the low-value-added and low-skill work that was being produced also provided few barriers to entry and was subject to automation. The resulting competition and price deflation mimicked the situation in manufacturing exports in developing countries and raised the likelihood that asymmetries of globalization could be repeated in services exports from developing countries. Dossani presents a multifaceted explanation of India’s success with offshoring services. Local entrepreneurship and a high level of infant industry protection allowed the Indian IT industry to reach a high growth path and allowed local skills to keep pace with global changes. A key advantage appears to be widespread education in the English language. Other institutional advantages are India’s mature judicial system, its conformance with WTO obligations, and a history of successful private enterprise that provided the talent for initiating and managing complex service projects. In contrast, China has better infrastructure but lacks a history of private entrepreneurship, a large population with knowledge of English, and a mature judicial system. Dossani disagrees with the widely held view that spending by the government on education was a key contributor to the success of offshoring, and also argues that the global Indian diaspora has been largely noninfluential, except during the past few years. He highlights the tendency for higher-stage work to remain in the developed countries. This is due to the lack of domain skills in India—a consequence of protectionist policies and the fact that India has no domestic demand for highend services to promote the development of such skilled workforces. Dossani

xxii


does say that recent reforms intended to attract transnational corporations and transnationally trained diaspora could change the environment by introducing domain skills. Additional obstacles are technology, the need to protect proprietary knowledge, economies of scale and time, and mission criticality. Even when offshoring is possible, the in-house multinational is a favored organizational form in most cases. This enables much of the rents from sophisticated work to be captured in the developed country even though the work may be done in India.

Part IV: Lessons from Industry Studies The next five papers provide a fascinating set of industry-level studies of offshoring in diverse service sectors. Clair Brown and Greg Linden examine semiconductor design, which is a frequently cited component of the new wave of services offshoring. Semiconductor (or chip) companies were among the first to invest in offshore facilities to manufacture goods for imports back to the United States. Because meaningful data about the impact of the offshoring of chip design (and even of manufacturing) are limited, Brown and Linden rely on a more qualitative analysis for their key points. They conducted dozens of interviews with engineers and managers at numerous semiconductor and related companies in the United States, Asia, and Europe over the past twelve years. Their research also incorporates the rich store of publicly available information in trade journals and company reports. After briefly describing the stages of semiconductor production and their analytical framework, they examine the offshoring of assembly jobs, manufacturing, and design jobs. They also discuss what their conclusions mean for the United States. Before addressing semiconductor design directly, Brown and Linden analyze the impact on the U.S. semiconductor industry of the offshoring of semiconductor assembly and fabrication. They argue that the initial concern about losing domestic jobs in both stages turned out to be unfounded, as the industry used the situation to its competitive advantage by becoming cost competitive (through assembly offshoring) and by developing the “fabless” (design-only) sector (through foreign outsourcing of chip fabrication or manufacturing). Brown and Linden then analyze the ongoing offshoring of design jobs and compare this stage with the two stages that preceded it in order to explore the possible impact on domestic jobs and the U.S. semiconductor industry.


xxiii

In the second industry study, Rosemary Batt, Virginia Doellgast, and Hyunji Kwon assess call centers through surveys of Indian and U.S. companies. To date, the depiction of this emerging sector has been based largely on anecdotal evidence rather than systematic empirical investigation of business practices— either in the United States or in India. Batt and her coauthors seek to map the extent of variation in service management and employment strategies among U.S. in-house, U.S. outsourced, and foreign offshore call centers that provide similar services to American customers. In addition, they test the impact of ownership status and firm strategies on worker turnover. The authors begin with a literature review, which shows that service management strategies and employment systems vary substantially—from professional approaches to service to highly transactional or cost-driven ones. In addition, work and employment systems typically are differentiated according to the level of education and training required; the level of discretion and collaborative problem-solving embedded in the design of work; and the level and type of compensation system designed to motivate effort. There are several straightforward implications from this literature on work design and employment systems. Companies are more likely to retain in-house those services that are more complex, involve customer transactions that are more nuanced or uncertain, and involve highly valued customers. In order to meet the demands of these types of products and customers, they are likely to focus on service quality and customization and to adopt a more professional approach to service than a subcontractor would. Centers operated by subcontractors, both in the United States and offshore, by contrast, are more likely to compete on costs by offering lower wages and benefits, using more standardized work processes, and closely monitoring performance. From the literature review, Batt, Doellgast, and Kwon develop several testable predictions differentiating U.S. in-house, U.S. outsourced, and foreign offshore establishments. In their sample of establishments, in-house centers tend to adopt a more coherent quasi-professional approach to service interactions than outsourced and offshore sites; in-house jobs are characterized by relatively higher levels of initial investments in training and pay, discretion, and problemsolving opportunities. In offshore centers, by contrast, workers have somewhat higher levels of formal education and receive more initial training than in-house centers but have fewer opportunities to make choices or solve problems. Further multivariate analyses show that U.S. outsourced and offshore centers have significantly higher quit rates. Ownership status is an important driver in the choice of management and employment practices, with U.S. outsourced and offshore

xxiv


centers more constrained to follow standardized operating procedures and performance monitoring. At the level of managerial policy, the authors conclude that the extensive use of routinized work processes in call centers leads to high turnover, which limits options for customization and is associated with lower service quality and productivity. Moreover, to the extent that call centers hire college-educated workers, the highly constrained and monitored work system creates an inefficient use of human capital. The underutilization of human capital represents a substantial loss for Indian subcontractors, who are paying for skills that they are not using. To the extent that companies have complex service offerings or want to compete on the basis of service differentiation, quality, or customer loyalty, they are likely to retain customer interactions in-house, consistent with the transaction costs perspective and core competency argument. To date, this appears to be what most U.S. corporations are doing: after two decades of rapid growth of U.S. call centers, most industry estimates are consistent with the authors’ finding that less than 15 percent of U.S. call centers are run by third-party subcontractors; and only a tiny fraction have moved offshore. However, for those transactions that are simple and codifiable, Batt and her coauthors predict that companies are likely to continue expanding their operations offshore. Their data suggest that the strategy of outsourcing operations to U.S. subcontractors is likely to be a transitory one, as the modest reductions in labor costs may be offset by the high costs of turnover and low levels of employee skill. According to this scenario, the U.S. subcontracting sector, which grew dramatically in the 1990s, will be the hardest hit by Indian competition. The scenarios also depend on human resource development. In India, there is evidence that demand is outstripping the short-run supply of skilled labor in call center cities such as Bangalore and Chennai. Thus, there is a need for the Indian government to invest in the human resource infrastructure. The next industry study, by Ravi Aron and Ying Liu, investigates the offshore outsourcing of business processes in financial services. Aron and Liu’s findings are based on four years of field research and data collection from firms that provide offshore outsourcing services in India, Mauritius, Singapore, and Thailand, and from clients that buy these services in the United States and the United Kingdom. Operational risk is central to considerations of offshoring. Although the business media often claim that process complexity—the converse of codifiability—is the primary reason for operational risk, Aron and Liu’s field research finds that process complexity actually means different things to different managers. Managers in the West perceive complexity very differently from man-


xxv

agers in the countries that provide offshoring services (India and Singapore). Managers in the United States and the United Kingdom tend to rate as complex those processes that call for analytical skills, algorithmic computational intensity, and many subtasks involving quantitative analysis. In contrast, managers in India and Singapore rate as highly complex processes where the work is not easily codifiable and requires judgment-driven communication and contextsensitive inference, and as less complex those that require quantitative analysis and algorithmic computational work. This astonishing finding leads Aron and Liu to ask if there might be a market for complexity arbitrage. Could offshoring release not just gains from wage arbitrage but also from classic specialization in relative comparative advantage? How much a firm gains from offshore outsourcing depends on how well it is able to manage operational risk. To measure this, Aron and Liu analyzs the factors that contribute to operational risk (from survey data). They use a “knowledge continuum” to describe the nature of different stages of information work and how that determines operational risk. They then regress the magnitude of observed operational errors against several of these attributes of the process, the outsourcing contract, and the workforce. Two factors have the greatest influence on operational risk. First, as the work involved in executing offshore processes becomes more codifiable, operational risk declines. Processes that are not easily codifiable or for which the agents need deep context-sensitive understanding of how the process is to be executed are more prone to operational errors. Second, when the buyer and provider of services can agree on a precise and unambiguous set of metrics of process quality, the resulting operational risk is low. However, when process quality itself is open to subjective interpretation, the operational risk is higher. These findings help to shed light on the optimal governance structure for sourcing different kinds of processes. Aron and Liu propose a governance structure that they call the extended organizational form (EOF), where (1) the buyer contracts the production process to the provider; (2) the buyer can inspect the provider’s output quality after production; and (3) the buyer’s managers can also exercise partial managerial control over the provider’s agents by monitoring the quality of process execution during the production. This hybrid mechanism allows buyers to exercise some managerial control across the boundaries of the firm without waiting for a process cycle to be completed. The paper concludes by comparing the efficiency in sourcing offshored services of the traditional in-house hierarchy, market-based outsourcing, and the EOF hybrid. The analysis shows that for relatively complex processes (as rated by the providers), the EOF is indeed the optimal governance choice and holds

xxvi


the potential to unlock gains from wage and complexity arbitrage associated with offshore outsourcing. This organizational form contrasts with the more traditional hierarchical form used for the offshoring of call center work but may also discourage innovation that would improve quality and efficiency. In his paper, Ashish Arora examines the growth and evolution of services offshoring in the global software industry. This experience is particularly interesting because software was one of the first skill-intensive industries for which production moved to relatively low-wage countries. This short paper draws from Arora’s recently coauthored book, which provides extensive additional detail. Broadly speaking, software activities can be divided into design, coding, and maintenance. Arora argues that the latter two of these are analogous to production and entail relatively low-end tasks. These, not design, account for most of the offshoring to date. He distinguishes between those who work in the software industry and the much larger number who work in software occupations outside of the “core” computer equipment and software services industries. After providing an overview of the global software industry in terms of employment, sales, and exports, Arora describes how three countries, India, Ireland, and Israel, have emerged as centers of offshored software service through a combination of excess skilled labor, key innovations at the level of the firm, and good timing. In contrast, Brazil and China, the other two newcomers, have pursued a very different strategy, relying considerably less on exports, at least to date. Brazil has relied on a sophisticated domestic banking industry to generate demand, hoping that it will lead to the creation of an internationally competitive software industry. China appears to be following a more traditional importsubstitution model. Arora does not believe that the fast growth of export-oriented sectors in lower-wage countries yet threatens the supremacy of the United States as a producer of technology services. The continued importance of a close relationship between the producer and end-user of a software service, as well as the U.S. advantage in innovation-spurring institutions such as venture capital, suggests to him that offshored software services will remain concentrated in relatively lowvalue-added activities. But he recognizes that the sensitivity of this conclusion to factors specific to software production may leave workers in other potentially offshorable occupations feeling less secure. In “Offshoring and Radiology,” Frank Levy and Ari Goelman conclude that the much discussed reading of radiology images offshore by “cheap foreign doctors” is, to date, no more than an urban legend. Unlike software professionals, production workers, and call center operators, U.S. doctors (including radiologists) determine who qualifies as a doctor. Many radiology images are out-


xxvii

sourced, some to offshore locations. But because of radiologists’ power, virtually all of these images are read by “nighthawk” firms that employ U.S.-boardcertified radiologists who are also accredited in the state and hospital in which the image was created. The typical nighthawk customer is a small U.S. hospital whose emergency room generates only a few images in a night. For these hospitals, hiring full-time or on-call radiologists would be prohibitively expensive, and it is often more efficient and cost-effective to contract with a nighthawk firm offshore. The authors also point out that radiologists’ power to restrict foreign competition is reinforced by other factors, including the cognitive nature of the work. In 2004 Levy and his coauthor Richard J. Murnane argued that tasks are easiest to offshore when they can be performed using deductive or inductive rules, a condition similar to that required to program a task for a computer. But the reading of most radiology images cannot be expressed in rules—for example, few images can be scanned by a computer. Thus, a radiologist’s output is hard to monitor, placing extra emphasis on the radiologist’s credentials. This emphasis on credentials interacts with the threat of malpractice litigation: few doctors would want to explain to a jury why an image was interpreted by an unlicensed radiologist. Correspondingly, a nighthawk firm cannot purchase malpractice insurance unless it can prove it uses board-certified radiologists. Levy and Goelman point out that 75 percent of the fee reimbursed by most health plans is a “technical fee” paid to the entity that owns the scanning equipment and that only 25 percent goes to the radiologist who reads the image. An insurer seeking to limit aggregate costs might in the future focus on limiting the number of scans through benefits management rather than trying to certify foreign radiologists. By contrast, the doctors themselves—radiologists and nonradiologists who own their own scanning equipment—may be a future source of foreign demand. The authors close by explaining why mammography might be a candidate for this kind of offshoring.

Part V: What Role for Policy? The final two papers focus on the implications of offshoring for U.S. policy. Lael Brainard, Robert E. Litan, and Nicholas Warren argue that there is a strong case for helping to insure the livelihoods of the widening pool of American workers who face insecurity associated with structural shifts in order to preserve the benefits of an open and innovative economy. They propose a new wage loss insurance program to provide incentives for more rapid reemployment and on-

xxviii


the-job-training—a program that insures against earnings losses for permanently displaced workers who secure reemployment. Brainard and her coauthors estimate that it would cost roughly $3.5 billion a year to insure permanently displaced full-time workers (who secure reemployment) for 50 percent of their earnings loss up to a cap of $10,000 a year for two years. The authors argue that although the U.S. labor market ranks second to none in job turnover, America’s safety net for easing job transitions is one of the weakest among the wealthy economies. The main federally mandated unemployment insurance (UI) program contains so many restrictions that today only about 40 percent of all jobless workers receive benefits. Meanwhile, workers have long found it difficult, time-consuming, and expensive to prove that they are entitled to extended unemployment benefits under the nation’s Trade Adjustment Assistance (TAA) program. Despite important reforms in 2002, TAA has helped fewer than 75,000 new workers per year, while denying more than 40 percent of all employees’ petitions. And remarkably, the Department of Labor has interpreted the TAA statute as excluding service workers displaced by trade. Arguing that workers’ firm-specific skills are losing value at an accelerating pace in the face of offshoring and technological change, Brainard and her coauthors advocate supplementing the existing safety net with a new program that insures against wage loss, not just unemployment, for permanently displaced workers. Wage insurance would encourage workers to broaden their employment search and go back to work more quickly, while defraying the cost to employers of hiring and providing on-the-job training to new employees from different sectors. With wage insurance, the economy as a whole would benefit from shorter spells of joblessness and more efficient reskilling for workers. A chief goal of wage insurance is to speed the reemployment of workers who have been permanently displaced. Wage insurance is most likely to have overall positive economic benefits if it targets workers whose earnings would otherwise fall dramatically as forces outside their control devalue their firm-specific skills. A Canadian pilot wage insurance program reduced unemployment durations by 4.4 percent, on average, according to research by the Social Development and Research Corporation. This could amount to hundreds of millions of dollars in annual savings on unemployment insurance payments in the U.S. context. The authors emphasize that wage insurance also serves as a training subsidy for the worker’s new employer. The retraining and new skills that a displaced worker receives on a new job benefit both the worker and the new employer. Finally, evidence suggests that wage insurance encourages workers to broaden their job search to new types of jobs in new sectors.


xxix

The authors estimate that the net cost of a fairly generous program that provides 50 percent replacement of lost earnings with a $10,000 annual payment cap for two years would be $3.5 billion a year, on average, which amounts to an insurance premium of roughly $25 per worker per year. This is substantially less than the $42.4 billion paid by the state and federal governments in unemployment insurance benefits in 2003. The authors project the costs of a wage insurance program for several different scenarios. Brainard and her coauthors argue that a comprehensive, incentive-based safety net for displaced workers that encourages rapid reemployment and onthe-job training is a benefit for workers and businesses alike. In the volume’s final paper, Kimberly Clausing examines the role of U.S. corporate tax policy on offshoring behavior. Under the current system, U.S. multinational firms are taxed on their worldwide income, although tax credits are granted for taxes paid to foreign governments. Since profits are only taxed upon repatriation to the United States, this system provides an incentive to locate real economic activity as well as profits in low-tax countries. In addition, there is an incentive to avoid locating (and earning profits) in high-tax countries, because U.S. tax credits are limited to the U.S. tax liability. Recent changes in tax law under the American Jobs Creation Act of 2004 strengthen these incentives by further lightening the taxation of foreign income and by granting a temporary tax holiday, lowering taxes on repatriations of dividends from low-tax countries. Clausing examines the incentives that are created by this system with respect to offshoring. Ceteris paribus, the U.S. tax system provides an incentive to offshore activities in low-tax countries and to offshore in-house rather than at arm’s length. Substantial empirical evidence documents that U.S. multinational firms are sensitive to tax rate differentials among countries in their decisions regarding where to invest; this responsiveness is increasing, in part because of the increasing globalization of U.S. business. In addition, the previous empirical evidence indicates that multinational firms are sensitive to tax differences among countries when they decide where to book profits. Such sensitivity has implications for government revenues in the United States and elsewhere. Clausing sees four potential goals for an international corporate tax system: enhancing efficiency, improving U.S. macroeconomic indicators, improving the competitiveness of U.S. multinationals, and generating government revenue. Most international tax systems reflect a balance of these goals, and these goals sometimes compete. For example, efforts to enhance the competitiveness of U.S.-based multinational firms may lead to an artificial tax preference

xxx


favoring economic activity and profits in low-tax countries. Also, some goals are easier to achieve with tax policy than others; attempts to use tax policy to enhance U.S. macroeconomic indicators or the position of U.S. financial balances may be misguided. Finally, Clausing discusses the merits of several policy alternatives in the context of the current U.S. system and the policy goals. These alternatives include the adoption of a territorial system of taxing international income, under which foreign dividends would be exempt from taxation; the elimination of deferral of U.S. taxation on income earned in low-tax countries; and greater international coordination. All of these major changes would require important tradeoffs and should be made with caution. In the meantime, pragmatic smaller changes are likely to improve the functioning of the current U.S. tax system. Proposals to lower the corporate tax rate, broaden the tax base, strengthen enforcement, and simplify the tax system deserve close attention, because these changes would likely improve the performance of the U.S. tax system. Such changes could also help ensure that offshoring activities occur in a manner that enhances efficiency and is consistent with the national interest.

JAMES R. MARKUSEN University of Colorado–Boulder

Modeling the Offshoring of White-Collar Services: From Comparative Advantage to the New Theories of Trade and Foreign Direct Investment

I

have always viewed trade theory as consisting of a portfolio of models.1 There are many underlying causes of or motives for trade, and it is probably more productive to have a series of models analyzing just a few of these at a time than to attempt one grand model that includes all possible bases for trade. At the other extreme, we could envision a model for every industry and every country pair and perhaps every multinational firm. But at this point, theory coincides with case-study analysis and we learn nothing of any generality. So a parsimonious set of models, the number of elements greater than one but less than say one thousand, is probably a good scientific objective. My first question in approaching my assignment for the Brookings Trade Forum is whether we can make good progress from off-the-shelf elements of our portfolio of models, or do we need an entirely new approach? The methodology I use to answer this question is to first ask another question: what are the important characteristics of the offshoring of white-collar services that we wish to capture in a theory model? Having identified a number of these characteristics, I am led toward the conclusion that we can indeed go a long way by drawing from our portfolio of models, mixing and matching elements to create a useful, empirically relevant, and productive subtheory for offshoring white-collar services. 1. I got this idea from Tjalling Koopmans’s (1957) Three Essays on the State of Economic Science, which is still of great value today. Koopmans used the term “sequence of models.”

1

2


I sketch the outlines of a number of candidate “template” models, each of which captures some aspect of the problem. From analytical insights and numerical simulations of these models, I am then able to answer questions about the effects and consequences of technological or institutional innovations that permit offshoring to arise. These include effects on the national income of each country, effects on the relative and real prices of skilled and unskilled labor in each country, and effects on the volume of trade in goods (for example, are trade in goods and trade in services complements or substitutes?). Before proceeding, I wish to emphasize that my goal here is to suggest ways of thinking about the issues in formal models. I was distressed following my presentation at the Brookings conference to find many people focusing on the results of some simulations, particularly with respect to “northern” welfare (read as U.S. welfare). All readers should understand that no theory says that a move from partial liberalization to full liberalization makes everyone better off. To push the point further, I am confident that I can concoct a model to generate any result desired by a reader with a deep pocketbook. I have tried hard to stick to reasonable and relevant structures and assumptions, but even so, qualitative results sometimes depend on specific parameter values, as we shall see. In the following section, I provide a brief overview of some of our theory portfolio and then identify some of the crucial aspects of offshoring we wish to capture. Finally I present a series of template models.

Our Theory Portfolio We can usefully draw from existing theories and models of trade in order to make progress on offshoring. I do not claim that the list is exhaustive or that alternative taxonomies might not be more useful; I just believe that these particular elements will prove useful. —Comparative advantage theories of trade in goods. Our traditional trade theory tends to focus on differences among countries as the primary motive for trade. The Ricardian model of trade, in which countries possess different technologies, is usually listed first. Second, the workhorse model of trade is factorproportions or Heckscher-Ohlin theory, in which differences in factor intensities among goods intersect with differences in factor endowments among countries to determine a pattern of comparative advantage and trade. This ever-popular approach not only gives an intuitive explanation for the direction of trade, but permits a detailed analysis of the distributional consequences of trade within countries and of aggregate gains from trade. Other country characteristics that fit

James R. Markusen

3

here include differences in market distortions among countries and country (internal market) size. —Non-comparative-advantage theories of trade. This category is largely the domain of the “new trade theory,” a term I dislike: “industrial organization approach” to trade is more apt (and avoids the awkward problem of what to call the theory that comes after the new theory). The principal motives for trade are scale economies, imperfect competition, and product differentiation. A subcategory of this branch of theory involves the existence of firm-specific assets, an especially useful approach to the theory of the multinational firm. These range from managerial and technology assets to brand names and trademarks. This approach has resurfaced more recently in heterogeneous firm models, in which (potential) firms get productivity “draws” from some distribution that make some firms more productive than others (Melitz 2003; Helpman, Melitz, and Yeaple 2004). Productivity in turn determines whether firms enter foreign markets and if so whether by exports or foreign production. —Trade in factors. While trade in goods has drawn the most attention in both theory and empirical analysis, the topic of trade in factors has always lurked in the background. Generalization of theoretical findings is difficult, but the loose consensus among trade economists is that trade in goods and trade in factors tend to be substitutes in comparative-advantage models. Indeed, Mundell’s (1957) early demonstration of this might explain the lack of interest in trade in factors. An elegant treatment of this substitutability is found in Jones, Coelho, and Easton (1986). Later it was shown that trade in goods and trade in factors tend to be complements for virtually any other causes of trade other than factor proportions (Markusen 1983) and even in some versions of factor-proportions models (Neary 1995). —Theories of foreign direct investment and arm’s-length trade in firm-specific assets. I think it is fair to say that, until the mid-1980s, FDI was just viewed as part of the theory of capital movements in a factor-proportions world. Eventually a huge amount of empirical evidence, most notably that most foreign direct investment (FDI) not only comes from but goes to other high-income capital-rich countries, led to new approaches to what we are now calling offshoring. Theory split into two branches. One could be called the vertical or resourceseeking approach, an early example of which is described by Helpman (1984). This is in fact a natural extension of factor-proportions comparative-advantage models in which activities differ in factor intensities and countries differ in factor proportions. The alternative is the horizontal or market-seeking approach, in which firms exploit firm-specific assets in multiple markets, an early example of which is described by Markusen (1984). The latter is more a part of non-comparative-

4


advantage theory and, while both approaches are important, does the job of explaining the large volume of intra-industry FDI among the high-income countries. I believe it is accurate to say that the overwhelming weight of empirical evidence, beginning with Brainard (1997), is more consistent with the horizontal approach. Intertwined with this literature on FDI is a long-standing literature on “internalization,” now being called by its inverse name, “outsourcing.” Both terms address whether firms keep certain activities internal to the firm or use arm’slength contractors to supply intermediates or to provide assembly, services, distribution, and so forth. Early analyses include that by Dunning (1977). Some more recent authors seem unaware of this large literature, but it is still a pertinent antecedent, and changing the name from internalization to outsourcing does not change that fact. This literature argues that the choice between internal and arm’s-length modes depends on issues such as moral hazard, adverse selection, hold-up, contract enforcement, and intellectual property protection. —Trade in business services (non-factor, non-trade-mediating services). There was an earlier wave of interest in trade in business services in the late 1980s, in Canada in particular. In my view (and I was a participant) the theory that came out of this was not very successful. Several authors got bogged down in trying to define services, an elusive goal, as Daniel Trefler (this volume) has so nicely indicated with a quote from Justice Potter Stewart. One traditional view of business services is that they are hard to trade, requiring the spatial and temporal proximity of supplier and customer. Herbert Grubel (1987) went so far as to argue that all trade in services is embodied in goods or persons. It is very clear from the topics we are considering today that this view is at best badly outdated. One area where progress has been made is in the theory of the multinational. The modern view is that parent firms are exporters of the services of knowledgebased assets to foreign subsidiaries (although goods and intermediates are often traded as well) (Markusen 1995, 2002). —Liberalization: trade expansion at the extensive margin. Much traditional trade theory involves liberalization expanding the volume of trade in existing traded goods. We could call this expansion at the intensive margin. But these models do not seem appropriate to the current discussion, in which we are looking at new things being traded. Some existing theory bears on this. In comparative-advantage models, liberalization expands trade at the intensive margin, but some “middle” goods can become traded as trade costs fall, as in the DornbuschFischer-Samuelson (1977) Ricardian continuum model. Yi (2003) has a neat Ricardian model in which goods are produced in distinct stages of production

James R. Markusen

5

that can be geographically fragmented. Other valuable empirical evidence is found in Hummels, Ishii, and Yi (2001). In the theory of the horizontal multinational, investment liberalization allows intrafirm trade in the services of knowledge-based assets, so more things are traded. For vertical multinationals and arm’s-length offshoring, innovations in technology, liberalization, or institutions (intellectual property protection) allow fragmentation of the production chain so that more things are traded: capitalintensive intermediates go out, labor-intensive assembly takes place abroad in labor-abundant countries, with much of the final output shipped back to the parent country. In all of these models, trade expands at the extensive margin.

Empirical Characteristics of Offshoring of White-Collar Services to Capture in Theory Models Here is a wish list of characteristics for theoretical models of offshoring of white-collar services. —Expansion of trade at the extensive margin: new things traded due to innovations in communications and technology. This poses a number of challenges to theory, especially the fact that we are talking about nonmarginal changes and discrete movements of something being nontraded to potentially lots of trade. Traditional comparative-statics analysis is of little use: it focuses on marginal changes in activities that are already in use in the benchmark. —Vertical fragmentation of production: the new traded services tend to be intermediates, and they may be upstream, downstream, or not part of a sequence. Traded white-collar services often have important characteristics that cannot be captured in the simplest off-the-shelf models, which assume a set of final goods. One is that they may be firm-specific rather than bought and sold in arm’s-length markets. Another is that they may form part of a particular production sequence, such as a well-defined upstream (design) or downstream (after-sales service) component of overall production. A third is that there may be crucial complementarities among different elements of the production chain, such as between skilled labor and telecommunications equipment and infrastructure. —Offshoring of medium-skilled or even highly skilled services to skilledlabor-scarce countries. Is this at odds with factor-proportions theories? The simplest off-the-shelf 2 x 2 Heckscher-Ohlin model is not going to offer insights into why relatively skilled-labor-intensive services are being offshored to very skilled-labor-scarce countries. One of the most important tasks of theory, in my

6


opinion, is to develop richer but empirically plausible models of why this is taking place. Yet the factor-proportions approach to trade does not necessarily have to be abandoned; but it must be enriched to include multiple goods or factors, or both, so that fragmentation and the complementarities just discussed can be analyzed. —Reversal in the direction of trade from existing multinationals models. New offshoring is exporting services back to high-income-country firms (intrafirm back to parents or via arm’s-length contracting). Trade in white-collar services is not new. The modern theory of the multinational has emphasized that parents are exporters of white-collar services, including management and engineering consulting, marketing, finance, and others, to their subsidiaries. One thing that is relatively new and that has generated much of the current interest is the reversal in the direction of trade that we are seeing. In some ways this is closely related to the previous point. —Firms, or specifically owners of knowledge-based assets, may offshore skilled-labor-intensive activities that are complements to these assets. A plausible worry is that skilled workers in the high-income countries are being hurt while their companies profit from offshoring. This cannot be dismissed and requires investigation. To me, it calls for at least a three-factor model, in which firms possess specific factors or other assets that are complements to skilled labor. One example is software engineers as complements to telecommunications equipment and network infrastructure, in which the third factor is physical capital. Or it could be that software engineers are complementary to managerial sophistication, organization infrastructure, and marketing channels. The complementary input is knowledge-based assets. Without services trade, you can train an engineer in India, but there will be no demand for his or her skills if there is nothing useful to do. The implication is that, in the absence of offshoring, these skilled workers are cheap even though they are relatively scarce in comparison with the availability of skilled workers in the country with the complementary factors. Offshoring that allows trade in the third factor causes that factor (or its services) to move to the skilled-laborscarce country to combine with cheap skilled labor there. This setup obviously has the elements of a story in which skilled labor is harmed in the high-income country, while owners of the complementary physical or knowledge-based assets benefit. This phenomenon is relatively easily modeled either in a competitive multifactor model or using Markusen’s knowledge-capital approach to the multinational. The latter approach has also proved a useful starting point for looking at

James R. Markusen

7

the internalization versus outsourcing decision in relation to the offshoring decision. The idea is that transacting in knowledge-based assets creates special problems for the owner (the multinational firm). For example, there are several laborturnover models in which workers in the host country absorb or learn the substance of the knowledge-based assets and can defect to start rival firms. Other issues that have been considered in the theory literature involve asymmetric information, reputations, and hold-up. The next few sections of the paper construct and analyze some simple template models that incorporate these features. All of the models presented have been coded into numerical simulation models using GAMS. Code for these models is available from the author. An appendix to the paper lays out the structure of model 1.

Models Model 1: A 3 x 2 x 2 Heckscher-Ohlin Model with Fragmentation Suppose we begin with a simple two-final-good, two-factor, two-country Heckscher-Ohlin model and then allow one good to fragment into two separate production activities, giving three in total. If we assume free trade, just considering free versus prohibitive fragmentation costs, we do not need to specify which is the upstream and which is the downstream activity. For a much more comprehensive treatment of this case, see Markusen and Venables (2005). For a more general approach, see Deardorff (2001, 2005a, 2005b). Here are the principal features of the model. (A) Two factors of production: skilled (H) and unskilled (U) labor (B) Two final goods, three production activities Y: unskilled-labor-intensive X: skilled-labor-intensive X: can fragment into high-tech manufacturing (M) and services (S) M: more skilled-labor-intensive than X S: middle skill intensity: less than X, more than Y (C) Two competitive, constant returns economies North: high-skilled-abundant South: low-skilled-abundant The service component of good X is thus chosen to have a middle factor intensity between integrated X and good Y; specifically, the complete ranking from most to least skill intensity is: M > X > S > Y. This choice definitely matters for

8


the results. We are thinking here of things like business process outsourcing or call centers, which are less skill intensive than the overall industry, but more than a developing economy’s traditional sector of comparative advantage.2 I will report and analyze the qualitative results of numerical simulations. Begin with fragmentation banned; that is, M and S used for X must be produced in the same location. For this case, I calibrated the model so that the two countries are initially specialized in X (North) and Y (South) in free trade: factor endowments have a bigger spread than factor intensities. Now allow for the geographic fragmentation of X production. This results in some or all of the middle-skill-intensive service activity switching from North to South, with services exported back to North or M exported to South to be combined with S. This does not really matter with free trade, except that measured changes in trade volume will depend on which is which. For our purposes, it is perhaps better to think of the services as exported back to North, where they are combined with M to produce the completed good.3 There is a fundamental tension that arises in general equilibrium when the ability to fragment manufacturing and services is introduced. 1. Services, which are middle-skill intensive, shift from North to South, increasing the relative demand for skilled labor in both countries. This is an idea familiar from Feenstra and Hanson (1996a, 1996b, 1997) and also arises in multinationals models (Zhang and Markusen 1999). North sheds an activity that is unskilled-labor-intensive from its point of view, but South gains an activity that is skilled-labor-intensive from its point of view. Thus we expect the real and relative price of skilled labor to rise in both countries. 2. However, general equilibrium is bedeviled by terms-of-trade (TOT) effects: North moves from integrated X production to exporting M and importing S. A fall in the relative price of M harms North, possibly outweighing efficiency gains for North. The ability to fragment X production has an effect loosely related to a technical improvement. South can produce S more cheaply than integrated

2. In his comments on this paper, Douglas Irwin quite properly wonders about the robustness of results based on one particular ranking of factor intensities, yet also wants to avoid sliding down the “slippery slope” into taxonomy. I agree with both thoughts. My decision is to concentrate on a central case that I find the most empirically plausible: the offshored service has a middle intensity between Y and integrated X production. 3. Furthermore, allowing the service to be provided by southern workers is close to the same thing as moving foreign workers to North. If allowing the service to move results in factor-price equalization, they are exactly the same (provided the welfare of each country is that of its original residents).

James R. Markusen

9

North, and North can produce M more cheaply. But as the countries begin to specialize, their relative size will do a lot to determine the relative price of M versus S. The equilibrium relative price of M to S is higher the smaller North is. When North is large, an adverse terms-of-trade change can make it worse off than before fragmentation despite the efficiency gain. When North is not large, efficiency gains outweigh the terms-of-trade shift and both countries benefit.4 Closely related to this are two results that emerge from the simulations. First, my results indicated that while skilled labor is the relative gainer in North, both factors could suffer an absolute loss of real income when North is large. This occurs with a low equilibrium relative price for M as just mentioned. Second, results indicate that skilled labor is an absolute gainer in South, but that it might be a relative loser when South is large. South shifts its output to a more skilledlabor-intensive sector, but that sector (services) suffers a price fall relative to the no-fragmentation case. The latter effect is large when South is large: skilled labor gains absolutely but loses relatively. Results from my simulations over a range of parameter values can be summarized as: MODEL 1 RESULTS 1. South gains, North loses if North is large; both countries gain if South is large.5 2. Skilled labor is the relative gainer in North (but a real-income loser if South is small: TOT effect dominates). 3. Skilled labor is a real-income gainer in South (but gains relatively less if South is large: TOT effect dominates). 4. Unskilled labor is a real-income gainer in South, a loser in North. 5. Volume of trade in goods increases: goods and services trade are complements (but can fall if South is small: South is self-sufficient in S, does not export). 4. An alternative intuition about the terms-of-trade effect is as follows. North has a factor endowment that is well suited to integrated X production. When fragmentation is allowed, the equilibrium price of X falls, harming North, which specializes in X. The question is whether shifting to specialization in M more than recoups this loss. The answer is yes if North is not large. Perhaps this intuition also shows why South always gains: this terms-of-trade effect against X must benefit South. 5. A finding that North can lose is not new. Gomory and Baumol (2004) note this in a model with increasing returns. Samuelson (2004) shows a case with constant returns and perfect competition. Many other such cases occur in models with multinationals (Markusen 2002). Note that some results “guaranteeing” gains from trade compare autarky to free trade with fragmentation (Deardorff 2005a). To me this is irrelevant: the relevant question is comparing free trade in goods to free trade in goods and services.

10


Despite having done countless runs of this model, I cannot guarantee that there are no other possibilities, and of course, reordering the factor intensities will change the results. What I can say is that it is easy to find ranges of parameters that generate these results, but we should all regard them as suggestive and not definitive. Model 2: A 3 x 3 x 2 Missing Input Model with Fragmentation My second model is designed to capture the idea that skilled labor can be cheap where it is scarce. It again has three production activities and two countries, but three factors. (A) Three factors of production: skilled (H) and unskilled (U) labor and know-how (K). K could be high-tech physical capital, such as telecommunications equipment and networks, or knowledge capital (managerial techniques, organization infrastructure, marketing channels). (B) Two final goods, three production activities Y: unskilled-labor-intensive X: skilled-labor- and know-how-intensive X: can fragment into high-tech manufacturing (M) and services (S) M: more skilled-labor-intensive than X S: skilled-labor- and know-how-intensive (C) Skilled labor and know-how are complements in the production of S (D) Two competitive, constant returns economies: North: high-skill- and know-how-abundant: South: low-skill-abundant, very know-how-scarce The complementarity between skilled labor and know-how in producing S is crucial. Specifically, this is modeled as a very low elasticity of substitution between H and K in producing S. When a country is very know-how-scarce, there is little for its skilled workers to do. You can train engineers, but there are no jobs for them. Assume initially that K (or its services) and skilled workers cannot move between countries. K is used with skilled workers largely in the North, which exports integrated X. The fundamental tension caused by introducing trade in K and S is now going to occur between northern and southern skilled workers. 1. Skilled labor is initially cheap in South (even though scarce) owing to a lack of K to work with. 2. Skilled labor in North and in South compete directly, introduction of trade in K moves K to South, shifting relative demand for skilled labor to South.

James R. Markusen

11

The introduction of the third factor, complementary to skilled labor, makes it straightforward to create a situation where skilled labor is initially cheap where it is scarce. In addition, the model is less sensitive to country-size issues, at least with respect to factor prices. But as in the previous model, there is a terms-oftrade issue for North. The ability of the owners of K to move their factor to South to work with cheap skilled labor there erodes not only the return to skilled labor in North, but also North’s implicit monopoly power over good X. The result in all of my simulations was that welfare decreases in North when trade in K and S is permitted and North is large. As in the previous model, both countries gain when North is not large. Here are my results for permitting trade in K and S. MODEL 2 RESULTS 1. South gains, North loses when North is large; both gain when North is not large. 2. Skilled labor is real-income loser in North, absolute gainer in South. 3. Real return to know-how rises in North, falls in South. 3. Unskilled labor is real-income gainer in North, loser in South. 4. Volume of trade in goods increases (complements services). Losses to North and northern skilled labor in particular are two of the things that analysts have worried about with respect to the offshoring of white-collar services. This model potentially validates the worry that northern business owners or owners of particular physical capital and knowledge-based assets will benefit considerably at the expense of northern skilled workers. Of course, the model is in part deliberately designed to do that, so this is hardly a coincidence. On the one hand, I cannot say with confidence that a thorough search would not lead to alternative models with quite different results. On the other hand, I would not have put this model forward if I did not find it empirically plausible and relevant. Model 3: A 3 x 2 x 2 Knowledge-Capital Model of Multinationals Now I would like to return to something close to model 1, but add in multinational firms following the knowledge-capital model of the multinational that I developed some time ago. This version of the model is based on Zhang and Markusen (1999). Markusen (2002) is the best source for the complete development of the theory, and this section is based on Chapter 9. (A) Two factors of production: skilled (H) and unskilled (U) labor (B) Two final goods, three production activities Y: unskilled-labor-intensive, constant returns, perfect competition

12


X: skilled-labor-intensive, increasing returns at the firm level: firm and establishment-level fixed costs, constant marginal cost X: can fragment into high-tech manufacturing (M) and services (S) M: more skilled-labor-intensive than X. By assumption, only North can produce M. S: middle skill intensity: less than X, more than Y (C) Cournot output competition by X firms, free entry and exit in two firm types National firms: produce M and S in North, export X to South Multinational firms: produce M in North, which is exported to South, where S is produced, or vice versa (D) Two economies North: high-skill-abundant South: low-skill-abundant The reader will see that this resembles model 1 insofar as X can fragment into a skilled-labor-intensive phase and a medium-skilled-labor intensive phase. I have modeled the S phase as largely unskilled-labor-intensive in marginal costs, but establishment fixed costs as having a sizable skilled-labor component. I do not think that this affects the results. A nice feature of this model, aside from its probable empirical relevance, is that it avoids the “curse of Stolper-Samuelson”6 and the terms-of-trade effects that are so important in the competitive, constant-returns models. Because of procompetitive effects leading to increased firm scale and lower markups, it is entirely possible that both countries and all factors gain following a liberalization.7 The way this works in the present model is straightforward. Again, begin in a situation where trade in disembodied S is not allowed: S and M must be produced together. This is equivalent here to not allowing multinationals to enter. Having to use North’s factor endowment for both M and S is a binding constraint on the world economy and limits the number of firms in free-entry equilibrium, which in turn leads to a high markup and a low output per firm (high average cost). When this constraint is relaxed by allowing firms to fragment X, much, perhaps all, of service production moves to South. This again tends to have the Feenstra-Hanson effect of raising the relative demand for skilled labor 6. I didn’t invent this phrase, though I wish I had. I think I heard it first from Ron Jones. 7. Alternatively, we could model the final goods or intermediate services as differentiated, using the now well-known large-group monopolistic-competition framework. As far as welfare is concerned, there is a benefit from increased variety analogous to the procompetitive effect of the oligopoly model that tends to generate large welfare gains. See Ethier (1982) and extensions by Markusen (1989).

James R. Markusen

13

in both countries. But now this also increases the profitability of the existing firms, which leads to entry, which in turn leads to lower markups and higher output per firm (lower average costs). There are, however, a lot of possibilities depending on relative endowments and intensities and again on country size. Chapters 8 and 9 of Markusen (2002) show that many outcomes are possible. I can say that it is easy to find parameter values for which allowing fragmentation leads to welfare increases for both countries and gains for skilled labor in both countries. I have to admit that I did not find a set of parameters for which the real prices of all four factors increase, however. I tended to find that the real return to unskilled labor in North fell following fragmentation and trade in services. Here are some typical, but not robust, results for the effects of allowing multinationals to enter, equivalent here to allowing trade in services. MODEL 3 RESULTS (for a range of parameterizations) 1. South gains, North gains. 2. Skilled labor is a relative and an absolute gainer in North and South. 3. Unskilled labor loses in North and gains in South. 4. Procompetitive effects lead to more firms, lower markups, higher output per firm. Model 4: A 3 x 3 x 2 Model That Combines the Knowledge-Capital Model with the Missing Factor Model Our fourth template combines the knowledge-capital model with the missing factor model. I take the skilled labor in North and assume that some portion of it is factor K, which is complementary to skilled labor in producing establishment fixed costs. In fact, I coded up this model first and then moved to model 3 by simply allowing the substitution between K and H in producing establishment fixed costs to move to infinity. Otherwise, the models are the same. In the initial equilibrium, trade in K (or the services of K) is not permitted; alternatively, multinational firms are not permitted to enter. These results are then compared with allowing trade in K, or equivalently allowing multinationals to enter. Again, a range of outcomes is possible. But for the same parameterization that model 3 just indicated, the liberalization here generates a stronger adverse terms-of-trade effect for North. North’s welfare declines if North is large. As with model 2, now the skilled labor in North competes directly with skilled labor in South. The introduction of multinationals moves K from North to work with initially cheaper skilled labor in South. This lowers the real return to skilled labor in North with the big beneficiary being owners of the factor K.

14


MODEL 4 RESULTS 1. South gains, North loses if North is big; both gain if South is big. 2. Skilled labor is relative and a real-income loser in North, a gainer in South. 3. Large gain in North for the owners of know-how. 4. Unskilled labor gains in both countries. 5. Procompetitive effects lead to more firms, lower markups, higher output per firm.

The Offshoring-Outsourcing (Location–Mode Choice) Relationship Internalization or its inverse, outsourcing, is a decision about the boundaries of the firm and what activities to keep inside or internal to the firm’s ownership structure and which to contract to arm’s-length firms. Multinationals offshore but do not outsource, keeping their foreign activities within owned foreign affiliates. Firms that contract or license in some way to foreign firms are engaging in both offshoring and outsourcing. As I indicated earlier, the internalization decision, also known as mode choice, has a long history of analysis, particularly in the international business literature. Its rediscovery under the name outsourcing has coincided with many researchers overlooking this long tradition. In any case, the traditional focus of the internalization/outsourcing decision has been on the various transactions costs, particularly when offshoring, of doing business at arm’s length rather than internal to the firm. It is important to keep offshoring and outsourcing decisions distinct: they are location choice and mode choice decisions, respectively. A factor that encourages outsourcing might at the same time discourage offshoring in favor of exporting from the home country or choosing a third country. There are in fact a number of factors associated with producing abroad that do precisely this: they encourage outsourcing but discourage offshoring. Some of these are: —restrictions on foreign investment —restrictions on the right of establishment —restrictions on immigration (generally temporary) of foreign business personnel —lack of intellectual property protection —lack of contract enforcement The first three in this list generally follow from the fact that offshoring requires setting up a foreign subsidiary, which in turn requires foreign investment and the use of home-country personnel in the host country for some period of time.

James R. Markusen

15

Thus, problems in any of these three areas would encourage a firm to outsource to a local firm, but they also discourage offshoring relative to other outside alternatives. The last two points involve various aspects of moral hazard and hold-up when firms make investments, both sunk physical capital and investments in training local workers, in connection with establishing a subsidiary. Again, they tend to encourage outsourcing but discourage offshoring. The problem of transferring knowledge and skills probably exists for both modes of offshoring, about which I will say more shortly. Two formal approaches found in the theory literature may be useful. The first I will call the “labor-learning model.” Variations of this are set forth by Ethier and Markusen (1996), Fosfuri, Motta, and Rønde (2001), Markusen (2001), and Glass and Saggi (2002). All of these papers discuss multinationals that make a foreign investment that is profitable owing to knowledge-based assets of the firm. However, workers in the host country acquire this knowledge themselves and can later defect to start a rival firm. If binding contracts cannot be written, then the firm will have to pay these workers a premium in subsequent periods to hold them in the firm. Thus, the multinational must share rents with local employees if contracts cannot be written or enforced. A second promising line of research involves the Grossman and Hart (1986) hold-up model, which has been developed in a series of papers by Antràs (Antràs 2003, 2005; Antràs and Helpman 2004). Here the idea is that the multinational firm and a local agent must each make sunk, relationship-specific investments in a project. In the absence of complete contracts or contract enforcement, this creates a bilateral ex post hold-up problem. The optimal mode of entry is generally that ownership, defined as residual rights in assets if bargaining breaks down, should go to the party with the larger sunk investment. As in the labor-learning model, this approach requires the multinational to share rents with a local agent whether or not that agent is the manager of an owned subsidiary or the owner of an arm’s-length contractor. Alternative assumptions can produce alternative correlations between offshoring and outsourcing. Suppose that a firm wishes to supply a product X in South. If the fixed costs of setting up a foreign plant are not too large but large relative to the sunk investments of the local partner, then the firm will tend to choose both offshoring (in preference to exporting) and internalization: a negative correlation between offshoring and outsourcing. A difference between the labor-learning model and the hold-up model is that, in the latter, both the multinational and the local manager make ex ante sunk investments that generate bilateral hold-up. In the labor-learning model, workers acquire bargaining power ex post as they learn. I think that both approaches

16


have something to contribute to the offshoring of white-collar services. There is no question that there are a lot of training costs for the foreign workers. Information technology (IT) and business process outsourcing (BPO) activities are often learned on the job by workers who already possess good general skills. It is my understanding that many call-center workers are trained by independent firms before landing a call-center job. Here is my suggestion for one approach that combines the labor-learning approach and the sunk-cost hold-up approach. Think of this as, perhaps, a model of business process outsourcing or call centers. Model 5: Template for an Integrated Outsourcing-Offshoring Approach (A) Begin with the “missing input model,” two time periods. (B) Interpret this as firm-specific knowledge capital à la Markusen’s knowledge-capital model: skilled workers in the host country are cheap because they lack crucial physical or knowledge-based inputs. (C) With appropriate technological and institutional conditions, this asset can be “exported” by a firm (used abroad) in combination with local skilled workers. (D) However, local workers “absorb” the relevant knowledge and are able to “defect” to start rival firms on their own in the second period. (E) Also assume a capital investment in land, structures, and telecommunications is needed. Whoever owns this defines whether the project is a subsidiary (internalized) or an arm’s-length relationship (outsourced). The cost must be borne by the multinational. I suppose that many researchers in the international business field would conjecture that given complete and enforceable contracting, the firm would prefer outsourcing on a simple cost basis, so let us make that assumption. MODEL 5 RESULTS (conjecture! this paper has not been written!) 1. Given complete and enforceable contracts, outsourcing is preferred (by assumption). 2. If contracts are not enforced, then the multinational will want to own the physical capital—that is, internalization is chosen by the Grossman-Hart-Antràs argument. 3. However, even if it is possible to contract for physical capital (local firm contracts to pay a mortgage), the firm may still want to own it if it is not possible to contract for the intellectual capital (skills) that is transferred to local workers, in order to reduce the ex post hold-up problem of skilled workers threatening to leave.

James R. Markusen

17

4. On the other hand, if the learned skills of the foreign manager are relationship-specific—that is, they are only useful to the contracting multinational—then there is limited hold-up from the workers and outsourcing would be preferred. Indeed, in this case it seems as though the firm would have ex post hold-up power, and so the manager would want to own the capital. Again, this is conjecture. I am working on this project, but not yet certain of the results. As a final point, recall again that the agency costs and rent-sharing costs to the firm, whether they be less in the internal or the outsourcing mode, also affect the firm’s offshoring choice. For a firm seeking to serve the local host market, these costs may lead the firm to choose exports rather than offshoring. For a firm seeking to serve its own home market, these costs may lead it to choose domestic outsourcing or search for a third supplier.

Summary and Conclusions I have argued in this paper that we can make good progress in understanding the offshoring of white-collar services at the theory level from our existing portfolio of models. Many important features of offshoring white-collar services can be modeled from a recipe that mixes and matches elements from the existing inventory. Useful elements from our portfolio include: —vertical fragmentation of production —expansion of trade at the extensive margin —fragments that differ in factor intensities, countries that differ in endowments —knowledge stocks of countries or firms that are complementary to skilled labor; these create missing inputs for countries otherwise well suited to skillintensive fragments —knowledge-based assets that create particular contractual and agency issues for firms engaging in international business. Existing models of laborlearning and hold-up are useful places to start in considering the outsourcing (mode) choice in relation to the offshoring (location) choice. These features allow us to construct relatively simple and tractable generalequilibrium models that predict changes induced by fragmentation on aggregate welfare, factor prices (income distribution), the location of production activities, and the direction and volume of trade. While I view this paper as listing a number of plausible and empirically relevant ways of modeling the offshoring of white-collar services, it was clear at the conference that many people were much more interested in specific results from these models. Unfortunately, it is hard to offer robust conclusions, especially

18


about the aggregate welfare of countries. Trade theorists are well aware of the underlying problem: any move from partial liberalization to more liberalization (allowing more things to be traded in our case) often does not result in Pareto improvements for the trading partners. It would be intellectually dishonest for me to report only cases in which everyone is better off. Overall, my simulation models suggest a clear gain for world welfare and for South in particular, but North may lose if it is large. This result is very familiar to all trade theorists in different contexts; for example, a large country may well prefer a Nash equilibrium in tariff rates to free trade with a small country. Stephen Magee and Kwang-Yeol Yoo (2005) have argued persuasively that the United States is not a large country in the sense of my models, which might give us some comfort. Results on factor-price changes are interesting and consistent with a range of existing literature. In the two-factor models, skilled labor is the relative and (usually) absolute gainer in both regions, as activities that are not skill intensive from North’s point of view are transferred to South, where they are. Results for unskilled labor are more mixed. I have been asked to indicate which model may fit reality the best, and I have to say that I think that the three-factor “missing input” model is my favorite, preferably with multinationals. I have called the third factor “knowhow”: it could be knowledge capital, high-tech physical capital, or highly skilled knowledge workers, including management. I started working with this model in connection with Central and Eastern Europe, where the productivity of workers with excellent skills in math, science, and engineering was very low: they were missing the crucial organizational, managerial, quality control, and marketing skills needed to complement their other skills. Many case studies I have read about East Asia suggest that the same circumstances prevail. You can educate scientists or engineers, but if there is nothing for them to do they will not be productive. I capture this by making North rich in know-how as distinct from general skilled labor, and by making know-how a strong complement to general skilled labor. The result is that skilled labor is cheap in South where it is scarce, a principal stylized fact that has generated much of the white-collar offshoring. This model sets up a tension between the general northern skilled labor and the southern skilled labor; perhaps routine programmers and routine businessprocess workers are examples. Allowing fragmentation moves know-how, or rather the services of the know-how, to South, generating big gains from the owners of the know-how and losses for general skilled labor in North. We should keep in mind, however, that much of the know-how is surely embodied

James R. Markusen

19

in the human capital of highly skilled tech workers, managers, marketers, and so forth, and this may suggest that second-level white-collar workers are the ones who are most at risk. I will close with another caveat about theory. Strong and robust findings about welfare gains from fragmentation are not forthcoming from a general approach to theory. Alternatively, it is usually possible to find some strange model that generates whatever result a client wants. I have tried to construct models that I feel are plausible and relevant, but some residual ambiguity remains. I hope I have at least left us with some sense of why this ambiguity exists.

Appendix Example of the Simulation Models Used in the Paper The models in the paper seem simple enough. The first model begins with the classic 2 x 2 x 2 workhorse model of trade theory. All students of economics learn the analytics of this model, many of these as undergraduates. Any economist can reasonably conjecture that introducing the ability to geographically fragment some activities should still permit analytical solutions. Unfortunately, it is not nearly that simple. Let us take model 1 as an example, our simplest model. We go from two to four production activities: Y, M, S, and final X instead of just Y and X. The number of possible production specialization patterns for a country goes from three to fifteen (this assumes that you have to produce some of something). In addition, there are a great many more possible trade patterns, the number going from two to fourteen. In other words, the dimensionality of the model increases greatly, making the simple analytical methods we are used to much less useful. Second, the entire model must now be formulated in terms of inequalities, not equations. We do not know which of these will hold for a particular set of parameters (for example, which production activities and which trade activities are slack), and, indeed, the set that holds with equality will typically change a lot when parameters are changed. Allowing fragmentation can, for example, reverse the direction of trade in X or Y, or both. The models are termed nonlinear complementarity problems in math programming language: each weak inequality is associated with a non-negative complementary variable. If an inequality holds as a strict equality in the solution, the complementary variable is positive; if it holds as a strict inequality (for example, marginal cost exceeds price), the complementary variable (quantity in this case), is zero. Traditional comparative-statics techniques used on sets of equalities are of no

20


use. If you read some of the existing literature on production fragmentation, you will then understand why almost none of it actually solves for a world general equilibrium. Thus, I have chosen to approach the template models using simulations. I use software from GAMS, which has the only powerful and robust nonlinear complementarity solver in the business. All models consist of three sets of inequalities and complementary variables. First, there are zero-profit inequalities for each production and trade activity: marginal cost is greater than or equal to price. The complementary variable is the output or “activity level” of that activity. A quantity variable is complementary to the price inequality. Second, there are market-clearing inequalities: the supply of a commodity (good, factor, import, or other) is greater than or equal to its demand. The complementary variable is the price of that commodity. A price variable is complementary to a quantity equation. Finally there is an income balance equation for each country. In this appendix, I give the set of inequalities for model 1. There are thirtyfour production and trade activities, twenty-four “commodities” (final goods, intermediate goods, imports and exports, and utility, which is treated as a good produced from inputs of X and Y in the code), and two income levels. Walras’s Law makes one equation redundant: I fix the world price of Y at 1 and drop the world market-clearing equation for Y. Thus the entire model consists of fiftynine weak inequalities, each with an associated non-negative variable. In the body of the paper, I introduced only the notation needed to describe the models in basic economic terms. Here are the definitions of additional notation needed for the formal model. pki pcki pk wji ck(...) EKi IKi Wi Ii

producer price (that is, marginal cost) of good k in country i (k = Y, X, M, S; i = North, South) consumer/import price of good k in country i world price of good k price of primary factor j in country i (j = U, H) unit cost of producing good k (includes “cost” of producing utility: the unit expenditure function) exports of good k by country i imports of good k by country i welfare of country i income of country i

Other:

James R. Markusen

21

(1) market-clearing inequalities make extensive use of Shepard’s lemma, in which the unit demand for a good or factor is the derivative of the unit cost function with respect to the price of that good or factor. (2) very small trade costs, 0.05 percent, are used to prevent “ties” or, more formally, model degeneracy. This prevents, for example, its being equally profitable to both import and export a good, which would lead to indeterminacy of gross trade flows and possibly a failure to solve (infinitely many solutions differing in gross trade flows). (3) solutions without fragmentation permitted are computed by constraining import and export activities for S and M to be zero. (4) all production activities using more than one input are Cobb-Douglas, with shares as follows: Y U: 0.70 H: 0.30 M U: 0.17 H: 0.83 S U: 0.60 H: 0.40 X M: 0.70 S: 0.30 implied shares of primary factors in X produced from domestic M and S are the inverse of share for Y. X U: 0.30 H: 0.70 (5) factor endowment ratios for North and South are symmetric: North H: 0.90 U: 0.10 South H: 0.10 U: 0.90 The model is thus symmetric between North and South and between X and Y without fragmentation. Endowment ratios have a bigger spread (9/1) than intensities (7/3). If countries are the same size, then the no-fragmentation equilibrium is symmetric with both countries specialized, a goods terms of trade of one, factor-price ratios that are inverses in the two countries, and equal welfare across countries.

22

Brookings Trade Forum: 2005 Inequalities zero profit inequalities

Complementary variable activity levels

description (no. of inequalities) production of Y in i (2) production of M in i (2) production of S in i (2) production of X i from Si , Mi (2) production of X i from Sj , Mi (2) production of X i from Si , Mj (2) production of X i from Sj , Mj (2) exports of M by i (2) imports of M by i (2) exports of S by i (2) imports of S by i (2) exports of X by i (2) imports of X by i (2) exports of Y by i (2) imports of Y by i (2) home supply of Yi to i (2) production of welfare in i (2) supply - demand of Y prod (2)

supply - demand of Y cons (2)

supply - demand for M i (2)

supply - demand for imported M (2)

James R. Markusen market clearing inequalities

23 prices

description (no. of inequalities) supply - demand for Si (2)

supply - demand for imported S (2)

supply - demand for X i (2) world market and price for Y (1) world market and price for M (1) world market and price for S (1) world market and price for X (1) supply - demand for utility (2) market for unskilled labor in i (2)

market for skilled labor in i (2)

Income balance inequalities

incomes

description (no. of inequalities) income balance in country i (2)

Comments and Discussion

Alan V. Deardorff: Jim Markusen has done an excellent job of putting the phenomenon of offshoring into perspective. He has done this by drawing upon models of international trade and foreign direct investment to tell a series of stories about why offshoring may occur and what effects it will have. These stories take the form of several explicit models, for each of which he explains the structure and reports results. The results reported are only qualitative, but they are based on what were apparently numerous quantitative simulations. As I read his paper, I found the explanations given very intuitive and helpful. I especially appreciated his emphasis on why scarce factors may be cheap in some countries despite their scarcity, a result of being complementary to another factor that is even scarcer. As he puts it, scarce factors have nothing to do. Offshoring to take advantage of these services may make the services of the complementary factor available, benefiting both them and the world. I like that story a lot. When it came to the results of the models, however, I found myself more confused. Each model had its own list of results, and I had to keep flipping pages to compare them. So the main contribution I will make in these comments is to put these results into one place. In table 1, I report both the assumptions of Markusen’s models 1–4 and the results, all taken directly from his paper. Below the table is the notation I use, supplementing Markusen’s factors H, U, and K and his goods/fragments Y, X, M, and S with factor prices s for skilled labor, u for unskilled labor, and r for know-how. Most of the assumptions, at the top of the table, are common to all four models, which are distinguished by whether they have two or three factors and by whether they assume perfect or Cournot competition. 24

James R. Markusen

25

Table 1. Markusen’s Models: Assumptions and Results for Countries and Factors Model H-O with fragmentation 1 Assumptions Final goods Fragments (of X) Skill intensity ranking North abundance South abundance Factors of production Competition in X

2 (H, U) Perfect

Results North (country) South (country) North (s/w) North (s) North (w) North (r) South (s/w) South (s) South (w) South (r)

–a + + –a – … –a + … …

Missing Input (MI) 2

Multinational (MNC) 3

2 (Y, X) 2 (M, S) M>X>S>Y H (and K) U 3 (H, U, K) 2 (H, U) Perfect Cournot

–a + … – + + … + – –

+ + + + – … + + + …

MI + MNC 4

3 (H, U, K) Cournot

–a + – – + + + + + …

a. If North is large and South is small. Factors H = skilled labor Y = unskilled labor K = know-how Factor prices (real) w = unskilled wage s = skilled wage r = return to know-how

Goods and fragments Y = unskilled-labor-intensive final good X = skilled-labor-intensive final good M = high-tech manufacturing fragment of X production S = service fragment of X production

The bottom portion of the table shows the results that Markusen reports for welfare of the two countries and the two or three factors (in the form of their real factor prices) in each country as a result of introducing offshoring/fragmentation. I present simple plus and minus signs, except for those results that depend on country size due to terms-of-trade effects, in which case a footnote indicates that these signs hold only if North is a large country relative to South. In all cases my understanding from Markusen’s paper is that these qualitative results may not be valid for all possible parameters, even within the assumptions stated here,

26


but that these signs represent the solutions for what he takes to be the most plausible range of parameters. The message I take from this collection of results is that almost anything can happen. Even within the constraints of the four models’ common assumptions, these apparently minor variations of assumed structure of competition (perfect versus Cournot) and number of factors (two versus three, with—importantly— the third factor complementary to one of the others) permits a mixture of both plus and minus signs in all but two of the main rows in this table. (I ignore the rows for return to know-how, since this factor does not appear in two of the models.) Only for country South as a whole and for skilled labor within South do the models yield an unambiguous conclusion: these two constituencies gain in all four models. Unskilled labor has mixed results in both countries, as does skilled labor in North and country North as a whole. This is not intended as a criticism of the models or of Markusen for presenting them. On the contrary, I think we need to know that sensible economic models do not provide a consensus message on some issues, and clearly that is the case for offshoring. As Markusen himself emphasizes, this should not be a surprise, since that has been true if one looks objectively at the literatures on other issues of trade theory, including even the gains from trade liberalization. In particular, it is indeed possible that the developed world, represented here by the country North, may lose from offshoring, and so especially may skilled labor within the developed world. This is not a new observation, and Markusen acknowledges others who have said it before, but it is at least helpful to have more light shed on the mechanisms by which this may happen, and Markusen’s models provide that light. There is one thing that Markusen does not do with his model that I wish he had, which would be to calculate the effect on welfare of the world as a whole. Admittedly, actual residents of the real world may not care about that, since their own welfare will be better tracked by the separate countries and factors. But a trade economist like myself would like to know the answer to this. I presume that the answer must be that the world as a whole must gain from the introduction of offshoring, since it represents an improvement in the efficiency with which resources are used worldwide, and the models here, or at least those with perfect competition, do not seem to have the sorts of market distortions that could render an improvement in efficiency harmful. But admittedly, the welfare effect on the world as a whole, even if it were reported, would not help us to resolve these ambiguities in effects on countries and factors. And that raises the question of how we might best go about resolving these ambiguities.

James R. Markusen

27

On that, I am even more at sea than Markusen’s models. Should we use case studies of individual episodes of offshoring for which more detailed information might be available? That would certainly be useful, but it could hardly be complete, since such case studies would probably be confined to single industries and they would miss the important general-equilibrium and terms-of-trade effects that drive models like Markusen’s. Alternatively, one might imagine econometric studies of offshoring together with national incomes that could provide estimates of the effect of the former on the latter. But surely, if we cannot even reach a consensus on the effects of trade more broadly on economic growth, then this approach does not seem very promising. Finally, Markusen’s own modeling suggests that perhaps we can build more elaborate models, incorporating the features he has here, but more deliberately replicating the data and parameters of the real world. Such computable general equilibrium (CGE) modeling has become commonplace for analysis of trade policy changes, and perhaps it can be applied productively here. But the truth is that such models rest very critically on assumptions made about model and market structure, which the data themselves are seldom able to inform. Seeing how sensitive Markusen’s simple models are to such assumptions, I doubt that CGE models applied to this problem could tell us much more than we already know here: that almost anything can happen. If that is the state of our knowledge about offshoring, what should be our policy advice? Knowing that losses from offshoring are possible, should we recommend that protectionist policies be employed to prevent it? Surely not, since in our ignorance we might as easily be depriving ourselves of benefits as of costs. Should we therefore advocate that offshoring be permitted to proceed unabated, regardless of the cost that it may impose? Perhaps, but if we are honest about our confidence that it will be beneficial, we may not be listened to. Maybe the best approach is not to condemn or to praise offshoring across the board, but to consider each example of offshoring on a case-by-case basis. That makes sense, except that I am not sure what information we should even want to have in order to judge it case-by-case. In the end, although I very much appreciate the insights that Markusen has given us with his series of models, I find myself knowing even less about the likely effects of offshoring than I did before I read his paper. Douglas A. Irwin: Let me begin by saying that I am very sympathetic to the sentiments expressed by James Markusen at the beginning of his paper—that trade theory is a portfolio of models, and that the art of economics is matching

28


the phenomena you wish to analyze with the structural assumptions that you build into a model. As John Maynard Keynes wrote in the early 1920s: “The theory of economics does not furnish a body of settled conclusions immediately applicable to policy. It is a method rather than a doctrine, an apparatus of the mind, a technique of thinking, which helps its possessor to draw correct conclusions” (see Moggeridge 1983). I think this is the spirit in which Jim Markusen presents his paper. Markusen is off to a very good start in moving the analytics of the debate over offshoring forward. The contending views of Samuelson (2004) and Bhagwati, Panagariya, and Srinivasan (2004) represent the first published intuitions of trade economists on the issue, but the phenomenon cries out for a more detailed theoretical analysis. This is what Markusen does, mixing elements of standard trade models to match what we think is happening in international trade in services, and in the offshoring of white-collar workers in particular. The simulations he conducts will give all trade economists a new benchmark for thinking about the possible effects of offshoring on factor incomes and welfare at home and abroad. The one danger in this approach, however, is the risk of sliding down the slippery slope toward taxonomy. The problem is that the underlying phenomenon of offshoring is not without ambiguity (is it simply newly tradable services?). This means that we do not necessarily know the best modeling strategy for approaching it. It is not even clear what the comparative static exercise should be (a reduction in trading costs for services?). Since we are not sure how to model offshoring correctly, perhaps the best strategy (adopted by Markusen) is to play with a variety of structural assumptions. But this leaves us with a multitude of possible outcomes, and the varying results may not be too informative. Even with this danger in mind, Markusen should be commended for trying to sort out what possible outcomes emerge from a standard calibrated trade model simulation. In this comment, I will not take issue with the particular simulations that he undertakes, but rather make three points that suggest that we should take the results with a healthy dose of skepticism. First, in line with the title of the conference, Markusen assumes throughout that it is moderately to highly skilled labor that gets offshored. He does not hide the fact that his results hinge on a specific factor-intensity ordering, and indeed suggests the ordering is crucial to his findings. Yet the types of personnel that are commonly discussed as being offshored include (a) call center workers and secretaries (presumably unskilled labor), (b) financial analysts and traders (presumably moderately skilled labor), and (c) radiologists and information technology personnel (presumably highly skilled labor). It seems that, across a

James R. Markusen

29

variety of service activities, many different types of labor are potentially ripe for offshoring. As a result, I am not sure that we can generalize strictly on the basis of factor intensity what sort of outcomes we might expect. It could be that some interaction with technology, rather than the skill-type of labor, is the defining characteristic of what activities can be offshored to other countries. (This can be considered a call for more empirical work on who and what gets offshored.) Second, Markusen finds that, under a range of modeling approaches, North tends to lose from the tradability of services because of adverse terms-of-trade effects. This consistent finding reminds me of the tendency for older-generation, single-country-computable general equilibrium models employing the Armington assumption to find that a unilateral tariff reduction would be welfare worsening owing to adverse terms-of-trade effects—even for seemingly small, pricetaking countries such as Israel or Malaysia. Later study showed that these results were an artifact of modeling strategy (the Armington assumption of product differentiation in trade). This assumption built into the model a degree of market power that the countries did not actually possess. I raise this point as a reminder that the dominant terms-of-trade effect could, to some extent, be model driven. Alternative modeling approaches would likely lead to a different set of outcomes. For example, much of what is offshored seems to fall into the category of intermediate business services. If these services were built into an Ethier type of production function, where product differentiation and variety are valued, then making these services tradable would allow firms to add to the array of intermediates that they could potentially consume. Like trade in intermediate goods, the possibility of trade in these intermediate services might have positive productivity effects that would redound to the benefit of consumers in both North and South. Third, the work of Alan Deardorff (2005a, 2005b) in this area is relevant. Deardorff finds that defining comparative advantage is difficult in a world of Ricardian trade with intermediate inputs. In a Ricardian framework, Deardorff finds that trade patterns exhibit a high degree of sensitivity to trade costs. As my first comment suggests, it could be that sensitivity to trade costs is the important defining characteristic of offshoring, not a specific type of factor intensity. By contrast, in the pure technology case considered by Deardorff, allowing trade in intermediate goods almost always results in an economic benefit to North. (Deardorff’s other work on lumpy countries might help to explain why there is offshoring to countries abundant in unskilled labor. Both coastal China and Bangalore, India, might be viewed as agglomerations of physical or human capital within larger polities abundant in unskilled labor.)

30


As a final point, the economic historian in me cannot help but observe that David Landes, in his book The Wealth and Poverty of Nations, noted that outsourcing and offshoring go back to thirteenth-century Europe. The putting-out system, whereby textile firms located in urban areas would employ cheap labor (primarily women) in the countryside as part of the production process, was opposed by the city guild officials and generated tremendous controversy at the time. I doubt that there were thirteenth-century modeling efforts to analyze the issue, but it is clear that offshoring and outsourcing are not necessarily new phenomena. General Discussion: Robert Feenstra launched the general discussion. He first raised the issue of whether India has abundant skilled labor. Recent data he has seen show that India’s share of the world’s scientists exceeds its share of world GDP. This is the correct criterion for factor abundance according to the Heckscher-Ohlin model. However, he noted that India is, of course, relatively much more abundant in unskilled labor. Feenstra also discussed the implications of offshoring for real wages of both skilled and unskilled labor in the North. In a model he developed with Gordon Hanson, it is possible for both types of real wages to rise, so that the North gains from offshoring. This reflects a Wal-Mart effect, in which offshoring raises world efficiency and reduces consumer prices. Thus, he questioned whether the results were as clear-cut as the models presented in James Markusen’s paper suggest. Shang-Jin Wei argued that a class of models was potentially missing from Markusen’s paper. He pointed out that services trade goes in both directions, with rich countries both importing from and exporting to developing countries. This feature is not captured in any of Markusen’s models. In fact, while media reports have been focusing on service outsourcing from the United States to India and other developing countries, the United States actually runs a surplus in services. He suggested that allowing for skill-intensive work to be offshored in both directions could provide a channel through which skilled labor in the North would gain. Dalia Marin discussed lessons from research, by herself and by others, that examines implications of offshoring to eastern Europe by Germany and Austria. The effects are difficult to tease out for Germany, because eastern Europe accounts for a very small percentage of its outgoing foreign investments. In contrast, most of the foreign investment from Austria goes to the region. She stated that the empirical analysis does support the conclusion that skilled Austrian labor has been hurt. In more recent work with more recent data she finds that German human capital has been hurt by outsourcing worldwide (including to eastern Europe). Her finding suggests that the distinction between manufacturing outsourcing and service outsourcing does not seem to be the right one, since Germany outsources

James R. Markusen

31

mainly in the manufacturing sector. What appears to matter is that the outsourced activity of German firms is almost three times as skill-intensive as the firms’activity in Germany. Catherine Mann stressed that severe measurement issues limit how much we actually know about the effects of trade in services for workers. Instead of simply jumping from trade to workers, she argued that we need to understand the mechanisms better: how services trade influences prices, which would affect production and technology, and through those channels affect employment. But we know very little about prices and costs for services, particularly in areas such as software, where services are often bundled together with goods. Many participants agreed that supporting the statistical community in finding ways to improve available measures should be a high priority.

32


References Antràs, Pol. 2003. “Firms, Contracts, and Trade Structure.” Quarterly Journal of Economics 118 (4): 1375–1418. ———. 2005. “Incomplete Contracts and the Product Cycle.” American Economic Review (September): 1054–73. Antràs, Pol, and Elhanan Helpman. 2004. “Global Sourcing.” Journal of Political Economy 112 (3): 552–580. Bhagwati, Jagdish, Arvind Panagariya, and T. N. Srinivasan. 2004. “The Muddles over Outsourcing.” Journal of Economic Perspectives 18 (Fall): 93–114. Brainard, S. Lael. 1997. “An Empirical Assessment of the Proximity-Concentration Tradeoff between Multinational Sales and Trade.” American Economic Review 87 (4): 520–44. Deardorff, Alan V. 2001. “Fragmentation in Simple Trade Models.” North American Journal of Economics and Finance 12 (2): 121–37. ———. 2005a. “Ricardian Comparative Advantage with Intermediate Inputs.” North American Journal of Economics and Finance 16 (1): 11–34. ———. 2005b. “A Trade Theorist’s Take on Skilled-Labor Outsourcing.” International Review of Economics and Finance 14 (3): 259–71. Dornbusch, Rudiger, Stanley Fischer, and Paul A. Samuelson. 1977. “Comparative Advantage, Trade, and Payments in a Ricardian Model with a Continuum of Goods.” American Economic Review 67 (5): 823–39. Dunning, John H. 1977. “Trade, Location of Economic Activity and the Multinational Enterprise: A Search for an Eclectic Approach.” In The International Allocation of Economic Activity, edited by B. Ohlin, P. O. Hesselborn, and P. M. Wijkman, pp. 395–418. New York: Macmillan. Ethier, Wilfred. 1982. “National and International Returns to Scale in the Modern Theory of International Trade.” American Economic Review 72 (3): 389–405. Ethier, Wilfred, and James R. Markusen. 1996. “Multinational Firms, Technology Diffusion and Trade.” Journal of International Economics 41 (1): 1–28. Feenstra, Robert C., and Gordon H. Hanson. 1996a. “Foreign Investment, Outsourcing, and Relative Wages.” In The Political Economy of Trade Policy: Papers in Honor of Jagdish Bhagwati, edited by R. C. Feenstra, G. M. Grossman, and D. A. Irwin, pp. 89–127. MIT Press. ———. 1996b. “Globalization, Outsourcing, and Wage Inequality.” American Economic Review 86 (2): 240–45. ———. 1997. “Foreign Direct Investment and Relative Wages: Evidence from Mexico’s Maquiladoras.” Journal of International Economics 42 (3): 371–93. Fosfuri, Andrea, Massimo Motta, and Thomas Rønde. 2001. “Foreign Direct Investment and Spillovers through Workers’ Mobility.” Journal of International Economics 53 (1): 205–22. Glass, Amy Joyce, and Kamal Saggi. 2002. “Multinational Firms and Technology Transfer.” Scandinavian Journal of Economics 104 (4): 495–513.

James R. Markusen

33

Gomory, Ralph E., And William J. Baumol. 2004. “Globalization: Prospects, Promise, and Problems.” Journal of Policy Modeling 26 (4): 425–38. Grossman, Stanford J., and Oliver D. Hart. 1986. “The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration.” Journal of Political Economy 94 (4): 691–719. Grubel, Herbert. 1987. “All Traded Services Are Embodied in Materials or People.” World Economy 10 (3): 319–30. Helpman, Elhanan.1984. “A Simple Theory of Trade with Multinational Corporations.” Journal of Political Economy 92 (3): 451–71. Helpman, Elhanan, Mark J. Melitz, and Stephen R. Yeaple. 2004. “Exports versus FDI with Heterogeneous Firms.” American Economic Review 94 (1): 300–16. Hummels, David, Jun Ishii, and Kie-Mu Yi. 2001. “The Nature and Growth of Vertical Specialization in World Trade.” Journal of International Economics 54 (1): 75–96. Jones, Ronald W., I. Coelho, and Stephen T. Easton.1986. “The Theory of International Factor Flows: The Basic Model.” Journal of International Economics 20 (3–4): 313–27. Koopmans, Tjalling. 1957. Three Essays on the State of Economic Science. New York: McGraw-Hill. Landes, David. 1998. The Wealth and Poverty of Nations: Why Some Are So Rich and Some Are So Poor. New York: W. W. Norton. Magee, Stephen P., and Kwang-Yeol Yoo. 2005. “The United States Is a Small Country in World Trade: Further Evidence and Implications for Globalization.” Working Paper. University of Texas. Markusen, James R. 1983. “Factor Movements and Commodity Trade as Complements.” Journal of International Economics 14 (3–4): 341–56. ———. 1984. “Multinationals, Multi-Plant Economies, and the Gains from Trade.” Journal of International Economics 16 (3–4): 205–26. ———. 1989. “Trade in Producer Services and in Other Specialized Intermediate Inputs.” American Economic Review 79 (1): 85–95. ———. 1995. “The Boundaries of Multinational Firms and the Theory of International Trade.” Journal of Economic Perspectives 9 (2): 169–89. ———. 2001. “Contracts, Intellectual Property Rights, and Multinational Investment in Developing Countries.” Journal of International Economics 53 (1): 189–204. ———. 2002. Multinational Firms and the Theory of International Trade. MIT Press. Markusen, James R., and Anthony J. Venables. 2005. “A Multi-country Approach to Factors Proportions Trade and Trade Costs.” Working Paper 11051. Cambridge, Mass.: National Bureau of Economic Research. Melitz, Mark J. 2003. “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity.” Econometrica 71 (6): 1695–1725. Moggeridge, Donald, ed. 1983. The Collected Writings of John Maynard Keynes. Cambridge, Eng.: Macmillan. Mundell, Robert A. 1957. “International Trade and Factor Mobility.” American Economic Review 47 (3): 321–35.

34


Neary, J. Peter. 1995. “Factor Mobility and International Trade.” Canadian Journal of Economics 28: S4–S23. Samuelson, Paul A. 2004. “Where Ricardo and Mill Rebut and Confirm Arguments of Mainstream Economists Supporting Globalization.” Journal of Economic Perspectives 18 (3): 135–46. Yi, Kei-Mu. 2003. “Can Vertical Specialization Explain the Growth of World Trade.” Journal of Political Economy 111 (1): 52–102. Zhang, Kevin Honglin, and James R. Markusen. 1999. “Vertical Multinational and Host Country Characteristics.” Journal of Development Economics 59 (2): 233–52.

DANIEL TREFLER University of Toronto

Service Offshoring: Threats and Opportunities

W

hen asked to provide a framework piece on offshoring, I decided it would be much easier to have the work done by an Indian consulting firm. A quick bit of research turned up a perfect corporate partner. Not surprisingly, the company has a London-based front end—it is a fact of the industry that many customers prefer to work through a Western intermediary. The company quoted the job at $63,000, no taxes. That’s about one-tenth of what an American management consulting firm would charge, but still too rich for my academic salary. So you are stuck with me. The experience taught me two things. First, you can outsource abroad just about anything, from which I conclude that all of our jobs are threatened. Second, the big money in offshore outsourcing goes to the OECD business analysts who help customers communicate their needs to business process outsourcers in low-cost countries. I conclude from this that offshoring brings remarkable opportunities to us all. Therein lies the paradox of offshoring: it is both a threat and an opportunity. In considering international offshoring, two trends scream out for our attention. The first is the rise of China as the world’s manufacturer. Surprisingly, many American firms have yet to wake up to this sea change in their sourcing possibilities. Better information about the strategic offshoring options available to American firms is desperately needed. Aside from this, the rise of China’s

I am indebted to Someshwar Rao and Prakash Sharma of Industry Canada for initiating this study and to both Runjuan Liu of the University of Alberta and Nathan Nunn of the University of British Columbia for help with completing this paper. Lael Brainard and Dani Rodrik provided many insightful comments, as did Susan Collins, who also provided extensive editorial direction. Belinda Lobo cheerfully provided secretarial support. Industry Canada graciously funded most of this study. I am also grateful to the Brookings Institution for funding.

35

36


manufacturing sector poses no new public policy issues. All the familiar arguments hold. On the one hand, international trade is disruptive for workers and firms engaged in import-competing industries. On the other hand, international trade provides the benefits of lower prices to consumers and offers new opportunities for producers (both workers and firms) to expand into foreign markets. In aggregate, the benefits outweigh the costs. What remains for policymakers is the crucial task of ensuring that we generously care for our most disadvantaged, since these unskilled workers are the ones who will bear the brunt of the Chinese offshoring onslaught. The second extraordinary development in international trade has been the rapid growth of traded services involving innovative, technology-intensive processes and employing high-paid white-collar workers. In the past it was unheard of for low-cost countries such as India to be exporting high-value-added services. Now it is common to find Indian software programmers customizing sophisticated software applications for businesses worldwide. This development fundamentally alters the way we must think about innovation-based corporate strategy and public policies that affect the flexibility of the white-collar labor market. The United States faces a choice. It can insulate itself from the global competitive pressures that come with offshoring to low-cost countries. Such policies will protect firms and workers in the short run. However, there is at least some weak evidence that protectionism retards growth.1 In addition, insulating policies will likely encourage foreign countries to deny us market access. Considering that the United States is a major supplier of traded services to the rest of the world, insular policies are about as useful as a blow-dryer in an igloo. Alternatively, the United States can pursue domestic framework policies that promote the competitiveness of U.S. firms and workers. These framework policies would encourage productivity-enhancing investments both by individuals (for example, in human capital) and by firms (for example, in R&D and advanced technologies). The building blocks for globally competitive American firms are domestic policies that encourage continual investments in upgrading and innovation by individuals and firms. When it comes to the U.S. public policy response to offshoring, my best advice is: think globally, invest locally. Finally, let’s not forget about compassion. The American government must be prepared to generously help its most disadvantaged, for they are at greatest risk from the downside of offshoring. 1. See Nunn and Trefler (2005) for support of this view and Rodriguez and Rodrik (2001) for a scathing rebuke of the openness-and-growth literature.

Daniel Trefler

37

What Is Offshoring? There is no universal definition of offshoring, and one task of the Brookings Trade Forum is to decide how broad a set of phenomena to examine. The approach taken by all commentators on offshoring is to attempt a careful definition. This is a natural but misguided approach. We must first start by identifying America’s broad public policy objectives and then identify which aspects of offshoring enhance or impinge on our ability to meet these objectives. In my view there are two complementary objectives: (1) promoting competitiveness and raising incomes; and (2) advancing core values of community and caring through redistributive policies. The most interesting policies are the few that promote both objectives. These objectives will help us delineate the boundaries of a discussion of offshoring by answering three definitional questions. Offshoring, Nearshoring, Inshoring, or All of the Above? As I have already stated, the most interesting aspects of new trends in the tradability of services is the offshoring of technology-intensive, high-end services to low-wage countries. There are two other phenomena of interest: (1) Nearshoring: Much of U.S. offshoring is nearshoring to Canada—for example, to a call center in Toronto that services customers in Chicago. Yet Canada is a country that is very close to the United States (whence nearshoring), and more important, a country that is hardly a low-wage producer. (2) Inshoring: The United States is a major supplier of traded services to the rest of the world. This exporting of services or inshoring cannot be ignored. (Slaughter [2004] calls this “insourcing.”) Offshoring, nearshoring, and inshoring must all be examined. Offshore Outsourcing or Foreign Direct Investment (FDI)? “Offshore outsourcing” describes an arm’s-length transaction between a U.S. firm and a foreign firm. In contrast, FDI describes a domestic firm with a controlling equity investment in a foreign establishment. Recent theories of international trade make it clear that the distinction between offshore outsourcing and FDI is intimately related to the question of whether the United States will retain the highest-paying jobs in the value chain or watch them migrate both to other OECD countries and to emerging low-cost countries such as China and India. One cannot understand this process without looking at what is called the make-

38


or-buy decision—that is, the decision about whether to produce in-house using FDI or to offshore outsource using arm’s-length transactions. In a nutshell, the new theories state that when a project is sufficiently routinized that it can be fully scoped or described, then outsourcing is the appropriate relationship with a foreign service provider. When the project is difficult to describe from its outset, it should be done in-house via FDI. For example, see Antràs (2005). The difficult-to-describe projects are typically the innovative projects that generate the highest value added. Thus we need to understand how firms choose between offshore outsourcing and FDI if we are to understand how to keep high-paying jobs in the United States. My suggestion is thus to study both offshore outsourcing and FDI. Not all economists will agree. For example, Bhagwati, Panagariya, and Srinivasan (2004) argue that we should only be thinking about offshore outsourcing. On this one point, I think that Bhagwati, Panamanian, and Sinicising are wrong.2 It is fitting to develop this discussion of offshoring by providing examples of its pervasiveness and the difficulties of further definitional refinements. Example 1. Traditional “mode 3” FDI in the service sector: Citibank sets up an office in Hong Kong that provides limited services to Chinese customers. The office is staffed primarily by Chinese, and most of the key decisions are made in the United States. Example 2. Traditional “mode 4” FDI in the service sector: A U.S. architectural firm sets up an office in Shanghai to bid and work on local contracts. The firm sends its American architects to Shanghai on a long-term basis to do the design work. What distinguishes this from the previous example is that the control of decisions is largely in the hands of Americans who have temporarily migrated to Shanghai. Example 3. The service-trade revolution using an FDI mode of entry: Verizon sets up an information technology (IT) center in Bangalore that hires Indian programmers to write software for Verizon’s U.S. operations. Example 4. The service-trade revolution with an offshore outsourcing mode of entry: Satyam (India) sets up a contact center that makes Wells Fargo VISA marketing calls to potential customers in Seattle. The use of the term “mode” comes from the IMF (2005) Balance of Payments Manual and is used by all OECD countries in presenting their data.3 2. These misspelled names were introduced by a Chinese student who typed up my corrections to this paper. The typos typify the monitoring and agency problems associated with offshore outsourcing. 3. The manual distinguishes four modes based on the location of the supplier and consumer of the traded service. Mode 1: The supplier is in one country and the consumer is in another. Each

Daniel Trefler

39

Table 1. Definitions of Export-Oriented FDI Projects Related to Offshored Services Contact center services Help desk Technical support/advice After-sales support Employee inquiries Claims inquiries Customer support and advice Market research Answering services Prospecting Information services Customer relationship management

Back-office services Claims processing Accounts processing Transaction processing Query management processing Customer administration processing HR and payroll processing Data processing IT outsourcing Logistics processing Quality assurance Supplier invoices

IT services Software development Application testing Content development Engineering and design Product optimization Other high-end services Regional headquarters Architectural services Biotech and pharmaceuticals R&D Radiology, X-ray Distance education

Source: UNCTAD (2004) and author.

Table 1 provides many more examples of the types of activities that I believe we should focus on. These examples are classified into four areas: contact centers or what are commonly called call centers, back-office services, IT services, and other high-end services. It is worth noting a problem with refining the definition of offshoring. Most of us would be comfortable with the following statement: “Manulife is offshore outsourcing development of its new human resources software to India, while the plastic products industry is importing shopping bags from China.” Why is one “offshore outsourcing” and the other “importing”? In both cases, products currently made in Asia were previously made in-house in America, and in both cases there has been phenomenal growth over the past five years. There are no good answers to this question. Given this problem of definition (and other problems as well), finer definition of offshoring seems impossible. I therefore adopt the approach of U.S. Supreme Court Justice Potter Stewart in his attempt to define pornography: I can’t define it, but I know it when I see it.4 stays in his or her own country. Mode 2: The consumer moves to the supplier’s country to obtain the service. Mode 3: The supplier sets up a foreign affiliate in the consumer’s country. Mode 4: The supplier supplies the service by moving to the consumer’s country. For more, see International Monetary Fund, Balance of Payments Manual (www.imf.org/external/np/sta/ bop/BOPman.pdf [2005]). 4. Jacobellis v. Ohio, 378 U.S. 184 (1964).

40


Occupations or Industries? What makes manufacturing interesting is the dramatic rise in manufacturing exports from low-cost countries, especially China. This export surge has already had a large impact on America’s least-skilled workers in industries such as garments. See Feenstra and Hanson (1996, 1999) and Trefler (1998). It is now poised to threaten America’s moderately skilled workers in such industries as auto parts. However, to my mind these developments pose no new public policy issues that have not already been discussed in the context of conventional import competition. The reason it is not new is that it affects occupations that have always been impacted by international trade. On the other hand, the revolution in the world’s ability to trade in services is something new. At least some of the new service trade involves highly skilled white-collar workers operating in low-cost countries such as India. Successful policy responses aimed at assisting skilled labor will likely be very different from policy responses aimed at assisting less-skilled labor. This distinction has had no play in the offshoring debate but is likely crucial for reasons to be explained below. Thus, service offshoring poses new policy challenges not raised by manufacturing offshoring because it involves white-collar workers. My view will find critics. Most researchers argue that the rise of China as the world’s manufacturer poses such important challenges that it must be included in every discussion of international trade policy. I look forward to a healthy debate of this point. Another problem with focusing on industries rather than occupations stems from recent changes in traditional manufacturing. With the offshore outsourcing of back-office jobs by manufacturing firms, we tend to think that the line between manufacturing and services is becoming cleaner. However, the opposite is also happening. When Microsoft introduced its Xbox game player, it hired Singaporebased Flextronics (the contract manufacturing giant) to build a factory in lowwage Guadalajara, Mexico, that was supplied with standardized parts from China. Design of the core proprietary technology was outsourced to Nvidia Corp. of the Bay Area and manufactured in Taiwan. Clearly, Xbox could not have been brought to market in this way without tremendous logistics support. As such, Xbox is a manufactured product that embodies a significant service component. This example is commonplace. Accenture (2004) reports that 43 percent of its customers outsource their supply chain management. This reflects the rise of contract manufacturers that both manufacture and provide manufacturing service support. Thus, in many respects traditional industry distinctions are blurring. Focusing on occupations is much cleaner and more useful for policy.

Daniel Trefler

41

White-Collar Workers and the New Trade Issues Raised What is most novel about the recent emergence of offshoring is that it affects white-collar workers employed in technology-intensive industries (be these services or manufacturing). We simply do not know what the net effects of this are because empirical trade economists have virtually no experience with this phenomenon. Three issues need to be researched: —Many (though not all) white-collar jobs are high-paying jobs, paying $70,000 or more a year. As a country, we are familiar with losing high-paying jobs (head-office service jobs and banking service jobs, for example) to other rich countries. What we are less familiar with is losing high-paying jobs to India. We certainly want to avoid losing these good jobs. However, these losses are somewhat offset by the jobs created for U.S. business analysts with IT expertise. These Americans work as highly paid intermediaries who interface between U.S. companies and Indian business service providers. —When a white-collar job is offshored, the value of an American worker’s industry-specific and firm-specific knowledge is destroyed. This stands in contrast to what happens when an unskilled worker is displaced. There is little valuable knowledge to be destroyed. It is unclear whether loss of such knowledge is an equity concern alone (because it hurts offshore-displaced workers) or whether it is also an efficiency concern (because it destroys valuable human capital). This needs to be investigated. —There is now a large literature showing that retraining programs are not effective for most displaced workers (see, for example, Baicker and Rehavi 2004). The argument is that unskilled workers are unskilled for a reason: they are missing the most fundamental of abilities, namely the ability to learn (see Heckman and Carneiro 2003 and Trefler 2004). This means that displaced unskilled workers need income transfers to handle trade shocks. In contrast, IT professionals are likely to be highly motivated individuals who would do well in retraining programs.

The 64,000 Job Question: Whither China and India? Behind the alarm about service offshoring is a sense that OECD countries are in danger of being overtaken by China, India, and a number of other developingcountry destinations for service offshoring. In the most alarming scenario, these countries have an infinite capacity to absorb OECD technologies and management strategies, to improve on them, and ultimately to compete head-to-head

42


with the OECD. Finally, in this scenario, China and India with their newly acquired high-tech status will continue to pay low wages for skilled labor and will use this advantage to create an economic steamroller that crushes all OECD countries. There are two reasons why this argument is flawed. First, there is an ironclad economic law that prevents one country from ever dominating world trade. Second, there are political-economic reasons to doubt the speed at which this scenario can unfold. I review these reasons in detail. The Ironclad Law of Comparative Advantage I am a better researcher than my secretary. Surprisingly, I am also a better typist than he is. That is, I have an absolute advantage over my secretary in both research and typing. Nevertheless, I find my secretary to be indispensable. That is because I am relatively better at research than typing. Thus if I typed an hour less a day I could write one page of this paper, whereas if my secretary typed an hour less a day he could only write one sentence of this report. In economic jargon, I have a comparative advantage in research, and my secretary has a comparative advantage in typing. In the most alarmist scenarios about China and India, these countries will soon have an absolute advantage in producing all goods and services. However, the United States will continue to have a comparative advantage in the most knowledge-intensive goods and services. Thus, even in the most alarmist scenario, the United States will continue to export knowledge-intensive goods and services to China and India. With their low wages, what prevents these countries from exporting everything and importing nothing? If they import nothing they will be giving their goods away for free. I doubt they would agree to this. In addition, the Americans will need yuan to buy Chinese goods. As we demand more of their currency it will rise in value. Eventually, the yuan will rise so much in value that Chinese wages are no longer so dominantly competitive. (This is exactly the problem the United States faces when its currency is strengthening.) In real life there are things China can do to slow this process down, but China cannot forever keep the yuan undervalued. This is an ironclad law. Countries such as Germany in the 1960s and Japan in the 1970s ran afoul of the comparative advantage police. They ran huge trade surpluses that threatened to destroy U.S. manufacturing. Over time, however, their currencies strengthened to the point where these countries ceased being low-cost producers. In this context it is important to remember that in 1959 Japan had a highly skilled and disciplined

Daniel Trefler

43

labor force that was paid 10 percent of U.S. wages. Japan in 1959 was, from the limited perspective of offshoring, not that different from China today. Yet Japan never was able to dominate world manufacturing. Why? Because Japan succumbed to the comparative advantage police by steadily revaluing the yen. The same will eventually happen to China. It does not matter that they have hundreds of millions of citizens ready to work for next to nothing. If we buy too much from them, their currency will rise to the point where their low yuandenominated wages are wiped out by the currency conversion. It does not matter that Chinese workers are paid 4 yuan an hour unchanged over the next hundred years. If the yuan strengthens, Chinese dollar-denominated wages will rise. Like the Mounties, the comparative advantage police always get their man. Institutions and the Mystery of Modern Economic Growth The comparative advantage argument has one significant limitation. It is possible that China and India will develop a comparative advantage in knowledgeintensive goods and services, leaving the United States to produce T-shirts for the Shanghai market. In this scenario, the United States continues to export to China according to the law of comparative advantage. However, the United States becomes poor relative to China and possibly poor even in absolute terms. The argument for absolute impoverishment was first made by Graham (1923) and has been repeated by Hicks (1953), Johnson and Stafford (1993), Gomory and Baumol (2000), and most recently by Samuelson (2004). While the argument is logically correct, fortunately for the United States it is irrelevant. The problem with the argument is that it presumes that China and India will become the world’s technological leaders. Such a presumption is in flagrant contradiction to what we know about the role of domestic institutions for promoting innovation. Current thinking about the determinants of long-term economic growth focuses on the central role of domestic institutions. See Helpman (2004) for a review of the literature. In this view, there are limits to what China and India can produce under their current political-legal-economic regimes. As China and India expand the range of services they provide, they will eventually enter into services that depend on constant innovation. In the new institutions-and-growth view, innovation cannot occur without institutions that protect property rights, that provide a fully functioning legal framework for arm’s-length transactions, that support thick equity and debt markets, and that balance the needs of inside innovators against those of outside investors. Srinivasan (this volume) argues that these institutional constraints on growth were made irrelevant in India’s IT sector because of special regulatory provisions

44


and protections afforded the sector. While I would certainly never want to disagree with Professor Srinivasan—wait, I think I already have once in this paper— my point is less about the development of a single sector and more about longterm, innovation-based, multisectoral, modern economic growth. In short, China and India will not be able to compete in innovation-intensive sectors without the “institutions of modern capitalism” (Rosenberg and Birdzell 1986) and its handmaiden, “the invention of invention”(Mokyr 1990). For China and India to compete over the very long haul, their institutions will have to look a lot more like OECD institutions. This is unlikely to occur even over a quartercentury horizon. Evidence on the Importance of Institutions for Long-Run Growth Figure 1 provides two examples of a now-burgeoning institutions-and-growth literature. The top panel plots GDP per capita in 1997 against the Kaufmann, Kraay, and Zoido-Lobaton (1999) rule-of-law index. This index ranks countries based on the degree of rent-seeking or opportunistic behavior that investors are exposed to. For example, when I make an equity investment in a U.S. company I have some confidence that I will see my money again—not always, but usually. In contrast, my equity investment in China is much more likely to be siphoned out of the company and forever lost to me. The figure 1 R 2 of 71 percent shows just how much rent-seeking behavior can retard growth. The bottom panel plots GDP per capita against the Gwartney and Lawson (2003) legal-quality index. This index captures the ability of firms to write enforceable contracts. The need for rule of law governing commercial transactions is obvious. Later in this paper I discuss how important it is for understanding offshoring. The bottom panel of figure 1 shows how important the quality of legal institutions is for growth. Of course, India and especially China have grown rapidly with weak institutions. But as Alwyn Young (1992, 1994) has pointed out, much of this growth is based on unsustainable factor supply growth rather than on productivity growth. It is the latter that is the basis for modern economic growth. For example, reforms in China and India have led to a movement of workers from farms to cities, thus providing manufacturing with an almost infinite supply of cheap labor. In contrast, long-run sustainable growth of the kind experienced in OECD countries is driven by innovation-induced productivity growth. And rock-solid institutions are the supporting architecture for innovation.

Daniel Trefler

45

Figure 1. Good Institutions Promote Growth Log real per capita GDP

10 9 Canada and USA

China

8

India R 2 = 0.71

7 6 5 –1.75

–0.75

0.25 Rule of law

1.25

2.25

Log real per capita GDP

10

9

Canada and USA

8

China India

R 2 = 0.60

7

6

3

5 Legal quality

7

Source: Data from Kaufmann, Kraay, and Zoido-Lobaton (1999) and Gwartny and Lawson (2003)

9

46


Institutions in China and India Rich countries have good institutions. The quality of Chinese and Indian growth-enhancing institutions is at best moderate. Historically, very few countries experience rapid improvements in their domestic institutions. Rather, institutions develop at a glacial pace, over a century or more. The idea that China or India can rapidly develop these institutions is a complete misread of the sources of modern economic growth. How does this pan out in the specific contexts of China and India? In personal conversation, Wendy Dobson of the Rotman School of Management has identified five weaknesses in Chinese and Indian institutions: (1) the role of the government—particularly state-owned enterprises and corrupt officials—in preventing the efficient reallocation of resources such as capital; (2) a weak financial system that leaves firms under-resourced; (3) a social safety net that leads to labor market inflexibilities; (4) a lack of an endogenous capability to innovate, in part because entrepreneurs are hemmed in by the rent-seeking behavior of bureaucrats; (5) a one-party state in China and a corrupt alliance between bureaucrats and politicians in India that retards the development of a local entrepreneurial class. Although some of these institutional impediments are slowly evaporating, it will take decades before they all disappear. An Application to the Worldwide Software Industry To make the argument about institutions less abstract, consider how it plays out in the emerging centers of the worldwide software industry, that is, in China, India, Brazil, Ireland, and Israel (see table 2). The industry is very large in India, China, and Brazil. The combined employment of these three countries is 600,000, approaching the U.S. level of 1,024,000. On the other hand, sales per employee are very small in these countries. A U.S. software employee generates almost $200,000 of sales per employee, four times more than an Indian employee. This means that China, India, and Brazil are providing low-valueadded programming skills. Will the software industry in these low-cost countries grow and enter into higher-valued-added segments? Four significant institutional factors may prevent this. —Long-term software growth must be primarily driven by domestic developments. Apart from India, the software industries in these countries developed in response to local needs (Arora and Gambardella 2005; Arora, this volume). Banking and telecommunications drive software growth in Brazil and China,

Daniel Trefler

47

Table 2. The Software Industry Worldwide, 2002 or latest available year

Countries Brazil China India Ireland (multinational enterprises) Ireland (domestic) Israel United States Japan Germany

Sales ($U.S. billion)

Employment (1,000s)

Sales/ employment (1,000s)

Software sales/GDP (percent)

7.7 13.3 12.5

160 190 250

46 38 50

1.5 1.1 2.5

12.3 1.6 4.1 200.0 85.0 39.8

15 13 15 1,024 534 300

804 127 273 195 159 133

10.1 1.3 3.7 2.0 2.0 2.2

Source: Arora and Gambardella (2005).

software growth in Israel was driven by Israel’s high-tech sector, and software firms in Ireland developed by providing services to multinationals using Ireland to enter the European market. In each case, domestic factors drove the initial growth: exports came later. The message, then, is that the institutions that promote domestic-led growth must be in place. —Clusters. In order to have domestic-led growth, many pieces must fall into place simultaneously. For example, the weak financial systems in China, India, and Brazil leave firms under-resourced because insiders routinely steal from outside investors. Thus, firms in these countries are short not only on capital, but also on sophisticated financial advice provided by banks and venture capital firms. Further, downstream demanders of software such as banks are also underdeveloped because of poor national institutions. Therefore, software firms are missing sophisticated buyers who will push them to innovate and upgrade their products (Porter 1998). It is sometimes argued that R&D follows production. Thus, as the low end of the software industry migrates to India, product development will also migrate. Indeed, India’s National Association of Software and Service Companies (NASSCOM) boasts many new products. However, available evidence suggests strong limits to this process. Audretsch and Feldman (1996) show that as an industry matures and manufacturing moves to low-cost locations outside the cluster, R&D continues to occur inside the cluster. Jaffe, Trajtenberg, and Henderson (1993) explain why. Much of what is important for ongoing innovation involves the local exchange of tacit information, that is, information that cannot be codified and that can only be communicated face-to-

48


face. All of this implies institutional limits to the development of an increasingly sophisticated software industry in China, India, and Brazil. —National innovation systems. A skilled labor force is critical for the growth of a domestic software industry. China, Brazil, and India all have large and growing university systems. Each country turns out about as many natural science and engineering degrees as the United States (see, for example, Bardhan and Kroll 2003; Arora and Gambardella 2005; and Arora 2005). It is often argued that this provides these countries with cheap skilled labor. I am more skeptical. If skilled labor is so abundant, why are IT sector wages rising by 15 percent a year in India? The answer is that there is often a significant gap between what the university provides and what the private sector needs. The most successful country in the world at bridging this gap has been the United States. As is well known, the U.S. university system co-evolved with private sector needs. The development of the state university system is a typical example. As a result, the U.S. university system has an unparalleled curriculum vitality. Further, Rothschild (2003) argues that the continued success of the U.S. university system has been driven by competition. On the one hand, U.S. universities compete fiercely among themselves for the best faculty and ideas. On the other hand, the system has diverse revenue sources, and the many funders of U.S. university research compete among themselves to fund the best projects. As a result, there are no misdirected top-down injunctions about how to run engineering schools, and good ideas are rarely suppressed. Universities in China, Brazil, and India are able to crank out large numbers of graduates, but they will be unable to train the world’s best graduates for many decades to come. —International technology transfer. There can be little doubt that OECD multinationals are teaching China and India how to compete. There is also an argument that we are selling ourselves short by underpricing these technology transfers. However, for better or worse, in an open society it is virtually impossible to act differently than we are currently doing. How far will the process of international technology transfer go? Countries with strong protection of intellectual property rights are the favored destination of multinational enterprises (MNEs): these companies go where institutions are strong. Thus weak institutions in China, India, and Brazil will place a limit on technology transfer in the software industry. This section has demonstrated in the context of the software industry that weak Chinese, Indian, and Brazilian domestic institutions will prevent these countries from migrating too far up the software value chain. The United States need not worry that in the next twenty years we will be reduced to mending the socks of Chinese businessmen.

Daniel Trefler

49

The Determinants of Offshore Outsourcing: The Contracting Environment The rise of service offshoring has two main drivers: —Technological improvements in the information and communications technology (ICT) sector. These improvements launched what UNCTAD (2004) calls the “service tradability revolution.” While the financial sector has been using ICTs for fifteen years, developments of the past five years have dramatically reduced costs to the point where ICTs are cheaply available to all. —The new “openness” consensus among political coalitions in developing countries. In the spring of 1992, Deng Xiaoping used a tour of southern China to call for a radical opening up of the Chinese economy to both domestic and foreign competition. Since then, southern China has been growing at 25 percent a year. Likewise, the 1991 financial crisis in India led to the dismantling of tariffs and restrictions on FDI. Across the developing world there has been a wave of reforms aimed at integrating these low-cost countries into the world economy. The rise of manufacturing offshoring has also been greatly facilitated by reductions in transportation costs and improvements in transportation logistics. Conventional wisdom has it that firms go offshore to reduce costs, usually to low-cost countries. This is a misleading view. For one thing, 85 percent of U.S. service offshoring is with other OECD countries. For another, many firms want access to foreign markets in order to tap into new sources of skilled workers, to position themselves in rapidly growing markets, and to be closer to foreign customers. Accenture (2004) reports that lower costs is only third on the list of the most important factors in choosing an offshore outsourcing provider (see figure 2). The first two are service providers’ expertise or capability and service providers’ flexibility. What is most interesting about the list in figure 2 is that most of the items cannot be easily codified or written down in a contract. Mirroring this fact, less than a third of the firms in the Accenture study feel that their offshore outsourcing contract is the key framework for managing the offshore outsourcing relationship. The former CEO of one huge corporation related to me the story of the lengthy contract negotiations he had for a greenfield investment in China. Years of negotiating with the Chinese culminated in a party to celebrate the conclusion of the contract talks. At the party, the Chinese host turned to the CEO and candidly told him that the contract meant nothing to the Chinese partners and that it was only signed to make the CEO comfortable! For the Chinese partners, the important thing was that they trusted the CEO.

50


Figure 2. Importance to Firms of Factors in Choosing an Outsourcing Provider Percent Provider expertise/capability

86

Flexibility

81

Low cost/price

78

Industry knowledge

75

Ability to earn trust

74

Reputation

69

Culture fit

55

Creativity

51

Outsourcing team members

50

Provider's global reach

39

Prior relationship

36

Knowledge of company and key executives Offshore capabilities

34 30

Source: Data are from Accenture (2004).

The New Theories of Offshoring: Trade and Contracting The difficulty of writing and enforcing contracts has led to a new generation of theories about offshoring that focuses on contractual incompleteness. The core idea is that parties to a contract cannot specify all possible future contingencies, particularly when an American firm is operating in a foreign environment with which it is not entirely familiar. For concreteness, suppose that an Indian service provider is required to make an up-front investment in customizing software for a U.S. buyer’s human resource (HR) needs. Also suppose that there is only a single outcome of interest, namely the “quality” of the software. I make the extreme assumption that a court cannot judge quality or observe anything that might be informative of quality. The contract is incomplete in the sense that the court cannot properly enforce it. In addition, the contract may not be enforceable because the Indian court is corrupt or lacks the expert judges needed to properly adjudicate the dispute. As a result, after the customization investment is made, there is a bilateral hold-up problem. The buyer would like to offer a lower price for the software than initially agreed to, perhaps arguing that the customization is incomplete.

Daniel Trefler

51

Figure 3. Ex Post Renegotiation Leads to Inefficient Underinvestment in Customization Buyers (continuum) Service provider (finite number)

Customization (relationship-specific investment)

Of course, the Indian service provider is no fool. He fully anticipates that the buyer will renegotiate and so takes steps to protect himself. In particular, the Indian service provider will underinvest in customization. Figure 3 illustrates this point. There is a continuum of buyers spread out on a circle. Each point on the circle represents one buyer’s ideal HR software needs. The number of Indian service providers is finite, three in figure 3. A buyer wants to find a service provider who is a perfect match, but usually will not find one. Instead, the buyer will have to ask the service provider to make a relationship-specific investment in customization. There are several steps in the timeline of this analysis: —The U.S. buyer enters India in search of a local service provider. —The buyer and service provider match. —The buyer chooses an organizational form. That is, the buyer decides whether to offshore outsource or to vertically integrate using FDI to buy the service provider’s firm. —The service provider chooses a level of relationship-specific investment in customization. —The buyer renegotiates. The question is, should the buyer use offshore outsourcing or FDI as the mode of securing customized HR software? The answer depends on the outside options of the service provider. If the service provider can turn around and find

52


another buyer whose HR software needs are similar to the original buyer’s needs, then the service provider can walk away from the old relationship and start up a new one at little cost. In this case of good outside options, the service provider is not overly concerned with hold-up problems and so makes most of the necessary customization investments. This means that the buyer does not have to provide incentives to the service provider to make up-front investments. Logic dictates that in this scenario the buyer should offshore outsource. In contrast, if the service provider’s outside options are poor, he will be concerned about hold-up, will not make the customization investments, and will provide low-quality service. The buyer will then have to use FDI if the buyer wants to control up-front investments in customization. Thus, the decision to offshore outsource or use FDI depends on the degree of hold-up, which in turn depends on (a) the outside options available to the service provider and (b) the quality of contract-enforcement institutions such as the legal system and government rentseeking behavior. A key issue is the question of precisely how FDI provides the right incentives for the service provider to invest in customization. The earliest forms of these models were based on what is called the transactions cost theory of the firm (see Coase 1937; Williamson 1975, 1985; Klein, Crawford, and Alchian 1978; and Grossman and Helpman 2002, 2003). A problem with this approach is that it assumes that vertical integration (FDI) magically eliminates hold-up problems within the firm. But how does this happen? After all, service providers within the firm still have incentives to underinvest by shirking. To address this concern, Grossman and Hart (1986) and others developed the property rights theory of the firm. In this theory, the focus is on how the service provider’s incentives are altered by allowing or not allowing the service provider control over the buyer’s core asset. In particular, control of the relationship-specific asset is given to the party whose effort most influences profits. If the buyer’s input into developing customized HR software is crucial, then it should be done using FDI. If the buyer can scope the project with precise specifications, then what is needed most is to provide high-powered incentives to the service provider. This is done by making the service provider the residual claimant on profits—that is, by offshore outsourcing. This insight has been built into models of offshoring by Antràs (2003, 2005), Grossman and Helpman (2005), and Antràs and Helpman (2004). Three related papers that are less about the inability to write complete contracts than about the unwillingness of courts in developing countries to enforce them appear in Ethier and Markusen (1996), Markusen (2001), and Nunn (2005).

Daniel Trefler

53

Empirical Evidence Supporting the New Trade Theories It is useful to review the two papers that combine theoretical insights with empirical support. These are Antràs (2003) and Nunn (2005). In Antràs (2003), both the buyer and the service provider make relationship-specific investments. The buyer invests capital and the provider invests labor. With offshore outsourcing, each party’s outside option in the renegotiation stage is 0, so there is underinvestment by both parties. With FDI, the buyer is allowed to take a fraction of the provider’s output. Thus the buyer’s outside option is and the provider’s outside option is 0. Thus, relative to offshore outsourcing, FDI induces more investment by the buyer and less investment by the provider. Restated, activities done via FDI will be relatively more capital-intensive than offshore outsourced activities. This yields an important empirical prediction. The larger capital’s share of an industry, the more sensitive profits are to the buyer’s capital underinvestment. Hence the property rights approach predicts that FDI will be the dominant organizational form. This is exactly what we see in figure 4 (see the notes to figure 4 for a complete explanation). Nunn (2005) changes the focus slightly. Instead of being interested in the inability to write complete contracts, he is interested in the extent to which a country’s legal system appropriately enforces contracts. In particular, in countries with poor contract enforcement institutions, buyers and service providers will be unwilling to make relationship-specific investments for fear that they will expose themselves in court to hold-up problems. Thus, goods requiring substantial relationship-specific investments will tend to be produced in countries with good contract-enforcement institutions. Figure 5 provides Nunn’s evidence on this mechanism (see the notes to figure 5 for a complete explanation). We tend to think that offshoring is driven almost exclusively by the search for low-cost labor. This is simply not true. Eighty-five percent of U.S. service offshoring is with other OECD countries. Many firms enter into service offshoring relationships in order to gain access to a skilled workforce, to be in a rapidly growing market, or simply to be closer to customers. For many firms, the problem of offshoring to low-cost countries is the contracting environment. These countries do not have the legal institutions that allow firms to write complete and enforceable contracts. As a result, opportunistic behavior by local entrepreneurs, bureaucrats, and politicians leads to hold-up problems, underinvestment in the relationship, and ultimately an unsatisfactory offshoring experience. China in the 1990s was massively subject to these hold-up problems. Clissold (2004) provides a vivid description of just how terrifying the weaknesses of China’s legal system were to foreign investors. Things are improving, but only slowly.

54


Figure 4. The Share of U.S. Imports Controlled by MNEs Rises with the Capital Intensity a Share of U.S. imports controlled by MNEs 1

y = –6.86 + 1.17x (1.02) (0.24) R2 = 0.54 0 0 Capital-labor ratio by industry Source: Data are from Antràs (2003). a. Each point in the plot is an industry. The data plot an industry’s capital-labor ratio against the share of U.S. imports for that industry that are imported by MNEs from their affiliates. The more capital-intensive the industry, the larger the share of U.S. imports involving MNEs. Note that this is manufacturing trade rather than services trade and that offshore outsourcing is any arm’s-length transaction.

Offshoring and Dynamic Comparative Advantage The best way to understand the economic aspects of the spectacular ascendancy of East and South Asia is that it has not been driven by high-tech innovation. Rather, it has been driven by kaizen, which means “improvement” or “idiotproofness” in Japanese and is translated into English as “total quality control.” The reason that Asian economies have stormed on to the scene one by one is that quality or reliability competition is discontinuous. Once a firm meets or surpasses the quality of its lead competitors, it grabs huge market share. What has been happening in East and South Asia has been a steady process of incremental innovation. This is Rosenberg’s (1982) unsung hero of modern economic growth. As Rosenberg has argued persuasively, incremental innovations lead year in and year out to the modest but steady productivity gains underlying modern economic growth. Thus, to understand offshoring one must understand

Daniel Trefler

55

Figure 5. Contract Enforcement and Comparative Advantqage a Quality of institutions .8

(beta coef = .38, t statistic = 4.81)

.4

0 0

.5 Relationship-specific investments in exports

1.1

Exports by good-institution countries/exports by bad-institution countries 2.7 (beta coef = .40, t statistic = 6.37)

0

–2.7 –.1

.5 Importance of relationship-specific investments

1.1

Source: Data are fom Nunn (2005). a. In the top panel each point is an industry. Countries with strong institutions (as measured by the rule of law) tend to export goods that require large relationship-specific investments. In the bottom panel each point is a pair of countries. Relative to countries with a weak rule of law, countries with strong rule of law have exports that are skewed toward goods requiring large relationship-specific investments.

56


Figure 6. Incentives: What Are Firms Trying to Achieve through Outsourcing? Incentive for continual performance improvement

53

Service provider to make greater investments up front

44

Create environment of pooled commitment

Groundbreaking endeavor

41

35

Source: Data are from Accenture (2004).

incremental growth, not pathbreaking innovations dominated by the Western countries that invented invention. For firms thinking about offshore outsourcing, the single most important incentive issue is how to encourage service providers to continually improve their performance (see figure 6). Improving performance is much more important to firms than “groundbreaking endeavor.” It is even more important than the up-front investments that we focused on in the previous section. The problem that most firms face is that what is being offshore outsourced is a small component of a larger system. This creates a tension. On the one hand, a buyer would like a service provider to contribute ever-improving component services. On the other hand, ironing out incompatibilities between interdependent components can be a drain on the buyer’s energies. The buyer can conserve its energies by tightly controlling the improvement process, but this may inadvertently stifle the service provider’s incentive to innovate. Puga and Trefler (2002) explore this tension using the novel concept of the imperfect substitutability of innovative effort. Imperfect substitutability is a measure of the costs imposed on one party (the buyer or service provider) by the innovative efforts of the other party (the service provider or buyer). To illustrate, consider the key component of a television, namely, the cathode ray tube (CRT). A CRT is basically an electron gun aimed at the phosphorcoated front screen of a vacuum tube. Rising consumer preference for flatter screens has created a tension between electron gun manufacturers such as

Daniel Trefler

57

Figure 7. The Cathode Ray Tube (CRT): A Complex System

Source: Puga and Trefler (2002).

Sony and vacuum tube manufacturers such as Asahi Glass. From the perspective of Asahi Glass, domes are better than flat surfaces at withstanding the implosion forces of the vacuum tube. Asahi would thus prefer the solution illustrated in figure 7. The CRT screen is flat from the viewer’s perspective but domed from the electron gun’s perspective. Sony would prefer a flat screen from the perspective of both the viewer and the gun because the variable thickness of the glass creates a prism effect that reduces the sharpness of the picture. This distortion can only be remedied by modifying the electron gun. Asahi’s solution imposes costs on Sony, while Sony’s solution imposes costs on Asahi. In our terminology, the innovative efforts of Asahi and Sony are imperfectly substitutable. Sony must decide in advance the conditions under which it will accept Asahi’s solution. The broader these conditions are, the more likely Asahi’s solution is to be adopted and the more resources Asahi will funnel into the project. That is, delegation of control over knowledge is an incentive device. What makes this view of how innovation is organized so useful is its implication for long-term growth. As multinationals from the developed world use more Chinese service providers, good matches, in the sense that relatively little customization is needed, are more likely. In these cases, the MNEs will delegate

58


control over knowledge to their Chinese service providers. This will give the Chinese incentives to do more incremental innovation, which in turn will make them more knowledgeable. In the next period, therefore, these Chinese service providers will have a greater ability to serve MNEs. Compare two markets, one in which relatively few MNEs enter (such as Indonesia or market 1) and one in which many MNEs enter (such as China or market 2). On average, the service providers in China will require less customization in order to meet the needs of MNEs. This will encourage MNEs in China to delegate control over knowledge creation. This will create more knowledgeable service providers in the next period, which in turn will make them even more attractive to MNEs in the period after that. This will lead to even more MNEs arriving in the next period and thus to even less need for customization. In short, the market becomes more and more attractive as a place for offshore outsourcing. Such a process is exactly what took Taiwan from being a country of original equipment manufacturers (OEMs) to being a country of original design manufacturers (ODMs), and it is moving China from being an auto-parts supplier to producing passenger cars for Southeast Asia and engine blocks for the United States. This analysis explains what is currently happening in China and India and offers further insights into how these developments are embedded in an institutional and organizational context.

Policy Challenges By any international yardstick the United States is a rich and successful economy. However, it could do better, and if it does not actively work on doing better it will fall behind. The problem is that offshoring has raised the stakes in the global competition game. The primary effect of offshoring is that it makes it all the more important for the United States to adopt productivity-enhancing domestic policies. What follows is a list of the key policy issues. I start with what the United States should not do. It is perhaps worth focusing on three policies, two of which receive inappropriate attention and one of which may be receiving too little attention. Two Dumb Ideas It is tempting to approach the problem of how to benefit from offshoring as a problem of designing an industrial policy that successfully picks winners.

Daniel Trefler

59

This is a dumb idea. We should not be in the business of subsidizing contact centers, management consultants, financial institutions, or insurance companies. Sure, China does it and Japan did it. But we forget the dumb mistakes that Japan made (see, for example, Saxonhouse 1998). And do we want a Chinesestyle command economy that is great at catching up but unproven at leapfrogging and horrible at allowing individuals the personal freedoms to make economic choices? Another dumb idea is to adopt a protectionist stance. This will help in the short run, but it will provide the wrong long-run incentives for investing in productivity. Without the spur of international competition, U.S. productivity in protected industries will languish, leading to even deeper structural problems. The Destruction of Human Capital The new competition from offshoring will lead to lost jobs and bankruptcies. Each time a worker is separated from her firm, firm-specific human capital is lost. This reduces the incentives of both managers and workers alike to invest in developing firm-specific knowledge. For example, a highly paid IT consultant will typically know much more than just IT. She will know about the unique needs of her firm. Offshoring leads to more frequent separations between workers and firms, thus destroying important dimensions of American human capital. There is solid evidence to support concerns about the destruction of human capital. Wasmer (2002) demonstrates that the major differences between European and U.S. labor markets stem from differences in the specificity of human capital investments. Martin and Moldoveanu (2003) offer substantial evidence of the rising importance of human capital for firm value. For example, in 2000 Cisco Systems employees earned between $5 and $8 billion in option profits alone at a time when the company only made $4.6 billion. It is unclear whether the loss of knowledge that arises from worker-firm separations is an equity concern alone (because it hurts offshore-displaced workers) or whether it is also an efficiency concern (because it destroys valuable human capital). It becomes an efficiency issue if there are incomplete contracts governing worker-firm relationships. Specifically, after relationship-specific training has occurred, workers cannot credibly commit to staying with the firm. This contractual incompleteness leads to an underinvestment in training relative to some unattainable first-best contract. The main point is that offshoring may be exacerbating this inefficiency by leading to more frequent separations. Ensuring that firm-relevant human capital continues to be created in the United States is always a key issue. Whether offshoring creates an environment

60


in which government intervention (new policies to promote human capital formation) is appropriate is an open question that demands to be researched. The policy issues that flow from this are simply not well understood. There is a tension between promoting long-term relationships and promoting flexibility. Flexibility describes how easy it is both for workers and for firms to terminate a relationship and find an alternative one. How do we design incentives for greater on-the-job training and formal job training programs in an environment where offshoring is likely to reduce the length and value of worker-firm relationships? How do we help workers carry accumulated skills across firms? Should corporate and personal taxes reflect our need to promote both greater specific investments as well as greater flexibility? Clearly, more research is needed in this area.

Policy Conclusions Most of the sensible policies aimed at fostering American competitiveness in the service offshoring market are investment-promoting framework policies. They encourage U.S. workers, firms, and governments to invest in building productive assets such as human capital and new technologies. Such framework policies address a whole host of domestic competitiveness issues and so are not unique to issues raised by service offshoring. Nevertheless, it would be a mistake to think that this makes framework policies less central to issues raised by service offshoring. Offshoring creates only a few new policy issues. First, it forces U.S. firms to be part of a global market and hence to compete globally. It thus makes framework policies that encourage investment and competitiveness all the more important. Second, it creates more churning among firms and workers, thus destroying human capital that is specific to worker-firm matches. We must think of policies that encourage these investments without at the same time creating the kinds of labor market inflexibilities that are the source of Euro-sclerosis. Third, it is central both politically and morally to find ways of helping workers displaced by service offshoring. In Trefler (2004), I offer one approach—investing in children at risk so they grow up with skills that allow them to escape the pressures of foreign competition. This and other redistributive policies are clearly affordable for the richest country on the planet.


Dani Rodrik: Dan Trefler’s paper is full of valuable nuggets, but the core of the paper rests on two assertions: (1) by the very logic of comparative advantage, it is impossible for China, India, and other newcomers to take over everything that the United States and other advanced countries are currently producing; and (2) the weakness of institutions in these newcomers will necessarily retard the rate at which they converge with technology levels in the West. The first of these points is unassailable in its logic and is hardly controversial (at least in a roomful of economists). The second seems also on target and serves as a useful reminder that China and India remain by and large very poor countries with lots of work still ahead of them, despite their prowess in certain tradable activities. But putting the problem this way somehow minimizes the ability of these countries (and their emulators) to compete head-on with the United States in global markets. I would like to suggest a somewhat different angle on this question. I have been doing some work recently (with my colleague Ricardo Hausmann) that attempts to measure the “quality” of the export baskets of different countries. We basically quantify the income level that is associated with each country’s exports, which we call EXPY. Without going into too much detail, we do this first by identifying the average income level of countries that are the main exporters of any six-digit level product. This gives us the income level associated with each traded product. Then we calculate EXPY as the weighted average of these values for each individual country. Figure 8 shows how EXPY stacks up against per capita GDP. As expected (and almost by construction), rich countries tend to export goods that other rich countries export. But upon closer look, two things are important in this figure. First, the range of EXPY is a lot narrower than the range of per capita GDP. Country export baskets exhibit considerably greater similarity 61

62


Figure 8. Relationship between Per Capita GDP and Income Level Associated with Exports Log EXPY

10 United States

China

9.5 India 9

8.5

8

6.5

7

7.5

8

8.5

9

9.5

10

10.5

Log GPD per capita Source: Author’s calculations.

than the underlying aggregate productivity of individual countries. Second, some of the key countries associated with outsourcing/offshoring have EXPY levels that are very high; in fact, several times higher than their GDP per capita. Consider, for example, the following three countries: the United States, China, and India. Their respective values (for 2003) are shown below. Overall productivity in the United States is about 7.5 times higher than that in China, and about 13 times higher than that in India. Trefler is right that these

United States China India a. Income associated with each country’s exports.

GDP per capita (US$)

EXPY (US$) a

35,484 4,726 2,732

15,977 13,575 10,701

Daniel Trefler

63

huge gaps will not close until institutional quality in China and India come to resemble that of the United States, which is unlikely to happen in our lifetime. But now look at the second column of numbers (EXPY), which shows the productivity level associated with each country’s exports. Here the differences are actually tiny in comparison with the previous gaps. The income level associated with U.S. exports exceeds that of China’s exports by only 18 percent, and that of India’s exports by about 50 percent. Furthermore, India’s software exports are not included in this comparison, since the EXPY are calculated for commodity exports only. The important lesson is that the nature of the competition that the United States faces (and will face in the future) is determined by the productivity not of the average foreign producer, but of the very best among them. What is special about international trade today is that the very best producers in these poor but huge economies are very good indeed. To the extent that one worries about such things (and it is not at all clear that one should), there is less reason to be complacent than Trefler would lead us to believe. My other disagreements with Trefler, to the extent that there are any, also relate to differences in emphasis. For example, I think he underestimates the role played by industrial policy in most of the success cases, and he downplays the need to think about intelligent industrial policies. As I have argued elsewhere, the trick in successful industrial policy is not to “pick the winners” (an impossible task if there ever were one), but to know how to “let the losers go” (a much less demanding standard).1 I also would have liked to see greater exploration of the circumstances under which the reduction in the value of job-specific human capital (due to outsourcing and offshoring) has an efficiency (as opposed to a purely distributional) consequence. After all, the key question in this debate is whether competition from China and India undermines technological dynamism in the United States. A plausible channel, to an economist at least, would be through the reduction in the incentive to invest in specific human capital. Trefler talks about this possibility but leaves us guessing as to how seriously we should take it.

1. Dani Rodrik, “Industrial Policy for the Twenty-First Century,” Harvard University, 2004 (http://ksghome.harvard.edu/~drodrik/UNIDOSep.pdf).

64


Pol Antràs: In his insightful paper Daniel Trefler makes three basic points. First, the “offshoring debate” should focus on the offshoring of services because the enormous recent increase in the offshoring of manufactured goods poses no significant new challenges to conventional trade policy. Second, service offshoring is distinct because it affects high-skilled workers. This is important because unlike low-skilled workers, the skill of high-skilled workers is endogenous, and thus service offshoring may well affect U.S. workers’ incentives to acquire human capital. Trefler’s third main point is that although service offshoring poses threats to certain types of white-collar workers, the United States is not likely to lose comparative advantage in high-tech goods to China or India any time soon. This is due to the weak nature of institutions governing economic transactions in these countries. This last point can be paraphrased using international trade jargon: high-tech goods are “contract intensive,” China and India are “institutions scarce” relative to the United States, and the process of “institution accumulation” is a very slow one. Let me next discuss each of these three points in more detail. I do not disagree with the claim that offshoring of manufacturing poses no new challenges to conventional trade policy. Still, I believe that the remarkable growth in the offshoring of manufacturing calls for a change in focus in trade policy debates. As a result of the increased offshoring of manufacturing processes, the role of intermediate inputs in the volume of international trade flows has grown in importance. Although the theoretical literature on the determinants of trade protection has not ignored the fact that trade taxes and subsidies fall not only on final goods but also on intermediate inputs (see, for instance, Grossman and Helpman 1994), these considerations have not been properly implemented in the empirical literature and are often not sufficiently stressed in policy discussions. Beyond its effect on trade policy, the increasing international fragmentation of manufacturing production also has significant implications for the optimal design of fiscal policy and monetary policy. To understand this, it is important to remember that a significant fraction of international offshoring is actually conducted within the boundaries of multinational firms. These firms thus have the ability to use transfer pricing (that is, the pricing of goods and services transacted within the firm) to shift profits across locations in order to minimize their tax burden. The public finance literature has studied this phenomenon, but more work is needed in this important area. A much less emphasized point is that transfer pricing may also have important consequences for the optimal design of monetary policy. In particular, transfer pricing may allow multinational firms to cope better with exchange rate movements (see Lawrence and Rangan 1999 for

Daniel Trefler

65

some suggestive evidence). This asymmetric pass-through between internal and external vertical relationships may influence the effectiveness of exchange rate devaluations, and through this channel it may affect the way monetary policy ought to be conducted in countries that host a large number of affiliates of foreign-based multinational firms.1 Having stressed that much more work is needed in the study of the policy implications of manufacturing offshoring, let me now turn back to service offshoring. This brings me to Trefler’s second main point, namely that service offshoring poses new challenges because it affects relatively skilled white-collar workers. He emphasizes the dynamic costs associated with the offshoring of services jobs, which stem from the reduced incentives to acquire skills that are specific to certain occupations that may eventually be offshored to lower-wage countries. This is certainly a reason for concern, but I would also stress the importance of static costs related to the loss of occupation-specific human capital. Every time a U.S. white-collar worker is displaced and his job is offshored, a significant fraction of the knowledge he acquired on the job gets lost. This destruction of occupation-specific human capital may have important aggregate effects. A challenge for future research is trying to quantify these negative productivity effects of service offshoring. A source of inspiration should be the recent work of Kambourov and Manovskii (2004), who have found that the increase in occupational mobility in the United States during the 1980s generated productivity effects able to account for up to 90 percent of the increase in wage inequality recorded over that period. Service offshoring is different from manufacturing offshoring not only because it involves different types of agents. It is also distinct because it involves the offshoring of relatively knowledge-intensive tasks. Traditional theoretical frameworks for understanding offshoring do not explicitly consider the transmission of knowledge inherent in the international offshoring of services. In Antràs, Garicano, and Rossi-Hansberg (2006) we provide a model of offshoring that places knowledge flows and communication technologies at center stage. We show that characteristics of offshoring and its consequences for “northern” workers are critically affected by the state of information and communication technologies (ICTs). For instance, in the model, an improvement in ICTs is associated with a larger amount of offshoring in the world economy, but also with larger increases in “northern” within-worker inequality resulting from the delocation of certain stages of production. Furthermore, and given that in equilibrium 1. I am grateful to Marc Melitz for suggesting this link between transfer pricing and monetary policy.

66


the industrialized North is a net exporter of knowledge services, we also show that if knowledge flows are not appropriately recorded in official statics, the trade balance of the North will tend to appear in deficit. We view this misrecording as a potential contributor to explaining recent U.S. trade imbalances. Finally, let me move to Trefler’s third main point. His argument rests on the premises that (a) innovation or high-tech sectors are “contract intensive” and that (b) China and India will not become relatively “institution abundant” for years to come. My knowledge of institutional economics is rather limited, so I do not have much to add to the latter point. This brings me to the question: what makes a production process contract-dependent? There is a burgeoning literature analyzing the effects of institutions on comparative advantage, from both a theoretical and an empirical perspective (see Levchenko 2004; Costinot 2005; Nunn 2005; and Antràs 2005). A caveat of these pioneering studies is that they fail to convincingly map the theoretical concept of contract dependence to some observable variables in the data. In Acemoglu, Antràs, and Helpman (2005) we model explicitly the frictions that emerge in multi-agent production in the absence of fully enforceable contracts and show that contract-dependence can naturally be related to task complementarities in production. A salient result of our analysis is that countries with good institutions will gain comparative advantage in production processes that feature high complementarities. In light of our results, Trefler’s first premise can be empirically validated by testing whether indeed innovation is a process that involves high levels of complementarity between the agents engaged in it. I conclude by noting another implication of service offshoring on which Trefler does not comment. In particular, service offshoring leads to a shrinkage of what in international macroeconomics is commonly referred to as the “nontradable sector.” Surely this effect is qualitatively identical to that created by manufacturing offshoring. But quantitatively, this further reduction of the nontradable sector may have important implications for several issues in international macroeconomics. In an influential paper, Obstfeld and Rogoff (2000) showed that six major puzzles in international macroeconomics could be jointly rationalized by appealing to the nontradable nature of certain economic transactions. If this were indeed the common explanation for all these puzzles, we would expect these puzzles to gradually vanish with the advent of service offshoring. Future research should establish whether this is indeed what we observe in the data. General Discussion: The formal presentations in this session launched a lively discussion. Alan Deardorff questioned the extent to which the types of services

Daniel Trefler

67

being offshored really do rely on firm-specific human capital that would be subject to the kinds of hold-up problems Daniel Trefler emphasized in his paper. Giving an example to the contrary, his impression was that workers who could speak English and had strong, up-to-date programming skills could move easily from job to job. He also noted that he did not believe that firm-specificity was likely to help explain what he views as one of the main puzzles raised by offshoring—that products that require skilled labor are being offshored to locations where that skilled labor is relatively scarce. Referring to results from his recent surveys of Indian firms, Rafiq Dossani pointed out that workers who undergo long and extensive training programs are not necessarily more valuable to the firm. For instance, a programmer with four years experience in C++ immediately earns more than someone who has gone through a year and a half of in-house training. Thus he argued that it was important to consider value to the firm when examining the role of firm-specific human capital. Thea Lee raised three sets of issues. First, she expressed the view that previous speakers had been overly dismissive of the potential role for trade policy responses to problems associated with offshoring. She stressed that many current trade rules were not designed to address trade in services. For example, the subsidies rules and countervailing duties remedy do not apply to services. And workers laid off from service sector jobs are not eligible for Trade Adjustment Assistance. If these tools are legitimate for goods, should they not also be legitimate for services? Second, she believed that Trefler’s analysis downplayed the threats posed by India and especially China. She saw his posing the scenario of China exporting everything and the United States exporting nothing as an unhelpful caricature. The question she sees as more relevant is whether China, through currency manipulation and industrial policies, might gain an edge in the products that the United States would wish to expand its exports of. Third, Lee focused on the distributional implications of Trefler’s analysis. She asked if there was evidence that the group of winners was narrowing to very highly skilled workers whose skills are nontransferable, as well as to owners of capital. She was particularly struck by the large share of the gains from trade that appeared to be going to intermediaries. She wondered whether middlemen were more prominent in services than in goods trade, and what this meant for the extent to which consumers gained through lower prices. Trefler responded that there have been large gains for consumers, and that this was reflected in lower import prices and improving U.S. terms of trade. As an example, he cited the PCs and other consumer electronics now commonplace in homes and offices. These have been standardized abroad and have become

68


substantially cheaper. Catherine Mann cautioned, however, that a lack of data on prices for trade in services makes it very difficult to assess what has happened to the terms of trade. Dossani also saw industrial policy in a much more positive light than Trefler’s portrayal in his presentation. Dossani pointed out that less-developed economies such as China can find successful models to follow, making industrial policy more promising and less risky than it would be for countries on the leading edge. Trefler agreed that there may be the potential for successful industrial targeting in developing countries attempting to catch up but reiterated his view that this approach was not appropriate for countries at the technological forefront, such as the United States. Claire Brown cautioned against making the assumption that the relevant skilled labor markets clear. Instead, drawing from interview results of firms in the semiconductor industry, she argued that they are in fact highly rationed. For example, her work suggested there were two distinct groups of engineers: Asian engineers who have been educated and trained in the United States and who could work in either the United States or Asia, and American engineers who do not have the option of repatriating to Asia. Although we talk about the rising skill and educational levels in countries such as India and China, they have a very small cadre of people who can run companies and manage projects. Thus, she noted, the distributional issues for these two groups are very different; and these characteristics imply that the labor markets may be far from functioning efficiently. Theodore Moran emphasized that offshoring of services is, in many key respects, very similar to the offshoring of manufactured products that occurred in previous decades. In particular, both affect white- as well as blue-collar workers, and both involve internationalization of production within multinational corporations. In his view, the two are part of the same phenomenon. Susan Collins noted that the initial title of the project and conference had been services offshoring. This was changed to offshoring of white-collar work to reflect the fact that manufacturing has also been important in explaining white-collar employment trends. Robert Feenstra raised some issues of interpretation about Dani Rodrik’s graphs, which he found very interesting. First, he assumed that the actual income levels reflected averages for each country as a whole. In China, income in the coastal provinces, which do most of the exporting, have been considerably higher than the country average, and thus closer to the constructed trade income level. Second, he wondered whether the observed catch-up of actual income to constructed trade income was really due to continued innovation in China’s most dynamic sectors and provinces, or simply to the spread of these gains to the rest

Daniel Trefler

69

of the country, causing income in the hinterlands to catch up with income in the coastal areas. Finally, Collins asked whether the increased fractionalization of production and trade might have implications for the institutional constraints stressed in Trefler’s paper. She raised the possibility that the ability to siphon off particular pieces of a production process could provide new opportunities for countries with less well developed institutional structures to expand and that this dynamic could limit the informativeness of historical data relating institutional development and composition of production and trade.

70


References Accenture. 2004. “Driving High-Performance Outsourcing: Best Practices from the Masters, Executive Survey Results.” Technical Report 2004. Acemoglu, Daron, Pol Antràs, and Elhanan Helpman. 2005. “Contracts and the Division of Labor.” Harvard University. Antràs, Pol. 2003. “Firms, Contracts, and Trade Structure.” Quarterly Journal of Economics 118 (November): 1375–1418. ———. 2005. “Incomplete Contracts and the Product Cycle.” American Economic Review (September): 1054–73. Antràs, Pol, and Elhanan Helpman. 2004. “Global Sourcing.” Journal of Political Economy 112 (3): 552–80. Antràs, Pol, Luis Garicano, and Esteban Rossi-Hansberg. 2006 (forthcoming). “Offshoring in a Knowledge Economy.” Quarterly Journal of Economics 121 (1). Arora, Ashish, and Alfonso Gambardella. 2005. “The Globalization of the Software Industry: Perspectives and Opportunities for Developed and Developing Countries.” Innovation Policy and the Economy 5: 1–32. Audretsch, David B., and Maryann P. Feldman. 1996. “R&D Spillovers and the Geography of Innovation and Production.” American Economic Review 86 (June): 630–40. Baicker, Katherine, and M. Marit Rehavi. 2004. “Policy Watch: Trade Adjustment Assistance.” Journal of Economic Perspectives 18 (Spring): 239–55. Bardhan, Ashok D., and Cynthia Kroll. 2003. “The New Wave of Outsourcing.” Working Paper 1103. University of California, Berkeley, Fisher Center for Real Estate and Urban Economics. Bhagwati, Jagdish, Arvind Panagariya, and T. N. Srinivasan. 2004. “The Muddles over Outsourcing.” Journal of Economic Perspectives 18 (Fall): 93–114. Clissold, Tim. 2004. Mr. China. London: Constable and Robinson. Coase, Ronald. 1937. “Nature of the Firm.” Economica 4 (November): 386–405. Costinot, Arnaud. 2005. “Contract Enforcement, Division of Labor, and the Pattern of Trade.” Princeton University. Ethier, Wilfred J., and James R. Markusen. 1996. “Multinational Firms, Technology Diffusion and Trade.” Journal of International Economics 41 (August): 1–28. Feenstra, Robert C., and Gordon H. Hanson. 1996. “Globalization, Outsourcing, and Wage Inequality.” American Economic Review Papers and Proceedings 86 (May): 240–45. ———. 1999. “The Impact of Outsourcing and High-Technology Capital on Wages: Estimates for the United States, 1979–1990.” Quarterly Journal of Economics 114 (August): 907–40. Gomory, Ralph E., and William J. Baumol. 2000. Global Trade and Conflicting National Interests. MIT Press.

Daniel Trefler

71

Graham, Frank D. 1923. “Some Aspects of Protection Further Considered.” Quarterly Journal of Economics 37 (February): 199–227. Grossman, Gene M., and Elhanan Helpman. 1994. “Protection for Sale.” American Economic Review 84 (4): 833–50. ———. 2002. “Integration versus Outsourcing in Industry Equilibrium.” Quarterly Journal of Economics 117 (1): 85–120. ———. 2003. “Outsourcing versus FDI in Industry Equilibrium.” Journal of the European Economic Association 1 (2–3): 317–27. ———. 2005. “Outsourcing in a Global Economy.” Review of Economic Studies 72 (1): 135–59. Grossman, Sanford J., and Oliver D. Hart. 1986. “Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration.” Journal of Political Economy 94 (4): 691–719. Gwartney, James, and Robert Lawson. 2003. Economic Freedom of the World: 2003 Annual Report. Vancouver: Fraser Institute. Heckman, James, and Pedro Carneiro. 2003. “Human Capital Policy.” Working Paper 9495. Cambridge, Mass.: National Bureau of Economic Research (February). Helpman, Elhanan. 2004. The Mystery of Economic Growth. Harvard University Press. Hicks, John R. 1953. “An Inaugural Lecture.” Oxford Economic Papers 5 (2): 117–35. Jaffe, Adam B., Manuel Trajtenberg, and Rebecca Henderson. 1993. “Geographic Localization of Knowledge Spillovers as Evidenced by Patent Citations.” Quarterly Journal of Economics 108 (3): 577–98. Johnson, George E., and Frank P. Stafford. 1993. “International Competition and Real Wages.” American Economic Review 83 (2): 127–30. Kambourov, Gueorgui, and Iourii Manovskii. 2004. “Occupational Mobility and Wage Inequality.” University of Pennsylvania. Kaufmann, Daniel, Aart Kraay, and Pablo Zoido-Lobaton. 1999. “Governance Matters.” Policy Research Working Paper 2196. Washington: World Bank. Klein, Benjamin, Robert G. Crawford, and Armen A. Alchian. 1978. “Vertical Integration, Appropriable Rents, and the Competitive Contracting Process.” Journal of Law and Economics 21 (2): 297–326. Lawrence, Robert Z., and Subramanian Rangan. 1999. A Prism on Globalization: Corporate Responses to the Dollar. Brookings. Levchenko, Andrei. 2004. “Institutional Quality and International Trade.” IMF Working Paper WP/04/231. Washington: International Monetary Fund. Markusen, James R. 2001. “Contracts, Intellectual Property Rights, and Multinational Investment in Developing Countries.” Journal of International Economics 53 (1): 189–204. Martin, Roger, and Mihnea C. Moldoveanu 2003. “Capital versus Talent: The Battle That’s Reshaping Business.” Harvard Business Review (July).

72


Mokyr, Joel. 1990. The Lever of Riches: Technological Creativity and Economic Progress. New York: Oxford University Press. Nunn, Nathan. 2005. “Relationship-Specificity, Incomplete Contracts, and the Pattern of Trade.” University of Toronto. Nunn, Nathan, and Daniel Trefler. 2005. “The Political Economy of Tariffs and Growth.” University of Toronto. Obstfeld, Maurice, and Kenneth Rogoff. 2000. “The Six Major Puzzles in International Macroeconomics: Is There a Common Cause?” In NBER Macroeconomics Annual 2000, edited by Ben Bernanke and Kenneth Rogoff. MIT Press. Porter, Michael E. 1998. “The Competitive Advantage of Nations.” In On Competition, edited by Michael E. Porter, pp. 155–95. Harvard Business School Press. Puga, Diego, and Daniel Trefler 2002. “Knowledge Creation and Control in Organizations.” Working Paper 9121. Cambridge, Mass.: National Bureau of Economic Research (August). Rodriguez, Francisco, and Dani Rodrik. 2001. “Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-National Evidence.” In NBER Macroeconomics Annual 2000, edited by Ben S. Bernanke and Kenneth Rogoff, pp. 261–325. MIT Press. Rosenberg, Nathan. 1982. “Technological Interdependence in the American Economy.” In Inside the Black Box: Technology and Economics, edited by Nathan Rosenberg, pp. 55–80. Cambridge University Press. Rosenberg, Nathan, and L. E. Birdzell. 1986. How the West Grew Rich: The Economic Transformation of the Industrial World. New York: Basic Books. Rothschild, Michael. 2003. “What Makes American Public Universities Great?” NBER Reporter (Winter). Samuelson, Paul A. 2004. “Where Ricardo and Mill Rebut and Confirm Arguments of Mainstream Economists Supporting Globalization.” Journal of Economic Perspectives 18 (3): 135–46. Saxonhouse, Gary R. 1998. “Structural Change and Japanese Economic History: Will the 21st Century Be Different?” American Economic Review Papers and Proceedings 88 (2): 408–11. Slaughter, Matthew J. 2004. “Insourcing Jobs: Making the Global Economy Work for America.” Working Paper. Tuck School of Business, Dartmouth College. Trefler, Daniel. 1998. “Immigrants and Natives in General Equilibrium Trade Models.” In The Immigration Debate: Studies on the Economic, Demographic, and Fiscal Effects of Immigration, edited by James P. Smith and Barry Edmonston, pp. 206–38.Washington: National Academy Press. ———. 2004. “Looking Backwards: How Childhood Experiences Impact a Nation’s Wealth.” University of Toronto. UNCTAD. 2004. World Investment Report. Geneva: United Nations.

Daniel Trefler

73

Wasmer, Etienne. 2002. “Interpreting Europe and U.S. Labor Market Differences: The Specificity of Human Capital Investments.” Discussion Paper 549. Bonn, Germany: IZA (August). Williamson, Oliver E. 1975. Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press. ———. 1985. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. New York: Free Press. Young, Alwyn 1992. “A Tale of Two Cities: Factor Accumulation and Technical Change in Hong Kong and Singapore.” In NBER Macroeconomics Annual, edited by Oliver Jean Blanchard and Stanley Fischer, pp. 13–54. MIT Press. ———. 1994. “Lessons from the East Asian NICs: A Contrarian View.” European Economic Review 38 (3–4): 964–73.

This page intentionally left blank

J. BRADFORD JENSEN Institute for International Economics LORI G. KLETZER University of California–Santa Cruz and Institute for International Economics

Tradable Services: Understanding the Scope and Impact of Services Offshoring

G

lobalization, particularly globalized production, is evolving and broadening from manufacturing into services. Services activities now account for a larger share of global trade than in the past. Services trade has almost doubled over the past decade: in the period 1992 to 2002, exports increased from $163 billion to $279 billion, and imports increased from $102 billion to $205 billion. These changes, and their implications for American firms and workers, have attracted widespread attention. Coincident with the broadening of global economic integration from manufacturing to services, the face of job displacement in the United States is changing. While manufacturing workers have historically accounted for more than half of displaced workers, over the period 2001–03, nonmanufacturing workers accounted for 70 percent of displaced workers.1 The share of job loss accounted for by workers displaced from information, financial services, and professional and business services nearly tripled, from 15 percent during the 1979–82 recession to 43 percent over the 2001–03 period. The industrial and occupational shift We appreciate the comments and suggestions of our Brookings Trade Forum discussants, Jared Bernstein and Robert Feenstra, as well as those of Andrew Bernard, Catherine Mann, Michael Mussa, Dave Richardson, Peter Schott, and seminar participants at the Institute for International Economics; the University of California, Santa Cruz; and the 2004 Empirical Investigations in International Trade conference. We gratefully acknowledge the support of the Alfred P. Sloan Foundation. 1. The shift in job loss from manufacturing and production workers toward service and whitecollar (nonproduction) workers has been in evidence since the recession of the early 1990s. At that time, concerns about downsizing and reengineering were coincident with a rise in the share of white-collar and service sector job loss. See Podgursky (1992); Farber (1993); Gardner (1993); and Kletzer (1995, 1998).

75

76


in job loss has been associated with a rise in the probability of job loss for moreeducated workers.2 Bringing these two trends together, the changing mix of industries exposed to international trade in services may have deep implications for the structure of U.S. industry and labor markets in the future. Currently, there is little clear understanding of the role of services globalization in domestic employment change and job loss. More fundamentally, there is little clear understanding of the size and extent of services offshoring, how large it is likely to become in the near-term future, or what impact it is having on the U.S. economy. Fueled by the 2004 presidential race and continued slack in the labor market, the services offshoring debate became headline material. The literature on services offshoring is expanding rapidly. A nonexhaustive list of recent contributors includes: Amiti and Wei (2004); Arora and Gambardella (2004); Bardhan and Kroll (2003): Bhagwati, Panagariya, and Srinivasan (2004); Brainard and Litan (2004); Bronfenbrenner and Luce (2004); Dossani and Kenney (2003, 2004); Kirkegaard (2004); Mann (2003); Samuelson (2004); and Schultze (2004). Despite the attention, relatively little is known about how many jobs may be at risk of relocation or how much job loss is associated with the business decisions to offshore and outsource. There are a few prominent projections, advanced mostly by consulting firms. The dominant and most widely quoted projection of future job losses due to movement of jobs offshore is Forrester Research’s estimate of 3.3 million.3 Others include: Deloitte Research’s estimate that by 2008 the world’s largest financial service companies will have relocated up to 2 million jobs to low-cost countries offshore; Gartner Research’s prediction that by the end of 2004 10 percent of IT jobs at U.S. IT companies and 5 percent of IT jobs at non-IT companies will have moved offshore; and Goldman Sachs’s estimate that 300,000 to 400,000 services jobs have moved offshore in the past three years, and that 15,000 to 30,000 jobs a month, in manufacturing and services combined, will be subject to offshoring in the future.4 It is clear that changes in technology are enabling more activities to be traded internationally. What is unclear is how large these trends are likely to become, 2. It is still the case that less-educated workers have the highest rates of job loss overall. Over the 2001–03 period, the rate of job loss for workers with a high school diploma or less was .141; for workers with at least some college experience, the rate of job loss was .096 (estimates from the 2004 Displaced Worker Survey). See Farber (2005) for a more detailed examination of worker characteristics and the risk of job loss. 3. See McCarthy (2002). The Forrester projection was updated in 2004 to 3.4 million. 4. See, in order, Gentle (2003); Gartner Research (2004); and Tilton (2003).

J. Bradford Jensen and Lori G. Kletzer

77

the sectors and occupations affected to date and going forward, and the impact on workers of the resulting dislocations. Without understanding the nature and scope of the changes, it is difficult to formulate effective public policy to address emerging needs. This paper develops a new empirical approach to identifying, at a detailed level, service activities that are potentially exposed to international trade. We use the geographic concentration of service activities within the United States to identify which service activities are traded domestically. We classify activities that are traded domestically as potentially tradable internationally. Using the identified industries and occupations, we develop estimates of the number of workers who are in tradable activities for all sectors of the economy. We compare the demographic characteristics of workers in tradable and nontradable activities and employment growth in traded and nontraded service activities. We also examine the risk of job loss and other employment outcomes for workers in tradable activities. To preview the results, we find considerable employment shares in tradable service industries and occupations. Based on our estimates, there are more workers in tradable professional and business service industries than in tradable manufacturing industries. We also examine the characteristics of workers in tradable and nontradable activities and find that workers in tradable sectors have higher skills and significantly higher wages. Within specific sectors like professional services, the earnings differentials are even larger, approaching 20 percent. When we examine employment growth trends across traded and nontraded activities, tradable activities have lower growth rates, due primarily to employment losses in manufacturing. Within services, tradable and nontradable activities have similar growth rates except at the lowest end of the skill distribution. Low-skill tradable industries and occupations have negative average employment growth, whereas employment growth in nontraded, low-skill services is positive (though low). We also examine worker displacement rates in tradable and nontradable service activities. We see some evidence that displacement rates are higher from tradable service industries than from nontradable. We also find higher displacement rates from tradable white-collar occupations than from nontradable. Consistent with the characteristics of employed workers, we find that workers displaced from tradable service activities are more educated, with higher earnings, than workers displaced from nontradable activities. Job loss from tradable and nontradable service activities is costly to workers in terms of earnings losses (comparing new job earnings to old job earnings). Taken together, the results are consistent with the view that economic activity within the United

78


States is moving toward a U.S. comparative advantage in services, similar to manufacturing. In the next section we describe our empirical approach to identifying tradable activities. The following sections describe the tradable and nontradable categories for both manufacturing and services activities; compare worker characteristics in tradable and nontradable services; explore the employment trends in tradable and nontradable services; and consider the most recent evidence on job displacement from tradable activities.

Empirical Approach Historically, services have been considered nontradable, with a paucity of empirical work examining trade in services relative to empirical work on manufacturing. To examine the potential impact of trade in services on the U.S. economy, we wanted to identify the size and scope of services trade at as detailed a level as possible. As many observers and researchers have noted, gathering detailed data on the extent of services offshoring is quite difficult. While the Bureau of Economic Analysis (BEA) provides data on international trade in services, the data on international trade in services that BEA publishes do not provide particularly detailed industry-level data. Table 1 shows the level of industry detail available from BEA. Our interest in examining trade in services in more detail than what is available through the BEA services trade data necessitated an alternative empirical approach to identifying tradable service activities. Our approach to identifying service activities that are potentially tradable is novel: we use the geographic concentration of service activities in the United States to identify industries and occupations that appear to be traded domestically. From this domestic information, we infer that service activities that can be traded within the United States are also potentially tradable internationally. Framework The economic intuition we rely on to develop our baseline measure of tradable services is that nontraded services will not exhibit geographic concentration in production. We observe that goods that are traded tend to be geographically concentrated (to capitalize on increasing returns to scale, access to inputs such as natural resources, etc.), while goods that are not traded tend to be more ubiquitously distributed. We apply this same intuition to service production.


79

Helpman and Krugman (1985) present a model that demonstrates this intuition. They model a world with two goods, two countries, and three industries, where the first industry is assumed to be a nontradable constant-returns sector, the second industry is an industry with differentiated varieties that are assumed to be costlessly traded, and the third industry is a tradable constant-returns sector. Helpman and Krugman derive the input vectors V(1), V(2), and V(3) for the integrated world equilibrium. With homothetic and identical tastes, if country j has a share s j of world income, it must allocate resources s j V(1) to the nontradable industry; that is, the production of the nontraded good must be allocated between countries in proportion to their shares of world income. Nontraded goods are distributed uniformly according to population and income. This intuition is revealed more descriptively by Paul Krugman, who notes, “In the late twentieth century the great bulk of our labor force makes services rather than goods. Many of these services are nontradable and simply follow the geographical distribution of the goods-producing population—fast-food outlets, daycare providers, divorce lawyers surely have locational Ginis pretty close to zero. Some services, however, especially in the financial sector, can be traded. Hartford is an insurance city; Chicago the center of futures trading; Los Angeles the entertainment capital; and so on. . . . The most spectacular examples of localization in today’s world are, in fact, services rather than manufacturing. . . . Transportation of goods has not gotten much cheaper in the past eighty years. . . . But the ability to transmit information has grown spectacularly, with telecommunications, computers, fiber optics, etc.”5 The idea is that when something is traded the production of the activity is concentrated in a particular region to take advantage of some economies in production. As a result, not all regions will support local production of the good, and some regions will devote a disproportionate share of productive activity to a good and then trade it.6 We use the geographic concentration of service activity within the United States as an indicator that the service is traded within the United States and thus potentially tradable internationally. The “locational Gini” referred to by Krugman is one of several ways to measure geographic concentration.7 The measures compare a region’s share of 5. Krugman (1991, p. 65). 6. The relationship between geographic concentration of production and trade, particularly exports, has a long tradition in both economic geography (where the measure used is the location quotient) and trade analysis (where the measure used is revealed comparative advantage). The measures of economic concentration used in this paper are different from the location quotient and revealed comparative advantage measures, but all the measures have a similar flavor in that they compare the share of production (or exports) in a particular region to an “expected” baseline. 7. Among the different empirical approaches to measuring geographic concentration and agglomeration are Duranton and Overman (2004).

80


Table 1. Private Services Trade by Type, 2002 Millions of dollars Trade type

Exports, 2002

Imports, 2002

Travel Overseas Canada Mexico Passenger fares Other transportation Freight Port services Royalties and license fees Affiliated U.S. parents’ transactions U.S. affiliates’ transactions Unaffiliated Industrial processes Other Other private services Affiliated services U.S. parents’ transactions U.S. affiliates’ transactions Unaffiliated services Education Financial services Insurance services Telecommunications Business, professional, and technical services Accounting, auditing, and bookkeeping services Advertising

66,547 54,772 6,268 5,507 17,046 29,166 12,330 16,836 44,142 32,218 29,066 3,152 11,924 3,900 8,024 122,594 43,500 25,194 18,306 79,094 12,759 15,859 2,839 4,137 28,799 360 633

58,044 44,494 6,489 7,061 19,969 38,527 25,973 12,554 19,258 15,132 2,958 12,174 4,126 1,935 2,192 69,436 32,367 17,529 14,838 37,069 2,466 3,665 15,348 4,180 10,732 716 1,360 (continued)

employment in or output of an activity with the region’s share of overall economic activity. We make use of two common measures of geographic concentration; but before turning to those measures we address one more conceptual issue. Demand-Induced Agglomeration and Intermediate Services Measures of geographic concentration are a way to implement the intuition described above. Most measures of concentration use the region’s share of employment in an industry relative to the region’s share of total employment. The measures of concentration do not differentiate the reasons activity is concentrated. It does not matter whether production is concentrated because of the


81

Table 1. Private Services Trade by Type, 1992–2002 (Continued) Millions of dollars Trade type Agricultural, mining, and on-site processing services Agricultural and mining services Waste treatment and depollution services Architectural, engineering, and other technical services Computer and data processing services Construction, architectural, engineering, and mining services Construction Data base and other information services Industrial engineering Installation, maintenance, and repair of equipment Legal services Management, consulting, and public relations services Medical services Miscellaneous disbursements Operational leasing Research, development, and testing services Sports and performing arts Trade-related services Training services Other business, professional, and technical services Other unaffiliated services

Exports, 2002

Imports, 2002

366 346 20 1,916 3,004

273 259 14 312 1,057

n.a. 654 2,426 749 4,992 3,270 1,696 1,901 623 3,573 1,086 175 353 501 430 14,700

n.a. 226 236 185 812 768 1,188 n.a. 1,522 190 1,040 110 95 361 283 679

Source: Bureau of Economic Analysis. n.a. = not available.

location of natural resources, increasing returns in production, or spillovers due to the agglomeration of workers; the concentration of production indicates that the good or service is produced in a location different from where it is consumed. So, in general, the reason for the concentration does not matter to us, except in one instance. If a service is nontradable and demand for the service is concentrated (that is, if industries that use the nontraded service are geographically concentrated), the service industry will be geographically concentrated and we would incorrectly infer that the service is tradable. To incorporate this case into our approach, we extend the intuition from the framework. If a nontradable industry provides intermediate inputs to a downstream industry, we would expect the geographic distribution of the nontraded intermediate industry to follow the distribution of the downstream industry. Instead of being distributed with income, the nontraded good is distributed in proportion to the geographic distribution of demand for that industry.

82


We construct region-specific measures of demand for each industry using the 1999 input-output use tables produced by the Bureau of Economic Analysis.8 This measure of industry demand share (IDSi,p ) represents how much geographic concentration there is in demand for a good or service i in a particular region p. We construct the demand for industry i in Place of Work Metro Area p by: IDSi,p = j (Yi,j /Yi * InEMPj,p /InEMPj ),

(1)

where Yi,j = the output of industry i used by industry j (including government and private households as “industries”); Yi = total output of industry i; InEMPj,p = industry j employment in region p; InEMPj = total employment in industry j. We include both direct use and investment in the “use” of industry i output by industry j. To construct the region-specific measures of demand for each occupation, we use the industry-region-specific demand measures described above and weight those by the share of occupation employment in an industry. ODSo,p = j (IDSj,p * OcEMPo, j /OcEMPo),

(2)

where IDSj,p = industry demand share for industry j in region p; OcEMPo,j = occupation o employment in industry j; and OcEMPo = total employment in occupation o. These adjustments take account of the concentration of downstream industry concentration and adjust the “denominator” in the geographic concentration measures that follow. Measuring Geographic Concentration The first measure of economic concentration, as described in Ellison and Glaeser (1997), is: ECi = p (si,p – xp)2.

(3)

8. For more information, see www.bea.doc.gov/bea/dn2/i-o.htm. We aggregate some BEA input-output (IO) industries to a level consistent with the industry classification used by the Census Bureau on the 2000 Decennial PUMS (Public Use Micro Sample).


83

The measure is an index for comparing a region’s share of industry employment (si,p) with the area’s share of aggregate activity/employment (xp). When an area’s employment share in an activity is significantly greater than the area’s share of aggregate employment, this is interpreted as indicating a concentration, or specialization, in the given activity. The index EC provides a national index for each industry, and measures of EC indicating geographic concentration are interpreted as indicative of trade in that activity, in the sense that “local” employment exceeds “local” demand in some areas and the difference is traded outside the area. We modify the EC measure to look at the difference between the region’s share of industry employment and the region’s share of industry demand, as noted above: ECi = p (si,p – IDSi,p)2.

(4)

The new measure of EC is an index for comparing a region’s share of an industry’s employment (si ) with the region’s share of demand for that industry (IDSi,p). We do not make the Herfindahl adjustment that Ellison and Glaeser (1999) use in their index of agglomeration because we are not interested in agglomeration (the co-location of different firms in the same industry), but are interested in pure geographic concentration (whether the concentration is due to one firm or a number of firms). If economic activity is concentrated because significant scale economies are captured within a firm, we do not want to discount this concentration. The second measure of geographic concentration we use is the Gini coefficient. The Gini coefficient (G) for the concentration of industry activity is given by: Gi = | 1 – p (Yi,p – 1 + Yp) * ( Xi,p – 1 – Xp ) | ,

(5)

where p’s index regions (sorted by the region’s share of industry employment), Yi,p is the cumulative share of industry i employment in region p, Yi,p – 1 is the cumulative share of industry i employment in the region (p – 1) with the next lowest share of industry employment, Xp is the cumulative share of total employment in region p, and Xp – 1 is the cumulative share of total employment in region p – 1. We modify the Gini measure to: Gi = | 1 – p (Yi,p – 1 + Yi,p) * (IDSi,p – 1 – IDSi,p) | , where IDSi,p is the region’s share of demand for industry i.

(6)

84


Implementation We implement these measures using employment information from the 2000 Decennial Census of Population Public Use Micro Sample (PUMS) files. We use as our geographic entity the Consolidated Metropolitan Statistical Area or the Metropolitan Statistical Area where an individual reports working.9 We construct the measures of geographic concentration for each industry. Industries that are geographically concentrated are considered tradable. We recognize that the use of worker-level data to investigate economic concentration is somewhat unusual. We pursue this strategy because we are interested in both industrial concentration and occupational concentration. The ability to identify both industries and occupations that are tradable is an important feature of the empirical strategy because many of the service activities that are reportedly being globally sourced are tasks within the service “production” process (for example, a bank’s customer service/call center component may be moved offshore, but not the banking relationship); occupations correspond more closely to these types of activities than industries do. We construct the adjusted G and EC measures for both industries and occupations. The correlation between the EC measure and the G measure is quite high, .713 for industries and .732 for occupations. For the remainder of this paper, we focus on the G results.

Classifying Industries and Occupations as Tradable or Nontradable An important task in our empirical approach is to identify the level of geographic concentration that indicates that an industry or occupation is “tradable.”10 We started exploring where to impose the tradable/nontradable threshold with industries because we have a much better sense of which industries are

9. For regions, we use the Place of Work Consolidated Metropolitan Area (POWCMA5) field on the Decennial PUMS. When POWCMA is coded as a nonmetropolitan area or a mixed metro/nonmetro area, we concatenate the Place of Work state code with the POWCMA5 code. For more information on the 5 percent sample PUMS, see www.census.gov/Press-Release/www/ 2003/PUMS5.html. 10. While choosing the threshold for nontradable and tradable is inherently arbitrary, we ran a number of robustness checks on the results reported in the paper. With the exception of the share of employment in the tradable sector (which decreases as the threshold rises), the results are robust to the choice of threshold.


85

Figure 1. Geographic Concentration of Industries 0.9

0.8

0.7

Gini Coefficient

0.6

0.5

0.4

0.3

0.2

0.1

0 0

100

200

300

400

500

600

700

Retail

Transportation

800

900

1000

NAICS Industry Ag

Mining

Utilities

Construction

Manufacturing

Wholesale

Services

Public Admin

tradable, particularly goods-producing industries. We initially placed industries into three roughly equal groups: Gini class 1 (least geographically concentrated) when the industry Gini was less than .1; Gini class 2 when the industry Gini was between .1 and .3; Gini class 3 (most geographically concentrated) when the Gini coefficient was greater than or equal to .3. Approximately 36 percent of industries are in Gini class 1, about 37 percent are in Gini class 2, and 27 percent are in Gini class 3. Figure 1 plots the Gini coefficients for all industries by two-digit NAICS code. The pattern exhibited in figure 1 is generally consistent with our priors that tradable industries will be geographically concentrated. For example, industries in the goods-producing sectors of Agriculture, Mining, and Manufacturing are typically in the top two Gini classes. Only five of the ninety-two industries in these sectors are in Gini class 1: Cement and Concrete; Machine Shops; Miscellaneous Manufacturing n.e.c.; Structural Metals and Tanks; and Printing and Related Activities. All of these industries seem to be either nontraded because of a high weightto-value ratio (such as Cement and Concrete), or they are categories that include a range of potentially dissimilar activities (Miscellaneous Manufacturing n.e.c.) that make them appear to be broadly geographically distributed. Most agriculture,

86


mining, and manufacturing products are considered tradable; so as a first-order approximation, classifying the lowest geographic concentration category (Gini class 1) as nontradable seems appropriate for these sectors.11 Using a Gini coefficient of .1 as the threshold for tradable seems to make sense in other sectors as well. Industries in the retail trade sector are primarily classified as nontradable. Industries in the Transportation sector are mostly classified as tradable. In Public Administration, most activities are nontradable; Public Finance and the military are exceptions. In the Service sector, industries are balanced between nontradable and tradable. Table 2 provides a complete list of service industries by 2-digit NAICS sector and the industry’s Gini class.12 Table 3 shows the share of employment classified in tradable industries by major NAICS group. Again, the employment shares across categories and industries conform to our priors. All employment in the Agriculture and Mining sectors is classified as tradable (in one of the top two Gini classes). In Manufacturing, most employment is in the tradable sector.13 Utilities are mostly nontradable and Construction is entirely nontraded. For the remainder of the paper, we categorize industries with a Gini coefficient below .1 as nontradable and industries with a Gini coefficient greater than or equal to .1 as tradable. Size and Scope of Tradable Service Industries We use the categorization of industries as tradable and nontradable to develop estimates of the employment potentially affected by trade in services. Table 4 shows the share of total employment in tradable and nontradable industries by major NAICS group. In contrast to traditional characterizations of services as predominantly nontradable, our categorization suggests that a significant share of 11. Another check on the industry classification is to examine the correlation of geographic concentration of manufacturing industries with the level of trade intensity in those industries. The mean industry trade share [(imports + exports)/domestic production] for Gini class 1 = .40, Gini class 2 = .57, Gini class 3 = .71. If Manufacturing Machinery n.e.c. is removed from Gini class 1 (by virtue of its not being a consistent industry), the mean trade share for that class falls to .35. The pattern revealed is one of a positive correlation between Gini class and mean trade share, with some notable variation within class. 12. Higher education may appear to stand out in table 2 as a nontradable service industry. U.S. colleges and universities, particularly research institutions, have an acknowledged global comparative advantage and attract many foreign students. The sector also includes community colleges that are, by design, geographically dispersed. The types of specialized scientific occupations associated with research institutions (the most likely to “export” educational services) are geographically concentrated and thus considered tradable. 13. Alternatively, if we modify the cutoff and use .2 as the break between tradable and nontradable, 28 percent of manufacturing employment would be in the nontradable sector.


87

Table 2. Service Industries, Gini Coefficient Class 2-digit NAICS

Industry description

Gini coefficient class

51 51 51 51 51 51 51 51 51 51 51

Information Newspaper publishers Radio and television broadcasting and cable Libraries and archives Wired telecommunications carriers Data processing services Other telecommunication services Publishing, except newspapers and software Other information services Motion pictures and video industries Sound recording industries Software publishing

1 1 1 2 2 2 2 3 3 3 3

52 52 52 52 52

Finance and insurance Savings institutions, including credit unions Banking and related activities Insurance carriers and related activities Nondepository credit and related activities Securities, commodities, funds, trusts, and other financial investment

1 1 2 2 3

53 53 53 53 53

Real estate and rental and leasing Video tape and disk rental Other consumer goods rental Commercial, industrial, and other intangible assets rental and leasing Real estate Automotive equipment rental and leasing

1 1 2 2 2

54 54 54 54 54 54 54 54 54 54

Professional, scientific, and technical services Veterinary services Accounting, tax preparation, bookkeeping and payroll services Architectural, engineering, and related services Other professional, scientific, and technical services Legal services Specialized design services Computer systems design and related services Advertising and related services Management, scientific, and technical consulting services Scientific research and development services

1 1 2 2 2 2 2 2 2 3

55

Management Management of companies and enterprises

2

56 56 56 56 56

Administrative support Waste management and remediation services Business support services Services to buildings and dwellings Landscaping services Employment services

1 1 1 1 2 (continued)

88


Table 2. Service Industries, Gini Coefficient Class (Continued) 2-digit NAICS



56 56 56

Other administrative and other support services Investigation and security services Travel arrangement and reservation services

2 2 2

61 61 61 61

Education Elementary and secondary schools Colleges and universities, including junior colleges Other schools, instruction, and educational services Business, technical, and trade schools and training

1 1 1 2

62 62 62 62 62 62 62 62 62 62 62 62 62 62 62

Health care and social services Hospitals Nursing care facilities Vocational rehabilitation services Offices of physicians Outpatient care centers Offices of dentists Offices of optometrists Residential care facilities, without nursing Child day care services Home health care services Other health care services Office of chiropractors Individual and family services Community food and housing, and emergency services Offices of other health practitioners

1 1 1 1 1 1 1 1 1 1 1 1 1 2 2

71 71 71 71

Arts, entertainment, and recreation Bowling centers Other amusement, gambling, and recreation industries Museums, art galleries, historical sites, and similar institutions Independent artists, performing arts, spectator sports, and related industries

72 72 72 72

Accommodation Drinking places, alcoholic beverages 1 Restaurants and other food services 1 Recreational vehicle parks and camps, and rooming and boarding houses 1 Traveler accommodation 2 (continued)

1 1 2 2

total employment is in tradable service industries. For example, more workers are in tradable industries in the services sector than in the manufacturing sector. The sum of the share of total employment in industries that are tradable in professional services (NAICS 51–56) is 13.7 percent and larger than the share of employment in tradable manufacturing industries (12.4 percent). There are sizable service sec-


89

Table 2. Service Industries, Gini Coefficient Class (Continued) 2-digit NAICS


81 81 81 81 81 81 81 81 81

Other services Beauty salons Funeral homes, cemeteries, and crematories Personal and household goods repair and maintenance Automotive repair and maintenance Barber shops Religious organizations Commercial and industrial machinery and equipment repair and maintenance Dry cleaning and laundry services Car washes Electronic and precision equipment repair and maintenance Civic, social, advocacy organizations, and grant-making and giving Nail salons and other personal care services Other personal services Business, professional, political, and similar organizations Labor unions Footwear and leather goods repair

92 92 92 92 92 92 92 92 92 92 92 92 92 92 92

Public administration Justice, public order, and safety activities Administration of human resource programs Other general government and support Executive offices and legislative bodies Military Reserves or National Guard Administration of economic programs and space research Administration of environmental quality and housing programs Public finance activities National security and international affairs U.S. Armed Forces, branch not specified U.S. Coast Guard U.S. Air Force U.S. Army U.S. Navy U.S. Marines

81 81 81 81 81 81 81


1 1 1 1 1 1 1 1 1 1 1 2 2 2 3 3 1 1 1 1 1 1 1 2 3 3 3 3 3 3 3

tors correctly characterized as having low shares of employment in tradable industries (education, health care, personal services, and public administration). However, because the service sector is much larger than the manufacturing sector, the number of workers potentially exposed to international trade in services is actually larger than the number of exposed workers in manufacturing.

90


Table 3. Share of Sector Employment by Gini Coefficient by NAICS Sector Percent NAICS sector 11 21 22 23 31 32 33 3M 42 44 45 4M 48 49 51 52 53 54 55 56 61 62 71 72 81 92

Description

Gini class 1

Gini class 2

Gini class 3

Agriculture Mining Utilities Construction Manufacturing Manufacturing Manufacturing Manufacturing Wholesale trade Retail trade Retail trade Retail trade Transportation and warehousing Transportation and warehousing Information Finance and insurance Real estate and rental and leasing Professional, scientific, technical services Management Administrative support Education Health care/social services Arts, entertainment, recreation Accommodation Other services Public administration

0 0 80.89 100.00 0 21.99 14.44 0 45.82 81.72 88.65 100.00 42.81 0 33.25 32.05 9.06 13.95 0 59.53 98.89 97.80 67.35 81.92 79.77 71.68

87.95 24.24 15.31 0 40.39 44.88 65.36 100.00 50.62 18.28 11.35 0 22.03 100.00 50.37 50.98 90.94 79.87 100.00 40.47 1.11 2.20 32.65 18.08 9.86 4.63

12.05 75.76 3.80 0 59.61 33.13 20.21 0 3.57 0 0 0 35.17 0 16.38 16.97 0 6.18 0 0 0 0 0 0 10.37 23.69

60.82

29.75

9.43

All Industries

Occupation Results We are also interested in categorizing occupations as tradable and nontradable. We are interested in identifying tradable occupations because, at least based on anecdotal reports in the press, some intermediate inputs into service production might be tradable even though the service industry is not (think computer programming for the banking industry). We use a similar methodology to classify occupations into tradable and nontradable categories. We construct a demand-weighted Gini coefficient for each occupation as described above and use the same Gini = .1 threshold for the nontradable/tradable categorization. Table 5 shows the share of employment by Major Standard Occu-


91

Table 4. Share of Total Employment in Tradable and Nontradable Industries by NAICS Sector Percent NAICS sector 11 21 22 23 31 32 33 3M 42 44 45 4M 48 49 51 52 53 54 55 56 61 62 71 72 81 92

Description Agriculture Mining Utilities Construction Manufacturing Manufacturing Manufacturing Manufacturing Wholesale trade Retail trade Retail trade Retail trade Transportation and warehousing Transportation and warehousing Information Finance and insurance Real estate and rental and leasing Professional, scientific, technical services Management Administrative support Education Health care/social services Arts, entertainment, recreation Accommodation Other services Public administration All industries

Nontradable

Tradable

0 0 0.76 6.86 0 0.81 1.16 0 1.66 5.90 2.91 0.62 1.32 0 1.04 1.64 0.16 0.82 0 1.99 8.75 10.90 1.12 4.52 3.76 4.14

1.36 0.39 0.18 0 2.17 2.86 6.86 0.53 1.96 1.32 0.37 0 1.76 1.27 2.08 3.47 1.63 5.08 0.06 1.35 0.10 0.25 0.54 1.00 0.95 1.63

60.82

39.18

pational Classification group by Gini class. The groupings are largely consistent with our priors. The occupational groups with large shares of employment classified as tradable include: Business and Financial Operations (68 percent); Computer and Mathematical Occupations (100 percent); Architecture and Engineering (63 percent), Legal (96 percent), and Life, Physical and Social Sciences (83 percent).14 The notable nontradable occupational groups include 14. Van Welsum and Reif (this volume) offer a list of U.S. occupations (at the 3-digit level) identified as “potentially affected by offshoring” in table A-2. As explained in the chapter, their method relies on occupations having “offshorability attributes” that rely on the use of information and communication technologies, highly codifiable knowledge, and no face-to-face contact. There is overlap

92


Table 5. Share of Occupation Employment by Gini Coefficient by Major Occupation Category Percent SOC 2-digit code

Description

Gini class 1

Gini class 2

Gini class 3

11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55

Management Business/financial operations Computer/mathematical Architecture/engineering Life, physical, social sciences Community/social services Legal Education and library Arts, design, entertainment Health care practitioners/technicians Health care support Protective service Food preparation/serving Building maintenance Personal care service Sales and related Office/administrative support Farm, fish, forestry Construction/extraction Installation, maintenance, repair Production Transportation/material moving Military specific

34.48 31.73 0 36.04 16.32 100.00 3.78 99.54 17.13 86.56 96.73 59.83 95.68 98.54 82.64 75.41 93.14 0 61.37 90.00 80.30 89.20 0

61.15 65.96 73.07 58.31 58.61 0 96.22 0.46 75.02 13.10 3.27 40.17 4.32 1.46 7.22 21.82 6.66 81.01 36.18 8.89 17.15 5.86 0

4.37 2.32 26.93 5.65 25.08 0 0 0 7.85 0.34 0 0 0 0 10.13 2.77 0.20 18.99 2.45 1.11 2.55 4.95 100.00

71.66

24.86

3.47

All occupations

Education and Library (99 percent nontradable); Health Care Practitioners (86 percent); Health Care Support (97 percent), Food Preparation (96 percent). On the blue-collar side, 90 percent of employment in Installation, Maintenance, and Repair is classified as nontradable, as is 80 percent of Production and 89 percent of Transportation and Material Moving.15 between the two lists of occupations, although our method identifies a larger set of tradable occupations. Van Welsum and Vickery (2005) offer a list of U.S. industries potentially affected by offshoring, in table 6. Our detailed industry list shares similarities with theirs, but our list excludes a number of retail industries (dairy stores, liquor stores, and others) included in their list. 15. The geographic concentration results are at first counterintuitive for production occupations given the manufacturing industry results. Production occupations are typically not industryspecific but instead are functional activities and are thus distributed more broadly.


93

Table 6. Share of Employment in Tradable and Nontradable Occupations and Industries Percent Occupation category (SOC 2-digit code)

Nontradable occupations

Tradable occupations

Management occupations (11) Non-tradable industries Tradable industries

23.97 10.51

26.58 38.94

Business and financial operations occupations (13) Nontradable industries Tradable industries

14.11 17.61

27.72 40.56

Computer and mathematical occupations (15) Nontradable industries Tradable industries

0 0

24.22 75.78

Architecture and engineering occupations (17) Nontradable industries Tradable industries

8.46 27.59

13.30 50.66

Life, physical, and social science occupations (19) Nontradable industries Tradable industries

7.28 9.03

36.49 47.20

Legal occupations (23) Nontradable industries Tradable industries

3.54 0.24

18.89 77.33

50.03 21.64

10.79 17.54

All occupations Total nontradable industries Total tradable industries

The last two rows of table 6 show for all occupations how many workers are in occupations classified as tradable in industries classified as nontradable. In the aggregate, the share of workers in tradable occupations and nontradable industries is not large, about 10 percent. However, for business and professional occupations, the share of workers in tradable occupations in nontradable industries is much larger. The typical professional occupation has about 25 percent of employment in tradable occupations in nontradable industries. To the extent that firms can vertically “disintegrate” the provision of these intermediate service inputs, workers in these tradable occupations are potentially vulnerable to trade even though their industry is not tradable. This suggests that for service activities the share of workers potentially vulnerable to trade is probably understated. Outside of education and health care occupations, the typical white-collar occupation involves a potentially tradable activity.

94


Table 7. Mean Earnings and Demographic Characteristics for Selected and All Industries Percent, unless otherwise noted Industry (NAICS code) Manufacturing (3x) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Information (51) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Finance and insurance (52) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age

Nontradable

Tradable

36,974 75.1 6.1 9.7 2.6 13.8 85.3 40.0

39,901 67.8 9.7 11.7 6.0 20.4 82.9 40.2

35,472 50.9 10.4 7.8 9.4 37.4 94.2 38.7

49,510 55.9 11.5 7.3 10.6 41.3 96.2 37.6

38,170 29.0 11.5 7.8 7.1 30.5 97.1 38.1

54,460 42.7 9.2 6.4 10.2 43.8 97.4 39.1 (continued)

Worker Characteristics Beyond mere employment counts, we also examine demographic characteristics such as education, age, gender, and earnings to identify whether there are differences between workers in tradable service activities and those in nontradable industries and occupations. These characteristics are available from the 2000 Decennial Census of Population Public Use Micro Sample (PUMS) 5 percent sample.16 Table 7 shows the demographic characteristics of workers in tradable industries and nontradable industries in aggregate. Workers in tradable industries have 16. For more information on the 5 percent sample PUMS see www.census.gov/PressRelease/www/2003/PUMS5.html.


95

Table 7. Mean Earnings and Demographic Characteristics for Selected and All Industries (Continued) Percent, unless otherwise noted Industry (NAICS code) Real estate and rental and leasing (53) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Professional, scientific, technical services (54) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Management (55) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Administrative support (56) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma All industries Employment income Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age

Nontradable

Tradable

23,056 58.1 9.1 10.8 1.9 13.3 84.7 31.1

42,915 51.1 8.6 9.7 6.7 29.7 90.6 42.4

42,246 35.3 5.1 5.0 16.6 52.5 97.1 39.5

57,959 57.1 5.5 5.6 25.7 59.5 97.8 39.3

… … … … … … … …

61,285 45.5 5.4 4.9 14.3 49.7 97.8 40.5

24,039 64.1 11.9 22.2 2.0 10.7 72.3 37.2

28,742 48.5 17.6 12.2 5.0 23.4 88.0 36.1

30,966 49.6 10.2 10.4 10.2 26.6 87.0 38.8

41,836 60.1 9.9 10.3 9.2 30.2 88.7 39.4

96


higher incomes, are more likely to be male, and are more likely to have a college degree (though not an advanced degree). The table also breaks out these same characteristics for selected service industries classified as tradable and nontradable. We present the results for the manufacturing sector as a benchmark for demographic characteristics typically associated with trade-affected workers. Workers in tradable service industries are higher paid and more skilled than workers in tradable manufacturing. Within services, the most striking feature of the service industry results is the difference in annual earnings. Across all major service sector groups, the differential in earnings between tradable and nontradable industries is large, with tradable services having appreciably higher wages. Service workers in tradable industries also tend to have attained a higher level of education and are more likely to be male and white. Table 8 shows the results for all occupations divided into tradable and nontradable groups. Individuals in occupations identified as tradable tend to have higher earnings, are more likely to be male and have more years of schooling. The table also shows the same characteristics for selected occupations. Again, as in the industry results, workers in tradable occupations earn more and are more highly educated than workers in nontradable service occupations. In tables 9–12, we estimate a number of regressions to examine whether the earnings differentials in tradable industries and occupations are the result of higher educational attainment. Table 9 shows regression results for all industries and NAICS 51–56 industries. Across all industries, controlling for observable demographic characteristics and industry (2-digit NAICS) and regional (POWCMA) fixed effects, workers in tradable industries have 6 percent higher wages. For workers in professional and business service industries, the differential associated with being in a tradable industry is even larger. Again controlling for observable demographic characteristics, in the professional service sector, workers in tradable industries have almost 15 percent higher wages than workers in nontradable industries in the same sector. Table 10 shows a similar specification for occupations. The first column reports the results for all occupations, and the second column reports the results for “high-end” service occupations.17 Across all occupations, workers in tradable occupations receive 9 percent higher wages than workers in nontradable occupations. For high-end service occupations, workers in the tradable sector receive almost 13 percent higher wages, even after controlling for demographic characteristics and occupation group (2-digit SOC) and region. 17. High-end service occupations include SOC major groups 11, 13, 15, 17, 19, 23, and 29. See table 8 for the names of the SOC major groups.


97

Table 8. Mean Earnings and Demographic Characteristics for Occupations Percent, unless otherwise noted Industry (NAICS code) Management (11) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Business and financial operations (13) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Computer and mathematical occupations (15) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Architecture and engineering occupations (17) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age Life, Physical, and Social Science Occupations (19) Employment income (dollars) Male Percent African American Percent Hispanic With advanced degree With bachelor’s degree With high school diploma Age

Nontradable

Tradable

51,399 56.2 8.3 6.8 19.9 46.5 95.2 41.8

69,029 67.3 4.7 5.0 15.7 49.6 95.8 42.6

42,813 41.3 10.3 6.9 10.5 44.0 97.6 40.4

51,998 48.0 8.3 5.4 16.2 61.6 98.6 40.2

… … … … … … … …

54,297 70.3 6.8 4.5 17.8 59.9 99.1 37.3

40,505 82.5 5.7 6.4 5.3 26.2 96.2 39.4

62,115 89.0 3.9 4.1 25.5 76.2 99.9 40.6

29,339 57.4 7.0 7.2 11.6 40.0 96.4 36.0

50,000 59.2 4.6 4.0 54.4 85.3 99.2 40.3 (continued)

98


Table 8. Mean Earnings and Demographic Characteristics for Occupations Percent, unless otherwise noted Industry (NAICS code) Legal Occupations (23) Employment income (dollars) Male Percent African American Percent Hispanic With advanced degree With bachelor’s degree With high school diploma Age Healthcare Practitioners and Technical Occupations (29) Employment income (dollars) Male Percent African American Percent Hispanic With advanced degree With bachelor’s degree With high school diploma Age Healthcare Support Occupations (31) Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age All Occupations Employment income (dollars) Male African American Hispanic With advanced degree With bachelor’s degree With high school diploma Age

Nontradable

Tradable

71,304 60.6 9.1 4.5 58.2 78.8 99.2 47.7

80,265 51.4 5.6 5.1 64.1 76.9 99.3 40.9

39,922 19.5 9.8 4.5 17.8 47.3 98.8 40.5

139,375 70.6 4.6 4.8 93.4 97.8 99.7 42.8

18,423 11.9 24.0 10.6 2.2 7.9 83.8 37.8

18,751 17.6 3.7 5.6 9.9 30.9 97.3 39.0

28,789 48.5 11.1 10.9 7.4 21.8 86.3 38.8

51,503 66.7 7.5 8.8 16.1 43.9 91.0 39.9

Table 11 examines whether the effects of being in a tradable industry and occupation are independent. Workers in tradable industries and tradable occupations are the omitted category. For all industries and occupations, workers in nontradable industries and nontradable occupations have 10 percent lower wages than workers in both tradable industries and occupations. Interestingly, the effect seems to be additive. Workers in either only a tradable industry or only


99

Table 9. OLS Regression Results, Tradable Industry Wage Differentialsa All industries Dependent variable: log (employment income) Tradable industry Male African American Hispanic Hours Weeks Advanced degree Bachelor’s degree Industry controls (2-digit NAICS) POWCMAb controls Summary statistics R2 N Weighted N

NAICS 50s

0.060 (0.0008) 0.214 (0.0006) –0.096 (0.0010) –0.215 (0.0010) 0.026 (0.0000) 0.040 (0.0000) 0.262 (0.0011) 0.380 (0.0008) Yes Yes

0.147 (0.0016) 0.225 (0.0014) –0.145 (0.0024) –0.218 (0.0026) 0.029 (0.0001) 0.039 (0.0001) 0.224 (0.0023) 0.325 (0.0017) Yes Yes

0.538 5,836,360 122,155,903

0.519 1,074,271 23,609,616

a. Standard error in parentheses. b. Place of Work Consolidated Metropolitan Area.

a tradable occupation receive wages about 5 percent lower than workers in both a tradable industry and a tradable occupation. In both professional service industries and “high-end” service occupations, the effect of being in a tradable industry and a tradable occupation is quite large. Workers in tradable industries and occupations in NAICS 50 sector receive wages 17 percent higher than workers in a nontradable industry and nontradable occupation within the same sector. For high-end service occupations, the differential is almost as large: workers in tradable industries and occupations make almost 16 percent more than workers in nontradable industries and occupations. These results demonstrate that tradable industries and occupations pay higher wages, even after controlling for observable characteristics. These effects appear to be independent: being in both a tradable industry and a tradable occupation is associated with a larger (almost double) income differential than being in either a tradable industry or occupation alone. The comparison of worker characteristics in tradable service activities suggests that tradable services are consistent with U.S. comparative advantage; they

100


Table 10. OLS Regression Results, Tradable Occupation Wage Differentialsa All occupations Dependent variable: log (employment income) Tradable occupation

0.091 (0.0008) 0.215 (0.0006) –0.061 (0.0010) –0.187 (0.0010) 0.026 (0.0000) 0.039 (0.0000) 0.216 (0.0011) 0.303 (0.0008) Yes Yes

0.127 (0.0014) 0.245 (0.0013) –0.112 (0.0023) –0.168 (0.0027) 0.020 (0.0001) 0.038 (0.0001) 0.227 (0.0016) 0.297 (0.0013) Yes Yes

0.545 5,836,630 122,155,903

0.396 1,446,158 30,803,183

Male African American Hispanic Hours Weeks Advanced degree Bachelor’s degree Occupation controls (2-digit SOC) POWCMAc controls Summary statistics R2 N Weighted N

High-end service occupations b

a. Standard error in parentheses. b. High-end service occupations are occupations in SOC major groups 11, 13, 15, 17, 19, 23, and 29. c. Place of Work Consolidated Metropolitan Area.

are high-skill and high-wage activities (relative to both manufacturing and nontradable service activities).

Changes in Aggregate Employment Growth Much of the recent attention to services offshoring has emphasized job losses in specific occupational categories. We examine recent employment growth trends using both aggregate industry data from the Census Bureau’s County Business Patterns program and aggregate occupation data from the Bureau of Labor Statistics’ Occupational Employment Statistics program.18 We present the 18. The County Business Patterns program is an establishment-based data collection program that uses primarily administrative data and thus has nearly universal coverage of in-scope estab-


101

Table 11. OLS Regression Results, Tradable Industry and Occupation Wage Differentialsa All industries and occupations

NAICS 50s

High-end service occupations b

Dependent variable: Log (employment income) Nontradable industry and nontradable –0.098 –0.174 occupation (0.0011) (0.0026) Nontradable industry and tradable –0.055 –0.072 occupation (0.0012) (0.0026) Tradable industry and nontradable –0.055 –0.045 occupation (0.0010) (0.0022) Tradable industry and tradable occupation —Omitted category— Male 0.205 0.205 (0.0007) (0.0015) African American –0.064 –0.111 (0.0010) (0.0024) Hispanic –0.173 –0.169 (0.0010) (0.0026) Hours 0.025 0.027 (0.0000) (0.0001) Weeks 0.039 0.038 (0.0000) (0.0001) Advanced degree 0.223 0.197 (0.0011) (0.0024) Bachelor’s degree 0.279 0.245 (0.0008) (0.0017) Industry controls (2-digit NAICS) Yes Yes Occupation controls (2-digit SOC) Yes Yes POWCMAc controls Yes Yes Summary statistics R2 N Weighted N

0.545 5,836,630 122,155,903

0.540 1,074,271 23,609,616

–0.159 (0.0022) –0.050 (0.0019) –0.087 (0.0021) 0.244 (0.0013) –0.111 (0.0022) –0.158 (0.0026) 0.020 (0.0001) 0.036 (0.0001) 0.232 (0.0016) 0.276 (0.0013) Yes Yes Yes 0.419 1,446,158 30,803,183

a. Standard error in parentheses. b. High-end service occupations are occupations in SOC major groups 11, 13, 15, 17, 19, 23, and 29. c. Place of Work Consolidated Metropolitan Area.

data broken out as tradable/nontradable and by sector. The results in the previous section indicate that tradable activities in general and tradable services in particular require higher skills than other activities. High-skill activities are consistent with U.S. comparative advantage, and we would expect that as trade

lishments. For more information on County Business Patterns see www.census.gov/epcd/cbp/ view/cbpview.html. The Occupational Employment Statistics program is also an establishmentbased program, but it is collected through a survey instrument. For more information on the Occupational Employment Statistics see www.bls.gov/oes/home.htm.

102


Figure 2. Industry Employment Growth, 1998–2002 0.8

Change in Log(Industry Employment)

0.6

0.4

0.2

0

-0.2

-0.4

-0.6

-0.8 0

100

200

300

400

500

600

700

800

900

NAICS Industry Ag

Mining

Utilities

Construction

Manufacturing

Wholesale

Retail

Transportation

Services

increases, economic activity would shift to activities consistent with U.S. comparative advantage. Thus, we would expect higher-skill industries and occupations to have higher rates of employment growth. We also break out the employment growth rates by industry and occupation skill quartile.19 Figure 2 shows the change in industry employment (log) for the period 1998–2002 by NAICS code.20 Overall, employment in manufacturing industries shrank, and employment in service industries grew. Table 12 presents mean industry employment growth by tradable and nontradable sectors. In the aggregate, the mean tradable industry experienced an employment loss of almost 6 percent, while the mean nontradable industry experienced an employment gain of 5.6 percent. The lower panels of table 12 break out industries by sector, tradable category, and skill quartile. The lower panels of table 12 show that the 19. Industry and occupation skill quartiles are created by placing industries and occupations into skill quartiles based on the share of employees within the industry with a bachelor’s degree. 20. We are constrained to use 1998 as our starting point because it is the first year that County Business Patterns was produced on a NAICS basis; 2002 is the most recent year available. Public Administration is not in scope for the County Business Patterns program, so employment change figures are not available for this sector.


103

Table 12. Industry-Level Employment Change, by Industry Characteristics, 1998–2002 Industry classification Nontradable Tradable Ag, Min, Mfga Services Ag, Min, Mfg

Tradable v. nontradable

Nontradable Tradable Nontradable Tradable Nontradable Tradable

Services

Nontradable

Tradable

Skill quartile

Number of industries

Mean

Standard deviation

Skill Q1 Skill Q2 Skill Q1 Skill Q2 Skill Q3 Skill Q4 Skill Q1 Skill Q2 Skill Q3 Skill Q4 Skill Q1 Skill Q2 Skill Q3 Skill Q4

88 149 5 83 91 85 3 2 32 24 16 11 24 23 20 24 7 16 31 31

0.056 –0.059 –0.116 –0.173 0.067 0.076 –0.067 –0.190 –0.191 –0.203 –0.114 –0.147 0.016 0.084 0.015 0.156 –0.006 0.112 –0.007 0.139

0.114 0.198 0.099 0.161 0.107 0.145 0.102 0.015 0.169 0.148 0.103 0.216 0.080 0.098 0.106 0.088 0.233 0.104 0.095 0.148

a. Agriculture, Mining, Manufacturing.

employment losses are, on average, concentrated in the goods-producing sector (and in the lower portion of the skills distribution).21 In the service sector, the average nontradable industry experienced 6.7 percent growth, and the average tradable service industry experienced 7.6 percent growth. In general, industries in the lower-skill quartiles have a lower rate of employment growth. Tradable industries do not seem to have dramatically different employment outcomes than nontradable industries, though at the low end of the skill distribution tradable industries had, on average, employment losses.22

21. These results are consistent with Bernard, Jensen, and Schott (forthcoming 2006). Bernard, Jensen, and Schott use detailed, plant-level data to examine the impact of imports from low-wage countries on U.S. manufacturing. The results show that activity in U.S. manufacturing is shifting to industries consistent with U.S. comparative advantage. 22. Using a t test to compare the lowest-skill quartile with the highest-skill quartile in the tradable services industry group, we cannot reject the null hypothesis that the means are the same at the 10 percent level.

104


Table 13. Occupation-Level Employment Change, by Occupation Characteristics, 1999–2003 Occupation classification Nontradable Tradable Ag, Prod, Ext, Cona Services Ag, Prod, Ext, Con

Tradable v. nontradable

Nontradable Tradable Nontradable Tradable Nontradable

Tradable

Services

Nontradable

Tradable

Skill quartile

Number of industries

Mean

Standard deviation

Skill Q1 Skill Q2 Skill Q3 Skill Q1 Skill Q2 Skill Q3 Skill Q1 Skill Q2 Skill Q3 Skill Q4 Skill Q1 Skill Q2 Skill Q3 Skill Q4

197 228 38 77 180 180 23 12 3 56 18 3 30 57 54 39 10 32 59 79

0.022 –0.004 –0.044 –0.141 0.036 0.059 –0.070 –0.026 0.056 –0.148 –0.150 0.014 0.005 0.037 0.021 0.078 –0.065 0.086 0.032 0.083

0.160 0.247 0.143 0.228 0.161 0.230 0.145 0.140 0.125 0.235 0.196 0.272 0.114 0.173 0.165 0.164 0.111 0.210 0.181 0.269

a. Agricultural, Production, Extractive, Construction b. Skill Q is Skill Quartile

Table 13 shows similar employment growth rates for 1999–2003 for occupation categories.23 Similar to industries, tradable occupations in aggregate have lower employment growth rates than nontradable industries on average. Also similar to industries, this is explained primarily by differences between production-related occupations and service activities. Tradable service occupations have, on average, higher employment growth rates than nontradable service occupations. It is interesting to note that, as in tradable industries, at the low end of the skill distribution tradable service occupations have negative employment growth. In comparison, the highest skill category has positive employment growth.24

23. We use 1999 as our starting year because it is the first year the Occupational Employment Survey was published on a Standard Occupational Classification basis. We use 2003 as the end point to have a four-year period consistent with the industry data. 24. Using a t test to compare the lowest-skill quartile with the highest-skill quartile in the tradable services occupation group, we can reject the null hypothesis that the means are the same.


105

The employment growth results are consistent with the comparative advantage framework. Employment is shifting toward activities that are consistent with U.S. comparative advantage. Industries and occupations that require higher skills are growing relative to low skill industries and occupations. In both tradable service industries and occupations, those in the lowest skill classes experience negative employment growth on average.

Evidence on the Risk of Job Loss and Characteristics of Displaced Workers The Displaced Worker Surveys (DWS) provide basic information on the scope and cost of involuntary job loss. The DWSs offer large sample sizes, are nationally representative, and allow several key elements to be investigated, including the incidence of job loss; the characteristics of workers affected; likelihood of reemployment; reemployment industry and occupation; and earnings changes.25 These surveys have been used extensively to study manufacturing job loss (see Kletzer 2001). The 2000 census provides the most up-to-date industry and occupational classifications of the services and white-collar jobs of primary interest. The need for updated detail on industry and occupation (currently) limits our use of the Displaced Worker Surveys to the most recent administration, in January 2004. Although we lose the ability to observe services and white-collar job loss over time, we gain the industry and occupational detail necessary for studying services offshoring. Job displacement from services Job loss rates by industry are reported in table 14, focusing on the 2001–03 period covered by the January 2004 Displaced Worker Survey. Remembering that this time period covered the dot.com bust and the most recent recession, the Information sector (NAICS 51) had a notably high rate of job loss (.232). Overall, the risk of job loss was lower in services than in manufacturing. As a reference point, table 14 includes job loss rates by industry for the period 1999–2001, from the 2002 Displaced Worker Survey. The industry classifications are different, reflecting the use of 1990 census codes for the 2002 survey. What is clear is that job loss rates increased from 1999–2001 to 2001–03, most 25. See the appendix for more information on the Displaced Worker Surveys.

Table 14. Job Loss Rates, by Industry Mean From the 2004 Displaced Worker Survey (2001–03) Industry Agriculture Mining Construction Manufacturing Wholesale and retail trade Transport and utilities Information Financial Professional and business services Education and health services Leisure and hospitality Other services Public administration Total Manufacturing, tradable Manufacturing, nontradable Nonmanufacturing, tradable Nonmanufacturing, nontradable

Total 2001–03 0.049 0.127 0.131 0.209 0.113 0.089 0.232 0.081 0.144 0.040 0.105 0.051 0.020 0.103

From the 2002 and 2004 Displaced Worker Surveys

Tradable

Nontradable

… … … … 0.077

… … … … 0.091

0.317 0.08 0.158 0.071 0.083 0.03

0.075 0.081 0.113 0.039 0.113 0.057

0.153

0.076

0.213 0.192 0.128 0.073

Dropping agriculture, mining, and construction Manufacturing, tradable 0.213 Manufacturing, nontradable 0.192 Nonmanufacturing, tradable 0.106 Nonmanufacturing, nontradable 0.054 Total

Industry

1999– 2001

2001– 03

Agriculture Mining Construction Manufacturing, durables Manufacturing, nondurables Transportation Communications Utilities and sanitary service Wholesale trade Retail trade Finance,insurance, and real estate Private household Business and repair services Personal services Entertainment and recreation Hospitals Other medical Educational services Social services Other professional services Forestry and fisheries Public administration

0.042 0.173 0.107 0.177 0.133 0.096 0.159 0.054 0.111 0.099 0.079 0.044 0.181 0.080 0.071 0.026 0.052 0.020 0.033 0.071 0.008 0.017

0.065 0.127 0.131 0.236 0.157 0.103 0.305 0.052 0.123 0.107 0.080 0.016 0.172 0.057 0.098 0.030 0.055 0.030 0.060 0.078 0.070 0.020

0.090

0.106

Total

0.126

0.058

Source: Authors’ calculations from the 2002 and 2004 Displaced Worker Surveys, using sampling weights.


107

notably in Communications (the former name of the sector for some of Information) and Manufacturing. When we apply our tradable/nontradable distinction to the overall economy, the rate of job loss is notably higher in tradable industries (.153) than in nontradable industries (.076). Within the broad sectors of manufacturing and nonmanufacturing, tradable industries also had higher rates of job loss. The tradable/nontradable distinction is small in manufacturing, with tradable industries having a job loss rate of .213, and nontradable (of which there are few) a rate of .192. Outside of manufacturing, the tradable distinction is large. Tradable nonmanufacturing industries have a rate of job loss of .128, and nontradable industries, .073. This difference is most notable in the Information sector, where the rate of job loss from tradable (3-digit) industries was .317 and the nontradable job loss rate was .075. Job loss rates by occupation are reported in table 15. The blue-collar occupations faced a higher rate of job loss (about .12) than the white-collar occupations (about .09). Workers in all occupational categories faced a higher rate of job loss in 2001–03 than in 1999–2001. Production workers faced the highest rate of job loss, .206 (the cross-occupation average was .106). Some of the white-collar occupational categories forecast to be at risk of services offshoring had high job loss rates (but lower than Production workers), including Business Operations Specialists (.143), Computer and Math (.177), and Architecture and Engineering (.128). In the overall economy, tradable occupations had a higher rate of job loss than nontradable occupations, with the greatest difference in white-collar occupations. White-collar workers in tradable occupations faced a job loss rate of .094, and workers in nontradable occupations faced a rate of .065. For blue-collar workers, the tradable job loss rate was .128 and the nontradable rate was .122. There is no clear pattern of exposure to the risk of job loss by tradability within detailed occupations. Parallel to our discussion of worker characteristics from the 2000 PUMS, table 16 reports demographic and educational characteristics for workers displaced from tradable and nontradable nonmanufacturing industries, with (tradable) manufacturing industries offered as a reference group. As noted by Kletzer (2001), workers displaced from nonmanufacturing industries are slightly younger, less tenured, less likely to be male, and considerably more educated than workers displaced from manufacturing. In tradable nonmanufacturing, 75 percent of displaced workers had at least some college experience. In manufacturing, the share of displaced workers with some college was 46 percent.

Table 15. Job Loss Rates, by Occupationa Mean From the 2004 Displaced Worker Survey (2001–03) Industry Management business, financial (white collar) Business operations specialists Financial specialists Professional and related (white collar) Computer and math Architecture and engineering Life, physical, and social science Service (white collar) Sales (white collar) Office and administrative support (white collar) Farming, forestry, fishery (blue collar)

From the 2002 and 2004 Displaced Worker Surveys Industry

1999– 2001

2001– 03

0.091 0.171 0.044

Executive, administrative, managerial Professional specialty Technician and related

0.086 0.059 0.088

0.094 0.066 0.110

0.109 0.177 0.113 0.057 0.072 0.123

0.033 n.a. 0.158 0.066 0.056 0.079

Sales Administrative support Private household Protective services Food, health, cleaning, personal Precision production, craft, repair

0.094 0.097 0.047 0.045 0.069 0.111

0.109 0.106

0.109

0.067

0.092

0.181

0.219

0.110

…

…

Operators, assemblers, inspectors Transportation and material moving equipment

0.103

0.112

Total 2001–03

Tradable

Nontradable

0.089 0.143 0.054

0.077 0.121 0.057

0.070 0.177 0.128 0.059 0.073 0.106

0.059 0.075 0.151

Construction and extractive (blue collar) Installation, maintenance, repair (blue collar) Production (blue collar) Transportation and material moving (blue collar)

0.149

…

…

0.112 0.206

0.117 0.163

0.083 0.169

0.117

0.057

0.096

0.102

0.101

0.078

Blue collar, tradable Blue collar, nontradable White collar, tradable White collar, nontradable

0.128 0.122 0.094 0.065

… … … …

… … … …

Full sample Blue collar, tradable Blue collar, nontradable White collar, tradable White collar, nontradable Full sample total

0.175 0.150 0.104 0.078 …

… … … … 0.122

… … … … 0.087

Total

Source: Authors’ calculations from the 2002 and 2004 Displaced Worker Surveys. a. Agriculture, Mining, and Construction omitted. n.a. Not available.

Handlers, cleaners, helpers Farming, forestry, fishery Armed forces Total

0.139 0.044

0.151 0.067

0.090

0.103

110


Table 16 also shows that in tradable nonmanufacturing industries, displaced workers were more educated, more likely to have health insurance, more likely to lose full-time jobs, and more likely to have higher predisplacement earnings than workers displaced from nontradable industries. The educational attainment differences are stark: 42 percent of workers displaced from nontradable nonmanufacturing industries, but 24 percent of workers displaced from tradable nonmanufacturing industries, had a high school diploma or less. The educational differences show up in predisplacement weekly earnings. Postdisplacement, reemployment rates (also reported in table 16) are higher for displaced nonmanufacturing workers than for manufacturing workers. Reemployment rates are .75 and .77 for nontradable and tradable nonmanufacturing workers, respectively, .64 for manufacturing workers. The earnings cost of job displacement, well established for manufacturing workers, also affected nonmanufacturing workers. For the 2001–03 period, with the weak job recovery from the recession, we see large earnings losses. Median earnings losses are smaller for nonmanufacturing than for manufacturing, and a larger share of nonmanufacturing workers experience no earnings loss. Consistent with lower predisplacement earnings, workers displaced from nontradable nonmanufacturing industries experienced smaller earnings losses than workers displaced from tradable nonmanufacturing industries. Table 17 reports worker characteristics and reemployment outcomes for three services sectors: Information; Financial, Insurance and Real Estate; and Professional and Business Services. For the most part, workers in tradable industries in these sectors have higher levels of educational attainment. In Information and Professional and Business Services, predisplacement weekly earnings were higher in tradable industries than in nontradable industries. Consistent with higher earnings, more workers displaced from tradable industries reported that they had health insurance coverage than workers displaced from nontradable industries. Reemployment outcomes (reemployment rates or average earnings losses) are similar within sector, across the tradability of the detailed industries. Table 18 reports a similar breakdown, by occupation, for sectors: Management, Business and Financial; Professional and Related; Office and Administrative Support. Workers from tradable occupations have higher levels of education, within occupational group, than workers from nontradable occupations. Their predisplacement earnings were higher, as was the availability of health insurance coverage. Men are more highly represented in the tradable occupations. Again, there is no clear pattern of reemployment outcomes by tradability. Earnings losses range from 3 percent to 16 percent, with 40 to 50 percent of reemployed workers reporting no earnings loss.


111

Table 16. Characteristics of Displaced Workers, by Industrial Sector and Tradability Worker characteristics Age (mean in years) Standard deviation Job tenure (mean in years) Standard deviation Job tenure > ten years

Manufacturing, tradable

Nonmanufacturing, Nonmanufacturing, tradable nontradable

41.60 11.20 7.11 8.43 0.23

39.60 11.10 4.40 5.60 0.12

38.10 11.70 4.26 5.61 0.14

0.14 0.40 0.24 0.22 0.61

0.05 0.19 0.30 0.45 0.54

0.11 0.31 0.33 0.25 0.45

0.75 0.96

0.66 0.90

0.47 0.82

342.70 300.54

443.18 383.08

294.91 271.21

Share reemployed Of reemployed, share full-time

0.64 0.80

0.77 0.78

0.75 0.72

All reemployed Change in ln earnings (mean) Standard deviation Median change Share with no loss in earnings

–0.32 0.89 –0.15 0.42

–0.30 0.98 –0.11 0.45

–0.14 1.02 –0.03 0.51

Full-time to full-time Change in ln earnings (mean) Standard deviation Median change Share with no loss in earnings

–0.21 0.76 –0.10 0.42

–0.21 0.69 –0.07 0.46

–0.12 0.97 –0.03 0.52

Educational attainment (share) High school dropout High school graduate Some college College + Male In predisplacement job Share with health insurance Full-time If full-time, real weekly earnings (dollars) Standard deviation (dollars)

Source: Authors’ calculations from the 2004 Displaced Worker Survey, using sampling weights. Agriculture, Mining, and Construction omitted.

Conclusions This paper develops a new empirical approach to identifying, at a detailed level for the entire economy, industries and occupations that are tradable. Using the methodology, we find substantial employment in tradable service industries and occupations. Workers in these industries and occupations are more highly skilled and have higher earnings than workers in the manufacturing sector and nontradable service activities. The higher earnings are not solely a result of higher skill levels: in regressions controlling for observable characteristics,

Table 17. Characteristics of Selected Service Sector Displaced Workers, by Industry and Tradability Information Tradable Job tenure (mean in years) Standard deviation Job tenure > ten years Educational attainment (share) High school dropout High school graduate Some college College + Male In predisplacement job Share with health insurance Full-time If full-time, real weekly earnings (dollars) Standard deviation (dollars) Share reemployed Of reemployed, share full-time All reemployed Change in ln earnings (mean) Standard deviation Median change Share with no loss in earnings Full-time to full-time Change in ln earnings (mean) Standard deviation Median change Share with no loss in earnings

Nontradable

Financial, insurance, real estate Tradable

Nontradable

Professional and business services Tradable

Nontradable

5.80 7.37 0.192

4.51 7.25 0.16

5.82 7.00 0.167

8.28 9.14 0.259

3.55 3.98 0.066

3.24 4.68 0.109

0.032 0.207 0.262 0.499 0.559

0.00 0.038 0.45 0.512 0.668

0.04 0.179 0.389 0.392 0.47

0.046 0.243 0.354 0.357 0.479

0.047 0.157 0.261 0.535 0.527

0.173 0.446 0.196 0.186 0.527

0.82 0.93 530.82 409.45 0.72 0.76

0.62 0.87 387.98 350.69 0.81 0.87

0.62 0.91 409.88 380.43 0.61 0.80

0.73 0.94 542.51 454.14 0.68 0.82

0.66 0.91 504.61 415.82 0.71 0.80

0.36 0.83 273.95 251.57 0.62 0.73

–0.57 1.07 –0.34 0.346

–0.72 2.97 –0.024 0.469

–0.16 1.09 –0.08 0.456

0.013 0.499 0.03 0.531

–0.34 0.96 –0.08 0.457

–0.18 0.93 –0.03 0.468

–0.40 0.82 –0.25 0.36

–1.003 3.328 –0.07 0.344

–0.15 0.51 –0.047 0.457

0.018 0.36 –0.007 0.508

–0.185 0.737 –0.034 0.49

–0.162 0.999 –0.029 0.489

Source: Authors’ calculations from the 2004 Displaced Worker Survey, using sampling weights.

Table 18. Characteristics of Displaced Workers in Selected Service Occupations, by Occupation and Tradability Management, business, and financial Worker characteristics Job tenure (mean in years) Standard deviation Job tenure > ten years Educational attainment (share) High school dropout High school graduate Some college College + Male In pre-displacement job Share with health insurance Full-time If full-time, real weekly earnings (dollars) Standard deviation (dollars) Share reemployed Of reemployed, share full-time All reemployed Change in ln earnings (mean) Standard deviation Median change Share with no loss in earnings Full-time to full-time Change in 1n earnings (mean) Standard deviation Median change Share with no loss in earnings

Tradable

Nontradable

Professional and related Tradable

Nontradable

Office and administrative support Tradable

Nontradable

6.72 8.04 0.204

5.03 4.99 0.143

4.82 6.09 0.111

4.30 5.25 0.109

5.31 6.69 0.176

4.57 5.74 0.136

0.008 0.132 0.269 0.591 0.466

0.012 0.272 0.28 0.436 0.633

0.003 0.092 0.198 0.708 0.717

0.026 0.115 0.328 0.53 0.248

0.051 0.331 0.438 0.18 0.306

0.05 0.339 0.406 0.204 0.241

0.775 0.965 554.78 434.23 0.786 0.791

0.588 0.927 426.02 336.05 0.72 0.726

0.794 0.93 523.24 369.44 0.80 0.805

0.632 0.791 323.60 226.58 0.801 0.707

0.616 0.896 299.45 254.48 0.691 0.758

0.577 0.865 261.96 198.07 0.755 0.763

–0.374 1.08 –0.127 0.492

–0.364 1.144 –0.165 0.389

–0.34 1.155 –0.084 0.455

–0.14 0.811 –0.037 0.507

–0.227 0.677 –0.15 0.443

–0.093 1.063 –0.045 0.512

–0.205 0.852 –0.045 0.528

–0.357 1.165 –0.109 0.351

–0.318 1.176 –0.068 0.462

–0.128 0.343 –0.029 0.515

–0.113 0.455 –0.068 0.471

0.012 0.704 –0.025 0.542

Source: Authors’ calculations from the 2004 Displaced Worker Survey, using sampling weights.

114


workers in selected tradable service activities earn 16–17 percent higher incomes than similar workers in nontradable activities in the same sector. Examining employment growth across industries and occupations, there is little evidence that tradable service industries or occupations grow more slowly than nontradable industries or occupations overall, though at the low end of the skill distribution employment growth is negative for tradable services. Highskill service activities have the highest employment growth rates. There is job insecurity associated with employment in tradable activities, including service activities. We find a higher rate of job loss from tradable industries than from nontradable industries, with the greatest difference outside of manufacturing. In comparison with an overall rate of job loss of .103 for 2001–03, tradable nonmanufacturing industries have a rate of job loss of .128 and nontradable industries .073 (though we note the possibility that these differences are driven by the tech bubble). Also within occupations, workers in tradable jobs faced a higher rate of job loss than workers in nontradable jobs, with the greatest difference within white-collar occupations. These results have several implications. First, it seems inappropriate to consider all service activities as inherently nontradable. The geographic concentration of some service activities within the United States is as great as in manufacturing and is consistent with the view that a number of service industries and occupations are tradable. The share of employment in tradable services is large enough that a better understanding of the forces shaping trade in services warrants our attention. At a minimum, more resources should be devoted to collecting and publishing considerably more detail on international service flows. Continuing to increase the amount of information collected on the use of intermediate service inputs within the United States would also increase our ability to track and understand developments in this large and growing sector. Second, the results presented in this paper suggest that tradable services are consistent with U.S. comparative advantage. While professional and business services jobs require higher skills and pay higher wages than manufacturing jobs in general, tradable services jobs in these sectors require even higher skills and are more highly paid than nontradable service activities. We would expect that as technological and organizational change increases the potential for trade in services, economic activity in the United States will shift to activities consistent with U.S. comparative advantage.26 It is therefore possible that further liberalization in international services trade would directly benefit workers and firms in

26. The United States maintains a positive trade balance in service activities; see table 1.


115

the United States. The policy community should devote more attention to understanding the impediments to services trade. Third, although tradable services have relatively high employment growth rates overall, at the low end of the skill distribution tradable service activities have negative employment growth. The potential for reallocation across activities in response to shifting trade patterns in services is real. Policymakers should prepare for additional reallocation among this group of workers. The process of adjustment to job displacement might be eased by service worker characteristics. For the most part, workers displaced from tradable services are different, in terms of job tenure and educational attainment, from workers displaced from (tradable) manufacturing industries. Generalizing from what we know from studies of manufacturing worker job loss, lower levels of job tenure and higher levels of educational attainment may be advantages in seeking reemployment. Given the current availability of data, it is too early to tell. We need data beyond the time period of the “jobless recovery.” We also need more information to discern whether workers in tradable activities face different reemployment outcomes than workers in nontradable activities. The evidence we do have tells us that job loss for services workers is costly. These costs underscore the need to have a less porous safety net (for example, by extending Trade Adjustment Assistance [TAA] to services workers and extending wage insurance beyond TAA). Lower rates of employment growth at the lower end of the skill distribution in tradable service activities may have implications for the retraining strategies and opportunities for displaced low-skill workers in both manufacturing and services.

Appendix: Displaced Worker Survey The Displaced Worker Survey is administered biennially as a supplement to the Current Population Survey (CPS). The first survey was administered in January 1984 and the most recent in January 2004. In each survey, adults (aged 20 years and older) in the regular monthly CPS were asked if they had lost a job in the preceding three- or five-year period due to “a plant closing, an employer going out of business, a layoff from which he/she was not recalled, or other similar reasons.”27 If the answer was yes, a series of questions followed concerning 27. For the 1984–92 surveys, the recall period was five years. Starting in 1994, the recall period was shortened to three years.

116


the lost job and the period of joblessness. Other causes of job loss, such as quitting and firing, are not considered displacements.28 This categorization is consistent with our common understanding of job displacement: it occurs without personal prejudice in that terminations are related to the operating decisions of the employer and are independent of individual job performance. This operational definition is not without ambiguity: the displacements are “job” displacements, in the sense that an individual displaced from a job and rehired into a different job with the same employer is considered displaced. A key advantage of the DWS is its large-scale representative nature. As part of the CPS, it draws on a random sample of 60,000 households, which is weighted to be representative of the U.S. workforce. As a result, the surveys yield responses from large numbers of displaced workers in a wide set of industries. In exchange for breadth of coverage, the DWSs have two weaknesses relevant to any study of the costs of job loss. The first is the relatively short-term horizon. Individuals are surveyed just once, providing information about one postdisplacement point in time, rather than about their experiences over time. The second weakness is the lack of a readily available comparison group of nondisplaced workers. Without such a comparison group, we cannot investigate what would have happened to these workers if they had not been displaced. The lack of a comparison group leads to some unavoidable errors in measuring outcomes such as postdisplacement reemployment and earnings losses. The rate of job loss reported in the tables is calculated as in Farber (1993, 2003, 2005): it is the ratio of the (weighted) number of reported displacements divided by the (weighted) number of workers who were either employed at the survey date or reported a job loss but were not employed at the survey date. See Kletzer (2001) for more discussion of the issues that arise when using the DWSs to measure the incidence of job loss.

28. Individuals who respond that their job loss was due to the end of a seasonal job or the failure of a self-employed business are also not included.


Jared Bernstein: Jensen and Kletzer have written a refreshingly clear and insightful paper that readers will find to be one of more useful contributions to the often fuzzy literature on offshoring. Much of this work has tried to identify the service or white-collar jobs at risk to offshore competition, but we have been stymied by the difficulty of using trade data on service flows for this purpose. These authors derive a clever method using geographical clustering for doing so, and while they may need to work a bit harder to convince skeptics, many will find their approach convincing, as I do. This innovative classification scheme sets the stage for the paper’s other main contribution: a description of the characteristics and earnings of those in tradable services relative to those in nontradable services. One criticism of the paper is that the title promises more than the authors, or anyone else for that matter, can yet deliver. That is, while they go further than others toward identifying the industries and occupations directly affected by offshoring, to truly capture the “scope and impact” of this growing competitive challenge, researchers need to go beyond the direct effects. The authors do point out that displaced workers in tradable services suffer large wage losses relative to other displaced workers, but (a) it is not clear that this is because they are in tradable services, and (b) surely the impact of offshoring goes beyond this subgroup. This latter point is critical. The implicit supply shock from adding millions of skilled workers to a relatively concentrated set of occupations and industries may have a significant negative impact on the wage structure of white-collar workers, much as the increase of trade in manufacturing goods with low-laborcost competitors has structurally altered the wage distribution of blue-collar 117

118


workers. In short, a white-collar worker needn’t get displaced to feel the impact of this growing phenomenon.

Using Geographic Clustering to Identify Tradable Services The credibility of their paper rests on the authors’ novel method for identifying tradable services. They point out that BEA data on international trade flows in services are not disaggregated enough by industry to serve this purpose. But the problem goes deeper than this. As my EPI colleague Josh Bivens points out, these data, especially the highly relevant parts relating to information technology, are getting a bit hard to believe, given what so many firms are telling us about their service imports and what some other countries’ service export data suggest. Take, for example, data on the value of imports of computer-related services, which includes software writing, from India. Even with recent large upward revisions, the tiny magnitudes of the BEA numbers—for example, $330 million in 2003— are hard to believe. The Indian tech trade group NASSCOM puts this value at $4.7 billion. This is not to suggest that NASSCOM’s data capacity is superior to BEA’s. Rather, if you’re out to identify service jobs affected by offshoring, most analysts are suspicious of the quality of our data on the import of some key services associated with offshoring. At any rate, Jensen and Kletzer use the assumption that tradable firms exhibit geographic concentration. This assumption comes from research on the goods sector, where returns to scale, access to transportation nodes, and proximity to natural resources lead goods producers to congregate near each other. Is it reasonable to extend this to service production? Empirically, we can, without much effort, observe this concentration, or lack thereof. Silicon valleys and “research triangles” have appeared in numerous places over the past decades. Meanwhile, bowling alleys and child-care centers are scattered pretty much all over the place. In this regard, their transporting of this method of identifying tradable industries from goods to services does not seem a big stretch. There are, however, some differences between goods and services that will lead some readers to wonder if scale economies and access issues loom large enough in services to motivate geographic clustering. For example, to transport cars or steel, manufactures have historically needed to locate near waterways. But it is hard to see why this constraint would hold for, to take a very relevant case, transmitting information across the Internet. In fact, it is the sharp decline


119

in such costs that has allegedly motivated service firms to offshore data to extremely distant places. So they may need to work a little harder to convince skeptics. What are the specific benefits they have in mind that motivate tradable services to locate near each other? Are there some case studies they could cite? As mentioned, it is not hard to point to areas where high-tech firms are concentrated, but there could be lots of reasons for that, including niche education and labor markets: California’s Silicon Valley and North Carolina’s Research Triangle, for example, are both near universities with specialties in computer science. And where I live, in northern Virginia, our silicon alley, out Route 66 in the Dulles corridor, likely grew out of the desire to be close to federal government contractors and purchasers. What is the connection to international trade? And why shouldn’t nongeographically clustered service industries offshore some of their jobs? Hospitals, for example, score in the authors’ least geographically concentrated category, presumably because they are pretty pervasive across localities in our economy. But anecdotes suggest that hospitals are beginning to offshore some of their accounting services, certainly a plausible scenario (anecdotes also suggest hospitals are offshoring high-tech functions, like radiology services, but as the conference paper by Frank Levy and Ari Goelman (this volume) finds, this does not appear to be occurring).1 While I encourage them to work a little harder to convince the reader that their classification scheme is up to the task, a close look at their tables and figures reveals strong face validity. There are a few industries, such as hospitals, that seem questionably classified as nontradable (accounting, tax preparation, bookkeeping, and payroll services is another), but no such system will be perfect. In the case of the two examples I just mentioned, they are services that by their nature tend to be demanded in most localities and thus fly under the radar of their test. So perhaps Jensen and Kletzer can think of an added filter that would help address such industries.2 They presumably pick up some of these jobs in their occupational analysis. Their table 8, for example, shows that 11 percent of total employment is in tradable occupations in nontradable industries. Still, the apparent misclassification of a few industries may unsettle some readers.

1. Levy and Goelman show that both gatekeeper actions by U.S. radiologists and malpractice regulations explain why hospitals are hard-pressed to offshore such services. 2. I doubt anyone would squawk if they just added a few industries like hospitals and tax preparation services that are widely reported to be tradable services, even though they are not geographically concentrated.

120


They are careful to avoid the following mistake: suppose a nontradable upstream service provides an intermediate service to a tradable downstream service industry. If the upstream firms need to locate near the downstream firms, they will be misclassified as tradable services. For example, if a computer firm both offshores programming tasks to India and outsources payroll services to a nearby firm, the authors could end up mistakenly labeling the upstream industry as a tradable service. To avoid this, they use input/output tables to parse the upstream services from the downstream ones. A final concern is in regard to the role of productivity growth in their method of using workers to identify where firms are clustered. If demand is constant, falling, or not growing too quickly, as was arguably the case over their period of study, firms with fast-growing productivity might be shedding workers. The impact of this on their analysis is not necessarily problematic, as long as the firms in such industries remain clustered (and it is hard to see why they would not). But this may be one reason why this type of analysis is usually based on more direct measures of industry output (one reason they are sticking with workers is because they want to examine occupations as well as industries).

Comparing the Characteristics of Workers in Tradable and Nontradable Jobs As one might have expected, given the anecdotes in the newspapers, jobs in tradable services pay more than those in nontradable services: a 35 percent annual earnings differential in tradable services, unadjusted for worker differences, and a large adjusted differential, discussed next. Such workers are also more likely to be male and have higher educational attainment. With a set of earnings regressions, the authors find a statistically and economically large premium associated with being in a tradable industry, a tradable occupation, and a combination of the two (in their later analysis on displacement, we see the downside of this—workers displaced from such jobs experience large relative losses). Relative to those in nontradable industries and occupations, the premium amounts to between 10 and 17 percent, depending on the sample. What is interesting here is that the impact of being in such industries and occupations is modeled as a sort of interaction, as the regressions already control for industries and occupations. The coefficient of interest thus tests whether an earnings premium exists above that already accounted for by the underlying industries or occupations that are also included in the tradable services indicator.


121

Such interactions are difficult to interpret. The descriptive statistics reveal that workers in tradable services have characteristics that by themselves are all associated with positive and significant coefficients in such regressions: they are disproportionately male, nonminority, and have higher educational attainment. Combine these characteristics and you get a fairly hefty wage boost beyond that accounted for by any one characteristic alone. Are such workers truly more productive, or are there other factors, such as bargaining power and discrimination, that might explain their premium relative to those who lack this set of characteristics? The result is also curious in relation to the tradable service categorization. One might expect that the wages of such workers face downward pressure from international competition relative to the wages of other workers with similar skill sets in nontradable industries and occupations. At least in these static regressions, that is not the case. It will be interesting to track the premium over time to see if this pressure develops. At any rate, the important point is that service workers exposed to trade competition have a lot to lose. The last section helps to quantify that point. This part of the paper includes two tables on changes in employment levels by industry and occupation. The goal here is to determine the extent to which job losses have occurred in recent years in tradable services, a question that is a bit of a holy grail, given the nervousness regarding the impact of offshoring services. As such, I thought the section got short shrift. This part of the analysis would have benefited from more discussion of the data and trying a little harder to separate out cyclical effects. On the first point, their sources for employment data are the Census Bureau’s County Business Patterns and the BLS Occupational Employment Statistics (OES). Neither of these sources is typically used to track aggregate employment changes, and readers will legitimately wonder whether they reflect the stylized facts of employment trends over the years in question (1998–2003). In fact, given the difference in employment trends between the two surveys that are universally used for such analysis—the BLS Establishment and Household surveys—some will question whether the facts are “stylized” at all. I took a cursory look at the total OES employment counts from 2000 through 2003, which seem to show a large growth of jobs over these years, which is hard to square with data from more reliable sources of aggregate employment growth (such as the Establishment survey). Also, one of the biggest challenges regarding the question of the impact of offshoring on job loss over recent years is separating an offshoring effect from that of the cycle. This is particularly tough given the burst of the IT bubble in late

122


2000 and the resulting spike in layoffs in this sector. In table 12, the authors examine changes from 1998–2002, a period including a strong run-up in employment growth (1998–2000) and a recession (2001) and jobless recovery (2002). At the least, the authors might consider breaking out these two periods to add some accounting for these cyclical effects. Better yet, given the caveats regarding these data sets for this purpose and the difficulty untangling cycle from offshoring effects, they might want to be more cautious about their claims here. For example, claims comparing the employment growth of tradable and nontradable services made it into their abstract and could be widely cited. There is also a claim here regarding employment losses at the lower end of the skill distribution in tradable services, but this change is essentially zero in table 12 and (if I calculated the standard error correctly) statistically insignificant (at the 5 percent level) in table 13.3

Displaced Workers in Tradable Services The final section of the paper uses the Displaced Workers Survey (DWS) to examine the extent to which being in a tradable job raises a worker’s chance of displacement. Because of coding changes on industries and occupations, the authors cannot do comparisons across this biennial survey. But using the most recent survey, covering the years 2001–03, they find that those in tradable services face significantly higher displacement rates than those in nontradable services. For example, 31.7 percent of those in the tradable sectors of information services were laid off (not for cause) over these years, but only 7.5 percent of those in the nontradable sectors. Here again, the concern is that we are catching the cycle and the bursting of the tech bubble in the analysis, and thus not really isolating an offshoring effect. Information services includes both newspaper publishing (a nontradable service) and Internet publishing (a tradable service), and it is surely the case that a postbubble, large negative spike in domestic demand affected the former more than the latter. A simple difference-in-difference estimator might help to difference out the cycle, say using the changes in displacement in services that were nontradable. The problem is the introduction of new industry and occupation codes in the most recent DWS. However, the BLS has a version of the monthly CPS with 3. I divided the standard deviation by the square root of the number of industries, both given in the table (0.111/3.16) for a standard error of 0.035, which returns a t statistic of –1.85.


123

new sectoral codes starting in 2000, and although they could not track displacements, the authors should see if these files might enable them to compare wage and employment changes in tradable and nontradable services controlling for the cycle. The DWS has long showed that among displaced workers who are reemployed at the time of the survey, blue-collar production workers take the biggest hit in wages (the pay gap between their old and new jobs is above the average loss). But Jensen and Kletzer find negative effects of a similar magnitude for displaced workers in tradable services. The difference between the old and new wage was, on average, about –30 percent for workers displaced from tradable jobs in both manufacturing and services, and about –14 percent for those displaced from nontradable services. So workers in tradable services were more likely to be displaced during the recent downturn/jobless recovery, and for those who found new jobs at the time of the survey, these displacements were quite costly relative to nontradable services.

Summary Faced with the question of how we identify service workers directly affected by offshoring, Jensen and Kletzer come up with an elegant solution: borrow the observation from the goods-producing literature that firms engaged in trade exhibit geographic concentration. While some might question how well this assumption travels across these different sectors, their results are, for the most part, intuitively satisfying and believable. This aspect of the paper makes a useful contribution to what has been a major stumbling block in this fledgling literature, namely, identifying affected workers in tradable services. The paper’s other major contribution is its documentation of the characteristics of these workers, including their relative earnings. The paper has two shortcomings, both of which are evident in much work on offshoring. First, barring some attempt to control for cyclical effects, it is hard to know whether the job and wage loss effects they identify for workers in tradable services are due to their exposure to offshoring competition or to the protracted labor-market downturn over this period. While they get some traction in this argument by comparing tradable and nontradable services, the problem is that the negative cyclical demand shock was particularly acute in some of the same industries and occupations that have heavy weights in their tradable service category (like IT).

124


Second, from a policy perspective, economists need to look far beyond those directly affected by offshoring to grasp the magnitude of the challenge it poses. Compared to the number who are and will be affected in some way by the competitive pressures from this form of trade, the number of workers who lose their jobs is surely very small. This by no means should lead us to give up on those who take the “direct hit”—workers displaced by service trade. Their needs are often the most acute, and in this regard, ideas like wage insurance and expanding Trade Adjustment Assistance are meritorious. But as Richard Freeman has discussed (this volume), the implicit supply shock from the introduction of millions of skilled workers into a relatively concentrated set of occupations and industries may have a significant impact on the wage structure of white-collar workers, just as the increase in trade in manufacturing goods has structurally altered the wage distribution of blue-collar workers, partially contributing to the post-1979 increase in wage inequality and real wage losses, particularly for men. In this sense, Jensen and Kletzer may be overstating the breadth of their work by giving their piece the subtitle: “Understanding the Scope and Impact of Services Offshoring.” They get us a long way, further than any previous forays, toward identifying the most visible victims of offshoring: those who lose their jobs. But if Samuelson and others are right about the impact of competitive pressures on the United States from trade with low-cost countries in sectors where we have held a comparative advantage, the scope and impact of offshoring could spill over far beyond those directly affected. Robert C. Feenstra: This is a good paper that introduces a new technique for classifying service industries as tradable and nontradable and then pursues a number of applications. The technique involves looking at the geographic concentration of service industries, using the idea that a more concentrated industry is most likely tradable. Geographic concentration is measured using population census data from the PUMS files, which also allow us to track individuals’ occupations as well as their industries of employment. So the paper not only introduces a new technique for measure of the tradability of industries or occupations, it also shows how it can be implemented on a dataset that is novel for trade economists. I actually thought of using the geographic concentration of industries to measure something about trade some years ago, when reading a Scientific American article (Landy 1999) dealing with the distribution of stars in the universe. The “cosmological principle” states that the universe overall is homogeneous, so galaxies have no particular pattern. That is true on a very large scale, but on


125

smaller scales, galaxies form into clusters that are fractal: even as the scale of observation is reduced, the basic pattern of galaxies is the same. The extent to which galaxies cluster together can be measured by their spatial correlation. When reading that article I thought that the same should be true of the location of economic activity: we could use spatial correlation or some other technique to measure the clustering of industries. That is exactly what the authors do here, using the Gini coefficient and a second measure of concentration. They find that the clustering or concentration of many service industries is just as strong as for manufacturing industries, implying that these service activities must be traded. While my reference to astronomy is just for fun, economists also use the concentration of industries to make conclusions about trade. Jean Imbs and Romain Wacziarg (2003) have shown, for example, that for developing countries the concentration of industries first falls and later increases as the countries mature, so the Gini coefficient follows a U-shaped pattern. For China, Alwyn Young (2000) found that after trade was opened the concentration of industries across provinces fell, which seemed to be contrary to comparative advantage, where we would expect regions to specialize. But later research found that industries in China later became more specialized across provinces, so the Gini coefficient also follows a U-shaped pattern in that country (see Naughton 2003; Poncent 2003). From these examples I conclude that using the concentration of industries to measure their trade orientation is well motivated and that the application to service industries is entirely new. Let us now consider the results of the paper. Using the Gini coefficients of geographic concentration, the authors divide industries into three groups: those with a Gini of less than 0.1 being the least concentrated, and therefore nontradable; those with a Gini above or equal to 0.3 being the most concentrated, and therefore tradable, and those with a Gini between 0.1 and 0.3 in an intermediate category, but also treated as tradable. The classification of industries into these three groups is appealing: there are only a handful of nontraded manufacturing industries, including cement and concrete, whereas service industries are evenly divided between nontraded and traded activities. There are some anomalies, however: the education sector is very diversified geographically, so it is classified as nontradable, despite the fact that it is a principal service export of the United States. The geographic diversification of education holds for elementary and high schools, as well as colleges and universities (see Jensen and Kletzer’s table 2), perhaps because of the land grant system in higher education. Because the authors use census data on individuals from the PUMS files, they can also distinguish tradable occupations as opposed to tradable industries.

126


That is, they can measure the geographic concentration of job titles rather than just industries. These job titles are unfamiliar to trade economists, so some further explanation would be desirable. For example, occupational titles within the life, physical, and social sciences are mostly tradable; that is, these persons are geographically concentrated in their employment (see table 5). About half of these persons work in nontraded industries (such as education, which is not concentrated in space), and another half work in traded industries (see table 6). So at this point I could use some examples to understand the classifications: how can most of the employment in the life, physical, and social sciences be concentrated, when a significant number of these individuals work in education, which is not concentrated? In the next part of the paper, the authors investigate the characteristics of workers as classified by the tradability of their industry and occupation. Workers in traded industries are more highly skilled and are paid more than in nontraded industries, and this is especially true in traded service industries. The same is true for occupations: workers in tradable occupations earn more and have more education than those in nontradable occupations. Even if we strip out the effect of higher education, a wage premium persists for the traded industries, especially for traded service industries: these workers command a premium over and above their education level and demographic characteristics. The premium is about 6 percent for traded manufacturing and 15 percent for traded professional service industries. These results reminded me of two other related studies. First, Jeffrey Sachs and Howard Shatz (1998) made the point that services really are more skillintensive than manufacturing. The characterization of service jobs as flipping hamburgers is not true on average, where the jobs are more likely to be professional. Second, I was reminded of the earlier studies on the wage premiums in manufacturing by Larry Katz and Larry Summers (1989a, 1989b). They found that capital-intensive industries in manufacturing pay higher wages, and since these industries have higher exports, there is a wage premium in exporting. Trade economists were always squeamish about this finding, since it runs the risk of implying that being an exporter leads to paying higher wages, therefore suggesting that a subsidy to exports might help. On the contrary, most of us would believe that being more productive at the plant level leads to being an exporter and paying higher wages, with little or no role for export subsidies (see Fernandez 1989). The authors then investigate the growth across industries and occupations. In this I did not agree with the their expectations regarding which sectors would grow the most. For example, they state: “High-skill activities are consistent with


127

U.S. comparative advantage, and we would expect that as trade increases, economic activity would shift to activities consistent with U.S. comparative advantage. Thus we would expect higher-skill industries and occupations to have higher rates of employment growth.” My difficulty with this logic is that it all depends on whether the United States is benefiting from increased export opportunities in the sectors where it has comparative advantage, or, on the contrary, whether it is facing new competition in those sectors. Paul Samuelson (2004) suggests that outsourcing could cause the United States to face competition in sectors where it formerly had comparative advantage. That is different from what Jensen and Kletzer have in mind. What they actually find is that service employment expanded during the period 1998–2003 and manufacturing employment contracted, and this shift holds regardless of whether one looks at traded or nontraded industries. So on the issue of employment growth, the methods developed in this paper to measure tradability just do not give us any extra explanatory power. We are back to the hypothesis advanced by James Harrigan and Rita Balaban (1999) and also by Bernardo Blum (2004): namely, that it is the rise in the service sector in the United States, combined with the skill-intensity of that sector (Sachs and Shatz 1998), that explains the rising relative wages of skilled workers. We still do not know whether this shift toward services comes from demand pressure, trade, productivity, or some other cause. It would have been nice if the tradability of service industries gave us extra insight on this issue, but that is not what the empirical results here show. In the final section of the paper the authors examine job loss and the characteristics of displaced workers. This is an issue that Lori Kletzer has written on extensively, and the results here complement her earlier findings. Workers in tradable industries face a notably higher rate of job loss than those in nontradables. That is particularly true in service industries and in white-collar occupations. Nevertheless, it is still true that production workers in the United States have a higher rate of job loss than those in nonproduction and white-collar occupations, including those occupations that we believe are being affected by services outsourcing. General Discussion: Many participants commended the authors for their extremely creative and useful paper. The discussion also raised a variety of issues of interpretation and suggestions for further work, with some questioning how well domestic geographic concentration could capture international tradability. Perhaps not surprisingly, a number of speakers found the results surprising for particular industries or occupations.

128


Lael Brainard highlighted two reasons why the authors’ concentration index approach to identifying tradability was particularly valuable. First, it can be applied across occupations as well as across industries. Second, it gets around the problem that direct measurement is more difficult for services than for goods production. Internationalization of services essentially entails linking domestic and foreign factors of production so that work moves between product or project teams. It is extremely difficult to quantify the value added from each step of the process. Brainard also wondered why the authors focused on a bivariate indicator (whether something was tradable or nontradable) in their empirical analyses instead of exploiting the continuous variable that they constructed. She and others saw their use of an essentially arbitrary threshold as throwing away potentially useful information. The revised version of the paper does provide the actual indicators for major sectors and occupations. Some participants suggested that it would be helpful to compare the results of the tradability measure constructed here with other available alternatives. This would be one way to explore how well it captures what we mean by tradability. Brainard noted that we have direct tradability indicators for merchandise. She expected to find that some highly tradable goods, such as sugar, are not particularly highly concentrated. Catherine Mann wondered whether the approach by Brad Jensen and Lori Kletzer had implications similar to the work by Frank Levy and Richard Murnane, which classifies tasks in terms of routinization. Susan Collins asked how similar it was to the classification by Desirée van Welsum and Xavier Reif. The issue of comparability is partially addressed in the introduction to this volume. Robert Lawrence advanced another way to look at the paper, focusing on agglomeration. The results show that even inside the United States, where firms are free to set up everywhere, they often choose not to, presumably because of the benefits of locating near one another. Clearly, if costs were different enough abroad, they would choose to relocate. But it may be that the more concentrated firms are now, the greater the agglomeration benefits and the less likely they are to move. From this perspective, we should see their concentration as comforting, not threatening. Lawrence also stressed that one should not jump from tradability in the sense of this paper to trade. For example, his work with Martin Baily finds considerable job loss in the computer industry, which is tradable. But their input-output table analysis concludes that this is overwhelmingly due to declines in domestic demand and that trade appears to have played a relatively minor role. Other participants elaborated on Robert Feenstra’s point that domestic tradability may be very different from international tradability. In particular, T. N.


129

Srinivasan noted that if transactions and transportation costs are much less for domestic than for international trade, then domestic tradability has no implications for international tradability. Srinivasan also raised the point that often the same industry can produce using different technologies. For example, steel, which is certainly tradable, can be produced using both integrated mills that are quite concentrated and the more recent electronic processing mini-mills, which tend to be quite dispersed. He argued that it is important to consider technology in assessing whether concentration provides a good indicator of tradability. He also pointed out that occupation, and perhaps to some degree industry, is a matter of choice. Thus he suggested controlling for selection when estimating the earnings regressions. Catherine Mann noted that regulations can play a very important role in some service sectors. This includes legal bar exams, state-specific insurance regulations, and others. There are also significant differences in cross-country regulations. Thus it would be interesting to explore whether changes in state-specific regulations that make a particular industry more easily traded have affected its occupational stratification or its concentration indicators. Changes in rules for interstate banking are one especially interesting recent example. Mann also asked what the results in the paper could tell us about the risk versus expected return associated with particular occupations. Job loss is certainly very costly. However, her casual impression was that the empirical estimates find a relatively large wage premium for jobs in risky service industries and occupations, and it was worth exploring how this compared with the probability and expected costs of job loss. In contrast, manufacturing jobs are also risky but have been commanding a much smaller wage premium. Lawrence Mishel raised concerns about drawing conclusions from simply comparing employment growth in traded and nontraded industries (or occupations) within a given time period. Because employment trends may be quite different, he thought it important to develop a more convincing counterfactual that incorporates information about previous trend behavior. David Richardson suggested that it would be interesting to consider other concentration measures. For instance, the Ellison-Glaeser measure comes very close to an indicator of revealed comparative advantage. He also noted that the authors should be looking for both industries and occupations with very low concentration and those with very high concentration, because unusually low ratios for production to state GDP are also an indicator of (domestic) tradability. Collins noted that it might be helpful to distinguish between different types of services, and that the domestic concentration approach could be more

130


appropriate for some types than for others. The General Agreement on Trade in Services (GATS) distinguishes among four modes by which services are traded. For example, mode 1 includes services supplied from one country to another, such as telephone calls, while mode 2 includes consumers who use a service in another country, such as tourists and students studying at a foreign university. It seemed to her that domestic concentration might be a better indicator of tradability for mode 1 services than for mode 2.


131

References Amiti, Mary, and Shang-Jin Wei. 2004. “Fear of Service Outsourcing: Is It Justified?” Working Paper WP/04/186. Washington: International Monetary Fund. Arora, Ashish, and Alfonso Gambardella. 2004. “The Globalization of the Software Industry: Perspectives and Opportunities for Developed and Developing Countries.” Working Paper 10538. Cambridge, Mass.: National Bureau of Economic Research. Bardhan, Ashok Deo, and Cynthia A. Kroll. 2003. “The New Wave of Outsourcing.” Working Paper 1103. University of California, Berkeley: Fisher Center for Real Estate and Urban Economics. Bernard, Andrew B., J. Bradford Jensen, and Peter K. Schott. Forthcoming 2006. “Survival of the Best Fit: Exposure to Low Wage Countries and the (Uneven) Growth of U.S. Manufacturing Plants.” Journal of International Economics 68 (1): 219–37. Bhagwati, Jagdish, Arvind Panagariya, and T. N. Srinivasan. 2004. “The Muddles over Outsourcing.” Journal of Economic Perspectives 18 (Fall): 93–114. Blum, Bernardo S. 2004. “Trade, Technology and the Rise of the Service Sector: An Empirical Assessment of the Effects on U.S. Wage Inequality.” University of Toronto (September). Brainard, Lael, and Robert E. Litan. 2004. “Offshoring Service Jobs: Bane or Boon— and What To Do?” Brookings Policy Brief 132 (April). Bronfenbrenner, Kate, and Stephanie Luce. 2004. “The Changing Nature of Corporate Global Restructuring: The Impact of Production Shifts on Jobs in the U.S., China, and around the Globe.” U.S.–China Economic and Security Review Commission. October. Dossani, Rafiq, and Martin Kenney. 2004. “The Next Wave of Globalization? Exploring the Relocation of Service Provision to India.” Working Paper 156. Stanford University: Asia Pacific Research Center. ———. 2003. “Went for Cost, Stayed for Quality? Moving the Back Office to India.” Stanford University: Asia-Pacific Research Center. Duranton, Gilles, and Henry G. Overman. 2004. “Testing for Localisation Using MicroGeographic Data.” London School of Economics. Ellison, Glenn and Edward L. Glaeser. 1999. “The Geographic Concentration of Industry: Does Natural Advantage Explain Agglomeration?” American Economic Review, Papers and Proceedings 89 (May): 311–16. ———. 1997. “Geographic Concentration of U.S. Manufacturing Industries: A Dartboard Approach.” Journal of Political Economy 105 (October): 889–927. Farber, Henry S. 2005. “What Do We Know about Job Loss in the United States? Evidence from the Displaced Worker Survey, 1984–2004.” Working Paper 498. Princeton University: Industrial Relations Section. ———. 2003. “Job Loss in the United States, 1981–2001.” Working Paper 471. Princeton University: Industrial Relations Section. ———. 1993. “The Incidence and Costs of Job Loss: 1982–1991.” BPEA, no. 1: 73–132. Fernandez, Raquel, 1989. Comment on “Industrial Wage Differential, International Competition and Trade Policy,” by Lawrence F. Katz and Lawrence H. Summers. In

132


Trade Policies for International Competitiveness, edited by Robert Feenstra, pp. 117–22. University of Chicago Press. Gardner, Jennifer M. 1993 “Recession Swells Count of Displaced Workers.” Monthly Labor Review 116 (June): 14–23. Gartner Research. 2004. “Worldwide IT Services Market Forecast, 2002–2007” (January 12). Gentle, Chris. 2003. “The Cusp of a Revolution: How Offshoring Will Transform the Financial Services Industry.” Deloitte Research. Harrigan, James, and Rita A. Balaban. 1999. “U.S. Wages in General Equilibrium: The Effects of Prices, Technology, and Factor Supplies, 1963–1991.” Working Paper 6981. Cambridge, Mass.: National Bureau of Economic Research. Helpman, Elhanan, and Paul R. Krugman. 1985. Market Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and the International Economy. Cambridge, Mass.: MIT Press. Imbs, Jean, and Romain Wacziarg. 2003. “Stages of Diversification.” American Economic Review 93 (March): 63–86. Katz, Lawrence F., and Lawrence H. Summers. 1989a. “Can Interindustry Wage Differentials Justify Strategic Trade Policy?” In Trade Policies for International Competitiveness, edited by Robert Feenstra, pp. 85–116. University of Chicago Press. ———. 1989b. “Industry Rents: Evidence and Implications.” In Brookings Papers on Economic Activity: Microeconomics 1989, edited by Martin N. Baily and Clifford Winston, pp. 209–75. Brookings. Kirkegaard, Jacob F. 2004. “Outsourcing—Stains on the White Collar?” Washington: Institute for International Economics (February). Kletzer, Lori G. 2001. Job Loss from Imports: Measuring the Costs. Washington: Institute for International Economics. ———. 1998. “Job Displacement.” Journal of Economic Perspectives 12 (Winter): 115–36. ———. 1995. “White Collar Job Displacement, 1983–91.” In Proceedings of the 47th Annual Meeting, pp. 98–107. Madison, Wis.: Industrial Relations Research Association. Krugman, Paul R. 1991. Geography and Trade. Cambridge, Mass.: MIT Press. Landy, Stephen D. 1999. “Mapping the Universe.” Scientific American 280 (June): 38–45. Mann, Catherine L. 2003. “Globalization of IT Services and White Collar Jobs: The Next Wave of Productivity Growth.” Policy Brief 03-11. Washington: Institute for International Economics. McCarthy, John C. 2002. “3.3 Million U.S. Services Jobs To Go Offshore.” San Francisco: TechStrategyTM Research, Forrester Research (November). Naughton, Barry J. 2003. “How Much Can Regional Integration Do to Unify China’s Markets?” In How Far across the River? Chinese Policy Reform at the Millennium, edited by Nicholas Hope, Dennis Tao Yang, and Mu Yang Li, pp. 204–32. Stanford University Press.


133

Podgursky, Michael. 1992. “The Industrial Structure of Job Displacement 1979–89.” Monthly Labor Review 115 (September): 17–25. Poncet, Sandra. 2003. “Measuring Chinese Domestic and International Integration.” China Economic Review 14 (1): 1–21. Sachs, Jeffrey D., and Howard J. Shatz. 1998. “International Trade and Wage Inequality: Some New Results.” In Imports, Exports, and the American Worker, edited by Susan M. Collins, pp. 215–40. Brookings. Samuelson, Paul A. 2004. “Where Ricardo and Mill Rebut and Confirm Arguments of Mainstream Economists Supporting Globalization.” Journal of Economic Perspectives 18 (Summer): 135–46. Schultze, Charles L. 2004. “Offshoring, Import Competition, and the Jobless Recovery.” Brookings Policy Brief 136 (August). Tilton, Andrew. 2003. “Offshoring: Where Have All the Jobs Gone?” U.S. Economic Analyst 03/38 (September 19). Goldman Sachs & Co. Van Welsum, Desirée, and Graham Vickery. 2005. “Potential Offshoring of ICTIntensive Using Occupations.” DSTI Information Economy Working Paper DSTI/ICCP/IE(2004)19/FINAL. Paris: Organization for Economic Cooperation and Development. Young, Alwyn. 2000. “The Razor’s Edge: Distortions and Incremental Reform in the People’s Republic of China.” Quarterly Journal of Economics 115 (November): 1091–35.


MARIA BORGA U.S. Bureau of Economic Analysis

Trends in Employment at U.S. Multinational Companies: Evidence from Firm-Level Data

B

ecause of their networks of foreign affiliates and their well-established trade channels, U.S. multinational companies (MNCs) would be expected to play a leading role in the offshoring of production of goods and services. As a result, there has been much discussion about the impact of offshoring on employment at U.S. MNCs. In this paper, I use firm-level data from the Bureau of Economic Analysis (BEA) to examine the evidence on the extent of offshoring by U.S. parent companies to their foreign affiliates and then to determine if, and how, offshoring is associated with changes in U.S. parent employment. While no standard definition of offshoring exists, the Government Accountability Office (GAO) recently defined the offshoring of services as generally referring to “an organization’s purchases from abroad (imports) of services that it previously produced in-house or purchased from another domestic source.”1 This definition could be adapted to apply equally well to goods. Such an expanded definition is applicable to all U.S. firms and is broad enough to encompass several different types of offshoring behavior. For example, it covers the case in which a U.S. parent company begins to purchase an input, either a good or a service, from a foreign affiliate or from an unaffiliated foreign supplier that it previously either produced in-house or purchased from a domestic supplier. It

The author wishes to thank this paper’s discussants, Robert Lawrence and Catherine Mann, and participants at the Brookings Trade Forum for helpful comments. The author also thanks her colleagues at the U.S. Bureau of Economic Analysis for their valuable suggestions and comments. The views expressed in this paper are those of the author and do not necessarily reflect those of the Bureau of Economic Analysis. 1. GAO (2004, p. 2). This report goes on to observe, “The term offshoring has also been used in the public debate to include several other types of international trade and foreign investment activities.”

135

136


would also cover the case of a U.S. parent company choosing to transfer to a foreign affiliate the production of a good or service produced domestically and exported to foreign markets. This paper does not attempt to deal with all types of offshoring, but instead focuses on a very specific form of offshoring: the case in which a U.S. MNC chooses to produce a good or service at a foreign affiliate and then to import that good or service back to the U.S. parent company to use in its production or to resell. This setup is a specific form of vertical foreign direct investment (FDI), with production located in other countries to take advantage of differences in relative factor costs or to exploit economies of scale.2 On the opposite end of the spectrum is horizontal FDI, in which MNCs establish foreign affiliates whose production processes replicate those of the parents.3 A single MNC may demonstrate characteristics of both vertical and horizontal FDI. For example, it may expand horizontally by establishing an affiliate in Europe to serve the local market, and also vertically by establishing an affiliate in Mexico to assemble and reexport parts fabricated in the United States or by establishing an affiliate in India to provide accounting services to the U.S. parent. Much of the discussion about offshoring by U.S. MNCs has centered on its impact on employment in the United States. It has been conjectured that as U.S. MNCs offshore there will be a negative impact on employment in the United States, either from production lost in the United States to foreigners or from forgone job creation. This conjecture, however, ignores the potential positive impact that offshoring can have on the MNC as a whole. If offshoring achieves cost savings, then the MNC’s total sales may expand, resulting in increased demand for the goods and services still provided by the U.S. parent as well as by its foreign affiliates. In an extreme case, the cost savings from offshoring may enable the MNC to survive. BEA has examined long-term trends in the aggregate data on U.S. MNCs.4 This analysis showed that worldwide operations of U.S. MNCs are concentrated in the United States, that the foreign operations of U.S. MNCs are centered in high-income countries, and that most of the output of foreign affiliates is sold in local or other foreign markets rather than exported to the United States. However, the aggregate data used in this analysis could mask diversity at the firm level by giving greater weight to larger firms. This paper builds on the earlier 2. Another form of vertical FDI occurs when the foreign affiliate of an MNC sells its output to another affiliate of the U.S. parent for use as an intermediate input rather than selling it back to the parent. For a discussion of vertical FDI, see Helpman and Krugman (1985). 3. See Markusen (2002). 4. U.S. Bureau of Economic Analysis (2004, pp. 52–56).

Maria Borga

137

work by examining the firm-level data. Whereas the earlier work portrayed aggregate patterns in the data, the present analysis is more directed at characterizing and explaining typical firm behavior. In general, it confirms the earlier findings and makes some additional ones as well: —While U.S. parents increased their reliance on purchased goods and services between 1994 and 2002, there was no significant association between increased reliance on purchased inputs and decreases in parent employment. However, changes in parents’ reliance on imports of goods from their foreign affiliates was significantly and negatively associated with changes in parent employment. In contrast, there was no significant correlation between changes in parents’ reliance on imports of services from their affiliates and changes in parent employment. —More of the change in parent employment between 1994 and 2002 is attributable to changes in output and in labor productivity than to changes in the use of purchased inputs. —Growth at U.S. parents and at their foreign affiliates is closely and positively linked. This growth can result from mergers and acquisitions as well as the expansion of existing operations. —On average, the share of sales by foreign affiliates to local markets increased between 1994 and 2002, suggesting that market access is an increasingly important reason for investing overseas. The increase in the share of sales sold locally coincided with an increase in the share of affiliate employment in low-income countries. This result suggests that for investment in low-income countries, market access, not just factor cost differences, is an important consideration. In the remaining sections of the paper I first describe the data set. I then present descriptive statistics from the panel data set. Next I estimate the relative impacts of the main sources of change in parent employment. I then seek to explain associations between changes in employment at U.S. parents and changes in selected characteristics of the U.S. parents, their affiliates, and the entire MNC. Finally, I offer conclusions and suggest some avenues for future research.

The Data Set The analysis is based on a panel of 1,117 U.S. parent companies that responded to both BEA’s 1994 benchmark survey and its 2002 annual survey of U.S. direct investment abroad. The panel thus excludes firms that entered or dropped out of the universe after 1994. By looking at a sample of long-lived

138


parents over an extended period, it is possible to distinguish changes in behavior that may be obscured in the aggregate data covering all parents, particularly because of the entry of new parents into the universe. It is mandatory for firms to respond to BEA’s direct investment surveys, which obtain data on the financing and operations of U.S. parents and their foreign affiliates. The analysis that follows covers only majority-owned foreign affiliates because U.S. parents are likely to exhibit a higher degree of management influence and control over majority-owned affiliates than minority-owned affiliates. Also, only nonbank affiliates are included because bank affiliates are not required to report financial and operating data to BEA for years not covered by benchmark surveys. One of the variables that is particularly relevant to a discussion of offshoring is the extent to which U.S. parents’ reliance on purchased goods and services has changed over time. Because the resale of purchased goods is the principal business of wholesalers and retailers, including them in the data set could obscure any changes in the reliance of parents in other industries on purchased inputs.5 Therefore all parents whose primary industry is wholesale or retail trade have been excluded from the data set.6 Mergers and acquisitions affected the companies that were included in the panel data set. Take the case in which a U.S. parent company (company A) that reported on the 1994 benchmark survey was acquired by another U.S. parent company (company B) before 2002. Company A (and its affiliates) was not included in the panel data set because it did not report on the 2002 annual survey. Company B was included in the panel data set because it did report in both 1994 and 2002. Company B’s reported data in 2002 would reflect the acquisition of company A and its affiliates. The acquisition would manifest itself as growth in employment (at both the parent and the affiliates), in sales, and so on. The variables included in the data set are employment, employee compensation, value added, sales, expenditures on research and development (R&D), and

5. In estimating the gross output of the wholesale and retail trade industries, the goods for resale are excluded from the value of intermediate inputs consumed in production by wholesalers and retailers because these goods are subject to only minimal processing, such as cleaning or packaging. 6. U.S. parents usually have operations in multiple industries but are classified in a primary industry on the basis of their sales data. Throughout the paper, the parent’s primary industry classification in 2002 will be used to classify the data. Because parents can have operations in multiple industries, those that have secondary activities in wholesale or retail trade remain in the data set.

Maria Borga

139

exports and imports of goods. Sales are disaggregated into goods, services, and investment income. Affiliates’ sales are disaggregated into six destinations: to the host country, to the United States, or to third countries, and then further for each of these into sales to either affiliated or unaffiliated parties. The sample accounts for 69 percent of employment and 68 percent of sales of all parents in 1994; by 2002 the percentage of employment covered by the sample had fallen to 63 percent, but the percentage of sales covered had increased to 69 percent. Most of the characteristics for the sample match the population. However, there are some differences. The sample shows slower employment growth than the population. Also, it slightly underrepresents services firms and overrepresents firms that rely on imports from their affiliates in their production. For further information about the data set and a complete comparison of the panel data set with the population, see the appendix.

Descriptive Statistics This section begins with an examination of changes between 1994 and 2002 in the mean values of selected characteristics of the firms in the sample data set. The characteristics examined describe the U.S. parents, their foreign affiliates, and the MNC as a whole, such as the distribution of employment between the parents and their affiliates. Several of these characteristics are expected to be associated with changes in U.S. parent employment, such as changes in sales, in labor productivity, in the share of purchased inputs in the parents’ production, and in the sourcing of inputs by parents from their foreign affiliates. Other characteristics, such as R&D spending, are less closely associated with changes in U.S. parent employment but are still of interest. Then the sample is divided between parents that lost employment and those that gained employment, and the characteristics of those two subsets are contrasted. Finally, the characteristics are examined by industry of the U.S. parent. Characteristics of the Firms in the Sample Table 1 shows the mean values of selected sample characteristics in 1994 and 2002. Average employment growth for the parents in the sample was 39 percent. Note that an increase in employment for a parent could result not only from an expansion of existing operations, but also from the merger with, or acquisition of, an existing U.S. company. By the same token, a decrease in a parent’s

140


Table 1. Characteristics of the MNCs in the Panel: Mean Values of Selected Variable, 1994 and 2002 1994 Variables U.S. parent variables Employment change (percent) Salesa (millions of dollars) Ratio of purchases to sales Share of services in sales International share of sales R&D intensity Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Affiliate variables Employment change (percent) Share of services in sales Share of sales to local market Share of sales to third countries Share of sales to U.S. Value added per employeeb (thousands of dollars) Compensation per employeeb (thousands of dollars) MNC variables Parent share of employment Parent share of sales Parent share of value added Share of affiliate employment in low-income countries Affiliated goods imports as a share of U.S. parent purchases Unaffiliated goods imports as a share of U.S. parent purchases Affiliated services imports as a share of U.S. parent purchases Number of observations

2002

Mean

Standard deviation

Mean

Standard deviation

… 2,378 0.60 0.22 0.14 0.031

… (7,650) (0.17) (0.39) (0.17) (0.05)

39 3,190 0.63 0.24 0.14 0.029

(1.02) (9,712) (0.18) (0.40) (0.18) (0.07)

86.58

(59.99)

89.43

(70.30)

55.00

(25.44)

61.65

(28.99)

… 0.21 0.70 0.19 0.11

… (0.39) (0.31) (0.26) (0.21)

231 0.24 0.78 0.14 0.08

(7.87) (0.41) (0.27) (0.08) (0.16)

n.a.

n.a.

89.60

(580.65)

n.a.

n.a.

44.59

(30.49)

0.77 n.a. n.a.

(0.20) n.a. n.a.

0.72 0.73 0.80

(0.22) (0.20) (0.98)

0.20

(0.32)

0.27

(0.32)

0.035

(0.09)

0.046

(0.11)

0.030

(0.08)

0.030

(0.09)

0.004

(0.04)

0.002

(0.02)

1,117

1,117

a. For 1994, rough estimates of U.S. parents’ sales, value added, and employee compensation in 2002 dollars were derived using the U.S. GDP deflator. b. For 1994, estimates of affiliates’ sales, value added, and employee compensation in 2002 dollars were not derived because of the difficulties of accounting for changes in exchange rates from 1994 to 2002. n.a. = Not available.

Maria Borga

141

employment could result either from the contraction of existing operations or from the selling off of some of the parent’s operations. U.S. PARENT CHARACTERISTICS. The ratio of purchased goods and services by U.S. parents to their sales measures the reliance of parents on inputs purchased from outside the firm rather than on production inside the firm.7 These purchases could be sourced from domestic companies or imported from foreign suppliers, both affiliated and unaffiliated. Thus, this measure encompasses both the outsourcing (inputs provided by a third party, either domestic or foreign) and offshoring (inputs provided by a foreign company, either affiliated or unaffiliated) behavior of the firms in the sample. Parents’ offshoring behavior is examined by looking at imports as a share of their purchased inputs. The data show that reliance on purchases of goods and services increased between 1994 and 2002, from 60 percent to 63 percent. The share of sales accounted for by services increased slightly between 1994 and 2002, as sales of services increased more rapidly than sales of goods. The share of U.S. parents’ sales accounted for by international sales remained constant, indicating that U.S. parents’ international orientation did not change between 1994 and 2002. R&D intensity (measured as the ratio of R&D expenditures to sales) fell slightly between 1994 and 2002. Mean labor productivity, as measured by value added per employee, grew between 1994 and 2002; however, its growth was less than the growth in mean compensation per employee. FOREIGN AFFILIATE CHARACTERISTICS. The mean change in employment at the foreign affiliates is 231 percent. There are some parents in the panel whose foreign operations were quite small in 1994 and whose expansion resulted in employment increases at their foreign affiliates of several orders of magnitude by 2002. These few outliers raise the mean significantly. The median increase in affiliate employment was 47 percent—much lower than the mean increase in affiliate employment, but still higher than the mean increase in parent employment. Furthermore, not all of these increases in affiliate employment are due to growth at existing affiliates or to the establishment or acquisition of new affiliates overseas. For example, some affiliate employment growth is due to the acquisition of an existing U.S. parent by another U.S. parent. Such acquisitions would result in the affiliates of the acquired U.S. parents being transferred to the 7. Purchases of goods and services are calculated residually as sales minus value added. Parents’ output is measured as their sales. Ideally, to accurately reflect output, sales should be adjusted for inventory change, but because BEA collects data on parents’ inventories only for benchmark years, and then only as of the end of the year, it is not able to estimate the annual change in inventories for parents.

142


acquiring U.S. parents and would be manifested as employment growth at the affiliates of the acquiring U.S. parents. The mean share of affiliates’ sales accounted for by services increased between 1994 and 2002, indicating that, as for parents, sales of services grew faster than sales of goods for affiliates. In 2002 an average share of 78 percent of affiliates’ sales was to the local market. The high share of sales to local markets demonstrates the importance of serving local customers in the parent’s decision to invest abroad. For an affiliate, the greater the share of local sales, the more likely the affiliate was established as part of horizontal expansion abroad by its parent rather than vertical expansion. Because offshoring is defined in this paper as the parents sourcing inputs from their affiliates, horizontal expansion is less associated with offshoring than vertical expansion. Between 1994 and 2002, the share of sales to the local market increased while the share of sales to third countries and to the United States fell. This suggests that serving local markets was an increasingly important motive for U.S. MNCs’ overseas expansion. MNC CHARACTERISTICS. For the MNC as a whole, the typical parent accounts for a majority share of MNC employment, sales, and production. However, the average parent share of MNC employment fell between 1994 and 2002, reflecting the faster growth rate in employment at affiliates than at parents. The average share of affiliate employment in low-income countries increased between 1994 and 2002.8 Countries in Asia and the Pacific and Latin America and Other Western Hemisphere accounted for the largest share of growth in the low-income countries. In Asia and the Pacific, China and India accounted for the largest increases in employment. In Latin America and Other Western Hemisphere, the largest increases were in Mexico and Brazil. Notably, the increase in the average share of affiliate employment in low-income countries coincided with an increase in the local share of sales for all affiliates, demonstrating that while factor costs differences may be important in the decision to invest in lowincome countries, serving those markets is also an important reason, and perhaps the paramount one. The three characteristics that describe parents’ offshoring behavior are the share of their purchased inputs accounted for by imports of goods from their affiliates, by imports of goods from unaffiliated parties, and by imports of services from their affiliates. Parents’ use of goods imported from their foreign affiliates grew more intensive between 1994 and 2002; the average share of par8. High-income countries are defined as all members of the Organization for Economic Cooperation and Development except the Czech Republic, Hungary, Mexico, Poland, the Slovak Republic, the Republic of Korea, and Turkey. All other countries are considered low-income countries.

Maria Borga

143

ents’ purchased inputs accounted for by affiliated imports of goods increased from 3.5 percent to 4.6 percent. In contrast, there was little change in the share of parents’ purchases accounted for by imports of goods from unaffiliated parties. The share of parents’ purchased inputs accounted for by imports of services from their affiliates fell from 0.4 percent in 1994 to 0.2 percent in 2002. Summing over the share of purchased inputs accounted for by imports that can be identified in the data shows that parents’ reliance on imports increased from 6.9 percent in 1994 to 7.8 percent in 2002. Even though data on unaffiliated imports of services are not available, it is clear that the vast majority of U.S. parents’ purchases were from domestic suppliers.9 Characteristics of MNC Parents That Gained or Lost Employment U.S. PARENT CHARACTERISTICS. Parents that gained employees outnumbered parents that lost employees by 689 to 428 (see table 2). For parents that gained employees, the average employment increase of 82 percent was slightly below the average increase in sales of 84 percent. In contrast, for those parents that lost employees, the average employment decrease of 30 percent outpaced the average decrease in sales of only 10 percent. For parents that hired additional employees, in order to reach levels of sales that outpaced their employment growth, either their workforce had to become more productive or the parents had to rely to a greater extent on purchased inputs. While labor productivity was virtually constant for those parents that gained employees, they did increase their reliance on purchased inputs from 59 percent to 62 percent of sales. For parents that lost employees, the much greater reduction in employment than in sales implies either increased labor productivity or greater reliance on purchased inputs. Unlike for the parents that gained employees, both labor productivity and reliance on purchased inputs increased for parents that lost employees— labor productivity increased 8.2 percent, and purchased inputs increased from 62 percent of sales to 66 percent of sales. Employees of parents that reduced the size of their workforce had larger increases in average compensation than employees of parents whose workforce grew. Finally, services accounted for a larger share of sales for those that gained employees. FOREIGN AFFILIATE CHARACTERISTICS. While affiliate employment grew for both sets of parents, those parents who gained employees had greater average

9. Purchases from domestic suppliers could embody imported goods and services. Currently, it is not possible to estimate the import content of purchases from domestic suppliers. However, BEA is exploring the possibility of estimating these indirect imports with the use of input-output tables.

144


Table 2. Characteristics of MNCs in the Panel for U.S. Parent Firms That Gained or Lost Employment: Mean Values for Selected Variables, 1994 and 2002

U.S. parent variables Employment Salesb (millions of dollars) Employment change (percent) Ratio of purchases to sales Share of services in sales International share of sales R&D intensity Value added per employeeb (thousands of dollars) Compensation per employeeb (thousands of dollars) Affiliate variables Employment change (percent) Services as a share of sales Share of sales to local market Share of sales to third countries Share of sales to the United States Value added per employeec Compensation per employeec

Parents with employment gains

Parents with employment losses

1994

2002

1994

2002

7,554.4 (19,576.0)a 1,815 (5,106) …

11,437.3 (26,258.3) 3,342 (9,353) 0.82 (1.09) 0.62 (0.18) 0.27 (0.42) 0.14 (0.18) 0.030 (0.07) 88.93 (74.10) 59.45 (29.40)

12,184.4 (33,641.3) 3,286 (15,106) …

7,566.4 (17,859.0) 2,945 (13,312) –0.30 (0.23) 0.66 (0.18) 0.20 (0.37) 0.14 (0.17) 0.028 (0.06) 90.23 (63.77) 65.18 (27.97)

0.59 (0.18) 0.24 (0.41) 0.14 (0.17) 0.031 (0.06) 88.54 (59.96) 56.01 (28.38) … 0.24 (0.41) 0.71 (0.32) 0.19 (0.26) 0.11 (0.21) n.a. n.a.

3.07 (9.24) 0.27 (0.42) 0.80 (0.27) 0.13 (0.22) 0.07 (0.15) 72.61 (549.19) 45.11 (32.88)

0.62 (0.16) 0.18 (0.35) 0.14 (0.15) 0.030 (0.05) 83.42 (59.97) 53.39 (19.66) … 0.17 (0.35) 0.68 (0.31) 0.20 (0.25) 0.11 (0.21) n.a. n.a.

1.09 (4.65) 0.21 (0.38) 0.74 (0.28) 0.16 (0.23) 0.10 (0.17) 116.97 (627.71) 43.74 (26.19) (continued)

growth in employment at their affiliates than parents who lost employees. All affiliates increased their focus on selling to the local market, but the increase in the share of local sales was greater for affiliates of parents with employment gains. Also, there was a decrease in the share of sales to the United States for affiliates of parents that gained employees, while that share was little changed for the other affiliates.

Maria Borga

145

Table 2. Characteristics of MNCs in the Panel for U.S. Parent Firms That Gained or Lost Employment: Mean Values for Selected Variables, 1994 and 2002 (Continued)

U.S. parent variables MNC variables Parent share of employment Share of employment in lowincome countries Affiliated goods imports as a share of parent purchases Unaffiliated goods imports as a share of parent purchases Affiliated services imports as a share of parent purchases Number of observations

Parents with employment gains

Parents with employment losses

1994

2002

1994

2002

0.78 (0.20) 0.19 (0.33) 0.034 (0.09) 0.029 (0.09) 0.006 (0.05)

0.76 (0.20) 0.26 (0.32) 0.039 (0.11) 0.027 (0.08) 0.002 (0.02)

0.76 (0.20) 0.21 (0.31) 0.037 (0.08) 0.031 (0.07) 0.0015 (0.01)

0.66 (0.24) 0.28 (0.33) 0.058 (0.13) 0.033 (0.09) 0.002 (0.01)

689

428

a. Standard deviations are in parentheses. b. Rough estimates of U.S. parents’ sales, value added, and employee compensation in 2002 dollars were derived using the U.S. GDP deflator. c. For 1994, estimates of affiliates’ sales, value added, and employee compensation in 2002 dollars were not derived because of the difficulties of accounting for changes in exchange rates from 1994 to 2002. n.a. = Not available.

MNC CHARACTERISTICS. Those parents that lost employees strengthened their ties to their affiliates by increasing their reliance on imported goods from them, from 3.7 percent of purchases to 5.8 percent of purchases. To a lesser extent, they increased their reliance on services imported from their affiliates, from 0.15 percent to 0.2 percent. These parents also increased their reliance on goods imported from unaffiliated parties. On the other hand, parents that gained employees had a smaller increase in reliance on imports of goods from their affiliates, from 3.4 percent to 3.9 percent of purchased inputs, and reduced their reliance on imports of services from their affiliates, from 0.6 percent to 0.2 percent. They also reduced their reliance on imports of goods from unaffiliated parties.

Characteristics by Industry of the U.S. Parent Tables 3, 4, and 5 show the mean values for the parent, the affiliate, and the MNC characteristics, respectively, using three broad industry categories for the U.S. parent: manufacturing, services, and “other” industries.10 10. For this paper, services are defined as the following NAICS sectors: information; professional, scientific, and technical services; and finance and insurance (except depository institutions because the data set consists of only nonbank parents and affiliates). “Other” industries are all remaining NAICS sectors except manufacturing.

146


Table 3. Characteristics of Parent Firms in the Panel by Industry: Mean Values of Selected Variables, 1994 and 2002 1994 Characteristics Parents in manufacturing Employment change (percent) Ratio of purchases to sales Salesa (millions of dollars) Share of services in sales International share of sales R&D intensity Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Number of observations Parents in services Employment change (percent) Ratio of purchases to sales Salesa (millions of dollars) Share of services in sales International share of sales R&D intensity Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Number of observations Parents in “other” industries Employment change (percent) Ratio of purchases to sales Salesa (millions of dollars) Share of services in sales International share of sales R&D intensity Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Number of observations

2002

Mean

Standard deviation

Mean

Standard deviation

… 0.62 2,080 0.02 0.16 0.033

… (0.14) (7,647) (0.10) (0.16) (0.04)

31 0.65 2,770 0.02 0.16 0.034

(0.93) (0.16) (9,812) (0.08) (0.16) (0.07)

(46.72)

85.51

(60.29)

(16.50)

57.97

82.18 52.26 822 … 0.55 4,113 0.83 0.09 0.039

… (0.25) (9,670) (0.33) (0.16) (0.10)

100.17 73.12

57 0.58 5,942 0.94 0.10 0.028

(86.78)

106.55

(48.40)

81.88

163 … 0.57 2,097 0.68 0.10 0.005 101.83 50.45 132

(23.50) 822 (1.23) (0.24) (11,903) (0.19) (0.21) (0.08) (83.90) (40.46) 163

… (0.22) (3,477) (0.42) (0.20) (0.02)

81 0.61 2,405 0.74 0.08 0.003

(91.59)

94.74

(21.03)

59.62

(2.63) (0.22) (3,926) (0.38) (0.19) (0.01) (101.00) (32.52) 132

a. Rough estimates of U.S. parents’ sales, value added, and employee compensation in 2002 dollars were derived using the U.S. GDP deflator.

Maria Borga

147

U.S. PARENT CHARACTERISTICS. There are 822 manufacturing parents in the panel, far outnumbering the 163 parents in services and the 132 parents in “other” industries (see table 3). Parents in “other” industries had the fastest average employment growth between 1994 and 2002, at 81 percent, but the smallest increase in average sales, at 15 percent. While their reliance on purchased inputs increased, from an average of 57 percent of sales to 61 percent, their average labor productivity fell. Parents in services had the second fastest average employment growth, at 57 percent, but the largest increase in average sales, at 44 percent. These parents increased their reliance on purchased inputs, from an average of 55 percent of sales to 58 percent, and the average productivity of their workers increased. Parents in manufacturing had the slowest average growth in employment, at 31 percent, but the second largest increase in average sales, at 33 percent. These parents increased their reliance on purchased inputs, from an average of 62 percent of sales to 65 percent, and their average labor productivity increased. Thus, for parents in each industry group, employment, sales, and reliance on purchased inputs increased. For parents in manufacturing and services, labor productivity increased, while it fell for parents in “other” industries. FOREIGN AFFILIATE CHARACTERISTICS. Affiliates of parents in services had the fastest employment growth between 1994 and 2002, followed by affiliates of parents in “other” industries and in manufacturing (table 4). The share of sales sold locally increased for affiliates of parents in manufacturing and in services. It fell for affiliates of parents in “other” industries, but it was the share of sales to third countries, not to the United States, that increased for affiliates of parents in “other” industries. Indeed, the share of sales to the United States fell for all three industry groups. MNC CHARACTERISTICS. Parents in manufacturing accounted for a smaller share of MNC employment and sales than other parents (see table 5). Parents in “other” industries had the largest share of affiliate employment in low-income countries, but parents in services had the largest increase in the share of employment in low-income countries. For parents in services and in manufacturing, the increases in affiliate employment in low-income countries coincided with increases in the share of sales to the local market, suggesting that serving these markets, rather than exploiting differences in factor costs, was the principal reason for investing in them. Turning to parents’ offshoring behavior, parents in manufacturing used imported goods the most intensively—the share of purchased inputs accounted for by goods imported from affiliates increased from 4.6 to 5.9 percent, and the share accounted for by goods imported from unaffiliated parties was constant at

148


Table 4. Characteristics of the Majority-Owned Foreign Affiliates in the Panel by Industry of the Parent: Mean Values of Selected Variables, 1994 and 2002 1994 Characteristics Affiliates of parents in manufacturing Employment change (percent) Share of services in sales Salesa (millions of dollars) Share of sales to local market Share of sales to third countries Share of sales to U.S. Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Number of observations Affiliates of parents in services Employment change (percent) Share of services in sales Salesa (millions of dollars) Share of sales to local market Share of sales to third countries Share of sales to U.S. Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Number of observations Affiliates of parents in “other” industries Employment change (percent) Share of services in sales Salesa (millions of dollars) Share of sales to local market Share of sales to third countries Share of sales to U.S. Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars) Number of observations

2002

Mean

Standard deviation

Mean

Standard deviation

… 0.03 n.a. 0.65 0.23 0.12

… (0.12) n.a. (0.31) (0.26) (0.22)

1.89 0.04 1,858 0.75 0.16 0.10

(6.91) (0.15) (9,822) (0.27) (0.22) (0.17)

n.a.

n.a.

79.96

(215.45)

n.a.

40.52

n.a. 822

(17.13) 822

… 0.79 n.a. 0.83 0.12 0.05

… (0.38) n.a. (0.26) (0.22) (0.13)

3.87 0.91 1,623 0.88 0.10 0.02

(10.37) (0.24) (6,146) (0.23) (0.21) (0.07)

n.a.

n.a.

100.22

(161.15)

n.a.

62.75

n.a. 163 … 0.64 n.a. 0.85 0.07 0.08

… (0.45) n.a. (0.27) (0.18) (0.20)

n.a. n.a. 132

(44.36) 163

2.99 0.69 731 0.83 0.10 0.07

(9.52) (0.43) (1,802) (0.28) (0.22) (0.17)

n.a.

162.28

(1,599.48)

n.a.

47.82

(56.23) 132

a. For 1994, estimates of affiliates’ sales, value added, and employee compensation in 2002 dollars were not derived because of the difficulties of accounting for changes in exchange rates from 1994 to 2002. n.a. = Not available.

Maria Borga

149

Table 5. Selected Characteristics of MNCs in the Panel by Industry of the Parent: Mean Values of Selected Variables, 1994 and 2002 1994 Characteristics MNCs with parents in manufacturing Parent share of employment Parent share of salesa Parent share of value addeda Share of employment in low-income countries Affiliated goods imports as a share of U.S. parent purchases Unaffiliated goods imports as a share of U.S. parent purchases Affiliated services imports as a share of U.S. parent purchases Number of observations MNCs with parents in services Parent share of employment Parent share of salesa Parent share of value addeda Share of employment in low-income countries Affiliated goods imports as a share of U.S. parent purchases Unaffiliated goods imports as a share of U.S. parent purchases Affiliated services imports as a share of U.S. parent purchases Number of observations MNCs with parents in “other” industries Parent share of employment Parent share of salesa Parent share of value addeda Share of employment in low-income countries Affiliated goods imports as a share of U.S. parent purchases Unaffiliated goods imports as a share of U.S. parent purchases Affiliated services imports as a share of U.S. parent purchases Number of observations

2002

Mean

Standard deviation

0.76 n.a. n.a.

(0.20) n.a. n.a.

0.71 0.72 0.77

(0.22) (0.19) (0.24)

0.20

(0.31)

0.26

(0.32)

0.046

(0.10)

0.059

(0.13)

0.038

(0.10)

0.038

(0.10)

(0.01)

0.0004

0.001

Standard deviation

Mean

822

(0.002) 822

0.83 n.a. n.a.

(0.17) n.a. n.a.

0.77 0.78 1.00

(0.20) (0.18) (2.50)

0.13

(0.25)

0.22

(0.29)

0.004

(0.03)

0.005

(0.04)

0.005

(0.03)

0.003

(0.03)

(0.09)

0.007

0.017 163

(0.04) 163

0.79 n.a. n.a.

(0.23) n.a. n.a.

0.75 0.75 0.74

(0.26) (0.26) (0.30)

0.33

(0.42)

0.32

(0.39)

0.007

(0.03)

0.014

(0.07)

0.009

(0.03)

0.009

(0.06)

(0.05)

0.008

0.008 132

(0.03) 132

a. For 1994, estimates of affiliates’ sales, value added, and employee compensation in 2002 dollars were not derived because of the difficulties of accounting for changes in exchange rates from 1994 to 2002. n.a. = Not available.

150


3.8 percent. The share of purchases accounted for by services imported from their affiliates fell, from 0.1 percent to 0.04 percent. The share of purchases accounted for by goods imported from affiliates increased for parents in “other” industries, from 0.7 percent to 1.4 percent. For parents in “other” industries, the share accounted for by imports of goods from unaffiliated parties and by imports of services from affiliates did not change. For parents in services, the share of purchases accounted for by services imported from their affiliates fell from 1.7 percent in 1994 to 0.7 percent in 2002.

Decomposition of Changes in Parent Employment To provide a starting point for a more formal analysis of, and a framework for, examining the impact of changes in offshoring behavior on employment at U.S. parent companies, the data reported by U.S. parent companies are used to apportion the actual change in U.S. parent employment among three factors: the change in output, the change in labor productivity, and the change in the use of purchased inputs in production. This framework allows changes in sourcing behavior, which are central to the issue of offshoring, to be viewed in isolation from other sources of change in employment and allows its importance relative to that of the other two factors to be gauged. U.S. parent employment can be expressed as an identity involving output, labor productivity, and the share of production performed in-house S(VA/S) E = —————— (VA/E) where E = employment, S = output (as measured by sales), and VA = value added. Output can be expressed as the sum of value added and purchased inputs, P, or, S = VA + P. Substituting this expression into the identity above yields S(1 – P/S) E = —————— (VA/E) Thus, employment is positively associated with changes in output and negatively associated with changes in the share of purchased inputs in output and with changes in labor productivity. Although each variable has determinants of its own, which would have to be investigated in seeking the root causes of

Maria Borga

151

Table 6. Decomposition of Changes in U.S. Parent Companies Employment between 1994 and 2002 Employment change

Share of total employment change

All U.S. parents Total change in employment Due to changes in output Due to changes in labor productivity Due to changes in the use of purchased inputs Residual

703.0 3,648.1 –1,436.6 –903.8 –604.7

5.19 –2.04 –1.29 –0.86

U.S. parents that lost employment Total change in employment Due to changes in output Due to changes in labor productivity Due to changes in the use of purchased inputs Residual

–1,972.3 –657.7 –825.6 –601.5 112.5

–0.33 –0.42 –0.31 0.06

U.S. parents that gained employment Total change in employment Due to changes in output Due to changes in labor productivity Due to changes in the use of purchased inputs Residual

2,675.3 4,305.8 –611.0 –302.3 –717.2

1.61 –0.23 –0.11 –0.27

changes in employment, they provide a framework for an initial examination of the changes in employment at U.S. parents. Table 6 attributes the change in parent employment to each variable, both for all parents and, separately, for parents that gained employment and parents that lost employment. In the decomposition, each variable is allowed in turn to change as it did between 1994 and 2002, while holding the other two variables constant at their 1994 levels. The residual is the portion of actual employment change not explained by the three calculations described above. The residual is due to the finite, rather than infinitesimal, changes in the reported data and can be viewed as reflecting the net interactions among the three variables. For all parents, employment increased by 703,040 employees. Increased output more than accounted for the increase in employment. Increases in labor productivity and in the use of purchased inputs had negative, and much smaller, impacts on parent employment. Parents that lost employees had a total decrease of 1,972,305 employees. The increase in labor productivity was the most important of the three variables, accounting for 42 percent of the total decline in employment, followed by the loss of output (33 percent) and by the increased

152


use of purchased inputs (31 percent).11 Those parents that gained employees added 2,675,345 employees. The increase in their output more than accounted for the increased employment. The increase in labor productivity had a much larger negative impact on their employment than the increase in the use of purchased inputs. Overall, increased reliance on purchases by U.S. parents, from 60 percent to 63 percent of their sales, reduced their workforce by 903,800 employees between 1994 and 2002. Because most of the inputs purchased by parents are supplied by domestic firms, rather than imported, the loss of jobs due to the increased use of imports is less than that. Between 1994 and 2002, the share of purchases accounted for by imports of goods from affiliates increased about one percentage point, from 3.5 percent to 4.6 percent. Thus, the increased use of imports of goods from affiliates accounted for about one-third of the total increase in purchases. The share of unaffiliated imports of goods was flat, and the share of affiliated imports of services—small to begin with—fell slightly, so the use of these imports had little impact on the change in parent employment.

Correlations Table 7 shows the correlations between changes in selected variables. The change in parent employment was significantly positively correlated with the change in output, as measured by sales, and significantly negatively correlated with the change in labor productivity, as measured by value added per employee. The change in parent employment was negatively correlated with the change in the ratio of purchases to sales, but this correlation was not significant at the 10 percent level. However, a change in the share of goods imported from affiliates in U.S. parent purchases was significantly negatively correlated with a change in employment at the parent. Therefore, while a change in the use of purchased inputs overall was not significantly associated with a change in employment, there was a significant association between a change in parents’ use of goods imported from their foreign affiliates and its employment. In contrast, a change in the use of goods imported from unaffiliated parties was significantly positively correlated with employment change. Changes in the use of services imported from affiliates were not significantly correlated with changes in parent employment. 11. The residual term was positive and accounted for 6 percent of the entire change in employment.

Maria Borga

153

A change in the ratio of purchases to sales was significantly negatively correlated with a change in the use of goods imported from affiliates. This indicates that when parents increased their reliance on purchased inputs they tended to increase their reliance on inputs from outside the firm more than on goods purchased from their foreign affiliates. Conversely, it suggests that when parents reduced their reliance on purchased inputs, they continued to rely on imports of goods from their foreign affiliates and reduced their reliance on inputs from other sources. There was a significant positive correlation between changes in parent employment and changes in affiliate employment, reflecting a close link between growth at affiliates and growth at their parents. The growth in employment at both the parents and their affiliates could be due to mergers and acquisitions as well as to the expansion of existing operations. This finding suggests that a complementary relationship exists between a U.S. parent’s employment and its affiliates’ employment.12 The change in the share of affiliate employment in low-income countries was also positively correlated with the change in parent employment, suggesting that the expansion of operations in low-income countries is associated with gains, not losses, in employment at the parent. The change in affiliate employment was significantly positively correlated with the change in the share of sales to the United States and significantly negatively correlated with the share of sales sold locally. Changes in the share of affiliates’ sales to the local market were significantly negatively correlated with changes in their parents’ sourcing of inputs from them. Changes in compensation per employee at parents were highly negatively correlated with changes in parent employment, and highly positively correlated with changes in labor productivity, confirming that those parents with the greatest employment losses are those with the greatest compensation and productivity per employee increases. It is interesting to note that changes in the use of purchased inputs were significantly negatively correlated with changes in value added per employee. If firms were choosing to purchase inputs that they were relatively inefficient at producing, then an increase in purchased inputs would result in an increase in labor productivity. However, there was no evidence of this in the sample data set. There was a significant positive correlation between an increased share of imports of goods from affiliates in parent purchases and increased compensation and value added per employee, indicating that employees of firms that used inputs from their affiliates more intensively tended to 12. For a discussion of the complex issue of whether U.S. parent and foreign affiliate labor are substitutes or complements and a formal statistical analysis, see Brainard and Riker (1997).

154


Table 7. Correlations between Changes in Selected Variables between 1994 and 2002 Change in:

Change in: Parent employment R&D intensity Real compensation per employeea Real value added per employeea Real salesa Share of services Ratio of purchases to sales Affiliate employment Local sales Sales to the United States Affiliates’ share of services Employment in low-income countries Affiliated goods imports Unaffiliated goods imports Affiliated services imports

Parent employployment

R&D intensity

1.00 0.02 (0.55)b –0.19** (0.0001) –0.08** (0.01) 0.23** (0.0001) –0.03 (0.33) –0.05 (0.12) 0.17** (0.0001) –0.01 (0.74) –0.01 (0.82) –0.02 (0.47) 0.06* (0.06) –0.12** (0.0001) 0.06** (0.04) –0.04 (0.23)

1.00 0.13** (0.0001) –0.05 (0.10) –0.01 (0.78) –0.004 (0.90) 0.02 (0.49) 0.01 (0.63) –0.08** (0.01) –0.001 (0.98) –0.01 (0.78) 0.04 (0.22) 0.11** (0.0003) 0.01 (0.65) –0.02 (0.58)

Real Real compen- value sation added per em- per employeea ployeea

1.00 0.43** (0.0001) –0.01 (0.72) 0.05 (0.12) –0.17** (0.0001) 0.04 (0.15) –0.07** (0.02) 0.03 (0.37) 0.03 (0.29) 0.02 (0.47) 0.08** (0.01) 0.02 (0.49) 0.09** (0.003)

Real salesa

Share of services

1.00 0.05* (0.08) 0.02 (0.60) –0.57** (0.0001) 0.02 (0.50) –0.05 (0.12) 0.01 (0.69) –0.03 (0.31) 0.05* (0.07) 0.09** (0.002) –0.01 (0.76) 0.01 (0.77)

1.00 –0.01 (0.69) 0.05 (0.13) 0.20** (0.0001) –0.06** (0.04) –0.01 (0.83) –0.04 (0.15) 0.03 (0.28) –0.01 (0.66) –0.02 (0.48) 0.01 (0.79)

1.00 –0.13** (0.0001) –0.03 (0.14) –0.001 (0.99) –0.01 (0.66) 0.53** (0.0001) –0.02 (0.42) –0.0001 (0.99) –0.01 (0.69) 0.03 (0.29)

* Significant at the 10 percent level. ** Significant at the 5 percent level. a. Rough estimates of U.S. parents’ sales, value added, and employee compensation in 2002 dollars were derived using the U.S. GDP deflator. b. Standard deviations are in parentheses.

become more productive and received higher compensation than employees of other firms. Finally, changes in the share of parent purchases accounted for by imports of services from affiliates were significantly positively correlated with changes in compensation per employee. Changes in R&D intensity were significantly positively correlated with changes in compensation per employee at parents but not with value added per employee at parents. Changes in R&D intensity were also significantly posi-

Maria Borga

Ratio of purchases to sales

Affiliate employment

155

Local sales

Sales to the United States

EmployAffiliment in Unaffiliates’ low- Affiliated ated Affiliated share of income goods goods services services countries imports imports imports

1.00 0.02 1.00 (0.61) –0.04 –0.10** 1.00 (0.14) (0.001) 0.07** 0.07** –0.48** 1.00 (0.02) (0.03) (0.0001) –0.08** –0.05* 0.02 0.01 (0.01) (0.07) (0.49) (0.77) 0.02 0.17** –0.04 0.11** (0.57) (0.0001) (0.18) (0.0003) –0.10** 0.07** –0.12** 0.15** (0.001) (0.01) (0.0001) (0.0001) –0.01 0.01 0.04 –0.02 (0.75) (0.81) (0.15) (0.58) –0.02 0.02 –0.13** –0.17** (0.50) (0.43) (0.0001) (0.0001)

1.00 0.01 1.00 (0.78) –0.004 0.01 (0.88) (0.79) –0.01 0.02 (0.71) (0.51) 0.07** –0.03 (0.02) (0.27)

1.00 0.03 (0.33) 0.05 (0.11)

1.00 0.01 (0.78)

1.00

tively correlated with changes in the share of parents’ purchases accounted for by affiliated imports of goods. This finding concurs with many studies of intrafirm trade that have found higher R&D intensity to be associated with greater intrafirm trade due to concerns about protecting intellectual property.13 13. See, for example, Lall (1978); Buckley and Pearce (1979); and Andersson and Fredriksson (2000).

156


Conclusions A panel data set consisting of U.S. parent companies and their foreign affiliates was used to examine evidence on the extent of offshoring by U.S. parent companies to their foreign affiliates, and then to determine if this offshoring is associated with changes in employment at U.S. parents between 1994 and 2002. The data were used to apportion the changes in U.S. parents’ employment to three sources: changes in output, changes in labor productivity, and changes in their reliance on purchased inputs. This decomposition showed that changes in output and in labor productivity had relatively larger impacts on parents’ employment than changes in the use of purchased inputs. Correlations between changes in employment at U.S. parents and changes in various characteristics of the parents, their affiliates, and the MNC were examined. It was found that parents have increased their reliance on purchased inputs, but there was no significant correlation between changes in their reliance on purchased inputs and changes in the size of their workforce. Indeed, the most significant factors affecting parent employment were found to be changes in output and in labor productivity. While there was no significant association between changes in parents’ reliance on purchased inputs and changes in their employment, there was a significant negative correlation between changes in the share of parent purchases accounted for by imports of goods from their affiliates and changes in their employment. However, the vast majority of U.S. parents’ purchased inputs are acquired from domestic sources, not imports. It was found that the fortunes of parents and their foreign affiliates were closely linked, with changes in parents’ employment being positively correlated with changes in employment at their affiliates. Overall, the majority of foreign affiliate sales were local, indicating the importance of market access in the decision to invest abroad. The share of sales sold locally increased between 1994 and 2002, indicating that market access may be becoming an increasingly important reason for investing abroad. Finally, the increase in the share of local sales coincided with an increase in the share of employment in low-income countries, suggesting that market access, and not just factor cost differences, is an important reason for investing in these countries. One drawback of the approach used in the paper is that each factor has been examined individually. The correlations between changes in the variables measure the associations between any two of the variables. However, it is not possible to say with any certainty whether any of those associations found to be significant would still be if other variables were taken into account. Likewise, it is

Maria Borga

157

possible that variables examined above that showed no significant relationship would become significant if other variables were taken into account. Thus, an obvious avenue for future research would be to adopt a more rigorous statistical framework for analyzing the data that would allow more definitive conclusions to be drawn. Another avenue for future research would be to investigate the impact that industry-specific factors and mergers and acquisitions have had on the changes in U.S. parents’ employment. In addition it would be useful to link BEA’s data on direct investment with BEA’s data on imports of services from unaffiliated foreigners. Linking these two data sets would allow a more complete portrait of the direct imports by U.S. parents to be drawn.

Appendix: The Data Set BEA’s financial and operating data provide a picture of the overall activities of foreign affiliates and U.S. parent companies using a wide variety of indicators of their financial structure and operations. These data cover items that are needed in analyzing the characteristics, performance, and economic impact of MNCs, such as sales, value added, employment, compensation of employees, and exports and imports.

Definitions A U.S. parent is defined as a U.S. resident that has a 10 percent or more ownership interest in a foreign business enterprise, where U.S. resident is defined in the broad legal sense to include individuals, business enterprises, trusts, and other entities. However, most U.S. parents are businesses. A foreign affiliate is any foreign business enterprise in which there is a U.S. direct investment. Employment is defined as the number of full-time and part-time employees on the payroll at the end of the company’s fiscal year. Employee compensation consists of wages and salaries and employee benefits. Value added is measured as the sum of costs incurred and the profits earned in production. The costs incurred fall into four categories: compensation of employees, net interest paid, indirect business taxes, and capital consumption allowance. Sales, or gross operating revenues, are disaggregated into goods, services, and investment income. Sales of goods are those typical of establishments in any of the following NAICS sectors: agriculture, forestry, fishing, and hunting

158


(except support activities for agriculture and forestry); mining (except support services); construction; manufacturing; wholesale trade; and retail trade. Sales of services are defined as those typical of establishments in the other fourteen NAICS sectors and the support activities for agriculture, forestry, fishing, hunting, and for mining.14 Information on investment income was collected to ensure that if it were included in sales or gross operating revenues it would not also be included in sales of services. Sales are also disaggregated into six destinations: the host country, the United States, third countries, and then further for each of these, into sales to either affiliated or unaffiliated parties. Expenditures for research and development (R&D) measure expenditures for R&D conducted by parents, whether for themselves or for others under contract. These expenditures exclude those for R&D conducted by others for the parents under contract.

Import Data The data on trade in goods collected on BEA surveys are generally comparable to the concepts and definitions used by the Census Bureau in compiling the data on U.S. trade in goods. Data on the imports of goods from both affiliated and unaffiliated parties are reported on forms covering the activities of the U.S. parent. In addition, trade data are reported on forms covering the activities of majority-owned foreign affiliates, including imports of goods to their U.S. parents. The data on parents’ imports from affiliated parties used in this paper are from the U.S. parents’ reports.15 BEA collects data on imports of services by U.S. parents from their foreign affiliates on its quarterly surveys of transactions between U.S. parents and their foreign affiliates. Because of the difficulties of matching data collected on those surveys with data on the 1994 benchmark and 2002 annual surveys (which were 14. For 1994, sales were divided into goods or services based on establishments using industry classifications derived from the 1987 Standard Industrial Classification system (SIC). For that year, sales of services were defined as those typical of establishments in the following SIC-based industry categories: services; finance (except depository institutions), insurance, and real estate; agriculture, mining, and petroleum services; and transportation, communication, and public utilities. 15. The only difference between the two data series is the inclusion of imports from minorityowned foreign affiliates in the data reported on the U.S. parents’ forms. These tend to be relatively small in value. For example, in 2002, imports from minority-owned foreign affiliates were estimated to be $11.7 billion while imports of goods from majority-owned foreign affiliates were estimated to be $171.6 billion.

Maria Borga

159

Table A-1. Imports of and Sales of Services to U.S. Parent Firms from Affiliates, 1994 and 2002 Millions of dollars

Total U.S. parent imports of other private services from their foreign affiliates Total sales of services to U.S. parents by majority-owned foreign affiliates Sales of services to U.S. parents by majority-owned foreign affiliates in panel data set

1994

2002

6,538

17,006

6,955

15,272

5,152

12,566

the source of the data used in this paper), data on imports of services from foreign affiliates were not used. Instead, the data on sales of services by majorityowned foreign affiliates to their U.S. parents from the annual and benchmark surveys were used. While there are some differences in definitions and coverage between the two data series, they are closely related.16 The first two rows of table A-1 compare the two data series, from BEA’s aggregate data, for the years 1994 and 2002. One concern about the composition of the panel used in this paper is that by excluding U.S. parents that entered the universe after 1994, the panel excluded parents that decided to invest abroad because of advances in technology (for example, better and cheaper telecommunications) that allow them to rely on services produced at their affiliates. To check this possibility, the last row of table A-1 shows the sales of services to U.S. parents by majority-owned foreign affiliates included in the panel data set. Comparing rows 2 and 3, the share of sales of services to U.S. parents accounted for by foreign affiliates included in the panel increased from 74 percent in 1994 to 82 percent in 2002. Thus, the panel does not appear to be missing large numbers of new entrants that rely to a greater extent on services from their affiliates than existing firms. Comparison of the sample data set with the population of firms. Differences due to sample selection can be identified by comparing the aggregate values for the sample data set with the aggregate values for the population of U.S. direct

16. An example of a difference in coverage is that the data on imports from affiliates include all foreign affiliates while those on sales of services include only majority-owned foreign affiliates. An example of a difference in definition is that the data on sales of services include settlement transactions between affiliated telecommunications carriers, but these transactions are excluded from the affiliated imports of services data.

160


Table A-2. Aggregate Values of Selected Characteristics for the Panel Data Set and the Population, 1994 and 2002 1994 Characteristics U.S. parent variables Employment (in thousands) Sales (millions of dollars)b Employment change (percent) Ratio of purchases to sales Share of services in sales International share of sales R&D intensity Value added per employeeb (thousands of dollars) Compensation per employeeb (thousands of dollars) Affiliate variables Employment (in thousands) Share of services in sales Share of sales to local market Share of sales to third countries Share of sales to U.S. MNC variables Parent share of employment Parent share of sales Parent share of value added Share of affiliate employment in low-income countries Affiliated goods imports as a share of parent purchases Unaffiliated goods imports as a share of parent purchases Affiliated services imports as a share of parent purchases

2002 Sample

Population a

15,059.9 3,885,676 … 0.65 0.30 0.13 0.026

11,118.3 3,563,318 0.07 0.70 0.31 0.13 0.028

17,531.0 5,130,846 0.16 0.69 0.38 0.11 0.026

89.56

91.38

97.00

91.66

56.80

56.19

61.51

57.65

3,894.9 0.13 0.67 0.23 0.10

5,153.3 0.13 0.67 0.23 0.10

5,104.9 0.16 0.62 0.27 0.11

7,138.2 0.18 0.62 0.27 0.11

0.73 0.70 0.73

0.75 0.71 0.75

0.69 0.65 0.70

0.71 0.69 0.74

0.34

0.35

0.40

0.39

0.059

0.047

0.056

0.046

0.038

0.037

0.033

0.033

0.003

0.003

0.004

0.004

Sample

Population

10,415.3 2,656,980 … 0.65 0.28 0.11 0.031

a

a. Aggregate values shown exclude U.S. parents in wholesale and retail trade because those parents were excluded from the sample data set. b. Rough estimates of U.S. parents’ sales, value added, and employee compensation in 2002 dollars were derived using the U.S. GDP deflator.

investment abroad. Table A-2 shows the values of selected variables for the sample data set and the population of US MNCs. The percentage change in aggregate employment of U.S. parents for the sample was about half that for the population as a whole, partly because the employment of new entrants to the population exceeded that of those parents that

Maria Borga

161

Table A-3. Aggregate Values of Selected Characteristics for MNC Deaths for 1994 and Births for 2002 Deaths (values in 1994)

Births (values in 2002)

3,740.5 3,740,504 0.63 0.31 0.10 0.020 76.61 45.32

5,913.9 1,355,504 0.70 0.41 0.06 0.020 68.79 43.18

Affiliate variables Employment (in thousands) Services as a share of sales Share of sales to local market Share of sales to third countries Share of sales to the United States

901.8 0.11 0.70 0.21 0.09

1,469.0 0.28 0.68 0.23 0.09

MNC variables Parent share of employment Parent share of sales Parent share of value added Share of affiliate employment in low-income countries Affiliated goods imports as a share of parent purchases Unaffiliated goods imports as a share of parent purchases Affiliated services imports as a share of parent purchases

0.81 0.74 0.79 0.25 0.030 0.037 0.003

0.80 0.82 0.86 0.34 0.023 0.052 0.002

Characteristics U.S. parent variables Employment (in thousands) Salesa (millions of dollars) Ratio of purchases to sales Services as a share of sales International share of sales R&D intensity Value added per employeea (thousands of dollars) Compensation per employeea (thousands of dollars)

a. Rough estimates of U.S. parents’ sales, value added, and employee compensation in 2002 dollars were derived using the U.S. GDP deflator.

dropped out. Table A-3 presents the aggregate values for 1994 for deaths and the aggregate values for 2002 for births.17 One issue that arises with a data set consisting only of parents that responded to both the 1994 and 2002 surveys of U.S. direct investment abroad is that parents that dropped out of the universe (deaths) or entered it (births) between 1994 and 2002 are excluded. The exclusion of deaths raises the possibility of survivor bias. However, it must be emphasized that the death of a parent due to the liquidation of the firm is rare. The more 17. Data from BEA’s 2002 preliminary estimates were used in constructing the sample data set. Some parents’ responses were still being edited at the time the panel data set was assembled, and these parents were dropped from the sample even though they were in the universe in both 1994 and 2002. Therefore the aggregates from the sample plus the aggregates from deaths or births do not equal the population aggregates.

162


common causes are that the parent is acquired by another U.S. parent company (in which case the first parent is included in the data but is now consolidated with the acquiring company); the parent is acquired by a foreign company that chooses to hold the foreign affiliates directly; the parent sells off its foreign operations; or, the foreign operations of the parent fall below BEA’s reporting requirements. For most of the other parent variables, the sample and the population characteristics are similar. One difference between the sample and the population is the share of parents’ sales accounted for by services, which is lower for the sample in 1994 and 2002, indicating an underrepresentation of services firms. The affiliates of parents included in the sample account for 76 percent of all affiliate employment in 1994, falling to 72 percent by 2002. The sample and the population are similar for most of the affiliate and MNC characteristics. One difference is that imports of goods from foreign affiliates account for a larger share of purchased inputs for parents in the sample than for the population, indicating an overrepresentation in the sample of parents that source inputs from their affiliates. Comparison of the aggregate and mean values for the panel data set. Comparing the aggregate values for the sample in table A-2 and the mean values from the sample in table 1 identifies differences that are due to the greater weight given to larger firms in the aggregate values. The unweighted data show much higher average employment growth, at 39 percent, than the aggregate value because the employment of smaller parents tended to grow faster than that of larger firms. In addition, smaller firms used purchased inputs less intensively than larger firms. The foreign affiliates of smaller parents tend to focus more on selling to the local market and less on exporting from the host country, and this tendency increased between 1994 and 2002. For the MNC as a whole, the data indicate that smaller parents have less extensive foreign operations than larger parents, and these operations tend to be more concentrated in high-income countries. In addition, smaller parents use imports of goods less intensively than larger firms.

Maria Borga

163

References Andersson, Thomas, and Torbjorn Fredriksson. 2000. “Distinction between Intermediate and Finished Products in Intra-firm Trade.” International Journal of Industrial Organization 18 (July): 773–92. Brainard, Lael S., and David A. Riker. 1997. “Are U.S. Multinationals Exporting U.S. Jobs?” Working Paper 5958. Cambridge, Mass.: National Bureau of Economic Research. Buckley, Peter J., and R. D. Pearce. 1979. “Overseas Production and Exporting by the World’s Largest Enterprises: A Study in Sourcing Policy.” Journal of International Business Studies 10 (March): 9–20. Government Accountability Office (GAO). 2004. “Current Government Data Provide Limited Insight into Offshoring of Services.” Washington: GAO-04-932 International Trade. Helpman, Elhanan, and Paul R. Krugman. 1985. Market Structure and Foreign Trade. Cambridge, Mass.: MIT Press. Lall, Sanjaya. 1978. “The Pattern of Intra-firm Exports by U.S. Multinationals.” Oxford Bulletin of Economics and Statistics 40 (3): 209–22. Markusen, James R. 2002. Multinational Firms and the Theory of International Trade. Cambridge, Mass.: MIT Press U.S. Bureau of Economic Analysis. 1998. U.S. Direct Investment Abroad: 1994 Benchmark Survey, Final Results, U.S. Department of Commerce. ———. 2004. “A Note on the Patterns of Production and Employment by U.S. Multinational Companies.” Survey of Current Business (March): 52–56. ———. 2005. U.S. Direct Investment Abroad: Operations of U.S. Parent Companies and Their Foreign Affiliates, Preliminary 2002 Estimates. U.S. Department of Commerce.


D E S I R É E VA N W E L S U M X AV I E R R E I F Organization for Economic Cooperation and Development

Potential Offshoring: Evidence from Selected OECD Countries

R

apid advances in information and communication technologies (ICTs), combined with continuing efforts to liberalize international trade and investment in services, have increased the tradability of services and created new types of tradable services. This, in turn, has led to a new wave of globalization in the services sector, with the offshoring of particular types of services activities now becoming increasingly common, as it has been for many years in manufacturing. New technological developments now allow many service activities to be carried out regardless of their geographic location, and their production and delivery no longer have to take place in the same location. As a result, many white-collar jobs that were shielded from international competition now face competition from abroad. Despite the widespread media attention given to the apparent offshoring of service sector jobs, little is known about the extent of this phenomenon, or the extent to which it is related to other economic and structural developments. This paper draws on and extends a previous detailed analysis of occupational data for

The opinions expressed and arguments employed in this paper do not necessarily reflect the official views of the organization or of the governments of its member countries. This paper draws on a larger body of work and a longer paper entitled “The Share of Employment Potentially Affected by Offshoring—An Empirical Investigation,” which will be published by the OECD (www. oecd.org/sti/offshoring). We are grateful to Nigel Pain of the OECD Economics Department and Ron Smith of Birkbeck College at the University of London for their help and advice in preparing this paper, as well as our colleagues in the Directorate for Science, Technology and Industry, in particular Graham Vickery; in the Directorate for Employment, Labour, and Social Affairs; and in the Trade Directorate. Finally, our thanks go to participants in the conferences at which earlier parts of this work have been presented, in particular Catherine Mann and Robert Lawrence.

165

166


selected OECD countries that sought to determine the share of total employment that could potentially be affected by the international sourcing of IT and ICTenabled services (van Welsum and Vickery 2005a). Including both the low- and the high-skill white-collar occupations potentially affected by global services sourcing, that analysis suggested that close to 20 percent of total employment could potentially be affected by offshoring. The work also found that sectors such as business services (for example, accounting and consulting), financial services, and research and development have a relatively high share of such employment. It is important to keep in mind that “potentially affected by offshoring” refers to activities that could be coming into a country as well as those that might leave a country. Incoming offshored services activities would bring about an increase in the share of employment potentially affected by offshoring, whereas services activities that leave a country would bring about a relative decline in the share. This paper takes this analysis one step further by examining the relationship between the share of employment potentially affected by offshoring and other economic and structural developments, using some simple descriptive regressions on a panel of OECD economies between 1996 and 2003. In particular, first estimates are provided of the statistical association between the share of employment potentially affected by service sector offshoring, trade in business services, and foreign direct investment.The analysis in this paper does not find any systematic evidence to support the popular belief that net outward investment or imports of business services are associated with significant declines in the share of employment potentially affected by offshoring, at least at the aggregate level. Exports of business services are found to have a positive statistical association with the share of employment potentially affected by offshoring, suggesting that increases in demand and production have also raised demand for these types of ICT-using occupations. Other factors positively associated with the share of employment potentially affected by offshoring are found to be the comparative size of the service sector, the growing share of ICT investment in total fixed investment, and human capital. Although there are no direct official data measuring the extent of offshoring, it is commonly believed that it has the potential to grow substantially beyond its current relatively small level. This paper aims to contribute to the debate surrounding offshoring by looking in detail at some of the trade in services and employment data that may reveal further insights about its current extent, as well as by performing a simple descriptive econometric analysis of the factors statistically associated with movements in the aggregate share of employment that could potentially be affected by offshoring.

Desirée van Welsum and Xavier Reif

167

The paper is organized as follows. The next section looks at what can be learned about offshoring on the basis of data on trade in services and occupational employment data, respectively. The subsequent section sets out the simple descriptive empirical model employed to examine what factors are associated with the share of employment potentially affected by offshoring and gives some initial results. The final section offers concluding remarks.

The Extent of Potential Offshoring Offshoring includes both international outsourcing (where activities are contracted out to independent third parties abroad) and international insourcing (to foreign affiliates). The cross-border aspect is the distinguishing feature of offshoring—that is, whether services are sourced within the domestic economy or abroad—not whether they are sourced from within the same company or from external suppliers (outsourcing). Offshoring is often confused with outsourcing, but only a part of offshoring consists of outsourcing. Offshoring is also often interpreted as referring to the purchase of intermediate services, even though the distinction between final and intermediate services is a difficult one to make in some instances and may not be very meaningful for certain types of services. To date there are no official data measuring the extent of offshoring, so it is necessary to use indirect measures such as data on trade in services, employment data, input-output tables, and trade in intermediates. Evidence from company surveys can also be a useful complement.1 However, while the offshoring of services activities should result in a flow of trade in services, not all trade in services is related to offshoring, and it is not possible to distinguish which part of it is. Similar problems apply to the analysis of foreign direct investment (FDI) because it is not possible to determine what share of FDI is directly related to offshoring. There are also no official and internationally comparable data on changes in employment resulting from the offshoring of services activities. This paper uses both trade and employment data to examine what is known about the extent of services offshoring and what can be said about its potential. Trade Data Trade in services provides one possible proxy for measuring ICT-enabled services offshoring. Because offshoring of activities takes place between residents of 1. See, for example, Marin (2004).

168


different countries, some of it should result in a flow of trade in services, exports from the country receiving the offshored activities, and imports for the country from which the offshoring originates. OECD (2004a) and van Welsum and Vickery (2005a) examine exports of services, while Schultze (2004) and van Welsum (2004) analyze offshoring and imports of services. It is important to bear in mind, though, that not all trade in services is related to offshoring, and it is not possible to identify the share of trade in services that is directly related to offshoring. The extent of international trade in IT and ICT-enabled business process services in international statistics is approximated by summing the IMF balance of payments categories “computer and information services” and “other business services” (see table A-1 for details on which services are included in these categories). These data contain information on international outsourcing and international insourcing combined, but it is not possible to identify the proportion of this trade that results directly from offshoring. Data on computer and information services are not available for all countries. For some, such as India, they are included under “other business services,” along with other services.2 The “other business services” category may have variable shares of IT and ICT-enabled services in different countries. Moreover, the data are reported in current U.S. dollars and will be affected by currency movements. Most exports of other business services and computer and information services still originate in OECD countries, although their share is slowly declining (from 80.3 percent of the total reported value in 1995 to 79.1 percent in 2002).3 The twenty countries that accounted for the largest-value shares in 2003, as well as some selected other economies, are shown in figure 1. OECD countries have the top seven shares of these services exports; China is in tenth and India is in fourteenth position. Nevertheless, some nonmember developing economies are experiencing rapid growth in exports (see figure 2), although most are starting from very low levels. Only Ireland is among the ten countries with the largest shares (in 2003) and the fastest export growth rates (1995–2003). The average annual growth rate of exports and imports of other business and computer and information services (in current U.S. dollars) over the period 2. For India, the category “other business services” includes all services except travel, transport, and government services. However, Indian firms are now extensively exporting ICT-enabled services and business process services, and the remaining services included in the category are likely to be small in comparison. Furthermore, data on overseas revenues from annual reports of top Indian export firms show patterns similar to the IMF data. 3. The share of some services-exporting countries may be understated because the data on trade in services they report to the IMF are not very accurate. Furthermore, other countries that export services may not be members and report to the IMF.


169

1995–2003 is given in figure 2 (the top twenty of selected countries). It shows that many of the countries often mentioned in the offshoring debate (such as India, China, and Brazil, but also eastern European countries such as Estonia) have indeed experienced rapid growth of these exports, which may confirm their emergence as offshoring locations in recent years. However, some of these countries are growing from a very low level, and some of the rapid growth is explained by their economic development. Furthermore, many countries experiencing rapid exports growth are also seeing rapid growth of their imports of these services (fifteen countries are among the twenty countries with the fastest growth of both exports and imports of these services). The trade balance (in current U.S. dollars) in the sum of the categories “other business” and “computer and information services” as a percentage of GDP is shown in figure 3. The balance share, though positive, is smaller and more stable in the United States than in the United Kingdom and Denmark. The share in the United Kingdom is increasing, in spite of the impression that may be given by the many reports, media reports in particular, on the extent of offshoring and related imports. Surprisingly, the deficit in Ireland is quite large. It remains difficult, however, to interpret these data and link them to different sourcing activities. It is not possible to tell what share of this trade results from international sourcing activities. Offshoring can include unaffiliated trade in services (from international outsourcing) and affiliated trade (from international insourcing), but some of it is also related to foreign direct investment and temporary migration, mode 4 trade in services (the temporary movement of natural persons) under the General Agreement on Trade in Services (GATS). But temporary migration is not captured by balance of payments trade data.4 Furthermore, the quality of the data may vary, and there can be very large discrepancies between reported exports and imports.5 Employment Data To get an idea of the “outer limits” of employment potentially affected by offshoring, van Welsum and Vickery (2005a) calculate the share of people employed who are performing mainly the types of functions that could potentially be carried out anywhere, using data on employment by occupation by industry. The classifications were not harmonized internationally, but the same 4. See van Welsum (2003) for a discussion. 5. See OECD (2004a, chap. 2) for an example using Indian data.

Figure 1. Share of the Value of Reported Total Exports of Other Business Services and Computer and Information Services, Selected Countries, 1995 and 2003a Percent in decreasing order of the total reported value share in 2003 1995

2002

14 12 10 8 6 4

Finland

Australia

Brazil

Korea

Singapore

Sweden

Canada

India

Austria

Spain

China

Japan

Ireland

Italy

Netherlands

France

Germany

United Kingdom

United States

2

Source: Authors’ calculations based on IMF balance of payments database (August 2005). a. The reported total for all countries does not necessarily correspond to a world total. For some countries, such as India, it is not possible to isolate other business services and computer and information services. As a consequence, for India the category includes total services, minus travel, transport, and government services (that is, including construction, insurance, and financial services as well as other business services and computer and information services).


171

Figure 2. Average Annual Growth in the Value of Exports and Imports of Other Business and Computer and Information Services, Selected Countries, 1995–2003a Compound annual growth rate Latvia Croatia Ireland Argentina Romania Lithuania Estonia India Peru Sweden China Brazil Israel Colombia Spain United Kingdom Morocco Norway Iceland Denmark 0

10

20

30 Exports

40

50

60

0

10

20

30 Imports

40

50

60

Latvia Estonia Lithuania Ireland Sweden Iceland Cyprus Switzerland Croatia India Brazil Denmark Ghana Morocco Spain United States Turkey Romania United Kingdom Israel

Source: Authors’ calculations based on IMF balance of payments database (August 2005). a. OECD member countries in dark shading. Data for Belgium, Luxembourg, and Mexico are not included.

Figure 3. Trade Balance in the Categories “Other Business Services” and “Computer and Information Services” as a Percentage Of GDP, Selected Countries, Various Years Percent 4.0 1995

2000

2001

2002

2003

3.0 2.0 1.0 0.0 –1.0 –2.0 –3.0 –4.0 –5.0

Source: Authors’ calculations based on IMF balance of payments database (February 2005).

ed

en

l ga rtu

s

ate

Sw

he et

Po

rla

nd

ur bo m N

xe Lu

s

g

ly Ita

nd la Ire

e ec re

an

d

ite

Un

n Ki

G

ce

d

om

gd

Fr

n

an nl Fi

ai

m

Sp

k ar

y en D

G

er

m

na

an

da

m Ca

iu lg

tri

a Be

us A

A

us

tra

lia

–6.0

Un

i

ted

St


173

methodology and rationale were applied to the individual country data sources.6 Because this analysis was carried out in order to obtain an order of magnitude on the share of people employed performing tasks that could potentially be carried out anywhere, no additional assumptions were made as to what proportion of each occupational group was actually likely to be affected by offshoring in practice. Thus the whole of each selected occupation was then included in the calculations. Occupations were selected by examining detailed occupational and task descriptions on the basis of the following four criteria, or “offshorability attributes”: (1) intensive use of ICTs, (2) an output that can be traded/transmitted enabled by ICTs, (3) high codifiable knowledge content, and (4) no face-to-face contact requirements. The occupational selections that resulted from this exercise are reported in the appendix (tables A-2 through A-5).7 This analysis, using occupational data for several OECD countries, suggests that around 20 percent of total employment carries out the kinds of functions that are potentially geographically footloose as a result of rapid technological advances in ICTs and the increased tradability of services, and could therefore potentially be affected by international sourcing of IT and ICT-enabled services.8 Nevertheless, because 6. The European data are Labour Force Survey data provided by Eurostat. The occupational classification system in those data is the ISCO, the International Standard Classification of Occupations; NACE, the industrial classification system of the European Union, is used for sectoral classification. For the United States, data from the Current Population Survey were used. The Current Population Survey collects information on both the industry and the occupation of the employed and unemployed. However, beginning with data from January 2003, the 1990 Census Industrial Classification System was replaced by one based on the North American Industry Classification (NAICS), and the 1990 Census Occupational Classification was replaced by one derived from the U.S. Standard Occupational Classification (SOC). Further information is available on the website of the U.S. Bureau of Labor Statistics at www.bls.gov/opub/hom/pdf/homch1.pdf [November 2004]: Chapter 1: Labor Force Data derived from the Current Population Survey. For Canada, labor force data provided by Statistics Canada were used. The occupational classification is in SOC91. For Australia, data from the Labour Force Survey provided by the Australian Bureau of Statistics were used. The occupational classification is in Australian Standard Classification of Occupations (ASCO), second edition. 7. For further details on the methodological background see van Welsum and Vickery (2005a), van Welsum and Vickery (2005b) and OECD (2004a). 8. Other studies have taken a similar approach. For example, Bardhan and Kroll (2003) produced estimates of 11 percent of total employment in the United States in 2001 as potentially affected by offshoring, and Forrester Research, as reported by Kirkegaard (2004), up to 44 percent of total employment. The differences in these estimates can be explained by the selection criteria that are applied to the occupational data. Thus Bardhan and Kroll (2003) only included occupations in which at least some offshoring was already known to have taken place yielding a more conservative estimate of the share of employment potentially affected, whereas the Forrester study used less detailed occupational categories resulting in a larger estimate of jobs potentially affected.

174


Figure 4. ICT-Intensive Using Occupations Potentially Affected by Offshoring as a Share of Total Employment: EU15, USA, Canada, and Australia, 1995–2003a Percent

19.5 19.0 18.5 18.0 17.5 17.0 EU15

USA

Canada

Australia

16.5

1995

1996

1997

1998

1999

2000

2001

2002

2003

Source: Authors’ calculations and van Welsum and Vickery (2005a), based on EULFS, U.S. Current Population Survey, Statistics Canada, and Australian Bureau of Statistics (2004/5). a. Includes estimates where a full data set was not available. Because of classification changes, the number for the United States for 2003 is also an estimate. There is a break in the data for Australia, with data for 1995 and 1996 in the ASCO first edition and subsequent data in ASCO second edition. Because of differences in classifications, the levels are not directly comparable.

classifications are not harmonized internationally, the country estimates are not directly comparable. The evolution over time of the share of employment potentially affected by offshoring is illustrated in figure 4. Even though the levels of these shares are not directly comparable, the evolution of the trends is interesting. The share of occupations potentially affected by offshoring in the EU15 increased from 17.1 percent in 1995 to 19.2 percent in 2003. For Canada it was more or less flat around 19.5 percent until 2001, after which it declined to 18.6 by 2003. For the United States the share declined by more than a percentage point from 19.2 percent in 1995 to 18.1 percent in 2002.9 In Australia, the share increased between 1997 and 2001, except in 1999, when it started to decline. Data for 1995 and 1996 were generated by a different classification system and therefore are not directly comparable. 9. The number for 2003 (just under 18 percent) is an estimate since both the occupational and industrial classification systems were changed in 2003 in the United States.


175

While it is difficult to draw inferences from these trends without further analysis, since the trends are affected by a multitude of factors, the evolutions shown in these trends are consistent with some casual observations on the ICTenabled offshoring that is taking place, such as Canada serving as an offshoring location, mainly from the United States, but less so more recently as other locations, such as India, have started to emerge. Similarly, Australia possibly also experienced competition for attracting, or keeping, activities that can be sourced internationally from India and other emerging locations in the region. Thus, the declining share in the United States, Canada, and Australia toward the end of the period could be consistent with the offshoring of IT-related and back-office activities (with some “potential offshoring” having become “actual offshoring”), for example, even though this is unlikely to account for all of the decline. Another possible explanation could be a differential pace of technological change with a relatively more rapid adoption and integration of new technologies, leading to relatively more jobs disappearing sooner as they become automated or digitized, or both.10 The increasing share for Europe is compatible with an overall increase in services employment as well as the finding from surveys that European firms tend to offshore within Europe.11 (At least one EU country, Ireland, is also a major destination for offshoring activities from the United States—IT-related activities in particular.) Other factors could also be important, such as cyclical developments and changes in labor supply and labor quality. The offshoring phenomenon does not necessarily have to result in a decline in services employment, though. Many existing services sectors have expanded, new services have emerged, and with ongoing technological developments and services trade liberalization it is likely yet more will be created. Furthermore, with the elasticity of demand of internationally traded services greater than one,12 rapid growth in countries such as India and China should also lead to reinforced exports from OECD countries. The offshoring phenomenon itself will also create new jobs in the domestic economy. However, certain types of occupations may experience slower growth than they otherwise might have. 10. A parallel can be drawn here with some of the work undertaken by Autor, Levy, and Murnane (2003) and Levy and Murnane (2004). These authors argue that the tasks most vulnerable to being replaced by technology are those where information processing can be described in rules. If a significant part of a task can be described by rules, this increases the likelihood of the task’s being offshored, since the task can then be assigned to offshore producers with less risk and greater ease of supervision. 11. See Millar (2002) and Marin (2004), for example. 12. See, for example, Pain and van Welsum (2004); van Welsum (2004); and Mann (2004).

176


There are several possible explanations for the changes in these trends, which are expressed in shares (see figure 4). For example, a decline in the share could be explained by an absolute decline in the number of people employed in the categories identified as potentially affected by offshoring. Alternatively, this selection of occupations could be growing at a slower pace than total employment. The relatively slower growth of employment potentially affected by offshoring is in fact what explains most of the declines observed in the trends, except for the United States, where the absolute number of people employed in the categories identified as potentially affected by offshoring has declined. These observations would therefore tend to support the idea that offshoring may lead to slower growth of employment in occupations potentially affected by offshoring and not necessarily to actual declines in employment. UNDERLYING TRENDS. This section describes the employment data underlying figure 4 in more detail, and the detailed graphs for the EU15 countries are given in figures A-1 and A-2. One caveat about these data is that there may be differences in the ICT content of occupations within and between countries. Similarly, any possible dynamic adjustments or changes in qualifications, skill requirements, and task descriptions that may take place within occupations over time are not taken into account. For the EU15 as a whole, the trend increases in all years, except in 1998. The year-by-year rate of change shows that employment potentially affected by offshoring grew faster than total employment in EU15 in all years except 1998, when total employment grew faster than offshoring. There was no absolute decline in employment potentially affected by offshoring. Figures A-1 and A-2 show the underlying data for the countries that make up the EU15. For Greece and Portugal the data quality is poor, especially early in the sample period. Furthermore, there appears to be a break in the data for Ireland between the 1995 and 1997 period and between the 1999 and 2003 period, with a missing data point in 1998. Nevertheless, the increase observed for the EU15 aggregate appears to be representative of the evolution in most of the underlying countries and does not appear to be driven by outliers. Most countries see an increase in the share of employment potentially affected by offshoring, and while there were some incidences of absolute decline at the individual country level, these were isolated and do not represent a trend. For the United States there was a downward trend in the period 1995–98 and 2001–03 (although 2003 is an estimate since both the industry and occupational classifications were changed that year). The year-by-year rate of change shows that total employment grew faster than employment potentially affected by offshoring in all years except 1999 and 2000. The absolute number in employment


177

potentially affected by offshoring declined in the United States in 1996 and in 2001, 2002, and 2003. This absolute decline was fairly generalized and not limited to a specific type of occupation or level of skills. All of the occupations selected as potentially affected by offshoring experienced at least one annual period of decline. Furthermore, forty-five of the sixty-seven occupations included in the U.S. selection experienced an absolute decline between 2001 and 2002, as did the overall selection of occupations potentially affected by offshoring and total employment. Finally, forty-seven of the selected occupations experienced at least three absolute declines between 1995 and 2002. For Canada the trend fell in 1995–96 and 1998–2003, except in 2000. The annual rate of change shows that total employment grew faster than employment potentially affected by offshoring except in 1997, 1998, and 2000. There was no absolute decline in employment potentially affected by offshoring. Finally, for Australia the trend fell in 1999 and 2001–03. The annual rate of change shows that total employment grew faster than employment potentially affected by offshoring in 1999 and 2002–03. There was no absolute decline in employment potentially affected by offshoring. Data for 2004 indicate that the trend continues to decline. Data for 1995 and 1996 are not directly comparable with those for the rest of the period because 1995 and 1996 are in the first edition of the Australian Standard Classification of Occupations (ASCO) and subsequent data in the ASCO second edition. These observations support the idea that it is not so much a decline in certain types of employment that can be expected, but rather slower employment growth in these types of occupations. Even though technology may account for at least some of the relative decline in the occupations potentially affected by offshoring (and for absolute declines in the case of the United States), the possibility that some of these jobs have been offshored cannot be ruled out. For example, Baily and Lawrence (2005) argue that at least some of the declines in low-wage ICTenabled occupations, a concept close but not equivalent to the group of clerical workers identified in the occupational selections of tables A-2 through A-5, took place as a result of activities being shifted overseas. Looking at IT specialist occupations, they also find that the net loss of computer programmers in the United States was most likely the result of offshoring. Nevertheless, even the largest projections of jobs to be offshored, as often reported in the media, are relatively small in comparison with annual job churning in OECD labor markets.13 Having examined some of the underlying trade and employment data, the next section presents a simple descriptive empirical model to provide a first indication 13. See OECD (2004b).

178


of the factors associated with aggregate changes in the share of potentially offshorable employment.

The Empirical Model and Some Results Using panel data estimation techniques, this paper attempts to identify those factors that are associated with the share of employment potentially affected by offshoring in total employment for the United States, Canada, Australia, and the EU15 countries (except Greece, Ireland, Luxembourg, and Portugal)14 over the period 1995–2003. The Model In the model, potentially offshorable employment as a share of total employment (OL) is a function of trade, investment, the industrial structure of the economy, a technology adoption/integration variable, a product market regulations indicator, an employment protection indicator, and human capital.15 The choice of variables is motivated by findings from a vast background literature, including studies of the factors determining the overall share of the service sector in the economy, studies of services sector employment, and studies of the effect of trade and technology on employment.16 Ideally, it would be appropriate to begin with a simple structural model of the factors affecting the relative demand for ICT-using occupations. Using the firstorder marginal productivity conditions from an (unknown) production function with two types of labor (ICT and non-ICT labour), such a model might be 14. These countries were excluded from the sample because of a lack of data. 15. Even though GDP per capita is a variable found to be an important determinant of the share of services sector employment (Messina 2004), it is not used here. In a time series context it does not make sense to include the level of GDP per capita in a regression of a bounded variable. The first difference in GDP per capita was found to be insignificant. This is not necessarily surprising since the countries in the sample all have relatively high levels of GDP per capita, so over the sample period (1995–2003) this variable is not found to have an impact on the share of employment potentially affected by offshoring. Nevertheless, with the exception of Austria, the countries with a relatively low share of employment potentially affected by offshoring were also those with the lowest levels of GDP per capita. The role of productivity growth is also not considered here. It is sometimes argued that the decline in certain types of employment, or the lack of new jobs (the jobless recovery), is the result of important productivity increases, but Baily and Lawrence (2004) argue that this is a mistake and that while productivity may have played some role, it should not be considered a fundamental cause. Time dummies pick up common cyclical effects. 16. See van Welsum and Reif (2005) for a review.


179

expected to include measures of the relative output and relative wages of ICTusing occupations. Control variables might also be included to pick up possible differences in the extent of (labor-augmenting) technical progress in the two broad types of occupations. As in the literature on the demand for skilled and unskilled labor, possible controls are indicators for both trade and technology. Unfortunately, although it is possible to control for output and technology effects directly, data on occupational wages are not readily available in most countries at the level of detail required. Their effect can be captured only indirectly by including variables that can be expected to have an influence on real wages. It should be noted that although it is not possible to estimate a full structural model, the estimates we show are not a pure reduced-form model either, since potentially endogenous current dated terms in output or in trade and technology (or both) remain in the model. OL = f(TRADE, FDI, STRUC, ICT, PMR, union, HK)

(1)

In particular, trade effects are approximated by including both imports and exports of other business and computer and information services as a share of GDP (current U.S. dollars, IMF balance of payments for trade data, OECD ANA database for GDP data). It is expected from the literature on trade-related displacement that imports may have a negative association with the share of employment potentially affected by offshoring, while exports are thought to have a positive relationship. Nevertheless, trade may not have an impact at the aggregate level but rather bring about shifts at the industry and occupation level.17 Net foreign direct investment is included as a share of GDP (current U.S. dollars, IMF balance of payments for stock data, and OECD ANA database for GDP data).18 The predictions from the literature are ambiguous about the direction of the relationship between these variables and the share of employment potentially affected by offshoring. Differential effects might be expected to occur for FDI in services and in manufacturing (similar to the way the relationship between trade and FDI depends on the level of aggregation),19 but such differences are hidden in the aggregate measures; only the net effect, which will be dominated by manufacturing FDI, can be picked up because much of the total inward and outward FDI stocks is in manufacturing, and there are relatively few detailed cross-country data that distinguish manufacturing from services FDI 17. See OECD (2005) for an overview. 18. This is done by imposing equal and opposite signs on outward and inward FDI, a restriction accepted by the data. 19. See Pain and van Welsum (2004) and van Welsum (2004).

180


over a long time period. However, in further research it will be attempted to include separate indicators for services and manufacturing FDI. The share of services sector20 value added in total value added and the share of high-tech industries21 value added in total value added are included as indicators of the industrial structure of the economy (OECD STAN database; missing values have been estimated using the “60-Industry Database” from the Groningen Growth and Development Centre of the University of Groningen (Netherlands), available at www.ggdc.net/dseries/60-industry.html [April 28, 2005]). Other things being equal, the larger the share of the services sector in the economy, the larger the relative demand for ICT-using occupations can be expected to be. To approximate technology adoption or integration, ICT investment (capital expenditure)22 as a share of gross fixed capital formation and as a share of GDP are included separately in different versions of the model.23 The ICT investment data are from an unpublished OECD database based on national account sources. The indicator of product market regulation is an average of indicators of regulation in selected nonmanufacturing industries.24 These indicators measure, on a scale of 0 to 6 (from least to most restrictive), restrictions on competition and private governance. The original version of these data is described in Nicoletti and Scarpetta (2003). This indicator is used as a proxy for competitive pressures in the economy. The weaker such pressures are, the less incentive there is for companies to adopt efficient new technologies and new, more productive, ways of working. This would imply that a negative relationship can be expected between the importance of product market regulations in the economy and the 20. ISIC Rev.3 categories: 50–55: wholesale and retail trade; repairs; hotels and restaurants; 60–64: transport, storage, and communications; 65–74: finance, insurance, real estate, and business services; 75–99: community, social, and personal services. 21. ISIC Rev.3 categories: 2423: chemicals excluding pharmaceuticals; 30: office, accounting, and computing machinery; 32: radio, television, and communication equipment; 33: medical, precision, and optical instruments; 353: aircraft and spacecraft. 22. ISIC Rev.3 categories: 30: office, accounting, and computing machinery; 3130: insulated wire and cable; 3210: electronic valves and tubes and other electronic components; 3220: television and radio transmitters and apparatus for line telephony and line telegraphy; 3230: television and radio receivers, sound or video recording or reproducing apparatus, and associated goods; 3312: instruments and appliances for measuring, checking, testing, navigating, and other purposes; 3313: industrial process control equipment; 5150: wholesale of machinery, equipment, and supplies; 6420: telecommunications; 7123: renting of office machinery and equipment (including computers); 72: computer and related activities. 23. Results from the regressions using ICT investment as a share of GDP are not reported here. See van Welsum and Reif (2005) for details. 24. We use a preliminary unpublished version of this product market regulation indicator.


181

share of employment potentially affected by offshoring. Similarly, Messina (2004) includes a measure of entry barriers to the creation of new firms in the economy as an indicator of product market regulations and finds a significant and negative effect on the share of services sector employment. Two variables are included to capture institutional and supply-side influences on (unobserved) real wages: union density and human capital. Trade union density indicators may of course provide information about the degree of flexibility in national labor markets, as well as the relative strength of workers in wage bargaining.25 A number of existing papers suggest that union density rates are related to the growth of service sector occupations. For example, Messina (2004) finds that a fall in union density rates is associated with an increase in services sector employment. Similarly, Nickell, Redding, and Swaffield (2004) find evidence that countries with higher levels of employment protection were slower to reallocate resources from declining sectors (agriculture, manufacturing, and other production) to the services sector, possibly because stronger employment protection makes shedding labor in declining sectors more costly. The analysis in this paper does not consider employment at the sectoral level, but an analogy can be drawn since labor market inflexibilities are likely to affect occupational shifts as well as sectoral changes. The a priori effect of this variable is ambiguous, though, because it can both prevent a reallocation of resources into ICTintensive using occupations and hinder the speed at which existing ICT-intensive using jobs can be transferred abroad, maintaining the share at a higher level than it would otherwise have been. Finally, human capital is approximated by the average years of education per person.26 It is expected that this variable is positively related to the share of potentially offshorable occupations since increases in human capital are positively correlated with increases in the supply of ICT-literate people in the workforce. Such increases in supply should help to restrain the growth of real wages of workers in ICT occupations and hence support demand. Nickell, Redding, and Swaffield (2004) find a strong positive effect of increases in educational attainment on the output share of the “other services” sector in the economy in Australia, Canada, France, Italy, Japan, the Netherlands, Sweden, West Germany, the United Kingdom, and the United States.27

25. The data on trade union density rates come from OECD Labor Force Statistics Indicators and OECD 2004c (table 3.3). Factors other than union density rates, including union coverage and hiring and firing restrictions, are also important. 26. See de la Fuente and Doménech (2002a, 2002b) and OECD (2003). 27. But in the sector “business services” they found a greater role for changes in relative prices.

182


Some Results The results using fixed-effects and instrumental variables estimation techniques on a sample excluding Greece, Ireland, Luxembourg, and Portugal are reported in table 1.28 The estimation for the basic fixed-effects models is for a sample of fourteen countries over the period 1996–2003. The instrumental variables estimates are for the same countries, but over the period 1997–2003. Columns 1 and 3 of table 1 show the standard fixed-effects results, and columns 2 and 4 show the results obtained when reestimating these models using instrumental variables. A year is dropped from the estimation period for these latter regressions to allow higher-order lagged variables to be used as instruments. All current dated terms, with the exception of the product market regulation indicator, are instrumented in columns 2 and 4. For these variables only instruments dated t-2 are included in the instrument set. The Sargan tests of the overidentifying restrictions provide support for the validity of the instrument set employed in both models. In each of the four models (columns 1 to 4), exports are found to have a positive and significant association with the share of employment potentially affected by offshoring, as expected. The coefficient on imports is negatively signed, as expected, but is not significant at the conventional 5 percent level in any of the models. Thus there is no significant evidence that increasing imports of other business and computer and information services are associated with a reduction in the share of employment potentially affected by offshoring at the aggregate level. Care is needed in drawing strong conclusions from these results, though, as the trade variables may be endogenous, especially if companies’ decisions about international sourcing and employment are made simultaneously. However, as shown in columns 2 and 4 of table 1, and in van Welsum and Reif (2005), the basic findings remain even when an instrumental variables estimator is employed. Net FDI is found to have a positive and significant association with the share of employment potentially affected by offshoring. Thus, contrary to popular belief, there is no evidence that outward investment or net FDI reduces the share of this type of employment at the aggregate level. This effect can probably be explained by the fact that manufacturing activities are much more important in total FDI than they are in the overall share of activities in host and home economies. An increase in the outward stock of FDI can also be expected to increase the relative share of occupations in support functions, as well as mar28. Country fixed-effects and year time dummies are included in all models. See van Welsum and Reif (2005) for further results.


183

Table 1. Estimation Results Using Fixed-Effects and Instrumental Variables Dependent variable: employment potentially affected by offshoring as a share of total employment (OLt ) 1 (X/GDP)t (M/GDP)t (NETFDI/GDP)t-1 (ICTI/INV)t-1 SERVICESt-1 HTECHt-1 PMRt UNIONSt-1 HKt-1 Sample period Observations Log likelihood R2 Standard error Time dummies (p value of joint deletion) Sargan test (p value)

2

3

4

0.9086 (5.8)* 0.9298 (2.6)* 0.8977 (5.6)* 1.3139 (3.0)* –0.2246 (1.4) –0.1309 (0.3) –0.3099 (2.0)* –0.7119 (1.4) 0.0384 (3.3)* 0.0435 (3.2)* 0.1132 (1.8)† 0.0984 (0.8) 0.0968 (1.5) 0.0992 (0.8) 0.1649 (3.6)* 0.1716 (3.5)* 0.1852 (3.5)* 0.1961 (3.6)* 0.1592 (0.7) 0.1760 (0.6) 0.2382 (1.1) 0.3056 (1.1) –0.1614 (0.7) 0.0171 (0.0) –0.0348 (0.1) –0.0105 (0.0) –0.1252 (2.9)* –0.1298 (2.6)* –0.0952 (2.1)* –0.1145 (2.1)* 1.1719 (3.7)* 1.2913 (3.2)* 1.3954 (4.2)* 1.4404 (3.3)* Fixed effects Fixed effects IV Fixed effects Fixed effects IV 1996–2003 1997–2003 1996–2003 1997–2003 112 98 112 98 –70.145 –74.863 0.963 0.960 0.960 0.957 0.542 0.563 0.562 0.583

0.193

0.795 0.112

0.609

0.853 0.611

Notes: (X/GDP) is the share of exports of other business and computer and information services in GDP; (M/GDP) is the share of imports of other business and computer and information services in GDP; (NETFDI/GDP) is the net stock of foreign investment (outward-inward) as a share of GDP; (ICTI/INV) is the share of ICT investment in total fixed investment; SERVICES is the share of the services sector in total value added; HTECH is the share of high-tech industries in total value added; PMR is a product market regulations indictor; UNIONS are trade union density rates; and HK is the average years of education per person. The additional instruments used are drawn from a set comprising (X/GDP)t-2, (M/GDP)t-2, OLt-2, (ICTI/INV)t-2, PMRt-1, PMRt-2, UNIONSt-2 and (NETFDI/GDP)t-2. * Significant at the 5 percent level. † Significant at the 10 percent level.

keting, design, and general headquarters services. Inward investment is found to be negatively related to the share of employment potentially affected by offshoring. With manufacturing also having a comparatively heavy weight in the activities of inward investors, it is not necessarily surprising that the relative share of employment in the types of occupations identified as potentially affected by offshoring is reduced. Further research will attempt to disentangle the effects of services investment from manufacturing investment. There are many different factors that might be reflected in the coefficients on the FDI variables. It is also true that FDI data can, at times, be a poor measure of the scale of activities that multinational companies undertake. Although this in itself is not a reason for omitting the FDI variables, it is prudent to repeat the regressions without them to ensure that their inclusion is not serving to significantly bias the coefficients on the other explanatory factors. The results, shown

184


in columns 3 and 4 of table 1, suggest that the net FDI variable is largely orthogonal to the remaining regressors, with the possible exception of the imports term, whose coefficient becomes more negative. However, it remains insignificant, at least at the 5 percent level. The share of ICT investment in gross fixed capital formation is positively signed, but is not especially significant. The share of services sector value added as a percentage of total value added has a significant positive association with the share of employment potentially affected by offshoring, as expected, with many services having high shares of ICT-using occupations, but there is no significant relationship between the dependent variable and the high-tech industry value added as a percentage of total value added (though the coefficient is positively signed). The indicator of the importance of product market regulations in the economy is negatively signed (except in column 3) but is not significant. The two variables that are most likely to affect wages—union density and human capital—both have coefficients of the sign expected given the assumption that wages have a negative effect on employment. Higher levels of union density are associated with slower adjustment into the types of occupations potentially affected by offshoring, and the average years of education per person is significantly positively associated with the share of potentially offshorable employment, consistent with the observation that many such occupations are comparatively skill-intensive. Overall, the results appear fairly robust to different estimation techniques and specifications of the model. The most stable coefficients appear to be those on the ratio of exports of other business and computer and information services to GDP, net foreign direct investment stocks as a share of GDP, the share of the services sector in value added, and the average years of education per person. The full interpretation of these results must await further study. In particular, the development of corresponding data on relative wages should help to separate out demand and supply influences more clearly. Nevertheless, the results from the descriptive regressions in the present paper provide some useful indications of the statistical associations that are found between the variables examined and provide guidance for further work in this area.

Conclusions Despite the widespread media attention given to the apparent offshoring of service sector jobs, little is known about the extent of this phenomenon, or the


185

extent to which it is related to other economic and structural developments. In particular, an explicit link is often made between trade, the activities of multinational firms, and changes in employment, but this has not been founded on any solid evidence. This paper builds on previous detailed analysis of trade and occupational data. Trade data show that many of the countries frequently cited as beneficiaries of offshoring have seen rapid growth of their exports of other business and computer and information services. However, many have also seen rapid growth of imports of these services, and the bulk of exports of these types of services still come from OECD countries, although their export share is declining. The analysis of occupational employment data for selected OECD countries sought to determine the share of total employment that could potentially be affected by the international sourcing of IT and ICT-enabled services, drawing on van Welsum and Vickery (2005a). It suggested that close to 20 percent of total employment could potentially be affected by offshoring. This paper also makes an initial attempt to examine the relationship between the share of employment potentially affected by offshoring and other economic and structural developments using some simple descriptive regressions on a panel of selected OECD economies between 1996 and 2003. In particular, first estimates are provided of the statistical association between the share of potentially offshorable employment and trade in business services and international direct investment. The results indicate that exports of other business services and computer and information services are positively associated with the share of employment potentially affected by offshoring. This suggests that increases in demand and production have led to a relative increase in the types of ICT-using occupations identified in the analysis. Furthermore, contrary to popular belief, no evidence is found of a significant negative association between imports of these services and the share of employment potentially affected by offshoring. Similarly, no evidence is found that net outward investment reduces the employment share of the ICT-intensive using occupations identified as potentially affected by offshoring. Other key factors associated with cross-country differences in the employment share are found to be the comparative size of the service sector, the growing share of ICT investment in total fixed investment, and human capital. These results suggest that, in the OECD countries analyzed, ICT-enabled services offshoring (as proxied by trade and investment) has not yet led to a relative decline in the occupational share of location-independent ICT-using occupations. This implies that in the long run the positive benefits of services off-

186


shoring outweigh the costs, even though the adjustment process may occasionally be difficult in the short run. Policy responses to services offshoring should reflect these positive aspects. These would include policies that seek to contribute to the overall competitiveness of the economy and improve the macroeconomic framework, to contribute to a sound investment climate, and to improve the skills base and flexibility of the workforce. It is important to interpret these results with caution, though, because they are not drawn from the empirical testing of a formal theoretical model of the underlying structural relationships. Thus, it is not possible to separate out completely the effects from demand and supply-side developments. However, the results provide guidance on the statistical associations between the variables included in these descriptive regressions and to this extent can be used to shape further work and analysis. This could include improvements to the underpinnings of the empirical model, such as the use of separate indicators for services and nonservices FDI, and examination of whether different factors affect different groups of ICT-using occupations, such as clerical and nonclerical occupations. It would also be useful to develop an indicator of business adoption of ICTs to try to control for differences in “the use of ICT” or the “ICT content of occupations” across countries.

Desirée van Welsum and Xavier Reif Table A-1. IMF Balance of Payments Categories 7 7.1 7.2 7.2.1 7.2.2

Computer and information services Computer services Information services News agency services Other information provision services

9 9.1 9.1.1 9.1.2 9.2 9.3 9.3.1 9.3.1.1 9.3.1.2 9.3.1.3 9.3.2 9.3.3 9.3.4 9.3.5 9.3.5.1 9.3.5.2 9.3.6 9.3.7

Other business services Merchanting and other trade-related services Merchanting Other trade-related services Operational leasing services Miscellaneous business, professional, and technical services Legal, accounting, management consulting, and public relations Legal services Accounting, auditing, bookkeeping, and tax consulting services Business and management consulting, and public relations Advertising, market research, and public opinion polling Research and development Architectural, engineering, and other technical services Agricultural, mining, and on-site processing services Waste treatment and depollution Agricultural, mining, and other on-site processing services Other business services Services between related enterprises, n.i.e.a

Source: OECD (2002). a. Not included elsewhere.

187

188


Table A-2. Europe: Occupations Potentially Affected by Offshoringa 3-digit ISCO-88 categories 123 211 212 213 214 241 242 243 312 341 342 343 411 412 422

Other specialist managers Physicists, chemists, and related professionals Mathematicians, statisticians and related professionals Computing professionals Architects, engineers, and related professionals Business professionals Legal professionals Archivists, librarians, and related information professionals Computer associate professionals Finance and sales associate professionals Business services agents and trade brokers Administrative associate professionals Secretaries and keyboard-operating clerks Numerical clerks Client information clerks

Source: Van Welsum and Vickery (2005a), based on European Union Labor Force survey (www.eds-destatis.de/en/microdata/ microlfs.php [December 2004]). a. Shaded occupations are classified as clerical.


189

Table A-3. United States: Occupations Potentially Affected by Offshoringa CPS categories 23 24 25 26 43 44 45 46 47 48 49 53 54 55 56 57 58 59 63 64 65 66 67 68 69 73 74 75 76 77 78 79 83 164 165 166 173

Accountants and auditors Underwriters Other financial officers Management analysts Architects Aerospace engineers Metallurgical and materials engineers Mining engineers Petroleum engineers Chemical engineers Nuclear engineers Civil engineers Agricultural engineers Engineers, electrical and electronic Engineers, industrial Engineers, mechanical Marine and naval architects Engineers, n.e.c.b Surveyors and mapping scientists Computer systems analysts and scientists Operations and systems researchers and analysts Actuaries Statisticians Mathematical scientists, n.e.c. Physicists and astronomers Chemists, except biochemists Atmospheric and space scientists Geologists and geodesists Physical scientists, n.e.c. Agricultural and food scientists Biological and life scientists Forestry and conservation scientists Medical scientists Librarians Archivists and curators Economists Urban planners

183 184 195 227 229 233 243 253 254 255 257 304 305 306 308 309 313 315 318 335 336 337 338 339 343 344 348 383 385 386

Authors Technical writers Editors and reporters Air traffic controllers Computer programmers Tool programmers, numerical control Supervisors and proprietors, sales occupations Insurance sales occupations Real estate sales occupations Securities and financial services sales occupations Sales occupations, other business services Supervisors, computer equipment operators Supervisors, financial records processing Chief communications operators Computer operators Peripheral equipment operators Secretaries Typists Transportation ticket and reservation agents File clerks Records clerks Bookkeepers, accounting, and auditing clerks Payroll and timekeeping clerks Billing clerks Cost and rate clerks Billing, posting, and calculating machine operators Telephone operators Bank tellers Data-entry keyers Statistical clerks

Source: van Welsum and Vickery (2005a), based on U.S. Current Population Survey. a. Shaded occupations are classified as clerical. b. Not elsewhere classified.

Table A-4. Canada: Occupations Potentially Affected by Offshoringa SOC91 Canada categories A121 A122 A131 A301 A302 A303 A311 A312 A392 B011 B012 B013 B014 B022 B111 B112 B114 B211 B212

Engineering, science, and architecture managers Information systems and data processing managers Sales, marketing, and advertising managers Insurance, real estate, and financial brokerage managers Banking, credit and other investment managers Other business services managers Telecommunication carriers managers Postal and courier services managers Utilities managers Financial auditors and accountants Financial and investment analysts Securities agents, investment dealers and traders Other financial officers Professional occupations in business services to management Bookkeepers Loan officers Insurance underwriters Secretaries (except legal and medical) Legal secretaries

C012 C013 C014 C015 C021 C031 C032 C033 C034 C041 C042 C043 C044 C045 C046 C047 C048 C051 C052 C053

Chemists Geologists, geochemists, and geophysicists Meteorologists Other professional occupations in physical sciences Biologists and related scientists Civil engineers Mechanical engineers Electrical and electronics engineers Chemical engineers Industrial and manufacturing engineers Metallurgical and materials engineers Mining engineers Geological engineers Petroleum engineers Aerospace engineers Computer engineers Other professional engineers, n.e.c. Architects Landscape architects Urban and land use planners

B213 B214 B311 B312 B412 B512 B513 B514 B521 B522 B523 B524 B531 B532 B533 B534 B553 B554 C011

Medical secretaries Court recorders and medical transcriptionists Administrative officers Executive assistants Supervisors, finance and insurance clerks Typists and word processing operators Records and file clerks Receptionists and switchboard operators Computer operators Data entry clerks Typesetters and related occupations Telephone operators Accounting and related clerks Payroll clerks Tellers, financial services Banking, insurance and other financial clerks Customer service, information and related clerks Survey interviewers and statistical clerks Physicists and astronomers

Source: van Welsum and Vickery (2005a), based on Statistics Canada. a. Shaded occupations are classified as clerical.

C054 C061 C062 C063 C152 C172 E012 E031 E032 E033 F011 F013 F021 F022 F023 F025 G131

Land surveyors Mathematicians, statisticians, and actuaries Computer systems analysts Computer programmers Industrial designers Air traffic control occupations Lawyers and Quebec notaries Natural and applied science policy researchers, consultants, and program officers Economists and economic policy researchers and analysts Economic development officers and marketing researchers and consultants Librarians Archivists Writers Editors Journalists Translators, terminologists, and interpreters Insurance agents and brokers

192


Table A-5. Australia: Occupations Potentially Affected by Offshoringa ASCO 4-digit categories 1221 1224 1231 1291 2111 2112 2113 2114 2115 2119 2121 2122 2123 2124 2125 2126 2127 2211 2212 2221 2231 2292 2293 2294

Engineering managers Information technology managers Sales and marketing managers Policy and planning managers Chemists Geologists and geophysicists Life scientists Environmental and agricultural science professionals Medical scientists Other natural and physical science professionals Architects and landscape architects Quantity surveyors Cartographers and surveyors Civil engineers Electrical and electronics engineers Mechanical, production, and plant engineers Mining and materials engineers Accountants Auditors Marketing and advertising professionals Computing professionals Librarians Mathematicians, statisticians, and actuaries Business and organization analysts

2299 Other business and information professionals 2391 Medical imaging professionals 2521 Legal professionals 2522 Economists 2523 Urban and regional planners 2534 Journalists and related professionals 2535 Authors and related professionals 3211 Branch accountants and managers (financial institution) 3212 Financial dealers and brokers 3213 Financial investment advisers 3294 Computing support technicians 3392 Customer service managers 3399 Other managing supervisors (sales and service) 5111 Secretaries and personal assistants 5911 Bookkeepers 5912 Credit and loan officers 5991 Advanced legal and related clerks 5993 Insurance agents 5995 Desktop publishing operators 6121 Keyboard operators 6141 Accounting clerks 6142 Payroll clerks 6143 Bank workers 6144 Insurance clerks 6145 Money market and statistical clerks 8113 Switchboard operators 8294 Telemarketers

Source: van Welsum and Vickery (2005a), based on Australian Bureau of Statistics. a. Shaded occupations are classified as clerical.


193

References Autor, David H., Frank Levy, and Richard J. Murnane. 2003. “The Skills Content of Recent Technological Change: An Empirical Exploration.” Quarterly Journal of Economics 118, no. 4: 1279–1333. Baily, Martin N., and Robert Z. Lawrence. 2004. “What Happened to the Great U.S. Job Machine? The Role of Trade and Electronic Offshoring.” BPEA, no. 2: 211–84. Bardhan, Ashok D., and Cynthia Kroll. 2003. “The New Wave of Outsourcing.” Fisher Centre Research Report 1103. University of California Berkeley, Fisher Centre for Real Estate and Urban Economics. de la Fuente, Angel, and Raphael Doménech. 2002a. “Educational Attainment in the OECD, 1960–1995.” Discussion Paper 3390. Centre for Economic Policy Research, London. ———. 2002b. “Human Capital in Growth Regressions: How Much Difference Does Data Quality Make? An Update and Further Results.” Discussion Paper 3587. Centre for Economic Policy Research, London. Kirkegaard, Jacob Funk. 2004. “Outsourcing—Stains on the White Collar?” Washington: Institute for International Economics. Levy, Frank, and Richard J. Murnane. 2004. The New Division of Labor. Princeton University Press and the Russell Sage Foundation. Mann, Catherine L. 2004. “The U.S. Current Account, New Economy Services and Implications for Sustainability.” Review of International Economics 12, no. 2: 262–76. Marin, Dalia. 2004. “‘A Nation of Poets and Thinkers’—Less So with Eastern Enlargement? Austria and Germany.” Department of Economics Discussion Paper 2004-06. University of Munich. Messina, Julian. 2004. “Institutions and Service Employment: A Panel Study for OECD Countries.” Labour: Review of Labour Economics and Industrial Relations 19, no. 2: 343–72. Millar, Jane. 2002. “Outsourcing Practices in Europe.” STAR Issue Report 27, www.databank.it/star/list_issue/e.html. Nickell, Stephen, Stephen Redding, and Joanna Swaffield. 2004. “The Uneven Pace Of Deindustrialisation in the OECD.” Paper prepared for the OECD Workshop on Services, November 15–16, based on CEPR Discussion Paper 3068. Nicoletti, Giuseppe, and Stefano Scarpetta. 2003. “Regulation, Productivity and Growth: OECD Evidence.” Economic Policy (April): 9–72. Organization for Economic Cooperation and Development (OECD). 2002. The Manual on Statistics of International Trade in Services, joint publication of the United Nations, the International Monetary Fund, the OECD, the European Commission, the United Nations Conference on Trade and Development, and the World Trade Organization. An electronic version of the manual is available free of charge at www.oecd.org/std/trade-services.

194


———. 2003. The Sources of Economic Growth in OECD Countries. Paris. ———. 2004a. OECD Information Technology Outlook 2004. Paris. ———. 2004b. OECD Economic Outlook 2004/1, no. 75. Paris. ———. 2004c. OECD Economic Outlook 2004. Paris. ———. 2005. OECD Employment Outlook 2005. Paris. Pain, Nigel, and Desirée van Welsum. 2004. “International Production Relocation and Exports of Services.” OECD Economic Studies 2004/1, no. 38. Schultze, Charles L. 2004. “Offshoring, Import Competition, and the Jobless Recovery.” Policy Brief 36. Brookings. van Welsum, Desirée. 2003. “International Trade in Services: Issues and Concepts.” Birkbeck Economics Working Paper 2003/4. Birkbeck College, University of London. ———. 2004. “In Search of ‘Offshoring’: Evidence from U.S. Imports of Services.” Birkbeck Economics Working Paper 2004/2. London: Birkbeck College, University of London. van Welsum, Desirée, and Xavier Reif. 2005 (forthcoming). “The Share of Employment Potentially Affected by Offshoring—An Empirical Investigation.” DSTI Information Economy Working Paper DSTI/ICCP/IE(2005)8/FINAL. Paris: OECD. www.oecd. org/sti/offshoring. van Welsum, Desirée, and Graham Vickery. 2005a. “Potential Offshoring of ICTIntensive Using Occupations.” DSTI Information Economy Working Paper DSTI/ICCP/IE(2004)19/FINAL. Paris: OECD. www.oecd.org/sti/offshoring. ———. 2005b. “New Perspectives on ICT Skills and Employment.” DSTI Information Economy Working Paper DSTI/ICCP/IE(2004)10/FINAL. Paris: OECD. www. oecd.org/sti/ICT-employment.


Robert Z. Lawrence: The strength of these two papers is that they both carefully scrutinize the data to see what we can learn about offshoring. Their weakness is that they both do so without adopting a theoretical framework that could help them sort out what they find. Data mining can sometimes be useful when the data clearly allow us to reject hypotheses, but when they do not, the limitations become apparent and the need for a framework more obvious. I will elaborate on these points in discussing each paper in turn. In the first paper, Maria Borga uses firm-level data to explore trends in employment in the parents and affiliates of U.S. multinational companies. Her strategy is to explore the degree to which firms changed their offshoring behavior between 1994 and 2002 and then to explore the association between offshoring and other variables, in particular employment. As the author herself points out, this approach allows only for a consideration of bivariate relationships and excludes the possibility of detecting more complex interactions. The paper does yield a number of insights. Let me point out three I found striking. First, the most relevant for our topic is the finding that between 1994 and 2002 imports of services from majority-owned foreign affiliates actually played a very small role in the growth of parent companies’ overall purchases of goods and services. In fact, services imports actually fell from 0.4 percent of parent purchases in 1994 to 0.2 percent in 2002. If these data are correct, it suggests that whatever may happen in the future, offshoring has had very little to do with U.S. employment changes in the past. This would be a very surprising result for anyone listening to the nightly news in 2004. Second, the data also suggest less direct outsourcing in goods than might have been expected. The growth in parents’ use of merchandise imports from affiliates (up from 3.5 to 4.6 percent of purchases) was relatively small, while 195

196


there was no growth in the share accounted for by imports from nonaffiliated firms. Firms are outsourcing more (up from 60 to 63 percent of sales), but apparently most of this rise is from domestic firms. Thus, there is some support for the view that firms are slimming down to their core competencies, but not for the view that the globalization of production is a hugely important phenomenon. Third, a conspicuous feature of the recent U.S. recession was the sustained growth in labor productivity. This could be due either to the cold bath effect of eliminating the least productive parts of the firm—that is, downsizing—or to technological innovation. Borga’s paper suggests downsizing could be important. The picture we obtain from the paper is that the firms that were reducing employment were disproportionately laying off their low-wage (least productive) workers. We see this in the rise in average compensation and relatively faster productivity growth in these firms. These firms also significantly increased their purchases of goods and services as a share of output (increasing purchased inputs from 62 to 66 percent of sales—much of which represented purchases from the rest of the world up from 6.95 to 9.3 percent of purchases). The foreign affiliates of firms with declining employment growth also have slower employment growth, suggesting complementarity rather than substitutability in employment at home and abroad. This is a useful paper, but it is just a start because it does not allow us to explain why we obtain these results. As we know, correlations do not tell us about causation. Consider the finding that quantitatively the most significant factors affecting parent employment were changes in output and in labor productivity and that outsourcing in general, and offshoring in particular, appear much less important. We know that all three of these variables are jointly determined and are not exogenous. It is certainly possible that the ability to source foreign inputs actually stimulates output and helps to induce greater productivity. In this case, even though an ex post accounting exercise of the type performed in the paper would attribute only a small part of the change to outsourcing, it would clearly be wrong to infer that outsourcing is not important. In fact, the more powerful the impact of outsourcing on output and productivity, the less important it would appear to be in the decomposition exercise that is performed here. On the other hand, if the variables have predominantly independent effects, it is also possible that the conclusions drawn here are correct. It is therefore appropriate that Maria Borga is cautious in the claims that she makes about these findings and that she calls for further research on these issues. The paper by Desirée van Welsum and Xavier Reif uses international occupational data to construct an estimate of jobs that could potentially be offshored.

Borga / Van Welsum and Reif: Comments

197

As the authors are aware, these estimates are subject to considerable qualification because national classification methodologies differ. In addition, even a casual inspection of the occupational categories suggests that there is considerable misclassification. Even though their jobs have been identified here as all potentially affected by outsourcing, for example, it is highly likely that many secretaries, librarians, bank tellers, real estate agents, air traffic controllers, editors, and reporters need to be in the location where their services are used. This means that measurement error is introduced into the dependent variable and could seriously impair the results. The authors report on the behavior of this variable (let’s call it the PO variable) over the period 1995 to 2003 in different OECD countries. The PO employment experience has been mixed. There are pronounced declines in the United States and Canada and more recently in Australia, but strong upward growth in the European Union. This could be evidence that the offshoring of ITrelated and back-office activities has had an important impact on North American employment. But this is not something the authors are able to pin down. The reason is that they are interested in explaining PO, not actual offshoring. The main contribution of the paper is to explain the behavior of the PO variable using a cross section regression analysis of fourteen OECD countries. The results indicate that PO employment is significantly and positively affected by exports of information services, net FDI, the employment share of services, and human capital, and negatively affected by a high share of unions. Think about what these findings imply for the United States. I presume, for example, that the United States exports lots of information services, has a high share of services employment, a highly educated labor force, and a low share of unions. This would lead me to expect, on the basis of the regression results in the paper, that the United States should have a high share of PO employment! But apparently, as of 2003, the United States actually had a low share of PO. This result could mean something else; perhaps, as the authors suggest, it is that offshoring has caused the U.S. share to be low. But they do not prove or even explore this hypothesis. In fact, taking these results at face value increases the puzzle as to why the share of employment potentially affected by offshoring has declined in the United States. My main problem with the paper is that it gets to the regressions too quickly and does not lay out an adequate theoretical framework. If this is to be a paper about PO, I would like a step-by-step explanation of the relationship between PO and the variables that are introduced as independent. Assume, for a moment, that there is no offshoring—that all information services have to be provided locally.

198


We could then build and estimate a model that explained the share of employment in such services. The reduced form of such a model would probably have several of the independent variables that are in the regression the authors use; some would relate to the demand for these services (specific types of spending such as investment, exports, services in general, and imports) and some would relate to the supply of these services (such as human capital). As long as there was no offshoring we would be able to estimate this model using data for actual information technology services employment. Now assume that because of the Internet or improvements in communication offshoring can take place. We would then have to come up with a model that reflected the determinants of this offshoring. (Presumably this would require modeling the variables driving relative costs of undertaking services activities at home and abroad—variables that seem to be missing in this paper.) Actual services employment would then reflect potentially affected employment—the variable explained by the previous model—minus employment that is actually offshored. If we now want to explain actual services employment, therefore, we see that we have to include the independent variables from both the first and the second model. Which of these two worlds do we live in? Presumably the one in which outsourcing is already taking place. This means that the variable used by the authors as the dependent variable is also affected by the determinants of offshoring, but this is not explicitly taken account of in the regression. To the degree that offshoring is already an important source of variation in IT service employment, this could be a serious omission, and it could introduce bias into the results. Indeed, the essence of the debate about offshoring is that the relationship between certain types of domestic production and (actual) employment in IT services is changing. If offshoring has become important, exports, investment, and other types of spending now lead to fewer domestic employment opportunities in IT. But that is not something we really can learn about from these regressions. In sum, both these papers take first steps to deal with the questions we are interested in at this conference, but neither really succeeds in presenting a convincing explanation for its findings. In both cases, it seems, the main problem stems from moving too quickly to the data, before laying out the theory of what really drives offshoring in the first place. Catherine L. Mann: Recent media and political hype has focused on the negative impact on white-collar employment in industrial countries of cross-border


199

trade in services, particularly by multinational corporations. These papers address this theme, but try to evaluate the simplistic stories through the more robust and well-founded research lens of how multinational corporations engage with the global economy through trade and direct investment, and how the decision to go abroad is importantly linked to both the domestic and foreign environments. These two papers are complementary in their research focus and method. Borga focuses on U.S. multinational firms and on types of outsourcing in both manufacturing and services activities. Van Welsum and Reif address a range of industrial country experience, paying attention primarily to services occupations. In keeping with new availabilities, each uses some data that are quite detailed. Borga uses firm-level data for trade, direct investment, and employment. Van Welsum and Reif use detailed occupation data. The authors of both papers recognize that their results would be enhanced by more rigorous matching at the detail level of the data. For example, Van Welsum and Reif need to use FDI disaggregated at least to the level of manufacturing and services and could stratify occupations by above- and below-average wage. Borga could match the sector of the firmlevel data to occupational data, which might give a glimmer of the different impacts of outsourcing on production and on nonproduction employment. Each paper uses some statistical methods to carve the data into subsamples (by industry, or by occupation, or by expanding and contracting firms). A more systematic strategy of fixed-effects or random-effects econometric analysis (by country or occupational group in the case of Van Welsum and Reif, and by industry sector and occupational group in the case of Borga) would perhaps appear a bit less ad hoc to the reader and would highlight important differences across sectors or countries or occupations. On balance, however, the findings of these two papers are important additions to our understanding of both “old” outsourcing and “new” offshoring. In keeping with most previous work on global engagement (which could have been more liberally referenced by the authors), these papers find that trade, FDI, and domestic employment are positively related, particularly for competitive and expanding industries. Moreover, factors that underpin a rising standard of living (such as IT and human capital investment) go hand-in-hand with greater exposure to international market forces. More detailed comments on the findings and methods of each paper individually follow. A key contribution and a key aspect of the Borga paper is in distinguishing between outsourcing and offshoring. The value of a multinational parent’s “purchased inputs” plays a key role in the paper. Purchased inputs include outsourcing (purchasing from a domestic or foreign unrelated party) and offshoring

200


(purchasing from a foreign party, either affiliated or unrelated). The data show that dependence on purchased inputs rose from 60 to 63 percent (1994 to 2002). But the imported share of the purchased inputs rose only from 6.9 to 7.8 percent. Hence, even for multinationals, the bulk of purchased inputs (such as outsourcing) comes from home, not from abroad. Borga stratifies the sample into growing parents and shrinking parents, as measured by employment over the period 1994 to 2002. It may come as a surprise, but of the 1,117 multinational parent firms, 689 expanded and 428 shrank. The most notable difference between expanding parents and contracting ones was in sales. Sales grew in expanding parent firms and declined in contracting ones. In many other respects the two are similar: Both increased employment at their affiliates abroad as well as the share of those affiliates’ sales in their own markets, although expanders increased by more both the share of employment abroad and the share of affiliate sales abroad. Both relied more on purchased inputs. So, other than in sales growth and decline, how do expanders and shrinkers differ? Expanders reduced their reliance on imported purchased inputs, whereas shrinkers increased their reliance on imported inputs, particularly from their own affiliates (although keep in mind that the affiliate imports as a share of purchased inputs was still small, just 6.0 percent in 2002). When all data are aggregated, the finding that FDI and domestic employment are complements, not substitutes, is affirmed. However, when the expanders and the shrinkers are disaggregated, the increase in imported inputs is statistically negatively correlated with employment, both because expanders reduced imports and because shrinkers increased imports. All told, however, the shrinkers look to be in sectors experiencing secular decline in demand for their products that no form of outsourcing (either purchasing domestically or from affiliates abroad) can remedy. It would be very interesting to know the sectors of the expanders and shrinkers, even if only at the level of textiles and apparel, electronics, and so on. In this paper, Borga only pursues the limited classification of manufacturing, services, and “other,” which is not very enlightening. The most important attribute of the van Welsum and Reif paper is the detailed work with occupation-level data across several different OECD countries. Occupations are stratified by characteristics including ICT intensity in use; digitizable output; codifiable knowledge content; face-to-face contact not required. The stratification requires a detailed assessment of each country’s labor statistics. It would be nice if a more econometrically based approach could back up the stratification, but the necessary data (such as IT by sector, or wages by sector and occupation) are not available. At the end of the exercise, van Welsum and Reif find that about


201

20 percent of employment is exposed to technology-based tradability. Note that this tradability could yield domestic outsourcing or foreign offshoring. Van Welsum and Reif find that the share of such exposed employment rises in the European Union and is stable or falling in the United States, Canada, and Australia. They undertake preliminary econometric analysis to explain the time series evolution of the share using country-specific international data on services trade flows and net total FDI, and domestic structural characteristics including intensity of ICT investment, product and employment regulations, and human capital. As in the Borga paper, exports and FDI are positively associated with the share of employment exposed to tradability. The FDI variable, being both net and inclusive of manufacturing and services, needs additional disaggregation to better match the detail of the data, as the authors agree. With respect to the structural variables, it is notable that the factors that underpin a rising standard of living (such as IT and human capital investment) go hand-in-hand with greater exposure to international market forces. The results of both papers yield the observation that technological change and global forces engender trend and structural change for businesses and labor. Gaining from these forces may require that workforce and business change what they do to take advantage of new growth and occupational opportunities, whether those be at home or abroad. General Discussion: Robert Litan reflected that some of the results reported by Desirée van Welsum and Xavier Reif may come from the vulnerability of particular occupations to technological changes, and not to their offshorability. He suggested this as a possible explanation for the finding that the share of employment vulnerable to offshoring in the United States has been declining, a finding that he saw as counterintuitive. If so, these jobs would be disappearing, with or without offshoring. Perhaps not surprisingly, some participants had questions about the particular classification of some occupations in the paper by van Welsum and Reif. For example, Litan had not thought of air traffic controllers as vulnerable to offshoring. Van Welsum explained that there had been extensive discussions about how to classify some of the occupations. In some countries, air traffic controllers can be located up to a thousand kilometers away from the airport for which they work, hence the classification of this occupation as offshorable. Litan also pointed out that the United States uses somewhat different occupational classifications than the European countries reported, which makes cross-country comparisons difficult.

202


Brad Jensen noted that his look at the data does suggest that importing from affiliates can help a multinational firm to survive. In this sense, he concurred with points made by Maria Borga. A number of participants made suggestions regarding the empirical analyses in the two papers. Some expressed concerns about possible sample selection issues such as survivor bias, in the paper by Borga. An appendix has been added to the revised version of the paper to address these concerns. Others noted that it would be interesting to explore recent developments using price and wage data.

T. N . S R I N I V A S A N Yale University

Information-Technology-Enabled Services and India’s Growth Prospects

D

uring the first three decades (1950–80) of India’s planned insular economic development, real GDP grew at an annual average rate of around 3.75 percent. The 1980s saw a limited opening of the economy and hesitant reforms. The growth rate accelerated to 5.7 percent, fueled by fiscal profligacy financed in part by external borrowing at high interest rates. A severe macroeconomic and balance-of-payments crisis in 1991 following the first Gulf War, the collapse of the Soviet Union (which was not only India’s model for planned economic development but also its arms supplier, a partner for barter trade, and a supporter of India’s interests in the Security Council of the United Nations), and the fear of being left behind by the rapid growth of China since its opening in 1978 led Indian policymakers to break away from its inward-oriented, state-directed, and controlled development strategy and open the economy to external competition and investment. After addressing the crisis with the assistance of the International Monetary Fund and the World Bank, the policymakers launched a process of systemic economic reforms that is still in progress. The economy responded to the reforms and quickly rebounded from the crisis-induced fall in growth of GDP to 1.3 percent in 1991–92. The growth rate accelerated, peaking at 7.8 percent in 1996–97. Subsequently it has fluctuated, falling to a low of 4.0 percent in 2002–03, largely because of a severe drought-induced decline in agricultural output, and rising to a peak of 8.5 percent the very next year, in large part owing to the recovery of agricultural output (MOF 2005, appendix table 1.6). The latest available data show that in the fiscal year of 2004–05, GDP growth was estimated at 6.9 percent,

I thank Lael Brainard, Susan Collins, Kanwal Rekhi, AnnaLee Saxenian, and Nirvikar Singh for their comments.

203

204


and in fiscal year 2005–06 it is also expected to be around that level (RBI 2005a, table 1). India’s record of sustained growth since 1980 is second only to China’s among large economies, so much so that, in discussions about global economic prospects generally or about global demand for natural resources (including, most important, fossil fuels), the impact of Chinese and Indian growth is explicitly mentioned. However, India has succeeded only modestly in raising its share of world merchandise trade to 0.8 percent in 2004 from a low of 0.5 percent in 1983, while China’s share has more than quintupled, from 1.2 percent to 6.4 percent, during the same period. China ranked third from the top in the share of world merchandise trade in 2004, while India ranked a distant thirty-first in 2003 (WTO 2005, appendix tables 1 and 3; WTO 2004, table I.3). In global trade in commercial services (in which trade in information-technology-enabled services [ITES] and business process outsourcing [BPO] are included),1 India has done relatively better, with a share of 1.5 percent and a rank of twenty-one in 2004 as compared to China’s 2.8 percent share and ninth rank (WTO 2005, appendix tables 2 and 4). Both China and India are expected to gain a significant share of the global market of textiles and apparel with the expiration of the infamous Multifibre Arrangement. Apparel imports from China are already being targeted for restrictions by the United States and the European Union, and China itself is restraining exports in anticipation. Similar actions against Indian exports are possible, although India as a founding member of the World Trade Organization (WTO) is not subject to the special provisions of China’s Agreement of Accession to the WTO that have been used to restrict China’s exports. India’s software services exports in 2003–04 amounted to $12.2 billion, or nearly half the total services exports of $24.9 billion. Earnings from ITES and BPO accounted for another $3.6 billion (RBI 2005b, p. S343; MOF 2005, p. 111).2 A high-level strategy group set up by the All India Management Association (AIMA), comprising leaders from industry, academia, and the government, deliberated on the opportunities for providers in India in the growing global market for remote services (IT services such as software, ITES, telemed1. As will be clear from my subsequent discussion, a range of services is included under the umbrella of ITES. BPO is a significant enough category of ITES to be broken out separately; IT services themselves, which are treated in a different category, are also inherently “IT enabled.” 2. Data from Indian sources on India’s exports of computer and information systems to the United States and data on imports of the same services by the United States differ. On the reasons for this difference and adjustments to narrow it, see WTO (2005, box 2, p. 280). There is a presumption that Indian data on exports include the earnings of its nationals working in the United States on a temporary basis. They are also apparently included in the total employment of the sector in the Indian data.

T. N. Srinivasan

205

icine, and learning) and services sought by visitors to India such as conventional tourism, health care, and education.3 The group consulted wide segments of society and decisionmakers, and its findings were processed by a task force comprising AIMA, the Confederation of Indian Industry, and the Boston Consulting Group to formulate an action program. The task force concluded that lack of coordination among various stakeholders toward achieving a common goal was why India had been relatively slow in availing itself of the emerging opportunities (AIMA 2003). It views its report as a first step in a process of bringing about coordination and aligned actions. For the purposes of this paper, it is enough to note the growth and employment implications of the opportunities that the task force identifies. It finds that, by 2020, India can hope to generate $139–$365 billion of additional revenue from the supply of remote services for foreign residents and in situ services for visitors, pushing up the GDP growth rate by an additional 0.6–1.5 percent per year between 2002 and 2020. It estimates the direct and indirect employment generated by this additional growth to be between 20 million and 72 million.4 To put these numbers in perspective, India is targeting GDP growth of at least 8 percent per year in the next two decades. Its labor force, estimated at 363 million persons in 1999–2000 (MOF 2004, table 10.7), is expected to grow by 1.5 percent per year in the next two decades. If that happens, 125 million persons would be added to the labor force during the period. IT job growth projected by AIMA could provide jobs for a significant share of these additions to the labor force, assuming that each IT worker is fully employed. The methodology of projection by the task force is not explained by AIMA. The possibility that the projections of revenues and employment are very optimistic cannot be ruled out. However, for the very near term, official projections are also available (MOF 2005, box 6.2). The Ministry of Finance expects value added by the IT sector (including ITES) to grow to 7 percent of GDP by 2008 from around 2.64 percent in 2003–04. Exports of this sector are expected to be between $57 billion and $65 billion, accounting for 35 percent of total exports 3. The last category does not directly come under the umbrella of ITES, but just as the development of IT services created positive spillovers for ITES, the latter, by improving information flows and infrastructure, support the development of tourism and health and education services for nonresidents. Thus this third category is also discussed in this paper. 4. AIMA (2003, p. 16). WTO (2005, p. 283 and appendix table 8, p. 301) cites India’s National Association of Software Service Companies (NASSCOM) as its source for data on employment in India’s software industry. It reports employment of 813,000 in 2003–04 and employment growth of 21 percent in that year alone. If this rate of growth is sustained over sixteen years, employment would grow to 17.2 million, generating additional employment of 16.4 million in the software industry alone between 2003–04 and 2019–20.

206


in 2009, a rise of 14 percent from 2003–04. ITES/BPO exports rose from $0.57 billion in 1999–2000 to $3.6 billion in 2003–04 and are expected to rise to $21–24 billion by 2008. The fact that India’s share of world IT spending was only 3.4 percent in 2003–04 suggests that there is still considerable potential for significant further growth. Services imports from and BPO to India have attracted and continue to attract media attention in the United States. They have also evoked a protectionist response: “State legislators have introduced at least 112 bills in 40 states to restrict outsourcing till March 17 this year [2005]. . . . The New Jersey bill, which awaits the governor’s decision to sign or veto, would be the most farreaching anti-outsourcing measure in the country by prohibiting all state contract work from being performed overseas.”5 All this is indicative of the fact that not only within India, but also in the rest of the world, India is expected to be a major player in the IT industry in general and in BPO in particular. The emerging consensus is that India will continue to grow rapidly in the next several decades, and that its IT sector, broadly speaking, will contribute significantly not only to GDP growth but also to employment generation and poverty alleviation. In what follows, I trace the development of India’s IT sector and the continuing role of the Indian IT diaspora in the Silicon Valley in the United States. I then briefly discuss the possible role of IT in the growth process and as a source of dynamic comparative advantage and look in some detail at the prospects of and possible constraints on India delivering the high expectations about its IT sector.

The Development of India’s IT Sector Prima facie, it is a surprise that India has been able to achieve as much as it has in IT development.6 India is still a low-income country with gross national income per capita of $540 (ranked 159th from the top) using World Bank’s Atlas method of calculating exchange rates, or $2,880 at purchasing power parity exchange rates in 2003 (146th from the top). Even the IT indicators for India are not impressive: with 7.2 personal computers per 1,000 people in 2002, India is 5. Suman Guha Mozumdar, “BPO Scare Intact: 112 Anti-Outsourcing Bills Moved,” India Abroad, May 6, 2005, p. A26. The United States is among thirty or more members of the WTO who are signatories to the plurilateral code on government procurement. The New Jersey measure could be in violation of the code. However, since India is not a signatory to the code, it cannot avail itself of the provisions of the code to dispute the measure. 6. This section draws on Kapur (2002) and Saxenian (2002a).

T. N. Srinivasan

207

at just about the average of 6.9 for low-income countries and one-quarter of China’s 27.6. Indian has 17 Internet users per 1,000 people, just about the average of 16 for low-income countries, but still only one-quarter of China’s 64. India spent 3.7 percent of its GDP on IT in comparison with China’s 5.3 percent (World Bank 2005, tables 1.1 and 5.11).7 Given this apparent backwardness, Kapur rightly asks why India emerged as a leader “in a leading edge industry when, despite strenuous (and, in retrospect, misguided) policies, it failed to achieve such leadership in any other technology-intensive sector” (2002, p. 93) (with the possible exception of pharmaceuticals). He is again right in rejecting as inadequate and incomplete the explanation that the onerous economic control regime that was in place, certainly during 1950 –80 and arguably until the reforms of 1991, had not intervened in the software sector, and given India’s endowment of science and technology manpower, comparative advantage considerations would have enabled the development of the IT sector anyway. At a deeper level, why indeed the state’s role eventually became more facilitative than constraining in this sector remains to be answered. Saxenian (2002a) points out that it is not entirely appropriate to conclude that the self-sufficiency-oriented, insular development strategy of India since 1950 did not affect the development of the IT sector adversely; indeed it did, by restricting imports of computer hardware (even if the importer committed to exporting a certain amount of software) through high tariffs and limits on foreign exchange allocations, and above all by insulating Indian industry from its global counterparts. She notes that IBM was forced to depart from India in 1978, primarily because of its refusal to comply with the requirements of India’s draconian Foreign Exchange and Regulation Act. Kapur (2002) suggests that the departure of IBM and heavy protection of domestic hardware raised the relative costs of hardware and technology acquisition.8 The rising costs induced the industry to develop software skills in 7. According to UNDP’s (2005) technology achievement index (a composite of disparate indexes ranging from patents granted to residents to gross tertiary education enrollment ratio), India ranked sixty-third, behind Trinidad and Tobago with a rank of forty-one and China with a rank of forty-five! India, though included among the group of dynamic adopters, barely managed to escape being included in the groups of marginalized countries (Nicaragua, ranked sixty-fourth, leads the group of marginalized). This says more about the dubious value of this index and others (such as the Human Development Index) put together by the UNDP than about the technological achievement of the countries ranked and groups. 8. It is interesting that the Chinese prime minister started his April 2005 visit to India at Bangalore, India’s IT capital, and there were euphoric statements at official and unofficial levels of the possibility of joint efforts to capture a large share of the global market by capitalizing on China’s capabilities in hardware and India’s strength in software.

208


response. While this is plausible, until the disincentives of the autarkic system were attenuated (and indeed reversed and turned into incentives), the skills so developed would not have led to the spectacular growth of software exports in particular, and of the IT sector in general. Both Kapur and Saxenian note that there was a dramatic turning point in the policy environment for India’s software and IT industries after Rajiv Gandhi became prime minister in 1984. The major policy changes made by his regime included recognition in the Computer Policy announcement of November 1984 of software as an “industry” entitled to the investment and other incentives available for domestic industries; lowering of import tariffs (from 100 percent to 60 percent) on software and personal computers; and the announcement in 1986 of the Computer Software, Development and Training Policy, which liberalized access to the latest technologies and software tools for promoting the domestic software industry, with the expectation of its becoming globally competitive, moving up the value-added ladder, and capturing a significant share of global software exports. The policy allowed import of software in any form, invited foreign investment, and promised access to venture capital. There cannot be a more dramatic departure than this policy from the strategy of technological selfreliance, import substitution across the board (from intermediates to capital goods), and export pessimism. Another important event of the 1980s was the visit to New Delhi in September 1989 by Jack Welch, then chairman of General Electric (GE), and his breakfast meeting with Sam Pitroda, the chief technology adviser to Prime Minister Rajiv Gandhi. It led to GE’s technology partnership with India, which began in 1991.9 The 1984 and 1986 policy initiatives were enabling in the sense of removing policy-created barriers to the growth of the software sector, but the policies did not become proactive until after the reforms of 1991. Saxenian quotes an indus9. “India today earns more than $17 billion from corporations world-wide seeking low-cost overseas talent to do everything from write software to collect debts to design semiconductors. GE in large measure stoked the phenomenon, playing an unheralded role as the Johnny Appleseed on India Inc. and reaping billions in savings for itself along the way. But the strategy has been pivotal for GE. In 2000, it inaugurated a Jack F. Welch Technology Centre in Bangalore that employs thousands of researchers working on everything from new refrigerators to jet engines. This year, the conglomerate plans to spend about $600 million on computer-software development from Indian companies, according to a recent company report. The company estimates that similar products would cost it as much as $1.2 billion in the US. GE also recently unleashed a big new player in Indian outsourcing. In November, it sold a controlling interest in GE Capital International Services, or Gecis, a company with about 17,000 employees that GE started in 1997 to answer mail from its credit-card customers.” Jay Solomon and Kathryn Kranhold, “Early Investments Helped Fuel Tech and Service Sectors,” Wall Street Journal, March 23, 2005.

T. N. Srinivasan

209

try analyst: “Until 1991–92 there was virtually no policy support at all for the software sector. Even the term ‘benign neglect’ would be too positive a phrase to use in this connection” (2002a, p. 172). She notes the lack in the 1980s of international communication links for software exports and cites the example of Texas Instruments which, in setting up the first earth station in Bangalore in 1986, had to negotiate with the authorities for the removal or breaking of twentyfive different government rules relating to export of data via satellite links. The absence of reliable telecommunications links forced Indian firms to be primarily “body shoppers,” who provided programming services on-site, typically in the United States, to customers under contract. All of this changed in the 1990s. Telecommunications reforms, which began in the mid-1980s and included privatization and the creation of a regulatory agency, were very successful, although the process was not smooth because of the resistance of state-owned monopolies. Telephone calls within and from India are perhaps the least expensive in the world. The Department of Electronics introduced the scheme of Software Technology Parks (STPs) in the early 1990s. An STP is the analogue of an export-processing zone. Firms in STPs were allowed tax exemptions, guaranteed access to high-speed satellite links, and provided with reliable electric power and basic infrastructure, including core computer facilities, ready-to-use office space, and communications facilities. They were allowed to import equipment duty-free and without import licenses. Full (100 percent) foreign ownership was permitted in exchange for an export obligation. Firms were also allowed to repatriate capital investment, royalties, and dividends freely once they paid the taxes due.10 The STPs played a major role in the development of the IT sector in the 1990s. The share of units located in STPs in India’s software exports rose dramatically, from 8 percent in 1992–93 to 81 percent ten years later (WTO 2005, box 1, p. 274). However, there is no hardheaded social cost-benefit analysis of the use of public resources in the creation of STPs and the provision of other incentives. There were also several somewhat fortuitous factors. First, as Kapur notes, the restrictions on large business houses entering new fields under the Monopolies and Restrictive Trade Practices Act (which was repealed de facto only after 1991) prevented most of them from exploiting the newly emerging opportunities in the IT sector. Only one of the very successful IT enterprises (namely, Tata 10. China provided all of these incentives and more (in particular, complete flexibility in hiring and firing) in its special economic zones (SEZs) to attract investment, particularly foreign direct investment that was oriented toward export markets. But India’s STPs focused only on software. India’s later embrace of SEZs did not attract much FDI or lead to rapid growth of exports, in contrast to what happened in China.

210


Consulting Services) belonged to a large industrial house. Although large houses had substantial capital and other resources at their command, and precluding or restricting them could have been crippling, this did not happen in the emerging software and IT service industry since it was not very capital-intensive. Besides, small and medium-sized enterprises are usually the most innovative elsewhere in the world, and the Indian IT enterprises were no exception. Second, India had invested disproportionately in higher education in general and, in particular, had established the elite Indian Institutes of Technology (IITs). Although IIT graduates emigrated in significant proportions to the United States and accounted for less than 5 percent of the engineering graduates in India, the impact of émigrés on the development of India’s IT sector, particularly in the 1990s, was significant (more on this in a later section). As in the United States, India’s IT sector is also concentrated, located in clusters in Bangalore, Chennai, Hyderabad, Mumbai, New Delhi, and Pune. These cities also had the highest concentration of public sector R&D establishments (especially defense) as well as publicly funded engineering colleges. Five of these (other than New Delhi) are in the West and in South India. The states in which the five are located, namely Maharashtra (Mumbai and Pune), Karnataka (Bangalore), Tamil Nadu (Chennai), and Andhra Pradesh (Hyderabad), together accounted for 64 percent of the annual intake of engineering colleges in 2003, even though their share of India’s population is only 27 percent (Forbes 2003, table 3). Indeed, these were the states that had the largest expansion of engineering colleges since 1983 when, in a major liberalization, privately funded institutions without state aid were encouraged. In Forbes’s view: “It is this expansion of engineering education that fueled India’s software boom, and it is no accident that the states with massive private expansion of engineering education are precisely those where the software industry is located” (Forbes 2003, p. 8). The Y2K crisis brought prominent global attention to the skills of Indian software engineers. Indians had become good at converting old mainframe and minicomputer applications to Unix applications in the 1980s and early 1990s. It was essentially a tedious task that Americans were not willing to undertake. However, it directly prepared Indians for Y2K by giving them expertise in applications and also the reputation for reliable work. Y2K work accelerated the process, but the previous application conversion work was the base on which the whole thing was built.11 According to DataQuest, 11. I thank Kanwal Rekhi for pointing this out in a private communication.

T. N. Srinivasan

211

Y2K . . . triggered off a chain of events. Exports grew on the strength of Y2K and never looked back. The local training industry boomed partially because of it. . . . Locally, full-page government ads in leading dailies, asking companies to become “Y2K-compliant,” helped the paranoia along. By the end of 1999, the industry was on an all-time high. IPOs of software companies were getting oversubscribed several times. . . . This gave rise to a minor scandal that rose and fell—fly-by-night operators who often had no business at all started putting IT in their names and entering the IPO market. They made money; a lot of investors lost theirs.12 Whether or not it was fortuitous, the 1986 report of the Rangarajan Committee on Modernization of India’s largely state-owned banking sector recommended “standardizing banking systems on Unix, then an unperfected operating system when compared with MS-DOS. The government floated a tender for 400 Unix systems and set off a scramble among Indian companies to come up with a Unix platform. Though the local part of the contract eventually went to Sunray Computers, the report led local vendors into the Unix arena and eventually India’s transformation into a ‘Unix country.’”13 It is possible that Unix would have been attractive to a developing country such as India even had the committee not recommended it: it was a semi-open source developed for larger, more powerful computers than PCs, for which the totally proprietary MS-DOS was developed. Other competitors for Unix were also totally proprietary. Later, in the 1990s, Unix turned out to be ideal for networked computing, and Unix-based systems still dominate the Internet server realm.14 The sources of the spectacular development of India’s IT sector are diverse. There is no doubt that the foundation of a skill base existed for its development in the 1980s, in large part due to public investment in higher education and the creation of elite engineering schools. The public policy regime mattered, both negatively in restricting the potential development of the sector until the mid-1980s, and positively when it changed gradually from enabling its development before the reforms of 1991 to proactively supporting its growth thereafter. Fortuitous or serendipitous factors, depending on one’s point of view, contributed as well. From its beginning in body-shopping and routine programming, the industry has grown

12. DataQuest, “The Hot Verticals: The Great Indian Software Revolution,” December 23, 2002, p. 5 (www.dqindia.com/content/20years/102122306.asp [September 13, 2005]). 13. DataQuest, “The Hot Verticals,” p. 3. 14. I thank Nirvikar Singh for pointing this out.

212


Table 1. The Indian IT Industry, 2003–04 Domestic 2003–04

Growth, 2003–04 (percent)

Exports Value, Growth 2003–04 a (percent)

Units

Valuea

Units

Value

Hardware Systems: Servers Total non-PC servers Total PC servers Total servers

4,393 52,609 57,002

1,170 912 2,082

14 50 46

13 34 21

… … …

… … …

Systems: Workstations Personal workstations Traditional workstations Total workstations

12,078 3,100 15,178

99 122 221

73 29 61

13 44 28

… … …

… … …

Systems: Single-user systems Desktops 2,691,823 Notebooks 90,680 Total PCs 2,782,503

8,014 736 8,750

21 88 22

8 80 12

… … …

… … …

Total systems

2,854,683 11,053

23

14

…

–100

Peripherals Total printers/MFD Other peripherals Total peripherals

1,382,993 … …

1,333 3,163 4,496

21 … …

33 54 47

… 2,200 …

… 72 …

… 2,978 … 18,527 … 3,142

… … …

32 23 39

… 2,300 10

… 59 –67 (continued)

Networking Total hardware Total hardware services

in depth and scope. I conclude this section with a description of the industry as of 2003–04. The hardware component of the industry is still small, accounting for a little over one-fifth of the total value of output (see table 1). Exports accounted for nearly two-thirds of total output, of which software and BPO services had shares of 67 percent and 28 percent, respectively. It employs fewer than a million workers out of a labor force of 363 million. According to India Today, “Still, the buoyant BPO sector is absorbing English-speaking graduates in the thousands. In 2005, IT & ITES will be the biggest job generator, creating more than 2.75 lakh [275,000] jobs. India’s huge cost advantages with quality assurances and large pool of skilled manpower will keep the going smooth. More than 250 of the Fortune 500 firms outsource their IT needs to India. There is more growth

T. N. Srinivasan

213

Table 1. The Indian IT Industry, 2003–04 (Continued) Domestic 2003–04

Growth, 2003–04 (percent)

Exports Value, Growth 2003–04 a (percent)

Valuea

Units

…

1,710

…

–10

27,755

18

… … …

3,300 1,951 6,961

… … …

30 69 24

1,586 10,309 39,650

–8 11 15

…

1,450

…

53

16,380

45

Training Services Total training services Total services

… 992 … 12,545

… …

–12 26

110 56,150

29 22

Grand total

… 33,374

…

24

59,550

24.5

Units Software Services Customized software Turnkey projects Consulting/others Total software services Business process outsourcing (BPO)

Value

Source: DataQuest, “Intellectual Property: India: Sleeping IP Giant” (www.dqindia.com/dqtop 20/2004 [December 2005]). a. Rupees Crore (107 rupees); US$1= 48 Rs.

in store” (India Today, March 2, 2005, p. 14). The same story reports that the top five IT and ITES firms pay an annual salary of $4,500–$6,250 for an engineering graduate, roughly ten to twelve times the per capita income of the country, one-sixth of the annual average U.S. salary in 2000, but less than that of an employee of the occupations at risk of being offshored in the United States (see also Bardhan and Kroll 2003, table 4). The Indian industry is no longer confined to producing and exporting low-end software products and services. Several multinational companies (MNCs), including many leading ones, have established software development centers in India. DataQuest reports that such MNC centers are filing for patents in large numbers (1,108 in 2002–03).15 It suggests that intellectual property revenues would constitute a major chunk of a software company’s revenue in the future, and Indian companies (other than MNCs), including some of the large ones, have not yet started preparing for it. Leading Indian IT firms, such as Infosys and Wipro, are multinational with offices around the world and employ nationals in these countries. Infosys has alliances with the world’s leading firms, including 15. DataQuest, “Intellectual Property: India: Sleeping IP Giant” (www.dqindia.com/dqtop20/ 2004 [December 7, 2005]).

214


IBM, Intel, Microsoft, and Oracle, and also has made strategic acquisitions of foreign firms. The story is similar for Wipro; it is aggressively looking at companies to acquire in the enterprise, finance, and consulting areas. NASSCOM (2004, p. 9) documents the increasing maturity of the industry following a large number of mergers and acquisitions in 2002. It noted that traditional IT service players have added ITES-BPO portfolios to their existing offerings in order to provide a complete umbrella of end-to-end services. Multivendor and build-operate-transfer (BOT) contracts that offer customers advantages such as low risk, scalability, and competitive pricing have increased. Indian vendors are expanding the spectrum of their service offerings in client locations and even setting up facilities in other low-cost ITES-BPO destinations such as China and the Philippines in order to tap those markets. They are also moving up the value-added ladder to offer high-end services such as equity research and analytics, as well as insurance and technology support and development. Moreover, Indian vendors have moved far beyond call centers into financial services, telecom, retailing, and automotive segments of the ITES-BPO sector. In financial services, Indian companies are offering customers services centered on accounting, billing and payment services, and transaction processing. Over the past few years, some Indian service providers have also been offering highervalue services to customers in the areas of insurance claims processing and equity research support. They expect to gain from offshore outsourcing of customer and technical support and product development by the global telecom industry; transaction processing, billing, telemarketing, and inventory management for large retailers; and engineering activities, such as computer-aided product and tool design, claims processing, and accounting processes for the automobile industry (NASSCOM 2004, p. 10). The report also benchmarked the performance of Indian industry on key operational issues with global benchmarks. It found that Indian industry is able to deliver at levels comparable to their international counterparts on parameters such as quality, customer satisfaction, and between quality and customer satisfaction. Finally, let me turn to some relatively recent developments in the provision of services to tourists and visitors to India. As the home of an ancient civilization and of several ethnic and religious groups (Hindus, Buddhists, Jains, Sikhs, Muslims, Christians, Jews, and Zoroastrians) over millennia, India has many ancient monuments, including churches, mosques, synagogues, and temples besides the Taj Mahal. Of course, the diversity of its flora and fauna, the peaks of the Himalayas, and other natural beauty also attract tourists. However, because of the inadequacy of affordable quality hotel rooms, transport, and communications, India failed to attract as many tourists as even much smaller

T. N. Srinivasan

215

countries of the region. This is changing: tourist arrivals grew by 24 percent in 2004, and India is fast emerging as one of the top ten tourist destinations in the world, according to the Reserve Bank of India (RBI 2005a, p. 80). However, unconventional tourism is also on the rise. In particular, medical tourism is growing. According to several recent newspaper reports, the availability of worldclass treatment at a fraction of the cost in the United States or Europe is attracting up to 100,000 to 150,000 foreign patients to quality hospitals across India, representing a tenfold increase over just five years ago. Many of these hospitals were established and are staffed by Indian doctors who had emigrated to and practiced for a long time in the United States and Europe. Another unconventional service is education—institutions of higher education are attracting students from other South Asian countries. Recently, two premier institutions, the Indian Institute of Management at Bangalore and IIT in Mumbai, have started programs in Singapore in collaboration with the National University of Singapore. Also, drug trials and sponsored research in the pharmaceutical industry are being outsourced to India.16

The Indian Diaspora and ITES Sector The IT revolution, particularly the development of the Internet, has spawned networks of engineers and scientists who are transferring technology, skill, and know-how between distant locations, and more flexibly than most corporations.17 Among foreigners in the United States, Indians were second only to South Koreans as recipients of U.S. Ph.D.’s in engineering and science in 2003. Indian engineers were at the helm of a significant and growing number of Silicon Valley–based technology companies. The proportion of companies run by Indians has grown from 3 percent of those started between 1980 and 1984 to 10 percent of those started between 1995 and 2000, and is probably even higher 16. India’s success in producing generic equivalents of patented drugs is owed in part to India’s decision since 1970 to grant only process and not product patents in the pharmaceutical industry. This success became internationally visible when Indian companies became able to supply generic antiretroviral drugs to treat AIDS far more cheaply than the multinational companies. With India becoming a signatory of the TRIPS (Trade Related Aspects of Intellectual Property Rights) agreement, Indian patent law had to be amended in March 2005 to grant product patents as well. There is a real danger that the generic industry will be destroyed by this amendment (see Abbott, Kapczynski, and Srinivasan 2005). On the other hand, because the Indian pharmaceutical industry has built a reputation for quality, drug trials and sponsored research are being outsourced to India. 17. This section draws on Saxenian (2002b).

216


among those started after 2000. Further, data from the U.S. Bureau of the Census show that the Indian share of Silicon Valley’s science and engineering grew from 2 percent (400 workers) in 1970 to a significant 13 percent (20,000 workers) in 2000, most of the increase having taken place in the 1990s.18 Members of the two professional associations of Silicon Valley, The Indus Entrepreneur (TIE) and Silicon Valley Indian Professionals, and those who held senior positions in U.S. companies, were instrumental in convincing their senior management to establish operations in India. As India built a reputation as a supplier of software, most Silicon Valley technology companies established their own development centers in India. The dot.com bust resulted in some Silicon Valley–based Indians returning to India and setting up enterprises of their own. Saxenian (2002b) rightly stresses the growing influence of the Silicon Valley Indian community in Indian policymaking. One of the doyens of the community, often called the “sage of Silicon Valley,” Kanwal Rekhi, former chief technology officer at Novell, has been a vocal advocate of policy change in India. In his regular visits to India, he has met with senior policymakers at the central and state levels. He deserves much of the credit for the breathtaking scope of India’s telecommunications reform and its success. But for his very public attack on the entrenched bureaucracy of the Department of Telecommunications (DOT), it is unlikely that the reform would have proceeded very far.19 K. B. Chandrasekhar, another Silicon Valley entrepreneur, led a committee in 1999 on venture capital (VC) for the Securities and Exchange Board of India. The committee’s report provided a comprehensive vision of the growth of India’s VC industry. The industry has had remarkable growth since the report. In 2004 nonresident Indian (NRI) entrepreneurs led by the Silicon Valley giants prepared an action plan for attracting substantial foreign direct investment (FDI) to India and presented it to the government in January 2005. It has been well received and some of its recommendations have already been implemented. There is no doubt that their success in the global market is the main, if not the only, reason for their not only being heard with respect but actively sought after by policymakers. Three prominent Silicon Valley entrepreneurs, Vinod Khosla (of Sun Microsystems fame), Kumar Malavalli (of Brocade Communications Systems), and Kanwal Rekhi, realized that to change policy successfully, three ingredients are essential: sound, policy-oriented research from academics to provide the foun18. I thank AnnaLee Saxenian for drawing my attention to these data. 19. He coined the Hindi slogan “DOT Hatao-Desh Bachao,” which means “Curb the DOT and Save the Nation!”

T. N. Srinivasan

217

dation for policy advice; discussion and debate of the proposed policy advice in a forum consisting of academic researchers, prominent politicians, bureaucrats, and media representatives; and monitoring of the implementation of the advice. The three took the initiative to provide significant resources to Stanford University’s Center for International Development for sponsoring and undertaking policy research on India’s economic reform and organizing a conference at which research findings and policy recommendations following from them are presented. Five annual conferences have been held since 2000; two conference volumes have been published, and a third is in press. The sixth conference was held in June 2005. Besides supporting research at U.S. universities, the Silicon Valley community has also been active in upgrading Indian institutions, starting from the IITs, of which many of them are alumni, and also creating new institutions, such as a business school in Hyderabad whose faculty includes regular visitors from top U.S. business schools. Some Silicon Valley entrepreneurs, hailing from different states in India, are realizing that policy reforms are urgently needed at the state level in India. A group of TIE members interested in Kerala have formed a Kerala Support Group. Others are likely to follow. Finally, at the request of India’s prime minister a group of entrepreneurs and business persons in the Indian diaspora, including veterans of Silicon Valley, got together in fall 2004 and came up with practical suggestions for improving FDI flows to India. Their proposals were presented to the prime minister and the Planning Commission in January 2005. The subsequent removal of some dysfunctional restrictions on FDI and the legal action creating special economic zones were in part the result of these proposals. It is clear that the influence of the Indian IT diaspora is spreading beyond the narrow confines of the Indian IT industry to influence India’s economic development and growth more broadly. The demonstrable success of India’s IT and software sector in the global market is having a profound impact on Indian industry and also, importantly, on the educated youth. From a fear of competition, as evidenced, for example, by the demand for antidumping measures (ADMs) against cheap imports, particularly from China (a demand that the government was too willing to agree to, making India the largest user of ADMs among members of the WTO in the last couple of years), there is now a growing sense of confidence among Indian industries of being able to compete and a desire to view China as a large and growing market to export to, with the result that trade with China has grown rapidly. An important aspect of achieving and maintaining global competitiveness is attention to product and service quality. Again, the reputation for quality that IT vendors have built has not gone unnoticed. India is now poised to become a major destination for outsourcing of

218


manufactured products, as China is already. This is already happening in auto components—Indian suppliers, like the Chinese, are competitive both in cost and quality (Sutton 2005). Indian automakers have recently begun exporting passenger cars: from 25,000 cars in 1998–99, exports grew to 121,000 in nine months of 2004–05 (MOF 2005, table 7.5). Last, there is a sense of confidence among the educated youth about the future and in their ability to compete successfully with the best of their cohort from anywhere in the world, a confidence that was noticeably absent not so long ago. These intangible effects are no less important than their tangible ones on the growth of exports, GDP, and employment. Among the most important intangible effects of India’s perceived economic success broadly, and in high-technology sectors including IT and the pharmaceutical sectors in particular (both achieved without in any way compromising its vibrant democratic political system), one must include a vast improvement in India’s international standing. There is a recognition that India is a major Asian power, along with China and Japan. It is no coincidence that, in a short span of six weeks in March and April of 2005, high-level visitors to India included the Chinese prime minister, the U.S. secretary of state, the president of Pakistan, and the UN secretary general. India is a serious contender for permanent membership on the UN Security Council if the proposed expansion of the council comes about, though this seems unlikely in the foreseeable future.

The Analytics of the IT Sector in the Growth Process There are many possible channels through which IT could affect an economy’s output (its level and possibly rate of growth in a steady state). Singh (2004) focuses on one, namely the reduction of transaction costs through the use of information technology.20 In his model, a reduction in transaction costs increases the number of intermediate goods that are produced and in turn influences growth. A representative household supplies labor inelastically and has a logarithmic instantaneous utility function for a single consumption good produced by identical competitive firms with constant returns to scale technology using labor and a symmetric composite of differentiated varieties of intermediate goods best viewed as perishable producer services. The composite is a con20. Information technology can reduce transaction costs in several ways, for example, by reducing search and matching costs, by speeding up and making more reliable the completion of transactions, by substituting long-distance communications for physical transportation, and by improving tracking and logistics of delivery. These benefits can be obtained in transactions for intermediate or final goods and services.

T. N. Srinivasan

219

stant elasticity of substitution (CES) aggregate of available varieties at each point in time, with an elasticity of substitution greater than one. This implies that aggregate production must be increasing in the number of varieties of intermediates (in a symmetric monopolistically competitive equilibrium in which all the available varieties cost the same and are used in the same amount). All varieties are produced with the same technology with a fixed start-up cost and a constant variable cost, both in units of labor. Finally, transaction costs are modeled à la Samuelson’s iceberg transportation costs: intermediate goods producers receive only an exogenously set fraction of what the producers of the consumer goods pay them; the remaining fraction melts away as a transaction cost. Thus any fall in this exogenous fraction is a reduction in transaction cost. There are no forces in this model that can generate sustained growth, since the labor force is given, and in contrast to Helpman’s (1990) model, there is no learning effect that continuously reduces the fixed cost of producing a variety of intermediates as their cumulative output grows. Thus Singh focuses on the properties of a steady state, if it exists. It turns out that the relative value of the elasticity of substitution between labor and the intermediates aggregate in the production of the consumption good and that between different varieties in the intermediate aggregate, , is what matters. Both are assumed to exceed one. If ≤ , there is a unique value n* for the number of intermediates n, such that if the economy starts with n0 < n*, it converges to n* over time, and if it starts with n0 > n*, it stays at n0. Thus initial conditions do not matter too much, in the sense that any economy starting from n0 below n* converges to it. However, n* is a decreasing function of the transaction cost. If > , there are three possibilities depending on other parameters, including the utility discount rate and transaction costs: (i) n stays at n0 , so that any initial n0, is also a steady state; (ii) there are two values n *L and n *H with n *L < n *H such that if n0 ≤ n*L, it stays at n0 ; if n *H ≥ n0 > n *L, n converges to n *H ; and if n0 > n*H, n stays at n0 . Thus any initial n0 in the interval (0, n *L ] or [ n *H ,) is a possible steady state. However, for an initial n0 in ( n *L , n *H) the steady state is n *H ; and (iii) again, there are two values n *L and n *H as in (ii), but in this case, even if n 0 < n *L , the n can converge to n *H so that for any initial n0 in (0, n *H ] including the value n *L, the steady state is n *H . Summarizing, transaction costs have three impacts: first, the standard deadweight loss; second, with higher transaction costs, few intermediates are produced, leading to lower output of the consumer good and hence lower welfare; and third, they may arrest the process of change in the sense of keeping the number of varieties of intermediates at their initial level and/or reduce their long-run level. Although the model is useful in illustrating the possible cost of high transaction costs, as noted earlier, its structure precludes the analysis of growth effects.

220


The model in Singh (2002), though not as fully worked out as the above, addresses the growth impacts of IT (including nondigital methods of storing and communicating information) through an extension of a model of recombinant growth developed by Weitzman (1998). The Weitzman model captures the simple concept that new ideas are formed through combinations of old ideas. Singh follows Weitzman in focusing on this concept to the case in which ideas can be combined in pairs. Given the possible combinations of pairings, the number of ideas grows exponentially. By decomposing the total stock of knowledge or ideas into the stock of IT knowledge and non-IT knowledge, Singh captures the special role of IT knowledge. He specifies that the stock of IT knowledge affects the growth in stocks of both types of knowledge in the same manner, so that IT stock gives the growth process an “extra kick” even beyond the exponential growth that the Weitzman model produces.21 Singh conjectures that the model has a steady-state growth rate, but its comparative statics with respect to the parameters of the “extra kick” term are yet to be worked out. Let me conclude this section by listing the mechanisms through which IT could affect the level and growth of output. First, as in the Singh models, IT services are in effect universal intermediates (like energy) that are essential to any production activity and possibly most, if not all, consumption activities. Thus any technical progress in the IT sector reflects itself first in productivity gains (or cost reductions) in the IT sector. Then, as the changes in the IT technology diffuse, as the rest of the economy makes the appropriate investments in equipment and processes to take advantage of the new lower-cost IT technology, other sectors of the economy will experience productivity gains. Because the diffusion process is likely to be gradual, the total factor productivity (TFP) gains will be spread over time. The debate in the United States about the contribution of computers and IT to productivity growth is in large part about this diffusion effect (see Jorgenson 2005; Gordon 2000). In India also, the diffusion that has begun would yield TFP gains over an extended period of time. The mechanisms through which IT affects the level and growth of output are essentially aspects of the diffusion process. First, as IT diffuses, improvements in the efficiency of resource use would affect the level of output growth both sec21. The justification of this special role for IT comes from the importance of being able to store, process, and communicate information effectively: without writing, without telephones, without the Internet, the success rate of converting potential new ideas into actual additions to the stock of knowledge would be lower. As an illustration, the use of IT makes the IT sector itself more efficient and innovative (for example, using software for automated testing during software development) as well as providing this benefit to other sectors (for example, using IT to improve the workings of call centers, or to reduce mistakes in medical transcription).

T. N. Srinivasan

221

torally and, even more important, intersectorally, as gainful transactions (market and nonmarket) that were previously unavailable became available. Already, improvement in market price information, and hence higher revenues, are being realized in some parts of rural India. Another area that positively affects both the level and growth of output is IT-based delivery of education at all levels.22 In principle, an extensive and effective use of IT in public administration could enhance transparency, as well as reduce avoidable time lags and transaction costs. It is also possible that increased transparency in itself could reduce administrative corruption and malfeasance. It is too early to judge from several ongoing e-governance pilot projects whether these expected beneficial effects are being realized (Singh 2002, p. 18). However, it is early enough to recognize the existence and rapid growth of several private (profit-oriented as well as nonprofit) initiatives in addition to publicly funded schemes in India. Singh mentions several of them. There is reason to hope that the diffusion of IT will gather steam and contribute significantly to accelerating growth in the not too distant future. Because IT goods and services are tradable internationally, productivity gains from their diffusion to the entire economy should occur regardless of whether such goods and services are locally produced. Obviously, whether a country produces an intermediate good domestically or imports it is the aggregate outcome of the decisions of its producers of final products to buy the good from local or foreign suppliers. Indeed, these decisions are related to the organization of firms in the sense that a vertically integrated firm producing a final product produces the intermediate products it needs, whereas less integrated firms would purchase some of the intermediates they need from others (in particular from suppliers who specialize in the production of intermediates); that is, they would outsource such intermediates. When such outsourcing leads to offshoring, the intermediates would be imported. The offshore producer could also be a subsidiary of the offshoring multinational firm, so the purchase transaction would be internal to the firm. Thus, in general the problem of whether to outsource or offshore has to be embedded in an analysis in which the organization of firms (whether to be vertically integrated, or whether to become multinational—that is, to engage in FDI) is endogenous. In a series of papers Grossman and Helpman model the determinants of the location of subcontracted (that is, outsourced) activity in a general equilibrium model of outsourcing as trade, integration versus outsourcing in the equilibrium 22. Singh (2002) cites the projects “e-choupal,” linking the Indian farmer with national and international markets, and TARAhaat as examples.

222


organizational structure of industries, and outsourcing versus FDI (2002a, 2002b, and 2003). An essential feature of these models is that outsourcing of an intermediate requires searching for a partner and making relation-specific investments that are governed by incomplete contracts. Naturally, finding a suitable partner is easier if the market for intermediates is thick (and thickness can differ between domestic and foreign markets). Also, whether the intermediate has to be customized to the specification of a single user (in which case, once it is produced, the supplier has no other user for it than the one for whom it was customized) or is usable by many, matters for contracting, since the possibility of the supplier, once he has produced the customized intermediate, being “held up” by the user can arise. IT products, such as some software, have characteristics that often involve customization. Moreover, such characteristics are at best observable by the supplier and purchasers, but not verifiable by third parties, thus precluding contracts between suppliers and purchasers that stipulate a given price for an agreed quantity to be purchased. For a closed economy Grossman and Helpman (2002b) analyze domestic outsourcing in such a context, based on the tradeoff between costs of running a larger and less specialized organization in a vertically integrated firm and the costs arising from the search for a suitable partner to outsource and imperfect contracting, were it to outsource. Turning to FDI, Grossman and Helpman (2003) set forth three possible equilibria. In one, all firms choose to invest in a subsidiary abroad to produce their intermediates there. In another, all choose to purchase them from independent suppliers in the foreign country. In the third, some firms opt to engage in FDI and others in purchase. Which of the three is relevant depends on the values of three parameters representing, respectively, the price elasticity of demand for any variety of the differentiated final product (assumed to be greater than one), the marginal cost of producing the substitute in a subsidiary abroad, and the difficulty of contracting with an independent foreign supplier. With marginal cost sufficiently low, elasticity of demand sufficiently low (that is, close to one), and the difficulty of contracting sufficiently great, in equilibrium all firms engage in FDI. By contrast, with marginal costs neither too low nor too high, elasticity of demand above 1 but by not too much, and a moderate difficulty in contracting, in equilibrium some firms do engage in FDI and others purchase. It is conceivable that the search and contracting costs will be lower if both supplier and purchaser happen to be national firms. If this is the case, the gains from having domestic firms supplying and other domestic firms purchasing IT products could be large in comparison with domestic purchasers buying and contracting with foreign firms. For this reason, it is likely that firms in non-IT sectors in India would pur-

T. N. Srinivasan

223

chase their IT services and products from existing domestic IT firms rather than import them. Also, if contracting difficulties are relatively fewer and marginal cost of production relatively higher in India than in China, the Grossman and Helpman (2003) model would predict more FDI in China than in India.

Future Prospects of the Indian IT Sector: Conclusions The spectacular growth of software exports as well as ITES-BPO has been underpinned both by the relative abundance of skilled workers and by their relatively low cost. NASSCOM (2004, p. 15) points out that the success of the Indian ITES industry is in large part due to the country’s immense pool of English-speaking skilled workers. NASSCOM (2005) estimates the stock of graduates (engineering degree and diploma holders, degree holders in the arts, commerce, and science) at 22 million in 2003, with approximately 2.5 million graduates in 2004 from about 275 universities and 14,000 colleges in the country. Of course, those numbers do not adjust for differences in the quality of skills and are not broken down by the skills required for working in different segments of the IT sector. For example, the skills required to be employed in a call center are surely different from those for developing software. NASSCOM (2005) estimates that 250,000 engineering degree and diploma holders entered the workforce during 2003–04, with a large segment believed to have joined the IT industry. The industry is estimated to have employed 841,500 professionals in 2003–04, of which 270,000 worked in the export sector and 253,000 in the BPO sector. Still, if one assumes that the share of the stock that is employed in the industry does not change, the implied annual growth rate of the stock at 11.3 percent (the ratio of graduates to stock) is far below the growth of 35 percent that NASSCOM (2004, p. 15) projects for the value of India’s ITES/IT services output for the period 2003–12. Thus, unless the ITES/IT sector’s share of the stock grows or the stock itself grows more rapidly, there could be excess demand for labor in the sector, and the wages and salaries of ITES/IT workers would have to increase to wipe out the excess demand: While demographic studies have suggested that India could be one of the few countries with a surplus of personnel within the employable age group by 2020, there is a possibility of a shortage in terms of availability of skilled personnel for ITES/IT, even in the medium term. This gap could be to the tune of 235,000 for IT services and 262,000 for IT-enabled services and could increase in 2012 in the absence of any

224


special efforts to meet the manpower requirements. (Government of India 2003, p. 6) Any increase in emoluments of ITES/IT workers in India with no change in their productivity would cut into the cost competitiveness of India. It is not easy to obtain reliable and well-documented unit costs of IT sources and products for Indian suppliers relative to their competitors in other countries. The following data from reports of Evalueserve, a full-service business intelligence, market research, and intellectual property services firm, are suggestive: the costs reported do not include profit margins, which run around 10–15 percent in India, but do include marketing and sales costs: For IT services, costs range from $11 per hour to $19 per hour for work being done in India. $11 per hour essentially corresponds to the lower end IT work that is being done by a college graduate (with a bachelor’s in computer architecture) or a graduate engineer. On the other hand, $19 per hour is for higher-end work being done by people with four or five years of experience. For ITES the costs range from $9 per hour to $22 per hour (for work being done in India); $9 per hour is essentially credit card processing and other low-end, nonvoice work. For low-end voice work (such as call centers), this goes up to $11 per hour. On the other hand, $22 per hour is for higher-end work like investment banking research, intellectual property research, etc. In either case, Indian companies typically charge for 2,050 to 2,100 working hours per year per full-time-equivalent (FTE), although an average FTE in India is currently working 2,300 hours, which is the same as in South Korea (Alok Aggarwal of Evalueserve, private communication with the author, August 15, 2005). A significant component of unit cost is labor. Table 2 puts labor cost differences in perspective. At the low end of the skill spectrum (the first three rows of occupations in table 2), India is likely to remain competitive for the foreseeable future. At the higher end of the spectrum (the last three rows), India’s competitive edge is likely to be eroded, with Indian wages rising relative to the U.S. wage. On the other hand, productivity and quality improvements that will mitigate, if not completely offset, cost increases are also taking place. Saxenian, in a private communication of May 5, 2005, points out that in interviews Indian IT professionals informed her of “the process and quality improvement in the largest Indian firms (TCS, Wipro, Infosys, etc.), especially in the period since

T. N. Srinivasan

225

Table 2. Hourly Wages for Selected Occupations, United States and India, 2002–03 U.S. dollars Occupation

Hourly wage, United States

Hourly wage, India

Telephone operator Health record technologists/medical transcriptionists Payroll clerk Legal assistant/paralegal Accountant Financial researcher/analyst

12.57 13.17 15.17 17.86 23.35 33.00 –35.00

Under 1.00 1.50 –2.00 1.50 –2.00 6.00 –8.00 6.00 –15.00 6.00 –15.00

Source: Bardhan and Kroll (2003, table 3).

2000 when substantially more work was shifted to India from the U.S., allowing for the accumulation of skill and learning on the job.” Her own research found that there are a “surprisingly large number of Indian firms that are CMM [capability maturity model, developed by the Software Engineering Institute at Carnegie Mellon University] Level 5 certified. There are none in China at Level 5 yet, and few even in the U.S.” Another interesting set of cost comparisons relates to medical tourism. Jay Solomon reported in the Wall Street Journal (April 26, 2004) that India’s “Apollo [hospital] offers cardiac surgery for about $4,000, compared with at least $30,000 in the U.S. Apollo’s orthopedic surgeries cost $4,500, less than one-fourth the U.S. price.” John Lancaster reported in the Washington Post (October 21, 2004, p. A01) on a patient with a life-threatening heart condition who would have had to undergo surgery at a cost in the United States of $200,000; instead he flew to New Delhi and had it done at Escorts Heart Institute and Research Centre. It cost him $10,000, including airfare and a side trip to the Taj Mahal. Not only are costs lower, but quality may be higher as well at India’s private centers of excellence, such as Escorts. Lancaster quotes Naresh Trehan (a former assistant professor at New York University’s Medical School) of the Escorts Centre as saying that the “death rate for coronary bypass patients at Escorts is 0.8 percent. By contrast, the 1999 death rate for the same procedure at New York-Presbyterian Hospital, where former president Bill Clinton recently underwent bypass surgery, was 2.35 percent, according to a 2002 study by the New York State Health Department.” A third report, by Saritha Rai (New York Times, April 7, 2005) speaks of a patient from England who needed a coronary bypass operation but would have had to wait six months to get it from British National Health Services. He traveled to Wockhardt Hospital in Bangalore for the surgery following a chance

226


meeting with a businessman who had gone to India for surgery. His surgeon there had trained in London, and the surgery cost $8,400, including travel. Moreover, according to the patient, his surgeon gave him his cell phone number and was available twenty-four hours a day. He concluded that in the British Health System “you are just a number, but here you are a person.” Such stories are frequently reported in the U.S. media. As I noted earlier, 100,000–150,000 foreigners currently visit India for medical treatment annually. A report by McKinsey & Co. projects a revenue of $2.3 billion by 2012 from medical tourism to India. It is not out of the realm of possibility that India could go beyond offering inexpensive but high-quality surgery at hospitals and develop as a destination for the elderly to live out their retirement years in a warmer climate and with better health care than they could obtain in the United States or Europe. Although the prospects for substantial growth in the ITES/IT sector, as well as in in situ services such as medical tourism, are very bright, realistically speaking there are several constraints besides a potential manpower shortage that could preclude their full realization. Reliable electric power, efficient and inexpensive telecommunications, and access to venture capital are essential infrastructures for the IT sector. Although telecommunications infrastructure has vastly improved, as noted earlier, there are still some unresolved issues relating to the authority of the regulatory agency (Telecommunications Authority of India, TRAI) vis-à-vis the Department of Telecommunications and the stateowned providers. The electric power situation continues to be abysmal. In fact, the large IT firms, like other large enterprises, have had to invest in their own captive power generation facilities. To the extent that the unit cost of power from small-scale captive plants is much higher than it would be from an efficient large-scale utility, the failure of India’s public power system adds an avoidable cost to doing business and dampens the competitiveness of its IT sector. Narayana Murthy, the CEO of India’s leading IT firm, Infosys, and Sandeep Raju of the same firm pointed out in 2002: Efficient commercialization of cutting-edge output from research labs, entrepreneurship forums at universities, highly efficient alumni networks, close links between leaders in academia and business, risk appetites of venture capitalists, synergies between science/engineering schools and business schools, collaborative research among universities, keiretsus bringing together businesses and venture capitalists, angels with the willingness to nurture talent, the abundance of forums where youngsters may put forth their ideas and interact with industry leaders, opportunities for

T. N. Srinivasan

227

collective learning—all these are differentiators that put the Valley several notches above other hi-tech habitats. In sum, Silicon Valley operates in a vibrant market economy that reveres innovation. That Bangalore entrepreneur, on the other hand, does not have easy access to all these resources. However, it must be borne in mind that the information revolution is a fairly recent phenomenon in Bangalore. (Murthy and Raju 2002, p. 201). Things have improved substantially since they wrote. The venture capital supply has increased, and some of the networks they mention are beginning to emerge. But significant close links between academia and business are yet to be forged. Young workers in the IT industry are conscious of their market value, so much so that turnover of workers in the ITES sector, particularly in smaller firms operating at the lower end of the quality spectrum, is high. In call centers, annual turnover is reported “to exceed 50 percent. High staff turnover is reported even among the more established, employee-friendly IT companies, some of whom offer stock options and residential accommodations to entice employees to stay on” (Chithelen 2004, p. 1023). High turnover indicates an excess demand and makes it more expensive for Indian firms to maintain and improve their competitiveness and quality of service.23 Although India’s labor force is very large, and more workers would like to acquire the skills to enter the IT sector, setting up training facilities would be costly and involve years of planning and implementation. Chithelen notes that the explosive growth in BPO demand has attracted new entrants to the business, and the resulting competition is reported to have bid down fees from $20 per programming hour for U.S. entrants currently from over $60 in 2000. He fears that greater competition among vendors, some of whom are aggressive but inexperienced new entrants, coming on top of labor supply constraints, could compound the decline in quality of service. India’s labor and bankruptcy laws could be counterproductive in the IT sector, as they are in other sectors of the economy. A report in 2000 by the Subject Group of Knowledge-Based Industries in the Prime Minister’s Council recommended exempting the IT sector from some of the draconian provisions of labor laws. Whether it is wise to exempt one sector from a dysfunctional law rather than repealing it is arguable. In any case, political support for a repeal is not there

23. Kanwal Rekhi points out in his private communication that the high turnover in the ITES sector is in the nature of the beast. This industry essentially employs youngsters who do not consider it a career. Many of them go on to higher learning. He cites the increase of Indian students in U.S. universities over the past several years as evidence of that.

228


yet. However, some de facto exemptions do exist. For example, in ITES, states often exempt call centers from working-hour restrictions, allowing women to work at night. Also, it is likely that programmers in large firms are not subject to the same provisions as industrial workers.24 Moving away from domestic to external constraints, I have already mentioned the protectionist backlash against outsourcing in the United States and elsewhere. Apart from this, the outcome of negotiations on services in the Doha Round of world trade talks is also important. In the parlance of the General Agreement on Trade in Services (GATS), there are four modes of services trade. From the perspective of outsourcing and trade in ITES/IT services more generally, the two modes that are particularly relevant for India are mode 1, which covers outsourcing—that is, the supply of a service from the territory of one member of the WTO to the territory of another member without movement of the use and provider of the service from where they are located; and mode 4— that is, trade by a service supplier of one member, through the presence of natural persons of a member in the territory of any other member. In plain language, mode 4 involves temporary migration of labor from the territory of one member to that of another. Members undertake commitments under each mode of supply. Thus far, members have made most cross-border commitments for mode 3, relating to foreign establishment. Commitments for mode 1 are much narrower and more limited, with a range of restrictions involving nationality, residency, authorization, and local authentication requirements. Commitments on mode 4, perhaps the most relevant mode for most developing countries, are the fewest and most restrictive. Many of the IT professionals from India in the United States have been admitted to the United States as temporary immigrants under H-1B visas. The annual number of such visas to be issued is determined by Congress, in part on the basis of market conditions for labor with the requisite skills and in response to lobbying. It is to be hoped that before the WTO ministerial meeting in Hong Kong in December 2005 the lacunae in GATS will be addressed. India is not the only country with a pool of English-speaking workers available for employment in the ITES/IT sector. Other countries with such populations include Bangladesh, Ireland, Pakistan, Sri Lanka, and the Philippines. Except in Bangladesh, the wage costs are higher than India’s in the other countries. The ability to speak English can, of course, be acquired, and as such, potential future competition for India from countries currently without a significant pool of English-speaking workers cannot be ruled out. Prominent among such 24. I thank Nirvikar Singh for pointing this out.

T. N. Srinivasan

229

countries is China. Yahya (2002) notes that there are two competing schools of thought among Indian policymakers about China’s efforts to expand its software sector: “One school believes that China is a serious competitor and that India should not assist its progress in any way. . . . The other school argues that India has a lead of two to three years over China and notes that India should use China as a market and collaborate with it” (Yahya 2002, pp. 114–15). I have already noted that, given the various agreements signed and the press releases during Chinese premier Wen Jiabao’s visit to India in April 2005, the collaborationist school seems to have won. In fact, as Yahya points out, China’s former premier, Zhu Rongji, during his visit to India in January 2002, also expressed the view that China and India could dominate the world IT market if they combined forces. The report of the India-China Joint Study Group on Comprehensive Trade and Economic Cooperation, presented to the prime ministers of the two countries by their commerce and industries ministers during Wen Jiabao’s visit, recommended the following: Companies from India and China should continue to explore each other’s markets. The two countries can use their core competencies in hardware and software to increase their share of [the] world’s trade and gain greater access in third country markets. Industry association such as NASSCOM may closely interact with its counterpart organization in China to promote co-operation in this sector. Both countries should work together to enforce copyright and reduce piracy. Moreover, the software enterprises of both countries can easily adapt to the ever changing high-tech market and strengthen their status in [the] global market by having more joint research projects, enhancing the exchange of technological personnel and cooperation in training.25 The report also noted significant scope for collaboration between the two countries in the market for services in several areas, including health, accounting and auditing, education, finance, and advertising. India’s IT and ITES-BPO sectors have firmly established a global reputation, and the potential for sustaining their rapid growth is bright. The success of the industry’s premier organization, the National Association of Software and Service Companies (NASSCOM), is being emulated by associations in the pharmaceutical, computer hardware, and auto parts manufacturing industries and 25. “Report of the India-China Joint Study Group on Comprehensive Trade and Economic Cooperation” (www.hindu.com/thehindu/nic/0041/report.pdf [May 4, 2005]).

230


others. Importantly, like NASSCOM, they are succeeding not only in lobbying but also in raising standards and quality.26 The growth of the IT sector could contribute significantly to accelerating GDP growth and add a large number of wellpaid jobs. Faster GDP growth and the indirect employment created by IT sector growth would also contribute to poverty reduction. There are, however, some serious domestic and external constraints that, if left unaddressed, could substantially diminish the realization of the potential growth.

Conclusion The visible success of the IT sector raises two important questions. First, can India’s future growth be IT-led, in the sense that the IT sector would not only be the prime engine of growth but also, because of its increasing share of GDP, compensate for slower growth in the nonservice sectors such as agriculture and large-scale manufacturing? Second, to the extent that the success of the IT sector could be linked to policy reforms—in telecommunications, for example— would it broaden and strengthen political support for reforms in other sectors? In attempting to answer the first question it is useful to think about the backward and forward linkages in IT sector growth. Its backward linkages arise from the growth in demand for inputs of goods, services, and factors. In particular, the demand will rise rapidly for labor with appropriate skills, for institutions that can educate and train workers for employment in the IT sector, and for telecommunication services and power. Its forward linkages could be significant as well, if other sectors of the economy invest in hardware and software to reap the productivity advantages of using IT services in their operations. It was noted earlier that the Ministry of Finance projects that value added by the IT sector will reach 7 percent by 2008, and at that rate a share of 25 percent is likely by 2020. Taken together, the prospects for the IT sector’s becoming the leading sector of the economy in the next decade or so are bright. However, accelerating the rate of GDP growth to 8 percent or more per year and sustaining it for several decades is a necessary, though not sufficient, condition for achieving the overarching objective of India’s development, namely, the eradication of poverty. As is well known, a large majority of India’s population is rural, and more that half of the country’s labor force still depends on agriculture and informal sector employment for a livelihood. And among the states 26. Kanwal Rekhi points this out and rightly emphasizes that such quality consciousness bodes well for the future of Indian industry.

T. N. Srinivasan

231

there is a significant variation since 1980 in economic performance, as measured by growth of state domestic product. The IT sector is concentrated in the cities of only a few states. Although the IT sector’s spectacular past and likely future growth will almost surely help to accelerate aggregate growth, translating faster growth into more rapid eradication of poverty would require widening, deepening, and accelerating economic reforms and liberalization. The response to the second question, whether the prospects for bringing about the required reforms are likely to be enhanced by the contribution of past reforms to the visible success of the IT sector, is a qualified yes. The reasons for qualification are many. First, the link between reforms and success of the IT sector is not clearly seen even by reform-minded politicians. Second, a vast majority of the population, particularly those rural areas, has yet to experience the benefits of efficient and inexpensive IT services in production and consumption activities, in part because IT services have not yet reached this population. Although this is changing, it is not changing rapidly enough to affect a large share of the population. Third, although there is no political pressure to reverse the reforms enacted thus far, it is evident that the remaining items of the reform agenda to be adopted would require persuasion of and bargaining with the parties not in the ruling coalition and its allies. Unlike in China, with its authoritarian, single-party government, success in this effort of persuasion in the competitive democratic polity of India requires foresighted leadership and the commitment of all parties. Just as the evident success of early agricultural reforms in China and the participation of a large share of the population in the fruits of its success created a constituency for later reforms, one can expect a similar effect in India once the benefits of IT success, and reforms more broadly, diffuse to the larger Indian population. It is encouraging and augurs well for the future that no political party (except perhaps the unreconstructed Stalinist elements of the left) is demanding a reversal of the reforms already enacted, and the differences among parties are more on the pace and sequencing of further reforms than on their content.

Comment and Discussion

Anne Krueger: As always, T. N. Srinivasan has provided an excellent analysis of the Indian economy’s performance—this time of the role of ITES (information-technology-enabled services) in past and prospective economic growth. His analysis of the growth of the IT sector is masterly, and I have only a few comments on it. I will then focus on the insights that the experience of the IT sector provides for Indian economic policy and growth more broadly. Turning first to the IT sector, its performance has been spectacular by any standard, especially in contrast to the rather sluggish performance of many Indian economic activities. But it should be noted that there are several unique characteristics of the IT sector that may have enabled it to grow so rapidly. First and foremost, IT is less heavily dependent on infrastructure than most industries are. Srinivasan notes that twenty-five different government rules had to be changed or removed in order to set up the first earth station in Bangalore in 1986. But the fact is that the industry could rely on an earth link and avoid many of the cumbersome aspects of Indian infrastructure: it hardly needed Indian roads or railroads, telecommunications (minimal as they then were), or Indian ports. For many other industries, rules would have had to be altered and infrastructure and other bottlenecks removed. While the IT sector had to live with the same constraints regarding other aspects of infrastructure as other industries, the effect on their cost was arguably considerably less. A significant feature of the IT sector is that most of the major firms have campuses in Bangalore and a few other cities, and these campuses are virtually self-sufficient. They have their own generators and do not depend on public provision of power. Company-owned buses even take people to and from work so 232

T. N. Srinivasan

233

they don’t have to rely on public transportation. Indeed, in Bangalore, some of the large firms are concerned that they may be unable to sustain their rapid growth because, as their employment increases, road congestion will be too great to absorb additional buses. The IT firms perceive the need to fight with the local authorities for capacity expansion for virtually every publicly provided urban service for which private investment cannot substitute. Infrastructure was less of a constraint for IT, but another factor was important as well. The IT sector was “encouraged,” as Srinivasan notes, but mostly by removing the disadvantages that other industries have. The sector was “discriminated for” in the sense that it did not suffer the disadvantages the government imposes. For one thing, the industry was new, and in a sense it outran the government by growing rapidly before regulations could be put in place. But, as Srinivasan also notes (all too modestly, since he had a significant role in it), the telecommunications reforms of the 1990s and later were highly significant. The IT sector probably could not have grown as rapidly as it did in the absence of those changes. One can only wonder what industry or industries would experience comparably rapid growth if the transport, or the labor market, or the power, bottlenecks, and regulations were removed for other sectors of the economy. One final comment on the IT sector’s growth prospects. Srinivasan has a table on wage differentials and notes that wages are very low by international standards. The relevant comparison should, of course, be unit labor costs, and a comparison there would be highly worthwhile. During the NAFTA debate in the early l990s, for example, much was made of the fact that average factory wages in the United States were some ten times the average Mexican factory wage. But once the calculation of unit labor costs was undertaken, it was found—not surprisingly—that unit labor costs were quite similar, and indeed several percentage points lower in the United States than in Mexico. Clearly, there is great scope for productivity improvement in India, in IT and elsewhere, and that can enable rising wages and sustained economic growth. Moreover, one would have to guess that in many IT services (where the physical capital input is very low) unit labor costs are significantly lower in India than in the major industrial countries and that, for that reason, there is some scope for real compensation increases in India in the context of rapid IT expansion. But I want to focus on overall Indian growth. A first point relates to the excellent Indian Institutes of Technology, which have provided a steady flow of world-class engineers since they were first set up in the late 1950s. When they were established, it was in the belief that India had a “shortage” of engineers.

234


The number of student places in the institutes was determined by estimating the number of “needed” engineers and providing sufficient places to generate the “needed” stock in a fairly short time. Not surprisingly, before too long there was an excess supply of highly qualified engineers. That is a partial explanation for the large diaspora of Indian engineers in Silicon Valley and elsewhere, as well as of the advantage the IT sector had in recruiting its personnel. A strong case can be made that India overinvested in higher education for engineers and that overall growth may have been significantly more rapid had some of those resources been allocated to increasing places in primary, secondary, and technical education. This brings me to my second point. India remains a country with a very large quantity of unskilled labor. No matter how successful the IT industry is, India is going to have to use its abundant supply of unskilled labor more productively and provide primary and secondary education and training for an increasing fraction of its labor force. To achieve rising living standards, that will be essential, especially in light of the fact that 70 percent of India’s population is still located in rural areas. To utilize its unskilled labor more productively, innumerable further policy reforms are needed. There are considerable labor market rigidities, which undoubtedly serve as a disincentive for firms to hire unskilled labor: in the “organized sector,” as it is called, firing workers is illegal (although some businesses have learned that if they do not pay their electric bills their electricity is shut off and they are forced to close, thus solving their labor problems). There are requirements for training workers, for provision of housing, and for other services. There was a “small-scale reservation” (SSR) policy, under which more than 800 small, labor-intensive industries were identified as eligible for privileges (such as tax exemptions) provided that they did not grow large. They engaged in activities such as candle-making, radio assembly, and production of batteries, and large firms were forbidden to enter these activities. Interestingly, these industries, and exports from them, grew much more slowly than would have been expected.1 Logically, though, small-scale firms cannot be expected to have the resources or the capacity to develop international markets. While enterprising businessmen were able to have many “companies” owned by various relatives side by side in one building, and thus circumvent the small-scale requirement to some extent, it was still a major barrier and disincentive to expansion and exporting. 1. See Mohan (2002).

T. N. Srinivasan

235

Ironically, when China began its rapid industrial expansion, based largely on unskilled labor-intensive goods, Chinese firms were able to penetrate the Indian market in many of the SSR industries. In the past several years, the government of India has begun to remove industries from the SSR list, but it still includes about 350 reserved industries. I have already noted the poor quality of Indian infrastructure. It would probably be absolutely insane to plan a business without including provision for the costs of generators. And transport costs are high; port delays are long; and there is still excess demand for airplane flights and services, though they have greatly improved in recent years.2 And although the government of India has invested in education, there are many problems. Teachers are not in school; they are absent and tutoring for higher pay than they could obtain in the schools (but still collecting their pay from the school). In a recent survey, the state with the lowest rate of teacher absenteeism, Maharashtra, had an absentee rate of 15 percent. In some states, it was more than 40 percent. With rates such as those, of course, parents may decide to have their children work at home, or in the field, since even if they go to school, they might simply have to return home.3 The litany of ills could continue, but I will stop after mentioning two more. The first is the regulatory environment that still exists (despite improvements) and that is surely a major productivity-reducing block for most private sector activities.4 Bureaucratic red tape and delays constitute a major obstacle and deterrent to efficient production and expansion.5 The second is the very large fiscal deficit of the general government (almost 10 percent of GDP in 2004) with India’s debt-to-GDP ratio already above 80 percent. Yet the fiscal deficit must be addressed at the same time that the Indian government finds means for improving infrastructure. That is in part because much of the existing pattern of expenditures is inefficient, with untargeted subsidies intended to benefit the poor going largely to the rich, and the need for thorough-going tax reforms.6 In my judgment, India has tremendous growth potential. Reforms have proceeded, albeit much more slowly than might have been desirable to attain significantly higher growth rates. They have been undertaken in a functioning 2. See Forbes (2002). 3. See Kremer and others (2005). 4. See World Bank (2006), p. 129. 5. See Shourie (2004). 6. See Srinivasan (2000).

236


democracy, which is certainly a huge plus for India. But much more is needed by way of reform, and soon. While the IT sector will surely continue to contribute to growth, it cannot absorb the greater part of India’s abundant labor force. IT can increase growth somewhat over the coming decades, but the rapid increase India needs will depend on the success of other reforms. Certainly, Indian economic prospects for the coming decade or two are brighter than they were around 1990. A growth rate of around 6 percent is probably sustainable if reforms continue at their present rate. But with more reforms, the 8 or 9 percent growth that India needs would become attainable. The IT sector’s development is not only a major success story; it is also an indication of what could be achieved. General Discussion: Anne Krueger’s comments stimulated an animated discussion on the pros and cons of the enclave-driven growth strategy taken by India relative to the counterfactual of a broader-based growth path. Robert Litan pointed out that India’s growth story might not be deemed a success if the full opportunity cost of broader policy reform and social investment were taken into consideration. For instance, the significant resources allocated to the engineering schools and elite education might have had higher social returns if directed instead to broader education or more infrastructure. But Litan also allowed that this is not altogether evident, owing to the considerable waste that is likely to accompany broad social spending because of Okun’s leaky bucket. Alan Deardorff wondered whether we should credit India’s “awful policies” for having created a comparative advantage in software that would not have existed in a more benign overall economic policy environment. In a similar vein, Kimberly Clausing noted that she was struck by the evidence presented in T. N. Srinivasan’s paper of beneficial growth effects from a significant concentration of engineering graduates, against the backdrop of significant illiteracy in the overall population. This circumstance diverges sharply from the general emphasis in development economics on investing first in universal access to primary education and working up the education ladder only as a country gets richer. Lael Brainard asked whether the apparent economic spillovers from enclavedriven growth in this case were also apparent in the domain of political economy. India’s last election certainly did not appear to confirm that the IT sector had spawned a strong political voice for broader economic reform. Rafiq Dossani endorsed Anne Krueger’s notion that policy reforms had had a redemptive effect, noting that many of the reforms starting in the mid-1980s were mainly redeeming bad preexisting laws and policies. He cited a recent

T. N. Srinivasan

237

change to a 1952 law that prohibited women from working at night—the activity that is the mainstay of the entire call center industry. Robert Feenstra questioned the potential magnitude of economic spillover effects from the IT industry’s success into the critical agricultural sector—still the main source of employment in India. He cited a discussion with India’s secretary of agriculture, who mentioned that certain parts of the country could grow three crops a year, but were prevented from doing so by constraints on the water supply. Attention then shifted to the role of the diaspora and elite educational institutions in the success of India’s software industry. Rafiq Dossani raised doubts about the significance of graduates of the elite Indian technology and management institutes as well as the Indian diaspora in the rise of India’s software industry. He said that none of the top ten IT companies in 1970, 1980, 1990, or 2000 were started by someone who did his undergraduate degree at one of the Indian Institutes of Technology or by a member of the diaspora. Dossani refuted the notion that Silicon Valley Indians played a large role in India’s IT boom, noting that they invested little in India’s IT industry. Indeed, he said, it appeared Silicon Valley Indians did not have the right expertise, since success in the Valley depended on developing technology products, while the India IT industry initially prospered by delivering customized software to the financial services industry. As the Indian industry is becoming more technology oriented, however, the involvement of the diaspora is increasing. Martha Laboissiere largely agreed. She pointed out that the Indian diaspora is found mainly in senior positions in the United States or Europe, and has contributed to India’s software industry by elevating India as the location of choice for offshoring, rather than through direct participation in the Indian industry. This contrasts with members of the Chinese diaspora, who are more likely to return home as middle managers. Laboissiere also underscored concerns about demand outstripping the supply of suitable engineers in India. Despite a high number of engineering graduates each year in India, headhunters consider only about 15 to 20 percent of these suitable for recruitment by offshore service providers. As a consequence, she predicted that demand would outstrip supply in India sooner than the 2012 timeframe predicted by NASSCOM, and that the crunch would come as early as 2008 in some places, such as Hyderabad. Moreover, Laboissiere emphasized that middle managers are in particularly tight supply in India, in part because of recruitment in India by eastern European companies. Srinivasan responded by reiterating his preference for broad-based over enclave-driven growth—for making all of India a zone for economic development

238


rather than relying on special economic zones. Nonetheless, he reasserted his optimism that the IT sector might be driving growth through important indirect as well as direct effects because of its role as an essential intermediate input. He noted that India probably should prefer South Korea’s path—6 to 7 percent growth for several decades followed by a short but sharp financial crisis—to the path it actually took: no financial crises and three decades of 3.5 percent growth only recently giving way to growth above 5 percent. Srinivasan observed that India’s investment in engineering education had been driven by Nehru—following the Soviet dictum that investing in engineering and electricity would solve the problems of development—rather than by any rational cost-benefit analysis. He agreed with Rafiq Dossani that the elite engineering and management schools were quantitatively small, and he believed they produce no more than 5 percent of the 250,000 engineers joining India’s labor market each year. But he contended that they have much greater impact on policymaking and the highest levels of business than at the managerial or entrepreneurial levels. Finally, Srinivasan agreed with Feenstra’s concerns about agriculture, noting that credit constraints and lack of domestic market integration are important impediments, along with policy distortions that lead to inefficient water use.

References Abbott, Frederick, Amy Kapczynski, and T. N. Srinivasan. 2005. “The Draft Patent Law,” Hindu, March 12, 2005 (www.thehindu.com/2005/03/12/stories/ 2005031201151000.htm). AIMA. 2003. India’s New Opportunity—2020. Report of the High Level Strategic Group. New Delhi: All India Management Association and Boston Consulting Group. Bardhan, Ashok Deo, and Cynthia A. Kroll. 2003. “The New Wave of Outsourcing.” Research Report. Berkeley, Calif.: Fisher Center for Real Estate and Urban Economics. Chithelen, Ignatius. 2004. “Outsourcing to India: Causes, Reaction and Prospects.” Economic and Political Weekly, March 6, pp. 1022–24. Forbes, Naushad. 2003. “Higher Education, Scientific Research and Industrial Competitiveness: Reflections on Priorities for India.” Paper presented at the Fourth Annual Conference on Indian Economic Reform held at Stanford University, June 5–7, 2003. http://scid.stanford.edu/events/India2003/Priorities_India.pdf. ———. 2002. “Doing Business in India: What Has Liberalization Changed?” In Economic Policy Reforms and the Indian Economy, edited by Anne O. Krueger, pp. 129–67. University of Chicago Press. Gordon, Robert J. 2000. “Does the ‘New Economy’ Measure Up to the Great Inventions of the Past?” Journal of Economic Perspectives 14 (4): 49–74.

T. N. Srinivasan

239

Government of India. 2003. Ministry of Communications and Information Technology Department of Information Technology. “Task Force on Meeting the Human Resources Challenge for IT and IT Enabled Services.” Grossman, Gene, and Elhanan Helpman. 2003. “Outsourcing versus FDI in Industry Equilibrium.” Journal of the European Economic Association 1 (2–3): 317–27. ———. 2002a. “Outsourcing in a Global Economy.” Working Paper 8728. Cambridge, Mass.: National Bureau of Economic Research. ———. 2002b. “Integrating versus Outsourcing in Industry Equilibrium.” Quarterly Journal of Economics 117 (1): 85–120. Helpman, Elhanan. 1990. “Monopolistic Competition in Trade Theory.” Special Papers in International Finance 16. Princeton, N.J.: Princeton University Department of Economics. Jorgenson, Dale. 2005 forthcoming. “Accounting for Growth in the Information Age.” In Handbook of Economic Growth, edited by Philippe Aghion and Steven Durlauf. Amsterdam: North Holland. Kapur, Devesh. 2002. “The Causes and Consequences of India’s IT Boom.” India Review 1 (2): 91–110. Kremer, Michael, Nazmul Chaudhury, F. Halsey Rogers, Karthik Muralidharan, and Jeffrey Hammer. 2005. “Teacher Absence in India: A Snapshot.” Journal of the European Economic Association 3 (2–3): 658–67. MOF. 2005. Economic Survey 2004–2005. New Delhi: Ministry of Finance, Government of India. ———. 2004. Economic Survey 2003–2004. New Delhi: Ministry of Finance, Government of India. Mohan, Rakesh. 2002. “Small-Scale Industry Policy in India: A Critical Evaluation.” In Economic Policy Reforms and the Indian Economy, edited by Anne O. Krueger, pp. 213–67. University of Chicago Press. Murthy, N., R. Narayana, and Sandeep Raju. 2002. “Comment on Chapters 4 and 5.” In Economic Policy Reforms and the Indian Economy, edited by Anne O. Krueger, pp. 194–203. University of Chicago Press. NASSCOM. 2005. “IT Industry Communiqué for the Academic Fraternity.” An IT Workforce Development Initiative 1. New Delhi: National Association of Software and Service Companies (www.nasscom.org/download/issue-2-july-sep-05.pdf. [September 29, 2005]). ———. 2004. Indian ITES-BPO Industry Handbook 2004. New Delhi: National Association of Software and Service Companies. RBI. 2005a. Macroeconomic and Monetary Developments in 2004–05. Mumbai: Reserve Bank of India. ———. 2005b. Reserve Bank of India Bulletin, April 13, 2005. Mumbai: Reserve Bank of India. Saxenian, AnnaLee. 2002a. “Bangalore: The Silicon Valley of Asia?” In Economic Policy Reforms and the Indian Economy, edited by Anne O. Krueger. University of Chicago Press.

240


———. 2002b. “Transnational Communities and the Evolution of Global Production Networks: The Cases of Taiwan, China and India.” Industry and Innovation 9 (3): 183–202. Shourie, Arun. 2004. Governance and the Sclerosis That Has Set In. New Delhi: ASA Publications. Singh, Nirvikar. 2004. “Transaction Costs, Information Technology and Development.” Revised version of paper presented at a conference in honor of Pranab Bardhan’s contribution to the Journal of Development Economics and to the larger field of development economics, September 24–25, 2004, Harvard University. ———. 2002. “India’s Information Technology Sector: What Contribution to Broader Economic Development?” Paper prepared for the conference “The IT/Software Industries in Indian and Asian Development,” November 11–12, 2002, Chennai, India. Srinivasan, T. N. 2000. Eight Lectures on Indian Economic Reforms. New Delhi: Oxford University Press. Sutton, John. 2005. “The Globalization Process: Auto-Component Supply Chains in China and India.” In Are We on Track to Achieve the Millennium Development Goals? edited by François Bourguignon, Boris Pleskovic, and André Sapir. Washington: World Bank. UNDP. 2005. Technology Achievement Index. http://hdrc.undp.org.in/hds/rgnl/TAI.htm. Weitzman, Martin. 1998. “Recombinant Growth.” Quarterly Journal of Economics 113 (2): 331–60. World Bank. 2006. Doing Business in 2006. Washington: World Bank and IFC. ———. 2005. World Development Indicators. Washington: World Bank. WTO. 2005. World Trade Report. Geneva: World Trade Organization. ———. 2004. International Trade Statistics. Geneva: World Trade Organization. Yahya, Faizal. 2002. “The Dragon Arises: China’s Challenge to India in Software Development.” India Review 1 (4): 91–122.

RAFIQ DOSSANI Stanford University

Globalization and the Offshoring of Services: The Case of India

T

he overwhelming majority of exports from developing countries to developed countries consist of agricultural commodities and manufactured goods. Such goods are usually produced under contract to a buyer from a developed country, the buyer managing design, marketing, and sales, while the seller handles production. Intermediate steps, such as accessing finance, technology, and raw materials, managing currency risks, and maintaining quality control, are shared between the buyer and the seller in special arrangements. Typically, in the introductory stages, the buyer assumes more control and takes more risk than in later stages. As the seller matures, it shares more of the risks and rewards. Nevertheless, history shows that the share of rewards (that is, the economic rents, if any) tend to remain with developed-country buyers. A new phase of globalization, international trade in services, has been emerging for at least a decade, and by now it looms important within the total value of World Trade Organization (WTO) trade. Developing countries around the world, particularly in Asia, have become large producers of services for developed countries. The range of such services is impressive. It includes back-office services such as payroll; customer-facing services such as call centers and telemedicine; design services such as the design of application-specific integrated circuits; research services such as conducting clinical trials; venture capital provision, from Taiwan to Silicon Valley, for example; software services such as programming; and IT and infrastructure outsourcing such as the managing of

The author thanks Arvind Panagariya, an anonymous referee, and the participants at the 2005 Brookings Trade Forum for helpful comments.

241

242


corporate e-mail systems and telecommunications networks. These new fields of service exports join the traditional fields of tourism and labor migration. As a result, services are becoming an important component of international trade for both developing and developed countries.1 Hence, studies of changes in the composition and trends of trade in services between developed and developing countries and the analysis of sharing of risks and rewards between buyers and sellers can provide insights into an important component of international trade and the new challenges and opportunities for economic development in the developing world. Some have even argued that the ability of countries like China and India to undertake high-end services work, such as in semiconductor design and information technology (IT), can threaten employment in developed countries, if productivity gains in developing countries are sufficiently high (for example, Samuelson 2004). In this paper, I use India as a case study of growth and value addition. India has catapulted itself in recent years into the leadership position in services exports from developing countries, and the study of India’s experience with exporting services in information technology could provide important insights into the workings of this new and promising area of international trade. Services appear to be key to its future growth (see table 1). Some components of the Indian experience (such as software services) are over three decades old, enabling us to draw useful insights about how a developing country can succeed in exporting services to developed countries. This is likely to have implications for policy in institutional development (such as educational and risk management institutions) and physical infrastructure. I argue that (1) local entrepreneurship and a high level of infant industry protection enabled the Indian IT industry to reach a high growth path and allowed local skills to develop rapidly to keep pace with global changes; (2) protectionism—in force until 1990—hurt the industry by forcing the work done to focus on low-value-added components; and (3) the rising importance of multinational firms from the late 1990s onward was the result of the country’s adopting a more welcoming legal and regulatory structure and led to a rise in the sophistication of work done. As a result, the contractual relationship between India-based IT providers and developed-country buyers has changed over the past three decades. Ini-

1. According to WTO statistics, services accounted for about 20 percent of global trade in 2003. The fastest growing category, commercial services, accounts for half of all services trade and grew at 15 percent in 2003, somewhat faster than merchandise trade (13 percent). WTO tables (www.wto.org/english/res_e/statis_e/its2004_e/its04_toc_e.htm).

Rafiq Dossani

243

Table 1. India’s IT Exports as a Share of Total Exports, 1999–2000 to 2003–04 Billions of U.S. dollars, unless otherwise indicated

Total exports IT-enabled services software exports Share of total (percent)

1999–2000 a

2000 –01

2001–02

2002–03

2003–04

37.54

44.89

44.91

52.51

63.2

4.52 12.0

7.14 15.9

9.14 20.3

12.04 22.9

15.8 25.0

Source: NASSCOM (2004), pp. 26, 63; Indian Economic Survey (2004), p.101, and www.dgciskol.nic.in/. a. The Indian financial year is from April 1 to March 31.

tially, it was characterized by “body-shopping”: the provider recruited labor and the buyer decided how to use it (Wortzel and Wortzel’s stage 1, see table 2). Later, applications programming became more prominent (stage 2). Today, the relationship is characterized by a mix of the two, though stage 1 work remains dominant.2 Do these findings suggest that developed countries are likely to be only marginally threatened by the globalization of services? After all, if the Indian software industry case is typical, then high-end work is likely to stay with developed countries. The problem for developed countries is that not everyone in developed countries can readily shift to high-end work. The shift of developed nations’ economies toward service-based economies beginning in the 1960s certainly increased the number of highly skilled service workers, but there was an even greater rise in the number of less-skilled service workers. This is partly a consequence of the nature of many services as linked, inseparable sets of activities provided with different levels of skill, with a pyramid of labor requirements— that is, lower-level work employs more people than high-end work. In manufacturing, the unemployment created by the reduction in demand for blue-collar labor in developed countries was offset by the absorption of much of the surplus labor into services, often with minimal training. But the shift from low-end service workers to high-end workers will require a longer period of reeducation and may have significant interim consequences on unemployment rates and costs. The threat to developed countries is increased by the fact that, apart from software, the largest growth in offshoring is happening in business services. These are also the sectors with the largest growth in U.S. employment. Goodman and Steadman (2002, p. 3) found that in 2002 more than 97 percent of the jobs added to U.S. payrolls were in services. Of these, business services and health care accounted for more than half of the growth. 2. See Wortzel and Wortzel (1981).

244


The threat to developed countries may grow if India’s success with IT exports and with other business processes is replicable by other developing countries. India’s history of private sector entrepreneurship has continued uninterrupted for many years, and, along with it, many institutions have been created—legal institutions, stock markets, educational institutions, and others—that have helped it succeed.3 To the extent that such entrepreneurship and institutions are weak in some developing countries (including China), those countries will have difficulty replicating India’s success. In this paper, I first discuss the two main services that have been offshored to India, business processes and software services.4 I then analyze the factors that enabled India to succeed in the business and conclude by examining the impact of offshoring on employment patterns, organizational structure of work, and value retention in developed countries.

The Globalization of Services The novelty of services in international trade warrants some definitional classifications as basic as what a “service” is. Most would agree that manufacturing is a process that involves the transformation of a tangible good. It would also generally be agreed that manufacturing does not require face-to-face contact between buyer and seller. Usually, manufacturing creates a good that can be stored, thereby allowing a physical separation of the buyer and the seller. Services have been defined by the opposites to these qualities: as transactions involving intangible, nonstorable goods, requiring that client and vendor be face to face while the service is being delivered. For example, Gadfrey and Gallouj (1998, p. 6) define services as goods that are “intangible, cosubstantial (i.e., they cannot be held in stock) and coproduced (i.e., their production/consumption requires cooperation between users and producers).” This was obviously true 3. Whitman (1990), quoted in Schware (1992, p. 148), argues that the factors for success in software are: computerization in industry and schools; university R&D in software and direct interactions with industry; skilled labor; funding sources such as venture capital and government contracts; support services such as telecommunications infrastructure; social networks among players such as engineers, managers, marketers, and funders; an entrepreneurial culture; an attractive, low-cost work environment and access to market channels through joint ventures and cooperative arrangements. India has several of these factors in place (Dossani 2004) 4. I collaborated with several others on various parts of the research presented here: with Martin Kenney on business process offshoring and with Anita Manwani on the Agilent case (see references). These sections of this paper reflect joint work.

Rafiq Dossani

245

when the service required face-to-face experience, such as receiving a haircut (a “customized” service, in the phraseology of Yotopoulos [1996]), but also true when the service experience did not require customization, such as when a bank’s client wanted to check the bank’s home loan offerings, or did not require proximity, as when she wanted to check her bank balance. For these reasons services are intrinsically more difficult to offshore than manufactured goods. Indeed, examined as a totality, services appear to resist relocation. Very few service operations can be done only on the computer (the modern form of “mundane work” in the terminology of Thompson, Warhurst, and Callaghan [2001]). Most services require at least some level of face-to-face interactivity, either among coworkers or with persons outside the organization, such as vendors and clients. The new twist in the provision of services is that the required interaction between the seller and the consumer has been substantially limited. Advances in information technology made possible the parsing of the provision of certain services into components requiring different levels of skill and interactivity. As a result, certain portions of the serviced activity—which might or might not be skill-intensive but required low levels of face-to-face interactivity—could be relocated offshore. The sequence of events that enabled this process is the following. First, the digital age allowed (or at least revolutionized) the conversion of service flows into stocks of information, making it possible to store a service. For example, a legal opinion that earlier had to be delivered to the client in person could now be prepared as a computer document and transmitted to the client by e-mail, or better yet, encoded into software. Easy storage and transmission allowed for the physical separation of the client and vendor as well as their separation in time. It also allowed the separation of services into components that were standardized and could be prepared in advance (such as a template for a legal opinion) and other components that were customized for the client (such as the opinion itself) or remained nonstorable. Taking advantage of the possibility of subdividing tasks and the economies that come with the division of labor, these developments reduced costs by offering the possibility of preparing the standardized components with lower-cost labor and, possibly, at another location. The second fundamental impact of digitization was the conversion of noninformation service flows into information service flows. For example, a buyer can often examine virtual samples of tangible goods over the Internet instead of visiting a showroom. Once converted to an information flow, a service may then be converted into a stock of information, as noted earlier, and subjected to

246


the forces of cost reduction through standardization of components and remote production. Third, by enabling low-cost transmission of the digitized material, the digital age accelerated the offshoring of services. Services such as writing software programs that were offshored to India in the early 1970s were enabled by digitized storage and, in the 1980s, by the standardization of programming languages. Still later, as digital transmission costs fell in the 1990s (just as digital storage costs had fallen earlier), even nonstorable services, such as customer care, could be offshored. The sequence of events that enabled offshoring did not happen all at once and may not even be standardized. Consider the evolution of sophistication in services as an analog to manufacturing. Table 2 provides a framework of changes in the relationships between a developed-country (DC) buyer and the less-developedcountry (LDC) seller. At each higher stage the process becomes increasingly decommodified with the increase in the exporter’s reputation. This and other similar frameworks, such as the “global value chain” framework,5 do not, however, (a) imply that moving to new stages is automatic, happens in the same sequence, or is time-bound; (b) provide conditions for movement to new stages; or (c) predict that the exporter will capture a rising share of the economic rents. At the very least, the need for costly global coordination can hinder movement to higher stages. Such coordination will initially be done by the developedcountry buyer and enable it to earn a rent for doing so. In addition, much of the market-related coordination and networking will occur through developedcountry institutions, enabling a further retention of value in the developed country. This may be why some countries in Asia have failed to go beyond the initial stages, such as the Philippine back-office industry. Even the much-vaunted Taiwanese semiconductor industry specializes in subcontracting for developedcountry designers of chips—that is, it lies in Wortzel’s stage 2. Examples of rising through stages are known, such as the Israeli software industry and the Korean cell phone manufacturers; and some East Asian economies have seen increased employment, rising wages overall, and reduced poverty as a result of exporting services;6 more commonly, though, developingcountry industries have not reached parity with their developed-country clients. In fact, few go beyond stage 3 work.7 The inference is that certain key aspects, such as deciding on the product and its specification, design, marketing, and 5. Gereffi and others (2001). 6. Humphrey (2004, p. 5). 7. Scholte (2000).

Rafiq Dossani

247

sales, are usually retained by the importer. When that happens it appears to hinder the exporter’s ability to rise to new stages of growth and the rewards are overwhelmingly gathered by the developed-country buyer. Note, however, that stagnation within a stage of growth and stagnation in value share are not identical. For example, OPEC’s members largely have stagnated in stage 1, but their value share has fluctuated with time relative to higherstage work such as refining and marketing petroleum products, owing to fluctuations in world oil prices. More generally, stagnation within a stage can result from a lack of skills, capital, infrastructure, and institutions relative to international competitors, technology, or other interdependence at higher stages that makes separation of higher stages from the developed-country location too costly (lack of “modularization”).8 Stagnation within a stage is likely to lead to a reduction in value share. This is because, to enable outsourcing, a production process first needs to be modularized; that is, the costs of coordination have to be manageable, and the technical and other relevant parameters need to be codified. Implicit in modularization is a shift from proprietary to standard inputs, design, and fulfillment techniques. To the extent that standardization (or, as it is often termed, “commodification”) lowers barriers to entry, it will reduce the share of value through competition. On the other hand, stagnation in value share can happen even if the developing country’s industry rises to higher stages of growth. This can happen if developed-country buyers have market power vis-à-vis their vendors (which may or may not be linked to owning nonstandard processes), so that they obtain better terms of trade regardless of their vendors’ level of sophistication. Hence, moving to higher stages of growth may not be associated with rising value share.

The Offshoring of Business Processes Business processes (BPs) is the catchall term used for the myriad white-collar processes that any bureaucratic entity undertakes in servicing its employees, 8. Modularization is defined as the conversion of a component of the production process with one or more proprietary inputs, design, or fulfillment techniques into a component with standardized inputs, design, and fulfillment techniques. For example, understanding end-user needs is a requirement for a developing-country firm to move from stage 3 to stage 4. If such work is only possible through pilot projects undertaken on clients’ sites, the management of the pilot project must be an integral part of the requirements analysis. In this case, modularization is not feasible, so it may not be possible for the developing country to do such work.

Table 2. Stages of Growth in Offshoring

Stage Stage 1 (assembly) Stage 2 (original equipment manufacture) Stage 3 (original equipment manufacture) Stage 4 (original design manufacture) Stage 5 (original brand manufacture)

Importer’s rolea (A)

Transferred to exporter (B)

Shared roles (C)

Exporter’s roles outside the initial vendor relationship (D)

Custom software industry’s analog to column B (E)

Design, input sourcing and quality control External design and specification for internal design External design

Production capacity

None

None

Body-shoppingb

Sourcing, application of internal design

Quality control

None

Programming of applications

Specification for internal design

Quality control

Add customers

System design and integration

Purchase from exporter’s catalog

Product selection, external design, quality control Product innovation

None

Add sophistication to product range

System architecture and R&D

Competitor

None

System consulting and business development

Source: Wortzel and Wortzel (1981) and Hobday (1995) for stages and columns A–D. Column E by author, based on table 6. a. The rows indicate the stages of product development; the columns show the sharing of work between the importer and exporter. Columns A–D are relevant to manufacturing; column E provides the analogs to service for the particular case of software production. b. Body-shopping refers to providing the client with programmers. The domestic firm recruits the programmers, and the client decides how to use them, often on the client’s own sites in the developed country.

Rafiq Dossani

249

Table 3. Exports of Information Technology–Enabled Services (ITES) from India, 1999–2000 to 2003–04

ITES exports (millions of $) Employment (number) Revenue/employee ($) U.S.-firm comparable revenue/employee ($)

1999–00

2000 –01

2001–02

2002–03

Average growth 2003–04 (percent)

565 42,000 13,452

930 70,000 13,286

1,495 106,000 14,104

2,500 171,000 14,620

3,600 245,500 14,664

…

…

…

…

58,598

59 55 2.2 …

Source: Nasscom (2004, pp. 63, 64, 186).

vendors, and customers.9 These include human resources, accounting, auditing, customer care, telemarketing, tax preparation, claims processing, document management, and many other chores necessary for the firm to function. In the 1990s, India became the largest provider of such offshored services. As table 3 shows, the cost savings to companies that use these services can be significant. The services most offshored are call center services, employing about 200,000 persons at the end of 2003, or about 70 percent of the BPs offshored to India since 2000. This is a large industry in the United States: one recent study estimated that call centers alone employ as much as 3 percent of the U.S. workforce; some estimate that the share will increase to 5 percent by 2010.10 Most of the offshored jobs are relatively low-skilled jobs. For example, the single largest category is outbound calling for the financial services industry to sell financial services such as mortgages or to collect on overdue receivables. The work is routine; workers follow scripts that pop up on the computer screen in response to prompts. However, there is evidence that higher-skilled jobs can be offshored or might evolve on their own. For example, the author interviewed an Indian firm that had initially been contracted by an American firm to call its clients with overdue

9. We define a business process as a complete service, such as handling a customer complaint, processing a medical claim, or processing a purchase order. Completing a process requires undertaking a set of activities. For example, in handling a customer complaint it is necessary to understand the complaint, decide on a course of action, undertake the action, and follow up to ensure that the action solved the complaint. Each of these activities is potentially separable from the others. 10. “The Customer Care Workforce: Driving More Profitable Customer Interactions,” CRM Project, vol. 3, October 30, 2002 (www.crmproject.com/documents.asp?grID=293&d_ID=1578 [July 2003]).

250


Figure 1. India-to-United States Bandwidth Pricing, 1996–2000 Dollars per megabyte

50

40

30

20

10

1996

1997

1998

1999

2000

Source: Andy Grove, “The Coming Software X Curve,” presentation to the Business Software Alliance, October 9, 2003.

credit card payments. The Indian firm later graduated to purchasing the receivables from its client and assuming the collection risk itself. Another firm interviewed, Wipro Spectramind, managed the radiology services of Massachusetts General Hospital for its second and third shifts. American radiologists, who earn $315,000 a year on average, were replaced by Indian radiologists, who earn $20,000 a year on average. Offshoring such work began in 1993 when American Express, an American bank, started using its Indian operations to provide bookkeeping support to its other Asian operations. The business grew rapidly after 1999, when the telecommunications infrastructure in India improved and costs fell (Dossani 2002b) (see figure 1). With relatively few regulatory restrictions on ownership or types of work, a difference in work types related to ownership appears to be evolving. Multinationals that do their own work focus on the back-office, higher-stage, and more sensitive work (time-sensitive work such as payroll, or work involving confidential data). Outsourcers, whether owned by Indian nationals or foreigners, do mostly call center work.

Rafiq Dossani

251

Figure 2. Evolution of Financial Services at Agilent Gurgaon Business process reengineering; ERP support;b global projects; PO to payment process;c process redesigns; policy formulations; solutions and business integration

H e a d c o u n t

Invoice processing Basic accounting; data entry; basic troubleshooting

0

Mar 2002

Complex VP;a problem solving; payments

Supplier calls

Customer collections; accounts reconciliation; year- and month-end close; Six Sigma Program

Dec 2002 Mar 2003 Jun 2003

Dec 2003 TIME

Feb 2004

a. VP = Vendor payables. b. ERP = Enterprise resource planning. c. PO = Purchase order.

It is clear from Indian business process offshoring (BPO) that complex work can be efficiently done in India. A recent case study of a large multinational firm, Agilent, which in 2002 began offshoring about half of its head count in finance and accounting to Gurgaon, India, shows the evolution of complexity and these gains (see figure 2). The company began with the simplest work of data entry in March 2002. Up to December 2003, the work did not change in complexity as new staff were recruited to take over the work of Agilent’s offices around the world. However, by December 2003 Agilent felt that the Gurgaon office was sufficiently mature to enable a move to the next step. This was achieved rapidly and successfully; by February 2004 about half of Agilent’s global head count for the finance and accounting function were employed in Gurgaon. The complexity of work rose significantly to include customer collections and policy formulation support work, and finally to managing relations with suppliers. Several of these functions were unanticipated benefits when the offshoring relationship began. Also, as table 4 shows, the work was done with a gain in both costs and staff time; that is, the Indian operations were more efficient than the unconsolidated global

252


Table 4. Productivity Gain in the Vendor Payables Division at Agilent Technologies Number of workers Head count Activity

January 2003

April 2004

Head count reduction (percent)

19 75 11 27 9 1 46 188

13 56 9 19 6 1 34 138

46 34 22 42 50 38 35 37

Order supplier processing Invoice processing Employee reimbursement Help desk/CRC Payment Reporting metrics Management and administration Total Source: Agilent documents.

operations were earlier by a factor of 37 percent. This was attributable to the benefits of consolidation—that is, economies of both scale and scope—as well as new technology platforms.11

The Offshoring of Software Services A more mature example than business process offshoring is the offshoring of software services (different from software products) to India, which began in 1974.12 The Indian software services exporting industry employs over 250,000 people (see table 5). How the global software services industry’s scale and components compare with India’s is shown in table 6. Different service processes require different skills and levels of interaction with others in the supply chain. Consulting and system integration require high

11. Dossani and Manwani (2005). 12. Software is either (1) written for general use and intended to be replicated in its original form for many users or (2) customized to a client’s needs. The former is termed a software product or package. The latter is termed custom software and is part of a larger category called software services (see table 5). Services are more constrained by geography than products since products can be shrink-wrapped and transported physically or over the Internet, whereas services are delivered to order. Hence, most countries export software products, India being an exception. There are three types of software: (1) system-level software: programs that manage the internal operations of the computer, such as operating systems software, driver software, virus scan software, and utilities; (2) tools software: programs that help applications to work better, such as data-

Rafiq Dossani

253

levels of interaction, as does IT education and training. However, and this was important for the development of India’s software industry, the work of applications development requires only programming skills and limited interaction. This makes it possible for one firm to design and specify an application and for another firm to develop it, with almost no interaction needed between the two. Thus, a user might contract with a U.S.-based consulting firm for system specification and then transfer the work to an Indian programming firm to develop the applications. This separability of programming from other processes has been key to the work being done in India since the early days. But the need for high levels of interaction with clients and their customers and vendors may make it difficult to do consulting and other complex work offshore even though the skills might be available. Initially, the work done in India largely kept pace with the U.S. software industry. However, with the United States now mostly doing the highly interactive and most sophisticated work, the gap has lengthened (see table 7). Unlike the BPO business in India, the software business began with an Indian firm, TCS, in 1974. However, this was due to protection: in 1973 the Indian government required multinationals operating in India to reduce their shareholding to 40 percent. Many multinationals, including IBM (India’s largest IT firm at the time) preferred to close shop rather than divest. This created an opportunity for Indian firms to provide software services. Multinationals were allowed to reenter beginning in 1985, although the playing field was not leveled for at least another decade. Texas Instruments was the first to enter, in 1985. Multinationals’ presence in India remained small until recently, although since 2001 their growth rate has been greater than that of Indian firms. Their 2004 market share has been estimated at 13 percent.13 As tables 2 and 6 show, Indian software services are at the early stages of work. The relatively low value of their work may be due to the state’s protectionist policies and the resulting absence of multinationals in the early years, though there is some indication that in recent years the presence of multinationals has contributed to a rise in Indian value added (see table 5). The entry of base management software; (3) applications software: programs that deliver solutions to the end user, such as word processing software and financial accounting software. Systems software is the most complex because it manages the interfaces with both hardware and higher-level software; applications software is the least complex. All system-level software are products (although this was not always the case). However, the more varied end users’ needs are, the more likely the software is to be customized. Since variations in needs appear most at the stage of applications, most customized software is applications software. 13. NASSCOM (2004)

Table 5. Software Services Sales of the Indian IT Industry, 1996–97 to 2003–04 Millions of U.S. dollars, unless otherwise indicated

Software Domestic Exports Total Export (percent) Exports employment (number) Export revenue/ employee

1996–97 a

1997–98

1998–99

1999–2000

2000 –01

2001–02

2002–03

2003–04

Average growth, 1996–97 to 2003–04 (percent)

759 1,100 1,859 59.2

1,177 1,759 2,936 59.9

1,411 2,600 4,011 64.8

1,575 3,399 4,974 68.3

2,081 5,287 7,368 71.8

2,311 6,152 8,463 72.7

2,769 7,045 9,814 71.8

3,374 8,600 11,974 71.8

23.8 34.2 30.5 ...

n.a.

n.a.

n.a.

110,000

162,000

170,000

205,000

260,000

24.0

n.a.

n.a.

n.a.

30,900

32,635

36,188

34,366

33,077

1.8

Source: NASSCOM (2004, pp. 23, 26, and 64). a. The Indian financial year is from April 1 to March 31. n.a. Not available.

Rafiq Dossani

255

Table 6. Global Software Services Spending and Indian Share by Category of Work, 2003

Work categorya Consultingb Applications developmentc System integration: hardware and software deployment and supportd System integration: applications, tools, and operating systems (O/S)e IT education and trainingf Managed servicesg Total

Global software services spending (billions of $) (1)

Percentage share (2)

U.S. wage rate ($/hour) (3)

Indian exports (billions of $) (4)

India’s global market share (percent) (5)

41.5

11.6

80 –120

0.11

Brookings Trade Forum 2005 (Brookings Trade Forum)

Climate Change, Trade, and Competitiveness: Is a Collision Inevitable? (Brookings Trade Forum)

India Policy Forum 2005-06

Food Crises and the WTO: World Trade Forum

Food Crises and the WTO: World Trade Forum

Brookings Papers on Economic Activity 2005 (Brookings Papers on Economic Activity)

International Mining Forum 2004

Brookings papers on education policy, 1999

Dienstleistungscontrolling: Forum Dienstleistungsmanagement

Brookings Papers on Economic Activity: Fall 2009

The Buddhist Forum

India Policy Forum 2004

International Trade and Human Rights: Foundations and Conceptual Issues (World Trade Forum, Volume 5) (Studies in International Economics)

Brookings Papers on Economic Activity: Spring 2010

Brookings Papers on Economic Activity: Spring 2009

Democratic Faith (New Forum Books)

Trade Secrets

Skin Trade

Skin Trade

Skin Trade

Skin Trade

Skin Trade

Wicked Trade

Skin Trade

The Trade

Skin Trade

Lustrous Trade

Cogan's Trade

Brookings Trade Forum 2005 (Brookings Trade Forum)

Climate Change, Trade, and Competitiveness: Is a Collision Inevitable? (Brookings Trade Forum)

India Policy Forum 2005-06

Food Crises and the WTO: World Trade Forum

Food Crises and the WTO: World Trade Forum

Brookings Papers on Economic Activity 2005 (Brookings Papers on Economic Activity)

International Mining Forum 2004

Brookings papers on education policy, 1999

Dienstleistungscontrolling: Forum Dienstleistungsmanagement

Brookings Papers on Economic Activity: Fall 2009

The Buddhist Forum

India Policy Forum 2004

International Trade and Human Rights: Foundations and Conceptual Issues (World Trade Forum, Volume 5) (Studies in International Economics)

Brookings Papers on Economic Activity: Spring 2010

Brookings Papers on Economic Activity: Spring 2009

Democratic Faith (New Forum Books)

Trade Secrets

Skin Trade

Skin Trade

Skin Trade

Skin Trade

Skin Trade

Wicked Trade

Skin Trade

The Trade

Skin Trade

Lustrous Trade

Cogan's Trade

Recommend Documents