Advances in Management Accounting, Volume 12 (Advances in Management Accounting)

CONTENTS LIST OF CONTRIBUTORS ix EDITORIAL BOARD xi STATEMENT OF PURPOSE AND REVIEW PROCEDURES EDITORIAL POLICY AND ...

Author: John Y. Lee

230 downloads 2814 Views 3MB 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

CONTENTS LIST OF CONTRIBUTORS

ix

EDITORIAL BOARD

xi

STATEMENT OF PURPOSE AND REVIEW PROCEDURES EDITORIAL POLICY AND MANUSCRIPT FORM GUIDELINES INTRODUCTION Marc J. Epstein and John Y. Lee NEW DIRECTIONS IN MANAGEMENT ACCOUNTING RESEARCH: INSIGHTS FROM PRACTICE Frank H. Selto and Sally K. Widener

xiii

xv

xvii

1

THE PROFIT IMPACT OF VALUE CHAIN RECONFIGURATION: BLENDING STRATEGIC COST MANAGEMENT (SCM) AND ACTION-PROFIT-LINKAGE (APL) PERSPECTIVES John K. Shank, William C. Lawler and Lawrence P. Carr

37

THE MEASUREMENT GAP IN PAYING FOR PERFORMANCE: ACTUAL AND PREFERRED MEASURES Jeffrey F. Shields and Lourdes Ferreira White

59

v

vi

AN EMPIRICAL EXAMINATION OF COST ACCOUNTING PRACTICES USED IN ADVANCED MANUFACTURING ENVIRONMENTS Rosemary R. Fullerton and Cheryl S. McWatters

85

THE INTERACTION EFFECTS OF LEAN PRODUCTION MANUFACTURING PRACTICES, COMPENSATION, AND INFORMATION SYSTEMS ON PRODUCTION COSTS: A RECURSIVE PARTITIONING MODEL Hian Chye Koh, Khim Ling Sim and Larry N. Killough

115

COMPENSATION STRATEGY AND ORGANIZATIONAL PERFORMANCE: EVIDENCE FROM THE BANKING INDUSTRY IN AN EMERGING ECONOMY C. Janie Chang, Chin S. Ou and Anne Wu

137

ACCOUNTING FOR COST INTERACTIONS IN DESIGNING PRODUCTS Mohamed E. Bayou and Alan Reinstein

151

RELATIONSHIP QUALITY: A CRITICAL LINK IN MANAGEMENT ACCOUNTING PERFORMANCE MEASUREMENT SYSTEMS Jane Cote and Claire Latham

171

MEASURING AND ACCOUNTING FOR MARKET PRICE RISK TRADEOFFS AS REAL OPTIONS IN STOCK FOR STOCK EXCHANGES Hemantha S. B. Herath and John S. Jahera Jr.

191

CONNECTING CONCEPTS OF BUSINESS STRATEGY AND COMPETITIVE ADVANTAGE TO ACTIVITY-BASED MACHINE COST ALLOCATIONS Richard J. Palmer and Henry H. Davis

219

vii

CHOICE OF INVENTORY METHOD AND THE SELF-SELECTION BIAS Pervaiz Alam and Eng Seng Loh

237

CORPORATE ACQUISITION DECISIONS UNDER DIFFERENT STRATEGIC MOTIVATIONS Kwang-Hyun Chung

265

THE BALANCED SCORECARD: ADOPTION AND APPLICATION Jeltje van der Meer-Kooistra and Ed G. J. Vosselman

287

LIST OF CONTRIBUTORS Pervaiz Alam

College of Business Administration, Kent State University, Ohio, USA

Mohamed E. Bayou

School of Management, University of Michigan, Dearborn, Michigan, USA

Lawrence P. Carr

F. W. Olin Graduate School of Management, Babson College, Massachusetts, USA

C. Janie Chang

Department of Accounting and Finance, San Jose State University, California, USA

Kwang-Hyun Chung

Lubin School of Business, Pace University, New York, USA

Jane Cote

School of Accounting, Information Systems and Business Law, Washington State University, Washington, USA

Henry H. Davis

Lumpkin College of Business and Applied Sciences, Eastern Illinois University, Illinois, USA

Rosemary R. Fullerton

School of Accountancy, Utah State University, Utah, USA

Hemantha S. B. Herath

Department of Accounting and Finance, Brock University, Canada

John S. Jahera Jr.

College of Business, Auburn University, Alabama, USA

Larry N. Killough

Virginia Polytechnic Institute and State University, USA

Hian Chye Koh

Nanyang Technological University, Singapore ix

x

Claire Latham

School of Accounting, Information Systems and Business Law, Washington State University, Washington, USA

William C. Lawler


Eng Seng Loh

Business Intelligence Group, Caterpillar Inc., USA

Cheryl S. McWatters

University of Alberta, Canada

Jeltje van der Meer-Kooistra

University of Groningen, The Netherlands

Chin S. Ou

Department of Accounting, National Chung Cheng University, Taiwan

Richard J. Palmer

Lumpkin College of Business and Applied Sciences, Eastern Illinois University, Illinois, USA

Alan Reinstein

School of Business, Wayne State University, Michigan, USA

Frank H. Selto

University of Colorado at Boulder, Colorado, USA and University of Melbourne, Australia

John K. Shank


Jeffrey F. Shields

School of Business, University of Southern Maine, Maine, USA

Khim Ling Sim

School of Business, Western New England College, Massachusetts, USA

Ed G. J. Vosselman

Erasmus University Rotterdam, The Netherlands

Lourdes Ferreira White

Merrick School of Business, University of Baltimore, Maryland, USA

Sally K. Widener

Rice University, Texas, USA

Anne Wu

School of Accounting, National Chengchi University, Taiwan

EDITORIAL BOARD Thomas L. Albright Culverhouse School of Accountancy, University of Alabama

George J. Foster Stanford University James M. Fremgen Naval Postgraduate School

Jacob G. Birnberg University of Pittsburgh

Eli M. Goldratt Avraham Y. Goldratt Institute

Germain B. Boer Vanderbilt University

John Innes University of Dundee

William J. Bruns, Jr. Harvard University

Fred H. Jacobs Michigan State University

Peter Chalos University of Illinois, Chicago

H. Thomas Johnson Portland State University

Chee W. Chow San Diego State University

Larry N. Killough Virginia Polytechnic Institute

Donald K. Clancy Texas Tech University

Thomas P. Klammer University of North Texas

Robin Cooper Emory University

C. J. Mc Nair Babson College

Srikant M. Datar Harvard University

James M. Reeve University of Tennessee, Knoxville

Nabil S. Elias University of North Carolina, Charlotte

Jonathan B. Schiff Fairleigh Dickinson University

K. J. Euske Naval Postgraduate School

John K. Shank Dartmouth College

Eric G. Flamholtz University of California, Los Angeles

Barry H. Spicer University of Auckland xi

xii

George J. Staubus University of California, Berkeley

Lourdes White University of Baltimore

Wilfred C. Uecker Rice University

S. Mark Young University of Southern California

STATEMENT OF PURPOSE AND REVIEW PROCEDURES Advances in Management Accounting (AIMA) is a professional journal whose purpose is to meet the information needs of both practitioners and academicians. We plan to publish thoughtful, well-developed articles on a variety of current topics in management accounting, broadly defined. Advances in Management Accounting is to be an annual publication of quality applied research in management accounting. The series will examine areas of management accounting, including performance evaluation systems, accounting for product costs, behavioral impacts on management accounting, and innovations in management accounting. Management accounting includes all systems designed to provide information for management decision making. Research methods will include survey research, field tests, corporate case studies, and modeling. Some speculative articles and survey pieces will be included where appropriate. AIMA welcomes all comments and encourages articles from both practitioners and academicians. Review Procedures AIMA intends to provide authors with timely reviews clearly indicating the acceptance status of their manuscripts. The results of initial reviews normally will be reported to authors within eight weeks from the date the manuscript is received. Once a manuscript is tentatively accepted, the prospects for publication are excellent. The author(s) will be accepted to work with the corresponding Editor, who will act as a liaison between the author(s) and the reviewers to resolve areas of concern. To ensure publication, it is the author’s responsibility to make necessary revisions in a timely and satisfactory manner.

xiii

EDITORIAL POLICY AND MANUSCRIPT FORM GUIDELINES 1. Manuscripts should be type written and double-spaced on 8 21 by 110 white paper. Only one side of the paper should be used. Margins should be set to facilitate editing and duplication except as noted: (a) Tables, figures, and exhibits should appear on a separate page. Each should be numbered and have a title. (b) Footnote should be presented by citing the author’s name and the year of publication in the body of the text; for example, Ferreira (1998), Cooper and Kaplan (1998). 2. Manuscripts should include a cover page that indicates the author’s name and affiliation. 3. Manuscripts should include on a separate lead page an abstract not exceeding 200 words. The author’s name and affiliation should not appear on the abstract. 4. Topical headings and subheadings should be used. Main headings in the manuscript should be centered, secondary headings should be flush with the left hand margin. (As a guide to usage and style, refer to the William Strunk, Jr., and E. B. White, The Elements of Style.) 5. Manuscripts must include a list of references which contain only those works actually cited. (As a helpful guide in preparing a list of references, refer to Kate L. Turabian, A Manual for Writers of Term Papers, Theses, and Dissertations.) 6. In order to be assured of anonymous review, authors should not identify themselves directly or indirectly. Reference to unpublished working papers and dissertations should be avoided. If necessary, authors may indicate that the reference is being withheld for the reason cited above. 7. Manuscripts currently under review by other publications should not be submitted. Complete reports of research presented at a national or regional conference of a professional association and “State of the Art” papers are acceptable. 8. Four copies of each manuscript should be submitted to John Y. Lee at the address below under Guideline 11. 9. A submission fee of $25.00, made payable to Advances in Management Accounting, should be included with all submissions. 10. For additional information regarding the type of manuscripts that are desired, see “AIMA Statement of Purpose.” xv

xvi

11. Inquires concerning Advances in Management Accounting may be directed to either one of the two editors: Marc J. Epstein Jones Graduate School of Administration Rice University Houston, Texas 77251–1892

John Y. Lee Lubin School of Business Pace University Pleasantville, NY 10570–2799

NEW DIRECTIONS IN MANAGEMENT ACCOUNTING RESEARCH: INSIGHTS FROM PRACTICE Frank H. Selto and Sally K. Widener ABSTRACT Although the “new economy” once again resembles the old economy, the drivers of success for many firms continue to be intangible or service-related assets. These changes in the economic basis of business are leading to changes in practice which are creating exciting new opportunities for research. Management accounting still is concerned with internal uses of and demands for operating and performance information by organizations, their managers, and their employees. However, current demand for internal information and analysis most likely reflects current decision making needs, which have changed rapidly to meet economic and environmental conditions. Many management accounting research articles reflect traditional research topics that might not conform to current practice concerns. Some accounting academics may desire to pursue research topics that reflect current problems of practice to inform, influence, or understand practice or influence accounting education. This study analyzes attributes of nearly 2,000 research and professional articles published during the years 1996–2000 and finds numerous, relatively

Advances in Management Accounting Advances in Management Accounting, Volume 12, 1–35 Copyright © 2004 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12001-7

1

2

FRANK H. SELTO AND SALLY K. WIDENER

unexamined research questions that can expand the scope of current management accounting research. Analyses of theories, methods, and sources of data used by published management accounting research also describe publication opportunities in major research journals.

DATA AVAILABILITY Raw data are readily available online, and coded data are available upon request from the authors.

INTRODUCTION AND MOTIVATION While some aspects of the “new economy” reflected an unrealistic bubble, many firms continue to be driven by intangible assets, the highly competitive global economy, and increasing technological change to forge changes in what accountants have thought of as their “traditional” accounting responsibilities. In many cases, accountants and financial staff are leading the way in changing their internal roles. Accountants find themselves managing new business practices, such as outsourcing, focusing more on cost control and process re-engineering, and expanding their involvement with strategic planning and implementation. The expansion of accountants’ duties beyond traditional budgeting and reporting is occurring rapidly and is creating numerous opportunities for academic management accountants to conduct innovative research. According to a recent IMA study of practicing “management accountants” (IMA, 2000),1 apparently no management accountants are left in practice. Professionals in practice overwhelmingly have favored job titles such as financial analyst, business advisor, and consultant over “cost accountant” or “management accountant.” Perhaps this is not a purely cosmetic change. The IMA study also shows that current job titles reflect broader duties than traditionally executed by accountants. Instead of viewing this change as the end of management accounting, a more optimistic viewpoint is to see this as an opportunity to broaden management accounting, both in education and in research. This opens doors for exciting new research opportunities. Some accounting researchers conduct research that is explicitly oriented to or has application to practice. Others might seek to do so. Several related motivations or objectives for practice-oriented research include desires to: (1) gain increased understanding of why organizations use certain techniques and practices; (2) gain increased understanding of how and which techniques used in practice impact organizational performance; (3) inform practitioners; (4) increase the applicability

New Directions in Management Accounting Research

3

of accounting textbooks, coursework, and programs; (5) satisfy personal taste; and (6) increase consulting opportunities. While researchers pursuing any of these might find this study interesting and helpful, this study is explicitly motivated by the first four objectives. One desirable outcome of practice-oriented research may be a positive impact on accounting enrollments. Many university accounting programs in the U.S. have been in decline, perhaps because of: (1) increased education requirements for accounting certification in many states; (2) relatively greater employment opportunities and salaries in other business fields, such as finance; (3) competitive educational efforts by industrial and professional firms; (4) focused financial support of only select universities by employers of accounting graduates; and (5) perceived greater job-relevance of other courses. Many of the factors that can contribute to enrollments in accounting are beyond the control of accounting academics. Because research surely informs teaching, accounting faculty might help increase accounting enrollments by managing what is researched.2

Research Objectives Management accounting research, researchers, and education (and perhaps other accounting sub-fields by analogy) might benefit from identifying interesting, less researched topics that reflect issues of current practice. More influence on external constituents might lead to greater prestige, esteem, and resources for researchers, and, perhaps, improvements in practice (e.g. Anderson, 1983). The objective of this study is to use observed divergences between management accounting research topics and issues of practice to identify interesting, practice-oriented research questions. The study assesses and interprets correspondence (or lack thereof) between published research topics and topics of the practice literature. High correspondence can be misleading because it might represent good synergy, coincidence, or little interest. Low correspondence might present opportunities for interesting new research. Thus, this study examines both types of topics as potential sources of interesting research questions. Finally, the study then addresses the equally important issue of matching these research questions with theory, data, and research methods. Without these matches, management accounting research will have difficulty moving beyond pure description or endless theory building. It also might be possible to increase the probability of publication of these new questions by assessing journals’ past publication histories. This study is unlike recent, more focused reviews of management accounting research, which include Covaleski and Dirsmith (1996) – organization and

4


sociology-based research; Elnathan et al. (1996) – benchmarking research; Shields (1997) – research by North Americans; Demski and Sappington (1999) – empirical agency theory research; Ittner and Larcker (1998) – performance measurement research; and Ittner and Larker (2001) – value-based management research. The present study is in the spirit of Atkinson et al. (1997), which seeks to encourage broader investigations of management accounting research topics. The present study extends Atkinson et al. by documenting and identifying practice-oriented, innovative research questions in major topic areas based on observed divergences between practice and research.

RESEARCH DESIGN AND METHOD The study’s research design is to first compare topic coverage of research and professional publications. Differences between research and practice topics are indications of correspondence between the domains of inquiry. The study measures correspondence by levels and changes in relative topic coverage. The study further analyzes research articles’ use of theory, sources of data, and methods of analysis, which are sorted by topic and publication outlet. The remainder of this section describes the study’s research domain, sampling plan, data collection, and data analysis.

Research Domain The study’s research domain is limited to published articles that address conventional management accounting topics (i.e. as reflected in management accounting textbooks) plus several that additionally are salient in the professional financial and accounting literature (described in the next section). Both published research and practice topics are assumed to be reasonable proxies of issues and questions of interest to researchers and practitioners. Several problems arise in the use of these proxies. (1) It is well known that time between completion and publication of articles differs between the research and practice literatures. This study examines various time lags between research and practice topics to account for the publication lag. (2) Not all research efforts or practice issues appear in the published literature. This study assumes that unpublished research articles do not meet academic quality standards, although some researchers might harbor other explanations. This study also compares the practice literature to the IMA’s study of practice to confirm conformance between the practice literature and issues expressed by surveyed practitioners (see Note 2 and the later discussion of aggregate results).


5

The study considers an article to be of direct interest to “management accountants” if it addresses one or more of the following topics: Accounting software. Improving profits. Budgeting. Internal control. Business process improvement. Management accounting practices. Cash management. Management control. Compensation plans. Outsourcing. Cost accounting. Performance measurement. Cost management. Research methods. Effects of financial reporting on Shareholder value. internal systems. Effects of information technology on internal systems. Sampling The study analyzes articles that appeared in print during the years 1996–2000. This five-year period witnessed dramatic changes in technology, business conditions, and the responsibilities of financial and accounting professionals. There is no reason to believe that future years will be any less volatile. The study further defines the domain of management accounting research as articles fitting the above topics that were published in the following English-language research journals: Academy of Management Journal Journal of Accounting and (AMJ) Academy of Management Review

Economics (JAE) Journal of Accounting Research

(AMR) Accounting and Finance (A&F)

(JAR) Journal of Management Accounting

Accounting Organizations and

Research (JMAR) Management Accounting Research

Society (AOS) Advances in Management

(MAR) Review of Accounting Studies (RAS)

Accounting (AIMA) Contemporary Accounting

Strategic Management Journal (SMJ)

Research (CAR) Journal of Accounting, Auditing,

The Accounting Review (TAR)

and Finance (JAAF) We assume that the research literature in other languages either covers similar topics or is not related to the practice literature aimed at English-speaking professionals.3

6


Similarly, the study defines the domain of management accounting practice to be articles fitting the topical boundaries that were published in English-language professional magazines and journals aimed at financial managers, executives, and consultants. We, therefore, assume that articles published in the professional literature accurately reflect issues of importance to professionals themselves. The professional literature sources include: Strategic Finance (SF) Sloan Management Review (SMR) Management Accounting Harvard Business Review (HBR) (MA-U.S. and U.K.) Journal of Accountancy (JOA) Financial Executive (FE)

Business Finance (BF)

Data Collection The study uses the online, electronic contents of the abstracts of management accounting articles from research and practice journals published during the years 1996–2000 as its source of data. The study includes the entire contents of explicitly named management accounting journals (e.g. Advances in Management Accounting, Strategic Finance) and selected articles from other journals and magazines if articles matched the topic domain. The database of management accounting articles consists of information on: 373 research articles; 1,622 professional or practice articles.

Data Analysis Qualitative Method The study uses a qualitative method to label, categorize, and relate the management accounting literature data (e.g. Miles & Huberman, 1994). The study uses Atlas.ti software (www.atlasti.de), which is designed for coding and discovering relations among qualitative data.4 The study began with predetermined codes based on the researchers’ expectations of topics, methods, and theories. As normally happens in this type of qualitative study, the database contains unanticipated qualitative data that required creation of additional codes. This necessary blend of coding, analysis, and interpretation means that the coding task usually cannot be outsourced to disinterested parties. Thus, this method is unlike content analysis, which counts pre-defined words, terms, or phrases.


Table 1. Article Database Codes. ARTICLE article-ABSTRACT article-AUTHOR(S) article-DESCRIPTORS article-journal-A&F article-journal-AIMA article-journal-AMJ article-journal-AOS article-journal-CAR article-journal-JAAF article-journal-JAE article-journal-JAR article-journal-JMAR article-journal-MAR article-journal-RAS article-journal-SMJ article-journal-TAR article-TITLE article-VOLUME & PAGES article-YEAR article-year-1996 article-year-1997 article-year-1998 article-year-1999 article-year-2000 buddgeting-soc psych budgeting-agency budgeting-contingency budgeting-econ-other budgeting-jdm budgeting-org change budgeting-soc justice budgeting-systems GEOGRAPHY geography-ANZ geography-ASIA geography-EUROPE geography-INTERNATIONAL geography-LATIN AMERICA geography-NORTH AMERICA METHOD method-ANALYSIS method-analysis-analytical method-analysis-qualitative method-analysis-statistical

7

8


Table 1. (Continued ) method-ARCHIVAL method-EXPERIMENT method-FIELD/CASE STUDY method-LOGICAL ARGUMENT method-SURVEY THEORY theory-AGENCY theory-CONTINGENCY theory-CRITICAL theory-ECONOMIC CLASSIC theory-INDIVID/TEAM JDM theory-ORGANIZATION CHANGE theory-POSITIVE ACCOUNTING theory-SOCIAL JUSTICE/POWER/INFLUENCE theory-SOCIAL/PSYCH theory-SYSTEMS theory-TRANSACTION COST TOPIC topic-BUDGETING topic-budgeting-activity based topic-budgeting-capital budgeting topic-budgeting-general topic-budgeting-participation topic-budgeting-planning&forecasting topic-budgeting-slack topic-budgeting-variances topic-BUSINESS INTELLIGENCE topic-BUSINESS PROCESSES topic-business processes-credit management topic-business processes-fixed assets topic-business processes-inventory management topic-business processes-procurement topic-business processes-production management topic-business processes-reengineering topic-business processes-travel expenditures topic-CASH MANAGEMENT topic-cash management-borrowings topic-cash management-collections topic-cash management-credit policies topic-cash management-electronic banking topic-cash management-electronic exchange topic-cash management-foreign currency topic-cash management-investing topic-cash management-payments topic-COMPENSATION


Table 1. (Continued ) topic-compensation-accounting measures topic-compensation-design/implementation topic-compensation-executive topic-compensation-pay for performance topic-compensation-stock options topic-CONTROL topic-control-alliances/suppliers/supply chain topic-control-complementarity/interdependency topic-control-cost of capital topic-control-customers/customer profitabilty topic-control-environmental topic-control-information/information technology topic-control-intangibles topic-control-international/culture topic-control-JIT/flexibility/time topic-control-org change topic-control-quality topic-control-R&D/new product develop topic-control-risk topic-control-smart cards/purchasing cards topic-control-strategy topic-control-structure topic-control-system topic-COST ACCOUNTING topic-cost accounting-environmental topic-cost accounting-general topic-cost accounting-standards topic-cost accounting-throughput topic-COST MANAGEMENT topic-cost management-ABC topic-cost management-ABM topic-cost management-benchmarking topic-cost management-cost efficiency/reduction topic-cost management-cost negotiation topic-cost management-costing topic-cost management-process mapping topic-cost management-quality/productivity/tqm topic-cost management-shared services topic-cost management-strategy topic-cost management-target costing topic-cost management-theory of constraints/capacity topic-ELECTRONIC topic-electronic-business topic-electronic-commerce topic-electronic-intranet topic-electronic-processing

9

10


Table 1. (Continued ) topic-electronic-web sites topic-electronic-xml/xbrl topic-EXPERT SYSTEMS topic-FINANCIAL ACCOUNTING topic-financial reporting-accounting standards/SEC topic-financial reporting-depreciation topic-financial reporting-drill downs topic-financial reporting-e reporting topic-financial reporting-environmental topic-financial reporting-general topic-financial reporting-international topic-financial reporting-open books topic-financial reporting-realtime accounting topic-INTERNAL CONTROL topic-internal control-controls topic-internal control-corporate sentencing guidelines topic-internal control-data security/computer fraud topic-internal control-ethics topic-internal control-fraud awareness/detection topic-internal control-internal audit topic-internal control-operational audits topic-MANAGEMENT ACCOUNTING-practices topic-OTHER topic-OUTSOURCING DECISION topic-PERFORMANCE MEASUREMENT topic-performance measurement-balanced scorecard topic-performance measurement-business process topic-performance measurement-EVA/RI topic-performance measurement-evaluation/appraisal topic-performance measurement-group topic-performance measurement-incentives topic-performance measurement-individ topic-performance measurement-manipulation topic-performance measurement-nonfinancial topic-performance measurement-productivity topic-performance measurement-strategic topic-performance measurement-system topic-PRICING topic-PROFITABILITY topic-PROJECT MANAGEMENT topic-RESEARCH METHODS topic-SHAREHOLDER VALUE topic-SOCIAL RESPONSIBILITY topic-SOFTWARE topic-software-ABC/product costing topic-software-accounting technology (general)


11

Table 1. (Continued ) topic-software-budgeting topic-software-costing topic-software-credit analysis topic-software-data conversion topic-software-database topic-software-decision support topic-software-document management topic-software-erp topic-software-fixed assets topic-software-graphical accounting topic-software-groupware topic-software-human resources/payroll topic-software-internet topic-software-mindmaps topic-software-modules topic-software-operating system topic-software-project accounting topic-software-purchasing topic-software-reporting topic-software-sales/C/M topic-software-selection/accounting platforms/implementation topic-software-spreadsheets topic-software-t&e topic-software-warehousing/datamarts/intelligent agents topic-software-workflow topic-software-year2000 compliant topic-TRANSFER PRICING topic-VALUATION topic-VALUE BASED MANAGEMENT topic-VALUE CHAIN

Table 1 contains the complete list of research-literature codes used in this study. The practice literature codes are identical except for journal codes. Codes shown in capital letters (e.g. ARTICLE) are major codes, or “supercodes,” that contain related minor or subcodes (e.g. article-ABSTRACT). An “other” code collects topics that apparently are of minor interest at this time. Figure 1 displays sample information related to one of the data records. The left-hand panel shows a typical article’s data, while the right-hand panel contains the codes applied by the researchers to the data. An article may cover several topics and use several methods and theories; thus the numbers of topic, method, and theory codes exceeds the number of articles in the sample. The software’s query features allow nearly unlimited search and discovery of relations among coded data. These queries form the analyses that follow in this study.

12


Fig. 1. Example of Coded Article Data.

Measures of Correspondence The study measures correspondence between research and practice to capture different dynamics of information exchange between the realms of inquiry. The study defines differences in changes and levels of topic frequency as measures of correspondence. Research and practice topic frequencies are scaled by the total number of research or practice topics to control for the relative sizes of the two outlets. The study examines contemporaneous and lagged differences, as the data permit, for evidence of topic correspondence. Furthermore, the study investigates whether research topic frequency leads or lags practice. Validity Issues One researcher coded all of the practice article abstracts in the database and a 5% random sample of the research abstracts. Another researcher coded all of the research abstracts and a 5% random sample of the practice abstracts. Interrater reliability of the overlapped coding was 95%, measured by the proportion of coding agreements divided by the sum of agreements plus disagreements from the


13

5% random samples of articles in the research and practice databases.5 Because the measured inter-rater reliability is well within the norms for this type of qualitative research (i.e. greater than 80%) and because hypothesis testing or model building is not the primary objective of the study, the researchers did not revise the database to achieve consensus coding. Aggregate Analysis Figure 2 shows the most aggregated level of analysis used in this study, which reflects the levels of research and practice frequencies of major topics. The three most frequent practice topics in Fig. 2 are: (1) software; (2) management control; and (3) cost management. 6 The Institute of Management Accountants (IMA) analyzed the practice of management accounting (1997, 2000) in part by asking respondents to identify critical work activities that are currently important and that are expected to increase in the future. The IMA reports that 21% of respondents identified computer systems and operations as one of the five most critical current work activities and 51% believe that this work activity will increase in importance in the future. Eighteen percent of respondents in the IMA practice analysis state that control of customer and product profitability is one of the most critical work activities; however, 59% of respondents believe that this is one of the work activities that will increase

Fig. 2. Major Topic Frequencies, 1996–2000.

14


in importance in the future. The topic code “management control” includes sub-topics related to control of customers, customer profitability, quality, and new products. Finally, the IMA practice analysis found that 25% of respondents stated that “financial and economic analysis” was one of the most critical current work activities. Forty-two percent believed it would be more important in the future. The topic code “cost management” includes cost reduction, efficiency, activity-based management, and activity-based costing. The aggregate results of applying this study’s coding scheme to the practice literature are consistent with those of the IMA’s practice analysis. The similarity of aggregate results from this study and the IMA’s survey of practice support the validity of this study’s coding scheme.

ANALYSIS OF TOPIC FREQUENCY CHANGES The qualitative software enables several types of “drill-down” analyses at major topic and subtopic levels. These analyses support the statistical and graphical analyses that follow. The basic analysis in Fig. 2 guides all subsequent analyses. Relatively large differences in overall topic frequency are evident in this graph (e.g. budgeting, management control, performance measurement, and software), but more detailed analyses are used to identify less researched questions. Associated changes in topics can be evidence of information exchange between research and practice. If researchers and practitioners are communicating about topics of mutual interest, one expects changes in topic frequency (contemporaneous or lagged) to be closely associated over time. Creating tables of topic frequencies for each year (by disaggregating the data underlying Fig. 2) supports an investigation of contemporaneous and lagged topic changes. The study finds no significant correlations (␣ = 0.10) between changes in research and practice topic-frequencies that are either contemporaneous or lagged (plus or minus one year). Analysis of topic levels finds numerous opportunities for communication and exchange of findings between research and practice.

ANALYSIS OF TOPIC FREQUENCY LEVELS Contemporaneous Frequency Levels Analysis of contemporaneous levels shows some evidence of topic correspondence. For example, a glance at Fig. 2 shows visual correspondence. The contemporaneous overall correlation coefficient, which equals 0.45, is highly significant (p < 0.0001). We obtain similar overall results for individual years


15

(0.3 < R < 0.6). Note that these annual correlations do not reflect a monotonic increase of correspondence over time. However, the data show that modest contemporaneous correspondence of research and practice topics exists.

Lagged Frequency Levels Analysis of lagged topic frequency levels also shows similar correspondence. Examining whether practice leads research by 1 year yields an overall correlation coefficient (rounded) of 0.4 (p < 0.0001). Annual correlation coefficients range between 0.3 and 0.5 for each lagged year. These are also highly significant and reflect a “U” shaped pattern over time. Testing if research leads practice by 1 year generates an overall correlation coefficient of 0.5 and 0.3 < R < 0.6 for each lagged year (all highly significant). Furthermore, coefficients of research leading practice increase monotonically, suggesting increasing correspondence over time. Thus, this study finds mixed evidence of correspondence between research and practice: Analysis of lagged topic frequency levels suggests increasing correspondence, but changes in topic frequency show no evidence. This suggests that evidence of correspondence may reflect coincidence rather than active or causal exchange of information between researchers and professionals. To resolve this ambiguity we look more closely at topic levels.

ANALYSIS OF CORRESPONDENCE OF TOPIC LEVELS One can observe many instances in Fig. 2 where topic frequency differences are less than 5%, which indicate high correspondence between research and practice. Most of these topics apparently are of relatively minor interest to both researchers and professionals (i.e. total frequency of either practice or research is less than 5%). While these low frequency topics may represent emerging areas for both realms, we focus here on topics that also have at least 5%7 of the total article coverage in either practice or research. The only major topic meeting these criteria is “cost management.”

Cost Management Topics coded as cost management comprise approximately 14% of all practice topics and 13% of research topics, leaving only a 1% difference. Is this high

16


Fig. 3. Cost Management Sub-Topics, 1996–2000.

correspondence the result of coincidence or cross-fertilization? To answer that question, one can drill down into the database to contrast cost-management subtopics. The result of this analysis is shown in Fig. 3. A close look at Fig. 3 indicates that general cost-management correspondence is questionable. Benchmarking is the only subtopic with appreciable topic frequency and relatively high correspondence, comprising roughly 13% of practice and 9% of research subtopics. Examination of benchmarking-research articles shows they are evenly split between prescription and statistical analyses of the properties of benchmarks. There are, however, no research studies of the impacts of benchmarking. Practice articles are either prescriptions or self-reports of implementation or reports of organizational improvements attributed to benchmarking. Benchmarking Questions Several benchmarking research questions seem obvious, including: What are the costs and benefits of benchmarking at the process, service, or firm levels? One should be able to measure costs of benchmarking activities, but, as is usually the case, benefits may be more elusive. Attributing improvements in processes to benchmarking may be more feasible than attempting to explain business unit or firm-level financial performance. What are the attributes of successful or unsuccessful design and implementation of benchmarking? Addressing this question perhaps should follow the first unless one wants to proxy costs and benefits with user satisfaction measures.


17

Given that apparent correspondence at the cost-management subtopic level yielded new research questions, examination of other cost management subtopics also might bear fruit. For example, only 3% of research studies exist in the area of activity-based management (ABM) – a difference of nearly 8% – which seems surprising given this topic’s high profile over the past decade, and no research on shared services – a difference of 6%. Several interesting research questions for these topics include: Activity-Based Management Questions Several ABM research questions from practice are: Does ABM lead to observable improvements in processes, products, services, and financial performance? Self-reports indicate that ABM delivers improvements, but one suspects that these self-reports are censored and most reports of failures are either not written or published. What are determinants of successful ABM efforts? Determinants may include communication, team structure, management style, and management support and involvement. How can organizations successfully move from ABM pilot projects to wider deployment? Most ABM self-reports reflect results of limited pilot projects. Does implementation of ABM spread? How?

Shared Services Questions The topic of shared services refers to centers that provide business services, such as finance, human resources and legal. This service center would contract with business units, much as an outsourced-service provider would. Questions include: What are the efficiencies of locating business services in shared service centers? Cost savings are part of the equation, but effects on usage and quality of service also are important. This leads to related considerations of transfer pricing and performance evaluation. Is ABM necessary to justify shared or outsourced services? Is ABM the tool to use to identify opportunities, communicate rationales, and ease transition to shared service centers? What are the organizational impediments and arguments for shared services vs. distributed or outsourced services? Alternative organizational structures and contracting may have inertia and power considerations as well as economic.

Relatively more research than practice exists in several cost-management areas, including activity-based costing (ABC) – a difference of 29% – and strategy – a difference of 5%. However, perhaps surprisingly, numerous practice-oriented questions remain relatively un-researched. Activity-Based Costing Questions Practice-oriented research questions in this area include: What is the optimal complexity of ABC systems? Costs and benefits of complexity include design and maintenance costs, cognitive complexity, and value of finer information. Standard costing systems are notoriously expensive to maintain; are ABC systems even more so? Is it

18

FRANK H. SELTO AND SALLY K. WIDENER possible to prescribe optimal complexity or describe the complexity of apparently successful ABC systems? Are different levels of complexity appropriate for different purposes (e.g. product costing vs. strategic decision making)? Are objectivity and precision of measurement incompatible with efficient ABC systems? What are the information quality tradeoffs? This has added importance for ABC systems that are intended to serve multiple purposes, including reporting, costing, decision making, and performance measurement.

Strategy Questions Most research studies use measures of strategy as independent variables to explain performance or other organizational outcomes. Practical concerns related to strategy include: What are appropriate ratios or indicators to measure whether an organization is meeting its strategic goals, which may be heavily marketing and customer oriented? Are these indicators financial, non-financial, or qualitative? Numerous practice articles argue that strategic management is possible only with the “right indicators.” But what are they? How are they used? With what impact? Is the balanced scorecard an appropriate tool for performance evaluation, as well as for strategic planning and communication? The BSC is offered as a superior strategic planning and communication tool. Many organizations are inclined to also use the BSC (or similar, complex performance measurement models) as the basis for performance evaluations. What complications does this extension add? With what effects? Are scenarios from financial planning models effective tools for strategic management? Financial modeling is an important part of financial and cost management. Do strategic planners need or use scenarios from these models? Why or why not? With what effects?

Relatively less research than practice exists in the area of cost reduction/efficiency – a difference of 22%. Practice coverage of this topic is fairly uniform over the 5-year study period. Although research coverage peaked in 1998, some coverage continued into 2000. Cost Reduction/Efficiency Questions Perhaps accounting researchers consider issues of cost reduction and efficiency to be too basic for research inquiry. Practice is concerned with interesting developments that could benefit from a research perspective. Sample questions from practice include: What is the efficiency of spending to increase customer satisfaction? Though a number of researchers have addressed this issue (e.g. tests of statistical relations between customer satisfaction and profitability), it is of continuing interest to practice, particularly at the level of developing guidelines for efficient management of customer relations. What are the effects of IT and organizational changes on efficiency? There appears to be no research that addresses this type of question. Relevant contexts abound, including telecommuting, outsourcing, and internal support services.


19

What are the effects of IT on total costs and productivity? The information systems literature commonly focuses on measuring IT-user satisfaction while larger issues of efficiency remain under-researched. Application contexts include finance, human resources, procurement, payables, travel, payroll, customer service.

Perhaps surprisingly, observed major-topic correspondence yields much evidence of low correspondence and many interesting, less researched questions. Even when researchers and professionals address similar topics, they focus on different questions. An examination of topics with more obvious low correspondence yields even more research opportunities.

ANALYSIS OF LOW CORRESPONDENCE OF TOPIC LEVELS Low correspondence is defined as topic-frequency differences in excess of 5%. These differences include topic areas where research exceeds practice: budgeting (6.5% difference), management control (14% difference), and performance measurement (16% difference). The data also reveals major topic areas where practice exceeds research, including business processes (5% difference), internal control (6% difference), electronic business (7% difference), and software (19% difference). Perhaps these are areas of particularly abundant new research opportunities.8 The study presents analysis of low correspondence where budgeting research exceeds practice and where practice writings on electronic business issues exceed research. The appendix contains similarly identified research questions from each of the other topic areas. Budgeting (Research > Practice) Budgeting is a venerable management accounting research topic, and one might think that there are few un-researched questions remaining. If so, have researchers not communicated results to practice, or are they pursuing less practice-oriented topics? As before, one can drill down into the data to the budgeting sub-topic level to assess budgeting correspondence. Figure 4 presents topic frequencies of budgeting sub-topic coverage. The data contain no practice publications in topic areas of budget slack (Difference = 10%) and budget variances (Difference = 16%). More research than practice exists in the areas of capital budgeting (Difference = 11%) and participative budgeting (Difference = 21%). These topics appear to be of little

20


Fig. 4. Budgetting Sub-Topics, 1996–2000.

current, practical interest, but they continue to attract research efforts, perhaps because of tradition and the interesting theoretical issues they present. It also is possible that researchers’ long concern with budgetary slack still leads practice. For example, excess budget slack conceivably might be included with other dysfunctional actions designed to manipulate reported performance and targeted for elimination by financial reforms. Conversely, the data contain no research in topic areas of activity-based budgeting (Difference = 10%) and planning & forecasting (Difference = 65%). The latter area, planning and forecasting, has a large topic difference and has grown in practice coverage each year of the study period. Planning and Forecasting Questions Just a few questions from practice include: What are the determinants of effective planning and forecasting? Effective planning and forecasting can be defined as: (1) accurate, timely, and flexible problem identification; (2) communication; and (3) leading to desired performance. Researchers may find environmental, organizational, human capital, and technological antecedents of effective planning and forecasting methods and practices. Whether these are situational or general conditions would be of considerable interest.


21

What exogenous factors affect sales and cost forecasting? This includes consideration of the related question, What is a parsimonious model? Nearly every management accounting text states that sales forecasting is a difficult task. Likewise, cost forecasting can be difficult because of the irrelevancy or incompleteness of historical data. Yet both types of forecasting are critical to building useful financial models and making informed business decisions. What are effects of merging BSC or ABC with planning & forecasting? ABC and the balanced scorecard represent current recommendations for cost and performance measurement. However, the research literature has not extensively considered the uses or impacts of these tools, which may be particularly valuable for planning and forecasting. What are the roles of IT & decision-support systems in improving planning & forecasting? Most large organizations use sophisticated database systems, and accessing and using information can be facilitated by intelligent interfaces and decision support systems. Yet we know little about the theoretical and observed impacts of these tools in general and almost nothing about their effects on planning and forecasting.

Electronic Business (Practice Research) Topics coded as electronic business comprise approximately 7% of all practice topics, yet there is no management accounting research in this area. To determine if perhaps researchers are investigating electronic business issues and publishing in journals outside of mainstream management accounting journals, we also reviewed Information Systems research journals (MISQ, JMIS, JIS) and found no evidence that research in this area is being conducted in these journals. Due to the increasing emphasis of the role that technology plays in business in today’s competitive, global and fast-changing world, a 7% difference in this topic with no research seems surprising. Surely, electronic business is a research topic guaranteed to generate practical interest. An in-depth look at Fig. 5 shows that there are several main categories of sub-topics within electronic business. Approximately 31% of electronic business topics are articles of a general nature, 25% are related directly to issues on electronic commerce, 22% are concerned with the internet and websites, 16% are about processing transactions electronically, and the remaining 6% focus on the use of XML (extensible markup language). Electronic Business (General) Questions General electronic business issues center on the reengineering of business processes and business models to take advantage of electronic means of transacting business and creating efficiencies and enhanced performance for the firm. This generates opportunities for research questions related to the successful start-up of e-ventures, the changes in underlying business models, and the use of technology to reduce costs.

22


Fig. 5. Electronic Business Sub-Topics, 1996–2000. What are appropriate management controls, internal controls, and performance measures for E-business ventures? This includes the related question, Do they differ from conventional business? Doing business in the “New Economy” has impacted the underlying business model of most firms thus impacting the design of the firm’s management control system, internal control environment, and performance measurement system. What technologies drive enhanced productivity and efficiencies in the firm? Firms must be able to perform cost/benefit analysis weighing the potential benefits to be gained from employing new technologies against the cost of implementing that technology and reengineering the business process. Two related questions are What is the optimal capital budgeting model for electronic business? and Which business processes lend themselves to a reengineering process that would result in increased efficiencies and reduced costs?

Electronic Commerce Questions Practitioners appear to be primarily concerned with the management of costs in the electronic commerce space and the proper tracking and measurement of performance. These concerns lead to several promising research questions: What is the optimal amount for web-retailers to spend on customer acquisition costs? The prevailing business model among e-tailers was to increase traffic on the website and worry about revenues later. But what is too much to pay to acquire a new customer? How do e-tailers know how much to spend on customer acquisition costs? What performance metrics do firms need to track to effectively manage electronic customers and the integration of e-commerce with their current business model? How does a firm evaluate investments in e-commerce? What metrics are appropriate for measuring the performance of e-commerce initiatives? Not too long ago metrics focused on traffic, now firms are more focused on the generation of revenue. Identifying the appropriate drivers, outcome measures, and the timing and pattern of associations between the two are interesting areas for potential research.


23

How does electronic data interchange affect the management control system? Electronic commerce is changing traditional business practices in areas such as increased use of bar coding of transactions and inventory, and the use of electronic procurement. How do these new business practices impact the design of the MCS?

Internet and Website Questions The internet both facilitates the timeliness, exchange, and availability of information. Practice is particularly concerned with the impact the internet has on the reporting and use of financial information. What are the characteristics of an effective website? Firms are implementing intranets and websites for communication of information within the firm. This question includes a related question: What characteristics of websites facilitate effective exchange of accounting information between the firm and its investors? Or between users and/or business units within the firm? What is the impact of displaying financial information on a web site on the firm’s business risk? Firms now distribute financial accounting information on company websites. How does this practice impact the firm’s internal control environment? How does it impact a company’s risk of litigation? Are there controls that the firm can implement to reduce the associated risk?

Electronic Processing The processing of accounting transactions can be a tedious and time-consuming task. Electronic processing of transactions can create efficiencies within organizations. Questions of interest are primarily related to how electronic processing can improve firm performance and efficiencies. How should accounting workflows and transaction processing be reengineered to take advantage of electronic processing? Firms need to know how to integrate an environment that traditionally generates lots of paper and incorporates many formal controls with an electronic processing environment that may not generate any paper and dispenses with some of the traditional controls. What is the impact of electronic processing on the firm’s control environment? With the potential for increased efficiencies arising from the reduction of traditional paper documents, there may not be a paper trail left to substantiate and document transactions. What is the impact on internal control? Is a paperless environment cost effective?

Extensible Markup Language Extensible markup language (XML) (also, extensible business reporting language, XBRL, and extensible financial reporting markup language, XFRML) is fast becoming the language of accounting. XML is used for a multitude of purposes including reporting accounting information to investors via the firm’s website, uploading of SEC files and so forth. This leads to the question: How do accountants successfully use extensible markup language (XML) to facilitate the exchange and communication of accounting information?

24


OPPORTUNITIES FOR PUBLICATION This section of the study addresses designing research for publishability. Because the major portion of the study has focused on identifying new research questions, it seems only prudent to anticipate the opportunities to publish this novel research. It is one thing to recommend that researchers take risks and tackle new research questions, but it might be quite another to get these efforts published in quality research journals. The analysis that follows finds that some research journals, which published management accounting articles during the period of study, have specialized but others have been more general. Certainly, publication history might be an imperfect predictor of future publications opportunities, but a Bayesian might condition estimates of publication probability with priors based on history. Prudence (or a strategic approach to conducting research) also suggests that researchers conceive and design their efforts to meet target journals’ revealed preferences. The study next analyzes the management accounting research database for topic coverage by major journal. The study also analyzes each journal’s past publication practices regarding underlying theory, sources of data, and methods of analysis. This analysis is not intended to be a cookbook, but rather it is intended as realistic guidance based on historical evidence. Figure 6 displays coverage of major management accounting topics by journal. Figure 7 shows theories used in management accounting articles by journal. Similarly, Fig. 8 shows sources of data by journal. Finally, Fig. 9 shows methods of analysis by journal.

Fig. 6. Management Accounting Topics by Journal, 1996–2000.


25

Fig. 7. Theories Used in Management Accounting Articles by Journal, 1996–2000.

Fig. 8. Methods of Analysis Used in Management Accounting Articles by Journal, 1996–2000.

26


Fig. 9. Sources of Data in Management Accounting Articles by Journal, 1996–2000.

Topic Coverage Figure 6 shows some evidence of journal specialization by topic, although all the research journals have published at least some articles addressing these topics. For example, management control topics have appeared most often in AOS and MAR, both U.K.-based journals. Performance measurement issues have appeared most often in the North American journals, JAR, CAR, TAR, and JAE, and these journals also publish management control articles. This concentration may reflect editorial policies or results of years of migration of topics. Apart from concentration of performance measurement and management control, it appears that all the surveyed journals are open to publishing various management accounting topics.

Theories Figure 7 shows some strong evidence of theory specialization by journals. For example, virtually the only theories used in management accounting articles published in JAR, etc. are economic in nature (agency or microeconomic theories). This is true also for management accounting articles published in predominantly management journals, SMJ, etc. Nearly the only outlets for papers


27

using contingency theory are the U.K. journals, AOS and MAR. These journals plus AIMA and JMAR appear to be the broadest in using alternative theories. Methods of Analysis As shown in Fig. 8, articles in JAR, etc. tend to use either analytical or statistical methods, but almost never use qualitative analyses. On the other hand, management accounting articles in other journals rarely use analytical methods, though they often use statistical methods. For example, articles in AIMA, AOS, and JMAR most often use statistical methods. Qualitative analysis appears mostly in the U.K. journals, AOS and MAR, followed by AIMA and JMAR. Sources of Data Figure 9 shows specialization by journals in their uses of alternative data sources. JAR, etc. articles predominantly use archival data, though data from laboratory experiments also appear in JAR, etc. Field study and survey data appear most often in AOS and MAR, the U.K. journals. AIMA appears to be the most balanced in its data sources. JMAR, though a small player, publishes papers with a wide range of data, as does MAR and, to a lesser degree, AOS. Conclusions about Publication Opportunities Authors want to place their work in the most prestigious journals (a designation that varies across individuals and universities) and also want to receive competent reviews of their work. Thus it seems sensible (or perhaps explicitly strategic) to design research for publishability in desired outlets. As a practical matter, this strategic design perspective may lead researchers, who themselves specialize in theories and methods, to design practice-oriented management accounting research for specific journals. Historical evidence indicates that all surveyed journals may be open to new topics. Although several journals seem open to alternative theories and methods (AIMA, JMAR, MAR, AOS), the major North American journals have been more specialized. This may reflect normative values and practical difficulty of building and maintaining competent editorial and review boards. Thus, if one wants to pursue a new topic in research aimed at JAR, etc., for example, one perhaps should use a theory, source of data, and method that these journals have customarily published.

28


CONCLUSION There is no shortage of interesting, potentially influential management accounting research questions. From an analysis of published research and practice articles, this study has identified many more than could be reported here. Even where research and practice topics appear to correspond, considerable divergence in questions exists. Identified research questions offer opportunities for ALL persuasions of accounting researchers. Synergies between management accounting and accounting information systems seem particularly obvious and should not be ignored. Furthermore, research methods mastered by financial accountants and auditors can be applied to management accounting research questions. Even with efforts to design practice-oriented management accounting research for publishability, challenges to broader participation and publication might remain. Some of the challenges to publishing this type of management accounting research might include lack of institutional knowledge of authors, reviewers, and editors. To be credible, authors must gain relevant knowledge to complement their research method skills. For example, research on management control of information technology and strategic planning should be preceded by knowledge of the three domains, in theory and practice. Furthermore, editors and reviewers who want to support publication of practice-oriented research should be both knowledgeable of practice and open minded, particularly with regard to less objective sources of data. However, it does not seem necessary or desirable to lower the bar on theory or methods of analysis to promote more innovative research. In summary, we hope that this paper encourages management accounting researchers to take on the challenges of investigating interesting, innovative questions oriented to today’s business world and practice of management accounting.

NOTES 1. See http://www.imanet.org/content/Publications and Research/IMAstudies/moreless. pdf. 2. Although the data are available, we have resisted the temptation to classify the practice orientation of management accounting researchers or educational institutions. 3. Some management accounting researchers are placing work in other management and operations journals, such as Management Science. Omitting these articles could be a source of sampling bias if this is a growing trend. 4. Malina and Selto (2001) describe this qualitative method in more detail. 5. Ninety-seven article abstracts (containing 126 supercodes) were dual coded by both researchers. Five articles contained multiple codes of which one super code in each article was not in agreement between researchers.


29

6. The term “software” reflects selection, implementation, and management of software systems and the hardware to run them. “Cost management” refers to activities to create more value at lower cost and is distinguished from cost accounting, which measures costs. 7. The 5% cutoffs are arbitrary but retain the great majority of research and practice articles for study. Without some cutoff, the research would resemble an annotated bibliography of 2,000 articles. We do run the risk of ignoring particularly interesting but relatively unreported topics. 8. In concept, one might prefer to separate practice descriptions of emerging problems from advocacy for preferred solutions. Some also might argue that research naturally investigates different topics than practice. This study regards all differences as opportunities for interesting research.

ACKNOWLEDGMENTS We acknowledge and thank Shannon Anderson, Phil Shane, Naomi Soderstrom and participants at the 2003 Advances in Management Accounting Conference, 2002 MAS mid-year meeting, a University of Colorado at Boulder workshop and the AAANZ-2001 conference for their comments and suggestions for this paper.

REFERENCES Anderson, P. F. (1983). Marketing, scientific progress, and scientific method. Journal of Marketing, 47(4, Fall), 18–31. Atkinson, A. A, Balakrishnan, R., Booth, P., Cote, J., Groot, T., Malmi, T., Roberts, H., Uliana, E., & Wu, A. (1997). New directions in management accounting research. Journal of Management Accounting Research, 79–108. Demski, J. S., & Sappington, D. E. M. (1999, March). Summarization with errors: A perspective on empirical investigations of agency relationships. Management Accounting Research, 10(1), 21–37. Elnathan, D., Lin, T., & Young, S. M. (1996). Benchmarking and management accounting: A framework for research. Journal of Management Accounting Research, 37–54. Institute of Management Accountants (2000). Counting more, counting less. http://www.imanet.org/content/publications and research/IMAstudies/moreless.pdf. Ittner, C. D., & Larcker, D. F. (1998). Innovations in performance measurement: Trends and research implications. Journal of Management Accounting Research, 205–238. Ittner, C. D., & Larcker, D. F. (2001). Assessing empirical research in managerial accounting: A value-based management perspective. Journal of Accounting and Economics. Malina, M. A., & Selto, F. H. (2001). Communicating and controlling strategy: An empirical study of the effectiveness of the balanced scorecard. Journal of Management Accounting Research, 13, 47–90. Miles, M., & Huberman, A. (1994). Qualitative data analysis: An expanded sourcebook. Thousand Oaks, CA: Sage. Shields, M. D. (1997). Research in management accounting by North Americans in the 1990s. Journal of Management Accounting Research, 3–62.

30


APPENDIX: PRACTICE-ORIENTED RESEARCH QUESTIONS Major Topic

Sub-Topic

Selected Research Questions

Budgeting

Activity based Capital

What are effects of merging ABC with planning & forecasting? What is the optimal capital budgeting model for electronic business? What are the determinants of effective planning and forecasting? What are effects of merging the BSC with planning & forecasting? What exogenous factors affect sales and cost forecasting? What is a parsimonious model? What are the roles of IT & decision-support systems in improving planning & forecasting? Which activities in the finance function can be eliminated leading to reduced costs while maintaining high levels of support and integrity in the accounting information? Under what business conditions (e.g. size, industry, strategy, organization type, etc.) can the “lean support model” be effectively implemented in the finance function? Is the reduction in finance costs as a percent of sales “real” or have the costs and related work simply been shifted to other areas of the organization? There is anecdotal evidence supporting initiatives (e.g. corporate purchasing cards, single consolidated credit card system, online marketplace) designed to streamline the procurement system in order to increase efficiency and reduce costs. Which is most effective? What are the determinants of effective procurement initiatives?

Planning & forecasting

Business processes


31

Appendix (Continued ) Major Topic

Management control

Sub-Topic

Alliances

Customers/ Information Intangibles

Smartcards

Structure

Cost accounting

Environmental

Standards

Cost management

ABC

Selected Research Questions How does technology play a role in successfully reengineering business processes such as the travel and expense process? What is the nature of optimal business partner or exclusive supply contracts? What are the determinants of successful or failed alliances? What are appropriate controls for outsourced services or manufacturing? How do controls differ between using sole sources and a portfolio of suppliers? What are the costs, benefits, and risks of sharing information with customers? What is the influence of corporate culture on control, frequency and success of alliances? How does decentralizing purchasing authority via smartcards affect the control environment? What is the control tradeoff between keiretsu organizations and independence? How can firms measure current and future environmental costs and liabilities of products, processes, and projects? Can ABC and ABM reduce environmental costs and liabilities? How do cost accounting standards affect entry, exit, and profitability of affected firms? Do cost accounting standards result in improved contracting and performance? What is the optimal complexity of ABC systems?

32



Sub-Topic

ABM

Benchmarking

Cost reduction/ efficiency

Shared services

Strategy

Selected Research Questions Are objectivity and precision of measurement incompatible with efficient ABC systems? What are the information quality tradeoffs? Does ABM lead to observable improvements in processes, products, services, and financial performance? What are determinants of successful ABM efforts? How can organizations successfully move from ABM pilot projects to wider deployment? What are the costs and benefits of benchmarking at the process, service, or firm levels? What are the attributes of successful or unsuccessful design and implementation of benchmarking? What is the efficiency of spending to increase customer satisfaction? What are the effects of IT and organizational changes on efficiency? What are the effects of IT on total costs and productivity? What are the efficiencies of locating business services in shared service centers? Is ABM necessary to justify shared or outsourced services? What are the organizational impediments and arguments for shared services vs. distributed or outsourced services? Is the balanced scorecard (BSC) an appropriate tool for performance evaluation, as well as for strategic planning and communication?


33


Electronic business

Sub-Topic

Commerce

Internet/WWW

Processing

XML

Internal control

Controls

Data security/ computer fraud

Selected Research Questions Are scenarios from financial planning models effective tools for strategic management? What are appropriate management controls, internal controls, and performance measures for E-business ventures? Do they differ from conventional business? What is the optimal capital budgeting model for electronic business? How do on-line auctions affect transaction costs? What are the impacts of intranet exchange of accounting information? What determines effective intranet exchange of accounting information? What is the impact of electronic processing on the firm’s control environment? How do accountants successfully use extensible markup language (XML) to facilitate the exchange and communication of accounting information? Is the implementation of controls a cost savings or a cost expenditure (e.g. what are the costs and benefits of certain control procedures? Which decision models can facilitate this decision?) How do small firms, perhaps without necessary resources to support data security, implement a strong control environment?

34



Sub-Topic


Ethics

What are the characteristics of an effective ethical environment (e.g. training programs, codes of conduct, ethical standards, tone at the top, etc.)? Does implementing and adhering to the Internal Control Framework reduce the prevalence of fraud and positively impact firm performance? What are the observable impacts of BSC implementation and use? What are the effects of alternative means of BSC implementation? How does the design and structure of incentive compensation systems relate to incidences of fraud? Can action-profit-linkage chains be designed to capture affects of investments in intangibles such as training, information technology and employee satisfaction? What are the appropriate ratios or indicators to measure whether an organization is meeting its strategic goals, which may be heavily marketing and customer oriented? Are these indicators financial, non-financial, or qualitative? What are the costs and benefits of maintaining dual costing systems to satisfy demands for information related to both strategic costing management and operational improvements? What steps can be taken to ensure that a software conversion is performed competently and efficiently?

Fraud awareness/ detection Performance measurement

BSC

Systems

Strategic

Software

Accounting


35


Sub-Topic


ERP

What are the benefits and costs of implementing an enterprise resource system? What are the attributes of efficient data mining? What are the effects of HR software on improvements in hiring, training, retention and evaluation (e.g. linked to BSC performance models)? What characteristics of accounting software packages affect the efficiency of the organization?

Human resources

Selection

THE PROFIT IMPACT OF VALUE CHAIN RECONFIGURATION: BLENDING STRATEGIC COST MANAGEMENT (SCM) AND ACTION-PROFIT-LINKAGE (APL) PERSPECTIVES John K. Shank, William C. Lawler and Lawrence P. Carr ABSTRACT An important management topic across a wide spectrum of firms is reconfiguring the value delivery system – defining the boundaries of the firm. Profit impact should be the way any value chain configuration is evaluated. The managerial accounting literature refers to this topic as “make versus buy” and typically addresses financial impact without much attention to strategic issues. The strategic management literature refers to the topic as “level of vertical integration” and typically sees financial impact in broad “transaction cost economics” terms. Neither approach treats fully the linkages all along the causal chain from strategic actions to resulting profit impact. In this paper we propose a theoretical approach to explicitly link supply chain reconfiguration actions to their profit implications. We use the introduction by Levi Strauss of Personal Pair™ jeans to illustrate the theory, evaluating the management choices by comparing profitability for one pair

Advances in Management Accounting Advances in Management Accounting, Volume 12, 37–57 © 2004 Published by Elsevier Ltd. ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12002-9

37

38

JOHN K. SHANK ET AL.

of jeans sold through three alternative value delivery systems. Our intent is to propose a theoretical extension to the make/buy literature which bridges the strategic management literature and the cost management literature, using A-P-L and SCM, and to illustrate one application of the theory.

THE PROBLEM SETTING More explicitly, managing and even re-engineering relationships all along the supply chain has become a central element of strategy for many organizations in a wide range of industrial contexts (Beamish & Killing, 1997; Gulati, 1995; Mowery, 1988; Nohria & Eccles, 1992). Several rationales for more formal “partnering” have been offered:

achieving production efficiency (Womack & Roos, 1990); sharing R&D risks (Westney, 1988); gaining access to new markets and skills (Kogut, 1988); reducing the time to market in the development of new products (Clark & Fujimoto, 1991); searching for new technological opportunities (Hagedoorn, 1993). It has been reported that firms which have restructured their value delivery system have experienced lower overhead costs, enhanced responsiveness and flexibility, and greater efficiency of operations (Lorenzoni & Baden Fuller, 1995). Furthermore, from the single-firm perspective, alliances and partnerships have created new strategic options, induced new rules of the game, and enabled new complementary resource combinations (Kogut, 1991). From the initial emphasis on joint ventures, a wider spectrum of forms of network alliances has emerged. The basic idea in “outsourcing strategies” is to transform the firm’s value chain to reduce the assets required and the number of traditional functional activities performed inside the organization, resulting in a much different configuration of the corporate boundaries. This can challenge the firm to carefully reconsider its core capabilities (Prahalad & Hamel, 1994). Alliances and partnerships create leverage in strategic maneuvering, shaping the so-called “intelligent firm” (Quinn, 1992) or the “strategic center” (Lorenzoni & Baden Fuller, 1995). In “networked organization” design, the key choice is which activities to perform internally and which to entrust to the network (Khanna, 1998). The choice can free resources from traditional supply chains to focus on core competencies that foster the firm’s competitive advantage. One example is Dell’s de-emphasis of manufacturing in favor of web-enhanced direct distribution in business PCs in the 1990s. Williamson (1975) frames this choice as how much “market” and how

The Profit Impact of Value Chain Reconfiguration

39

much “hierarchy” and sees it as a cost-benefit trade-off. If transaction costs in the marketplace exceed the benefits of the outsourcing, the vertical integration option (hierarchy) is preferred. Accounting information should play a fundamental role in evaluating the placement of the boundaries of the firm. Indeed, cost estimation for the resources employed and for the related benefits should shape how managers make decisions about the rational level of vertical integration. But management accounting and strategic management studies to date do not provide full financial analysis for this trade-off. Typically, “make/buy” decisions are framed in managerial accounting from a short-run, differential cost perspective. This approach typically ignores long-run costs and asset investments related to the management activities involved and also ignores linkages with customers and suppliers. In contrast, the strategic management literature frames such decisions in terms of transaction costs which emerge in using markets instead of hierarchies. The former approach includes very little strategy, while the latter includes very little accounting. We will argue here that they are not alternative viewpoints but rather potentially reinforcing partial lenses. We support the view that the two approaches can and must be joined (Johnson, 1992; Johnson & Kaplan, 1987). The strategic cost management (SCM) framework uses cost information to develop and implement strategies to acquire or sustain competitive advantage (Shank & Govindarajan, 1993). Strategic cost management is a comprehensive cost analysis framework which explicitly considers the firm’s competitive positioning in light of the value creation process all along the value chain (Shank, 1989; Shank & Govindarajan, 1992). A useful extension to this approach is to model the profit implications of causal links all along the supply chain with the A-P-L framework proposed by Epstein et al. (2000).

THEORETICAL BACKGROUND The Transaction Cost Approach Transaction cost economics (TCE) defines the rational boundaries of the firm in terms of the trade-off between the costs of internally producing resources and the costs associated with acquiring resources in an external exchange. Williamson (1975) developed this approach, drawing upon the institutional economics studies of Coase (1937). A “transaction” is defined as the exchange of a good or a service between two legally separated entities. Williamson (1985) holds that such exchanges can increase costs, relative to internal production, because of delays, lack of communication, transactor conflicts, malfunctions or other maladjustments.

40


With the purpose of avoiding such transactions costs, firms set up internal governance structures with planning, coordination and control functions. However, these structures, themselves, also cause resource consumption and thus costs. Transaction costs derive from managing supplier and buyer activities both ex ante in searching, learning, and negotiating (or safeguarding) agreements, and ex post in organizing, managing and monitoring the resulting relationship. The cost of transacting may become higher than the production cost savings because of increases in opportunistic behavior, bounded rationality, uncertainty, transactor conflicts or asset specificity. Firms then tend to prefer the integrated organization over the market transaction. The vertically integrated organization may provide numerous benefits: controllability of the actors, better attenuation of conflicts, and more effective communications (Williamson, 1986). But, a great many people believe today that markets provide better incentives and limit bureaucratic distortions more efficiently than vertical integration. The market may also better aggregate several demands, yielding economies of scale or scope (Williamson, 1985). Vertical integration, as a generalization, is in decline. For further theoretical literature on vertical integration see D’Aveni and Ravenscraft (1994), D’Aveni and Illinitch (1992), Harrigan (1983), Hennart (1988), or Quinn et al. (1990).

The Management Accounting Approach The management accounting approach to evaluating outsourcing is basically to compare the short-run differential costs and benefits of the different make or buy options (Atkinson et al., 1997; Horngren et al., 1997; Shillinglaw, 1982). This approach, in principle, takes into consideration all the “economic” aspects of the decision. In practice, however, the analysis tends to be rather narrow for at least two reasons. First is the typical assumption that the relevant time frame is the short run where management support costs and asset structures are considered fixed and thus irrelevant for the choice (Horngren et al., 1997). The second limitation is that accounting systems typically do not provide explicit cost information related to all of the activities that would emerge or disappear with the decision (Anthony & Welsch, 1977). Recently, management accounting has become more cognizant of these “management systems” costs through activity based costing (Atkinson et al., 1997; Horngren et al., 1996; Kaplan & Cooper, 1998; Shank & Govindarajan, 1993). Activity based costing (ABC) focuses cost analysis on all the activities required to produce a product or support a customer. In principle, to emphasize transaction costs issues, ABC could be linked to a full value chain framework. This would facilitate a broader conception of the costs and benefits associated with make or buy decisions. In particular, it


41

could allow managers to better estimate the “cost of ownership” (Carr & Ittner, 1992; Kaplan & Atkinsons, 1989; Kaplan & Cooper, 1998). Cost of ownership would include not only purchase price, but also those costs related to purchasing activities (ordering, receiving, incoming inspection), holding activities (storage, cost of capital, obsolescence), poor quality issues (rejections, re-receiving, scrap, rework, repackaging), or delivery failures (expediting, premium transportation, lost sales owing to late deliveries). Cost of ownership is obviously dramatically higher than purchase price alone. For example, Carr and Ittner (1992) note that Texas Instruments increased its estimate of the cost of ownership of an integrated circuit from $2.50 to $4.76 when considering poor system quality. In another survey, Ask and Laseter (1998) found that in selected commodities such as office supplies, fabrication equipment and copy machines, total cost of ownership was, respectively, 50, 100, and 200% higher than purchase price alone. Clearly, ABC is a necessary augmentation to the traditional management accounting conception of the make/buy decision, but the result is still internally focused on the firm rather than the full supply chain.

The Strategic Cost Management Perspective Strategic Cost Management is the view that cost analysis and cost management must be tackled broadly with explicit focus on the firm’s strategic positioning in terms of the overall value supply chain of which it is a part. Strategic Positioning For sustained profitability, any firm must be explicit about how it will compete. Competitive advantage in the marketplace (Porter, 1985) ultimately derives from providing better customer value for equivalent cost (differentiation) or equivalent customer value for lower cost (low cost). Occasionally, in a few market niches, a company may achieve both cost leadership and superior value simultaneously, for awhile. Examples include IBM in PCs in 1986 or Intel in integrated circuits in 1992. In general, shareholder value derives more clearly from differentiation, since the benefits of low cost are ultimately passed more often to customers than to shareholders. Value Chain Analysis The Value Chain framework sees any business as a linked and interdependent progression of value-creating activities, from basic raw material acquisition through to end-use customers. Each link in the chain is strategically relevant. Where along the chain is value generated and where destroyed? More broadly, each firm is part of an industry which also is a linked system of multiple chains with comparable issues

42


about value creation and destruction at each stage (Shank et al., 1998). In carefully analyzing its internal value chain and the industry chain of which it is a part, a firm might discover that economic profits are earned in the downstream activities, such as distribution or customer service or financing, but not upstream in basic manufacturing. For example, Shank and Govindarajan (1992) show that in the consumer liquid packaging industry, much higher returns to investment are earned downstream at the filling plant than upstream in package manufacturing. In such a case, any incremented resource allocation upstream would require very strong justification. Many businesses today are showing that value is moving downstream in the chain (Slywotzky, 1996). General Electric and Coca Cola have experienced the benefits of moving downstream (Slywotzky & Morrison, 1997). In the U.S. auto industry, a very high percentage of overall profit is in after-market services such as leasing, insurance, rentals, and repairs. New car sales and auto manufacturing show low profit (Gadiesh & Gilbert, 1998). At the deepest level, value chain analysis allows managers to better understand their activities in relation to their core competencies and to customer value. Many firms have discovered that streamlining the chain can reduce costs and enhance the value provided to customers (Hergert & Morris, 1989; Normann & Ramirez, 1994; Porter, 1985; Rackharn et al., 1996; Womack & Jones, 1996).

The Action-Profit-Linkage (APL) Model Epstein et al. (2000) argue that evaluating the impact of any strategic initiative requires assessing the profit implications all along the linked set of causal steps that make up the initiative. Too often, they say, the linkages from a decision to the related action variables to the related intervening system variables to profit are not clearly identified and quantified. They propose and illustrate a theoretical model (APL) to make such linkages explicit. The APL model is intended to promote an integrative and systemic approach to evaluating strategic choices and to propose a new performance metric – full supply chain profitability – for use in monitoring strategy implementation. We believe that APL is a very appealing extension of the SCM framework for evaluating strategic choices and we incorporate it here.

Lean Thinking The work by Womack and Jones (1996) on supply chain reconfiguration provides a context to apply SCM and APL to the decision by Levi Strauss to introduce


43

Personal Pair™ jeans. Although Levi’s has long been a dominant brand in apparel, Levi Strauss is primarily a manufacturing company, selling its products to wholesalers or retailers rather than end-use customers. As the apparel business continues to evolve, firms all along the industry value chain are continually presented with opportunities to create new ways to compete. This requires the firm to carefully position itself within the industry structure, avoiding or mitigating the power of competitors. Successful firms have the ability to differentiate an idea from an opportunity and can quickly marshal physical resources, money, and people to take advantage of windows of business opportunity. Competitive advantage is always a dynamic concept, continually shifting as firms either reposition within industries or position in such a manner that existing industry boundaries are redrawn (D’Aveni, 1993). Womack and Jones have studied this process since the mid-1980s, starting with the auto industry. They later expanded their research base in an attempt to identify “best-in-class,” across industries (Womack & Jones, 1996). They termed their point of view “lean thinking.” In their cross-industry study, they demonstrate that many companies have been able to create substantial shareholder wealth by challenging the way they implement their strategies through a five-step process. First, identify the value criterion from the customer viewpoint at a disaggregated level – a specific product to a specific customer at a specific price at a specific place and time. Second, map the value chain’s three elements: the physical stream which originates with the first entity that supplies any raw input to the system and ends with a satisfied customer, regardless of legal boundaries; the information stream that enables the physical stream; and the problem solving/decision stream which develops the logic for the physical stream. Third, focus on continuous flow and minimize disruptions such as those in a typical “push-based, batch-and-wait” system. This is accomplished by the fourth step – creating “pull,” such that the customer initiates the value stream. And fifth, strive for continuous improvement (Kaizen) by creating a “virtuous circle” where transparency allows all members to continually improve the system. In their study of world-class lean organizations, the authors cite results such as 200% labor efficiency increases, 90% reductions in throughput time, 90% reductions in inventory investment, 50% reductions in customer errors and 50% reductions in time-to-market with wider product variety and modest capital investment. An APL model is necessary to tie down the profit implications of improvements in these leading performance indicators. In the next section of the paper, we present the Levi’s Personal Pair™ business initiative as one example of a management innovation demonstrating attention to all five of these steps. In Section IV, we present for the example a full value chain profitability impact assessment using APL.

44


LEVI’S “PERSONAL PAIR”TM JEANS In 1995, women’s jeans was a $2 billion fashion category in the U.S. and growing fast. Levi’s was the market leader with more than 50% share of market, but their traditional dominant position was under heavy attack. Standard Levi’s women’s jeans, which were sold in only 51 size combinations (waist and inseam) had been the industry leading product for decades, but “fashion” was now much more important in the category. Market research showed that only 24% of women were “fully satisfied” with their purchase of standard Levi’s at a list price of about $50 per pair. “Fashion” in jeans meant more styles, more colors, and better fit. All of these combined to create a level of product line complexity that was a nightmare for manufacturing-oriented, push-based companies like Strauss which depend on independent retailers to reach consumers. Recognizing a need for better first-hand market information, in the early 1990s Strauss opened a few retail outlets, Original Levi’s stores. By 1994, Strauss operated 19 retail outlets across the country (2,000 to 3,000 square foot mall stores) to put them in closer touch with the ultimate customers. But this channel was still a tiny part of their overall $6 billion sales which were still primarily to distributors and independent retailers. Strauss was as aggressive as most apparel manufacturers and retailers in investing in process improvements and information technology to improve manufacturing and delivery cycle times and (pull-based) responsiveness to actual buying patterns. But the overall supply chain from product design to retail sales was still complex, expensive and slow. In spite of substantial improvements in recent years, including extensive use of Electronic Data Interchange (“EDI”), there was still an eight-month lag, on average, between receiving cotton fabric and selling the final pair of Levi’s jeans (see Fig. 1). The industry average lag was still well over twelve months in 1995. Custom Clothing Technology Corp. (CCTC), a small Newton, MA-based software firm, offered Levi’s a very innovative business proposal in 1994 based on an alternative value chain concept. CCTC specialized in client/server applications linking point-of-sale custom fitting software directly with single-ply fabric cutting software for apparel factories. CCTC suggested a joint venture to introduce women’s Personal Pair™ kiosks in 4 of the Original Levi’s stores. The management of CCTC had solid technology backgrounds but little retail experience. They were, however, convinced of the attractiveness of their new process, which operates as follows: (1) The Personal Pair™ kiosk is a separate booth in the retail store equipped with a touch screen PC.


Fig. 1. The Conventional Supply Chain for Levi’s Jeans. 45

46


(2) A specially-trained sales clerk uses a tape to take three measurements from the customer (waist, hips and rise) and record them on the touch screen. There are 4,224 possible combinations of these three measurements. Inseam length is not yet considered. (3) The computer flashes a code corresponding to one of 400 prototype pairs which are stocked at the kiosk. The sales clerk retrieves the prototype pair for the customer to try on. (4) With one or two tries, the customer is wearing the best available prototype. Then the sales clerk uses the tape again to finalize the exact measurements for the customer (4,224 possible combinations) and to note the inseam length desired. (5) The sales clerk enters the 4 final measurements on the touch screen and records the order. The system was available only for the Levi’s 512 style, but 5 color choices were offered in both tapered and boot-cut legs. (6) The customer pays for the jeans and pays a $5 Fed Ex delivery charge (per pair). Delivery is promised in not more than three weeks. (7) Each completed customer order is transmitted by modem from the kiosk to CCTC where it is logged and retransmitted daily to a Levi’s factory in Tennessee. (8) At the factory, each pair is individually cut, hand sewn, inspected and packed for shipment. Each garment includes a sewn-in bar code unique to the customer for easy re-ordering at the store where the bar code is on file in the kiosk. (9) There is a money-back guarantee of full satisfaction on every order. Using Lean Principles to Analyze the CCTC Offer As is immediately obvious in Fig. 1, the Original Levi’s store system is the antithesis of lean! Due to uncertainty in demand forecasting and inconsistent supply chain lead times, large investments in inventories are necessary (raw, WIP and finished). This, in turn, necessitates investments in logistics support assets such as warehouses, IT systems, and vehicles. For the business that CCTC targeted, the five elements of lean thinking can be summarized as follows: (1) Value. Although Levi Strauss was a very profitable and very large firm, only 24% of women were satisfied with the fit of their new jeans. This “opportunity” in women’s jeans that Levi’s was missing illustrates the need to apply the lean thinking methodology on a disaggregated level. (2) Value stream. It is unclear how concerned Levi’s was about the cumbersome value stream for this particular product. Their approach was typical in the


47

industry, and Levi’s corporate ROE averaged a very robust 38% for the threeyear period, 1993 to 1995. (3) Continuous flow. Given the eight month denim-to-sale cycle, with frequent inventory “stops,” there is clearly very little “flow.” (4) Pull. Likewise, this is the classic push system. The customer initiates nothing. All production and distribution activity is driven by sales forecasts and production lot-sizing. (5) Kaizen. Again, it seems that Levi’s satisfaction with a high overall ROE may have led them to miss the lack of transparency in the women’s jeans chain which blocks the opportunity for continuous improvement.

PERSONAL PAIRTM IMPACT ON WEALTH CREATION Although Levi’s does not publish financial results for women’s jeans sold through the Original Levi’s channel, we were still able to analyze this channel with some degree of confidence using field site visits, industry averages, benchmark company comparisons, and interviews with industry participants.

Fig. 2. The Levi Strauss Aggregate Financial Footprint (1993–1995).

48


Publicly available information allows us to construct the overall financial footprint for Levi Strauss shown in Fig. 2. One key element of our analysis here is converting this aggregate level financial information for Levi Strauss to a disaggregated unit – one pair of Personal Pair™ jeans. This is the basis for building plausible profitability estimates for one pair of jeans sold in different distribution channels.

The Wholesale Distribution Channel A breakdown of the profitability impact along each stage of the chain for the normal wholesale channel, using the APL model, is shown in Fig. 3. As noted earlier, the retail list price for a pair of jeans is approximately $50. Assuming a typical retail gross margin of 30%, the Levi’s wholesale price is close to $35. In addition, historically, approximately 1/3 of Levi’s jeans are sold at markdowns averaging approximately 30% off list. This equates to average price allowances of

Fig. 3. An APL Formulation of the Profitability of Women’s Jeans to Levi Strauss as a Wholesale Supplier.


49

about $5 per pair (1/3 × 30% × $50). About 60% of this, or $3 per pair, is made good by Levi’s in some type of co-op agreement. The result is a net sales price for Levi’s of $32 ($35 – $3). The footprint gross margin in Fig. 2 for Levi’s as the manufacturer are about 40%. This implies that cost of goods sold for one pair of jeans is about $19 (60% × $32). From research, we know that denim costs about $5 per pair and conversion another $5. This leaves approximately $9 for distribution logistics. Overall S,G&A is 25% of sales per Fig. 2, which would be $8 per pair, based on net sales of $32. We estimate S,G&A to be moderately higher ($9 per pair) for women’s jeans because of the more complex supply chain for a fashion item. Pre-tax profit per pair is thus $4 ($32 – $19 – $9). Note that a significant part of cost is directly due to the “push” system in place ($3 in markdowns, the additional $1 in S,G&A, and $9 in distribution costs). The investment per pair can also be estimated from the financial footprint. Using the average inventory turnover of 4.73, we estimate the inventory investment to be about $4 ($19/4.73). Accounts payable (27 days) for this channel rounds to $1, yielding a net inventory requirement of $3 for every pair sold. The collection period for women’s jeans should not be that much different from the overall Levi’s collection period of 51 days, which translates to $4 in accounts receivable for each pair. In a like manner, the 5.33 fixed asset turn gives us a total of $6 per pair ($32/5.33) in property assets. Our field research indicates that this plant investment for the normal channel is mostly in the factory, rather than distribution. In total, we estimate that, for this channel, every pair sold requires capital of approximately $13 ($4 – $1 + $4 + $6). With the above pre-tax operating profit of $4, this is an overall very healthy ROIC of about 31%. The figure is marginally less than the corporate average because of extra downstream costs for a women’s fashion item.

The Own Store Channel We next make the adjustments necessary to convert this analysis to one pair of women’s jeans sold through Original Levi’s stores. The profitability impact at each stage along this chain is shown in the APL model in Fig. 4. Basically, the financial consequences of adding a retail outlet can be derived from databases for retail clothing companies. The only difficult element to estimate is the in-store investment. A visit to a local Levi’s outlet revealed the following information: Building investment – 3,000 square feet leased for about $240,000 per year. The lease rate of $80 per foot per year is typical for high-end malls. We capitalized

50


Fig. 4. An APL Formulation of the Profitability of Women’s Jeans Sold Through the Original Levi’s Channel.

the lease cost at 10% to estimate an investment of approximately $2,400,000 in the building space. Volume – The average Original Levi’s store has 1,000 SKUs and 20,000 pairs of pants on hand, on average. This inventory turns approximately 6 times per year, yielding an annual store volume of approximately 120,000 pairs. Store investment per pair sold – $2,400,000/120,000 pairs = $20 per pair. This calculation assumes Levi’s to be the implicit owner of this space. Whether Levi Strauss should actually own or rent retail space is beyond the scope of this analysis.


51

Comparing the normal wholesale channel with the owned retail channel, profitability (ROIC) for women’s jeans falls by about 50%, from 31 to 16%. Levi Strauss is paying a high price to gain customer intimacy in this segment!

The Personal Pair™ Channel Our next step is to estimate how Personal Pair™ changes the profitability analysis. This requires that we first understand the CCTC value chain, its impact on the aggregate financial footprint, and its impact on each pair of jeans. Based on the CCTC proposal outlined earlier, a reasonable estimate of the new value chain is as shown below.

As is obvious in comparing the original Levi’s and Personal Pair™ value chains, the CCTC system is indeed much more “lean.” It adds “fit” value for the customer. It has a well-defined value stream, including not only the physical and information flows, but also the decision-making. The flow is interrupted only at the transportation nodes and is initiated by “pull” from a customer order. Although perfection can never be achieved, the areas for kaizen seem obvious given the transparency and simplicity of the system. All five lean thinking criteria have been markedly improved. But, how is profitability affected? Specific financial information with respect to this system is difficult to estimate because of the short history for CCTC and the vastly different structure of the chain. However, our research enabled us to make the estimates summarized in Fig. 5. Again, the APL framework is used to show profit impact all along the causal chains. If, indeed, CCTC can deliver what it has promised, the results are dramatic. More customer satisfaction implies an opportunity for higher selling prices. Levi’s priced each Personal Pair™ $15 higher initially, with very little customer resistance. One year later, the premium was cut to $10 based on the estimated price elasticity. Custom fit also eliminates mark-downs, driving up the net price. Distribution costs are transferred to Fed Ex for which the customer pays separately. Operating costs per pair in the store are cut by half, assuming half the orders are

52


Fig. 5. An APL Formulation of the Profitability of Personal Pair™ Women’s Jeans. Note: The normal $8 for Strauss plus the normal $10 for the store (increase in personal selling offset by decrease in space costs) but divided by 2 for 50% repeat orders by mail, plus $3 for CCTC ($8 + $ 10/2 + $3 = $16).

repeat business with zero store contact. Selling costs for the first pair would increase given the time spent on measuring and fitting each customer. But, this happens only once (until body dimensions change.). The inventory and retail store investment decrease substantially with only an offset of CCTC investment in computers and software. Overall, the CCTC opportunity is very enticing with very high ROIC. Given that this model did not have an established record in retail merchandising, a test phase was probably the wise choice for Levi’s.


53

ASSESSING THIS CHANGE IN THE SUPPLY CHAIN The Personal Pair™ kiosks were very popular almost immediately. The experiment was extended to seven stores by the Summer of 1995. About one half the sales were repeat orders which greatly simplify the point of sale process. In October of 1996, Heidi LeBaron-Leupp, marketing director for the Personal Pair™ program, declared it a “phenomenal success.” For the styles affected, unit sales were up 49%. Two years into the program (Fall, 1994 to Fall, 1996), the company’s experience was that “Personal Pair” resulted in no change in raw material or conversion cost (per pair), but virtually eliminated distribution costs and distribution investment. Non-material manufacturing and distribution costs are cut by an amazing 47% (from $15 to $8). Approximately eight months of inventory has been almost completely eliminated (except for raw material). Warehouses, intermediate handling, insurance, shrinkage, transportation logistics and vehicle depreciation and maintenance are now only memories of the past with this channel. Still, the CCTC stage of the chain does add some infrastructure costs and investment. Although the kiosks are more labor intensive for the first pair sold, the in-store cost is substantially less per pair sold after allowing for repeat orders at zero in-store cost. The net result, given the increase in price to $60, is a 467% increase in pre-tax profit (from $6 to $34 per pair) which is truly remarkable. Equally as remarkable is the impact on asset investment. Inventory of $12 per pair sold (reflecting the eight months of average “pipeline”) is reduced to $1 (reflecting only raw material requirements). Accounts receivable is now negative, since these jeans are prepaid for the period between sale and delivery. Non-manufacturing PP&E is reduced by 40%, from $22 per pair in Fig. 4 to $13 in Fig. 5. When the impact on the numerator and the denominator of ROIC are combined, the promises of lean thinking are fulfilled. A kiosk yields a greater than ten fold increase in profitability over an Original Levi’s store (from 16% ROIC to 200%), while still accomplishing the strategic purpose of the store – a research lab for putting Levi’s in closer touch with the end-use customer. By 1999, there were 60 Personal Pair™ kiosks across the U.S. and Canada, one in each of the Original Levi’s stores. The program in 1997 was responsible for 25% of all women’s jeans sales in the 30 U.S. company-owned stores. Delivery was averaging only 3–4 days. The “promise date” was cut from three weeks to two. All orders are shipped via Fed Ex which picks up daily at the factory located near the Fed Ex hub in Memphis. Levi Strauss acquired CCTC for more than $2 million in October of 1995. The acquisition insured that CCTC would continue to work with Levi’s, and only with Levi’s, to expand and improve this niche segment which

54


is based on computer-based custom fit, custom manufacturing, and direct distribution. In the Fall of 1996, Levi’s introduced Personal Pair™ in two stores in London as the first overseas locations. The price was £19 higher than the regular £46 price. Manufacturing was still in the U.S. with distribution via Federal Express. During 1997, the Personal Pair™ cutting and sewing operations were both moved into the main Levi’s production plant and are now side-by-side with the still-traditional bulk manufacturing business. Levi’s has made a decision not to roll out the Personal Pair™ concept to its independent distributors. Evaluating the rational for that choice is beyond the scope of this paper. We believe that Levi’s roll out of the Personal Pair™ concept demonstrates successful use of the lean thinking management principles. Women’s jeans was a $2 billion industry with only 24% customer satisfaction. The principles of lean thinking helped Levi’s seize this opportunity. CCTC, not burdened by the “push” focus of the existing Levi’s value chain, was able to design a system much leaner and “pull”-based, initiated by customer response. CCTC was able to create a process flow that reflected value from the customer’s perspective. But lean thinking theory, alone, does not directly address profitability. For that, an APL model is needed that draws on SCM concepts to encompass the entire supply chain. Conceptually, we argue that an SCM-based APL model is the way to demonstrate profitability impact. The example here provides one preliminary validation of the SCM/APL theory.

CONCLUSION Clearly, value chain reconfiguration is a central topic for many firms today in many industries. Although value chain analysis can be framed in accounting terms as the classic make/buy problem, the traditional accounting literature with its focus on relevant cost analysis (RCA) does not provide much help with the strategic aspects of the dilemma. The management literature is rich in discussing those strategic issues, but is very thin on the related financial analysis. This literature’s focus on transaction cost economics (TCE) is very appealing conceptually, but not of much pragmatic help. In this paper, we propose a new theoretical approach which extends conventional RCA and TCE analysis to a full cost ROIC basis spanning the entire value chain. This approach disaggregates the level of analysis from the firm as a whole to an individual product sold to a particular customer segment. It couples the SCM framework with an APL model to explicitly address the profit impact of the managerial actions at each stage along the supply chain. We apply this new


55

theoretical model to one particular decision context to demonstrate its practical, as well as conceptual, usefulness. Although our financial comparison across the wholesale channel, the Original Levi’s channel and the Personal Pair™ kiosk is based upon estimates, limited public information and best judgment, we feel that it is directionally correct and approximately accurate. We believe the SCM/APL framework is a very useful way to study changes in value delivery systems, particularly when the unit of analysis can be framed in end-use customer terms. This paper is intended as a first step in establishing the new approach. We believe the next steps are further studies to demonstrate its applicability in other supply chain configuration contexts. We encourage such further research.

REFERENCES Anthony, R. N., & Welsch, G. (1977). Foundamentals of management accounting. Irwin. Ask, J. A., & Laseter, T. M. (1998), Cost modeling: A foundation purchasing skill. Strategy and Business, 10. Booz Allen & Hamilton. Atkinson, A. A., Banker, R. D., Kaplan, R. S., & Young, S. M. (1997). Management accounting. Englewood Cliffs, NJ: Prentice-Hall. Beamish, P. W., & Killing, P. J. (Eds) (1997). Cooperative strategies. European perspectives. San Francisco: New Lexington Press. Carr, L. P., & Ittner, C. D. (1992). Measuring the cost of ownership. Journal of Cost Management, Fall. Clark, K. B., & Fujimoto, T. (1991). Product development performance. Boston, MA: Harvard Business School Press. Coase, R. H. (1937). The nature of the firm. Economica, 4. D’Aveni, R. A. (1993). Hypercompetition. New York, NY: Free Press. D’Aveni, R. A., & Ilinitch, A. V. (1992). Complex patterns of vertical integration in the forest products industry: Systematic and bankruptcy risk. Academy of Management Journal, 35. D’Aveni, R. A., & Ravenscraft, D. J. (1994). Economies of integration vs. bureaucracy costs: Does vertical integration improve performance? Academy of Management Journal, 37. Epstein, M. J., Kumar, P., & Westbrook, R. A. (2000). The drivers of customer and corporate profitability: Modeling, measuring and managing the causal relationships. Advances in Management Accounting, 9. Gadiesh, O., & Gilbert, J. L. (1998). Profit pools: A fresh look at strategy. Harvard Business Review, May–June. Gulati, R. (1995). Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Academy of Management Journal, 38. Hagedoorn, J. (1993). Understanding the rationale of strategic technology partnering: Interorganizational modes of cooperation and sectoral differences. Strategic Management Journal, 14(5). Harrigan, K. R. (1983). Strategies for vertical integration. Lexington, MA: Heath & Lexington Books. Hennart, J. F. (1988). Upstream vertical integration in the aluminum and tin industry. Journal of Economic Behavior and Organization, 9.

56


Hergert, M., & Morris, D. (1989). Accounting data for value, chain analysis. Strategic Management Journal, 10. Horngren, C. T., Foster, G., & Datar, S. (1997). Cost accounting: A managerial emphasis. Englewood Cliffs, NJ: Prentice-Hall. Johnson, T. H. (1992). Relevance regained. New York, NY: Free Press. Johnson, T. H., & Kaplan, R. S. (1987). Relevance lost. The rise and fall of management accounting. Cambridge, MA: Harvard Business School Press. Kaplan, R. S., & Atkinsons, A. A. (1989). Advanced management accounting. Englewood Cliffs, NJ: Prentice-Hall. Kaplan, R. S., & Cooper, R. (1998). Cost & effect. Using integrated cost systems to drive profitability and performance. Boston, MA: Harvard Business School Press. Khanna, T. (1998). The scope of alliances. Organization Science, May–June, Special Issue. Kogut, B. (1988). Joint ventures: Theoretical and empirical perspectives. Strategic Management Journal, 9(4). Kogut, B. (1991). Joint ventures and the option to expand and acquire. Management Science, 37. Lorenzoni, G., & Baden Fuller, C. (1995). Creating a Strategic center to manage a web of partners. California Management Review, 37. Mowery, D. C. (Ed.) (1988). International collaborative ventures in U.S. manufacturing. Cambridge, MA: Ballinger. Nohria, N., & Eccles, R. G. (Eds) (1992). Networks and organizations: Structure, form, and action. Boston, MA: Harvard Business School Press. Normann, R., & Ramirez, R. (1994). Design interactive strategy. From value chain to value constellation. Chichester, UK: Wiley. Porter, M. E. (1985). Competitive advantage. New York, NY: Free Press. Prahalad, C. K., & Hamel, G. (1994). Competing for the future. Boston, MA: Harvard Business School Press. Quinn, J. B. (1992). Intelligent enterprise. A knowledge and service based paradigm for industry. New York, NY: Free Press. Quinn, J. B., Doorley, T. L., & Paquette, P. C. (1990). Technologies in services: Rethinking strategic focus. Sloan Management Review, 3(1). Rackharn, N., Friedman, L., & Ruff, R. (1996). Getting partnering right. How market leaders creating long term competitive advantage. New York, NY: McGraw-Hill. Shank, J. K., & Govindarajan, V. (1992). Strategic cost analysis: The value chain perspective. Journal of Management Accounting Research. Shank, J. K., & Govindarajan, V. (1993). Strategic cost management: The new tool for competitive advantage. New York, NY: Free Press. Shank, J. K., Spiegel, E. A., & Escher, A. (1998). Strategic value analysis for competitive advantage. An illustration from the petroleum industry. Strategy and Business, 10. Booz Allen & Hamilton. Shillinglaw, G. (1982). Managerial cost accounting. Homewood, IL: Irwin. Slywotzky, A. J. (1996). Value migration. How to think several moves ahead of the competition. Boston, MA: Harvard Business School Press. Slywotzky, A. J., & Morrison, D. J. (1997). The profit zone. How strategic business design will lead you to tomorrow’s profits. New York, NY: Times Business. Westney, D. E. (1988). Domestic and foreign learning curves in managing international cooperative strategies. In: F. J. Contractor & P. Lorange (Eds), Cooperative Strategies in International Business. Lexington, MA: Lexington Books.


57

Williamson, O. E. (1975). Markets and hierarchies. New York, NY: Free Press. Williamson, O. E. (1985). The economic institutions of capitalism. New York, NY: Free Press. Williamson, O. E. (1986). Economic organization. Brighton: Wheatsheaf Books. Womack, J. P., & Jones, D. T. (1996). Lean thinking. New York, NY: Simon & Schuster.

THE MEASUREMENT GAP IN PAYING FOR PERFORMANCE: ACTUAL AND PREFERRED MEASURES Jeffrey F. Shields and Lourdes Ferreira White ABSTRACT What is measured gets managed – especially if rewards depend on it. For this reason many companies (over 70% in this survey) have upgraded their performance measurement systems so as to include a mix of financial and non-financial metrics. This study compares how companies currently measure performance for compensation purposes with how their managers think performance should be measured. We find significant measurement gaps between actual and preferred measures, and we find that larger measurement gaps are related to lower overall performance. The choice of performance measures for compensation purposes is also related to the attitudes of managers towards manipulation of reported results.

INTRODUCTION Performance measures are powerful means of conveying which aspects of performance are important to a company and which areas a manager needs to focus on to be evaluated as a top performer. Managers direct their attention to those measures that most strongly influence their compensation. Recognizing the motivational effects of performance measures, many companies have implemented major changes to improve their performance measurement systems. According to the Advances in Management Accounting Advances in Management Accounting, Volume 12, 59–83 © 2004 Published by Elsevier Ltd. ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12003-0

59

60

JEFFREY F. SHIELDS AND LOURDES FERREIRA WHITE

2001 survey of the Cost Management Group of the Institute of Management Accountants, 80% of the respondents reported that their organizations invested in significant changes in their performance measures in the last three years; 33% of the respondents described those changes as a “major overhaul” or “new performance measurement system” (Frigo, 2001). For many companies revising their performance measurement systems, there has been an increase in the relative weight placed on non-financial, strategic measures in incentive compensation contracts. But just as the percentage of companies that link pay to financial and non-financial performance consistently increases, there is little agreement on how to measure performance effectively. Even within organizations, managers in different hierarchical levels often disagree as to how performance should be measured for compensation purposes. These issues motivated us to conduct a survey, among managers, concerning actual company practices and their managers’ preferences regarding performance measures. While other surveys have provided general evidence regarding the performance measures that companies are currently using, we designed this survey to address four research questions. First, which performance measures are currently being used for the specific purpose of incentive compensation? Second, which measures do managers think should be used to calculate their own compensation? Third, is the choice of performance measures related to overall performance (especially in cases where actual measurement practices deviate from what managers would prefer)? Finally, what is the connection between the choice of performance measures and the likelihood that manages will manipulate the results?

LITERATURE REVIEW Recent surveys on performance measurement have documented managers’ widespread dissatisfaction with current performance measures. For example, the Institute of Management Accountants’ annual surveys on performance measurement practices have consistently shown, since the 1990s, that more than half of the respondents rate their companies’ performance measurement systems as poor or, at best, adequate (see summary of the recent survey in Frigo, 2002). The IMA survey in 2001 indicated that non-financial metrics related to customers, internal processes and learning and growth (three perspectives proposed in the balanced scorecard framework developed by Kaplan & Norton, 1992, 1996, 2001) received lower ratings than financial metrics. A large-scale study conducted by the American Institute of Certified Public Accountants showed that only 35% of the respondents regarded their company’s

The Measurement Gap in Paying for Performance

61

performance measurement system as effective (AICPA & Maisel, 2001). Consequently, about a third of the respondents stated that their organizations have implemented changes in their performance measurement system in the last two years. Both the IMA and AICPA surveys revealed that managers believed that more extensive use of non-financial measures would improve their organizations’ ability to assess and enhance performance in strategically critical areas such as customer performance, product innovation and employee capabilities. In the 2001 IMA survey, 87% of the respondents argued that non-financial metrics should be used more extensively in their businesses. A common theme in these surveys is the assumption that non-financial performance measures are more future-oriented than traditional financial measures, so they should assist managers in making decisions that will benefit their organizations in the long run. Non-financial measures are often deemed to be better leading indicators or drivers of future performance, while financial measures serve the purpose of providing lagging information about past performance (Epstein et al., 2000; Ittner & Larcker, 1998a; Kaplan & Norton, 1996, 2001). In addition, non-financial metrics tend to capture input or process measures, as opposed to outcomes (Simons, 2000). A balanced use of financial and non-financial measures, carefully linked to strategic objectives and integrated through cause-and-effect relationships, should facilitate strategy implementation and motivate superior performance, as argued by the increasing number of supporters of the balanced scorecard approach originally developed by Kaplan and Norton (1992). Yet, the literature on the linkages between the choice of performance measures and actual overall performance is only beginning to produce affirmative results (e.g. Banker et al., 2000b; Malina & Selto, 2001). For example, recent studies have shown that sales performance is positively driven by customer satisfaction (Ittner & Larcker, 1998b), on-time delivery (Nagar & Rajan, 2001) and employee satisfaction (Banker et al., 2000a). Some causal relationships among sets of performance measures are beginning to surface: for example, Rucci et al. (1998) provide evidence that an increase in employee attitude drives an increase in customer satisfaction which, in turn, drives an increase in revenue growth. So far, however, despite many calls for changes in performance measurement systems, there is little systematic evidence on how financial and non-financial performance metrics have been integrated in practice in the design of incentive compensation plans, and even less evidence on their impact on organizational performance. This study addresses this issue, and proposes a new potentially fruitful area of research: we investigate the gap or lack of fit between current performance measurement systems and managers’ preferred performance measures, and we explore whether this fit is related to managerial performance and attitudes towards manipulating results to achieve desired performance targets.

62


THEORETICAL DEVELOPMENT The person-organization fit literature is well-established in the human resource management field (e.g. Chatman, 1989; O’Reilly et al., 1991; Posner, 1992). Fit is defined in that literature as “the congruence between the norms and values of organizations and the values of persons” (Chatman, 1989). Going beyond simply measuring fit, organizational researchers have studied how individual values interact with situations (e.g. incentives) to affect the attitudes and behaviors of people in the workplace (O’Reilly et al., 1991). Person-organization fit has been shown to influence vocational and job choices, job satisfaction, work adjustment, organizational commitment and climate, and employee turnover (Chatman, 1989, 1991; O’Reilly et al., 1991). The importance of person-organization fit for the ethical work climate has also received much attention by researchers in business ethics (e.g. Sims & Kroeck, 1994). Yet, in the accounting literature, person-organization fit has received only limited attention. Management accounting researchers have mainly focused on fit from a macro perspective, using either contingency theory (e.g. Merchant, 1984) or national culture (Chow et al., 1996, 1999) to examine the effectiveness of actual control system characteristics. Those studies have contributed additional evidence to suggest that lack of fit can potentially increase the costs of attracting and retaining employees, and may induce behaviors contrary to the firm’s interests. In this study we attempt to extend prior research on management control system fit in three main ways. First, we take a more focused perspective and study fit at the level of interactions between individual managers and the specific performance measurement systems that govern their work. Second, instead of assuming what management preferences might be, we actually asked managers to state their preferences for seventeen performance metrics commonly used in performance-based incentive plans. This allowed us to directly quantify the measurement gap or lack of fit. Third, we address a common criticism of previous performance measurement research by investigating the empirical relationships between selection of performance measures and actual performance. We expect that existing performance measurement systems vary in how closely they match managers’ preferences for performance metrics. When organizations make decisions about which control system practices they will adopt, their choices reflect primarily the values and preferences of those in charge of designing control systems. There is little guarantee that those choices will be similarly valued by all managers subject to such control systems, as demonstrated in previous studies of the person-organization fit in the performance measurement and incentive compensation areas (Bento & Ferreira, 1992; Bento & White, 1998). While selection, socialization and turnover are all powerful mechanisms to improve


63

person-organization fit, significant disagreements between individual managers’ preferences and actual performance measurement system characteristics may still prevail and are thus worth investigating. We also expect, based on person-organization interactional research (Chatman, 1989) that organizational performance and gaming attitudes will depend not only on situational variables (the characteristics of extant performance measurement systems implemented by companies) but also on personal preferences (based on the values held by individuals working in those organizations). As performance measurement systems more closely match managers’ preferences, a series of positive effects (increased motivation, reduced job-related tension, and enhanced organizational commitment) will take place, leading to improved performance. By contrast, when a conflict is established by the lack of fit between managers’ preferences and existing performance measurement systems, there will be an increased chance that managers will respond by engaging in various gaming behaviors, in order to reach short-term performance targets without pursuing the actions that the performance metrics were designed to induce. These expectations are summarized by the following research proposition: The magnitude of fit between a manager’s preference for performance measures and his or her company’s actual use of such measures for incentive pay purposes is significantly associated with performance and the ethical climate in the organization.

While the purpose of this exploratory study is not to provide a direct and exhaustive test of this proposition, the results from this survey do provide initial empirical evidence on the relationships among performance metrics, fit, and organizational performance.

METHOD Sample We obtained support for this study from a sample of 100 managers in the midAtlantic area. We conducted initial interviews with the managers to explain the nature of the project, to ensure participation, and to verify that their positions in their firms included budget responsibility. The sample covered a wide range of industries (30% in manufacturing, 70% in the service sector). To ensure confidentiality, we asked each participating manager to complete the survey questionnaire and mail it to us anonymously. We received a total of 64 completed questionnaires, yielding a 64% response rate. We attribute this unusually high response rate (relative to other survey studies) to

64


our efforts to ensure participation prior to distribution of the questionnaires. The managers in our survey are responsible for units with average budgeted revenues of $40,000,000 and have been in their position for an average of five years. Measures Performance Measures The survey questionnaire included nine financial measures (sales growth, volume of orders, shipments, expense control, profitability, receivables management, inventory management, efficiency gains, cash flow) and eight non-financial measures (quality of products or services, customer satisfaction, new product introduction, on-time delivery, accomplishment of strategic objectives, market share, and employee satisfaction) selected from the literature on performance measurement. Gap Measure We asked the managers to rate, for each financial and non-financial performance measure, the extent to which it actually affected his or her compensation (actual measures). We also asked the managers to rate, for each measure, the extent to which he or she would like that measure to affect his or her compensation (preferred measures). Both types of ratings used a Likert-type scale ranging from “1” (very little) to “7” (very much). Consistent with the literature on person-organization fit (e.g. Chow et al., 1999), we computed the “measurement gap” or lack of fit by subtracting actual use from preferred use of performance measures for compensation purposes. A positive measurement gap indicates that managers would prefer more emphasis on those measures. A negative gap indicates that managers would prefer less emphasis on those measures. Performance We measured performance using a nine-item instrument that has been widely applied by management accounting researchers (e.g. Brownell & Hirst, 1986; Kren, 1992; Nouri et al., 1995). This instrument, while relatively subjective, was employed because more objective performance indicators are usually not available for lower-level responsibility centers. Matching self-evaluations with supervisor evaluations was not possible given that the scope of this study (which included earnings management issues) required that the various responsibility center managers could rely on absolute survey confidentiality within their firms. Gaming We described various scenarios in which managers decide to improve reported profits in the short-term, and asked the respondents to rate the likelihood that they


65

would make such decisions. We adapted the scenarios from previous studies on earnings management (Bruns & Merchant, 1989, 1990). For example, the scenarios described decisions directed at either increasing sales (e.g. shipping earlier than scheduled in order to meet a budget target), or reducing expenses (e.g. postponing discretionary expenses to another budgeting period).

PERFORMANCE MEASURES FOR COMPENSATION PURPOSES Table 1 presents the distributions of all performance measures (actual and preferred). All 17 measures included in the survey were used in the incentive plans of at least some of the managers surveyed. Based on the response scales ranging from 1 (very little) to 7 (very much), we calculated the percentage of respondents who reported high use or high preference for each performance measure as the cumulative percentage of respondents who rated actual usage or preference for that metric as 5 or higher. This convention is used in Fig. 1 through 4. Actual Measures Figures 1 and 2 show the financial and non-financial measures the respondents reported as the most commonly used for compensation purposes. At least half of the managers reported that their compensation is influenced largely by their performance measured by sales, receivables management, volume of orders, and profitability. Over 70% of the respondents reported that non-financial measures play a major role in determining their compensation. The three most widely used non-financial measures are new product introduction, customer satisfaction, and achievement of strategic objectives. This result is similar to the findings of the AICPA survey (AICPA & Maisel, 2001) and the Conference Board’s survey (Gates, 1999). Preferred Measures When asked what performance factors they would like to see being used to affect their own compensation, over half of the managers reported preferences for efficiency gains and profitability for financial measures (Fig. 3), while employee satisfaction, achievement of strategic objectives and quality are the three most preferred non-financial measures (Fig. 4). Their preferences may be explained by several different factors. Managers may feel that these are the aspects of

66

Table 1. Distributions of Actual and Preferred Performance Measures. Measures

Actual Measures

Preferred Measures

Theoretical Range

Actual Range Range

Mean

Standard Deviation

Actual Range

Mean

Standard Deviation

1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7

1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7

6.00 4.48 3.80 3.33 4.09 5.23 2.64 3.56 3.33 3.17 4.95 5.19 3.38 4.64 4.92 2.45 3.11

1.44 1.96 2.10 2.15 1.78 1.66 1.87 1.80 1.99 2.07 1.86 1.82 1.96 1.85 1.49 1.70 1.90

1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7 1–7

3.15 3.32 3.14 3.66 5.00 2.14 2.59 4.92 2.84 4.06 4.06 2.84 3.63 3.83 4.20 3.25 4.70

1.89 2.01 2.02 1.84 1.75 1.53 1.84 1.56 1.75 2.00 2.02 2.02 2.04 2.02 2.01 2.04 1.60

Note: Extent to which the performance measure affects the manager’s compensation (1 = very little, 7 = very much).


Sales growth Volume of orders Shipments Expense control Profitability Receivables management Inventory management Efficiency gains Cash flow Quality of products or services Customer satisfaction New product introduction On-time delivery Accomplishment of project milestones Achievement of strategic objectives Market share Employee satisfaction

Actual and Preferred


Fig. 1. Actual Use of Financial Performance Measures for Compensation Purposes. 67

68


Fig. 2. Actual Use of Non-Financial Performance Measures for Compensation Purposes.

performance that they can most directly control or they may consider that these factors most closely reflect the decisions that they make on a day-to-day basis. Preferences for particular performance measures may also be driven by the managers’ ability in those areas, so that managers seek a performance measurement system that closely matches their skill set. Interestingly enough, two of the most commonly used measures are reported among the least preferred: receivables management and new product introduction.

Measurement Gap We performed some additional analysis to calculate the level of fit between actual practices and managerial preferences with respect to performance measures affecting compensation. As explained in the method section above, the gap measure was obtained by subtracting actual use from preferred use of performance

Fig. 3. Preferred Use of Finanacial Performance Measures for Compensation Purposes.


69

Fig. 4. Preferred Use of Non-Financial Performance Measures for Compensation Purposes.

measures for compensation purposes. Figure 5 shows the measurement gap for financial measures, while the gap for non-financial measures appears in Fig. 6. Considering both financial and non-financial performance measures combined, the measurement gap is smallest for inventory management, expense controls, and on-time delivery; it is largest for receivables management, new product introduction, and sales. On average, there is greater disagreement (in absolute terms) with the use of non-financial than financial measurements when companies evaluate how much managers will get paid. The relative signs of the gap measure provide additional information: the respondents reported that they would rather have less emphasis on receivables management, new product introduction, and sales for compensation purposes. One possible interpretation is that managers would prefer less emphasis on these measures because they do not believe that these measures adequately capture their performance. Conversely, the managers responded that they would rather have more emphasis on employee satisfaction and efficiency gains.

Balancing Financial and Non-financial Measures In this survey we found that most companies are, in fact, trying to balance out the use of financial and non-financial performance measures. As the correlation

70


Fig. 5. Financial Performance Measurement Gap: Preferred Use Minus Actual Use.

matrix of Table 2 shows, there are some significant correlations among financial and non-financial metrics. For example, the respondents who reported that their compensation is largely influenced by sales also reported that their pay is contingent on achieving strategic objectives. Similarly, managers who are paid based on how well they perform with respect to efficiency gains also tend to be the ones who reported that employee satisfaction plays a major role in determining their rewards. Managers in those situations would have incentives to cut costs, but not in ways that would hurt employee morale so much that the short-term cost savings would end up hurting long-term profits and growth. Likewise, we found a strong correlation between the emphasis on cash flows and quality for compensation purposes. This combination encourages managers to invest in quality without losing sight of the need to


71

Fig. 6. Non-Financial Performance Measurement Gap: Preferred Use Minus Actual Use.

generate cash flows. Those results are consistent with existing evidence that companies are striving to get a more balanced picture of performance in order to avoid distortions caused by a concern directed exclusively toward short-term, historicbased financial measures (see evidence summarized by Malina & Selto, 2001). This balance depends on the availability of non-financial indicators that can deliver accurate, relevant and timely information on how well managers are doing in those areas. But in many companies those measures are not readily available. Besides measurement difficulties, companies also face serious constraints with respect to what their information systems can deliver. According to a survey by The Conference Board and the international consulting firm A. T. Kearney, Inc., 57% of the respondents said that their companies’ information technology limited their ability to implement the necessary changes in their current performance

72

Table 2. Correlation Matrix for Actual Financial and Non-Financial Performance Measures Used. Measures

7 8 9 10 11 12 13 14

15 16 17 ∗

Sales growth Volume of orders Shipments Expense control Profitability Receivables management Inventory management Efficiency gains Cash flow Quality of products or services Customer satisfaction New product introduction On-Time delivery Accomplishment of project milestone Achievement of strategic objective Market share Employee satisfaction

p < 0.10. p < 0.05. ∗∗∗ p < 0.01. ∗∗

2

3

4

5

6

7

8

0.37*** 0.20 0.07 0.10 0.28**

0.58*** 0.23* 0.25** 0.55***

0.55*** 0.15 0.49*** 0.27** 0.30**

0.44***

0.07

0.32***

0.36*** 0.44***

0.26** 0.45***

−0.03 0.05 0.28** 0.59*** ** 0.28 0.60***

0.05 0.02 0.30** 0.17 ** 0.27 0.32***

0.02 0.55*** −0.03 0.59*** 0.02

0.19 −0.03 0.08

0.06 0.06 0.09

9

10

11

12

13

14

15

16

0.61***

0.35*** 0.26**

0.44*** 0.38***

0.28** 0.24**

0.16

0.22*

0.33***

0.23*

0.30** 0.23*

0.40*** 0.30**

0.26** 0.27**

0.26**

0.18

0.11

0.18

0.34*** 0.57*** 0.25** 0.13

0.40*** 0.15 0.26** 0.38***

0.19 0.20

0.26** 0.13

0.10 0.31**

0.19 −0.01 0.29** 0.50***

0.75*** 0.28**

0.24*

0.20

0.18

0.14

−0.06

0.11

−0.04

0.10

0.39*** 0.40*** 0.37*** 0.10

0.22* 0.05

0.22* 0.05

0.23* −0.08

0.18 0.00

0.16 −0.00

0.51*** 0.03 0.65*** −0.08

0.18 −0.06

0.53***

0.07 0.37*** 0.36*** 0.43*** 0.44*** 0.44*** 0.29**

0.37*** 0.31** 0.29** 0.30** 0.30** 0.20 0.46*** 0.20 0.38*** 0.33*** 0.28** 0.33*** 0.21* 0.28** 0.50***

17


1 2 3 4 5 6

1


73

measurement systems (Gates, 1999). Similarly, more than half of the respondents to the AICPA survey anticipated technology changes in their organizations in the next year to 18 months, and 79% rated the quality of information in their performance measurement systems as poor to adequate (AICPA & Maisel, 2001).

IMPACT ON PERFORMANCE Firms invest large amounts of resources in designing, implementing and maintaining performance measurement systems. It is thus critical to assess whether the choice of performance measures has had any significant impact on actual overall performance and on managerial decisions. In this study we used Pearson correlation coefficients to evaluate the relationship between performance measures and managerial performance. Table 3. Correlations Between Actual Use of Performance Measures and Performance. Measures

Correlation Coefficient with Performance

2-Tailed Significance (p Values)

Panel A: Correlation between actual financial measures used for compensation purposes and performance Sales growth 0.13 (0.319) Volume of orders 0.04 (0.782) Shipments 0.17 (0.178) Expense control 0.15 (0.224) Profitability 0.17 (0.169) Receivables management −0.03 (0.801) Inventory management 0 (0.971) Efficiency gains 0.26 (0.039)* Cash flow −0.09 (0.483) Panel B: Correlation between actual non-financial measures used for compensation purposes and performance Quality of products or services 0.16 (0.207) Customer satisfaction 0.19 (0.128) New product introduction 0.23 (0.063)* On-time delivery 0.16 (0.198) Accomplishment of project milestones 0.14 (0.258) Achievement of strategic objectives 0.37 (0.003)* Market share 0.33 (0.008)* Employee satisfaction 0.04 (0.736) ∗ Significant

at p < 0.10.

74


Panels A and B of Table 3 show that managers who have their compensation primarily tied to efficiency gains, new product introduction, achievement of strategic objectives, and market share tend to be the ones who perform the best overall. This result is consistent with Kaplan and Norton’s central argument that individual performance measures must be directly tied to strategy (Kaplan & Norton, 2001). This evidence also confirms what many companies have learned, albeit the hard way: there has to be a strong link between financial and non-financial performance measurements, lest the two may work against each other. For example, a company that may be doing well in terms of achieving certain strategic objectives, introducing new products and controlling a significant market share may still face financial difficulties, if efficiency is not properly considered when evaluating and rewarding its managers. In Panels A and B of Table 4 the correlations between managerial preferences for performance measures and overall performance are reported. The three Table 4. Correlations Between Preferred Use of Performance Measures and Performance. Measures



Panel A: Correlation between preferred financial measures used for compensation purposes and performance Sales growth 0.25 (0.049)* Volume of orders 0.32 (0.009)* Shipments 0.14 (0.283) Expense control 0.27 (0.028)* Profitability 0.15 (0.244) Receivables management 0.08 (0.506) Inventory management −0.05 (0.696) Efficiency gains −0.04 (0.738) Cash flow 0.31 (0.014)* Panel B: Correlation between preferred non-financial measures used for compensation purposes and performance Quality of products or services 0.28 (0.023)* Customer satisfaction 0.30 (0.018)* New product introduction 0.14 (0.254) On-time delivery 0.28 (0.024)* Accomplishment of project milestones 0.47 (0.000)* Achievement of strategic objectives 0.36 (0.003)* Market share 0.11 (0.371) Employee satisfaction −0.06 (0.654) ∗ Significant

at p < 0.10.


75

strongest correlations between preferred performance measures and performance are found in the accomplishment of project milestones, the achievement of strategic objectives, and the volume of customer orders. In other words, top-performing managers tend to prefer these measures when choosing which factors should be used to determine their compensation. Again, we observe the relevance of non-financial measures for the identification of superior performance.

Consensus and Performance One major challenge in trying to implement changes in the performance measurement system is to ensure buy-in from business-unit managers and other employees (AICPA & Maisel, 2001; Kaplan & Norton, 1996; Leahy, 2000). But this challenge does not necessarily mean that companies should select the performance measures Table 5. Correlations Between Measurement Gap (Absolute Differences Between Preferred and Actual Use of Performance Measures) and Performance. Measures



Panel A: Correlation between measurement gap in financial measures used for compensation purposes and performance Sales growth −0.25 (0.044)* Volume of orders −0.42 (0.000)* Shipments −0.03 (0.827) Expense control −0.06 (0.662) Profitability −0.14 (0.255) Receivables management −0.11 (0.401) Inventory management −0.15 (0.226) Efficiency gains −0.21 (0.092)* Cash flow −0.08 (0.506) Panel B: Correlation between measurement gap in non-financial measures used for compensation purposes and performance Quality of products or services 0.02 (0.860) Customer satisfaction −0.19 (0.123) New product introduction −0.18 (0.148) On-time delivery 0.05 (0.707) Accomplishment of project milestones 0.06 (0.630) Achievement of strategic objectives −0.35 (0.004)* Market share −0.15 (0.233) Employee satisfaction −0.09 (0.461) ∗ Significant

at p < 0.10.

76


that managers want to see as factors influencing their compensation. Some experts argue that striving too hard to achieve consensus may paralyze an organization’s effort to implement a new performance measurement system. Moreover, an excessive concern about consensus may lead to an incoherent measurement system, one that uses diverse measures to please groups with different interests but has no clear link to overall strategy. We explored the issue of consensus by conducting further correlation analysis on the relationship between the level of disagreement with the performance measurement system and performance. The results in Table 5 show that the magnitude of the disagreements do matter. Disagreements with the use of volume of customer orders, strategic objectives, sales, and efficiency gains for compensation purposes are associated with lower performance. One possible explanation is that managers who disagree strongly about having order volume, strategic objectives, sales, and efficiency gains influence their compensation lack the motivation necessary to achieve high performance levels. These measurement Table 6. Correlations Between Actual Use of Performance Measures and Gaming. Measures

Correlation Coefficient with Gaming


Panel A: Correlation between actual financial measures used for compensation purposes and gaming Sales growth 0.18 (0.145) Volume of orders 0.02 (0.862) Shipments 0.01 (0.929) Expense control 0.06 (0.638) Profitability −0.05 (0.666) Receivables management 0.12 (0.356) Inventory management −0.04 (0.738) Efficiency gains 0.16 (0.220) Cash flow −0.02 (0.875) Panel B: Correlation between actual non-financial measures used for compensation purposes and gaming Quality of products or services 0.05 (0.694) Customer satisfaction −0.10 (0.422) New product introduction 0.01 (0.933) On-time delivery −0.12 (0.329) Accomplishment of project milestones 0.08 (0.526) Achievement of strategic objectives 0.30 (0.018)* Market share 0.21 (0.096)* Employee satisfaction 0.11 (0.398) ∗ Significant

at p < 0.10.


77

gaps and their relationship with performance suggest that companies need to devote more attention to the preferences of participants in pay-for-performance plans, and involve them in the process of selecting performance measures to be used for compensation purposes. This result is in sharp contrast with one finding in the Conference Board survey mentioned above: while recognizing the potential resistance to change in performance measurement systems, only 9% of the respondents said they would “consider identifying key stakeholder reasons for resisting the strategic performance measurement effort” (Gates, 1999, p. 6).

EARNINGS MANAGEMENT Critics of non-financial performance measures often point out that these measures are more easily manipulated than financial measures, since they are neither audited Table 7. Correlations Between Preferred Use of Performance Measures and Gaming. Measures



Panel A: Correlation between preferred financial measures used for compensation purposes and gaming Sales growth 0.13 (0.313) Volume of orders 0.18 (0.158) Shipments 0.07 (0.607) Expense control −0.13 (0.299) Profitability 0.10 (0.417) Receivables management 0.03 (0.794) Inventory management −0.04 (0.751) Efficiency gains 0.06 (0.664) Cash flow 0.21 (0.090)* Panel B: Correlation between preferred non-financial measures used for compensation purposes and gaming Quality of products or services −0.00 (0.984) Customer satisfaction 0.21 (0.089)* New product introduction 0.38 (0.766) On-time delivery 0.20 (0.108) Accomplishment of project milestones 0.38 (0.002)* Achievement of strategic objectives 0.28 (0.027)* Market share 0.08 (0.516) Employee satisfaction 0.06 (0.650) ∗ Significant

at p < 0.10.

78


nor subject to generally accepted accounting principles. Hence we used Pearson correlation coefficients to examine the extent to which the type of performance measure used for compensation purposes is associated with managers engaging in earnings-management practices to achieve short-term results. Tables 6 and 7 show the correlations between the managers’ responses concerning performance measures used for compensation purposes and the likelihood that they would engage in gaming the results. In Tables 6 and 7, a positive correlation coefficient means that a greater emphasis on those measures is associated with a greater likelihood of gaming. As shown in Table 6, managers whose compensation actually depends largely on the achievement of strategic objectives and market share reported a higher likelihood that they would engage in earnings-management practices to manipulate the results. Interestingly, these same two measures were significantly correlated with performance (see Table 3). In addition, Table 7 shows that managers who reported a higher preference for Table 8. Correlations Between Measurement Gap (Preferred Minus Actual Use of Performance Measures) and Gaming. Measures



Panel A: Correlation between measurement gap in financial measures used for compensation purposes and gaming Sales growth −0.01 (0.931) Volume of orders 0.15 (0.225) Shipments 0.06 (0.665) Expense control 0.15 (0.228) Profitability 0.14 (0.269) Receivables management −0.08 (0.535) Inventory management 0.00 (0.983) Efficiency gains −0.10 (0.425) Cash flow 0.19 (0.138) Panel B: Correlation between measurement gap in non-financial measures used for compensation purposes and gaming Quality of products or services −0.04 (0.759) Customer satisfaction 0.30 (0.018)* New product introduction 0.02 (0.854) On-time delivery 0.27 (0.029)* Accomplishment of project milestones 0.28 (0.025)* Achievement of strategic objectives 0.06 (0.666) Market share −0.13 (0.306) Employee satisfaction −0.07 (0.589) ∗ Significant

at p < 0.10.


79

the use of project milestones, strategic objectives, customer satisfaction, and cash flow for compensation purposes also reported that they are more likely to manage earnings. This result suggests that some managers may prefer those measures precisely because they believe that they can manipulate their reported performance in those areas. Again, preferences for all four of these measures were found to be significantly related to performance (see Table 4). Furthermore, the results in Table 8 show that disagreements regarding the use of customer satisfaction, project milestones, and on-time delivery as non-financial performance factors that influence compensation are associated with a higher likelihood of gaming. These correlations suggest that a larger measurement gap in the use of these non-financial measures is positively related to managers reporting that they are more likely to manipulate results. Disagreements related to the use of financial measures were not found to be associated with gaming (see Panel A of Table 8). One possible explanation for this surprising result is that financial measures are perceived as less susceptible to manipulation; yet, recent corporate scandals involving deceptive financial reporting have proven this explanation less plausible.

SUMMARY AND RELEVANCE OF THE FINDINGS Recent studies show that companies are implementing significant changes in their performance measurement systems. Those systems are becoming more complex as organizations try to balance several measures, both financial and non-financial, to figure out who the top performers are, and to motivate and reward their continued performance. Over 70% of the companies that we surveyed place a strong emphasis on non-financial performance measures when calculating how much to pay their managers. Our results indicate that several non-financial measures are significantly related to superior overall performance. In particular, we find that the achievement of strategic objectives is a metric significantly correlated with performance. This measure is one of the three most commonly used non-financial measures, and one of the three most preferred measures. Managers who are most critical of the degree of emphasis placed on volume of customer orders, strategic objectives, efficiency gains, or sales for compensation purposes show the lowest performance. Given that those measures are among the ones that most often influence managerial compensation, companies should pay special attention on how they are used, or their intended incentive effects may never materialize. Instead, managers’ dissatisfaction with the use of those measures may actually hamper performance. We also find that managers who disagree

80


the most with the degree of importance given to customer satisfaction, on-time delivery, and project milestones are the ones most likely to manipulate reported performance. The results we obtained relating the measurement gap (the difference between preferred and actual measures) to overall performance and gaming suggest that the process of designing a measurement system may be at least as important as the actual measures selected in the end. Companies need to involve those who will participate in pay-for-performance plans in the process of selecting appropriate performance measures. Teams in charge of designing measurement systems could benefit from an awareness of managers’ preferences, as those preferences may reveal the measures that most closely reflect what managers can control. A well-managed design process should increase motivation and commitment, promote improved performance, and reduce the incidence of a gaming attitude. This is particularly true in companies that value participation and consensus. The design process should encourage open, honest dialogue on strategic objectives, and it should involve as many incentive plan participants as possible. When designers and participants reach an agreement about strategic objectives, then they can select specific measures based on those objectives without risking a loss of focus. Managing this design process presents quite a challenge for an organization’s executive team, for this team must strive to stimulate open discussion while maintaining a coherent vision. The design process may be seriously constrained by company politics. This process may become even more difficult when conflicts arise concerning how non-financial measures should be defined. In our survey we found that managers on average disagree more with their company’s use of nonfinancial measures than with their use of financial measures. Such disagreement results at least in part from inexperience with non-financial measurements. Our finding of a high use of non-financial measures in pay-for-performance plans presents an important challenge to the management accounting profession. While most management accountants have been trained to produce, analyze and communicate financial information, few are ready for the technical and human adjustments necessary to deal with non-financial information that is useful for managers. Not surprisingly, the AICPA survey anticipated that the level of effort finance professionals will be devoting to performance measurement will increase in the near future (AICPA & Maisel, 2001). Furthermore, the need to develop and implement non-financial measures brings additional pressure to bear on information systems to provide reliable information on how well managers are doing in a wide range of areas. In this information age, it is increasingly important to have information systems capable of supporting the complexities of cutting-edge performance measurement systems. Current surveys show that such information systems are in short supply.


81

Future research into the effectiveness of performance measurement systems should benefit from examining management’s preferences regarding the choice of performance metrics, as we introduced in this study. In relating performance metrics to managerial performance or other outcome variables, management accounting researchers need to pay adequate attention to the “measurement gap” variable we proposed in this study. As suggested by the results from this survey, managers’ disagreements with the performance measures chosen to determine their pay significantly influence performance and gaming. Omission of the measurement gap variable may well explain some of the conflicting evidence in the existing literature on the performance effects of performance metrics included in incentive plans. The results of this study also indicate that caution is in order before inclusion of non-financial performance measures in incentive plans. We found that managers who reported a high preference for certain non-financial measures also reported they would most likely engage in games to manipulate the short-term results and thereby increase their pay. Hence designers of performance measurement systems should carefully define these measures in order to ensure that they are as clear and objective as possible, and that all of the parties involved understand and accept them. This increased awareness and communication should also help ensure that, as companies change their performance measurement systems for compensation purposes, they will be able to narrow the measurement gap.

ACKNOWLEDGMENT Research support from a grant by the Merrick School of Business is gratefully acknowledged.

REFERENCES American Institute of Certified Public Accountants & Maisel, L. (2001). Performance measurement practices survey results. Jersey City, NJ: AICPA. Banker, R., Konstans, C., & Mashruwala, R. (2000). A contextual study of links between employee satisfaction, employee turnover, customer satisfaction and financial performance. Working Paper. University of Texas at Dallas. Banker, R., Potter, G., & Srinivasan, D. (2000). An empirical investigation of an incentive plan that includes nonfinancial performance measures. The Accounting Review (January), 65–92. Bento, R., & Ferreira, L. (1992). Incentive pay and organizational culture. In: W. Bruns (Ed.), Performance Measurement, Evaluation and Incentives (pp. 157–180). Boston: Harvard Business School Press. Bento, R., & White, L. (1998). Participant values and incentive plans. Human Resource Management Journal (Spring), 47–59.

82


Brownell, P., & Hirst, M. (1986). Reliance on accounting information, budgetary participation, and task uncertainty: Tests of a three-way interaction. Journal of Accounting Research, 241–249. Bruns, W., & Merchant, K. (1989). Ethics test for everyday managers. Harvard Business Review (March–April), 220–221. Bruns, W., & Merchant, K. (1990). The dangerous morality of managing earnings. Management Accounting (August), 22–25. Chatman, J. (1989). Improving interactional organizational research: A model of person-organization fit. Academy of Management Review, 333–349. Chatman, J. (1991). Matching people and organizations: Selection and socialization in public accounting firms. Administrative Science Quarterly, 36, 459–484. Chow, C., Kato, Y., & Merchant, K. (1996). The use of organizational controls and their effects on data manipulation and management myopia: A Japan vs. U.S. comparison. Accounting, Organizations and Society, 21, 175–192. Chow, C., Shields, M., & Wu, A. (1999). The importance of national culture in the design of management controls for multi-national operations. Accounting, Organizations and Society, 24, 441–461. Epstein, M. J., Kumar, P., & Westbrook, R. A. (2000). The drivers of customer and corporate profitability: Modeling, measuring, and managing the casual relationships. Advances in Management Accounting, 9, 43–72. Frigo, M. (2001). 2001 Cost management group survey on performance measurement. Montvale, NJ: Institute of Management Accountants. Frigo, M. (2002). Nonfinancial performance measures and strategy execution. Strategic Finance (August), 6–9. Gates, S. (1999). Aligning strategic performance measures and results. New York: Conference Board. Ittner, C., & Larcker, D. (1998a). Innovations in performance measurement: Trends and research implications. Journal of Management Accounting Research, 10, 205–238. Ittner, C., & Larcker, D. (1998b). Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction. Journal of Accounting Research (Suppl.), 1–35. Kaplan, R., & Norton, D. (1992). The balanced scorecard: Measures that drive performance. Harvard Business Review (January–February), 71–79. Kaplan, R., & Norton, D. (1996). The balanced scorecard. Boston: Harvard Business School Press. Kaplan, R., & Norton, D. P. (2001). The strategy-focused organization. Boston: Harvard Business Press. Kren, L. (1992). Budgetary participation and managerial performance. The Accounting Review, 511–526. Leahy, T. (2000). All the right moves. Business Finance (April), 27–32. Malina, M., & Selto, F. (2001). Communicating and controlling strategy: An empirical study of the effectiveness of the balanced scorecard. Journal of Management Accounting Research, 47–90. Merchant, K. (1984). Influences on departmental budgeting: An empirical examination of a contingency model. Accounting, Organization and Society, 9, 291–307. Nagar, V., & Rajan, M. (2001). The revenue implications of financial and operational measures of product quality. The Accounting Review (October), 495–513. Nouri, H., Blau, G., & Shahid, A. (1995). The effect of socially desirable responding (SDR) on the relation between budgetary participation and self-reported job performance. Advances in Management Accounting, 163–177. O’Reilly, C., Chatman, J., & Caldwell, D. (1991). People and organizational culture: A profile comparison approach to assessing person-organization fit. Academy of Management Journal, 487–516.


83

Posner, B. (1992). Person-organization values congruence: No support for individual differences as a moderating influence. Human Relations, 351–361. Rucci, A., Kirn, S., & Quinn, R. (1998). The employee-customer-profit chain at Sears. Harvard Business Review (January–February), 83–97. Simons, R. (2000). Performance measurement & control systems for implementing strategy: Text and cases. Upper Saddle River: Prentice-Hall. Sims, R., & Kroeck, K. (1994). The influence of ethical fit on employee satisfaction, commitment and turnover. Journal of Business Ethics, 13, 939–947.

AN EMPIRICAL EXAMINATION OF COST ACCOUNTING PRACTICES USED IN ADVANCED MANUFACTURING ENVIRONMENTS Rosemary R. Fullerton and Cheryl S. McWatters ABSTRACT Despite arguments that traditional product costing and variance analysis operate contrary to the strategic goals of advanced manufacturing practices such as just in time (JIT), total quality management (TQM), and Six Sigma, little empirical evidence exists that cost accounting practices (CAP) are changing in the era of continuous improvement and waste reduction. This research supplies some of the first evidence of what CAP are employed to support the information needs of a world-class manufacturing environment. Using survey data obtained from executives of 121 U.S. manufacturing firms, the study examines the relationship between the use of JIT, TQM, and Six Sigma with various forms of traditional and non-traditional CAP. Analysis of variance tests (ANOVA) indicate that most traditional CAP continue to be used in all manufacturing environments, but a significant portion of world-class manufacturers supplement their internal management accounting system with non-traditional management accounting techniques.


85

86

ROSEMARY R. FULLERTON AND CHERYL S. McWATTERS

INTRODUCTION Firms competing in a global arena and adopting sophisticated manufacturing technologies, such as total quality management (TQM) and just-in-time (JIT), require a complementary management accounting system (MAS) (Sillince & Sykes, 1995; Welfe & Keltyka, 2000). The MAS should support advanced manufacturing technologies by providing integrated information to interpret and to assess activities that have an impact on strategic priorities. The adoption of advanced manufacturing practices suggests a shift away from a short-term, financiallyoriented product focus towards a modified, more non-financial, process-oriented focus that fits operations strategies (Daniel & Reitsperger, 1991) and integrates activities with strategic priorities (Chenhall & Langfield-Smith, 1998b). Previous studies have reported that organizations using more efficient production practices make greater use of non-traditional information and reward systems (Banker et al., 1993a, b; Callen et al., 2002; Fullerton & McWatters, 2002; Ittner & Larcker, 1995; Patell, 1987); yet, little empirical evidence exists that cost accounting practices (CAP) of a firm’s MAS also are changing. The objectives of traditional product costing and variance analysis seemingly operate contrary to the strategic goals of continuous improvement and waste reduction embodied in advanced manufacturing production processes. It is argued that the benefits from JIT and TQM implementation would be captured and reflected more clearly by the parallel adoption of more simplified, non-traditional CAP. However, minimal evidence exists to support the assessment that current accounting practices are harmful to the improvement of manufacturing technology. In fact, most studies that have examined this issue find that companies continue to rely on conventional accounting information, even in sophisticated manufacturing environments (e.g. Baker et al., 1994; Cobb, 1992; McNair et al., 1989). Zimmerman (2003, p. 11) suggests that managers must derive some hidden benefits from continuing to use “presumably inferior accounting information” in their decision making. For example, management control over operations provided by the existing MAS may outweigh the benefits of other systems that are better for decision-making purposes. The objective of this study is to explore whether specific CAP are changing to meet the information needs of advanced manufacturing environments. To examine the CAP used in advanced manufacturing environments, a survey instrument was sent to executives representing 182 U.S. manufacturing firms. Data from the 121 survey responses were analyzed to determine whether the use of non-traditional CAP is linked to the implementation of JIT, TQM, and Six Sigma. The results show that there have been minimal changes in the use of traditional CAP. However, evidence exists that rather than replacing the traditional, internal-accounting practices, supplementary measures have been

An Empirical Examination of Cost Accounting Practices

87

added to provide more timely and accurate information for internal planning and control. Perhaps much of the criticism of CAP is unfounded, and the emergence of supplemental financial and non-financial information, combined with traditional accounting techniques, equips management with the appropriate decision-making and control tools for an advanced manufacturing environment. This paper provides some of the first empirical evidence of what CAP actually are being used in conjunction with JIT, TQM, and Six Sigma. The remainder of this paper is organized as follows: The following section examines the prior literature related to advanced manufacturing practices and CAP, and identifies the research question. The next section describes the research method. The following two sections present and discuss the empirical results. The final section briefly summarizes the study and provides direction for future research.

RESEARCH QUESTION DEVELOPMENT Information Needs for Advanced Manufacturing Environments Advanced Manufacturing Environments Defined World class manufacturing (WCM) was defined initially by Hayes and Wheelwright (1984) and Schonberger (1986) as a competitive strategy employing the best practices in quality, lean production, and concurrent engineering. Voss (1995, p. 10) defines “Best Practices” as the most recent of three accepted paradigms of manufacturing strategy. Encompassed within this paradigm are the world-class elements of “JIT manufacturing which has evolved into lean production, TQM, and concurrent engineering.” According to McNair et al. (1989, p. 9), the “manufacturing competitiveness of the future is dependent upon a rapid development of JIT and other advanced manufacturing technology principles.” Both TQM and JIT have been described as organization-wide philosophies that focus on continuous improvement. They often are linked as complementary manufacturing strategies (see Banker et al., 1993a, b; Dean & Snell, 1996; Flynn et al., 1995; Kristensen et al., 1999; Vuppalapati et al., 1995). TQM concentrates on “systematically and continuously improving the quality of products, processes, and services.” JIT’s emphasis on waste elimination and productivity improvement is expected to increase organizational efficiency and effectiveness (Yasin et al., 1997) through better quality, lower inventory, and shorter lead times. Six Sigma is a more recent manufacturing technique that also has been described as “an organization methodology for achieving total quality throughout the company” (Witt, 2001, p. 10). Like TQM, Six Sigma’s focus is on the customer’s definition of quality, as it establishes objectives for reductions in defect rates (Linderman

88


et al., 2003). In comparison to TQM, Six Sigma is a rigorous measurement tool that depends on full commitment of top management, and evaluates its effectiveness in terms of bottom-line results (Ellis, 2001; Hendricks & Kelbaugh, 1998). Its success is dependent upon the adoption of and commitment to lean production practices (Voelkel, 2002). Inadequacies of Traditional CAP Considerable literature asserts that existing CAP do not support a WCM environment (i.e. Drury, 1999; Green et al., 1992; Harris, 1990; Hedin & Russell, 1992; Johnson, 1992; Kaplan, 1983, 1984; McNair et al., 1989, 1990; Sillince & Sykes, 1995; Wisner & Fawcett, 1991). Holzer and Norreklit (1991) state that the “greatest impact on cost accounting in recent years has come about through rapid advances in manufacturing technology and the adoption of a JIT management philosophy.” The classical model of cost accounting was designed for a different environment and encourages inappropriate behavior (Holzer & Norreklit, 1991; Howell & Soucy, 1987). Cobb (1992) warns that if management accountants do not respond to the information needs of the new manufacturing environment, they will be relegated to a historical recording role, and another function within the organization will be the forefront provider of internal information. Daniel and Reitsperger (1996, p. 116) reiterated this view: The skill-specific isolation of management accountants seems to be a significant constraint in adjusting management control systems in the U.S. Our evidence shows that progressive management accountants are scrambling for a ‘new purpose’ in a rapidly changing manufacturing environment, while traditionalists in the profession desperately seek to maintain systems that are incompatible with the dynamism of JIT manufacturing.

Research studies support the increasing lack of reliance upon management accounting. For example, case studies report that operations managers prepare their own cost accounting data to support the implementation of new manufacturing practices, with little communication or respect for the internal accountants (Chenhall & Langfield-Smith, 1998a; Sillince & Sykes, 1995). More than a decade ago, standard costing and variance analysis were declared inadequate and their continued use potentially contributing to lost competitiveness and other dysfunctional consequences. Standards often encompass waste and encourage mediocrity, focusing on conformance to standards, rather than continuous improvement (Hendricks, 1994). Unless updated and reported frequently, standards lack timeliness. Achieving standard direct cost ‘efficiency’ targets leads to larger batches, longer production runs, more scrap, more rework, and less communication across process . . . In any plant striving to achieve manufacturing excellence, standard cost performance systems are anathema –


89

especially if incentive compensation is geared to controlling standard-to-actual variances (Johnson, 1992, p. 49).

Research Question The benefits that firms reap from implementing advanced manufacturing techniques appear to be enhanced by complementary changes in their internal accounting measures (Ahmed et al., 1991; Ansari & Modarress, 1986; Barney, 1986; Ittner & Larcker, 1995; Milgrom & Roberts, 1995). “Quality improvement advocates argue that the organizational changes needed for effective TQM require new approaches to management accounting and control,” with more comprehensive distribution of new types of information that measure quality and team performance (Ittner & Larcker, 1995, p. 3). This study examines the association between the adoption of advanced manufacturing techniques and specific CAP in terms of the following research question: Do firms that implement advanced manufacturing techniques such as JIT, TQM, and Six Sigma use more non-traditional cost accounting practices?

RESEARCH METHOD Data Collection Survey Instrument To explore the research question, a detailed survey instrument was used to collect specific information about the manufacturing operations, product costing methods, information and incentive systems, advanced manufacturing practices employed, and characteristics of the respondent firms.1 The majority of the questions on the survey instrument were either categorical or interval Likert scales. Factor analysis combined the Likert-scaled questions into independent measures to test the research question. To evaluate the survey instrument for readability, completeness, and clarity, a limited pretest was conducted by soliciting feedback from business professors and managers from four manufacturing firms that were familiar with advanced manufacturing practices. Appropriate changes were made to reflect their comments and suggestions. Sample Firms The initial sample firms for this study constituted 253 pre-identified executives from manufacturing firms that had responded to a similar survey study in 1997 (see Fullerton & McWatters, 2001, 2002). Of the original 253 firms, 66 (26%) were no longer viable, independent businesses. Over half of the initial respondents

90


in the 187 remaining firms were no longer with the company. Replacements were contacted in all but the five firms that declined to participate. Thus, 182 manufacturing executives were contacted a maximum of three times via e-mail, fax, or mail.2 One hundred twenty-one usable responses were received, for an overall response rate of 66%. The majority of the respondents had titles equivalent to the Vice President of Operations, the Director of Manufacturing, or the Plant Manager. They had an average of 19 years of management experience, including 12 years in management with their current firm. The respondent firms had a primary two-digit SIC code within the manufacturing ranges of 20 and 39. As shown on Table 1, the majority (64%) of the respondent firms is from three industries: industrial machinery (SIC-35, 16%), electronics (SIC-36, 28%), and instrumentation (SIC-38, 20%).3

Advanced Manufacturing Practices Discrete Measures for JIT, TQM, and Six Sigma The advanced manufacturing production processes examined in this study are JIT, TQM, and Six Sigma. Survey respondents were asked to indicate “yes” or “no,” whether or not they had formally implemented JIT and TQM, and whether or not they were using Six Sigma. Thirty-seven of the firms (31%) indicated that they were not using any of the three techniques. As Table 1 demonstrates, many of the firms had implemented a combination of these practices. Of the 121 respondents, only 15, 13, and 5 firms had exclusively implemented JIT, TQM, or Six Sigma, respectively. Almost half of the firms had adopted TQM, 43% identified themselves as JIT firms, while 35% indicated they were using Six Sigma techniques. Adopting JIT and TQM together (one quarter of the firms) was the most robust combination of advanced manufacturing processes. Only 16% of the sample firms are using all three techniques. Survey Measures for Level of JIT Implementation In addition, respondents were asked to indicate the level to which their firms were using JIT methods as measured by lean manufacturing practices, quality improvements, and kanban systems. These data enabled the analysis of a broader perspective of JIT implementation, rather than as an either/or proposition. Although a universal set of JIT elements remains to be specified within the research literature (Davy et al., 1992; White & Ruch, 1990), different practices deemed important to successful JIT adoption are suggested in several studies (Koufteros et al., 1998; Mehra & Inman, 1992; Moshavi, 1990; Sakakibara et al., 1993; Spencer & Guide, 1995; Yasin et al., 1997). White and Ruch (1990) found

Industry

JIT Firms

TQM Firms

Six Sigma Firms

TQM & JIT Firms

TQM, JIT, & Six Sigma Firms

No JIT, TQM, or Six Sigma Firms

Total Sample

20 – Food 22 – Textiles 25 – Furniture & fixtures 26 – Paper & allied products 27 – Printing/publishing 28 – Chemicals & allied products 30 – Rubber products 32 – Nonmetallic mineral products 33 – Primary metals 34 – Fabricated metals 35 – Industrial machinery 36 – Electronics 37 – Motor vehicles & accessories 38 – Instrumentation 39 – Other manufacture

0 0 5 0 0 0 1 0 3 3 11 14 2 12 1

0 0 3 1 1 4 1 1 4 3 10 16 3 11 1

1 0 2 1 0 0 0 1 5 1 5 16 2 7 1

0 0 3 0 0 0 1 0 2 2 8 7 2 7 1

0 0 2 0 0 0 0 0 2 1 2 7 2 2 1

2 1 0 0 0 5 0 0 3 2 6 9 1 7 1

3 1 5 2 1 9 1 1 9 6 19 34 4 24 2

Totals

52

59

42

33

19

37

121


Table 1. Distribution of Sample Firms by Production Processes and Two-Digit SIC Codes.a

Supplemental information: 15 firms implemented JIT exclusively; 13 firms implemented TQM exclusively; 5 firms implemented six sigma exclusively; 13 firms implemented only TQM and six sigma; 5 firms implemented only JIT and six sigma. a Classification of production processes were self-identified by survey respondents.

91

92


ten consensus JIT elements identified in the work of established JIT authors (e.g. Hall et al.). These consensus elements used by White (1993), White et al. (1999), White and Prybutok (2001), Fullerton and McWatters (2001, 2002) and Fullerton et al. (2003) as JIT indicators are designated as follows: focused factory, group technology, reduced setup times, total productive maintenance, multi-function employees, uniform workload, kanban, JIT purchasing, total quality control, and quality circles.4 Factor Analysis Using the above-noted JIT indicators, eleven survey questions asked respondents to identify their firm’s level of JIT implementation on the basis of a six-point Likert scale, ranging from “no intention” of implementing the identified JIT practice to its being “fully implemented.”5 Using the principal components method, these items were subjected to a factor analysis. Three components of JIT with eigenvalues greater than 1.0 were extracted from the analysis, representing 61% of the total variance in the data.6 The first factor is a manufacturing component that explains the extent to which companies have implemented general manufacturing techniques associated with JIT, such as focused factory, group technology, uniform work loads, and multi-function employees. The second JIT factor is a quality component that examines the extent to which companies have implemented procedures for improving product and process quality. The third JIT factor identified is one of uniquely JIT practices that describe the extent to which companies have implemented JIT purchasing and kanban. For the results of the factor analysis, which are similar to those of previous studies by Fullerton and McWatters (2001, 2002), see Table 2.

Cost Accounting Practices McNair et al. (1989) identified three trends that differentiate traditional and non-traditional MAS: (1) a preference for process costing; (2) the use of actual costs instead of standard costs; and (3) a greater focus on traceability of costs. Traditional variance reports support maximum capacity utilization, which contradicts the JIT objective to produce only what is needed, when it is needed. Traditional CAP encourage production of goods with a high contribution margin and ignore constraints or bottlenecks. For performance evaluation, shop-floor cost accounting measures emphasize efficiency, and encourage large batches and unnecessary inventory (Thorne & Smith, 1997). Cooper (1995) suggests that the missing piece to the puzzle of Japanese cost superiority in lean enterprises is the role of cost management systems. Western enterprises use cost accounting systems, rather than cost management systems, which have different objectives.


93

Table 2. Factor Analysis (VARIMAX Rotation) Factor Loadings for JIT Variables.a Factor 1 JITMANUFb Focused factory Group technology Reduced setup times Productive maintenance Multi-function employees Uniform work load Product quality improvement Process quality improvement Kanban system JIT purchasing

Factor 2 JITQLTYc

Factor 3 JITUNIQUEd

0.700 0.761 0.618 0.727 0.509 0.637 0.887 0.848 0.738 0.787

Notes: n = 121. All loadings in excess of 0.400 are shown. a Respondents were asked to indicate the extent to which their firm had implemented the individual JIT elements. Possible responses were: No intention = 1; Considering = 2; Beginning = 3; Partially = 4; Substantially = 5; Fully = 6. b Cronbach’s Alpha – 0794. c Correlation Coefficients significant at p < 0.001: 0.772. d Correlation Coefficients significant at p < 0.001: 0.438.

Cost accounting systems report distorted product costs without helping firms manage costs. Cost management systems control costs by designing them out of products through such techniques as target costing and value engineering. Process costing (PROCESS), which simplifies inventory accounting procedures by reducing the need to track inventory, is generally considered to be better suited to the JIT environment than job-order costing. When Hewlett-Packard introduced JIT, all products were produced and accounted for in batches. JIT gradually transformed the manufacturing to a continuous process. The MAS changed to process costing, and CAP were simplified and streamlined (Patell, 1987). Over half of the sample firms in Swenson and Cassidy (1993) switched from job-order costing to process costing after JIT implementation. Related to the physical flow of inventory through the system is the parallel recording to account for it. In a process environment, the traditional costing methods have tracked work in process inventory per equivalent units. In an advanced manufacturing environment where both raw materials and work in process inventory is minimized, detailed tracking of inventory can be unnecessary (Scarlett, 1996). The simplification of product costing and inventory valuation in a JIT environment calls for backflush accounting (BCKFL), where inventory costs are determined at the completion of the production cycle (Haldane, 1998).

94


Although still widely used, standard costing (STDRD) was developed in a different business environment than currently exists. Standards in a traditional standard costing system can incorporate slack, waste, and downtime, without encouraging improvement. This system also allows for manipulation and mediocrity and may not be appropriate for advanced manufacturing environments (McNair et al., 1989). Rolling averages of actual performance, as benchmarks to monitor performance, are preferred to estimates of what is expected (Green et al., 1992; Hendricks, 1994). Haldane (1998) claimed that some uses of standard costing were “pernicious” and actually “enshrine waste.” However, standard costing may be a good tool if it is used properly to monitor trends for continuous improvement (Drury, 1999). Case studies have shown that lean-manufacturing Japanese firms use standard costs, but often adapt them continually to include Kaisen improvements (Cooper, 1996). Related to standard costing is the use of variance analysis (VARAN). Variances identify results that differed from initial expectations, not what caused the deviation to occur. Avoidance of negative variances actually can impede the implementation of lean manufacturing practices (Najarian, 1993). The use of the traditional labor and machine utilization as volume and efficiency criteria encourages overproduction and excess inventories (Drury, 1999; Fisher, 1992; Haldane, 1998; Hendricks, 1994; Johnson, 1992; McNair et al., 1989; Wisner & Fawcett, 1991). Standard-costing data used in traditional variance analysis lack relevancy and can lead to defects before variances are noted and problems corrected (Hendricks, 1994). In addition, the actual collection of variance information is a non-value-added activity that increases cycle time (Bragg, 1996). It is natural to conclude from this literature that the relevance of cost accounting information would increase if managers spent more time checking the accuracy of product costs, rather than reading variance reports. Use of the absorption method (ABSORP) for inventory costing is required in many jurisdictions to meet GAAP financial reporting requirements. Absorption costing encourages inventory building by attaching all manufacturing overhead costs to inventory and postponing the recording of the expense until the product is sold. Building and storing inventory is contrary to lean manufacturing objectives, yet can enhance net income. Traditional overhead allocation focuses on “overhead absorption,” rather than on “overhead minimization” (Gagne & Discenza, 1992). A suggested alternative for restraining the motivation to build inventory is the replacement of absorption costing with variable (direct) or actual costing for internal reporting of inventories (McWatters et al., 2001). Since the mid-1980s, activity-based-costing (ABC) has been cited as a remedy for the deficiencies of traditional cost accounting in advanced manufacturing environments (Anderson, 1995; Cooper, 1994; Shields & Young, 1989). ABC promotes decisions that are consistent with a lean manufacturing environment


95

through an enhanced focus on tracing and measuring costs of activities that consume resources (Cooper & Kaplan, 1991; Gagne & Discenza, 1992). The balanced scorecard (BSC) approach for measuring all key aspects of a business, both financial and non-financial, was developed by Kaplan and Norton. Atkinson et al. (1997) referred to it as one of the most significant developments in management accounting. Examining measures from four perspectives, financial, customer, internal process, and learning and growth, the scorecard can provide managers with an integrative framework for controlling management operations (Clinton & Hsu, 1997; Hoque & James, 2000). The internal business perspective of the balanced scorecard is especially conducive to evaluating JIT and TQM environments. Clinton and Hsu (1997, p. 19) pointed out that the implementation of JIT is a major change in manufacturing control that needs the support of a change in management control to avoid an “incongruent state that results in inconsistent performance evaluation and dysfunctional behavior.” Frigo and Litman (2001) explained that the downfall of companies is often their failure to execute their business strategy, rather than the business strategy itself. They propose the implementation of Six Sigma, lean manufacturing, and balanced scorecard as practices that will help align business strategy with its execution. Target costing (TGCST) for new product development concentrates on waste and cost reduction in the initial planning stages of a product. The greater emphasis by Japanese firms on cost management, rather than cost accounting is reflected in their focus on target costing over product costing (Cooper, 1995, 1996). Lean enterprises use target costing to encourage lower production costs, with higher quality and faster response times (Cooper, 1996). Case studies have indicated that simplification of the product-costing system and better integration of costing with strategic objectives can occur when firms rely on target costing (Mackey & Hughes, 1993; Sillince & Sykes, 1995). Economic Value Added (EVA) is a variant of residual income. Developed by Stern Stewart and Company,7 EVA adjusts the financial-reporting net-income figure to take into account costs such as research and development and marketing that are treated as period costs according to GAAP. It also deducts the weightedaverage cost of capital investment from the traditional net income figure to capture economic value that is represented on both the income statement and the balance sheet. The resulting calculation is said to capture the true value of shareholder wealth (Tully, 1993). Pettit (2000) states that EVA is “an integrated performance measurement, management, and reward system for business decision making.” He demonstrates how EVA measures can gauge the effectiveness of Six Sigma and lean production initiatives. Life-cycle costing (LCC) is described as the tracking and accumulation of product costs from a product’s inception to its abandonment. Proponents claim

96


that it better matches revenues and expenses, because it defers and allocates to future periods all initial costs of research, marketing, and start-up to the period in which the actual production of the units occurs and the benefits from these prior activities are expected to be received. Activities are expensed based on the number of units expected to be sold. Similar to management accounting in the world-class Japanese manufacturing firms, the MAS should be integrated with corporate strategy, and LCC should be integrated with the MAS (Ferrara, 1990). LCC provides better decision-making information, because it is more reflective of today’s advanced manufacturing environments (Holzer & Norreklit, 1991; Peavey, 1990). In fact, prediction of life-cycle costs is a requirement for the quality-initiative, ISO 9000 certification process (Rogers, 1998). Life-cycle costing and value chain analysis (VCA) are related concepts. Value chain analysis focuses on all business functions for a product, from research and development to customer service, whether it is in the same firm or different organizations (Horngren et al., 1997, p. 14). For example, TQM, business process re-engineering, and JIT may not be as successful as anticipated, because the necessary changes to support these processes have not been replicated for all of the firms along the supply chain. It is not effective to try to optimize each piece of the supply chain. All successive firms in the manufacturing process, beginning with the customer order, back to the purchase of raw materials, must be evaluated and integrated. Using “lean logistics,” activities should be organized to move in an uninterrupted flow within a pull production system (Jones et al., 1997). Survey Measures for CAP The 11 CAP categorical variables as outlined above (PROCESS, BCKFL, STDRD, VARAN, ABSORP, ABC, BSC, TGCST, EVA, LCC, and VCA) are evaluated from questionnaire responses. All of the variables except for PROCESS and ABSORP were answered on the questionnaire as a “yes” or “no” with respect to their use. For PROCESS, respondents indicated that they used either a process costing or job-order costing system. For ABSORP, respondents chose between the use of absorption or direct (variable) costing for internal estimation of inventory cost. All of the variables are coded as “0” or “1,” with “1” representing the cost accounting “preferred choice” in an advanced manufacturing environment.

Contextual Variables Four contextual variables are also examined. Firm size (SIZE) affects most aspects of a firm’s strategy and success; therefore, each sample firm’s net sales figure, as obtained from COMPUSTAT data, is used to examine firm size. Whether a firm follows a more innovative strategy can affect its willingness to make


97

changes. Innovation (INNOV) is measured by a firm’s response to the five-point Likert-scaled question on the survey instrument as to whether the firm is a leader or a follower in product technology, product design, and process design (Fullerton & McWatters, 2002; Ittner & Larcker, 1997). Top management commitment has been discussed as a necessary ingredient for successful implementation of JIT, TQM, and Six Sigma. Several descriptive survey studies found the lack of support from top management to be a serious problem in the JIT implementation process (Ansari & Modarress, 1986; Celley et al., 1986; Im, 1989; Lee, 1997). For Six Sigma to be successful, it must become a part of the organizational culture with the unwavering support of top management (Ellis, 2001; Hendricks & Kelbaugh, 1998; Witt, 2002). Top management commitment (COMMIT) is measured by the responses to three five-point Likert-scaled questions on the survey instrument that ask how supportive (from indifferent to highly supportive) top management is in initiating change programs, implementing lean manufacturing practices, and providing training for new production strategies. The last contextual variable asks the respondents to identify how satisfied they are with their firm’s management accounting system (MAS), from “1” not at all, to “5” extremely.

RESEARCH RESULTS ANOVA Results for CAP and Advanced Manufacturing Practices On the survey instrument, respondents indicated whether or not (yes or no) their firm had formally implemented the advanced manufacturing techniques of JIT, TQM and Six Sigma. In addition, they were asked to rate their manufacturing operations on a scale from 1 to 5, with 1 being traditional and 5 world class. The sample was separated into JIT and non-JIT, TQM and non-TQM, and Six Sigma and non-Six Sigma users to compare the firm mean differences in the use of specific CAP practices. The sample was further segregated into firms using both JIT and TQM or neither, as well as firms that had implemented one of the three WCM methods in comparison to firms that had none of the three methods in place. The ANOVA results for the five different classifications are fairly similar, which may indicate that the same measurement and information tools are used to support most advanced manufacturing practices (Tables 3–7). Of the 11 CAP practices examined, three of them consistently show significantly more use in advanced manufacturing environments than in traditional manufacturing operations. These are EVA, VCA, and LCC. There is limited and inconsistent significant means differences among the remaining eight CAP practices evaluated. Process costing, as opposed to job-order costing, is used more in JIT firms than in non-JIT firms, along with backflush costing. TQM and Six Sigma environments

98


Table 3. Comparison of CAP, Contextual Factors, and Production Processes Means Between JIT Firms and Non-JIT Firms. Full Sample Means n = 121 Cost accounting practices (CAP) Process or backflush costing (1), Job-order costing (0) Backflush costinga Direct (variable) costing (1), Absorption costing (0) Standard costingb Balanced scorecarda Target costinga Variance analysisb Activity based costinga Economic value addeda Value chain analysisa Life cycle costinga

JIT Firms Means n = 46

Non-JIT Firms Means n = 75

ANOVA F-Value

Sig. F

PROCESS

0.624

0.726

0.546

4.044

0.047

BCKFL ABSORP

0.231 0.609

0.373 0.615

0.121 0.600

11.028 0.028

0.001 0.868

STDRD BSC TGCST VARAN ABC EVA VCA LCC

0.076 0.218 0.835 0.096 0.571 0.382 0.348 0.242

0.058 0.297 0.854 0.058 0.532 0.476 0.477 0.378

0.091 0.146 0.820 0.127 0.603 0.298 0.222 0.120

0.450 2.621 0.228 1.575 0.535 3.022 6.713 9.230

0.504 0.110 0.634 0.212 0.466 0.086 0.011 0.003

3.712 3.657 3.111 961.240

4.135 3.853 3.157 1612.635

3.379 3.508 3.076 480.980

17.272 2.796 0.267 7.204

0.000 0.097 0.606 0.008

0.496 0.342 3.202

0.635 0.453 3.558

0.388 0.254 2.925

7.445 5.361 17.238

0.007 0.022 0.000

Contextual factors Management commitment∗ Innovation strategy∗∗ MAS Satisfaction∗∗∗ Firm sizec

COMMIT INNOV MAS SIZE

Production processes Total quality managementa Six Sigmaa World class manufacturing∗∗∗∗

TQM 6-SIG WCM

Notes: Survey possible responses: ∗ How supportive is top management? Indifferent = 1 . . . 2 . . . 3 . . . 4 . . . 5 = Highly Supportive. ∗∗ What represents your firm’s strategy related to product and process innovation? Follower = 1 . . . 2 . . . 3 . . . 4 . . . 5 = Leader. ∗∗∗ How satisfied are you with your management accounting system? 1 = not at all; 2 = little; 3 = somewhat, 4 = considerably, 5 = extremely. ∗∗∗∗ How would you rate your manufacturing operations? Traditional = 1 . . . 2 . . . 3 . . . 4 . . . 5 = World Class. a yes = 1; no = 0. b yes = 0; no = 1. c Information provided from COMPUSTAT database (net sales in millions of dollars).


99

Table 4. Comparison of CAP, Contextual Factors, and Production Processes Means Between TQM Firms and Non-TQM Firms. Full Sample Means n = 121 Cost accounting practices (CAP) Process or backflush costing (1), Job-order costing (0) Backflush costinga Direct (variable) costing (1), Absorption costing (0) Standard costingb Balanced scorecarda Target costinga Variance analysisb Activity based costinga Economic value addeda Value chain analysisa Life cycle costinga

TQM Firms Means n = 59

NonTQM Firms Means n = 62

ANOVA F-Value

Sig. F

PROCESS

0.624

0.621

0.633

0.020

0.888

BCKFLSH ABSORP

0.231 0.609

0.293 0.579

0.167 0.644

2.687 0.511

0.104 0.476

STDRD BSC TARGET VARAN ABC EVA VCA LCC

0.076 0.218 0.835 0.096 0.571 0.382 0.348 0.242

0.069 0.250 0.907 0.070 0.654 0.500 0.475 0.370

0.082 0.180 0.750 0.121 0.482 0.271 0.240 0.120

0.071 0.571 4.892 0.839 3.242 5.182 5.656 8.798

0.791 0.452 0.029 0.362 0.075 0.025 0.020 0.004

Contextual factors Management commitment∗ Innovation strategy∗∗ MAS satisfaction∗∗∗ Firm size c


3.712 3.657 3.111 961.240

3.977 3.472 3.922 3.383 3.086 3.133 1504.484 432.472

7.209 7.122 0.093 6.172

0.008 0.009 0.761 0.015

Production processes Just-in-timea Six-sigmaa World class manufacturing∗∗∗∗

JIT 6-SIG WCM

7.445 22.032 11.941

0.007 0.000 0.001

0.437 0.342 3.202

0.559 0.542 3.466

0.317 0.164 2.934


100


Table 5. Comparison of CAP, Contextual Factors, and Production Processes Means Between Six Sigma Firms and Non-Six Sigma Firms. Full Sample Means n = 121 Cost accounting practices (CAP) Process or backflush costing (1), Job-order costing (0) Backflush costinga Direct (variable) costing (1), Absorption costing (0) Standard costingb Balanced scorecarda Target costinga Variance analysisb Activity based costinga Economic value addeda Value chain analysisa Life cycle costinga

Six Sigma Firms Means n = 42

Non-Six Sigma Firms Means n = 79

ANOVA F-Value

Sig. F

PROCESS

0.624

0.610

0.636

0.080

0.788

BCKFLSH ABSORP

0.231 0.609

0.268 0.575

0.208 0.632

0.548 0.348

0.461 0.556


0.076 0.218 0.835 0.096 0.571 0.382 0.348 0.242

0.073 0.480 0.921 0.100 0.649 0.556 0.556 0.364

0.077 0.093 0.778 0.092 0.522 0.302 0.254 0.175

0.005 18.325 3.626 0.019 1.573 5.381 8.132 4.352

0.942 0.000 0.060 0.891 0.213 0.023 0.005 0.040

3.712 4.133 3.657 3.683 3.111 3.244 961.240 1758.545

3.506 3.628 3.039 540.151

10.338 0.064 1.620 7.390

0.002 0.801 0.206 0.008

0.346 0.3967 3.076

22.032 5.361 4.791

0.000 0.022 0.031

Contextual factors Management commitment∗ Innovation strategy∗∗ MAS Satisfaction∗∗∗ Firm sizec


Production processes Total quality managementa Just-in-timea World class manufacturing∗∗∗∗

TQM JIT WCM

0.496 0.442 3.202

0.762 0.585 3.44



101

Table 6. Comparison of CAP, Contextual Factors, and Production Processes Means Between Firms that have Implemented both TQM and JIT and Firms that have not Implemented Either. Full Sample Means n = 121 Cost accounting practices (CAP) Process or backflush costing (1), Job-order costing (0) Backflush costinga Direct (variable) costing (1), Absorption costing (0) Standard costingb Balanced scorecarda Target costinga Variance analysisb Activity based costinga Economic value addeda Value chain analysisa Life cycle costinga Contextual factors Management commitment Innovation strategy MAS satisfaction Firm sizec Production processes Six-sigmaa World class manufacturing

JIT & TQM Firms Means n = 29

Non-JIT & Non-TQM Firms Means n = 92

ANOVA F-Value

Sig. F

PROCESS

0.624

0.688

0.605

0.677

0.412

BCKFLSH ABSORP

0.231 0.609

0.438 0.645

0.151 0.600

11.730 0.192

0.001 0.662


0.076 0.218 0.835 0.096 0.571 0.382 0.348 0.242

0.063 0.292 0.967 0.031 0.621 0.560 0.577 0.444

0.085 0.182 0.775 0.119 0.546 0.308 0.250 0.159

0.106 1.181 5.803 2.082 0.478 5.056 9.478 9.313

0.745 0.280 0.018 0.152 0.491 0.027 0.003 0.003


3.712 3.657 3.111 961.240

4.400 3.984 3.063 2249.886

3.477 3.528 3.128 526.790

20.823 3.879 0.142 12.902

0.000 0.051 0.707 0.000

0.342 3.202

0.594 3.677

0.258 3.034

12.712 13.764

0.001 0.000

6-SIG WCM


employ target costing. Firms using the Six Sigma approach demonstrate a strong preference for using the balance scorecard measures over firms that do not use Six Sigma. Little difference exists in the sample firms’ use of a standard costing system, with over 90% of all the classifications indicating that standard costs were used internally for estimating product costs. Also, around 90% of the three

102


Table 7. Comparison of CAP and Contextual Factors Means between Firms with no Advanced Manufacturing (AM) Processes and Firms with some AM Processes. Full Sample Means n = 121 Cost accounting practices (CAP) Process or backflush costing (1), Job-order costing (0) Backflush costinga Direct (variable) costing (1), Absorption costing (0) Standard costingb Balanced scorecarda Target costinga Variance analysisb Activity based costinga Economic value addeda Value chain analysisa Life cycle costinga

No AM Processes Firms Means n = 39

Some AM Processes Firmsc Means n = 82

ANOVA F-Value

Sig. F

PROCESS

0.624

0.568

0.654

0.809

0.370

BCKFLSH ABSORP

0.229 0.609

0.108 0.722

0.284 0.563

4.547 2.683

0.035 0.104


0.076 0.218 0.835 0.096 0.571 0.382 0.348 0.242

0.108 0.083 0.794 0.114 0.531 0.241 0.207 0.034

0.061 0.273 0.842 0.086 0.581 0.443 0.410 0.328

0.802 3.622 0.373 0.218 0.222 3.441 3.651 10.438

0.372 0.061 0.543 0.642 0.683 0.067 0.059 0.002

Contextual factors Management commitment∗ Innovation strategy∗∗ MAS satisfaction∗∗∗ Firm sized


3.712 3.657 3.111 961.240

3.297 3.284 3.081 195.007

3.811 3.811 3.124 1278.641

9.294 5.876 0.065 5.312

0.003 0.017 0.799 0.023

Production processes World class manufacturing∗∗∗∗

WCM

3.202

2.676

3.434

22.686

0.000

Notes: Survey possible responses: ∗ How supportive is top management? Indifferent = 1 . . . 2 . . . 3 . . . 4 . . . 5 = Highly Supportive. ∗∗ What represents your firm’s strategy related to product and process innovation? Follower = 1 . . . 2 . . . 3 . . . 4 . . . 5 = Leader. ∗∗∗ How satisfied are you with your management accounting system? 1 = not at all; 2 = little; 3 = somewhat, 4 = considerably, 5 = extremely. ∗∗∗∗ How would you rate your manufacturing operations? Traditional = 1 . . . 2 . . . 3 . . . 4 . . . 5 = World Class. a yes = 1; no = 0. b yes = 0; no = 1. c Firms that have implemented JIT, TQM, or Six Sigma exclusively or some combination of them. d Information provided from COMPUSTAT database (net sales in millions of dollars).

advanced manufacturing production processes support their standard costing system with some type of efficiency, price, or volume variance analysis. Direct costing is used similarly by the majority of the sample firms in all manufacturing environments to estimate internal inventory cost. Slightly over half of the firms use an activity-based costing system, but there is only a marginally significant


103

difference in its use in TQM firms compared to non-TQM firms. In fact, although not a significant difference, more non-JIT firms than JIT firms are using ABC. This agrees with the study by Abernethy et al. (2001) that concluded firms adopting advanced manufacturing technologies may not be well served by a sophisticated, hierarchical ABC system. The results support the argument that most firms identified as world-class manufacturers will adopt a combination of advanced manufacturing techniques. Each of the advanced manufacturing processes (JIT, TQM, and Six Sigma) is implemented significantly more in all of the advanced manufacturing environments examined. In addition, the sample firms’ self-evaluated ratings as a world-class manufacturer are significantly higher in JIT, TQM, and Six Sigma environments. The contextual environments for firms practicing advanced manufacturing techniques are very similar. These firms are larger in size, and the respondents perceive their firms to be leaders in process and product development. They also report significantly more support from top management in initiating and implementing change, as well as providing necessary training in new production strategies. All of the firms appear to be “somewhat satisfied” with their MAS. No evidence exists that user satisfaction is dependent upon the type of manufacturing environment in which the MAS is operating.

ANOVA Results for CAP and the Level of JIT Implemented To more broadly examine the JIT manufacturing environment, comparisons were made between CAP and the level of implementation of specific JIT practices. As a result of the factor analysis, these practices were identified as a manufacturing component, a quality component, and a uniquely JIT component. In addition, the three factors were combined to represent a comprehensive JIT component.8 Each of the CAP shows some significant relationship with the level of implementation of at least one of the JIT components, except for ABC. However, the use of absorption and standard costing systems is contrary to the expected relationships. Those sample firms that have implemented higher levels of JIT practices are using a standard costing system, and absorption costing more than direct (variable) costing for internal decision making. Although the differences of the means shown in Table 2 between the JIT and non-JIT firms for the STDRD and ABSORP variables are not significant, the means of the preferred cost accounting practice in a JIT environment is higher for the non-JIT firms. Similar to the JIT/non-JIT results, the accounting practices of VCA and LCC consistently demonstrate the most significant differences between lower and higher levels of JIT implementation. The results also reinforce the relationship of JIT to the other advanced manufacturing practices of TQM and Six Sigma (Table 8).

104


Table 8. ANOVA Comparison of CAP Means with Level of JIT Implementation. n = 121 Full sample means Cost accounting practices (CAP) Process or backflush costing Job-order costing Backflush costing Yes No Direct (variable) costing Absorption costing

JIT Quality

JIT Unique

JIT Combined

3.241

4.889

3.560

3.897

PROCESS

3.405 2.938*

4.932 4.810

3.590 3.466

3.976 3.739

BCKFL

3.615 3.111* 3.083 3.500

5.241 4.781** 4.725 5.211**

4.482 3.258*** 3.364 3.886

4.099 3.422*** 3.718 4.205**

ABSORP

JIT Manufacturing

Standard costing Yes No

STDRD

3.306 2.444*

4.940 4.167*

3.638 2.438*

3.973 2.840**

Balanced scorecard Yes No

BSC

3.719 3.156

5.250 4.903

4.125 3.298*

4.348 3.801*

Target costing Yes No

TGCST

3.301 2.816

4.933 4.842

3.681 3.079

3.990 3.579*

Variance analysis Yes No

VARAN

3.347 2.303**

4.885 4.900

3.617 3.227

3.947 3.439

Activity based costing Yes No

ABC

3.333 3.085

4.975 4.733

3.441 3.739

3.927 3.850

Economic value added Yes No

EVA

3.581 3.062*

5.303 4.777**

3.956 3.491

4.290 3.789**

Value chain analysis Yes No

VCA

3.962 2.936***

5.317 4.776**

4.450 3.339***

4.579 3.680***

Life cycle costing Yes No

LCC

3.971 3.064***

5.386 4.847*

4.250 3.562*

4.537 3.829***

JIT

3.706 2.865***

5.226 4.609***

4.480 2.849***

4.458 3.445***

Production processes Just-in-time Yes No


105

Table 8. (Continued ) n = 121

JIT Manufacturing

JIT Quality

JIT Unique

JIT Combined

Total quality management Yes TQM No

3.515 2.962**

5.219 4.558***

3.711 3.408

4.148 3.645**

Six Sigma Yes No

3.421 3.136

5.088 4.776

4.150 3.263**

4.222 3.727**

6-SIG

Notes: Respondents were asked to indicate the extent to which their firm had implemented the individual JIT elements. Possible responses were: No intention = 1; Considering = 2; Beginning = 3; Partially = 4; Substantially = 5; Fully = 6. ∗ p < 0.05. ∗∗ p < 0.01. ∗∗∗ p < 0.001.

DISCUSSION OF THE RESULTS This study seeks to determine if firms that adopt innovative manufacturing technologies also adopt complementary, innovative internal cost accounting practices. The results of a survey of New Zealand manufacturing firms (Durden et al., 1999) indicated no differences between the CAP of JIT and non-JIT firms. Moreover, Yasin et al. (1997) reported that management accountants had the least impact of all the plant departments examined in encouraging JIT implementation. Similar to the findings of Lillis (2002), our research shows that world-class manufacturers continue to use traditional CAP practices, but unlike their competitors, they are supplementing and expanding their MAS. Despite criticisms, standard costing and variance analysis still are widely used in advanced manufacturing environments. In their recent study on the design of control systems for a changing manufacturing system, Banker et al. (2000) hypothesized that the monitoring of direct labor variances would be negatively associated with the adoption of new manufacturing processes. However, their research did not support this hypothesis. Their sample plants continued to use direct labor standards and variance analysis while adopting advanced manufacturing techniques. Drury (1999) points out that standard costing to monitor trends can be consistent with the lean concept of continuous improvement. If used appropriately, it continues to be an effective measurement tool. Cooper (1996) found that variance analysis often was used in relatively traditional ways for operational control in JIT firms. In addition, some Japanese firms embedded Kaisen improvements into their standards and used

106


variance analysis to monitor Kaisen objectives. Cooper also found that despite Japanese firms’ strong emphasis on cost management, their cost systems were relatively traditional, rather than technically advanced. Traditional accounting techniques may be more inadequate than outdated, and need to be supplemented with new planning and control tools, such as target costing, life-cycle costing, and value chain analysis. The research results suggest that advanced manufacturing environments require information beyond what is provided by traditional production-costing methods. For proper planning and control, firms need to understand the full gamut of costs from product inception to disposal, including costs for research and development, change initiatives, and marketing. Advanced manufacturing systems, such as JIT, must be highly flexible to respond to customer demand. In order to effectively operate a lean, pull system that focuses on continuous improvement, firms must integrate and coordinate their operations with both their suppliers and customers and have assurance that similar quality initiatives will be exercised and supportive all along the value chain (Kalagnanam & Lindsay, 1998). In evaluating their financial success, advanced manufacturing firms want to examine the net added value to their firms using measurement techniques such as EVA and make strategic decisions accordingly. Life-cycle costing and value chain analysis perspectives are supported by the tenets of target costing. According to this study’s results, target costing is being used for new product development in TQM and Six Sigma environments. Target-costing principles, which assist in the planning and control of costs during the initial stages of product and process design, are considered an important Japanese accounting tool (Ansari et al., 1997; Sakurai & Scarbrough, 1997), as demonstrated by Cooper’s (1996) study of Japanese lean manufacturers. An interesting result is the strong correlation between the use of Six Sigma and the balanced scorecard. Six Sigma is a data-driven process that is highly tied to the bottom line, but it is also much broader in its application than the measurement of profitability alone. It supports continuous improvement through management, financial, and methodological tools that improve both products and processes (Voelkel, 2002). A balanced scorecard analysis would assist in evaluation of Six Sigma efforts by providing information not only about profitability measures, but also about internal manufacturing operations, customer satisfaction, and employee contributions and retention. As earlier studies have reported, top management support is key to making changes and successfully implementing lean strategies (Ahmed et al., 1991; Ansari & Modarress, 1986; Celley et al., 1986; Im, 1989; Willis & Suter, 1989). The survey results support this idea, as those sample firms that have adopted JIT, TQM, and/or Six Sigma have a significantly higher level of support from top management for change initiatives and new strategies compared to firms that


107

have not implemented these techniques. In addition, larger firms that have more resources to allocate to these practices are more successful in adopting them. An innovative environment that would provide flexibility and empowerment also appears to facilitate the adoption of advanced production processes. The results indicate that world-class manufacturers integrate a combination of advanced manufacturing techniques. Six Sigma is implemented to support the use of TQM and JIT. The results support evidence argued by Vuppalapati et al. (1995) that JIT and TQM should be viewed as integrated, rather than isolated strategies. The two are complementary in their focus, “making production more efficient and effective through continuous improvement and elimination of waste strategies.” Research further shows that the integration of TQM, JIT, and TPM leads to higher performance than does the exclusive implementation of each technique (Cua et al., 2001).

Research Limitations Specific research limitations might reduce the generalizability and applicability of the findings. As in all survey research, a necessary assumption in data collection is that the respondents had sufficient knowledge to answer the items and that they answered the questions conscientiously and truthfully. Respondents might have been unfamiliar with some of the questionnaire terms, and reluctant to take the necessary time to examine the attached glossary explaining the terminology. In addition, the sample firms are a subset of a previous research sample. Thus, the sampling approach was not random. The diversion from random sample selection may limit the generalizability of the results to other U.S. manufacturing firms.

SUMMARY Strategic use of information resources help in customer service, product differentiation, and cost competition (Narasimhan & Jayaram, 1998). Although the MAS should support organizational operations and strategies, evidence in this study, as in similar research (Banker et al., 2000; Clinton & Hsu, 1997; Durden et al., 1999; Yasin et al., 1997), indicates that CAP are not changing substantially to support lean practices. However, the results demonstrate that world-class manufacturing firms are integrating additional, non-traditional information techniques into their MAS, such as EVA, life-cycle costing, and value chain analysis. In a survey of 670 UK firms, Bright et al. (1992) cited the following reasons given for the lack of substantial change in cost accounting systems: (1) The

108


benefits do not appear to outweigh the costs. (2) New techniques have not proven to provide better information. (3) There is already too much change and change is difficult; they are comfortable with what they have. (4) The current system is adequate; it just needs to be better utilized. (5) There is a lack of integration between the factory and accounting information; thus, non-accountants do not use accounting information for decision making. Also, there is a low expectation of what accountants can offer. “If companies have simplistic cost systems, it may well be that there is no need for a better system or that an existing need has not yet been recognized. In most cases, systems will be improved before poor cost information leads to consistently poor decisions” (Holzer & Norreklit, 1991, p. 12). When surviving firms retain the same procedures over time, it is implicit that the benefits derived therefrom exceed the costs. Moreover, the MAS has many uses. It is plausible that control aspects of the system yield benefits that are overlooked by those who decry the system’s inadequacy for decision making. Further research is needed to determine the extent to which the implementation of advanced manufacturing production processes motivates changes in a firm’s internal accounting practices, as well as the impact of more extensive CAP on firm profitability and competitiveness. While this study provides evidence that some change is taking place in the CAP of WCM firms, it appears to be from an expansion of traditional information techniques, rather than from their replacement. Advanced manufacturing firms must be experiencing benefits from the continued use of existing internal accounting measures. The alleged limitations of these methods might stem more from their application and not from the methods themselves. Rather than abandon practices that have endured through decades, what is needed is greater acceptance of management accounting’s role to support organizational strategy.

NOTES 1. The survey instrument was either available over the Internet, or hard copies were faxed or mailed. The executives contacted were asked to choose which alternative they preferred for responding to the questionnaire. Initially, 128 of the sample firms were contacted via the Internet of which 97 responded. Forty-two were faxed and 12 were mailed the questionnaire initially, with 17 and 6 respondents, respectively. 2. To check for non-response bias, the analyses were performed on the late responders. No significant differences were found in the results. 3. The industry distribution for the respondent firms is similar to the total sample industry distribution. Sixty-two percent of the firms sampled were from these same three industries: industrial machinery, electronics, and instrumentation. 4. A definition of these terms was supplied with the questionnaire and can be found in Fullerton et al. (2003).


109

5. Total quality control is represented by two questions on the survey: one is related to process quality and the other to product quality. 6. All of the 11 elements loaded greater than 0.50 onto one of the three constructs except for number 11, asking about the use of “quality circles.” Thus, this question was eliminated from further testing representing JIT. 7. EVA is a registered trademark of Stern Stewart & Company. 8. Cronbach’s alpha (1951) for the combined measure and the three individual JIT factors exceed the standard of 0.70 for established constructs (Nunnally, 1978).

ACKNOWLEDGMENTS This research was made possible through a Summer Research Grant provided by Utah State University. We gratefully acknowledge comments received from participants at the AIMA Conference on Management Accounting Research in Monterey, California (May 2003).

REFERENCES Abernethy, M. A., Lillis, A. M., Brownell, P., & Carter, P. (2001). Product diversity and costing system design choice: Field study evidence. Management Accounting Research, 12(1), 1–20. Ahmed, N. U., Runc, E. A., & Montagno, R. V. (1991). A comparative study of U.S. manufacturing firms at various stages of just-in-time implementation. International Journal of Production Research, 29(4), 787–802. Anderson, S. W. (1995). A framework for assessing cost management system changes: The case of activity-based costing at General Motors, 1986–1993. Journal of Management Accounting Research (Fall), 1–51. Ansari, A., & Modarress, B. (1986). Just-in-time purchasing: Problems and solutions. Journal of Purchasing and Materials Management (Summer), 11–15. Ansari, S. L., Bell, J. E., & CAM-I Target Cost Core Group (1997). Target costing: The next frontier in strategic cost management. Burr Ridge, IL: Irwin. Atkinson, A. A., Balakrishnan, R., Booth, P., Cote, J. M., Groot, T., Malmi, T., Roberts, H., Uliana, E., & Wu, A. (1997). New directions in management accounting research. Journal of Management Accounting Research, 9, 80–108. Baker, W. M., Fry, T. D., & Karwan, K. (1994). The rise and fall of time-based manufacturing. Management Accounting (June), 56–59. Banker, R., Potter, G., & Schroeder, R. (1993a). Reporting manufacturing performance measures to workers: An empirical study. Journal of Management Accounting Research, 5(Fall), 33–53. Banker, R., Potter, G., & Schroeder, R. (1993b). Manufacturing performance reporting for continuous quality improvement. Management International Review, 33(Special Issue), 69–85. Banker, R., Potter, G., & Schroeder, R. (2000). New manufacturing practices and the design of control systems. Proceedings of the Management Accounting Research Conference. Barney, J. (1986). Strategic factor markets: Expectation, luck, and business strategy. Management Science (October), 1231–1241.

110


Bragg, S. M. (1996). Just-in-time accounting: How to decrease costs and increase efficiency. New York: Wiley. Bright, J., Davies, R. E., Downes, C. A., & Sweeting, R. C. (1992). The deployment of costing techniques and practices: A UK study. Management Accounting Research, 3, 201–211. Callen, J. L., Fader, C., & Morel, M. (2002). The performance consequences of in-house productivity measures: An empirical analysis. Working Paper, University of Toronto, Ontario. Celley, A. F., Clegg, W. H., Smith, A. W., & Vonderembse, M. A. (1986). Implementation of JIT in the United States. Journal of Purchasing and Materials Management (Winter), 9–15. Chenhall, R. H., & Langfield-Smith, K. (1998a). Adoption and benefits of management accounting practices: An Australian study. Management Accounting Research, 9, 1–19. Chenhall, R. H., & Langfield-Smith, K. (1998b). Factors influencing the role of management accounting in the development of performance measures within organizational change programs. Management Accounting Research, 9, 361–386. Clinton, B. D., & Hsu, K. C. (1997). JIT and the balanced scorecard: Linking manufacturing control to management control. Management Accounting, 79(September), 18–24. Cobb, I. (1992). JIT and the management accountant. Management Accounting, 70(February), 42–49. Cooper, R. (1994). The role of activity-based systems in supporting the transition to the lean enterprise. Advances in Management Accounting, 1–23. Cooper, R. (1995). When lean enterprises collide. Boston, MA: Harvard Business School Press. Cooper, R. (1996). Costing techniques to support corporate strategy: Evidence from Japan. Management Accounting Research, 7, 219–246. Cooper, R., & Kaplan, R. S. (1991). The design of cost management systems. Englewood Cliffs, NJ: Prentice-Hall. Cronbach, L. J. (1951). Construct validity in psychological tests. Psychological Bulletin, 297–334. Cua, K. O., McKone, K. E., & Schroeder, R. G. (2001). Relationships between implementation of TQM, JIT, and TPM manufacturing performance. Journal of Operations Management, 19(6), 675–694. Daniel, S. J., & Reitsperger, W. D. (1991). Linking quality strategy with management control systems: Empirical evidence from Japanese industry. Accounting, Organizations and Society, 16(7), 601–618. Daniel, S. J., & Reitsperger, W. D. (1996). Linking JIT strategies and control systems: A comparison of the United States and Japan. The International Executive, 38(January/February), 95–121. Davy, J., White, R., Merritt, N., & Gritzmacher, K. (1992). A derivation of the underlying constructs of just-in-time management systems. Academy of Management Journal, 35(3), 653–670. Dean, J. W., Jr., & Snell, S. A. (1996). The strategic use of integrated manufacturing: An empirical examination. Strategic Management Journal, 17(June), 459–480. Drury, C. (1999). Standard costing: A technique at variance with modern management? Management Accounting, 77(10), 56–58. Durden, C. H., Hassel, L. G., & Upton, D. R. (1999). Cost accounting and performance measurement in a just-in-time production environment. Asia Pacific Journal of Management, 16, 111–125. Ellis, K. (2001). Mastering six sigma. Training, 38(12), 30–35. Ferrara, W. L. (1990). The new cost/management accounting: More questions than answers. Management Accounting (October), 48–52. Fisher, J. (1992). Use of non-financial performance measures. Cost Management (Spring), 31–38. Flynn, B. B., Sakakibara, S., & Schroeder, R. G. (1995). Relationship between JIT and TQM: Practices and performance. Academy of Management Journal, 38(October), 1325–1353. Frigo, M. L., & Litman, J. (2001). What is strategic management? Strategic Finance, 83(6), 8–10.


111

Fullerton, R. R., & McWatters, C. S. (2001). The production performance benefits from JIT. Journal of Operations Management, 19, 81–96. Fullerton, R. R., & McWatters, C. S. (2002). The role of performance measures and incentive systems in relation to the degree of JIT implementation. Accounting, Organizations and Society, 27(8), 711–735. Fullerton, R. R., McWatters, C. S., & Fawson, C. (2003). An examination of the relationships between JIT and financial performance. Journal of Operations Management, 21, 383–404. Gagne, M. L., & Discenza, R. (1992). Accurate product costing in a JIT environment. International Journal of Purchasing and Materials Management (Fall), 28–31. Green F. B., Amenkhienan, F., & Johnson, G. (1992). Performance measures and JIT. Management Accounting (October), 32–36. Haldane, G. (1998). Accounting for change. Accountancy, 122(December), 64–65. Harris, E. (1990). The impact of JIT production on product costing information systems. Production and Inventory Management Journal, 31(1), 44–48. Hayes, R. H., & Wheelwright, S. C. (1984). Restoring our competitive edge. New York: Collier Macmillan. Hedin, S. R., & Russell, G. R. (1992). JIT implementation: Interaction between the production and cost-accounting functions. Production and Inventory Management Journal (Third Quarter), 68–73. Hendricks, C. A., & Kelbaugh, R. L. (1998). Implementing six sigma at GE. The Journal for Quality & Participation (July/August), 48–53. Hendricks, J. A. (1994). Performance measures for a JIT manufacturer. IIE Solutions, 26(January), 26–36. Holzer, H. P., & Norreklit, H. (1991). Some thoughts on cost accounting developments in the United States. Management Accounting Research, 2, 3–13. Hoque, A., & James, W. (2000). Linking balanced scorecard measures to size and market factors: Impact on organizational performance. Journal of Management Accounting Research, 12, 1–17. Horngren, C. T., Foster, G., Datar, S. M., & Teall, H. D. (1997). Cost accounting. Scarborough, Ont.: Prentice-Hall. Howell, R. A., & Soucy, S. R. (1987). Cost accounting in the new manufacturing environment. Management Accounting (August), 42–48. Im, J. H. (1989). How does kanban work in American companies? Production and Inventory Management Journal (Fourth Quarter), 22–24. Ittner, C. D., & Larcker, D. F. (1995). Total quality management and the choice of information and reward systems. Journal of Accounting Research (Supplement), 1–34. Ittner, C. D., & Larcker, D. F. (1997). The performance effects of process management techniques. Management Science, 43(4), 522–534. Johnson, H. T. (1992). Relevance regained: From top-down control to bottom-up empowerment. New York: Free Press. Jones, D. T., Hines, P., & Rich, N. (1997). Lean logistics. International Journal of Physical Distribution and Logistics Management, 27(3/4), 153–163. Kalagnanam, S. S., & Lindsay, R. M. (1998). The use of organic models of control in JIT firms: Generalising woodward’s findings to modern manufacturing practices. Accounting, Organizations and Society, 24(1), 1–30. Kaplan, R. S. (1983). Measuring manufacturing performance: A new challenge for managerial accounting research. The Accounting Review (October), 686–705.

112


Kaplan, R. S. (1984). Yesterday’s accounting undermines production. Harvard Business Review, 62, 96–102. Koufteros, X. A., Vonderembse, M. A., & Doll, W. J. (1998). Developing measures of time-based manufacturing. Journal of Operations Management, 16, 21–41. Kristensen, K., Dahlgaard, J. J., Kanji, G. K., & Juhl, H. J. (1999). Some consequences of just-in-time: Results from a comparison between the Nordic countries and east Asia. Total Quality Management, 10(1), 61–71. Lillis, A. M. (2002). Managing multiple dimensions of manufacturing performance – An exploratory study. Accounting, Organizations and Society, 27(6), 497–529. Linderman, K., Schroeder, R. G., Zaheer, A., & Choo, A. S. (2003). Six Sigma: A goal-theoretic perspective. Journal of Operations Management, 21, 193–203. Mackey, J., & Hughes, V. H. (1993). Decision-focused costing at Kenco. Management Accounting, 74(11), 22–32. McNair, C. J., Lynch, R. L., & Cross, K. F. (1990). Do financial and non-financial performance measures have to agree? Management Accounting (November), 28–36. McNair, C. J., Mosconi, W., & Norris, T. (1989). Beyond the bottom line: Measuring world class performance. Homewood, IL: Irwin. McWatters, C. S., Morse, D. C., & Zimmerman, J. L. (2001). Management accounting: Analysis and interpretation (2nd ed.). New York: McGraw-Hill. Mehra, S., & Inman, R. A. (1992). Determining the critical elements of just-in-time implementation. Decision Sciences, 23, 160–174. Milgrom, P., & Roberts, J. (1995). Complementarities and fit: Strategy, structure and organizational change in manufacturing. Journal of Accounting and Economics, 19, 179–208. Moshavi, S. (1990). Well made in America: Lessons from Harley-Davidson on being the best. New York: McGraw-Hill. Najarian, G. (1993). Performance measurement: Measure the right things. Manufacturing Systems, 11(September), 54–58. Narasimhan, R., & Jayaram, J. (1998). An empirical investigation of the antecedents and consequences of manufacturing goal achievement in North American, European, and Pan Pacific firms. Journal of Operations Management, 16, 159–179. Nunnally, J. (1978). Psychometric theory. New York: McGraw-Hill. Patell, J. M. (1987). Cost accounting, process control, and product design: A case study of the Hewlett-Packard personal office computer division. The Accounting Review, 62(4), 808–839. Peavey, D. E. (1990). Battle at the GAAP? It’s time for a change. Management Accounting (February), 31–35. Pettit, J. (2000). EVA and production strategy. Industrial Management (November/December), 6–13. Rogers, B. (1998). Paperwork with a payoff. Manufacturing Engineering, 121(3), 16. Sakakibara, S., Flynn, B. B., & Schroeder, R. G. (1993). A framework and measurement instrument for just-in-time manufacturing. Production and Operations Management, 2(3), 177–194. Sakurai, M., & Scarbrough, D. P. (1997). Japanese cost management. Menlo Park, CA: Crisp Publications. Scarlett, B. (1996). In defence of management accounting applications. Management Accounting, 74(1), 46–52. Schonberger, R. J. (1986). World class manufacturing casebook: The lesson of simplicity applied. New York: Free Press. Shields, M. D., & Young, S. M. (1989). A behavioral model for implementing cost management systems. Journal of Cost Management (Winter), 17–27.


113

Sillince, J. A. A., & Sykes, G. M. H. (1995). The role of accountants in improving manufacturing technology. Management Accounting Research, 6(June), 103–124. Spencer, M. S., & Guide, V. D. (1995). An exploration of the components of JIT. International Journal of Operations and Production Management, 15(5), 72–83. Swenson, D. W., & Cassidy, J. (1993). The effect of JIT on management accounting. Cost Management (Spring), 39–47. Thorne, K., & Smith, M. (1997). Synchronous manufacturing: Back to basics. Management Accounting, 75(11), 58–60. Tully, S. (1993). The real key to creating wealth. Fortune (September 20), 38–50. Voelkel, J. G. (2002). Something’s missing: An education in statistical methods will make employees more valuable to Six Sigma corporations. Quality Progress (May), 98–101. Voss, C. A. (1995). Alternative paradigms for manufacturing strategy. International Journal of Production Management, 15(4), 5–16. Vuppalapati, K., Ahire, S. L., & Gupta, R. (1995). JIT and TQM: A case for joint implementation. International Journal of Operations and Production Management, 15(5), 84–94. Welfe, B., & Keltyka, P. (2000). Global competition: The new challenge for management accountants. The Ohio CPA Journal (January–March), 30–36. White, R. E. (1993). An empirical assessment of JIT in U.S. manufacturers. Production and Inventory Management, 34(Second Quarter), 38–42. White, R. E., Pearson, J. N., & Wilson, J. R. (1999). JIT manufacturing: A survey of implementations in small and large U.S. manufacturers. Management Science, 45(January), 1–15. White, R. E., & Prybutok, V. (2001). The relationship between JIT practices and type of production system. Omega, 29, 113–124. White, R. E., & Ruch, W. A. (1990). The composition and scope of JIT. Operations Management Review, 7(3–4), 9–18. Willis, T. H., & Suter, W. C., Jr. (1989). The five M’s of manufacturing: A JIT conversion life cycle. Production and Inventory Management Journal (January), 53–56. Wisner, J. D., & Fawcett, S. E. (1991). Linking firm strategy to operating decisions through performance measurement. Production and Inventory Management Journal (Third Quarter), 5–11. Witt, E. (2002). Achieving six sigma logistics. Material Handling Management, 57(5), 10–13. Yasin, M. M., Small, M., & Wafa, M. (1997). An empirical investigation of JIT effectiveness: An organizational perspective. Omega, 25(4), 461–471. Zimmerman, J. L. (2003). Accounting for decision making and control (4th ed.). Burr Ridge, IL: McGraw-Hill.

THE INTERACTION EFFECTS OF LEAN PRODUCTION MANUFACTURING PRACTICES, COMPENSATION, AND INFORMATION SYSTEMS ON PRODUCTION COSTS: A RECURSIVE PARTITIONING MODEL Hian Chye Koh, Khim Ling Sim and Larry N. Killough ABSTRACT The study re-examines if lean production manufacturing practices (i.e. TQM and JIT) interact with the compensation system (incentive vs. fixed compensation plans) and information system (i.e. attention directing goals and performance feedback) to reduce production costs (in terms of manufacturing and warranty costs) using a recursive partitioning model. Decision trees (i.e. recursive partitioning algorithm using Chi-square Automatic Interaction Detection or CHAID) are constructed on data from 77 U.S. manufacturing firms in the electronics industry. Overall, the “decision tree” results show significant interaction effects. In particular, the study found that better manufacturing performance (i.e. lower production costs) can be achieved when lean production manufacturing practices such as TQM and JIT are used along with incentive compensation plans. Also, synergies do


115

116

HIAN CHYE KOH ET AL.

result from combining TQM/JIT with more frequent performance feedback along with attention directing goals. These findings suggest that if organisational infrastructure and management control systems are not aligned with manufacturing practices, then the potential benefits of lean manufacturing (i.e. TQM and JIT) may not be fully realised.

INTRODUCTION In a US$5 million 5-year study on the future of the automobile, Womack et al. (1990) made an excellent summary of manufacturing practices since the late 1800s. At the beginning of the industrial age (and in fact even before that), manufacturing was dominated by craft production. It had the following characteristics: (1) highly skilled craftspeople; (2) very decentralised organisations; (3) generalpurpose machine tools; and (4) very low production volume. Although craft production had worked very well then, it had several drawbacks. These included high production costs (regardless of volume) and low consistency and reliability. The first revolution in manufacturing practices came after World War I in the early 1900s when Henry Ford introduced new manufacturing practices that could reduce production costs drastically while increasing product quality. This innovative system of manufacturing practices was called mass production (vis-à-vis craft production). The defining feature of mass production was the complete interchangeability of parts and the simplicity of attaching them to each other. This led to the division of labour in the production process (to repetitive single tasks) and the construction of moving assembly lines. Mass production spurred a remarkable increase in productivity (and a corresponding remarkable decrease in cost per unit output), a drastic improvement in product quality and a significant reduction in capital requirements. Mass production was eventually adopted in almost every industrial activity around the world. Problems with mass production started to become prominent in the mid-1900s. The minute division of labour removed the career path of workers and resulted in dysfunctions (e.g. reduced job satisfaction). Also, mass production led to standardised products that were not suited to all world markets. The production system was inflexibility and time-consuming and expensive to change. Finally, intense competition and unfavourable macro-economic developments further eroded the advantages of mass production. While mass production was declining in the mid-1900s, the second revolution in manufacturing practices took root in Toyota in Japan. The Toyota Production System, also referred to by Womack et al. (1990) as lean production, was established

The Interaction Effects of Lean Production Manufacturing Practices

117

by the early 1960s and since then has been incorporated by many companies and industries world-wide. Lean production brings together the advantages of craft production and mass production by avoiding the former’s high cost and the latter’s rigidity. It employs teams of multi-skilled workers at all levels of the organisation and uses highly flexible and increasingly automated machines to produce outputs in enormous variety as well as with high quality. Specifically, lean production is characterised by the following focus: (1) cost reductions; (2) zero defects; (3) zero inventories; (4) product variety; and (5) highly skilled, motivated and empowered workers. Lean production manufacturing practices have often been referred to by other names, such as total quality management (TQM) and just-in-time (JIT). In particular, these manufacturing practices (i.e. TQM and JIT) have been used by manufacturing firms striving for continual improvement. To date, however, mixed results have been reported – while some firms have excelled because of their use of TQM and JIT (e.g. Xerox and Motorola), other firms do not seem to have improved their manufacturing performance (see, for example, Harari, 1993; Ittner & Larcker, 1995). Although there is an expanding literature on lean manufacturing such as TQM or JIT implementation, there is little empirical evidence that provides reasons for these mixed results (Powell, 1995). Since the middle part of the 1980s, there is a growing interest in how management control systems could be modified to tailor to the needs of manufacturing strategies (Buffa, 1984; Hayes et al., 1988; Kaplan, 1990; Schonberger, 1986). Empirical evidence shows that higher organisational performance is often the result of a match between an organisation’s environment, strategy and internal structures or systems. By and large, most studies on management control focus on senior management performance or overall performance at strategic business unit levels (Govindarajan, 1988; Govindarajan & Gupta, 1985; Ittner & Larcker, 1997). Given that successes in manufacturing strategies are often influenced by activities at the shop floor or operational level, empirical studies linking management control policies to manufacturing performance at the operational level may provide useful information on the mixed findings related to lean manufacturing practices. Accordingly, the objective of this study is to examine whether manufacturing firms using lean production manufacturing practices such as TQM and/or JIT achieve higher manufacturing performance (i.e. lower production costs) when they accompany these practices with contemporary operational controls. More formally, the study investigates the performance effect (i.e. reduction in production costs) of the match between lean production manufacturing practices and compensation and information systems using a recursive partitioning model.

118


RESEARCH FRAMEWORK As mentioned earlier, TQM and JIT feature prominently in lean production. TQM focuses on the continual improvement of manufacturing efficiency by eliminating waste, scrap and rework while improving quality, developing skills and reducing costs. Along similar lines, JIT emphasises manufacturing improvements via reducing set-up and cycle times, lot sizes and inventories. These lean production manufacturing practices require workers who are highly skilled, motivated and empowered. In particular, workers are made responsible for improving manufacturing capabilities and product and process quality (Siegel et al., 1997), performing a variety of activities, and detecting non-conforming items. TQM and JIT implementation are expected to improve manufacturing performance. In addition, worker empowerment (which is an important part of TQM and JIT) is expected to indirectly improve manufacturing performance via greater intrinsic motivation (Hackman & Wageman, 1995). Among other things, improved manufacturing performance translates to reduction in production costs. Wruck and Jensen (1994) suggest that effective TQM implementation requires major changes in organisational infrastructure such as the systems for allocating decision rights, performance feedback and reward/punishment. Kalagananam and Lindsay (1998), on the other hand, suggest that a fully developed JIT system represent a significant departure from the traditional mass production systems. They advocate that manufacturing firms adopting JIT must abandon a mechanistic management control system and adopt an organic model of control. Taken together, the literature suggests that management control systems in lean manufacturing practices should be different from those under the traditional mass production system. Management control systems have been described as processes or tools for influencing behaviour towards attainment of organisational goals or objectives. A control system performs its function by controlling the flow of information, establishing criteria for evaluation, and designing appropriate rewards and punishment (Birnberg & Snodgrass, 1988; Flamhaultz et al., 1985). As such, this study focuses on the compensation system (incentive vs. fixed compensation plans) and the information system (i.e. attention-directing goals and manufacturing performance feedback).

Compensation System Much research in management control is based on economic models that assume that workers are rational, self-interested and utility-maximising agents.


119

Consequently, without monitoring and sanctions, self-interested workers will be risk averse and they will also exhibit shirking behaviour. Prior research findings indicate that incentives are a major motivator in this context (Fama & Jensen, 1983; Holmstrom, 1979; Jensen & Meckling, 1976). More generally, designing compensation systems to match the needs of lean manufacturing practices is consistent with the strategic compensation literature (see, for example, Gomez-Mejia & Balkin, 1992; Milkovich, 1988). While incentive compensation plans by themselves can be effective in enhancing manufacturing performance (Coopers & Lybrand, 1992; MacDuffie, 1995), Ichniowski et al. (1997) reported that workers’ performance were substantially higher when incentive plans were also coupled with supportive work practices. More recently, Sim and Killough (1998) found synergistic effects between TQM/JIT implementation and the use of incentive compensation plans. Given the above, it is hypothesised in this study that although incentive compensation plans can be effective independently of TQM and JIT implementation, it is the match between lean production manufacturing practices (i.e. TQM and JIT) and the compensation system that leads to higher synergistic manufacturing performance (e.g. lower production costs). In other words: H1. Lean production manufacturing practices (i.e. TQM/JIT) interact with the use of compensation plans to enhance manufacturing performance (i.e. to lower production costs).

Information System (Goals and Feedback) Locke and Latham (1990) found that goals positively influence the attention, effort and persistence of workers. This finding is consistent across many studies (e.g. Latham & Lee, 1986; Locke et al., 1981). Thus, if an organisation wants its workers to achieve particular goals, then prior research findings suggest that the presence of such goals can motivate workers to accomplish them. In practice, to help workers achieve better manufacturing performance, attention-directing goals or targets are often provided via the firm’s information system. Examples of such goals include customer satisfaction and complaints, on-time delivery, defect rate and sales returns, cycle time performance . . . etc. From a learning standpoint, providing performance feedback helps workers develop effective task strategies. Alternatively, feedback which shows that performance is below the target can increase the motivation to work harder (Locke & Latham, 1990). As a result, the combination of both goals and feedback often lead to better performance (Erez, 1977). Using a framework of management

120


controls, Daniel and Reitsperger (1991) concluded that manufacturing plants adhering to zero defect strategies are more likely to provide goals and frequent manufacturing performance feedback than manufacturing plants supporting a more traditional model. Similarly, Banker et al. (1993) found that the reporting of manufacturing performance feedback to workers is positively associated with TQM and JIT practices. Both studies, however, did not examine the impact of goals or performance feedback on manufacturing performance. The strategic literature shows that for manufacturing strategies to be effectively implemented, they must be integrated with day-to-day operational planning and control mechanisms. Unless this coupling exists, the strategic plan will become irrelevant with time (Gage, 1982). Thus, it can be argued that to improve manufacturing performance, it is imperative that the firm’s information system provides attention-directing goals and manufacturing performance feedback to workers. Conversely, it is expected that attention-directing goals and performance feedback will enhance manufacturing performance. In this aspect, Sarkar (1997) provided evidence that process improvement in quality is enhanced when information sharing is encouraged in the work place. It is further hypothesised in this study that although goals or performance feedback can be effective independently of TQM and JIT implementation, it is the match between lean production manufacturing practices (i.e. TQM and JIT) and the information system (i.e. attention-directing goals and manufacturing performance feedback) that leads to higher synergistic manufacturing performance in terms of reduction in production costs (see also Chenhall, 1997; Ittner & Larcker, 1995; Milgrom & Roberts, 1990, 1995). More formally, the research hypothesis in the study can be stated as follows: H2. Lean production manufacturing practices (i.e. TQM/JIT) interact with the use of information system (i.e. attention-directing goals and/or manufacturing performance feedback) to enhance manufacturing performance (i.e. to lower production costs).

MATERIALS AND METHODS To investigate the interaction effects between TQM/JIT and compensation and information systems (i.e. goal setting and feedback) on manufacturing performance, the following research materials and methods are employed. Sample Selection The electronics industry (SIC Code 36) was chosen as the primary industry for the study for the following reasons. Unlike the TQM concept, the JIT concept


121

works well mainly in repetitive manufacturing such as automobiles, appliances and electronic goods. Balakrishnan et al. (1996) reported that 68% of JIT firms (i.e. those that have adopted the JIT concept for a substantial portion of their operations) clustered within the three SIC codes 35, 36 and 38. The sample for the study was drawn from the electronics industry (SIC code 36) since it has the highest percentage of JIT firms. Questionnaire The questionnaire solicited information pertaining to manufacturing practices, workplace practices, as well as several aspects of manufacturing performance. Two stages were involved in a pilot test of the questionnaire. First, three production engineers from a semiconductor plant were asked to fill out the questionnaire. Since information provided was based on the same plant, the responses were compared, and found to be consistent. Next, the questionnaire was reviewed by four experts in the area of process improvement to check for relevancy or possible ambiguity in the instrument. Feedback from the pilot test resulted in no major changes, except for rephrasing of some statements. Appendix provides detailed information about the questionnaire. Letters requesting participation in the research study were sent to the directors of manufacturing of 1,500 randomly selected firms located within the United States with annual sales of ten million dollars and above. A total of 126 manufacturing plants agreed to participate in the study and three firms wished to review the research questionnaire prior to making a commitment to participate. As a result, 129 research questionnaires were mailed. About 50% of the plants replied within four weeks. Six weeks after the initial mailing of the research questionnaires, a status report together with a reminder was sent to all 129 plants. In total, 77 usable responses were received, giving a response rate of 59.7%. (For the analysis involving warranty costs, only 76 responses are used because of missing data.) The data collected pertain to a specific SBU and not the entire company. Dependent Variable Measures An effective management control system starts with defining the critical performance measures. Firms using lean production manufacturing practices such as TQM and JIT are expected to experience improved efficiencies such as lower manufacturing costs. Also, they are expected to have improved product quality and hence lower warranty costs. Thus, changes in manufacturing costs and warranty costs (collectively termed as production costs here) were used as dependent variables in the study. Respondents were requested to indicate the changes in

122


these production costs in the last three years, anchoring on a scale of 1–5, with 1 denoting “decrease tremendously,” 3 “no change” and 5 “increase tremendously.” It is noted that this study focuses only on one key aspect of lean production, namely, the reduction in production costs. This sole key aspect, however, captures a great (as well as important) part of lean production.

Independent Variable Measures Five independent variables (namely, total quality management, just-in-time, compensation system, goal setting and feedback [the last two representing information system]) were included in the study. Where a variable consists of multiple items, an average score across the items represents the score for that variable. Total Quality Management (TQM) Following Sim and Killough (1998), TQM was assessed using a modified Snell and Dean (1992) TQM scale anchored on a 7-point Likert scale. Eight of the ten original TQM items along with two additional items were used in the study. Both the deleted items were related to “information system,” which is already an independent variable in the study. The first deleted item relates to “information feedback” while the second deleted item relates to “the ability of a plant to keep track of quality cost.” The items added were “preventive maintenance to improve quality” and “quality related training provided to employees.” Just-in-Time (JIT) The JIT scale used in the study was a modified scale from Snell and Dean (1992) (see Sim & Killough, 1998). Snell and Dean (1992) developed a 10-item scale anchored on a 7-point Likert scale to measure JIT adoption. In this study, only eight of the ten original JIT items were used. The first omitted item relates to the extent to which the accounting system reflects costs of manufacturing. This item was loaded onto the TQM construct in the Snell and Dean (1992) study and did not seem to reflect a JIT construct. The second omitted item asked whether the plant was laid out by process or product. This item was also loaded onto the TQM construct and was deleted from the final TQM scale in Snell and Dean (1992). For this study, an item that represented “time spent to achieve a more orderly engineering change by improving the stability of the production schedule” was added. Compensation System The independent variable “compensation system” consisted of two categories, namely fixed compensation plans and incentive compensation plans. Specifically,


123

firms using fixed compensation plans were coded as “0,” while firms using incentive compensation plans were coded as “1.” That is, compensation system was measured as a dichotomous variable. Information System (Goal Setting and Feedback) To enhance manufacturing performance, contemporary information systems set goals/targets to achieve manufacturing performance as well as report manufacturing performance measures to workers. In the study, respondents were asked whether specific numeric targets and performance feedback were provided on eight manufacturing performance measures that were related to customer perceived quality, on-time delivery and waste. Also, following Daniel and Reitsperger (1992), frequency of feedback was anchored on a 5-point Likert scale, and goal setting was anchored between 0 and 1.

RESEARCH MODEL AND TESTING PROCEDURES To investigate the interaction effects between lean production manufacturing practices (i.e. TQM and JIT) and the compensation system and information system (i.e. goal setting and feedback) on manufacturing performance (i.e. changes in production costs in terms of manufacturing and warranty costs), decision trees are constructed on the sample data. In particular, the recursive partitioning algorithm using Chi-square Automatic Interaction Detection (CHAID) is employed. A recursive partitioning model provides evidence on interactions or non-linearity which otherwise may be difficult to capture using a conventional regression model with interaction terms. The objective of decision trees is prediction and/or classification by dividing observations into mutually exclusive and exhaustive subgroups. The division is based on the levels of particular independent variables that have the strongest association with the dependent variable. In its basic form, the decision tree approach begins by searching for the independent variable that divides the sample in such a way that the difference with respect to the dependent variable is greatest among the divided subgroups. At the next stage, each subgroup is further divided into sub-subgroups by searching for the independent variable that divides the subgroup in such a way that the difference with respect to the dependent variable is greatest among the divided sub-subgroups. The independent variable selected need not be the same for each subgroup. This process of division (or splitting in decision trees terminology) usually continues until either no further splitting can produce significant differences in the dependent variable in the new subgroups or the subgroups are too small for any further meaningful division.

124


The subgroups and sub-subgroups are usually referred to as nodes. The end product can be graphically represented by a tree-like structure. (See also Breiman, 1984, pp. 59–92; Ittner et al., 1999; Lehmann et al., 1998.) In the Chi-square Automatic Interaction Detection (CHAID) algorithm, all possible splits of each node for each independent variable are examined. The split that leads to the most significant Chi-square statistic is selected. For the purpose of the study, the dependent variables are dichotomised into two groups.

Results Table 1 – Panel A provides the job title of the respondents. A review of the respondents’ job titles shows that most respondents were closely associated with manufacturing operations, suggesting that they are the appropriate persons for providing shop floor information. Table 1 – Panel B shows descriptive statistics of lean manufacturing practices and the type of reward systems. Except for 24 (30) plants which did not have a formal TQM (JIT) program, respectively, the remaining plants have implemented some form of lean manufacturing. Also, additional analysis shows the majority of the sample plants had annual sales of between 10 and 15 million U.S. dollars. Further, manufacturing firms that made greater use of lean production manufacturing practices such as TQM and JIT also made greater use of incentive compensation plans, and more frequently set goals on operational performance and more frequently provided performance feedback to their workers. Results from the recursive partitioning algorithm are summarised in Fig. 1. As shown, goal setting (p-value = 0.0003) is significantly associated with the change in manufacturing costs. In particular, a higher level of goal setting is associated with decreasing manufacturing costs. In addition, there is also a significant interaction effect between goal setting and JIT (p-value = 0.0187). That is, a higher level of goal setting coupled with a greater use of JIT is associated with better manufacturing performance (in terms of decreases in manufacturing costs) vis-à-vis a higher level of goal setting and a lower use of JIT. The interaction effect of compensation plan can also be seen in Fig. 1. In particular, the combination of a fixed compensation plan with a high level of goal setting and a moderately high use of JIT is associated with a lower probability of decreases in manufacturing costs (p-value = 0.0362). Also, the combination of an incentive compensation plan with a high level of goal setting but a low use of JIT and either a very low use or a very high use of TQM (p-value = 0.0049) is associated with a very low probability of decreases in manufacturing costs. It is noted that the latter part of the findings is not consistent with conventional wisdom. Finally, there is persuasive evidence (p-value = 0.1216) that a high level of goal setting, a relatively


125

Table 1. Panel A: Job Title of Respondents Job Title Used by Respondents

Number of Respondents

Percentage

Plant manager, manufacturing manager, or operations manager VP of operations, VP of engineering, VP of manufacturing, or VP of quality Director of operations, director of manufacturing, or director of manufacturing and engineering CEO, president and CEO, executive VP, or president Miscellaneous titles used – e.g. materials manager, test manager, sourcing and fabrication manager, or product integrity manager No information on job title

24

31.1

22

28.6

13

16.9

5

6.5

7

9.1

Total respondents

77

6

7.8 100

Panel B: Descriptive Statistics – Lean Manufacturing and Type of Compensation Variables

Years of TQM experience Years of JIT experience Variables

Compensation type

No Formal Program

1–2 Years

3–4 Years

4 Years

24

20

18

15

30

22

12

13

Fixed Pay

Fixed + NonMonetary Reward

Fixed + IndividualBased Cash Reward

Fixed + GroupBased Cash Reward

12

25

34

6

low use of JIT and a moderately high use of TQM, coupled with both an incentive compensation plan and a high level of feedback is associated with decreases in manufacturing costs. This result does not come as a “surprise” – for example, the hallmark of lean manufacturing is efficient use of resources (see Womack et al., 1990). This means less space, fewer inventories, less production time (key aspects of JIT), less waste, better quality, and continuously striving for manufacturing excellence (key aspects of TQM). The literature suggests that JIT and TQM often

126 HIAN CHYE KOH ET AL.

Fig. 1. Decision Tree Results for Manufacturing Costs.


127

complement each other; also, a low JIT can be complemented with a high TQM, or vice versa. Figure 2 summarises the decision tree results for warranty costs. As shown in Fig. 2, a high level of TQM (p-value = 0.0139) is significantly associated with the change in warranty costs. In particular, a high level of TQM is associated with decreasing warranty costs. The results also show that a high use of JIT, even in the presence of a low use of TQM, is associated with decreasing warranty costs (p-value = 0.0237) (see earlier comments on the complementarity of JIT and TQM). Although not statistically significant, the effect of compensation plan can also be seen in Fig. 2. In particular, with a moderately high use of TQM, an incentive compensation plan seems to be more associated with decreasing warranty costs than the case of a fixed compensation plan (p-value = 0.2023). Also, there is a significant interaction effect between feedback and TQM and JIT practices (p-value = 0.0156). That is, with low levels of TQM and JIT, the probability of decreases in warranty costs is higher for a low level of feedback than for a high level of feedback. This finding illustrates that the best configurations of management control systems are often contingent upon the type of production systems. For example, for manufacturing plants which have not switched to lean manufacturing practices, it is not necessary for them to reconfigure the accounting systems to provide more timely feedback. Finally, there is some persuasive evidence of interaction effect between goal setting and TQM and compensation plan (p-value = 0.2069). In particular, in the presence of a moderately high level of TQM, the use of an incentive compensation plan coupled with a high level of goal setting appears to be associated with decreasing warranty costs. (It can be argued that some results are statistically insignificant primarily because of the relatively small sample size in the relevant nodes.) The above findings provide support for H1 and H2. Except for one unexpected result, the findings are consistent with the underlying theory, i.e. the best configurations of management control systems are contingent upon the type of production systems.

Discussion Several theoretical papers have motivated this study (see Hemmer, 1995; Ittner & Kogut, 1995; Milgrom & Roberts, 1990, 1995). In particular, Milgrom and Roberts (1990, 1995) provide a theoretical framework that attempts to address the issue of how relationships among parts of a manufacturing system affect performance. They suggest that organizations often experience a simultaneous shift in

128 HIAN CHYE KOH ET AL.

Fig. 2. Decision Tree Results for Warranty Costs.


129

competitive strategy along with various elements of organizational design when they move from mass production to lean manufacturing (i.e. JIT/TQM) or modern manufacturing. In other words, organization changes do not occur in isolation; this implies that synergies or complementarities often arise within clusters of the organizational design that improves manufacturing performance. For example, they argue that organizations are more likely to perform better if manufacturing is made-to-order. Such firms should keep a low level of inventory, employ production equipment with both low setup time and cost, and have shorter production runs, less waste and higher product quality (i.e. the notion of lean manufacturing). Also, it is equally important to have highly skilled workers, a mechanism where information flow and sharing are encouraged, and policies that support and reward positive behavior. In essence, Milgrom and Roberts’ (1995) framework suggests that successful implementation of lean manufacturing (TQM/JIT) requires complementary management control systems. Similarly, Hemmer (1995) suggests that characteristics of a production system and management control systems should be simultaneously designed. Given the complexity of the complementarities between manufacturing production systems and management control systems, it is expected that a recursive partitioning model should provide better insight into this still poorly understood issue. This study investigates if lean production manufacturing practices interact with compensation and information systems to enhance manufacturing performance (i.e. to lower production costs). Overall, results from the recursive partitioning model (i.e. CHAID) provide support for such interaction effects. In particular, the study found that better manufacturing performance (i.e. lower production costs) can be achieved when lean production manufacturing practices such as TQM and JIT are used along with incentive compensation plans. Also, synergies do result from combining TQM/JIT with more frequent performance feedback along with attention directing goals. These findings suggest that if organisational infrastructure and management control systems are not aligned with manufacturing practices, then the potential benefits of lean manufacturing (i.e. TQM and JIT) may not be fully realised. The findings are consistent with prior research that shows that poor performance of many modern manufacturing practices is partially attributable to the continued reliance on traditional management control systems that do not provide appropriate information or reward systems (Banker et al., 1993; Johnson & Kaplan, 1987; Kaplan, 1983, 1990). In interpreting the results of the study, the following limitations should be borne in mind. First, the sample size for the study (77 responses) is relatively small. This, however, is mitigated by the relatively high response rate of 59.7% in the final sample. Second, the usual caveats associated with a self-report questionnaire survey apply. Incidentally, results of sensitivity tests show no indication of

130


non-response bias by geographical region and 4-digit SIC code. Third, the present research design precludes inferences to be made with regards to the pattern of changes in the warranty costs or manufacturing costs. In this concluding section, it is appropriate to discuss some caveats to recursive partitioning models such as CHAID. For example, the splitting of the subgroups (or nodes) are results driven, i.e. they are not theory or decision driven. Nevertheless, when interpreting the findings, a common approach is to use some underlying theories to explain the results or patterns. Finally, since there is no model-fit statistic, there exists a risk of over-fitting the model. Despite some limitations, a recursive partitioning model such as CHAID is a potentially useful tool when examining complex relationships in the real world and has proved to be helpful in discovering meaningful patterns when large quantities of data are available (see, for example, Berry & Linoff, 1997, p. 5).

REFERENCES Balakrishnan, R., Linsmeier, T., & Venkatachalam, M. (1996). Financial benefits from JIT adoption: Effects of customer concentration and cost structure. The Accounting Review, 71, 183–205. Banker, R., Potter, G., & Schroeder, R. (1993). Reporting manufacturing performance measures to workers: An empirical study. Journal of Management Accounting Research, 5, 33–55. Berry, M., & Linoff, G. (1997). Data mining techniques – For marketing, sales, and customer support. Wiley Computer Publishing. Birnberg, J. G., & Snodgrass, C. (1988). Culture and control: A field study. Accounting, Organizations and Society, 13, 447–464. Breiman, L. (1984). Classification and regression trees. Belmont, CA: Wardsworth International Group. Buffa, E. S. (1984). Meeting the competitive challenge. IL: Dow Jones-Irwin. Chenhall, R. H. (1997). Reliance on manufacturing performance measures, total quality management and organizational performance. Management Accounting Research, 8, 187–206. Coopers & Lybrand (1992). Compensation planning for 1993. New York: Coopers & Lybrand. Daniel, S., & Reitsperger, W. (1991). Linking quality strategy with management control systems: Empirical evidence from Japanese industry. Accounting Organizations and Society, 17, 601–618. Daniel, S., & Reitsperger, W. (1992). Management control systems for quality: An empirical comparison of the U.S. and Japanese electronic industry. Journal of Management Accounting Research, 4, 64–78. Erez, M. (1977). Feedback: A necessary condition for the goal setting-performance relationship. Journal of Applied Psychology, 62, 624–627. Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of Law and Economics, 26, 301–325. Flamhaultz, E. G., Das, T. K., & Tsui, A. (1985). Towards an integrative framework of organizational control. Accounting, Organizations and Society, 10, 51–66. Gage, G. H. (1982). On acceptance of strategic planning systems. In: P. Lorange (Ed.), Implementation of Strategic Planning. NJ: Prentice-Hall.


131

Gomez-Mejia, L. R., & Balkin, D. B. (1992). Structure and process of diversification, compensation, strategy and firm performance. Strategic Management Journal, 13, 381–387. Govindarajan, V. (1988). A contingency approach to strategy implementation at the business-unit level: Integrating administrative mechanisms with strategy. Academy of Management Journal, 31, 828–853. Govindarajan, V., & Gupta, A. K. (1985). Linking control systems to business unit strategy: Impact on performance. Accounting, Organizations and Society, 10, 51–66. Hackman, J. R., & Wageman, R. (1995). Total quality management: Empirical, conceptual and practical issues. Administrative Science Quarterly, 40, 309–342. Harari, O. (1993). Ten reasons why TQM doesn’t work. Management Review, 82, 33–38. Hayes, R. H., Wheelwright, S. C., & Clark, K. B. (1988). Restoring our competitive edge: Competing through manufacturing. New York: Free Press. Hemmer, T. (1995). On the interrelation between production technology, job design, and incentives. Journal of Accounting and Economics, 19, 209–245. Holmstrom, B. (1979). Moral hazards and observability. Bell Journal of Economics, 10, 74–91. Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects of human resource management practices on productivity: A study of steel finishing lines. The American Economic Review, 87, 291–314. Ittner, C., & Kogut, B. (1995). How control systems can support organizational flexibility. In: E. Bowman & B. Kogut (Eds), Redesigning the Firm. New York: Oxford University Press. Ittner, C., & Larcker, D. F. (1995). Total quality management and the choice of information and reward systems. Journal for Accounting Research, 33(Suppl.), 1–34. Ittner, C., & Larcker, D. F. (1997). The performance effects of process management techniques. Management Science, 43, 534–552. Ittner, C., Larcker, D. F., Nagar, V., & Rajan, M. (1999). Supplier selection, monitoring practices, and firm performance. Journal of Accounting and Public Policy, 18, 253–281. Jensen, M., & Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs, and ownership structure. Journal of Financial Economics, 3, 305–360. Johnson, H., & Kaplan, R. (1987). Relevance lost: The rise and fall of management accounting. MA: Harvard Business School Press. Kalagananam, S. S., & Lindsay, R. M. (1998). The use of organic models of control in JIT firms: Generalising Woodward’s findings to modern manufacturing practices. Accounting Organizations and Society, 24, 1–30. Kaplan, R. S. (1983). Measuring manufacturing performance: A new challenge for managerial accounting research. The Accounting Review, 58, 686–705. Kaplan, R. S. (1990). Measures for manufacturing excellence. MA: Harvard Business School Press. Latham, G. P., & Lee, T. W. (1986). Goal setting. In: E. A. Locke (Ed.), Generalizing from Laboratory to Field Settings. MA: Lexington Books. Lehmann, D. R., Gupta, S., & Steckel, J. H. (1998). Marketing research. MA: Addison-Wesley. Locke, E., & Latham, G. (1990). Goal setting theory and task performance. New York: Prentice-Hall. Locke, E. A., Shaw, K. M., Saari, L. M., & Latham, G. P. (1981). Goal setting and task performance: 1969–1980. Psychological Bulletin, 90, 121–152. MacDuffie, J. P. (1995). Human resource bundles and manufacturing performance: Organizational logic and flexible production systems in the world auto industry. Industrial and Labor Relations Review, 48, 197–221. Milgrom, P., & Roberts, J. (1990). The economics of modern manufacturing: Technology, strategy and organization. American Economic Review, 80, 511–528.

132


Milgrom, P., & Roberts, J. (1995). Complementarities and fit strategy, structure, and organizational change in manufacturing. Journal of Accounting and Economics, 19, 179–208. Milkovich, G. T. (1988). A strategic perspective to compensation management. Research in Personnel and Human Resources, 9, 263–288. Powell, T. C. (1995). Total quality management as competitive advantage: A review and empirical study. Strategic Management Journal, 16, 15–37. Sarkar, R. G. (1997). Modern manufacturing practices: Information, incentives and implementation. Harvard Business School Working Paper. Schonberger, R. J. (1986). World-class manufacturing: The lessons of simplicity applied. New York: Free Press. Siegel, D. S., Waldman, D. A., & Youngdahl, W. E. (1997). The adoption of advanced manufacturing technologies: Human resource management implications. IEEE Transactions on Engineering Management, 44, 288–298. Sim, K. L., & Killough, L. N. (1998). The performance effects of complementarities between manufacturing practices and management accounting systems. Journal of Management Accounting Research, 10, 325–346. Snell, S., & Dean, J. (1992). Integrated manufacturing and human resource management: A human capital perspective. Academy of Management Journal, 35, 467–504. Womack, J. P., Jones, D. T., & Daniel, R. (1990). The machine that changed the world. New York: Macmillan. Wruck, K. H., & Jensen, M. C. (1994). Science, specific knowledge and total quality management. Journal of Accounting and Economics, 18, 247–287.

APPENDIX: QUESTIONNAIRE ∗ Reverse

Coding

Production Costs In this section, we are interested to know the extent to which the following performance attributes have changed during the past 3 years using the scale of 1–5 listed below: (1 = Decrease Tremendously, 3 = No Change, 5 = Increase Tremendously). Manufacturing Cost Warranty Cost

Just in Time (Anchored by 1 = Not at All or Very Little, 4 = To Some Extent, and 7 = Completely or A Great Deal)


133

(1) Are products pulled through the plant by the final assembly schedule/master production schedule? (2) How much attention is devoted to minimizing set up time? (3) How closely/consistent are predetermined preventive maintenance plans adhered to? (4) How much time is spent in achieving a more orderly engineering change by improving the stability of the production schedule? How much has each of the following changed in the past three years? (Anchored by 1 = Large Decrease, 4 = Same, and 7 = Large Increase) ∗ (5)

Number of your suppliers Frequency of the deliveries (7) Length of product runs ∗ (8) Amount of buffer stock ∗ (9) Number of total parts in Bill of Material ∗ (6)

Total Quality Management (Anchored by 1 = Very Little or None, 4 = Moderate, and 7 = A Great Deal or Consistent Use) (1) How much time does the plant management staff devote to quality improvement? (2) How much time is spent working with suppliers to improve their quality? ∗ (3) How would you describe your current approach to providing quality products? Built In

1

2

Some of Each

Post Production Inspection

3

5

4

6

7

(4) How much effort (both time and cost) is spent in preventive maintenance to improve quality? (5) How much effort (both time and cost) is spent in providing quality related training to the plant’s employees? (6)a What percentage of the plant’s manufacturing processes are under statistical % quality control?

134


(7)a What percentage of the plant’s employees have quality as a major responsibility? % How would you describe the level of use within your plant of the following quality improvement methodologies? (Anchored by 1 = Little or None, 4 = Moderate Use, and 7 = Consistent Use) (8) Quality Function Deployment (9) Taguchi Methods (10) Continuous Process Improvement numeric number reported was divided by 14.3 (i.e. 100/7 = 14.29%) in order to convert the percent to a scale of 1–7.

a The

Performance Feedback In this section, we are interested in the availability and frequency of performance feedback provided to the shop floor personnel. Please indicate the frequency of feedback by circling the appropriate number from 1 to 5. 1 = Never 2 = Occasionally 3 = Monthly 4 = Weekly 5 = Daily Customer Perception Customer perceived quality Customer complaints Delivery On-time delivery Quality Cost of scrap Rework Defect Warranty cost Sales return Cycle Time Product development time Manufacturing lead time Work station setup time


135

Goals Does your firm set specific numeric targets for the following performance measures? (Anchoring on “Yes” or “No”) Customer Perception Customer perceived quality Customer complaints Delivery On-time delivery Quality Cost of scrap Rework Defect Warranty cost Sales return Cycle Time Product development time Manufacturing lead time Work station setup time

The Reward System (1) How are plant workers currently being compensated? (please circle only one). (a) Strictly individual fixed pay only (b) Individual fixed pay + non-monetary reward (c) Individual fixed pay + individual-based monetary incentive (d) Individual fixed pay + group-based monetary incentive

COMPENSATION STRATEGY AND ORGANIZATIONAL PERFORMANCE: EVIDENCE FROM THE BANKING INDUSTRY IN AN EMERGING ECONOMY C. Janie Chang, Chin S. Ou and Anne Wu ABSTRACT To survive in the turbulent, global business environment, companies must apply strategies to increase their competitiveness. Expectancy theory indicates that salary rewards can motivate employees to achieve company objectives (Vroom, 1964). First, we develop an analytical model to predict that companies using a high-reward strategy could outperform those using a low-reward strategy. Then, we obtain archival data from banking firms in Taiwan to test the proposed model empirically. We control the effects of operating scale (firm size) and assets utilization efficiency (assets utilization ratio). Empirical results show that salary levels and assets utilization efficiency significantly affect banks’ profitability.

INTRODUCTION The banking industry has played a critical role in many countries’ financial operations. Since the early 1990s, many large banks in the world have been Advances in Management Accounting Advances in Management Accounting, Volume 12, 137–150 Copyright © 2004 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12006-6

137

138

C. JANIE CHANG ET AL.

involved in mergers and acquisitions to provide various services/products to their clients so that they can be competitive in the global market. For example, Sanwa Bank, Tokai Bank, and Asahi Bank, the three largest banks in Japan, have merged and become the third largest bank in the world; and Deutsche Bank acquired Dresdner Bank AG, becoming the largest bank in Germany (Cheng, 2000). Most significantly, in 1999, President Clinton signed the Gramm-Leach-Bliley Act that allows the operations of universal banking in the United States. Until the Act became effective, U.S. banks had to establish subsidiaries, so-called holding companies, to conduct their non-banking business activities such as insurance and security investments. Since the Act’s passage, many countries in emerging economies have passed similar laws to enhance the competitiveness of their banks. All these changes have reshaped the landscape of global banking industry. To survive and prosper in the turbulent and competitive business environment, players in commercial banking must develop strategies to attract highly qualified individuals to join the industry. Expectancy theory suggests that employees are motivated by the firm’s reward structure (Vroom, 1964). Kaplan and Atkinson (1998) state that “pay-for-performance is an artifact that you want to motivate people to pursue organization objectives.” Horngren et al. (2000) also indicate that reward systems are critical in any performance evaluation model. Many prior studies have examined the relationship between compensation strategy and organizational performance at executive levels (e.g. Hadlock & Lumer, 1997; Jensen & Murphy, 1990; Lambert et al., 1993; Warner et al., 1988). However, empirical evidence supporting this theory is sparse, when it is applied to the general employee level. According to Kaplan and Norton (1996), one important approach towards translating strategies into actions is to assure that the employee salary/reward system is closely linked to organizational performance. In this study, we explore whether salary rewards can be an effective motivator for employees to act in ways that promote their firms’ objectives. Rapid economic development in the Asian-Pacific region and the fact that many countries in that region have joined the World Trade Organization (WTO) mean that financial resources can be moved with relative freedom from one country to another. That is, emerging markets in this region have drastically increased their interactions with the advanced economies. Since many Asian-Pacific countries are at comparable stages of economic development and all have strong interests in developing their financial markets, examining one of the emerging markets in the Asian-Pacific region can provide insights into the other emerging markets in the region. The purpose of this study is to investigate empirically whether reward strategies may affect the organizational performance of the banking industry in an Asian-Pacific emerging market, Taiwan.

Compensation Strategy and Organizational Performance

139

Based on prior literature, we first develop an analytical model to prove that companies using a high-reward strategy could outperform those using a low-reward strategy. Then, we obtain archival data of 232 observations from banking firms in Taiwan to test the proposed model empirically. Our results fully support the conjecture we develop in the analytical model. The remainder of this paper is organized as follows. In the next section, we review the literature and develop the analytical model. Then, after describing our samples and variables, we report our empirical results. Finally, we discuss the findings, conclude the study and identify future research directions.

BACKGROUND AND ANALYTICAL MODEL DEVELOPMENT Background In general, there are two types of rewards for employees: intrinsic and extrinsic (Deci & Ryan, 1985; Kohn, 1992; Pittmaan et al., 1982). Intrinsic rewards relate to the environment within which the employee operates. Employees often have high job satisfaction in a collegial organizational culture with positive management styles. Extrinsic motivators include salaries, bonuses and financial incentives, such as stock options. A number of recent studies have examined whether different compensation schemes for top executives relate to corporate profitability and other measures of organizational performance. The results are quite consistent with Deci and Ryan’s prediction: there are only slight or even negative correlations between compensation and performance (e.g. Hadlock & Lumer, 1997; Jensen & Murphy, 1990; Lanen & Larcker, 1992). Conversely, other studies have shown that rewards should be based on individual performance and that the effects of such rewards can reflect on the company’s performance. In a study on making decisions about pay increases, Fossum and Fitch (1985) used three groups of subjects: college students, line managers, and compensation managers. All three groups agreed that the largest consideration should be given to individual performance – even over other relevant criteria, such as cost of living, difficulties in replacing someone who leaves, seniority, and budget constraints. In addition, in management accounting contexts, Kaplan and Atkinson (1998) argue that it is the responsibility of accounting professionals to evaluate whether employees’ rewards are appropriately associated with firm performance. Recently, Fey et al. (2000) conducted a survey using both managers and nonmanagers from 395 foreign firms operating in Russia. Their results show a direct positive relationship between merit pay for both managers and non-managers and

140


the firm’s performance. However, they used only non-financial measures, such as job security, internal promotion, employee training, and career planning, to evaluate firm performance. Schuster (1985) conducted a survey with 66 randomly sampled Boston-area high-tech firms; that survey’s results reveal a greater reliance on special incentives (e.g. bonuses, stock options, and profit-sharing plans) in financially successful high-tech companies than in unsuccessful ones. Many prior studies have reported that reward/incentive systems are positively related to firm performance (e.g. Arthur, 1992; Fey et al., 2000; Gerhart & Milkovich, 1990; Pfeffer, 1994; Schuster, 1985). However, these studies used either a non-financial measurement for firm performance or used a survey questionnaire and did not investigate actual firm profitability or financial performance. Neither did they control essential factors such as operating scale (e.g. firm size) and operating efficiency (e.g. asset utilization). Martocchio (1998) states that two aspects need to be examined to determine whether a firm’s compensation strategies are effective: in the short term, a compensation strategy is effective if it motivates employees to behave the way the firm expects them to; in the long term, the strategy should be able to boost the firm’s financial performance. Hence, we develop an analytical model to examine the impact of reward/incentive systems on a firm’s long-term performance. Proposed Analytical Model Similar to Ou and Lee (2001), we propose an analytical model to depict the association between a firm’s reward/salary strategy and its financial performance. We assume that there are two types (t) of workers on the labor market: type h with high skills/ability, and type l with low skills/ability (i.e. t = l, h). The productivity of type t workers is denoted as xt and x h > x l > 0. The probability of finding workers of types l and h is f and 1 − f, respectively. The above-mentioned information is available on the market. We further use the following notations: at : the effort put in by type t workers Yt : the value of the service provided by type t workers pt : the pays of type t workers when the value of the service they provide is Yt ␲t : the profit of the firm from the service provided by type t workers a 2t /2: type t workers’ personal costs incurred with effort at The value of service provided by a type t worker (Yt ) is determined by his/her productivity (xt ) and effort (at ). Their relationship can be described by Y t = x t + a t . While xt and et are a worker’s private information, the firm can observe Yt after hiring the worker and will pay the worker pt based on the observed Yt .


141

The a 2t /2 means that the cost to a worker increases with efforts in an increasing rate. Accordingly, the firm’s corresponding profit ␲t is (Y t − p t ) or (x t + a t − p t ). The objective function is to maximize the overall profit of the firm, which can be formulated using the following equations: Max

eh ,el ,ph ,pl

f (x l + a l − p l ) + (1 − f )(x h + a h − p h )

(1)

s.t. pl −

a 2l ≥0 2

(2)

a 2h [a l − (x h − x l )]2 ≥ pl − (3) 2 2 Equation (2) states the constraint that type l workers will take any job when the wage of the job is larger than or equal to the associated personal cost (a 2l /2). Equation (3) indicates that type h workers will take any job that pays them properly. That is, the personal benefit earned by a highly skilled worker taking ph is greater than or equal to the benefit from taking pl . Note that when a highly skilled worker earns ph , he/she must put in effort eh with personal costs of a 2h /2. We know that Y l = x l + a l , Y l = x h + x l + a l − x h = x h + a l − (x h − x l ). Therefore, when a highly skilled worker takes a low-paying job to produce Y l , the personal cost to the highly skilled worker will be a l − (x h − x l ). Hence, the personal benefit a highly skilled worker can have when he/she takes a low-wage job is p l − [a l − (x h − x l )]2 /2. Using this model, we would like to prove that firms using high rewards to attract high-skilled workers can outperform those using low rewards to attract less-skilled workers. We use the Lagrange multiplier to solve the above-mentioned objective function (Eq. (1)). Let L represent the Lagrangian; ␭ and ␬ represent the Lagrange multipliers. The Lagrangian can be shown as follows: ph −

a2 L = f (x l + a l − p l ) + (1 − f )(x h + a h − p h ) + k(p l − l ) 2 2 2 a [a l − (x h − x l )] + ␭ ph − h − pl + 2 2

(4)

By taking the derivatives of al , ah , pl , and, ph respectively, we get the following four equations: f − ␬a l + ␭[a l − (x h − x l ) = 0

(4a)

1 − f − ␭a h = 0

(4b)

142


−f + ␬−␭ = 0

(4c)

−(1 − f ) + ␭ = 0

(4d)

Besides, from Eq. (4), we know that: a2 ␬( p l − l ) = 0 2 a 2h [a l − (x h − x l )]2 ␭ ph − − pl + =0 2 2

(4e) (4f )

Solving Eqs (4a)–(4f), we get the following results: ␬=1, ␭ = 1 − f, and the values of pl , ph , al , ah are listed as Eqs (5)–(8). pl =

a 2l 2

(5)

(a l − x h + x l )2 1 a 2l + − 2 2 2 1−f al = 1 − (x h − x l ) f ph =

ah = 1

(6) (7) (8)

From Eqs (5) and (7), we know that the firm’s profit generated by a less-skilled worker is ␲l , and ␲l = x l + a l − p l = x l + 1 −

a2 1−f (x h − x l ) − l f 2

(9)

From Eqs (6) and (8), we know that the firm’s profit, ␲H , generated by a highly-skilled worker is (a l − x h + x l )2 1 a 2l ␲h = x h + a h − p h = x h + 1 − + − (10) 2 2 2 Comparing Eqs (9) and (10), we get the following equation. xh − xl 2 ␲h − ␲l = >0 2f

(11)

According to Eq. (11), we prove that firms using high rewards to attract highly-skilled workers can generate higher profits than those using low rewards to attract less-skilled workers. This becomes our empirical hypothesis. We have


143

empirically tested this hypothesis using banking firms in Taiwan. The following section describes the sample and variables used in the empirical study.

SAMPLE AND VARIABLES The Sample The sample consists of 232 observations of banking firms listed on the Taiwan Stock Exchange or in the Taiwan Over-the-Counter market from 1991 to 1999. Table 1 provides the descriptive statistics of the sample firms. On average, each firm has 1,645 employees. The means (standard deviations) of net income and salary expenses are NT$1,543,126,000 (NT$1,979,132,000) and NT$1,660,523,000 (NT$1,923,147,000), respectively.

Independent and Control Variables The purpose of this study is to examine the relationship between salary rewards and firm performance, especially profitability. The independent variable is a firm’s salary level (SL), which is the mean salary expense per employee (SE/NE). We include a couple of control variables in our empirical model. The well-known Du Pont model decomposes a firm’s operating performance into two components: profitability and efficiency. The efficiency component is the asset utilization ratio (AUR), which measures a firm’s ability to generate sales from investment in assets (Penman, 2001; Stickney & Brown, 1999). Bernstein and Wild (1998, p. 30) state that “asset utilization ratios, relating sales to different asset categories, are important determinants of return on investment.” One of the AURs suggested by Bernstein and Wild (1998, p. 31) is net sales to total assets ratio (NS/TA). Since our focus in the analytical model is firm profitability, we use this important variable to control firm efficiency when analyzing our data. Table 2 defines all the variables used in our empirical tests. In addition, we include operating scale (Log(Total Assets)) as the control variable in our model. Issues related to operating scale have continuously generated much interest in the academic community (e.g. Altunbas, Evans & Molyneux, 2001; Altunbas, Gardener, Molyneux & Moore, 2001; De Pinho, 2001). Theories of economies of scale suggest that the best efficiency and thus the highest performance can be obtained when a firm operates at the optimal scale. However, the operating scale (i.e. firm size) is difficult for individual employees to influence, so we have decided to use it as a control variable for firm differences.

144

Table 1. Descriptive Statistics of the Sample. Variables NS OI PTI NI SE TA NE

Mean

Std. Dev.

Maximum

3rd Quantile

Median

1st Quantile

Minimum

$16,995,306 $1,822,035 $1,825,293 $1,543,126 $1,660,523 $234,390,000 1.645

$17,624,233 $2,380,777 $2,294,861 $1,979,132 $1,923,147 $252,430,000 1.540

$80,714,132 $11,911,059 $11,911,059 $11,432,059 $8,432,252 $1,178,700,000 6.662

$19,481,054 $2,414,852 $2,782,093 $2,233,528 $1,825,000 $249,680,000 2.003

$10,560,396 $994,926 $1,042,781 $868,631 $916,080 $144,290,000 1.167

$5,332,513 $506,836 $503,978 $425,555 $482,589 $77,872,764 0.605

$568,302 $–4,950,170 $–4,938,785 $–3,515,260 $93,308 $3,112,100 0.237


Note: N = 232, Unit: 1000 (with NT$). NS = Net Sales, OI = Operating Income, PTI = Pretax Income, NI = Net Income, SE = Salary Expense, TA = Total Assets, NE = Number of Employees.


145

Table 2. Variable Definitions. Name

Definition

Dependent variables PM1 PM2 PM3 PM4

TL/TD (loan-to-deposit ratio) OI/NE (operating income per employee) PTI/NE (pretax income per employee) NI/NE (net income per employee)

Independent variable SL

SE/NE (salaries expense per employee)

Control variables OS AUR AUR LDR NE NI NS OI OS PM1 PM2 PM3 PM4 PTI SE SL TL TD TA

Log (TA) (log for total assets) NS/TA (net sales to total assets ratio) Assets utilization ratio Loan-to-deposit ratio Number of employees Net income Net sales Operating income Operating scale Performance measure of a bank’s potential profitability First accounting-based profitability measure (per employee) Second accounting-based profitability measure (per employee) Third accounting-based profitability measure (per employee) Pretax income Salary expenses Salary level Total loans Total deposits Total assets

Dependent Variables To evaluate bank performance concerning profitability, we use the loan-to-deposit ratio as one of the dependent variables. The traditional banking business generates interest income from loans. A high loan-to-deposit ratio indicates high operating performance in the banking industry. Banks often report increased earnings when they have increased the loan volume (ABA Banking Journal, 1992). In practice, the loan-to-deposit ratio has been used to monitor a bank’s potential profitability. For example, the American Bankers Association’s Community Bankers Council uses the loan-to-deposit ratio as one of several indicators to evaluate a bank’s loan demand and thus its potential earnings (Steinborn, 1994). Fin and Frederick (1992)

146


specify that “Banks that want a strategic earning advantage must strive for a strong loan-to-deposit ratio. They must cultivate loan business, maintain it, and attract new business. Increasing the loan-to-deposit ratio by one percentage point will likely add four or five basis points to net interest margins.” Hence, we use this important indicator as one of our performance measures. In addition, we use three accounting-based profitability measures as our dependent variables: operating income per employee (OI/NE), pretax income per employee (PTI/NE), and net income per employee (NI/NE). These measures are commonly used by financial analysts to evaluate a firm’s performance. Although prior research has suggested including market-based measures to evaluate firm performance, the focus was to examine the relationship between firm performance and executive compensations (Gorenstein, 1995; Jensen & Murphy, 1990; McCarthy, 1995; Stock, 1994). Since the purpose of this study is to explore a firm’s salary strategy on general employees, not on executives, we focus on accounting-based performance measures.

EMPIRICAL RESULTS Descriptive Statistics Table 3 presents the descriptive statistics of all the variables, including the means, standard deviations, maximum and minimum data values, and medians of the sample’s loan-to-debt ratio, operating income per employee, pre-tax income per employee, net income per employee, salary levels, assets utilization ratio, and operating scales. To assess the collinearity among independent and control variables in our regression models, we obtained the correlation matrix using Pearson and Spearman correlation coefficients. According to the results, none of the correlations is high (below 0.9). Thus, we do not have much concern about collinearity. Regression Models and Results Recall that we use four performance indicators to measure bank performance in profitability: loan-to-deposit ratio (Total loans/Total deposits), operating income per employee, pre-tax income per employee, and net income per employee. They are denoted as PMi , i = 1–4. That is, we use four regressions to examine our data. The regression model is as follows, PMi = ␤0 + ␤1 SL + ␤2 AUR + ␤3 OS + ␧

(i = 1–4)

(12)

where SL is salary level; AUR is assets utilization ratio; and OS is operating scale.

Variables PM1 (TL/TD) PM2 (OI/NE) PM3 (PTI/NE) PM4 (NI/NE) SL (SE/NE) AUR (NS/TA) OS (Log(TA))

Mean

Std. Dev.

Maximum

3rd Quantile

Median

89 1,037 1,065 900 889 0.0782 18.7619

19 1,565 1,583 1,489 260 0.0437 1.0819

213 19,526 19,526 18,741 1,760 0.4480 20.8876

95 1,332 1,342 1,143 1,027 0.0781 19.3356

87 1,034 1,023 853 822 0.0712 18.7873

1st Quantile 79 755 765 642 740 0.0676 18.1705

Minimum 7 −4,705 −4,611 −4,611 223 0.0436 14.9508


Table 3. Descriptive Statistics of Variables.

Note: N = 232. Unit: $1000 (with NT$).

147

148


Table 4. OLS Regression Results. Independent and Control Variables Intercept SL (SE/NE) AUR (NS/TA) OS (Log(TA)) Adj-R2 F-Value p-Value N

Dependent Variables PM1 (TL/TD) 28 (1.00) 0.000394*** (5.64) 0.03 (0.02) 0.03 (1.56) 0.25 25.88 0.0001 221a

PM2 (OI/NE)

PM3 (PTI/NE)

−1959 (−0.94) 1.92*** (4.01) 10930*** (4.56) 22.91 (0.19)

−1665 (−0.79) 2.00*** (4.12) 11162*** (4.61) 4.25 (0.04)

0.13 12.33 0.0001 231

0.13 12.61 0.0001 231

PM4 (NI/NE) −1859 (−0.93) 1.74*** (3.78) 9522*** (4.14) 24.66 (0.21) 0.11 10.65 0.0001 231

Note: Model: PMi = ␤0 + ␤1 SL + ␤2 AUR + ␤3 OS + ␧i (i = 1–4). t-Value in parentheses. a Due to missing values, there are only 221 samples in the model of PM1. ∗∗∗ Significant at 0.01.

According to the results in Table 4, a bank’s employee salary level is significantly related to all the profitability measures at 1% level. Our empirical results fully support the hypothesis we generated in the analytical model that compensation strategy is a significant managerial variable in determining a firm’s financial operating performance in the banking industry. In addition, the results reveal that the asset utilization efficiency is also an important factor explaining the variances in accounting-based profitability measures.

CONCLUSIONS The global business environment has been extremely turbulent and competitive. Companies must apply effective strategies to increase their competitiveness to survive and prosper in such an environment. Expectancy theory indicates that salary rewards can motivate employees to achieve company objectives. Accordingly, we develop an analytical model to prove that companies using a high-reward strategy could outperform those using a low-reward strategy. Then, we obtain archival data from banking firms in Taiwan to empirically test the proposed model. Using four different performance measures on profitability, we find that salary level and assets utilization ratio significantly affect Taiwanese bank performance. In this study, we have examined the salary strategies used by the banks in Taiwan on their profitability performance. To generalize our findings, future studies can look into this issue using firms in different industries and from different countries. Also, does the relationship between employee compensation strategies and firm performance depend on various firm characteristics, such as


149

service vs. manufacturing firms or labor-intensive vs. technology-intensive firms? It is also critical for firms to operate efficiently to be competitive. Hence, we should explore whether strategies of compensation for general employees affect firms’ operating efficiency as well as their profitability. Furthermore, there are different types of compensation, such as fixed salaries, variable commissions/bonuses, and stock options. Future studies can look into the impact of different types of compensation on firm performance. For example, employees’ fixed salaries may relate highly to a firm’s short-term performance, and strategies on employees’ bonus and stock options may affect a firm’s long-term performance. In addition, it will be beneficial to most firms if solutions can be derived to find an optimal combination of fixed and variable compensation for different types of employees (e.g. salespeople vs. engineers). In this study, we have used regression models to investigate the effects of the independent variables on several dependent variables. Most of the previous studies have employed a similar approach to examine the relationships between compensation variables (e.g. cash compensation or total compensation) and a specific performance indicator (e.g. an accounting-based or market-based performance measure). However, different performance measures should be incorporated simultaneously into the model, rather than separately tested, to fully understand their relationship to compensation strategies. Future studies can fill the gap by examining a simultaneous relationship between a set of compensation variables and a set of performance measures.

REFERENCES Altunbas, Y., Evans, L., & Molyneux, P. (2001). Bank ownership and efficiency. Journal of Money, Credit, and Banking, 33(4), 926–954. Altunbas, Y., Gardener, E. P. M., Molyneux, P., & Moore, B. (2001). Efficiency in European banking. European Economic Review, 45(10), 1931–1955. Arthur, J. B. (1992). The link between business strategy and industrial relations systems in American steel minimills. Industrial and Labor Relations Review, 45(3), 488–506. Bernstein, L. A., & Wild, J. J. (1998). Financial statement analysis. Boston: McGraw-Hill. Cheng, Y. R. (2000). The environmental change of money and banking system on the new development of Taiwan banking. Co Today – Taiwan Cooperative Bank, 26(5), 44–54. De Pinho, P. S. (2001). Using accounting data to measure efficiency in banking: An application to Portugal. Applied Financial Economics, 11(5), 527–538. Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York: Plenum Press. Fey, C. F., Bjorkman, I., & Pavlovskaya, A. (2000). The effect of human resource management practices on firm performance in Russia. International Journal of Human Resources Management, 11(1), 1–18. Fin, W. T., & Frederick, J. B. (1992). Managing the margin. ABA Banking Journal, 84(4), 50–53.

150


Fossum, J. A., & Fitch, M. K. (1985). The effects of individual and contextual attributes on the sizes of recommended salary increases. Personnel Psychology, 38(Autumn), 587–602. Gerhart, B., & Milkovich, G. T. (1990). Organizational differences in managerial compensation and financial performance. Academy of Management Journal, 33(4), 663–691. Gorenstein, J. (1995). How the executive bonus system works. The Philadelphia Inquirer (July 9). Hadlock, C. J., & Lumer, G. B. (1997). Compensation, turnover, and top management incentives: Historical evidence. Journal of Business, 70(2), 153–187. Horngren, C. T., Foster, G., & Darter, S. M. (2000). Cost accounting: A managerial emphasis (10th ed.). Englewood Cliffs, NJ: Prentice-Hall. Jensen, M. C., & Murphy, K. J. (1990). Performance pay and top-management incentives. Journal of Political Economy, 98(21), 225–255. Kaplan, R. S., & Atkinson, A. A. (1998). Advanced management accounting. Englewood Cliffs, NJ: Prentice-Hall. Kaplan, R. S., & Norton, D. P. (1996). The balanced scorecard: Translating strategy into action. Harvard Business School Press, 73–84. Kohn, A. (1992). No contest: The case against competition. Boston: Houghton Mifflin. Lambert, R. A., Larcker, D. F., & Weigelt, K. (1993). The structure of organizational incentives. Administrative Science Quarterly, 38, 438–461. Lanen, W. N., & Larcker, D. F. (1992). Executive compensation contract adoption in the electric utility industry. Journal of Accounting Research, 30(1), 70–93. Martocchio, J. J. (1998). Strategic compensation – A human resource management approach. Prentice-Hall. McCarthy, M. J. (1995). Top 2 UAL officers got $17 million in 94 stock options. Wall Street Journal (April 5). Ou, C. S., & Lee, C. L. (2001). A study of the association between compensation strategy and labor performance. Working Paper, National ChengChi University. Penman, S. H. (2001). Financial statement analysis security valuation. McGraw-Hill and Irwin. Pfeffer, J. (1994). Competitive advantage through people. Boston: Harvard Business School Press. Pittmaan, T. S., Emery, J., & Boggiano, A. K. (1982). Intrinsic and extrinsic motivational orientations: Reward-induced changes in preference for complexity. Journal of Personality and Social Psychology (March). Schuster, J. R. (1985). Compensation plan design: The power behind the best high-tech companies. Management Review, 74(May), 21–25. Steinborn, D. (1994). Earnings dip in first half. ABA Banking Journal, 86(9), 26–27. Stickney, C. P., & Brown, P. R. (1999). Financial reporting and statement analysis. Fort Worth: Dryden. Stock, C. (1994). Bottom lines: Did CEO earn his pay. The Philadelphia Inquirer (November 20). Vroom, V. H. (1964). Work and motivation. New York: Wiley. Warner, J., Watts, R. L., & Wruck, K. H. (1988). Stock prices and top management change. Journal of Financial Economics, 20, 461–492.

ACCOUNTING FOR COST INTERACTIONS IN DESIGNING PRODUCTS Mohamed E. Bayou and Alan Reinstein ABSTRACT Since quality cannot be manufactured or tested into a product but must be designed in, effective product design is a prerequisite for effective manufacturing. However, the concept of effective product design involves a number of complexities. First, product design often overlaps with such design types as engineering design, industrial design and assembly design. Second, while costs are key variables in product design, costing issues often arise that add more complexities to this concept. The management accounting literature provides activity-based costing (ABC) and target costing techniques to assist product design teams. However, when applied to product design these techniques are often flawed. First, the product “user” and “consumer” are not identical as often assumed in target costing projects, and instead of activities driving up the costs, managers may use budgeted costs to create activities to augment their managerial power by bigger budgets and to protect their subordinates from being laid off. Second, each of the two techniques has a limited costing focus, activity-based costing (ABC) focusing on indirect costs and target costing on unit-level costs. Third, neither technique accounts for resource interactions and cost associations.


151

152

MOHAMED E. BAYOU AND ALAN REINSTEIN

This paper applies the new method of associative costing (Bayou & Reinstein, 2000) that does not contain these limitations. To simplify the intricate procedures of this method, the paper outlines and illustrates nine steps and applies them to a hypothetical scenario, a design of a laptop computer intended for the college-student market. This method uses the well-known statistical techniques of clustering, Full Factorial design and analysis-ofvariance. It concludes that in product design programs, the design team may need to make tradeoff decisions on a continuum beginning with the designto-cost point and ending at the cost-to-design extreme, as when the best perceived design and the acceptable cost level of this design are incongruent.

INTRODUCTION Since quality cannot be manufactured or tested into a product but must be designed in, effective product design is a prerequisite for effective manufacturing (Cooper & Chew, 1996, p. 88; National Research Council, 1991, p. 7). But designing and developing new products are a complex and risky process that must be tackled systematically (Roozenburg & Eekels, 1995, p. xi). Ignoring this issue can adversely affect the nation’s competitiveness (National Research Council, 1991, p. 1). In this process, cost is a primary driver (National Research Council, 1991, p. 15; Ruffa & Perozziello, 2000, p. 1). Over 70% of a product’s life-cycle cost is determined during the design stage (Ansari et al., 1997, p. 13; National Research Council, 1991, p. 1). Michaels and Wood (1989, p. 1) elevate cost to “the same level of concern as performance and schedule, from the moment the new product or service is conceived through its useful lifetime.” Yet costing issues often add to the complexity of product design. For example, in the defense industry, Ruffa and Perozziello (2000, p. 161) report that aircraft manufacturers recently stressed the importance of adopting improved design approaches as a means to control product costs. However, cost advantages are often hard to discover, as they (p. 161) state: “Only, when we attempted to quantify the specific cost savings to which these [improved design approaches contributed], we were often disappointed. While it intuitively seems that broader benefits should exist, we found that they are not consistently visible.” How does the managerial accounting literature help in reducing this product-design and costing complexities? In managerial accounting, activity-based costing (ABC) and target costing are often touted as valuable methods of accounting for product design. They help management to develop cost strategies for product design programs to create new products or improve existing ones. While value engineering, value analysis and process analysis techniques identify, reduce or eliminate nonvalue-added

Accounting for Cost Interactions in Designing Products

153

activities during the product’s lifecycle, when applied to product design programs, these methods have serious shortcomings in theory and application. This paper explains the limitations of activity-based costing (ABC) and target costing in product design, and applies a more recent technique, i.e. associative costing (Bayou & Reinstein, 2000), to overcome such limitations. The first section of the paper explains the nature of product design since this concept is vaguely described in the engineering and accounting literatures. The second section discusses the shortcomings of ABC and target costing when applied to product design programs. The associative costing model is then applied to a product design scenario in the third section. Finally, a summary and conclusions are presented.

NATURE OF PRODUCT DESIGN Product design as discussed in the engineering literature needs clarification. Roozenburg and Eekels (1995, pp. 3, 53) define product design as the process of devising and laying down the needed plans for manufacturing a product. This definition shows that product design overlaps with several other types of design. Roozenburg and Eekels (1995, p. xi) consider product design a narrower concept than engineering design, which also includes designing chemical and physical processes. Conversely, product design is a wider concept than industrial design, which focuses on the usage and external appearances of products (Roozenburg & Eekels, 1995, p. xi). In Redford and Chal’s (1994, p. 77) framework, design for manufacturing (DFM) integrates product design and process planning into one common activity. A central element in DFM is the design for assembly (DFA), which addresses product structure simplification, since “the total number of parts in a product is a key indicator of product assembly quality.” Redford and Chal (p. 29) explain that product design affects assembly methods and processes. Since all parts have at least three operations associated with them – grasping, moving and inserting – reducing the number of parts must reduce the number of operations and must result in improved assembly. Understanding the product design concept requires clarifying the word “design,” which Borgmann (1995, p. 13) narrowly views as “the excellence of material objects.” Buchanon and Margolin (1995, p. ix) broaden the subject of design to include social, cultural and philosophic investigations. They argue that no one functional group can authoritatively speak for design; instead, many different disciplines can participate in developing design concepts and practices. In this paper, we adopt Akiyama’s (1991, p. 20) definition, which states that design is “the form and planning the content of some human-made object in accordance with goals or purposes of the object to be made.” He (p. 20) explains

154


that “design” is an activity that: (a) recognizes the goals or purposes of products or systems; (b) shapes its objects – creates their forms – in accordance with the goals or purposes of these objects; and (c) evaluates and determines the forms of its objects and makes their contents universally comprehensible. Both form and content are important in product and service design. “Appearance” is a closely related concept to form, which Niebel and Draper (1974, p. 21) assign a high value when they conclude: “Appearance must be built into a product, not applied to a product.” Appearance then is an important element for both product design (Niebel & Draper, 1974) and for industrial design (Roozenburg & Eekels, 1995; Wood, 1993). These product design issues have important implications for target costing and ABC techniques, discussed as follows.

LIMITATIONS OF TARGET COSTING AND ABC IN PRODUCT DESIGN When applied to product design programs, target costing and ABC have serious fundamental limitations, including the following.

Limiting Assumptions Target costing focuses on a target product. But the nature of this target product from the manufacturer’s viewpoint differs from that of its customers. For example, Morello (1995, p. 69) differentiates between the “user” and the “consumer.” He (p. 70) argues that “[b]oth user and consumer have an explicit or implicit ‘project’ to use with efficacy and efficiency . . . But the project of the user is a microproject, defined by many specific occasions, while the project of the consumer is, relatively, a macroproject, for every possible occasion of use.” He (p. 70) adds: “the only way to design properly is to have the user in mind; and the role of marketing . . . is to have in mind the true project of the consumer, which paradoxically, is not to consume but to be put in the condition to use properly.” Morello’s argument echoes the points made in 1947 by Lawrence D. Miles, the founder of modern value engineering (quoted in Akiyama, 1991, p. 9): Customer-use values and customer-esteem values translate themselves into functions as far as the designer is concerned. The functions of the product . . . cause it to perform its use and to provide the esteem values wanted by the customer. The use values require functions that cause the product to perform, while the esteem values require functions that cause the product to sell.

Therefore, a target-costing project cannot assume that the characteristics of “user” and the “customer-use value” always coincide with those of the “customer” and


155

“customer-esteem values.”1 Cost reduction focuses on driving this assumption in practice, since when the user and customer are the same, and when the customeruse value and customer-esteem value are identical, less need for function analysis, design and manufacturing arises. This assumption is also not inconsistent with such a view as: “Target costing is not a management method for cost control in a traditional sense, but it is one which intends to reduce cost” (Monden & Hamada, 1991, p. 18). It also underlies the well-known concept of “design to cost.” The essence of this concept boils down to “making design converge on cost instead of allowing cost to converge on design” (Michaels & Wood, 1989, p. 1). Nevertheless, this assumption may be responsible for many product problems. As Akiyama (1991, p. 18) explains, product problems increase when the gap between the manufacturer’s intention and customers’ desires widens, as an increasing number of companies have experienced. In turn, ABC has a fundamentally limiting assumption, especially since ABC stresses that activities are the drivers of costs as the name, activity-based costing, implies. However, in practice, the driving force may work in the opposite direction. Some managers, protective of their subordinates and determined to keep or increase their power through increasing their departmental budgets, may “create” activities for their employees to help justify their budget demands to top management. Hence, budgeted costs become the driver to create new activities or intensify the existing ones. That is, an activity-based costing may instigate budgeted cost-based activities as a reaction to the often-expected and sharp ABC recommendations: reduce (non-value-added) activities and consequently reduce the manager’s budget. In effect, two systems exist, an accounting (or consultant) ABC and an actual ABC. Paradoxically, an accounting ABC system installed to reduce unneeded activities may drive managers to create new (possibly unneeded) activities to keep or augment their power.

Limited Costing Focus ABC focuses primarily on indirect (manufacturing overhead) costs. While indirect costs often form a substantial portion of a product or service’s total cost, they exclude engineered costs that enter directly into the fabric of the product, e.g. direct raw materials, components, modules, interfaces and direct labor. These unitlevel costs play an essential role in designing the product form (appearance) and substance. Target costing inherits a similar limitation, but with a different focus. Value engineering, the primary tool of target costing, focuses primarily on direct (engineered) costs, and less on indirect costs. For example, value engineers determine the types

156


and combinations of parts of an automobile and may use time and motion studies to determine the standard direct labor time and cost allowed for assembling a vehicle. However, batch-level costs, e.g. machine setups, and facility-level costs, e.g. factory cafeteria, factory security, facility cleaning and maintenance costs, are difficult to incorporate into the design of a unit of a concrete product.2 In short, ABC focusing on indirect costs and engineered target costing focusing on unit-level costs render them, individually, incomplete costing systems for product design purposes.

Ignoring Strategic Interaction Effects ABC and target costing do not account for strategic interactions among resources, activities and their costs. For example, maintenance and testing activities frequently interact so much so that a reduction in maintenance activities can lead to more defective output units, which in turn may necessitate increased testing activities. Yet, for practical reasons, ABC models do not account explicitly for these interactions among activities. With only four groups of different activities, each at two levels, high and low, 11 interactions (24 – 4 activity groups – 1) among these activities to account for would arise, as explained in the following section. When considering that the median number of activity-area-based cost pools in practice is 22 (Innes et al., 2000, p. 352), accounting for the activity interaction effects becomes even more impractical. ABC also is a cost traceability model where costs are traced to the cost object. This leap of costing from input to output bypasses the manufacturing process where resources interact and costs associate (Bayou & Reinstein, 2000, p. 77). Similarly, value-engineering programs often do not account formally for input interactions and cost associations. This weakness of target costing systems when applied to product design arises, for example, with the type of metal (e.g. steel vs. aluminum) that enters into the production of an automobile, which must be associated with such other costs as fuel consumption, environmental problems, safety and price fluctuations of the metal (de Korvin et al., 2001). A method that does not contain these target-costing and ABC limitations is shown below.

AN APPLICATION OF THE ASSOCIATIVE COSTING MODEL The associative costing model focuses on main factors and interaction effects among these factors. This model “allows cost interactions to be designed, planned and controlled in order to help apply the application of process-oriented thinking


157

and realize its continuous improvement goals” (Bayou & Reinstein, 2000, p. 75). The ultimate goal of this model when applied to product design is to guide design engineers and management in determining the optimum product design on the basis of associating the most important factors, the interactions among these factors and the costs of their combinations. Following Bhote and Bhote (2000, p. 93), we label the most important factors Red X, the second-order factors Pink X, the third-order factors Pale X and the dependent variable, i.e. the output of each combination, Green Y. Table 1 lists the basic steps of applying this model. The associative costing model employs well-known statistical methods of clustering, classification and analysis of variance as illustrated by the following hypothetical scenario. The object of design is a new model of a laptop computer targeted for purchase by college students. The product has many elements that Table 1. Basic Steps of the Associative Costing Model Application. Step 1:

Factor list. Compile an exhaustive list of all factors to be considered in product design of a new product model or improving an existing one. Illustration: Components: {A1 , A2 , . . . An . . . B1 , B2 , . . . Bn , . . . Z1 , Z2 , . . . Zn }

Step 2:

Clustering: Using a clustering technique, classify the factors of Step 1 into homogeneous groups of factors. Illustration: Direct Material: {C1 , C2 , . . . C9 }

Step 3:

Choose an appropriate output (Green Y) measure. Quantify this measure using a 1–9 Likert scale. Illustration: Green Y: Degree of willingness to purchase the product.

Step 4:

For each cluster, identify the Red X, Pink X and Pale X factors on the basis of the Green Y measure using an appropriate methodology. Illustration: • Red X factors: C1 , C3 and C7 • Pink X factors: C2 , C4 and C6 • Pale X factors: C5 , C8 and C9

Step 5:

Apply the full factorial method if a cluster has four or fewer Red X factors. This step produces a list of experiments, each with a Green Y value (see Table 4).

Step 6:

Using two cost values, low and high, for each factor, determine the cost of each experiment combination (see Table 4).

Step 7:

Summarize the results of Step 5 in terms of Red X, Pink X and Pale X factors, their main effects, their significant interactions, their Green Y values and the corresponding cost of each experiment.

Step 8:

On the basis of Green Y and cost, choose the optimum combination.

Step 9:

Cluster of clusters design: Treating the optimum combination of each cluster as a factor, repeat Steps 5–8. The result of this step helps ranking the multitude of clusters in terms of importance on the basis of their Green Ys and costs.

158


can have low and high values, e.g. the RAM size, computing capability, number and kind of software packages installed on a unit, quality of material for the frame and carrying case, and the electronic screen. The following discussion applies the nine steps listed in Table 1 to the new laptop model.

Step 1: Factor List This step develops an exhaustive list of factors or components. The list can contain many components since many types of resources are needed to design, manufacture and deliver a product to customers. The product design team has several methods to compile this list. As a starting point, the method of reverse engineering of competitors’ products can provide insights to differentiate and improve on competitors’ models. Another method is the rapid automated prototyping (RAP), which is a new field that creates three-dimensional objects directly from CAD files, without human intervention. According to Wood (1993, p. 1), prototypes, which are integral to the industrial design cycle, have three purposes: (1) Aesthetic visualization – to see how the product appears, especially a consumer item that must look appealing when printed or packaged. (2) Form-fit-and-function testing – to ascertain that the part fits and interfaces well with other parts. (3) Casting models – to make a casting model around the part for full-scale production of replicas of the part. The prototypes can enhance the design team’s imagination and enliven their brainstorming sessions. To illustrate, this step develops a list of components: A1 , A2 , . . . An , B1 , B2 , . . . Bn , . . . Z1 , Z2 , . . . Zn (Table 1), as explained in Step 2.

Step 2: Clustering Clustering means grouping of similar objects (Hartigan, 1975, p. 1). Its principal functions include naming, displaying, summarizing, predicting and seeking explanation. Since “clustering” is almost synonymous with classification in that all objects in the same cluster are given the same name, data are summarized by referring to properties of clusters rather than properties of individual objects (Hartigan, 1975, p. 6). The concept of “similarity” among members of a cluster is crucial in a clustering approach (Kruskal, 1977, p. 17). Good (1977) provides many dimensions to describe alternative clustering approaches.


159

But accountants who emphasize the scientific statistical methods to identify and measure the Red X, Pink X, Pale X and their interaction effects for product design should note two major issues of clustering (Hartigan, 1975, p. 8). First, design experts may insist that fancy data manipulations are not better than subjective detailed knowledge. Second, clustering techniques themselves lack sound probability models and their results are poorly evaluated and unstable when evaluated. We argue that informed decision making requires using both a clustering approach that may rely greatly on detailed (subjective) knowledge and experience along with the statistical testing of the main effects and interaction effects among elements of chosen clusters, as demonstrated below. The hierarchical/nonhierarchical facet is one common dimension of clustering in Good’s (1977, p. 88) list. The value chain, which starts with R&D and ends with customer services during the product lifecycle is a hierarchical dimension. Each major section of the value chain is a cluster of items with similar functions. In turn, each cluster is divided into sub-clusters; each sub-cluster is segmented further into sub-clusters; and so on. For example, manufacturing is a major cluster in the value chain, consisting of direct material, direct labor, variable manufacturing overhead and fixed manufacturing overhead sub-clusters. In turn, the direct-material sub-cluster includes raw materials, in-house modules, outsourced modules, interfaces and other parts sub-clusters. Each such sub-cluster can branch out into more detailed sub-clusters, depending on the design team’s requested degree of detail. To sum, using the value chain’s hierarchical structure, Step 2 develops six clusters (R&D, Design, Manufacturing, Marketing, Administrative and Customer Service). The Manufacturing cluster is classified into four sub-clusters (Direct Materials, Direct Labor, Variable Manufacturing Overhead and Fixed Manufacturing Overhead). In turn, the Direct Material sub-cluster is analyzed into components, C1 , C2 , . . . Cn that enter into the design of the laptop computer (Table 1).

Step 3: Choosing and Quantifying the Green Y Green Y is the output whose selection and measurement depend on the design team’s views and corporate goals. To illustrate, the design team decides that in order to be consistent with the company’s goals to maximize sales revenues and market share, the output (Green Y) should be defined as the potential customer’s degree of willingness to buy the product. Using a 1–9 Likert scale, the responses are defined as follows: 1 = Definitely, Item Ci will NOT affect my decision to buy the product. 9 = Definitely, Item Ci will affect my decision to buy the product.

160


Respondents indicate their perception on the assumption that the price of the product in question is affordable.3

Step 4: Red X, Pink X, and Pale X Identification This step identifies critical factors in the chosen cluster. There are several ways to conduct this step. In each of the following methods, respondents receive a questionnaire seeking their degree of willingness to buy the product: (1) Descriptive method: The basis for respondents’ judgments is a description of several versions of the product, by varying one element at a time. This is the least expensive method; yet it is also the weakest since respondents do not physically examine the different versions of the product. (2) The CAD prototype method: Respondents examine several versions of a threedimensional CAD replica (Wood, 1993), on which they express their willingness to buy or not buy. This method is useful when the appearance of the product or its elements is a key factor in their purchasing decision. (3) The actual prototype method: Responses are based on an actual version of the product. This method is the most effective because perception is based on a real product; yet it is the most expensive since it requires producing several product versions and experiments, by varying one element at a time. For example, respondents compare two laptops, one basic with RAM of only 32 MB and one advanced, with 64 MB, then with 128 MB, then with 256 MB and so on. This is the method we apply in the following illustration since it is usually the most effective in designing such a relatively expensive (for many students) product as a laptop computer. We consider three samples of 30 college students each, from a private school (PS), a small state school (SS) and a large state school (LS). To test the degree of importance of each component (factor), Ci , in the component list of Step 2, a statistical inference for one mean with normal distribution and 95% significance level is applied to test the following hypothesis: Ho :

␮≤5

Ha :

␮>5

For Component C i ,

While a mean response equal to or less than 5 for the Green Y indicates a low degree of perceived importance, an average response greater than 5 denotes a high degree of importance. Table 2 applies this statistical procedure for Component C1 . A similar table can be developed for each of the C1 –C9 components of Step 2.


161

Table 2. Testing Component C1 Hypothesis. Statistical Inference: One Mean with Normal Distribution Input Data Respondent

Responses Groups PS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Statistical output: Sample statistics Sample means Sample Std. Dev. Sample size Point estimate

4 6 9 9 6 8 9 6 9 8 9 8 8 5 6 9 7 8 9 9 8 7 8 9 8 9 7 8 8 9 6.5 3.5355 30 6.5

SS 8 6 5 8 6 9 8 9 5 8 6 8 9 4 7 6 8 6 8 8 7 9 6 7 6 4 6 9 7 7

0.645

5 7 9 6 7 8 7 9 9 7 6 8 6 8 9 9 8 9 7 9 6 9 7 8 8 7 9 6 8 7

7.5 0.7071 30 7.5

6.0 1.414 30 6.0

0.129

0.258

Confidence interval: Confidence level critical zone Standard error of point estimate

LS

0.012

162


Table 2. (Continued ) Statistical Inference: One Mean with Normal Distribution Input Data Respondent

Responses Groups PS

SS

LS

Half width of conf. Lower level Upper level

0.008 6.492 6.508

0.002 7.498 7.502

0.003 5.997 6.003

Hypothesis test Test statistic Z-critical value

2.324 1.645 0.010

19.365 1.645 0.0

3.873 1.645 5.4E-0.5

Note: Significance level: 95%. PS = Private-school student sample; SS = Small state-school student sample; LS = Large state-school student sample. Responses: 1 = Defintely, Component C1 will NOT affect my decision to buy the product; 9 = Defintely, Component C1 will affect my decision to buy the product.

Table 2 shows that the hypothesis testing for Component C1 leads to rejecting the null hypothesis, which means that this component is significantly important. The results of the hypothesis testing of the nine components, C1 –C9 , in this step are as follows (Table 3):

Table 3. Mean and Significant Value of Each Product Component: Perception of Three Samples of Respondents. Component

C1

C2

Mean & p-value at 95% sig. level PS 6.5 5.0 0.00 0.00

C3

C4

C5

C6

C7

C8

C9

7.0 0.01

5.0 0.01

2.0 0.03

4.0 0.00

6.0 0.00

5.0 0.00

3.0 0.02

SS

7.5 0.00

5.0 0.04

8.0 0.00

6.0 0.03

3.0 0.00

5.0 0.01

8.0 0.03

4.0 0.00

3.0 0.03

LS

6.0 0.00 X

5.0 0.00

7.0 0.00 X

6.0 0.01

2.0 0.00

5.0 0.04

7.0 0.00 X

5.0 0.05

1.0 0.01

Red X Pink X Pale X

X

X

X X

X

X

Note: PS = Private-school student sample; SS = Small state-school student sample; LS = Large state-school student sample.


163

Red X factors: C1 , C3 and C7 Pink X factors: C2 , C4 and C6 Pale X factors: C5 , C8 and C9 Furthermore, the three student samples show consistent perceptions of the importance of these nine components.

Step 5: Full-Factorial Experimental Design Most design of experiment (DOE) experts consider the Full Factorial the most pure formal DOE technique because “it can neatly and correctly separate the quantity of each main effect, each second-order, third-order, fourth-order, and higher order interaction effect” (Bhote & Bhote, 2000, p. 282). The Full Factorial requires 2n experiments for a randomized, replicated and balanced design, where n is the number of factors. Thus, if n equals 3, 4, 5, 6 and 10, a Full Factorial design would respectively require 8, 16, 32, 64 and 1,024 experiments. Accordingly, the Full Factorial becomes impractical if the number of factors exceeds four (Bhote & Bhote, 2000, p. 282). The Full Factorial methodology requires selecting two levels for each factor, a low and high level. For n = 3 factors, the number of experiments equals 23 or 8 combinations, where each level of each factor is tested with each level of all the other factors (Bhote & Bhote, 2000, p. 234). Applying the Full Factorial method to the three Red X factors, C1 , C3 and C7 of Step 4, Table 4 shows the ANOVA data and results. To save space, only one sample is considered. The procedure should be repeated, however, for all samples of respondents. To examine the content of Table 4 more closely, respondents are asked to indicate their willingness to purchase each of the eight versions of a laptop on 1–9 Likert scale, where: 1 = Very low degree of willingness to purchase the unit 9 = Very high degree of willingness to purchase the unit The average response for each experiment (or combination) is listed in the output (Green Y) column. In experiment 1, the three Red X factors are held at low levels (each with a –1). This is a laptop version with its Red X factors at the lowest level. The student group’s examination derived an average response of 1, indicating no substantial potential customer demand for this laptop. Demand is strongest for the laptop version of experiment 6, with an average score of 8.

164

Table 4. Full Factorial Experimental Design. ANOVA (Full Factorial) Table Experiment #

Low level total High level total Net

Factor Interactions

C1

C3

C7

C 1 × C3

C1 × C7

C3 × C7

C 1 × C3 × C7

−1 1 −1 1 −1 1 −1 1

−1 −1 1 1 −1 −1 1 1

−1 −1 −1 −1 1 1 1 1

1 −1 −1 1 1 −1 −1 1

1 −1 1 −1 −1 1 −1 1

1 1 −1 −1 −1 −1 1 1

−1 1 1 −1 1 −1 −1 1

−18 21 3

−17 22 5

−11 28 17

−19 20 1

−19 20 1

−24 15 −9

−20 19 −1

Full-Factorial Costing Component C1 C3 C7 Total a

Low (−1)

High (+1)

$50 40 10

$250 180 70

$100

$500

To illustrate, the cost column is computed as follows for the first three experiments:

C1 C3 C7 Total cost

Experiment 1 $50 40 10 $100

Experiment 2 $250 40 10

Experiment 3 $50 180 10

$300

$240

Output (Green Y)

Costa in $

1 1 4 5 7 8 6 7

100 300 240 440 160 360 300 500


1 2 3 4 5 6 7 8

Factors


165

The data in the interaction columns are developed as follows. For the C1 × C3 interaction for experiment 1, the −1 in the C1 column multiplied by the −1 in the C3 column yields +1. For experiment 2, +1 multiplied by −1 under columns C1 and C7 , respectively, yields −1 for the C1 × C3 interaction column. Similar calculations are made for the remaining cells of the interaction columns. The bottom three rows of Table 4 are computed as follows for column C1 : Low-level total: (−1 × 1) + (−1 × 4) + (−1 × 7) + (−1 × 6) = −18 High-level total: (1 × 1) + (1 × 5) + (1 × 8) + (1 × 7) = + 21 Net main effect of factor C1 = +21−18 = +3 Similar calculations are performed for the remaining columns in Table 4. Table 4 shows that factor C7 with a net effect of 17 is clearly the most important Red X. Factors C1 and C3 , with a net result of 3 and 5, respectively, are less important than C7 . Except for the C3 × C7 interaction with a net result of –9, all interaction effects have small net values. This means that C3 and C7 interaction effects are the only ones that must be accounted for in the laptop design. Figure 1 shows the shape of these two factors interaction. Factor interactions are interpreted as follows (Bhote & Bhote, 2000, p. 292):

Fig. 1. Graphing the C3 × C7 Component Interaction.

166


Two parallel lines mean no interaction effect; Two nonparallel lines mean a decided interaction effect; and Two crossing lines (as in Fig. 1), mean very strong interaction effects. These interaction patterns have important implications for costing decisions. In Fig. 1, the best result occurs when components C7 is at the high level and C3 at the low level, as shown in the following steps.

Step 6: Full-Factorial Costing As indicated above, Table 4 shows the two costing levels low (−1) and high (+1) for factors C1 , C3 , and C7 and the results for only one sample of respondents. The last column of Table 4 shows the total cost of each experiment. To explain how these costs are computed, let us consider only three experiments, experiments 1, 2, and 6, in Table 4. The average response for the laptop version of experiments 1, 2 and 6 are 1, 1 and 8, respectively on the 1−9 Likert scale. The costs for these experiments are computed as follows: Component (Factor)

Experiment 1

Experiment 2

Experiment 6

C1 C3 C7

$50 40 10

$250 40 10

$250 40 70

Total cost

$100

$300

$360

Steps 7 and 8: ANOVA Results The ANOVA table (Table 4) shows that experiment 6 is the optimum in terms of the Green Y result. The average response of 8 indicates a high degree of willingness to purchase the laptop version. This experiment may be replicated with different sample groups to develop more confidence in this level of customer perception. The cost of this laptop version is $360 (Table 4). Simplifying these computations helps to illustrate the application of the associative-costing method. In some situations, the product design team may need to make tradeoff decisions where the product version with a high Green Y value may be too costly to produce, and the version with low cost is of too low Green Y value.4 In other words, a continuum exists that


167

begins at the design-to-cost point and ends at the cost-to-design point. Determining the optimum point on this continuum is a multivariate decision problem, which we recommend for further research.

Step 9: Cluster of Clusters Design In Step 8, we find that experiment 6 provides the best combination of factors in the direct-materials cluster. Through grouping the best combinations of the four manufacturing sub-clusters (direct materials, direct labor, variable manufacturing overhead and fixed manufacturing overhead) as explained above, we can perform a DOE on this group by following Steps 5−8. The end result ascertains the best combination of the manufacturing cluster. Similar procedures can determine the best combination for each marketing, delivery, and customer services cluster, thus helping management rank the various value-chain clusters in providing the highest values according to customers’ perception.

SUMMARY AND CONCLUSIONS Activity-based (ABC) and target costing management accounting techniques contain many limitations when applied to product design programs. First are limiting assumptions. Target costing often does not differentiate between differences in “users” and “customers” during the planning of product functions. While ABC assumes that activities drive costs, in practice, managers may create activities for their subordinates to help justify budget requests and protect such employees from losing their jobs; that is, budget costs may instigate the creation of activities. In effect, two ABC systems may exist, an accounting (or consultant) ABC, where activities drive up costs and an actual ABC, where budgeted costs drive managers to create activities. Second, both target costing and ABC have a limited costing focus. ABC focuses on indirect costs and target costing mostly on unit-level costs. This partial cost accounting is insufficient for product design projects where full costs are important considerations. Finally, both ABC and target costing methodologically do not account for cost interactions. This paper applies the new costing paradigm of associative costing (Bayou & Reinstein, 2000) that does not contain these limitations. This method accounts specifically for resource interactions and cost associations and uses the well-known statistical techniques of clustering, Full Factorial design and analysis-of-variance. Applying this method, the paper outlines and illustrates nine steps to design a hypothetical product, a new laptop computer intended for the college-student

168


market. This illustration simplifies the intricate procedures of associative costing. The paper concludes that a design team may need to make tradeoff decisions on a continuum that begins with the design-to-cost point and ends at the cost-to-design extreme. These decisions become necessary when the best perceived design and the acceptable cost level of this design are incongruent. A multivariate decision model to solve this issue is recommended for further research.

NOTES 1. According to Zaccai (1995, p. 6), during the first century of the industrial revolution, many peoples’ most basic needs were met by products narrowly designed by technical specialists. These products, although technically functional, often did not meet consumers’ requirement and “as a result could be undermined by more desirable alternatives.” This problem is magnified today by the variety of products consumers can acquire and the options and methods of financing or ownership (e.g. lease vs. buy) available to them. 2. Target costing is a concept mired in obscurity. First, an examination of Ansari et al.’s (1997, pp. 45–48) delineation of costs for target costing purposes reveals a vague, yet, incomplete set of cost perspectives, which includes:

Value chain (organizational perspective). Life cycle (time) perspective. Customers (value) perspective. Engineered (design) perspective. Accounting (cost type) perspective.

This list contains overlapping functions and illustrates the common problems of functional arrangements. One can add other perspectives, including micro (e.g. competition, demand elasticity, and substitutes) and macro (e.g. industry and the economy) perspectives. Second, the cost in the most common target-costing model, called “the deductive method” (Kato, 1993, p. 36) (where Target Cost = Price − Profit) is a difference, which does not exist for empirical measurement (Deleuze, 1994 [1964]; Sterling, 1970). (For a detailed explanation of the Deleuzian difference in accounting, see Bayou & Reinstein, 2001.) This means that a manufacturer does not and cannot measure (an empirical process) the target cost of a target product. It can only determine this cost. But cost determination is a result of calculation (a rational process) based on the design-to-cost view where design should converge to cost, rather than vice versa, as explained above. In short, target costing is a vague concept. 3. This assumption helps separate the product design from pricing issues. A target price, the ground for the target-costing deductive method, is often a vague notion of affordability, for several reasons. First, in some cases, this price may depend on the design, rather than the design depending on the price; in other cases, the price and design are locked into a circular interdependent relationship. Second, an affordable price is often a fuzzy variable that changes within a wide range (Bayou & Reinstein, 1998). Finally, positing an affordable price is valid when the target group of customers is homogenous in terms of tastes, preferences, income, demand elasticity and available means of financing in the short run and the long run.


169

In the West, this high degree of customer homogeneity is rarely found for many products and services. Shorter: the target price, which is a key element in the target-cost deductive method, is often a vague target. Separating design issues from pricing issues makes the marketing function and component suppliers more crucial in producing and selling the designed product at acceptable profits. 4. This tradeoff decision is consistent with Bayou and Reinstein’s (1997) discussion of the circular price-cost interaction and their “tiger chasing his tail” metaphor.

REFERENCES Akiyama, K. (1991). Functional analysis. Cambridge, MS: Productivity Press. Ansari, S. L., Bell, J. E., & the CAM-I Target Cost Core Group (1997). Target costing, the next frontier in strategic cost management. Chicago: Irwin. Bayou, M. E., & Reinstein, A. (1997, September/October). Formula for success: Target costing for cost-plus pricing companies. Journal of Cost Management, 11(5), 30–34. Bayou, M. E., & Reinstein, A. (1998). Applying Fuzzy set theory to target costing in the automobile industry. In: P. H. Siegel, K. Omer, A. de Korvin & A. Zebda (Eds), Applications of Fuzzy Sets and the Theory of Evidence to Accounting, II (Vol. 7, pp. 31–47). Bayou, M. E., & Reinstein, A. (2000). Process-driven cost associations for creating value. Advances in Management Accounting (Vol. 9, pp. 73–90). New York: JAI Press/Elsevier. Bayou, M. E., & Reinstein, A. (2001). A systemic view of fraud explaining its strategies, anatomy and process. Critical Perspectives on Accounting (August), 383–403. Bhote, K. R., & Bhote, A. K. (2000). World class quality (2nd ed.). New York: American Management Association. Borgmann, A. (1995). The depth of design. In: R. Buchanan & V. Margolin (Eds), Discovering Design: Explanation in Design Studies (pp. 13–22). Chicago: University of Chicago Press. Cooper, R., & Chew, W. B. (1996). Control tomorrow’s costs through today’s designs. Harvard Business Review (January–February), 88–97. de Korvin, A., Bayou, M. E., & Kleyle, R. (2001). A fuzzy-analytic-hierarchical-process model for the metal decision in the automotive industry. Paper N. 01IBECB-2, Society of Automotive Engineers (SAE) IBEC 2001 International Body Engineering Conference and Exhibition, Detroit, Michigan, October, 16–18. Deleuze, G. (1994). Difference and repetition. P. Patton (Trans.). New York: Columbia University Press. Good, I. J. (1977). The botryology of botryology. In: J. Van Ryzin (Ed.), Classification and Clustering (pp. 73–94). New York: Academic Press. Hartigan, J. A. (1975). Clustering algorithm. New York: Wiley. Innes, J., Mitchell, F., & Sinclair, D. (2000, September). Activity-based costing in the UK’s largest companies: A comparison of 1994 and 1999 survey results. Management Accounting Research, 11(3), 349–362. Kato, Y. (1993). Target costing support systems: Lessons from Leading Japanese companies. Management Accounting Research, 4, 33–47. Kruskal, J. (1977). The relationship between multidimensional scaling and clustering. In: J. Van Ryzin (Ed.), Classification and Clustering (pp. 17–44). New York: Academic Press. Michaels, J. V., & Wood, W. P. (1989). Design to cost. New York: Wiley.

170


Monden, Y., & Hamada, K. (1991). Target costing and kaizen costing in Japanese automobile companies. Journal of Management Accounting Research, 3(Fall), 16–34. Morello, A. (1995). ‘Discovering design’ means [re-]discovering users and projects. In: R. Buchanan & V. Margolin (Eds), Discovering Design: Explanation in Design Studies (pp. 69–76). Chicago: University of Chicago Press. Redford, A., & Chal, J. (1994). Design assembly: Principles and practice. London: McGraw-Hill. Roozenburg, N. F. M., & Eekels, J. (1995). Product design: Fundamentals and methods. Chichester: Wiley. Ruffa, S. A., & Perozziello, M. J. (2000). Breaking the cost barrier. New York: Wiley. Sterling, R. (1970). A theory of measurement of enterprise income. Iowa Printing. Wood, L. (1993). Rapid automated prototyping: An introduction. New York: Industrial Press. Zaccai, G. (1995). Art and technology: Aesthetics redefined. In: R. Buchanan & V. Margolin (Eds), Discovering Design: Explanation in Design Studies (pp. 3–12). Chicago: University of Chicago Press.

RELATIONSHIP QUALITY: A CRITICAL LINK IN MANAGEMENT ACCOUNTING PERFORMANCE MEASUREMENT SYSTEMS Jane Cote and Claire Latham ABSTRACT Performance measurement has benefited from several management accounting innovations over the past decade. Guiding these advances is the explicit recognition that it is imperative to understand the causal linkage that leads a firm to profitability. In this paper, we contend that the relationship quality experienced between two organizations has a measurable impact on performance. Guided by prior models developed in distribution channel and relationship marketing research (Cannon et al., 2000; Morgan & Hunt, 1994) we build a causal model of relationship quality that identifies key relationship qualities that drive a series of financial and non-financial performance outcomes. Using the healthcare industry to illustrate its applicability, the physician practice – insurance company relationship is described within the context of the model’s constructs and causal linkages. Our model offers managers employing a causal performance measurement system such as, the balanced scorecard (Kaplan & Norton, 1996) or the action-profit-linkage model (Epstein et al., 2000), a formal framework to analyze observed outcome metrics by assessing the underlying dynamics in their third party relationships. Many of these forces have subtle, but tangible

Advances in Management Accounting Advances in Management Accounting, Volume 12, 171–190 © 2004 Published by Elsevier Ltd. ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12008-X

171

172

JANE COTE AND CLAIRE LATHAM

impacts on organizational performance. Recognizing them within performance measurement theory adds explanatory power to existing performance measurement systems.

INTRODUCTION Performance measurement has benefited from several management accounting innovations over the past decade. Guiding these advances is the explicit recognition that it is imperative to understand the causal linkage that leads a firm to profitability. One of the most significant innovations is the integration of non-financial variables into performance measurement systems. Non-financial performance measures help firms recognize how specific actions or outcomes impact profitability. Some models such as the Action-Profit-Linkage (Epstein et al., 2000) and the Service Profit Chain (Heskett et al., 1997) develop quantitative causal models that demonstrate how a unit change in non-financial variables impact profitability and other financial variables. Other models, such as the Balanced Scorecard (Kaplan & Norton, 1996) represent the interrelationships among financial and non-financial performance variables as a set of causal hypotheses that serve as guideposts for managerial decision making. What all of these performance models have in common is the need for organizations to clearly understand the factors that drive performance outcomes. In this paper, we focus on one specific domain inherent in most of these performance models, inter-organizational transactions, to introduce a causal model that links the quality of inter-organizational relationships to financial and non-financial outcomes. Successful organizations develop exchange relationships with other organizations, that persist over time, to develop a network that provides reliable sources of goods or services. These relationships often develop to accommodate specific strategic goals such as when a manufacturer outsources portions of its value chain. Alternatively, inter-organizational exchange relationships also arise from unique interdependencies that compel organizations to cooperate to produce a product or service. In healthcare, for example, insurance companies and physician practices are dependent upon each other to provide healthcare services to patients. These relationships are recognized as important performance drivers by most causal performance measurement models. For instance, the Balanced Scorecard (BSC), in the internal-business-process perspective, acknowledges the critical role that vendors play in providing quality inputs efficiently. The Action-Profit-Linkage Model (APL) is a framework that “links actions taken by the firm to the profitability of the firm within its market environment” (Epstein et al., 2000, p. 47). A subset of firm actions are the inter-organizational

Relationship Quality

173

relationships developed to enhance the delivered product or service. As such, they form a link in the causal chain between firm actions and profitability which needs to be measured to recognize its role in the organization. Therefore, it is important to understand the causal forces that impact inter-organizational relationship quality. Relationship quality is a function of a series of subtle, but powerful variables that impact the working relationships between two organizations. Often a manager can identify arrangements with other organizations that are positive or negative but not be able to specifically point to the factors that make one relationship successful and the other troublesome. The statistics reported by the performance measurement system may be indicative of insidious actions and relationship structures that are not currently measured, hence overlooked as sources of tension within the relationship. As managers seek explanations for deviations from expected vendor (or any partnering organization) performance, management accounting theory does not guide them towards examination of relationship quality. It has not been an explicit component of performance measurement systems, not due to a failing of management accounting, but rather because the introduction of non-financial performance measures is relatively recent. Now that non-financial performance measures are widely accepted management accounting theory and practice, the opportunity exists to explore the causal factors that form the foundation of successful inter-organizational relationships. Substantial research examining inter-organizational relationships has been conducted in the marketing distribution channel literature (Frazier, 1999; Rindfleisch & Heide, 1997). Numerous dimensions affecting the channel relationships have been identified including, legal contracts, power, conflict, and communication. To explore the connection between distribution-channel research and management accounting performance measurement systems, we introduce an adaptation of the commitment-trust theory of relationship marketing (Morgan & Hunt, 1994) as a relevant model for measuring the qualitative factors that underlie management accounting metrics. The model defines the antecedents to trust and commitment between two organizations and recognizes the inherent causal link between trust and commitment and several outcome measures. As such, it provides a first step towards understanding the core structure and actions that form the foundation upon which inter-organizational relationships are developed and provides insight into how this structure impacts both financial and non-financial outcomes. Management accounting performance measurement systems can benefit from an interdisciplinary approach (Epstein et al., 2000). With respect to relationship quality, the rich research foundation developed in the marketing channel’s domain offers potential for management accounting researchers to enhance the knowledge of the fundamental causality that underlies outcome metrics commonly used

174


in current performance measurement contexts. Relationship quality is formed through a series of subtle forces, many of which are not cognitively recognized in a cohesive framework. Introducing them as a network of constructs that support traditional outcome measures offers managers opportunities to pro-actively examine the underlying structure of their inter-organizational relationships as drivers of the performance they expect from the arrangement. We demonstrate how developing a relationship quality framework can be beneficial in a managerial setting and how it can impact financial and non-financial performance measures using the healthcare industry as an example. Relationship quality among organizations in the healthcare value chain is at a tenuous state. Specifically, many physicians cite difficulties with insurance companies as a major impediment to patient care (Pascual, 2001; Sharpe, 1998a, b; Shute, 2002). Many of the problems center on procedure reimbursement and authorization. Prior research demonstrated that, within one physician practice, substantial variance in these performance metrics among insurance companies and that elements of relationship quality are associated with the outcomes (Cote & Latham, 2003). Thus, the physician-insurer relationship is a timely and relevant context to illustrate the applicability of a relationship quality framework in management accounting performance measurement systems. The remainder of the paper is organized as follows. The next section provides the conceptual foundation for the model of relationship exchange quality, identifying the key mediating variables, antecedents and outcomes and discussing hypothesized direction of effects. We then demonstrate applicability of the model to the healthcare exchange relationship between physicians and insurers, including the steps taken in the model refinement and validation process. The paper concludes with a discussion of potential directions for future research.

CONCEPTUAL FOUNDATION To build a model of relationship quality that augments current performance measurement frameworks, several studies drawn from the marketing literature provide direction. For instance, as a starting point, legal contracts are viewed as one of the primary governance structures that safeguard an exchange while maximizing benefits for the relationship partners. In their “plural form” thesis, however, Cannon et al. (2000) argue for viewing the contract as just one of a variety of mechanisms that provide the building blocks for governance structures in relationships; that focusing on the legal contract alone is a deficient approach to governing modern exchanges. Their research on purchasing professionals examines the interaction of contracts and relationship norms in various contexts and demonstrates that


175

legal bonds and social norms were effective, in combination, in enhancing performance. Expanding further the definitions and interplay of social norms, the commitment-trust model of relationship marketing developed in Morgan and Hunt (1994) provides the pivotal theory that can guide management accounting’s understanding of the correlation between such relationship building activities and performance metrics. For organizations to be competitive they must have a network of cooperative relationships comprised of partners such as suppliers, service people, customers, and investors (Solomon, 1992). Successful relationship building is advantageous because it lowers costs, improves quality and timely response to organizational needs. Morgan and Hunt (1994) identify commitment and trust as two key mediating variables in their model of relationship success. A new model of relationship quality is presented here which builds on the work of Cannon et al. (2000) and Morgan and Hunt (1994) that adds insights to existing management accounting performance measurement systems. Figure 1 identifies antecedent variables comprised of contracting and normative, tangible and intangible, constructs: legal bonds, relationship termination costs, relationship benefits, shared values and communication. These antecedents are shown influencing the core of the relationship, commitment and trust. Commitment and trust, as mediators, are then shown to have an effect on relationship quality outcomes that describe whether a relationship is viewed as successful or problematic: acquiescence, propensity to leave, cooperation, financial statement impact, functional conflict and uncertainty. A discussion of commitment and trust, the antecedents and their respective proposed influence on these mediators, as well as relationship outcomes follows.

Commitment and Trust: Mediating Variables Morgan and Hunt (1994) posit that the key mediating variables in a relational exchange are commitment and trust. Relationship commitment is defined as “an exchange partner believing that an ongoing relationship with another is so important as to warrant maximum efforts at maintaining it; that is, the committed party believes the relationship is worth working on to ensure that it endures indefinitely” (Morgan & Hunt, 1994, p. 22). Relationship trust exists when one exchange partner “has confidence in an exchange partner’s reliability and integrity” (Morgan & Hunt, 1994, p. 23). Drawing from the definitions above, these two constructs are positioned as mediating variables given their central role in influencing partners to: (1) preserve

176 JANE COTE AND CLAIRE LATHAM

Fig. 1. Trust and Commitment Model of Relationship Quality.


177

the relationship through cooperation; (2) favor a longer time horizon or working to ensure it endures; and (3) support potential high risk transactions in the exchange given the partners’ beliefs that neither will act in an opportune fashion. The authors further note that trust is a determinant of relationship commitment, that is, trust is valued so highly that partners will commit to relationships which possess trust. Thus, they theorize that the presence of both commitment and trust is what separates the successful from the failed outcomes. Building commitment and trust to reach relationship marketing success requires devoting energies to careful contracting, specific cooperative behaviors and other efforts that both partners invest. We now turn to our discussion of these antecedents.

Antecedents Legal Bonds Legal bonds or legal contracting refers to the extent to which formal contractual agreements incorporate the expectations and obligations of the exchange partners. A high degree of contract specificity, as it relates to roles and obligations, places constraints on the actions of exchange partners. It is this specificity and attention to detail that typically supports a willingness by partners to invest time in an exchange relationship. Exchange partners who make the effort to work out details in a contract have a greater dedication to the long-term success of the partnership (Dwyer et al., 1987). Thus, a higher degree of contract specificity is expected to have a positive influence on relationship commitment. Relationship Termination Costs Relationship termination costs refer to the expected losses from dissolution and such costs are widely defined in the literature. For example, Heide and John (1992), refer to specific investments made by the agent in principal-agent relationships and replacements of commission income. Anderson and Narus (1990) view it as a measure of relative dependence (e.g. there are other manufacturers available to firm x who sell product lines comparable to those of our company). Morgan and Hunt (1994) include non-economic costs as the loss of social satisfaction from the association as well as the socio-psychological costs of worry, aggravation and perceived loss of reputation or face. In essence, relationship termination costs are switching costs. A higher measure of switching costs presents a deterrent to ending the relationship and strengthens the perceived value of committing to the relationship. Hence, relationship termination costs will have a positive correlation with relationship commitment.

178


Relationship Benefits Firms that receive superior benefits from their partnership relative to other options will be committed to the relationship. As with relationship termination costs, partnership benefits have been measured along many dimensions. For example, Morgan and Hunt (1994) capture relationship benefits as an evaluation of the supplier on gross profit, customer satisfaction and product performances. Alternatively, Anderson and Narus (1990) discuss benefit as satisfaction from the perspective of whether the company’s working relationship with the exchange partner, relative to others, has been a happy one. Finally, Heide and John (1992) refer to the “norm of flexibility” where parties expect to be able to make adjustments in the ongoing relationship to cope with changing circumstances. Morgan and Hunt (1994) propose that benefits which relate to satisfaction and/or global satisfaction generally show a strong relationship with all forms of commitment. It is then expected that as the benefits to the relationship increase, relationship commitment will be stronger. Shared Values Shared values are “the extent to which partners have beliefs in common about what behaviors, goals, and policies are important or unimportant, appropriate or inappropriate, and right or wrong” (Morgan & Hunt, 1994, p. 25). Dwyer et al. (1987) note that contractual mechanisms and/or shared value systems ensure sustained interdependence. Shared values are shown to be a direct precursor to both relationship commitment and trust, that is, exchange partners who share values are more committed to their relationships. Communication Communication refers to the formal and informal sharing of “meaningful and timely information between firms” (Anderson & Narus, 1990, p. 44). Mohr and Nevin (1990) note that communication is the “glue” that holds a relationship together. Anderson and Narus (1990) see past communication as a precursor to trust but also that the building of trust over time leads to better communication. Hence, relationship trust is positively influenced by the quality of communication between the organizations. Opportunistic Behavior Opportunistic behavior is “self-interest seeking with guile” (Williamson, 1975, p. 6). Opportunistic behavior is problematic in long-term relationships affecting trust concerning future interactions. Where opportunistic behavior exists, partners no longer can trust each other which leads to decreased relationship commitment. We therefore expect a negative relationship between opportunistic behavior and trust.


179

In summary, trust and commitment is a function of specific efforts both organizations invest in the relationship to improve the value they derive from the arrangement. When a long term association is expected many organizations recognize the benefits that come from developing a strong bond of trust and commitment. For the effort to be worthwhile both must recognize substantial benefits from their joint association and have some common views related to the values they employ in business conduct. Perceptions of opportunism on either side will dampen the potential for trust within the relationship. Alternatively, where switching costs related to developing substitute relationships are substantial, partners will make more concerted efforts to maintain commitment to the existing dyad. Energies devoted to legal contracting and communication then serve to strengthen the commitment and trust bonds. We now turn to the outcomes observed through the presence of trust and commitment in the relationship.

Outcomes Acquiescence Acquiescence is the extent to which a partner adheres to another partner’s requests (Morgan & Hunt, 1994). This is an important construct in relationship quality because when organizations are committed to successful relationships, they recognize that the demands made by each other are mutually beneficial. Where requests are perceived as extraordinary, those in a committed relationship are willing to acquiesce because they value the relationship. Morgan and Hunt (1994) found support for higher levels of acquiescence in highly committed relationships. Propensity to Leave Commitment creates a motive to continue the relationship. The investments to create the committed relationship, described as the antecedents in the model, directly impact the perceptions that one or both partners will dissolve the relationship in the near future. Partners in relationships expected to terminate in the near term behave differently than those that perceive that both are invested in the relationship for the long term. Thus propensity to leave, resulting from the level of relationship commitment, is an outcome variable with performance implications. Financial Statement Impact Activity based costing has successfully demonstrated that inter-organizational arrangements have heterogeneous effects on profitability (Shapiro et al., 1987). Intuitively most managers recognize differential financial impacts among their third party interactions and recently many have begun to strategically structure terms with these organizations to enhance the financial benefits (Morton, 2002).

180


Similarly, relationship quality can be expected to have direct and indirect effects on revenues and expenses. Specifically, we propose that the levels of trust and commitment will have a positive impact on financial performance indicators. Trust has been previously defined as “confidence in an exchange partner’s reliability and integrity” (Morgan & Hunt, 1994, p. 23). With a trusting relationship, the partners do not need to continually verify adherence with agreed upon arrangements and procedures. Hence the need for costly monitoring systems is diminished. Likewise, commitment or “the enduring desire to maintain a valued relationship” (Morgan & Hunt, 1994, p. 23), can impact profitability. When a longer term relationship is expected, there are incentives for organizations to provide each other with favorable terms. For instance, favorable pricing, delivery or service terms may be present within committed relationships because the partners are confident that throughout the relationship a variety of benefits will flow in both directions. Alternatively, when relationship commitment is low fewer incentives exist to offer favorable financial terms or services. This behavior is evident in situations where one exchange partner is considered a “backup supplier,” contacted only when other more favorable exchange partners are not available (Kaplan, 1989). In these circumstances, managers must either negotiate to improve relationship commitment or they must evaluate the implications for creating an alternative working relationship. Both trust and commitment are important forces influencing profitability. The financial statement impact may enhance revenues or reduce costs depending on the environmental circumstances. Recognizing the role trust and commitment perform in a performance measurement system has the potential to improve the causal linkage between firm actions and profitability. Cooperation Cooperation refers to the exchange parties working together to reach mutual goals (Anderson & Narus, 1990). Even if partners have ongoing disputes concerning goals, they will continue to cooperate because both parties’ termination costs are high. Cannon et al. (2000) use the term “solidarity” which encompasses “the extent to which parties believe that success comes from working cooperatively together vs. competing against one another” (Cannon et al., 2000, p. 183). Though both are outcome variables, Morgan and Hunt (1994) point out that cooperation is proactive in contrast to acquiescence which is reactive. Organizations committed to relationships and trusting of their partners, cooperate to make the relationship work. Once trust and commitment is established, exchange partners will be more likely to undertake high-risk coordinated efforts (Anderson & Narus, 1990) because they believe that the quality of the relationship mitigates the risks.

Antecedent Variable

Mediating Variable

Direction

Impact on Mediating Legal bonds

Relationship commitment

+

Relationship termination cost


+

Relationship benefits


+

Shared values


+

Shared values

Trust

+

Communication

Trust

+

Opportunistic behavior

Trust

−

Mediating Variable

Outcome Variables

Impact on outcomes Relationship commitment

Acquiescence

+


Propensity to leave

−


Cooperation

+

Exchange partners having a higher degree of contract specificity have a greater commitment to the relationship. Exchange partners having a higher measure of relationship termination costs have a greater commitment to the relationship. Exchange partners possessing a higher measure of relationship benefits have a greater commitment to the relationship. Exchange partners possessing a higher measure of shared values have a greater commitment to the relationship. Exchange partners with a higher measure of shared values have greater relationship trust. Exchange partners with a higher degree of formal and informal communication have greater trust. Exchange partnerships where a higher degree of opportunistic behavior exists have less trust.


Table 1. Variables and Proposed Direction of Effect.

Direction

Exchange partners who have higher measure of relationship commitment are more willing to make relationship-specific adaptations (higher measure of acquiescence). Exchange partners who have a higher measure of relationship commitment are less likely to end the relationship. Exchange partners who have a higher measure of relationship commitment are more likely to cooperate.

181

182

Table 1. (Continued ) Antecedent Variable

Mediating Variable

Direction

Cooperation

+

Trust

Functional conflict

+

Trust

Decision-making uncertainty

−

Exchange partners who have a higher measure of trust are more likely to cooperate. Exchange partners who have a higher measure of trust are more likely to resolve disputes in an amicable manner (functional conflict). Exchange partners who have a higher measure of trust are less likely to have decision-making uncertainty.


Trust


183

Functional Conflict The resolution of disputes in a friendly or amicable manner is termed functional conflict is a necessary part of doing business (Anderson & Narus, 1990). Morgan and Hunt (1994) show that trust leads an exchange partner to believe that future conflicts will be functional, rather than destructive. When an organization is confident that issues that arise during the conduct of their arrangement with the other organization will be met with positive efforts to reach a mutual solution, they perceive the relationship quality to be higher and expect tangible benefits to result. Uncertainty Decision-making uncertainty encompasses exchange partners’ perceptions concerning relevant, reliable, and predictable information flows within the relationship. The issue relates to whether the exchange partner is receiving enough information, in a timely fashion, which can be then used to confidently reach a decision (Achrol, 1991; Morgan & Hunt, 1994). Cannon et al. (2000) conclude that uncertainty creates information problems in exchange. They further divide uncertainty into external and internal, where external refers to the degree of variability in a firm’s supply market and internal refers to task ambiguity. Morgan and Hunt (1994) support a negative relationship between trust and uncertainty. The trusting partner has more confidence that the exchange partner will act reliably and consistently. In summary, we have discussed the antecedent variables legal bonds, relationship termination costs, relationship benefits, shared values and communication. These antecedents have been shown to influence commitment and trust. Commitment and trust, as mediators, are then shown to have an effect on relationship quality outcomes acquiescence, propensity to leave, cooperation, financial statement impact, functional conflict and uncertainty. Table 1 provides an overview of this discussion of variables and direction of effect.

APPLYING THE MODEL TO THE HEALTHCARE INDUSTRY The relationship quality model has wide applicability in a variety of industry and organizational contexts. We demonstrate its relevance using the healthcare industry because currently, issues related to relationship quality and the implications for healthcare funding and access are primary concerns for many healthcare practitioners. Specifically, we select the physician practice – insurance company relationship as the inter-organizational dyad that best illustrates relationship quality and its impact on performance measurement metrics.

184


The physician – insurer relationship is symbiotic. Physician practices gain improved patient access and simplified accounts receivable procedures through their arrangements with insurance companies. Patients without health insurance often seek only emergency healthcare, complicating comprehensive patient care. Prior to the widespread use of health insurance, physician practices collected payment in full from each patient directly and many physicians were reluctant to demand payment from financially strapped patients causing erratic cash flow for the physician practice. Concurrently, the insurance companies benefit from developing a relationship with the physician practice because they achieve economies of scale through their integrated systems for authorization and payments. Though symbiotic, the relationship is replete with friction. Rapid changes in the healthcare funding system have led physicians to perceive loss of autonomy with corresponding demands to re-direct their attention from patients towards profits (Pascual, 2001; Sharpe, 1998a, b; Shute, 2002). Insurers, caught between employers seeking to control health insurance costs, physicians seeking to provide quality healthcare, and patients seeking access, have placed stringent controls on their funding systems to achieve operating efficiencies. These systems have created friction between the physician – insurer relationship that will require complex and comprehensive solutions. Prior research has established that relationship quality is a critical component that distinguishes successful from unsuccessful healthcare partnerships (Burns, 1999; Cote & Latham, 2003). It is then this inter-organizational dyad that provides a rich context for exploring the application of the relationship quality model as a causal link in performance measurement systems.

Antecedents To be successful, physician practices must contract with a broad selection of insurance providers. Each insurer has unique procedures and systems requiring separate legal contracts that detail the terms of the relationship. The contract forms the basis for each interaction requiring substantial investment from both sides to negotiate terms (Cannon et al., 2000; Cote & Latham, 2003; Leone, 2002). It is through this process that the physician practice and insurer define the legal level of commitment. Relationship benefits and termination costs become relevant constructs for physicians and insurers. From the physician’s perspective, the larger insurers cover a substantial fraction of the patients within their geographical area, necessitating willingness for the physician practice to invest substantial efforts to assure the relationship is successful. Likewise, there are often large physician groups that insurers need to be associated with in order to compete within a geographical


185

area. These environmental characteristics create substantial termination costs and relationship benefits that motivate the physician and insurers to develop a long term, committed relationship. Relationships between physicians and insurers often break down or endure substantial friction due to mis-matched values. Expectation gaps concerning procedure authorization, reimbursement, and general patient care are evidence that the physician and insurer do not completely share each others values in healthcare delivery. When it occurs physician practices often must make repeated oral and written contact to convince insurers to acquiesce to their position. As this conflict is replicated over a series of patients, trust begins to deteriorate and the physician practice begins to assess their level of commitment to the insurer. When values are aligned, both the insurer and physician practice are confident that judgments made by one side will be accepted by the other and the interactions are relatively seamless. Trust in the physician – insurer relationship is influenced both by communication and opportunistic behavior. Communication occurs frequently through procedure authorizations and receivable claims and periodically through practice management advice, processing updates, and office visits. Some insurers provide consistently accurate responses to physician practice inquiries, leading the practice to trust the insurer (Cote & Latham, 2003). Others give conflicting advice, dependent on the insurance representative responding to the inquiry. This destabilizes the relationship, forcing the practice to make multiple inquiries to a single issue and document each interaction precisely, creating decision-making uncertainty. Opportunistic behavior is exemplified in claims processing experiences. Receivable turnover is legally defined, in number of days, by most state insurance commissioners. An insurer must remit payment on a “clean claim” within the statutory period. Clean claims are those with no errors, regardless of the source of the error (Cote & Latham, 2003). If an error is detected, the statutory time period is reset to the beginning. Insurers acting opportunistically will return claims to the physician practice frequently with small errors or errors emanating from their own electronic processing system, thus extending the statutory receivable turnover period. When this happens consistently with an insurer, physician practices begin to doubt the sincerity of insurers’ behavior.

Outcomes Similar to the antecedent constructs, the physician practice – insurer relationship experiences the outcomes as pertinent issues driving the structure of both the relationship and operating systems. For instance, there is a trend whereby physician

186


practices eliminate their relationships with insurers, creating a practice structure that is analogous to a law firm (Pascual, 2001; Sharpe, 1998a, b; Shute, 2002). Patients pay a retainer for immediate access to the physician. The physician accepts cash for services and patients must seek insurance reimbursement on their own. This represents the extreme case where trust and commitment have dissolved and the physician has refused to acquiesce to insurers’ demands and completely left the system. Most physician practices have not resorted to such extremes, yet are still influenced by the model’s outcomes. Cooperation, functional conflict, and decision making uncertainty are ever present in the physician – insurer relationship. As stated earlier, the relationship is symbiotic; each needs to cooperate with the other to provide patient care. Often the physician practice administrators can trace specific issues related to cooperation and conflict back to the level of trust with the insurer (Cote & Latham, 2003). Patient care is complicated, with each patient having unique needs. In a trusting relationship where there is a high degree of confidence that the insurer is reliable and will respond faithfully to patient cases, the physician practice can predict how certain treatment options will be handled. Without trust, there is a degree of randomness in the responses from the insurer, making it difficult for the practice to prepare inquiries to the insurer and anticipate their success. Practice administrators acknowledge revenue and cost heterogeneity among insurers (Cote & Latham, 2003). For instance, approval for a particular medication, termed formulary, must be obtained from each insurance company to assure that it will be a covered expense. Some insurers require extensive paperwork prior to formulary approval, whereas others use a more streamlined approach. Time and paperwork create a measurable financial statement impact for the physician practice. Relationship quality as indicated by the levels of trust and commitment built within the relationship are often factors affecting the ease with which such authorizations are accomplished. In summary, the physician – insurer relationship is built on trust and commitment. This model describes the factors inherent in the relationship and provides a causal chain that indicates how these factors interact to form relationship quality. The model is particularly relevant to the healthcare industry because it is highly dependent on a network of inter-organizational alliances. From a performance measurement perspective, this model provides managers with the framework for diagnosing the root causes of observed performance metrics.

MODEL REFINEMENT AND VALIDATION Model refinement occurred through an iterative process, from practice consultation to literature review and back over approximately a two year time


187

frame. We worked with a physician-owned practice, which is one of the largest physician groups in its community. It is comprised of 8.5 FTE physicians (full time equivalents) who each see approximately 18–22 patients daily. Six insurance providers – half are preferred provider organizations (PPO’s) and the other half are health maintenance organizations (HMO’s) – represent the majority of their patient base that utilizes employer-provided insurance. Initially the global issue of third party relationships and performance measurement was presented to the practice manager and chief financial officer. The research was motivated by the growing trend towards physician practices terminating all association with insurance companies (Pascual, 2001; Sharpe, 1998a, b; Shute, 2002). Explanations for these strategic moves did not fit with a cost – benefit or transaction cost analysis argument (e.g. Williamson, 1985). Clearly there were underlying frictions affecting the relationship’s performance that were ill-defined. At this first stage the goal was to determine how the physician practice – insurance company relationship functions, the key issues they face in managing the interactions, and to assess correspondence with constructs from prior research. These initial interviews motivated an empirical analysis of billing records to identify the existence of factors differentiating insurance company performance (Cote & Latham, 2003). A second stage of interviews diagnosed the underlying causes of the factors leading to the differential performance among insurance companies. Further review of several parallel literature streams indicated the applicability of the distribution channel model and relationship marketing as analogous to the physician practice – insurance company relationship. In particular, the theory of commitment and trust in third party relationships (Morgan & Hunt, 1994) closely mapped the situations we were observing. Integrating this theory with our observations and corresponding research (e.g. Cannon et al., 2000; Epstein et al., 2000), we adapted prior research findings to develop the model presented in this paper. Further interviews were conducted to assess applicability and refine construct definitions. At this point we had considerable interaction with the physician practice side of the relationship. To ascertain the model’s applicability from the insurance company perspective we conducted an interview with insurance company executives at a large regional organization. This insurer is one of the largest insurance providers in their community, currently representing an average of approximately 20–25% of the patient population in medical practices in the geographical region. They concurred with the physician practice executives that the model was a fair representation of the causal network that defines the dyad’s relationship structure. Coincidently, they had a system and philosophy in place to build trust and commitment with local physician practices and hospitals, with the belief that these investments had a financial return to the insurance company.

188


This iterative process involving literature search and qualitative interviews is an integral step in theory building (Eisenhardt, 1989). It links theory to practice in an organized and thoughtful process. Empirical analysis can then be conducted using construct definitions that have been developed concurrently with observations in practice. Theory developed using this process “is likely to have important strengths like novelty, testability, and empirical validity, which arise from the intimate linkage with empirical evidence” (Eisenhardt, 1989, p. 548).

DIRECTIONS FOR FUTURE RESEARCH AND CONCLUSIONS Existing performance measurement systems, such as APL and BSC, have made two important contributions to management accounting: the integration of nonfinancial metrics and the development of a causal network that links performance outcomes into an interdependent system. We build upon this foundation by introducing a causal model that incorporates relationship constructs that are subtle, but compelling forces affecting organizational performance. As often stated, “if you can’t measure it, you can’t manage it” (Kaplan & Norton, 1996, p. 21). Thus, raising relationship dynamics from the manager’s subconscious to a measurable, interconnected level of awareness gives the manager opportunities to address the root causes directly. The model can be implemented numerous ways depending on the centrality of inter-organizational relationships within the firm. Where there is a strong direct relationship between the quality of inter-organizational relationships and financial success (e.g. the healthcare industry) the model may be employed as a stand alone tool that assists managers in understanding the sources of friction in their relationships. Where inter-organizational relationships are of secondary importance, the model can be nested within current BSC or APL frameworks to provide insight into the underlying causal links between key vendor or customer performance measures. Further research requires empirical testing to assess the strengths of the causal links in the model as well as integrating it with existing performance measurement systems. Generally within the existing literature related to inter-organizational relationships, studies examine one dyad with an industry that has the potential to generalize its findings across multiple environmental contexts. When replicated across multiple industry specific settings it is possible to assess the model’s robustness as well as the quality of the construct definitions and measures. Additional research could also consider the managerial implications. For instance, how does this model improve managerial decision making? Are there defined


189

environmental contexts where this model has greater relevance than others? The model represents an initial step towards measuring the role that relationship dynamics have in organizations by considering them as a part of performance measurement systems. Further research can bring refinements that will build an integrated and insightful perspective to performance measurement systems.

REFERENCES Achrol, R. (1991). Evolution of the marketing organization: New forms for turbulent environments. Journal of Marketing, 55(October), 77–94. Anderson, J. C., & Narus, J. A. (1990). A model of distributor firm and manufacturer firm working partnerships. Journal of Marketing, 54(January), 42–58. Burns, L. R. (1999). Polarity management: The key challenge for integrated health systems. Journal of Healthcare Management, 44(January–February), 14–34. Cannon, J., Achrol, R., & Gundlach, G. (2000). Contracts, norms and plural form governance. Journal of the Academy of Marketing Science, 28(Spring), 180–194. Cote, J., & Latham, C. (2003). Exchanges between healthcare providers and insurers: A case study. Journal of Managerial Issues, 15(Summer), 191–207. Dwyer, P., Schurr, H., & Oh, S. (1987). Developing buyer-seller relationships. Journal of Marketing, 51(April), 11–27. Eisenhardt, K. M. (1989, October). Building theories from case study research. Academy of Management Review, 14(4), 532–550. Epstein, M. A., Kumar, P., & Westbrook, R. A. (2000). The drivers of customer and corporate profitability: Modeling, measuring, and managing the causal relationships. Advances in Management Accounting, 9, 43–72. Heide, J. B., & John, G. (1992). Do norms matter in marketing relationships. Journal of Marketing, 56(April), 32–44. Heskett, J. L., Sasser, W. E. Jr., & Schlesinger, L. A. (1997). The service profit chain: How leading companies link profit and growth to loyalty, satisfaction, and value. New York, NY: Free Press. Kaplan, R. S. (1989). Kanthal (A). Harvard Business School Case #190–002. Harvard Business School Press. Kaplan, R. S., & Norton, D. P. (1996). The balanced scorecard. Boston, MA: Harvard Business School Press. Leone, A. J. (2002). The relation between efficient risk-sharing arrangements and firm characteristics: Evidence from the managed care industry. Journal of Management Accounting Research, 14, 99–118. Mohr, J., & Nevin, J. (1990). Communication strategies in marketing channels: A theoretical perspective. Journal of Marketing, 58(July), 20–38. Morgan, R. M., & Hunt, S. D. (1994). The commitment-trust theory of relationship marketing. Journal of Marketing, 58(July), 20–38. Morton, W. (2002). The unprofitable customer: How you can separate the wheat from the chaf. The Wall Street Journal (October 28), A1. Pascual, A. M. (2001). The doctor will really see you now. Business Week (July 9), 10. Rindfleisch, A., & Heide, J. (1997). Transaction cost analysis: Past, present, and future applications. Journal of Marketing, 61(October), 30–54.

190


Shapiro, B. P., Rangan, V. K., Moriarty, R. T., & Ross, E. B. (1987). Manage customers for profits (not just sales). Harvard Business Review, 65(September–October), 101–107. Sharpe, A. (1998a). Boutique medicine: For the right price, these doctors treat patients as precious. Their consultancy signals rise of a system critics say favors the wealthy. practicing HMO avoidance. The Wall Street Journal (August 12), A1. Sharpe, A. (1998b). Health care: Discounted fees cure headaches, some doctors find. The Wall Street Journal (September 15), B1. Shute, N. (2002). That old time medicine. U.S. News and World Reports (April 22), 54–61. Solomon, R. C. (1992). Ethics and excellence. Oxford: Oxford University Press. Williamson, O. (1975). Markets and hierarchies: Analysis and antitrust implications. New York, NY: Free Press. Williamson, O. (1985). The economic institutions of capitalism: Firms, markets, and relational contracting. New York, NY: Free Press.

MEASURING AND ACCOUNTING FOR MARKET PRICE RISK TRADEOFFS AS REAL OPTIONS IN STOCK FOR STOCK EXCHANGES Hemantha S. B. Herath and John S. Jahera Jr. ABSTRACT The flexibility of managers to respond to risk and uncertainty inherent in business decisions is clearly of value. This value has historically been recognized in an ad hoc manner in the absence of a methodology for more rigorous assessment of value. The application of real option methodology represents a more objective mechanism that allows managers to hedge against adverse effects and exploit upside potential. Of particular interest to managers in the merger and acquisition (M&A) process is the value of such flexibility related to the particular terms of a transaction. Typically, stock for stock transactions take more time to complete as compared to cash given the time lapse between announcement and completion. Over this period, if stock prices are volatile, stock for stock exchanges may result in adverse selection through the dilution of shareholder wealth of an acquiring firm or a target firm. The paper develops a real option collar model that may be employed by managers to measure the market price risk involved to their shareholders in offering or accepting stock. We further discuss accounting issues related to this contingency pricing effect. Using an acquisition example from U.S.


191

192

HEMANTHA S. B. HERATH AND JOHN S. JAHERA JR.

banking industry we illustrate how the collar arrangement may be used to hedge market price risk through flexibility to renegotiate the deal by exercising managerial options.

INTRODUCTION An important area of research in management accounting is the implementation of strategic management decisions and managerial behavior that focus on ways to improve management and corporate performance. With the increased reliance of firms on financial instruments to manage business risk, its measurement and disclosure has become increasingly important in accounting. This research looks at one element of business risk, specifically, the measurement of market price risk to shareholders in a merger and an acquisition transaction using an emerging capital budgeting tool – real option methodology and accounting issues of market price risk to the acquiring firm. Merger and acquisition activity has increased sharply since the 1990s. Strikingly, noticeable in the merger and acquisition activity in this decade is that companies are increasingly paying for acquisitions with stock rather than cash. Stock for stock offers typically takes more time from announcement to completion than cash offers. This is particularly noticeable in regulated mergers as in the banking industry. Over this extended time from announcement to completion, the target and acquiring firm stock prices can change dramatically due to various factors even though they are in the same industry. Especially, if the acquiring firms stock is highly volatile, it can significantly affect the value of the deal at consummation if there are no price protection conditions built into the deal. Published literature dealing with price protection in mergers and acquisitions is sparse. However, the contingent pricing effect on the value of a deal in a stock for stock exchange due to stock price volatility has important risk management implications for managers and both sets of shareholders. A practical way to provide price protection to both acquiring and target firm shareholders is to set conditions for active risk management by managers. For example, one possibility is to provide managers the flexibility to renegotiate the deal and hedge the market price risk by specifying a range within which the deal is allowed to fluctuate as in a collar type arrangement. This paper investigates how to better structure a stock for stock offer as a collar using real option methodology when stock prices are volatile and when there is considerable time lapse between announcement and final consummation. We propose that managers use real option analysis to measure the price risk involved to their shareholders. The main advantage of using real option analysis is that it can

Measuring and Accounting for Market Price Risk Tradeoffs

193

capture and measure the value of intangible such as maintaining flexibility when there is high uncertainty. We argue that explicit valuation of managerial flexibility including in the terms of interest may enhance deal value for both parties and enforce favorable managerial behaviors. We also discuss accounting issues related to business combinations and in particular accounting for these intangibles. We propose that perhaps these real options should be accounted as contingencies? Using a recent acquisition case example from U.S. banking industry, the paper illustrates how the proposed collar is used to hedge the market price risk and how this acquisition structure avoids earnings per share (EPS) dilution to both sets of shareholders. The paper is organized as follows. First, we discuss stock price variability and its valuation effects on stock for stock transactions. Second, we introduce real option theory and managerial flexibility in M&A decisions. Third, we present well-known formulas for optimal exchange ratios of a target and an acquiring firm. Fourth, we discuss the proposed real options collar model for valuing managerial flexibility in stock for stock transactions. Fifth, we discuss accounting issues related to business combinations and in particular accounting for contingencies. Sixth, we apply the real option collar model to value the recent acquisition decision of BankFirst Corporation by BB&T. Finally, we discuss our findings and future research.

STOCK PRICE VARIABILITY AND VALUATION EFFECTS Stock for stock transactions generally take more time to complete than cash transactions. If stock price variability of both buying and selling firm is high, the value of respective shares may change between time an exchange ratio is determined and acquisition date. If a fixed exchange ratio is used to determine a target’s compensation, an acquiring firm overpays when its stock price is higher on merger date than on agreement date or in an event that a target’s stock price is lower on merger date than agreement date. Alternatively, a target loses if an acquiring firm stock price is lower on merger date than on agreement date or when a target’s stock price has risen on merger date. Hence, the time to consummation and the underlying price volatility can result in over or underpayment. One solution to this problem suggested by Gaughan (1999) is to arrange a provision to adjust the exchange ratio if stock prices go above or below a certain threshold. Typically, an adjustment provision has to be negotiated between the two firms and agreed upon at time the deal is negotiated. Consequently, an adjustment provision in an agreement is more useful when one or both stock prices of a target and/or an acquiring firm are highly volatile. The above adjustment to the

194


exchange ratio is a na¨ıve approach although it seems to work similar to a collar type arrangement.

REAL OPTIONS – MANAGERIAL FLEXIBILITY Valuation of a M&A requires a bidding firm to forecast free cash flows of a target. A target’s cash inflows should exceed investment cost or price to be paid by the acquiring firm. Also, it should provide a return that is at least equal to or greater than current return to the acquiring firm. The operating cash flows of the combined company due to synergies are the fundamental value driver that determines the market value of the combined company (Herath & Jahera, 2001; Houston & Ryngaert, 1996; Rappaport & Sirower, 1999). Real option methodology is an emerging area of research in management accounting (Booth, 1999; Brabazon, 1999; Busby & Pitts, 1997; Stark, 2000; Zhang, 2000). In recent years, practitioners and academics have argued that the traditional discounted cash flow models do not adequately capture the value of managerial flexibility to delay, grow, scale down or abandon projects. For example, Myers (1977, 1984) and Kester (1984) argue that a substantial portion of the market value of a company arises not from assets in place but from the present value of future growth opportunities. Real option methodology combine strategy with valuation. The insight is that an investment opportunity can be conceptually compared to a financial option. Real options methodology is conceptually better than conventional capital budgeting techniques for M&A valuations since it takes into account managerial flexibility to respond to uncertainties (Copeland & Antikarov, 2001; Herath & Jahera, 2001; Lambrecht, 2001). Managerial flexibility in M&A decisions can be explicitly considered through innovative deal structures. For example, Herath and Jahera (2001) considered the value of managerial flexibility as an exchange real option to offer stock or cash for a target when an acquiring firm’s stock is highly volatile. In the real options framework a M&A investment decision is valued as the sum of two value components. The first value component is standard net present value (NPV) of a M&A investment decision which is the value without managerial flexibility. The second value component is the value of managerial flexibility to respond to uncertainties. In financial options terminology, NPV and managerial flexibility to delay/modify an investment are equivalent to an option’s intrinsic and time values. Consequently, the value of an M&A investment with managerial flexibility or strategic NPV can be considered as sum of two value components: Strategic NPV = Base NPV without flexibility + Value of managerial option(s)


195

OPTIMAL EXCHANGE RATIOS IN STOCK FOR STOCK TRANSACTIONS In this section we present well-known formulas for exchange ratios. In stock for stock transactions, an acquiring firm purchases a target firm by exchanging its stock for that of target. An acquiring firm may offer a premium to target to make an offer attractive. The exchange ratio is determined prior to acquisition date. In a conventional stock for stock transaction the exchange ratio remains fixed independent of stock price variability of both buyer and seller. An exchange ratio gives the number of shares of acquiring firm that will be exchanged for a target. The exchange ratio (F) is given by F=

No S B (1 + Z) = NB SA

where N o = the number of acquiring firm shares offered in exchange for target shares; N B = the number of shares outstanding of target firm; N A = the number of shares outstanding of acquiring firm; S A = market price of acquiring firm stock; S B = market price of target firm stock; Z = merger premium. While an exchange ratio is typically fixed when a deal is agreed upon, in fixing the ratio it is of interest to both an acquiring firm and target to determine the optimal exchange ratios that will not dilute shareholder wealth of the two firms. To determine optimal exchange ratios one assumes constant stock prices. An acquiring firm is interested in the largest exchange ratio (Fmax ) that will not reduce post acquisition share holder value or stock price. This ratio is given by: EA NA + EB NB P 1 F max ≤ − NA NB SA E c Target firm shareholders benefit from a higher exchange ratio resulting in more shares for each of its stock. The minimum exchange ratio (Fmin ) that preserves the post-acquisition target’s shareholder value is given by: SB NA F min ≥ (E A N A + E B N B )(P/E)c − S B N B where E A = the pre-merger earnings per share (EPS) of the acquiring firm; E B = the pre-merger earnings per share (EPS) of the target firm; (P/E)c = post acquisition P/E ratio. The expression for Fmax is an increasing linear function of post acquisition (P/E) ratio while the expression for Fmin is a decreasing convex function. The relationships between Fmax and (P/E)c and between Fmin and (P/E)c is shown in

196


Fig. 1. Dependency of Acquiring and Target Firm Wealth on Exchange Ratio and Post Merger P/E Ratio.

Fig. 1. Of particular interest is the region where shareholders of both firms will benefit. An exchange ratio f ∗ (F min ≤ f ∗ ≤ F max ) for some (P/E)c that lies in the optimal region will theoretically increase post acquisition share holder wealth of both parties. The minimum post completion price to earnings ratio (P/E)∗c is where the two expressions equate (F min = F max ). As shown in Fig. 1, an exchange ratio ( f ) that is greater than minimum exchange ratio (Fmin ) and less than maximum exchange ratio (Fmin ) for a any post completion (P/E)c ratio that is greater than minimum post completion price to earnings ratio (P/E)∗c should be negotiated.

MANAGERIAL FLEXIBILITY AS A REAL OPTION COLLAR ARRANGEMENT A better way to structure a stock for stock acquisition is to minimize consideration paid by a bidding firm and maximize deal value to the target over the period from announcement to completion of a deal. The theoretical value of an acquisition is defined as the deal value based on a fixed exchange ratio. Notice that the theoretical price will depend on acquiring firm stock price at consummation. Consequently, if stock price of acquiring firm increases or decreases, the theoretical deal value will either increase or decrease accordingly. In order to minimize the value of a deal, an acquiring firm could buy a call option or a cap, which guarantees a minimum


197

deal value. A target on the other hand could consider a put option or a floor, which ensures that holder, would receive the maximum deal value. Consequently, managerial flexibility to both parties can be structured as a collar arrangement, going long on a cap and shorting a floor. An acquiring firm should buy a real call option on theoretical value with a strike price equal to a deal value that cap the exchange ratio to a maximum exchange ratio that will not dilute post acquisition stock value for acquiring firm shareholders. The underlying asset is the theoretical deal value. The cap would guarantee that the deal value at any given time would be the minimum of the theoretical value or the deal value based on the maximum exchange ratio. On the other hand, to maximize its payment a target should hold a real put option on the theoretical value with a strike price based on the minimum exchange ratio. In this way the floor guarantees that a target would receive maximum of the theoretical value or the deal value using the minimum exchange ratio. In order to price the cap and the floor, we use the binomial lattice framework for pricing options on a stock. The real option prices are thus consistent with risk-free arbitrage pricing. Let SA and SB denote stock prices of an acquiring firm and a target firm at announcement of the deal. The stock price of firm (i), S(i) follows a random walk. The time between the announcement (t0 ) and the actual closing of the deal (t1 ) is denoted by (t) where (t = t 1 − t 0 ). Assume that there are four decision points (T = 0, 1, 2, 3) pertaining to when a deal may be closed. We divide the time period (t) into equal periods of length (T = t/3), which may be measured in weeks or months. In the binomial option pricing model, the formulas to compute risk neutral probability (p(i) ), and upward movement for stock price u (i) and downward movement d(i) for stock price (i) are as follows: u i = e␴i

√

and d i = e−␴i

T

√

T

and pi =

er f T − d i , ui − di

where, ␴(i) is the stock price volatility of firm (i). The short-term interest rate (rf ), stock price volatility of the acquiring firm (␴A ) and target firm (␴B ) are assumed to remain constant in the model. Upward and downward movement in stock price are represented by the state variable (k) and denoted by (+) and (−) respectively. Using above parameters we next develop a three period binomial lattices for movement in stock prices of an acquiring firm and a target firm as shown in Fig. 2.

198


Fig. 2. Three Step Binomial Lattices for Stock Prices.

For any exchange ratio ( f ) the deal value (P kT ) at any period (T ) in state (k), can be calculated as follows: P kT = f N B S kA As per our notation, the state variable (k) would be either + or −. The deal value based on the maximum and minimum exchange ratios are given by: U kT = N B S kA kF max and L kT = N B S kA kF min respectively. The theoretical deal value based on a fixed exchange ratio is given by: V kT = N B S kA F. Notice that value of U kT and L kT are also a function of acquiring firm stock price. Alternatively, a target and acquiring firm may agree upon minimum and maximum deal values based on the acquiring firm’s stock price at announcement. In this case, the maximum and minimum deal values will be equal to the following constant expressions U = N B S A F max and L = N B S A F min . Notice that stock prices of both target and acquisition firms can vary from the time a deal is announced to when it is completed. In addition, using the historic price volatility the ratio of the projected stock price of an acquiring firm to a target firm under each state (k) and time (T) can be computed. We refer to this variable exchange ratio as the critical exchange ratio defined by ( f k ) where fk =

S kB S kA


199

Notice that critical exchange ratio ( f k ) which is based on projected stock prices can be used to determine the target’s market value at any state k. The target’s market value (W kT ) based on variable exchange ratio at any period (T) in state (k), can be obtained by substituting f k in the expression for P kT . The target’s market value is W Tk = f k N B S kA Real Call Option – Cap A buying firm would want flexibility to minimize the theoretical deal value at completion. Thus a cap would guarantee that a deal is valued as the minimum of the theoretical deal value or the agreed upon value based on maximum exchange ratio given by min [U, V kT ] = min[NB SA Fmax , NB FS kA ]. An acquiring firm which holds the cap should pay a target a terminal payoff of max[NB FS kA − NB SA Fmax , 0] = max[NB FS kA − U, 0] to receive the benefit of having the right to pay min[NB SA Fmax , NB FS kA ]. Notice that U is analogous to the exercise price of the call option when the underlying asset in option terminology is the theoretical deal value V kT . Real Put Option – Floor A target should hold a floor which would guarantee the maximum of the theoretical value or the agreed upon deal value based on the minimum exchange ratio given by max[L, V kT ] = max[NB SA Fmin , NB FS kA ]. The floor would provide a target the managerial flexibility to maximize the value of the deal to its shareholders. A target should pay an acquiring firm a floor premium max[Fmin NB SA − NB FS kA , 0] = max[L − NB FS kA , 0] to have the right to receive benefit of max[NB SA Fmin , NB FS kA ]. Here L is analogous to the exercise price of the put option. Given that the theoretical post acquisition market price of a target and acquiring firm will depend on post acquisition earnings and the (P/E) ratio, the following optimal terminal cap and floor values can be identified. Proposition 1. If the exchange ratio is greater than maximum exchange ratio then post acquisition shareholder wealth will be lower; that is S A > S c . The call option (cap) to an acquiring firm has value (X kT > 0) and the put option (floor) to the target has zero value (Y kT = 0). Proof: To prove this result let us assume that f > F max EA NA + EB NB P 1 F max ≤ − NA NB SA E c

200

then


P − NA E c P − NA SA f N B S A > (E A N A + E B N B ) E c P ( f N B + N A )S A > (E A N A + E B N B ) E c EA NA + EB NB P SA > fN B + N A E c P SA > Ec E c

1 f> NB

EA NA + EB NB SA

SA > Sc Where the post acquisition earnings per share of the acquiring firm is given by: Ec >

EA NA + EB NB fN B + N A

When f > F max , an acquiring firm would prefer to cap a deal to maximum deal value of X kT = min[NB SA Fmax , NB fS kA ]. Since we assumed that both parties agreed on the fixed exchange at announcement by substituting f = F we obtain X kT = min[U, V kT ]. The terminal call option payoff is then equal to max[NB FS kA − NB SA Fmax , 0] = F NB S kA − NB SA F max > 0. For a target, the terminal payoff of its put option is Y kT = max[Fmax NB SA − NB FS kA , 0] = 0. Therefore if f > F max it is beneficial for the acquiring company to cap the deal since it will otherwise dilute post acquisition EPS of its shareholders. Proposition 2. If the exchange ratio is less than minimum exchange ratio then the deal does not preserve the post acquisition shareholder wealth of target shareholders. The call option (cap) to the acquiring form has no value (X kT = 0) and the put option (floor) has value (Y kT > 0). Proof: To prove this result let us assume that f < F min SB NA F min ≥ (E A N A + E B N B )(P/E)c − S B N B f
0. For an acquiring firm terminal payoff of its call option is X kT = max[F NB S kA − Fmin NB SA , 0] = 0. Therefore if f < F min it is beneficial for the target company to hold a floor to maximize its deal value. Otherwise it will dilute post acquisition EPS of target shareholders. Proposition 3. If the critical exchange ratio lies between minimum exchange ratio and maximum exchange ratio (F min < f < F max ) then both acquiring firm and target firm shareholder’s gain. There is no EPS dilution to either party and the collar has no value. Proof: To prove this result let us assume that f < F max EA NA + EB NB P 1 F max ≤ − NA NB SA E c then

P f N B S A < (E A N A + E B N B ) − NA SA E c P ( f N B + N A )S A < (E A N A + E B N B ) E c EA NA + EB NB P SA < f NB + NA E c

202


P SA < Ec E c SA < Sc The post acquisition stock price of acquiring firm is greater than its stock price before acquisition. Proof: To prove this result let us assume that f > F min SB NA F min ≥ (E A N A + E B N B )(P/E)c − S B N B SB NA (E A N A + E B N B )(P/E)c − S B N B P (E A N A + E B N B ) f > SB NA + SB NB f E c EA NA + EB NB P f > SB E c NA + NB f k P Ec f > SB E c f>

Substituting for the variable exchange ratio f = F = N 0 /N B we obtain the following: Pc No > NB SB The post acquisition shareholder wealth of a target firm is greater than its shareholder’s wealth before acquisition. Therefore, if F min < f < F max then there is no dilution in the post acquisition shareholder wealth of an acquiring firm or a target firm. By substituting f = F we obtain the following. When F < F max value of the cap given by X kT = max[F NB S kA − NB SA Fmax , 0] = 0. Similarly, when F > F min value of the floor given by Y kT = max[NB SA Fmin − NB FS kA , 0] = 0. The value of the collar can be found by holding a cap and shorting a floor; collar = cap – floor which is equal to zero. In Table 1 we summarize terminal payoffs of the cap and the floor.

Pricing the Cap and Floor Using terminal payoff values in Table 1, we can now price the cap and the floor based on risk-free arbitrage pricing. The payoff values (X K T ) of a call option in


203

Table 1. Terminal Payoffs to the Buyer and Seller.

F > Fmax and S kA > S A

Acquiring firm Real Call Option – CAP

Target Real Put Option – FLOOR max [U, V kT ] = F NB S kA , With fixed exchange ratio (F)

Objective Best value

min[U, V kT ] = Fmax NB SA , With maximum exchange ratio (Fmax ) max[F NB S kA − Fmax NB SA , 0] = NB S kA F − U min[L, V kT ] = F NB S kA , With fixed exchange ratio (F)

Savings

max[F NB S kA − L, 0] = 0

Objective Best value Payoff

F < Fmin and S kA > S A

max[Fmax NB SA − F NB S kA , 0] =0 max [L, V kT ] = Fmin NB SA , With minimum exchange ratio (Fmin ) max[Fmin NB SA − F NB S kA , 0] = L − NB S kA F

each period (T) and state (k) can be calculated as: X kT = max[N B S kA F − U, 0] The payoff values (Y K T ) of the put option in each period (T) and state (k) is given by: Y kT = max[L − N B S kA F, 0] In Fig. 3, we present payoff values in each state (k) and time (T) for a real call and a real put (as shown in box). Next, we employ a risk neutral approach to price the real call and real put options. Since we have assumed that the time between announcement and final consummation was divided into four decision points, we need to find managerial flexibility available to both buyer and seller in each period T = 1, 2 and 3. The value of managerial flexibility to the buyer can be priced as three European calls with respective maturity at time T = 1, 2 and 3. Let XT denote value of a T period call. For example, a two period call to offer minimum deal value in period T = 2 can be value as follows. The real call X2 is valued by finding terminal payoff values at T = 2, −+ −− X ++ 2 , X 2 , X 2 and folding back two periods. For the payoffs at T = 2 we obtain ++ X2++ = max[NB SA F − U, 0]

+− X2+− = max[NB SA F − U, 0] −− X2−− = max[NB SA F − U, 0]

204


Fig. 3. Payoff Values for the Real Call and Real Put (in Box). − Using the standard binomial approach, we next find values of X + 2 and X 2 at end of period 1 by risk-neutral discounting. Specifically,

X2+ = e−rf T [pA X2++ + (1 − pA )X2−+ ]

X2− = e−rf T [pA X2++ + (1 − pA )X2−− ] Thereafter, by applying risk-neutral discounting one more time, we compute the value of real call X2 as: − X 2 = e−r f T [p A X + 2 + (1 − p A )X 2 ]

Similarly, using the risk neutral procedure we can calculate flexibility to minimize deal values at T = 1 and 3, given by X1 and X3 respectively. The value of managerial flexibility to a target can be priced as three European puts with respective maturity at time T = 1, 2 and 3. Let YT denote the value of a T period real put option. For example, the one period put to receive maximum deal value in period T = 1, Y1 can be valued by finding the terminal payoff values


205

− at T = 1, Y + 1 , Y 1 and folding back one period. For the payoffs at T = 1 we obtain + Y1+ = max[L − NB SA F, 0] − Y1− = max[L − NB SA F, 0]

Using risk neutral discounting, we next find the value of a one period put option as − Y 1 = e−r f T [p A Y + 1 + (1 − p A )Y 1 ]

Similarly, by employing a risk neutral discounting procedure we can calculate the flexibility to maximize deal values at T = 2 and 3, denoted by Y2 and Y3 respectively.

Pricing the Collar In order to consider the value of managerial flexibility available to both the target and acquiring firm to benefit by renegotiating the deal at a future decision point, we can go long on the real call and short the real put. This would cap the theoretical deal value to a maximum exchange ratio to benefit the buyer while guaranteeing a deal value based on minimum exchange ratio to benefit the seller. Since the cap and floor have different strike prices U kT = N B S kA F max and L kT = N B S kA F min by holding a cap and selling a floor we can effectively create a collar. Therefore, the value of a collar at any period T can be computed as the difference between value of managerial flexibility to a buyer (call) and value of flexibility to a seller (put) given by ␾T = X T − Y T . Thus the deal value including managerial flexibility to both buyer and seller (V) can be calculated as: V = V0 +

3

␾T

T =1

Where (V0 ) is the acquisition value with out managerial flexibility agreed between the two parties at time T = 0 given by V 0 = FN B S A , where F is the fixed exchange ratio and ␾T is the net value of flexibility (value of flexibility to buyer) – (value of flexibility to seller). Notice that value of flexibility to a buyer will increase the purchase price of an acquisition while the value of flexibility to a seller would reduce it. Value of an acquisition if only buyer’s managerial flexibility to renegotiate the deal by exercising the real call option is considered equal to VA = V0 +

3 T =1

XT

206


Similarly, the value of an acquisition if only seller’s managerial flexibility to renegotiate the deal by exercising the real put option is considered equal to VB = V0 −

3

YT

T =1

A real option collar can alternatively be valued instead of separately pricing each cap and floor. In order to do so, one can find the value of collar (␾T ) at each time period (T ). Consider terminal collar payoffs at each state (k) and time (T ) which is equivalent to max[N B S kA F − U, 0] − max[L − N B S kA F, 0] and then price the collar at each period (T ) using risk neutral discounting. This method will give yield an equivalent value for the real option collar.

ACCOUNTING ISSUES: BUSINESS COMBINATIONS AND CONTINGENT CONSIDERATIONS Types of Business Combinations Mergers and acquisition transactions are treated for accounting purposes as business combinations. In a business combination transaction one economic unit unites with another or obtains control over another economic unit. Accordingly, there are two forms of business combinations; purchase of net assets, where an enterprise acquires net assets that constitute a business and; purchase of shares, where an enterprise acquires sufficient equity interest of one or more enterprises and obtains control of that enterprise or enterprises. The enterprise(s) involved in a business combination can be either incorporated or unincorporated. However, the purchase of some (less than 100%) of an entity’s assets is not a business combination. The form of consideration in a business combination could be cash or a future promise to pay cash, other assets, common or preferred stock, a business or a subsidiary or any combination of the above. In a purchase of assets, an enterprise may buy only the net assets and leave the seller with cash or other consideration, and liabilities. Alternatively, a buyer may purchase all the assets and assume all the liabilities. The more common form of business combination however is the purchase of shares. In a purchase of share transaction, the acquiring firm’s management makes a tender offer to target shareholders to exchange their shares for cash or for acquiring firm shares. The target continues to operate as a subsidiary. In both purchase of net assets and purchase of shares the assets and liabilities of the target are combined with the assets and liabilities of the acquiring firm. If the acquiring firm obtains control by purchasing net


207

Fig. 4. Purchase of Net Assets and Purchase of Share Transactions.

assets the combining takes place in acquirer’s books. If acquirer achieves control by purchasing shares combining takes place when the consolidated financial statements are prepared. The types of business combinations are summarized in Fig. 4.

Accounting for Business Combinations Prior to July 1, 2001, in the U.S., there were two alternative approaches to account for business combinations. The pooling method was used when an acquirer could not be identified. Stock deals were accounted for by the pooling method. The pooling method was only possible in a share exchange since if cash was offered then the company offering cash became the acquirer and the purchase method had to be used. Under the pooling method, the price paid was ignored, fair market values were not used, and the book values of two companies were added together. The pooling method avoided creating goodwill and ignored value created in a business combination transaction. Also reported earnings were higher. Since July 1, 2001, revised accounting standards in the U.S. allow only the purchase method to account for mergers and acquisitions. Regardless of the purchase consideration, if one company can be identified as the acquirer, the purchase method has to be used. Under this method, the acquiring company records net assets of the target at the purchase price paid. The purchase price may include cash payments; fair market value of shares issued, and the present value of promise to pay cash in the future. Goodwill is the difference between purchase price paid and fair market value of net assets of the target. Goodwill is reviewed for impairment and any impairment loss is charged against earnings. Future reported earnings under the purchase method are lower.

208


The purchase method is recommended due to the following limitations of pooling method (Hilton, 2003). In the pooling method, the fair market values are ignored and thus do not reflect the valued exchanged in a M&A transaction. Pooling information is less complete and often it is difficult to compare two methods of accounting for business combinations, under what is essentially the same economic event. Also, while future reported earnings would be different under the two methods, future cash flows are essentially the same. Finally, regardless of whether cash or stock is offered as purchase consideration, the method of recording an acquisition should be the same. After identifying the acquiring firm, the acquisition cost has to be allocated to the assets and liabilities that are acquired. The acquisition cost may constitute of cash paid, the present value of any promise to pay cash in future, and (or) the fair market value of shares issued. If the fair market value of shares cannot be ascertained, the fair market value of net assets is used. In addition, the acquisition cost may also include contingent considerations. We next discuss accounting for contingencies in business combinations. In the previous sections we illustrated how one could measure market price risk tradeoffs to acquiring and target firm as real options in a stock for stock exchange. We discussed why managerial flexibility to hedge market price risk by exercising the appropriate managerial options should be considered and reported. Lastly, we argue whether these real options to hedge market price risk should be treated for accounting purposes as contingencies in business combinations?

Contingencies In some situations the consideration promised in an acquisition may be adjustable depending upon future events. For example, business combination terms may require an additional cash payment or a share issue contingent on some specific future event. If the outcome is considerably certain to occur and can be estimated on date of acquisition, the estimated amount is recorded as part of purchase cost. However, if on the date of the combination, the contingent amount cannot be estimated and the outcome is uncertain, it is recorded at some future date when the outcome becomes payable. Whether the contingency amount is considered as part of purchase cost or as a capital transaction will depend on specific circumstances as illustrated below. If the contingency is based on earnings, it is considered an additional cost of purchase. The total acquisition cost will be the purchase cost that is incurred when the acquisition is complete plus the contingency when it is recorded. Since the contingency is recorded later a question may arise as to how to treat this


209

additional amount. The logical treatment would be to consider it as goodwill. If the contingency is based on future share prices, any consideration arising at the future date will be recorded at fair market value but as a reduction in amount of the original share issue. The consideration is then not considered as additional cost of the purchase.

CASE STUDY: BB&T CORPORATION’S ACQUISITION OF BANKFIRST CORPORATION (BKFR) In this section we present a recent acquisition that occurred in U.S. banking industry to illustrate the real options collar model. Herath and Jahera (2001) used this case to demonstrate how a more flexible deal structure to exchange cash for fair market value of a deal may have reduced dilution of value to acquiring bank shareholders. In August 23rd 2000, BB&T Corporation announced the acquisition of BankFirst Corporation (BKFR) for $149.7 million in stock. Based on BB&T’s closing price of $26.81, BKFR with $848.8 million in assets was valued at $12.21 per BankFirst share. The closing price of BKFR on announcement day was $11 21 . The exchange ratio was fixed at 0.4554 BB&T stock for each share of BankFirst Corporation. The following financial data pertaining to BB&T and BKFR acquisition given in Table 2 is used to determine the optimal exchange ratios for the real option collar model. Using the above data, the number of shares of BB&T that is offered for BKFR is equal to 5,579,216. The post acquisition EPS based on the combined earnings of the two firms is equal to $ 1.94 and post acquisition (P/E) ratio is found to be 13.75. Since the financial markets are efficient, bootstrapping of EPS is not possible and the post-acquisition (P/E) ratio will be the weighted average of the pre-acquisition (P/E) ratios. Incidentally, post acquisition (P/E) ratio is the minimum post completion price to earnings ratio (P/E)2c where the two expressions for optimal exchange ratios equate (F min = F max ). The theoretical Table 2. Financial Data for BB&T and BKFR. BB&T (A) Present earnings ($ million) Shares outstanding Earnings per share (EPS) Stock price (average) P/E ratio Target’s premium

767 395,360,825 $ 1.94 $ 26.71 13.9

BKFR (B) 8.754 12,260,500 $ 0.72 $9 12.50 35%

210


post-merger stock price can be computed using post-merger (P/E) ratio and the post-merger EPS relationship as $26.60. The deal announced on August 23, 2000, closed on December 28, 2000 for $216.2 million in stock. Accordingly, on a per share basis the deal was valued at $17.42 per BKFR share based on BB&T closing price of $38.25. Notice, that over the four-month period, from announcement to closing, the deal value increased by $66.6 million, a 44% increase. Theoretically, this increase is a hidden loss to shareholders of the acquiring bank since they are effectively paying more for relatively the same net assets if the deal had closed at the original value of $149.7 million. Notice that fundamental economics of the acquisition have not changed since expected post acquisition cash flows of target will remain unchanged. The prices paid in a stock swap are real prices and as such there is greater dilution of equity interest of acquiring firm shareholders. From a purely accounting perspective, however, it would not make a difference since the transaction would have been recorded using the pooling method at historic book values of net assets. What would have been the deal value, if both BB&T and BKFR’s managerial flexibility to renegotiate the deal by exercising real call and put options had been considered? How much would that flexibility be worth? The data for our model were obtained from company annual reports. The volatility of BB&T and BankFirst stocks were estimated using stock price data from August 1998 to December 2000. BankFirst Corporation went public in August 1998. Although for illustration purposes we used stock price data from August 1998 to December, 2000, notice that stock price data beyond the announcement date should not be used. Data for the model are summarized as follows: BB&T Corporation annual stock price volatility ␴ estimated using historic data A

is 35.7%. BankFirst Corporation annual stock volatility ␴B using historic data is found to be 33%. The binomial lattices for BBT and BKFR along with the actual stock prices (high) are presented in Appendix Exhibit 2. Volatility is measured as standard deviation of log returns of stock price over the period. A constant risk-free rate, r f = 6%, is assumed. Fixed exchange ratio is F = 0.4554 BBT stock for each BKFR share. Fair market value of BankFirst Corporation net assets is NB FSA = $149,500,000. Number of BKFR common stock outstanding is N B = 12,260,500 shares. The time from announcement to closing is T = 4 months since the deal was announced end of August and closed end of December. Thus T is divided in to 4 periods of equal length (T = 1 month or 0.0833 years). Market price of BB&T stock at announcement S A = 26.81.

Stock price data of BB&T and BKFR used to estimate the volatility, resulting binomial parameters and lattices for movements of stock prices for the acquiring


211

Table 3. Variable Exchange Ratio of BKRF/BBT Stock Price (f k ). T=0

T=1

T=2

T=3

T=4

0.4498

0.4463 0.4533

0.4428 0.4498 0.4568

0.4394 0.4463 0.4533 0.4604

0.4360 0.4428 0.4498 0.4568 0.4640

bank and target are shown in Appendix Exhibits 1 and 2. Once we develop binomial trees pertaining to each stock, we next compute the variable stock exchange ratio ( f k ) in each state (k). For example the variable stock exchange ratio at T = 0, is computed as f k = 12/26.68 = 0.4498. The variable exchange ratios in spreadsheet format are shown in Table 3. In spreadsheet format, an upward movement is shown directly to the right and a downward movement is shown directly to the right but one step down.

Fig. 5. Optimal Exchange Ratios.

212


In order to price each cap and floor, one needs to find maximum and minimum exchange ratios. The relationship between these two exchange ratios and the post merger (P/E) ratio is shown in Fig. 5. Notice that the fixed exchange ratio of 0.4554 will only benefit BB&T. Since the minimum post merger (P/E) ratio is 13.75, for both firm shareholders to benefit we selected a post-merger (P/E) ratio of 13.8, which falls in the optimal region. The corresponding maximum and minimum exchange ratios are 0.4591 and 0.3358. The optimal exchange ratios and resulting minimum and maximum agreed upon deal value are as follows:

Maximum exchange ratio F max = 0.4591. Minimum exchange ratio F min = 0.3358. Agreed upon minimum deal value L = 110.4 million. Agreed upon maximum deal value U = 150.9 million.

In order to price the four real call options (caps) and four real put options (floors) pertaining to each decision point T = 1, 2, 3 and 4 we use formulas for X kT and Y kT at each state (k) and time (T). The terminal payoff for the cap at T = 2, k = ++ = max{(12,260,500)(32.79)(0.4554) − 150,900,000, 0} = 32,155,266. is X ++ 2 Similarly we can compute terminal payoff of the floor at T = 4, k = − − −− as L −−−− = max {110,400,000 − (12,260,500)(17.67)(0.4554), 0} = 11,737,537. 1 The terminal payoff values for pricing the cap and floor are presented in Table 4. Using formulas developed in preceding section, we next calculate values of the four caps. This is done by finding expected terminal payoff values using risk-free

Table 4. Terminal Payoff Values. T=0

T=1

Cap payoff values (X kT ) 14,228,892 0

Floor payoff values (Y kT ) 0

T=2

T=3

T=4

31,155,266 0 0

52,027,630 14,228,892 0 0

74,057,231 32,155,266 0 0 0

0 0

0 0 0 1,029,571

0 0 0 0 11,737,537


213

probabilities and discounting by the short-term interest rate. More specifically, a two period cap is valued by first finding the risk neutral discounted payoff at T = 1 −(0.06)0.0833 X −− [0.499(32, 155, 266) + (0.501)0] = $15,950,563 1 =e −(0.06)0.0833 X −+ [0.499(0) + (0.501)(0)] = $0 1 =e

And then applying risk neutral discounting one more time to find the cap payoff at T = 0 as X2 X 2 = e−(0.06)0.0833 [0.499(15, 950, 563) + (0.501)(0)] = $7, 912, 249 Similarly, by using formulas developed in the preceding section, we can calculate values of the four floors (YT ). The value of managerial flexibility available to BB&T (XT ) and value of managerial flexibility available to BKFR (YT ) at each decision point T = 1, 2, 3 and 4 are presented in Table 5. By going long on a cap and shorting a floor we find the value of real option collar (␾T ). The value of real option collar at each decision point includes the value of managerial flexibility available to both (acquiring firm) BB&T and (target) BKFR. The four real option collar values are presented in Table 5. The conventional stock for stock offer agreed by BB&T and BKFR did not take into consideration real options available to both buyer and seller in structuring the acquisition. Thus value of the acquisition without flexibility to renegotiate was V0 = $149.5 million.

Table 5. Cap, Floor and Collar Values. T=1

T=2

T=3

T=4

Cap (XT ) Value of buyer’s flexibility

$7,058,217

$7,912,249

$11,591,413

$12,317,427

$38,879,306

Floor (YT ) Value of seller’s flexibility

0

0

$127,898

$727,536

$855,435

$7,058,217

$7,912,249

$11,463,515

$11,589,891

$38,023,872

Description

Collar (␾T ) Combined value of buyer and seller flexibility

Total

214


DISCUSSION The original deal, which was valued at $149.7 million, was closed on December 28th 2000 for $216.2 million in stock, a dilution of $66.6 million to BB&T shareholders. If the acquisition was structured to include only BB&T Corporation’s managerial flexibility to cap the deal value to hedge dilution to its shareholders, the deal would be valued at $188.6 million. Alternately, if the deal were structured with only BKFR’s flexibility to benefit its shareholders, it would be worth $148.9 million. Ideally, the acquisition could have been fairly valued by considering managerial flexibility available to both BB&T and BKFR’s. This would result in a deal valued at $187.72 million. The fair deal value based on the real option collar model is thus $ 28.5 million lower than actual closing value. Although we considered managerial flexibility to hedge market price risk available to both parties as a collar, at consummation of the deal only a cap or a floor will be exercised since they are mutually exclusive managerial actions. Therefore, when negotiating, the collar arrangement should clearly provide a contractual provision that allows for the final purchase cost to be computed as the deal value without flexibility plus the real call or put value that is relevant. For example, let us assume that the deal was consummated four months later when BBT’s stock price is $38.25. This would indicate that the acquiring firm’s management would exercise a four period call to cap the deal. The value of the four period call is $12 million and the purchase cost to the acquiring firm would be $161 million (deal value of $149.7 million without flexibility plus cost of the call option of $12 million). At the time of consummation the acquiring firm’s management will exercise the call option and cap the deal to the agreed maximum value of $150.9 million. By exercising the call option the acquiring firm’s has hedged a dilution of $65 million ($216 million less $150.9 million) at a cost of $12 million. The corresponding managerial actions for the two firms at each decision state (k) are shown in Table 6. The corresponding payoff diagrams of the collar are shown in Fig. 6.

Table 6. Optimal Exercise Decisions. T=0

T=1

T=2

T=3

CAP –

CAP – –

CAP CAP – FLOOR

T=4 CAP CAP – – FLOOR


215

Fig. 6. Hedging the Risks.

We have demonstrated that a better way to structure a stock for stock transaction with stock price variability is to consider value of managerial flexibility in the acquisition structure. Conventional stock for stock transactions ignores the value of managerial flexibility available to both parties, which may be significant. Using real option methodology we demonstrated how this managerial flexibility could be valued using market based data. In this paper we propose that these real options should also be accounted as contingencies in business combination transactions so that managers are held accountable. In summary, we have argued the case for real options as responses to anticipated managerial actions, which provide a mechanism to commit managers to desirable behaviors that mitigate EPS dilution over the period the exchange ratio is fixed and the acquisition is complete. Real option methodology is a significant step forward in conceptualizing and valuing managerial flexibility in strategic investment decisions. The real option methodology is conceptually superior especially when the there is high degree of uncertainty associated with investment decisions, but it also has limitations. While much of the academic research in real options to date has been done by corporate finance academics, there is scope for extending and applying real option methodology in management accounting areas such as capital budgeting for IT investments, research and development and performance measurement and evaluation.

216


ACKNOWLEDGMENTS We thank Harjeet Bhabra, Graham Davis, Bruce McConomy and Peter Wilson for their helpful suggestions. The paper also benefited from comments received at the 2002 Northern Finance Association meeting presentation in Banff, Canada, 2003 AIMA Conference on Management Accounting Research presentation in Monterey, California and 2003 CAAA Annual Conference presentation in Ottawa, Canada.

REFERENCES Booth, R. (1999). Avoiding Pitfalls in investment appraisal. Management Accounting (November), 22–23. Brabazon, T. (1999). Real options: Valuing flexibility in capital investment decisions. Accountancy Ireland (December), 16–18. Busby, J. S., & Pitts, C. G. C. (1997). Real options in practice: An exploratory survey of how decision makers in industry think about flexibility. Management Accounting Research, 8, 169–186. Copeland, T., & Antikarov, V. (2001). Real options – A practitioner’s guide. New York, NY: Texere Publishing Limited. Gaughan, P. A. (1999). Mergers, acquisitions and corporate restructuring (2nd ed.). New York, NY: Wiley. Herath, H. S. B., & Jahera, J. S., Jr. (2001). Operational risk in bank acquisitions: A real options approach to valuing managerial flexibility. In: Advances in Operational Risk (pp. 53–65). London: Risk Books. Hilton, M. W. (2003). Modern advanced accounting in Canada (3rd ed.). Toronto: McGraw-Hill. Houston, J. F., & Ryngaert, M. D. (1996). The value added by bank acquisitions: Lessons from Wells Fargo’s acquisition of First Interstate. Journal of Applied Corporate Finance, 9(2), 74–82. Kester, W. C. (1984). Today’s option for tomorrow’s growth. Harvard Business Review (March-April), 153–160. Lambrecht, B. M. (2001). The timing and terms of merger, stock offers and cash offers. Working Paper, University of Cambridge. Myers, S. C. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5(2), 147–176. Myers, S. C. (1984). Financial theory and financial strategy. Interface, 14(January–February), 126–137. Rappaport, A., & Sirower, M. L. (1999). Stock or cash? The trade-offs for buyers and sellers in mergers and acquisitions. Harvard Business Review (November–December), 147–158. Stark, A. W. (2000, April/May). Real options (dis)investments decision-making and accounting measures of performance. Journal of Business, Finance and Accounting, 27(3&4), 313–331. Zhang, G. (2000). Accounting information, capital investment decisions, and equity valuation: Theory and empirical implications. Journal of Accounting Research, 38(2), 271–295.


217

APPENDIX Exhibit 1. Monthly Stock Prices and Returns for BBT and BKFR (August 1998 to December 2000). Month

BBT Stock Closing Price $

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Mean return Standard deviation Annual standard deviation Annual mean

26.23 28.05 33.70 34.82 38.00 35.75 35.87 34.27 37.99 34.72 34.90 33.73 32.05 30.97 35.02 31.05 26.35 27.29 22.80 27.23 26.02 28.65 23.33 24.58 26.68 29.69 31.67 33.16 37.07

BKFR Stock (Target)

Return ri (%) – 0.0669 0.1835 0.0327 0.0874 −0.0611 0.0033 −0.0456 0.1032 −0.0900 0.0051 −0.0342 −0.0509 −0.0342 0.1226 −0.1204 −0.1639 0.0349 −0.1797 0.1774 −0.0454 0.0962 −0.2052 0.0519 0.0824 0.1066 0.0646 0.0460 0.1115 0.0123 0.1030 35.7% 14.8%

Closing Price $ 12.00 11.38 11.00 9.75 8.94 10.31 10.81 10.00 10.00 9.22 9.31 9.00 8.88 9.50 9.00 9.38 8.63 8.38 8.00 7.31 8.63 7.75 8.25 9.13 12.00 13.75 14.00 15.00 17.13

Return ri (%) – −0.0535 −0.0335 −0.1206 −0.0870 0.1431 0.0473 −0.0782 0.0000 −0.0813 0.0101 −0.0342 −0.0140 0.0681 −0.0541 0.0408 −0.0834 −0.0294 −0.0458 −0.0898 0.1650 −0.1070 0.0625 0.1008 0.2739 0.1361 0.0180 0.0690 0.1325 0.0127 0.0971 33.6% 15.2%

218


Exhibit 2. Binomial Parameters and Stock Prices in the Four-Step Lattice. Binomial Parameters Risk free rate (rf ) Volatility (␴) Subinterval (T) Proportion of upward movement (u) Proportion of downward movement (d) Growth rate during period T (a) Risk-free probability (p)

BBT

BKFR

6% 35.7% 0.0833 1.1086 0.9021 1.0050 0.499

6% 33% 0.0833 1.0999 0.9091 1.0050 0.502

T=0

T =1

T=2

T=3

T=4

26.68

29.58 24.07

32.79 26.68 21.71

36.35 29.58 24.07 19.58

26.68

29.68

31.67

33.16

40.29 32.79 26.68 21.71 17.67 37.07

12.00

13.20 10.91

14.52 12.00 9.92

15.97 13.20 10.91 9.02

12

13.75

14

15

Lattice for BBT

Actual stock price Lattice for BKFR

Actual stock price

17.57 14.52 12.00 9.92 8.20 17.13

CONNECTING CONCEPTS OF BUSINESS STRATEGY AND COMPETITIVE ADVANTAGE TO ACTIVITY-BASED MACHINE COST ALLOCATIONS Richard J. Palmer and Henry H. Davis ABSTRACT As manufacturers continue to increase their level of automation, the issue of how to allocate machinery costs to products becomes increasingly important to product profitability. If machine costs are allocated to products on a basis that is incongruent with the realities of machine use, then income and product profitability will be distorted. Adding complexity to the dilemma of identifying an appropriate method of allocating machine costs to products is the changing nature of machinery itself. Depreciation concepts were formulated in days when a machine typically automated a single operation on a product. Today’s collections of computer numerically controlled machines can perform a wide variety of operations on products. Different products utilize different machine capabilities which, depending on the function used, put greater or less wear and tear on the equipment. This paper presents a mini-case that requires management accountants to consider alternative machine cost allocation methods. The implementation of an activity-based


219

220

RICHARD J. PALMER AND HENRY H. DAVIS

method allows managers to better match machine cost consumption to products. Better matching of machine costs to products enables better strategic decisions about pricing, mix, customer retention, capacity utilization, and equipment acquisition.

INTRODUCTION This paper presents a mini-case that requires management accountants to consider alternative methods of machine cost allocation and how activity-based logic may assist modern businesses in connecting machine costs in fixed-cost intensive environments to products based on the demands products place on the machine. Better matching of machine costs to products enables better strategic decisions about pricing, mix, customer retention, capacity utilization, and equipment acquisition.

PAT’S CAR RENTALS Pat is the first person in the neighborhood to buy a new car. Impressed by the vehicle, all of Pat’s neighbors express an interest in renting the car to satisfy their various travel needs. Being a friendly (and entrepreneurial) neighbor, Pat wants to rent the car to each neighbor and establish “Pat’s Car Rentals.” However, while renting the car to each neighbor at the same price may maintain harmony on the block, Pat knows that each neighbor plans to use the car quite differently. For example: Neighbor A is a teenager who wants to borrow the car for dating purposes. Pat knows Neighbor A will drive at a high rate of speed about town, blast the CD player, and accelerate and decelerate quickly while making many stops to pick up friends. Neighbor B is wants to borrow the car for a short vacation driving in the nearby mountains. Neighbor B plans to attach his recreational camper by tow ball to Pat’s car for the vacation. Neighbor C plans to deliver telephone books throughout town. This driving will entail many stops and starts as well as ignitions and re-ignitions of the engine. Neighbor D wants to take the car on a trip to a large city near Pat’s town. In the town, Neighbor D will primarily be engaged in start and stop city driving. Neighbor E wants to use the car to take a vacation at the beach. This will entail a long straight drive on the flat Interstate highway from Pat’s town to the coast. Neighbor E will drive the car at the posted speed limits.

Connecting Concepts of Business Strategy and Competitive Advantage

221

Neighbor F want to use the car for short “off-road” adventures. Neighbor F will drive the car off of the main roads to various hunting and fishing locations within the region. Neighbor G is an elderly person who plans limited driving about town, primarily to nearby church social events. Pat estimates that the total automobile cost will equal $40,000. Pat believes that the useful life of the auto depends upon the type of miles driven. In other words, Pat recognizes that the diminution of the value and physical life of the car will differ depending upon the lessee’s pattern of use. If lessees put “easy miles” on the car, Pat estimates the car will go 200,000 miles before disposal; if “hard miles” are put on the car, Pat estimates the car has a useful life of 100,000 miles. Pat has evaluated the local rental car agency pricing scheme as a guide to determine how much should be charged per mile. The local rental agency charges a flat rate per mile to all lessees. However, Pat feels that the agency’s “one size fits all” mileage rate doesn’t make sense given the wide differences in vehicle usage by customers. Pat believes that knowledge of the lessees’ use of the automobile can be used to obtain a competitive advantage and that accounting records should reflect the economic fact that different customers consume the car to different degrees. For example, Pat could use knowledge about customer driving habits to offer a lower rental rate to Neighbor G (the grandparent) than that neighbor could obtain at the local rental agency. By doing so, Pat will attract more neighborhood customers who put “easy miles” on the automobile like Neighbor G. Conversely, Pat’s rental rates for Neighbor A (the teenager) and Neighbor B (heavy load puller) will likely be higher than the local car rental agency, driving these customers to the lower rates of the local car rental agency (which are, in effect, subsidized by neighbor G-type customers). If the local rental agency does not understand and use information about the automobile usage patterns of its customers for pricing decisions, they will eventually be driven out of business by their “hard miles” customer base. Pat reckons that variable costs to operate the vehicle are irrelevant, since these costs are borne by the lessee. Pat seeks your assistance in developing reliable estimate of the relevant costs associated with automobile use. First, Pat wants to develop a costing method that measures the diminution in value associated their particular auto usage patterns. Second, Pat’s Car Rentals must develop a method for monitoring the actual manner of car use by customers. Though presently confident of the neighbors’ planned uses for the automobile, Pat is not confident that potential customers will accurately reveal their usage patterns when different rental rates are applied to different types of auto usage.

222


Fig. 1. Machine and Machine-Related Cost Center Allocation Model.

Another friend provided Pat with a diagram of how an organization should, in theory, allocate machine-related costs (see Fig. 1). Pat doesn’t understand what is meant by intensity, duration, and transaction allocation methods, but has identified the following items as important determinants that decrease the useful life of an automobile: load pulled, number of starts and stops, number of ignitions, average speed, rate of acceleration, miles driven and off-road activities, or type of terrain traversed. In Pat’s mind, each type of automobile use is a “cost driver” that contributes equally to the diminution in the value of the vehicle (e.g. load pulling wears out a car in the same degree as start/stop traffic) and, if performed on a consistent basis, would reduce vehicle life by about 33,333 miles. Required (1) Explain the concept of activity based “cost drivers.” (2) Explain the concepts of: (a) intensity; (b) duration; and (c) transaction drivers. Classify Pat’s determinants of auto usage into one of these three categories and justify your classification scheme. (3) Develop a costing method that recognizes differential auto consumption due to different rental usage patterns.


223

(4) Compare and contrast the cost method developed in (3) with that traditionally assigned to rental cars. (5) Devise a method to accurately monitor actual auto usage patterns of renters. (6) How does Pat’s alternative costing/pricing method provide Pat with a competitive advantage in the rental car business?

PAT’S CAR RENTALS: TEACHING NOTES Background for Instructional Use Faced with the need to raise productivity to survive, especially against low cost competitors in nations such as China, North American companies are pushing toward fully automated production processes (known as “lights-out” or “unattended operations” manufacturing) (see, for example, Aeppel, 2002). As manufacturers continue their inexorable movement toward increased automation, the issue of how to allocate machine costs to products becomes increasingly important to organizational understanding of profitability dynamics. Ignorance of profitability dynamics is dangerous in highly competitive manufacturing contexts. If machine costs are allocated (i.e. depreciated) to products on a basis that is incongruent with the realities of the machine use, then income and product profitability will be distorted over the life of the asset. Pat’s Car Rentals presents a simple scenario about the increasingly important business and accounting problem of allocating fixed costs to products produced or services rendered. We have found that this case is easily comprehended by and provides useful insights to both undergraduate and graduate students, especially those enrolled in advanced cost or strategic cost management courses. There are two major hurdles to negotiate in teaching the case. The first hurdle is to get students to understand the analogy between the “car” and advanced manufacturing equipment. Like advanced manufacturing equipment, the car performs multiple operations based on the demands of different customers. Further, customer demands place differing amounts of stress on machinery (autos) – in ways not always commensurate with machine hours (driving time). Once this analogy is comprehended, students tend view depreciation of machinery quite differently. The second hurdle relates to the topic itself. The significance of depreciation is not adequately addressed in current accounting texts. Occasionally our students have argued that depreciation represents an irrelevant sunk cost and that any effort to allocate these charges is counterproductive.

224


The next sections provide a brief historical perspective on the concept of depreciation and modern application of activity-based principles to the case of Pat’s Car Rentals. Additionally, the machine cost assignment model presented in the case is described in greater detail as a mechanism to obtain a better match between machine resource consumption and machine cost assigned.

A BRIEF HISTORY OF THE MACHINE COST ALLOCATION DILEMMA Pat’s concerns about the relationship between the neighbor’s use and the physical life of the automobile are non-trivial and shared by managers of automation-intensive manufacturing operations. For some time, direct labor has been decreasing while manufacturing overhead (and in particular, machine and machine-related costs) has been an increasing component of product cost. The cost of goods sold at Ford Motor Company’s Romeo Engine Plant, for example, is comprised of 2% direct labor, 23% manufacturing overhead, and 75% direct material (Kaplan & Hutton, 1997). In high technology computer manufacturing, labor and material are even smaller components of cost (Cooper & Turney, 1988).1 The most common incarnation of inaccurate machine cost allocation, an artifact of the popular time-based depreciation method, is product costs that are unrealistically high and product line income figures that are unrealistically low in the early years of production.2 Since the movement toward increased automation is unlikely to abate (particularly given the positive impact of automation on quality), the problem of machine cost allocation and its impact on product costs will only increase in importance.3 Adding complexity to the dilemma of identifying an appropriate allocation of machine cost to products is the changing nature of the machinery itself. Depreciation concepts were formulated in the days when a machine largely automated one particular operation on a product. If all products undergo the exact same operation on the machine, allocating machine costs based on (for example) machine hours is a reasonable approach to assigning machine costs to products. However, today’s machines are very different from days past. Computer numerically controlled (CNC) machines perform a wide variety of operations on products. Collections of CNC machines (each designed for a broad range of functions), connected by a material handling system and controlled by a central computer system, comprise a flexible manufacturing system (FMS). Different products utilize different machine capabilities, which, depending on the capabilities used, put greater or less wear and tear on the equipment.


225

POPULAR DEPRECIATION FRAMEWORKS FOR MACHINERY Understanding machine resource consumption and the method of allocating the consumption of machinery among products are choices organizations make within the boundaries of accounting, legal, and regulatory boundaries. There are numerous ways to calculate depreciation, accounting rules only require that the method by which asset consumption is measured be rational and systematic (i.e. consistently applied across time periods). Until recently the measurement and allocation of depreciation charges to products typically was an immaterial issue. However, as noted above, increased levels of investment in machinery now make depreciation methodology a key element of accounting policy (see, for example, Brimson, 1989). The most popular machine cost allocation schemes are time-based or volumebased single factor models. However, both of these methods distort the cost of products produced by the machine. Specific problems with the use of a single factor method of depreciation, whether time or volume-based, are discussed below. Time-Based Depreciation Models Depreciation methods that recover costs over a fixed period of time assume that value added to the product is independent of individual products and the actual utilization of technology during the recovery period. Time-based models of depreciation dominate modern corporate accounting practice primarily because they guarantee (assuming that a conservative estimate of useful life is employed) that machine cost will be assigned (or “recovered”) by the end of a specified depreciable life. In addition, time-based methods are simple to calculate, require minimal on-going data collection, and are agreeable to virtually all regulatory reporting agencies.4 However, there are several important and well-known problems with the use of time-based depreciation models. First, time-based methods tend to increase current period expense (and hence product costs) compared to volume-based or activity-driven cost allocation schemes. This occurs because productivity often declines significantly in the first year when compared with the prior technology. The productivity drop in the first year is associated with the learning curve that companies must scale as they familiarize themselves with new machinery (Hayes & Clark, 1985; Kaplan, 1986). Hence, time-based depreciation charges are particularly onerous on individual product costs and return on investment calculations in the early years.5

226


Time-based approaches may also lead to dysfunctional behavior. Managers, conditioned to equate time with cost will fear that idle machinery is a drain on profitability (Brimson, 1989). Thus, managers may feel compelled to overproduce, resulting in excess inventory. Business cyclicality also highlights weaknesses in time-based depreciation methods. The problematic interaction of straight-line depreciation with business cycles was identified by a manager at Cummins Engines (Hall & Lambert, 1996) as follows: During the 1980s, the cyclicality of Cummins’ business became more frequent, more pronounced, and longer-lived, to the extent that straight-line depreciation of production equipment over 12 years no longer resulted in a systematic and rational method of allocating cost. The straight-line method over 12 years significantly understated product cost in years of high demand while overstating costs significantly in years of low demand (p. 33).

Volume-Based, Single-Factor Models A popular alternative to a time-based method is to depreciate machine cost on the basis of a measure of expected volume. Volume measures may be inputs (such as actual machine utilization) or outputs (such as standard machine hours for units produced). Volume-based approaches provide a better match of automation expenses to changing economic conditions, resulting in the lessening of the impact of economic cycles. There are, however, problems inherent with use of this methodology. Problems with depreciation methods that employ a single volume-based factor are both technical and behavioral. The technical issues relating to depreciation based solely on output volume measures were summarized by Staubus (1971): Rarely would one of these [output] measures be completely satisfactory because the units of service provided by a particular asset are not homogeneous. Every mile traveled by a motor vehicle is not identical to every other mile traveled; a mile is not a constant measure of service. This is more obviously true when the asset has a long life so that the conditions of its use change a great deal (p. 57, emphasis added).

By extension, all machine hours are not equal. The case of Pat’s Rentals, for example, recognizes that some customers will put “hard miles” (or hours) on the rented automobile while other customers will put “easy miles” (or hours) on his car. Likewise, in manufacturing contexts differences in speed and feed rates of CNC equipment subject the machinery to differing amounts of stress. Further, depreciation amounts based on any one volume-related factor fail to address situations where the machine is not in use. Typically, the machine loses service potential without regard to actual use because of deterioration or obsolescence.6


227

Dysfunctional employee behavior may also be encouraged by the use of any single volume-based metric for overhead allocation. Cooper and Kaplan (1999) argue, for example, that the use of machine hours alone to allocate depreciation encourages acceleration in speed and feed rates to increase production with a minimum number of machine hours. This behavior can damage machinery, reduce product quality, choke downstream bottleneck operations, and encourage outsourcing of machined parts that may result in a “death spiral” for the manufacturer. Staubus suggested that it may be possible to create a refined measure of depreciation that adjusts for unwelcome variations by either by refining the measurement unit itself or arbitrarily varying the weights of measured service units. An example of the former case is John Deere Component Works, where kilowatt-hours were multiplied by a calculated machine “load” factor to assign utility costs to four different machines (Kaplan, 1987). An example of the latter approach would be to use an “adjustment factor” that assigns higher costs to the early miles in the life of transportation equipment. Milroy and Walden (1960) recognized multiple causal factors associated with the consumption of capital resources and suggested “it may well be possible that a [scientific] method could be devised in which consideration would be given to variation in the contribution of service units to firm revenue” (p. 319). One means by which an organization can recognize several significant causes of depreciation of high technology machinery is to develop a multiple-factor model.

MULTIPLE COST DRIVERS FOR MACHINE COST ALLOCATION Over fifty years ago, Finney and Miller (1951) stated that no one factor alone is necessarily the primal causal factor of the consumption of the machine resource. Their support for the use of a multi-factor approach invoked the following logic: Although physical deterioration is undoubtedly affected by use, and although it is proper to adjust depreciation charges to give consideration to varying levels of operations, it does not follow that the depreciation charges should be exactly proportionate to use. The normal periodical depreciation charge may include provisions for: (a) deterioration caused by wear and tear; (b) deterioration caused by the action of the elements; and (c) obsolescence. If a machine which normally is operated for one eight-hour shift daily is, for some reason, operated on a twenty-four hour basis, the proper depreciation may be more or less than three times the normal charge (emphasis added, p. 452).

Historically, some innovative companies have used variations of multi-factor depreciation models. In early days in the railroad industry, for example, the cost of transportation equipment was allocated on the basis of ton-miles as opposed to

228


linear miles (Kaplan, 1985). More recently, Cummins Engines (Hall & Lambert, 1996) managers described their use of a modified units-of-production method that assumes that depreciation of an asset is a function of both time and usage. Consequently, depreciation in periods of extremely low production volume is a fixed amount; yet, as production volume increase above low levels, depreciation is increasingly attributable to volume. Though previous multi-factor depreciation models are both important and practical, they do not provide an overarching framework that can be applied to a broader set of business contexts. The comprehensive multi-factor depreciation model presented in Pat’s Rentals is built upon the activity-based costing logic of Cooper and Kaplan (1999). According to Cooper and Kaplan, there are three types of activity cost drivers: transaction, duration, and intensity. Transaction drivers, such as the number of setup, number of receipts, and number of products supported, count how often an activity is performed. Duration drivers, such as setup hours, inspection hours, and machine hours, represent the amount of time required to perform an activity. Intensity drivers directly charge for the resources used each time an activity is performed, such as an engineering change notice or creation of a pound of scrap. Intensity drivers are the most accurate activity cost drivers but are the most expensive to implement because they, in effect, require a direct charging via job order tracking of all resources used each time an activity is performed.7

A GENERAL MODEL FOR MACHINE COST ALLOCATION There are two significant challenges relating to machine depreciation: (1) identifying the appropriate measure of depreciation (as discussed in earlier sections of this paper); and (2) identifying the best way to assign depreciation (however measured) to the products that pass through a transformation process of the machine. Applying Cooper and Kaplan’s activity driver concepts to the machine cost allocation problem, the instructor can review the two stage causal model of machine resource consumption presented in Fig. 1. In the first stage, the consumption of the machine resource is measured by reference to the machine activities that have transpired (thus the dotted line between intensity, duration, and transaction machine cost activity drivers). In the second stage, the cost of the machine resource consumed is allocated to products or customers on the basis of duration, transaction, and intensity cost drivers associated with their use of the machine. Further definition of these activity cost drivers is discussed next.


229

Duration Drivers An important factor exhausting machine resource is volume – typically measured as the machine’s actual processing time or physical production. Machine utilization time probably is the most popular single activity factor prescribed by proponents of activity-based costing in machine-intensive environments (see, for example, Cooper & Kaplan, 1999). A CNC machine or machine center has an estimated number of productive hours under normal operating conditions. Thus, an hourly machine utilization rate for productive time is an important measure of the consumption of technology resources. However, as noted above, all machine hours are not equal. One factor that modifies the impact of the machine’s actual processing time on the consumption of the technology resource is the “intensity” (or “load”) placed on the machine. This aspect of machine utilization is discussed next.

Intensity Drivers The intensity of machine use (commonly referred to as “load”) is defined as the multiple of the speed and feed rates at which an operation is performed. Any departure from the average load will either place additional strain on or conserve the resource. Within an operation, the path, speed, and feed rate of tool heads are determined by operators. Identifying the intensity of machine use is analogous to understanding wear and tear factors on Pat’s rental automobile. Driving an automobile over mountainous terrain while pulling at trailer or driving at very high speeds will shorten the life of an automobile. Analogously, operating equipment at higher than normal prescribed speed and feed rates is a driver of inordinate consumption of any machine resource. Thus, an intensity factor must be combined with machine hours to address the weaknesses of depreciation attribution based solely on machine hours.8

Transaction Drivers Transaction drivers count how often an activity is performed. A machine operation is a set of machining tasks that can be completed without repositioning the part. The number of operations needed to produce a product is one driver of technology resource consumption. For example, flexible manufacturing systems are capable of performing many different operations. A typical operation contains several transformation activities, such as drilling, boring, and reaming. The ability to

230


perform multiple transformation activities on the same machine is one of the greatest benefits of FMSs. Some machines have as many as 500 different tools in their magazine. A flat charge rate could be established for each machine operation utilized to produce a product. Charging products on the basis of the number of operations would motivate the use of the full range of FMS capabilities by encouraging minimal product re-positioning within and across different operations. In addition, such charges would encourage simplification in product design and the manufacturing process since unneeded features or operations become more costly. This transaction activity consumes the resource by wear and tear associated with changing from one operation to another. In effect, it is a setup charge – the new workpiece has to be fetched by the machine and inserted automatically. Upon completion of that operation another ejection occurs, and another insertion of a workpiece occurs for another operation. A second transaction driver of machine resource consumption is the number of geometric movements a machine must make within an operation. This factor captures the physical complexity of the product. Every movement or shift within an operation places an additional element of wear and tear on the machinery. For example, one circuit board may require many movements (twists, turns, rotations, or shifts) or changes in the tool path to complete diode insertion. On the other hand, the product recipe of another circuit board may place much less demand on the machine. Therefore, a shift rate is a relevant factor to capture the wear-and-tear on an FMS machine. By allocating a portion of the machine cost on the basis of number of shifts or movements, programmers are given an incentive to program the geometric path of the machine more efficiently. In the case of Pat’s Rentals, transaction drivers would include the number of starts and stops and the number of ignitions.

AN ABC MACHINE COST ALLOCATION MODEL TO SOLVE PAT’S CAR RENTAL DILEMMA Step One: Identify duration, intensity, and transaction drivers and measures related to the machinery in use. Table 1 shows the activity drivers and measures in the case of Pat’s Rentals. Step Two: Calculate hourly machine cost allocation rates or rates per unit of output based on combinations of different machine utilization patterns. Table 2 shows how Pat’s rental could assign different depreciation costs of the automobile to different customers based upon duration, intensity, and transaction activities.


231

Table 1. Identification of Transaction, Intensity, and Duration Drivers, Cost Driver Categories, Asset Attributes, and Activities Measures Related to Automotive Machinery. Automobile Machinery Activity Drivers

Activity Cost Driver Category

Asset Attribute

Activity Measure

Amount of time car is driven Amount of miles driven Load being pulled

Duration Duration Intensity

Service life Physical life Service life

Off-road activities

Intensity

Physical life

Speed

Intensity

Service life

Rate of acceleration/deceleration Terrain traversed Start/stops Ignitions/re-ignitions

Intensity Intensity Transaction Transaction

Service life Physical life Service life Service life

Engine hours Odometer Average engine RPM; hookups to trailer Global positioning system (GPS) tracking/shock and strut sensors Global positioning system tracking Brake/tire sensors GPS/Average RPM Brake sensors Ignition sensor

Step Three: As shown in Table 3, apply the different rates calculated in Step Two to assign costs to customers and measure rental profitability. Pat will now show a profit on rental to Customer A if the rental charge includes a depreciation component that is above $0.40 per mile, while profitable rentals may be made to Table 2. Allocation of $40,000 Automobile Costs Depending on Transaction, Intensity, and Duration Drivers.a Automobile Consumption Score (a) Duration

(b) Intensity

(c) Transaction

(d) Cost Adjustment Score (a + b + c)

(e) Expected Automobile Life in Miles

(f) Depreciation Rate Per Mile on ($40,000 vehicle divided by (e))

0 0 0 1 0 1 1 1

0 0 1 0 1 1 0 1

0 1 0 0 1 0 1 1

0 1 1 1 2 2 2 3

200,000 (easy) miles 166,667 166,667 166,667 133,333 133,333 133,333 100,000 (hard) miles

$0.20 0.24 0.24 0.24 0.30 0.30 0.30 0.40

a For illustrative purposes only. See section on assumptions and limitations regarding the measurement

of resource consumption.

232


Table 3. Assigning Machine Costs Based on Customer Usage of Machine- The Case of Pat’s Rentals. Customer

Machine Activity Cost Driver

A (teenager) B (trailer tow in mountains) C (delivery service) D (city driving) E (high speed highway) F (off-road) G (grandparent)

Duration

Intensity

Transaction

1 0 0 0 1 0 0

1 1 1 0 0 1 0

1 1 1 1 0 0 0

Customer Use Score

Depreciation Charge Per Mile

3 2 2 1 1 1 0

0.40 0.30 0.30 0.24 0.24 0.24 0.20

Customer G for half that amount. Other customers fall somewhere in between, depending on usage patterns.

THE CHALLENGE OF MEASUREMENT Recommendations that add complexity to internal record-keeping systems are rarely welcome in the marketplace and the willingness of companies to bear the additional costs of data collection to support a multi-factor depreciation model – computer programming, data collection, and data analysis – are questionable. However, utilization of the proposed model can be justified on the basis of several important facts. First, machine costs are a very large component of total manufacturing charges in some industries. At high technology manufacturing organizations, machine cost allocations can make or break customer or product profitability and alter machine investment decisions. Second, product margins in globally competitive manufacturing are thin; thus increasing the need for companies to understand the connection between products and their consumption of machine resources has never been higher. Third, the cost of technology to measure machine activities continues to drop. Many of the commentators who identified the need of better machine cost allocation lived in the pre-computerized manufacturing world where the ability to gauge technology consumption was non-existent. That problem has slowly dissipated with increasingly sophisticated manufacturing and measurement tools.9 Fourth, the implementation a multi-factor model is easier to accomplish in advanced manufacturing environments since human resistance to change is minimized in highly automated contexts.


233

APPLICATION OF MODEL TO MACHINE-RELATED COSTS The concept that underlies the activity-based machine cost allocation model can be applied to other significant machine-related costs such as repairs and maintenance, tooling (and associated costs of purchasing, material handling, tool crib, etc.) and utilities (including compressed air, water, and power where not more accurately measured by separate metering).

LIMITATIONS AND ASSUMPTIONS The activity-based depreciation model presented in Fig. 1 assumes that each machine resource consumption factor contributes equally to the deterioration of the machine itself. Further, it assumes that there is no multiplicative impact associated with various combinations of the resource consumption factors. However, violation of these assumptions does not undermine the general model. The accountant simply needs to work with production personnel to understand the variety of possible production scenarios and the machine resource consumption pattern associated with each scenario. Engineering simulations may prove valuable in this context.

DISCUSSION AND CONCLUSION The purpose of this paper was to present an instructional mini-case that requires students to contemplate accounting thought on machine cost allocations and consider how activity-based logic may assist modern businesses in connecting machine costs in fixed-cost intensive operations to products based on the demands products place on the machine. Examination of the underlying issues revealed that, while problems with depreciation measurement are not new, their significance has grown commensurate with increases in the level of factory automation. Furthermore, advances in the technology by which to measure machine activity drivers enable organizations to apply activity-based accounting principles to more accurately connect product and customer demands with the consumption of machinery. A better understanding of how to match machine costs to products and customers will enable firms to make better strategic decisions relating to pricing, product mix, customer retention, capacity utilization, and equipment acquisition.

234


Ultimately, the logic embodied in this case should be employed to further the development of more refined models of machine cost allocation. Simulations of machine use and maintenance by engineers, analogous to that presented in this paper, could also provide a useful estimate of machine resource consumption. For purposes of depreciation estimation, accountants may one day defer judgment to these engineering simulations, much as they do now with geologists estimating the oil and gas “reserves” still in the ground. Provided the depreciation method is consistently applied on a uniform basis, such an accounting mechanism would pass muster for financial statement purposes and create no more additional work than is commonly found reconciling financial statement and tax income. Third, it should be noted that not every feature of the multi-factor model described in this paper needs to be employed. The model is intended to portray the need for a more comprehensive consideration of machine resource consumption. Companies would be best served to use the model as a general guide, customizing their own depreciation methodology in a manner consistent with their manufacturing realities.

NOTES 1. Further evidence of the significance of machine cost in business-to-business commerce can be found in special adjustments made for the cost of equipment in cost-based pricing situations (see, for example, the Shionogi and Company case by Cooper, 1996). 2. The unattractive product margins in the early years of the machine investment has been criticized by those who think that accounting practices discourage managers from investing in new machines as soon as necessary (see, for example, Porter, 1992). 3. While not within the scope of this paper, the well-publicized Enron debacle has increased scrutiny of accounting measures and increases the need for more accurate measures of current period income. For instance, Microsoft (Wall Street Journal, 2002b) has been chastised for under-reporting income and the SEC has suggested that adherence to GAAP is no justification for inaccurate financial statements (Wall Street Journal, 2002a). 4. The certainty of complete cost recovery by time-based models should be contrasted to production-based models that assign costs to products based upon, for example, machine hours. The production-based models require reasonably accurate estimates of future production activities. Errors in these estimates can create a situation where technology costs are illogically assigned to products or left unassigned to production entirely. 5. Managerial anticipation of an over-allocation of machine-related costs to products in the early year of a machinery investment (when time-based depreciation methods are employed) will have a significant impact on investment selection in “by the numbers” capital budgeting contexts. Managers who justify projects on the basis of expected ROI will either reject the project outright or accept the project provided that the technology cost be depreciated over an unjustifiably long time frame (Keys, 1986). Under the latter alternative, the understatement of depreciation in the early years will result in an overstatement of ROI


235

and income in those years. Unfortunately, when the technological (but not physical) life of the investment is over, managers have a double incentive to hold on to the old equipment – to avoid the loss on disposition and the higher depreciation charges associated with new equipment. Further, time-based approaches to depreciation typically are inconsistent with the assumptions used in the investment justification decision. For consistency with external reports, most companies use the arbitrary depreciable life span provided in the tax code, rather than depreciating equipment over the shorter of the technological or physical life of the equipment. This incongruity can become more distorted when the company has invested in a FMS that produces products with short product life cycles. 6. In addition, there is greater potential for denominator forecast error when a single volume factor is employed (Staubus, 1971). 7. Cooper and Kaplan (1999) state the some ABC analysts, rather than actually tracking the time and resources required for an individual product or customer, may simulate a duration or intensity driver with a weighted index approach that utilizes complexity indexes. This technique might, for example, entail asking employees to estimate the relative complexity of performing task for different products or customers. Thus, a standard product might get a weight of 1, a moderate complexity product or customer a weight of 3, and a very complex product or customer a weight of 5. 8. In fact, these limits are designed in by manufacturers of advanced manufacturing technologies. In most cases, information concerning these limits can be obtained from the manufacturer. 9. An example of the shrinking cost of measurement technology (applicable to Pat’s Rentals) is global positioning technology. Global positioning systems are now routinely used in the trucking industry and have been used by one car rental agency to penalize lessees by speeding violations (Wall Street Journal, 2001).

ACKNOWLEDGMENTS Special thanks to Marvin Tucker, Professor Emeritus of Southern Illinois University and Mahendra Gupta of Washington University in St. Louis for contributions to the concept in this paper.

REFERENCES Aeppel, T. (2002, November 19). Machines still manufacture things even when people aren’t there. Wall Street Journal, B1. Brimson, J. A. (1989, March). Technology accounting. Management Accounting, 70(9), 47–53. Cooper, R. (1996). Shionogi & Company, Ltd: Product and kaizen costing systems. Harvard Business School Case. Cooper, R., & Turney, P. B. B. (1988). Textronix: Portable Instruments Division (A) and (B). HBS Cases 188–142 and 143. Cooper, R., & Kaplan, R. (1999). The design of cost management systems (2nd ed.). Prentice-Hall.

236


Finney, H. A., & Miller, H. E. (1951). Principles of intermediate accounting. Englewood Cliffs, NJ: Prentice-Hall. Hall, L., & Lambert, J. (1996, July). Cummins engines changes its depreciation. Management Accounting, 30–36. Hayes, R. H., & Clark, K. B. (1985). Exploring the sources of productivity differences at the factory level. In: K. B. Clark, R. H. Hayes & C. Lorenz (Eds), The Uneasy Alliance: Managing the Productivity Dilemma. Boston, MA: Harvard Business School Press. Kaplan, R. S. (1985). Union Pacific (A). Harvard Business School Case 186–177. Kaplan, R. S. (1986, March–April). Must CIM be justified by faith alone. Harvard Business Review, 64(2), 87–93. Kaplan, R. S. (1987). John Deere Component Works (A) and (B). HBS Cases #9–187–107 and 108. Kaplan, R. S., & Hutton, P. (1997). Romeo Engine Plant (Abridged). HBS Case 197–100. Milroy, R. R., & Walden, R. E. (1960). Accounting theory and practice: Intermediate. Cambridge, MA: Houghton Mifflin Company. Porter, M. E. (1992). Capital choices: Changing the way America invests in industry. Research Report co-sponsored by Harvard Business School and the Council on Competitiveness. Staubus, G. (1971). Activity costing and input-output accounting. Homewood, IL: Irwin. The Wall Street Journal (08/28/2001). Big brother knows you’re speeding – Rental-car companies install devices that can monitor a customer’s whereabouts. The Wall Street Journal (2/12/02a). SEC accounting cop’s warning: Playing by rules may not ward off fraud issues. The Wall Street Journal (2/13/02b). SEC still investigates whether Microsoft understated earnings.

CHOICE OF INVENTORY METHOD AND THE SELF-SELECTION BIAS Pervaiz Alam and Eng Seng Loh ABSTRACT We examine the sample self-selection and the use of LIFO or FIFO inventory method. For this purpose, we apply the Heckman-Lee’s two-stage regression to the 1973–1981 data, a period of relatively high inflation, during which the incentive to adopt the LIFO inventory valuation method was most pronounced. The predicted coefficients based on the reduced-form probit (inventory choice model) and the tax functions are used to derive predicted tax savings in the structured probit. Specifically, the predicted tax savings are computed by comparing the actual LIFO (FIFO) taxes vs. predicted FIFO (LIFO) taxes. Thereafter, we estimate the dollar amount of tax savings under different regimes. The two-stage approach enables us to address not only the managerial choice of the inventory method but also the tax effect of this decision. Previous studies do not jointly consider the inventory choice decision and the tax effect of that decision. Hence, the approach we use is a contribution to the literature. Our results show that self-selection bias is present in our sample of LIFO and FIFO firms and correcting for the self-selection bias shows that the LIFO firms, on average, had $282 million of tax savings, which explains why a large number of firms adopted the LIFO inventory method during the seventies.

Advances in Management Accounting Advances in Management Accounting, Volume 12, 237–263 Copyright © 2004 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12011-X

237

238

PERVAIZ ALAM AND ENG SENG LOH

INTRODUCTION Management accounting provides critical accounting information for day-to-day managerial decision-making. The choice of the inventory method influences managerial behavior for purchasing, cash flow management, and tax management. For instance, based on managerial accounting information regarding expected cash flows on various products and services, managers may decide that it is optimal to forego LIFO (last-in, first-out) tax savings. Thus, managers need to have a good understanding of the expected cash inflows and outflows from various segments of the business. They should also have information on available tax saving vehicles, including depreciation, interest, tax losses and the affect of these variables on taxable earnings. Not only could management accounting provide information on the initial selection of inventory method but it could also assist in deciding to continue with the inventory accounting method currently in use. One important role of managerial accounting in this area is in monitoring inventory-levels to prevent LIFO layer liquidation in the event LIFO is used. Over the past twenty years, researchers in accounting have examined various issues arising from a firm’s choice of accounting methods. Much of this literature has been on the choice of the inventory costing method. Early research on inventory selection method estimated the tax effects of the LIFO vs. FIFO (first-in, first-out) method under the assumption that operating, financing, and investing activities remain unaffected as a result of a change in inventory method. Researchers have previously recognized that this ceteris paribus assumption ignores endogeneity and self-selection of LIFO and FIFO samples (see Ball, 1972; Hand, 1993; Jennings et al., 1992; Maddala, 1991; Sunder, 1973). The endogeneity problem in the choice of LIFO vs. FIFO method is particularly important because the tax effects of the inventory method may affect firm valuation (see Biddle & Ricks, 1988; Hand, 1995; Jennings et al., 1992; Pincus & Wasley, 1996; Sunder, 1973). However, it is not possible to observe what the managerial decision would have been had they used an inventory method different from the method currently in use. Hence, a number of studies have developed “as-if” calculations to estimate the tax effects of LIFO vs. FIFO (see Biddle, 1980; Dopuch & Pincus, 1988; Morse & Richardson, 1983).1 These studies estimate a LIFO firm’s taxes as if it was a FIFO firm and LIFO taxes for a FIFO firm. The “as-if” approach assumes that a firm’s managerial decisions would have remained unchanged with the use of an alternative method. The purpose of this study is to re-examine the choice of LIFO vs. FIFO by using Heckman (1976, 1979) and Lee’s (1978) two-stage method that incorporates

Choice of Inventory Method and the Self-Selection Bias

239

self-selection and endogeneity. The Heckman-Lee approach is used to incorporate the endogenous choice of inventory method in the tax estimation equation. The explicit inclusion of the inter-dependence of the choice of the inventory method and the tax effects of the choice distinguishes this paper from those extant in the literature. We use Lee and Hsieh’s (1985) probit model for the first-stage inventory choice model and develop a tax estimation function using prior literature for the second-stage analysis. Finally, we compare Heckman-Lee based measure of tax savings to those developed using ordinary least squares (OLS). Our analysis leads to the following results. First, following Lee and Hsieh (1985), we derive the reduced-form probit estimates for the inventory choice method. The significance and sign of the coefficients of our probit estimates are generally similar to those of Lee and Hsieh. Second, the self-selection bias is a significant factor in analyzing whether firms choose to use LIFO or FIFO. Firms, on average, choose the inventory method that gives them the largest tax benefit. Third, the results of structured-probit show (not reported) that the predicted tax savings has a significant positive coefficient suggesting that firms are likely to select LIFO when predicted FIFO taxes are more than actual LIFO taxes. Fourth, correcting for selfselection bias enables us to infer that, on the average, LIFO firms would pay more taxes if they were using FIFO, and FIFO firms would pay lower taxes if they were using LIFO. Our results for selectivity-adjusted approach show $282.2 million of mean (median $297.4 million) tax savings for LIFO firms and $12.3 million (median $1.2 million) of tax savings foregone for FIFO firms. Results based on ordinary least squares (OLS) estimate indicate that both LIFO (means $40.6 million, median $2.6 million) and FIFO firms (mean $11.3 million, median $5.0 million) had foregone tax savings by choosing the method they were using. In other words, these firms would have had tax savings had they used the alternative method. We have greater confidence in the results of the selectivity approach because these results are econometrically derived and are based on variables, which are largely accepted in the LIFO and tax literature. The larger tax savings of LIFO results under the selectivity-based approach explains why a large number of firms had adopted LIFO inventory valuation method during the seventies. We also recognize that FIFO firms could have obtained sizable tax savings had they switched to LIFO. However, firms do operate under various restrictive conditions, which suggest that tax minimization is not the only objective function a firm’s management desires to achieve (Scholes & Wolfson, 1992). Hence, it appears that a firm’s choice of the inventory method is a rational economic decision even when an alternative method could have produced larger tax savings.2

240


PRIOR RESEARCH A number of studies have attempted to explain why firms do not use the LIFO method in periods of rising prices and thus forego the opportunity of potential tax savings (see Adel-Khalik, 1985; Hunt, 1985; Lee & Hsieh, 1985). One explanation advanced is that FIFO firms are concerned about the drop in stock prices upon the adoption of LIFO. Still another explanation is that the cost of LIFO conversion may be more than the tax benefits of adoption. For instance, using 1974–1976 LIFO data, Hand (1993) estimated that the cost of LIFO adoption for his sample of firms was as high as 6% of firm value, a sizable cost for most firms. Empirical studies on whether LIFO tax saving is valued by investors has been extensively studied. Many of these studies suffer from the problems of event date specification, contaminating events, firm size, etc. (Lindahl et al., 1988). Kang’s (1993) model predicts that positive price reaction to LIFO adoption occurs only when the expected LIFO adoption costs are less than expected LIFO tax savings. He argues that the positive stock price reaction will occur because the switch to LIFO will recoup previously lost LIFO tax savings. Some studies demonstrate positive stock price reaction surrounding LIFO adoption (Ball, 1972; Biddle & Lindahl, 1992; Hand, 1995; Jennings et al., 1992; Sunder, 1973) while other studies have reported negative market reaction to LIFO adoption (see Biddle & Ricks, 1988; Ricks, 1982). Pincus and Wasley (1996) results show some degree of market segmentation. They found positive market reaction to OTC firms and negative market returns for NYSE/ASE firms. Finally, Hand’s (1993) results indicate that the LIFO adoption or non-adoption decision resolves uncertainty regarding LIFO tax savings. Contracting cost theory also provides reasons why firms may not adopt the LIFO inventory method. LIFO adoption decreases asset values and net income, potentially causing some firms to be in violation of debt covenants. Furthermore, managers on bonus contracts may not want lower LIFO earnings because the use of LIFO may reduce their total compensation. Adel-Khalik (1985), Hunt (1985), and Lee and Hsieh (1985) provide some evidence that debt covenants help to explain the choice of inventory method but compensation plans do not.3 Another reason why firms decide to use a particular method is that firms differ systematically in the nature of the production-investment opportunity set available to them. Therefore, the LIFO method is an optimal tax reporting choice for some firms and not for others. A common empirical approach to estimate the amount of tax savings firms could have obtained from an alternate method, other than their observed choice of inventory accounting method, is the as-if method (see Biddle, 1980; Biddle & Lindahl, 1982; Dopuch & Pincus, 1988; Morse & Richardson,


241

1983; Pincus & Wasley, 1996). We implement an alternative approach, which relies on the work of Heckman (1976, 1979) and Lee (1978).

CONCEPTUAL FRAMEWORK This section presents the conceptual basis for the empirical analysis that follows. Assume firms have only two inventory valuation methods available: LIFO or FIFO. A typical firm’s decision to adopt LIFO depends on its own assessment of the benefits to be gained and the costs that must be incurred. As previously stated, LIFO costs are associated, among others, with implementation, negative market reaction, and contracting costs. In addition, there may be LIFO layer liquidation costs resulting from price decline (e.g. electronics industry). As a result, a firm would rationally choose to use LIFO only if the expected benefits outweigh the expected costs.4 Otherwise, it would remain as a FIFO firm. With the LIFO or FIFO status thus determined, the firm’s LIFO benefits depend on their operating and financial characteristics, the nature of the industry, and the provisions of the tax code. Let the benefit of adopting the LIFO method be measured by the tax savings received by the firm. We assume that tax savings is a benefit to the firm because the increased cash flow widens the set of feasible production-investment opportunities and, thus, improves the firm’s long-term prospects. However, foregoing tax benefits may be an optimal strategy for a firm in a framework of Scholes and Wolfson (1992) or when the inventory valuation method is used to signal firm value.5 Let the total taxes T paid by each type of firm be written as: T L = aX L + e L

(1)

T F = bX F + e F

(2)

where the subscripts L and F denote the LIFO and FIFO firms, X is a vector of explanatory variables common to both groups of firms, a and b are vectors of coefficients, and e is the random error term. Firms choose the type of inventory valuation method that will maximize their overall tax benefits given other constraining factors. Thus, they choose the LIFO method if TS = (T F − T L ) > C

(3)

where TS is the positive dollar tax savings (assuming T F > T L ), and C is the associated dollar cost, of choosing LIFO. There is unlikely to be a single observed number in the firm-level data set to represent the cost of choosing LIFO, although, in practice, many reasons can be found to justify the assumption that choosing LIFO is not cost-free. Prior literature suggests that the dollar cost of LIFO adoption may

242


reduce management compensation (Hunt, 1985), lead to possible violation of debt covenants (Hunt, 1985; Morse & Richardson, 1983), and increased fixed costs of computing ending inventory value (Hand, 1993; Morse & Richardson, 1983). Using these arguments, we assume that there is a systematic relationship between LIFO costs (C) and these factors; that is, C = cY + n

(4)

where Y is a vector of regressors, c is a vector of coefficients, and n is the error term. Substituting Eq. (4) in Eq. (3), we get T F − T L > cY + n

(5)

This expression may be written as a probit equation: I = LIFO I = FIFO

if I ∗ > 0 if I ∗ ≤ 0

where I ∗ = ␣0 + ␣1 (T F − T L ) + ␣2 Y − ␮

(6)

where ␮ is the error term of the probit function. In practice, we observe the tax payments of either the LIFO or FIFO firm, but never both simultaneously, implying that the probit Eq. (6) cannot be estimated directly. One way to proceed is to estimate the tax Eqs (1) and (2), via ordinary least squares (OLS) and use the predicted values in Eq. (6). However, because of the self-selected nature of the LIFO and FIFO firms, the expected mean of the error terms in the tax equations are non-zero, i.e. E(e L |I = LIFO) = 0 and E(e F |I = FIFO) = 0. Thus, the OLS estimation of the tax equations leads to inconsistent results. There is no guarantee that the estimated coefficients will converge to the true population values even in large samples. To avoid this bias, we proceed via the two-stage method suggested by Heckman (1976, 1979) and Lee (1978).6 We begin by estimating the reduced form probit equation found by substituting Eqs (1) and (2) into Eq. (6), I = LIFO

if I ∗ > 0

I = FIFO

if I ∗ ≤ 0

where I ∗ = ␥0 + ␥1 X + ␥2 Y − (␮ − e).

(7)


243

Equation (7) is the form of the estimation equation commonly found in the LIFO determinants literature. After estimating (7), we derive the inverse Mills’ ratios, −␸(u) ␭L = , (8) (u) ␸(u) ␭F = (9) 1 − (u) where u is the predicted value of the error term from the reduced form probit, ␸ is the standard normal probability density function (pdf) for u, and its cumulative density function (cdf). In the second stage, the coefficient of the lambda terms in the tax functions serve as sample covariances between the tax function and the criterion I∗ (i.e. a 2 = ␴L␮ and b 2 = ␴F␮ ). Hence, the following the tax functions are obtained: T L = a 1 X L + a 2 ␭L + ␧L

(10)

T F = b 1 X F + b 2 ␭F + ␧F .

(11)

Equations (10) and (11) are then estimated using the self-selectivity and the OLS approaches to construct predicted FIFO tax payments for observed LIFO firms and predicted LIFO taxes for observed FIFO firms.7 In essence, the Heckman-Lee twostage procedure treats the self-selection bias as arising from a specification error: a relevant variable ␭ is omitted from each of the tax equations. Statistical significance on a2 and b2 shows that these covariances are important and that management selection of the LIFO or the FIFO inventory valuation method is not random. In short, self-selection is present. Interpretation of the firms’ behavior depends on the signs of ␴L␮ and ␴F␮ , which may be either positive or negative. For instance, if ␴L␮ < 0 then firms whose expected LIFO taxes are lower than average, should have a lower chance of being a FIFO firm. Similarly, if ␴F␮ < 0 then firms whose expected FIFO taxes are lower than average should have a lower chance of being LIFO firms. Although these covariances can bear any sign, model consistency requires that ␴L␮ > ␴F␮ (see Trost, 1981). This condition of the covariance term (␴L␮ > ␴F␮ ) ensures that the expected FIFO taxes of FIFO firms will be less than their expected taxes if they switched to LIFO status. Similarly, the expected tax payments of LIFO firms will remain less than their expected tax payments if they switched to FIFO. An important assumption of the Heckman-Lee model is that the error terms in the structural equations (␧L , ␧F , and ␮) are joint-normally distributed. Inconsistent estimates result if the underlying population distribution is non-normal (although, strictly speaking, this problem exists with any parametric model). Given the perceived rigidity of the joint-normality assumption, researchers have suggested

244


alternative approaches based on nonparametric and semi-parametric estimators (see, for example, Duncan, 1986; Heckman, 1976, 1979; Manski, 1989, 1990; Ming & Vella, 1994). Unfortunately, these approaches are often limited in their applicability. For instance, the nonparametric bounds model in Manski (1990) is defined for only two regressors. Thus, one might gain robustness in estimates but may not be able to exploit the breadth of data available. This issue remains unsettled at the present time, leaving the Heckman-Lee model as the accepted dominant vehicle for the empirical analysis of selection bias. In this study, we use Pagan and Vella (1989) for the joint normality test. The results of this test are reported in Note 23. In order to assess tax savings under different regimes we use three different approaches: (1) the estimated tax savings is calculated as the difference between predicted taxes less actual taxes and is used as an independent variable in a structured probit; (2) the coefficients from selectivity-adjusted tax equations are used to calculate alternative dollar FIFO tax savings for observed LIFO firms, and vice versa; and (3) the coefficients from OLS tax equations are used to compute LIFO (FIFO) dollar tax savings for FIFO (LIFO) firms.8

MODEL SPECIFICATION AND DATA SELECTION Determinants of Inventory Method Selection We use Lee and Hsieh’s (1985) model for purposes of estimating the reduced-form probit model described in Eq. (7). We select their model because of the comprehensiveness of the variables examined and the theoretical justification for the selection of those variables. They test the joint effect of political cost, agency theory, and Ricardian theory on the LIFO-FIFO decision, by using eight proxy variables and an industry dummy to capture the features of the production-investment opportunity set that are pertinent to the choice of the inventory accounting method. The variables they use are: firm size, inventory variability, leverage, relative firm size, capital intensity, inventory intensity, price variability, income variability, and industry classification.9 Thus in this study, the inventory choice model is expressed as follows: I = ␥0 + ␥1 LGTASSTit + ␥2 INVVARit + ␥3 LEVit + ␥4 RELASSTit + ␥5 CIit (+)

(?)

(−)

(+)

+ ␥6 INVMit + ␥7 CPRICEit + ␥8 INCVARit + ␥9 IDNUMt ␧it (−)

(+)

(−)

(?)

(+)

(12)


245

where: I = 1 for LIFO firms and I = 0 for FIFO firms; LGTASST = firm size computed as the log value of total assets; INVVAR = inventory variability computed as the coefficient of variation (variance/mean) for year-end inventories; LEV = agency variable derived as the ratio of long-term debt less capitalized lease obligations to net tangible assets; CI = capital intensity variable computed as the ratio of net fixed assets to net sales; RELASST = relative firm size derived as the ratio of a firm’s assets to the total industry assets; INVM = inventory intensity computed as the ratio of inventory to total assets; CPRICE = price variability derived as the relative frequency of positive price change for each four-digit SIC industry code; INCVAR = accounting income variability as the coefficient of variation (variance/mean) of before tax accounting income; and IDNUM = captures the industry effect by assigning a dummy variable to each of the 19 two-digit industries. Thus, IDNUM is a vector of 19 two digit SIC codes. Table 1 provides further description of the variables used in Eq. (12). The table lists the variables used, their description, and the Compustat data items used to derive the variables. A brief description of the reasons for the selection of the regressors used in the probit function (Eq. (12)) follows.10 Ceteris paribus, larger firms (proxied by LGTASST) are likely to adopt LIFO because of their comparative advantage in absorbing costs of LIFO conversion and related bookkeeping and tax-reporting costs. The high inventory variability (INVVAR) suggests that the cost of inventory control may be higher because of the possibility of liquidation of LIFO layers or possible excess inventories. On the other hand, it is likely that adopting LIFO may lead to lower inventory variability in order to maintain LIFO layers. Hence, it is difficult to predict the association of the INVVAR variable to LIFO use. The leverage variable (LEV) serves as an agency proxy. Firms with higher leverage are more likely to default on debt covenant restrictions (Smith & Warner, 1979), driving them to choose income-increasing accounting methods. Hence, the leverage variable is likely to be negatively related with the use of LIFO method. The relative firm size (RELASST) is a measure of size with respect to industry. It is expected that relatively larger firms in an industry will have a comparative advantage in using LIFO. Firms with high values of capital intensity (CI) generally possess necessary resources to engage in extensive financial and production planning needed to

246


Table 1. Operationalization of Variables Used for Probit and Tax Functions. Variable

Definition

I

Dependent variable used in probit estimation where I = 1 if the firm uses LIFO and 0 if the firm uses FIFO (59). Dependent variable used in tax functions. Defined as the natural logarithm of total income tax expenses (16). Firm size measured as log of total assets (6). Coefficient of variation (variance/mean ratio) for year-end inventories. Leverage ratio computed by dividing long-term debt less capitalized lease obligations to net tangible assets (9−84)/(6−33). Ratio of a firm’s assets to the total of industry assets based on the SIC four-digit industry codes. Capital intensity measured as net property, plant, and equipment divided by net sales (8/12). Inventory materiality computed as a ratio of inventory to total assets (3/6). Relative frequency of positive price changes for each SIC four-digit industry code over the sample period. Producer Price Index was obtained from the publications of the U.S. Department of Commerce. Coefficient of variation (variance/mean ratio) of before tax accounting income. Property, plant, and equipment, net (8) divided by the market value of equity at the fiscal year-end (24 × 199). Non-debt tax shield computed as the ratio of the sum of deprecation and investment tax credits (103 + 51) to earning before interest, taxes, and depreciation (172 + 15 + 16 + 103). Available tax savings measured as natural logarithm of tax loss carryforwards (52) multiplied by cost of goods sold (41). Inventory to sales measured as inventories (3) to net sales (12). Net operating tax loss carryforward measured as a ratio of net operating tax carryforward to net income before interest, taxes, and depreciation expenses (52/(172 + 15 + 16 + 103)).

LNTXPAY LGTASST INVVAR LEV RELASST CI INVM CPRICE

INCVAR FIXED NDTS

LATS INVTS TLCF

Note: Compustat data item numbers in parentheses.

use LIFO. Hagerman and Zmijewski (1979), Lee and Hsieh (1985), and Dopuch and Pincus (1988) suggest that large capital-intensive firms have a comparative advantage in adopting LIFO. The inventory to total assets ratio (INVM) serves as a proxy for measuring how efficiently the inventory has been managed. Following Lee and Hsieh (1985), INVM is expected to be negatively associated with the use of the LIFO method. The price variability (CPRICE) variable is a proxy for inflation. The higher the inflation rate, the higher the likelihood that firms would adopt LIFO. Lee and Hsieh (1985) argue that production-investment opportunity sets will vary from industry to industry. Therefore, a dummy variable is assigned to each of the two-digit SIC industries.


247

Specification of Tax Functions The regressors for the tax functions were identified based on the review of the relevant tax literature (see Biddle & Martin, 1985; Bowen et al., 1995; Dhaliwal et al., 1992; Trezevant, 1992, 1996). The tax functions are listed below: T L = ␦0 + ␦1 FIXEDLt + ␦2 NDTSLt + ␦3 LATSLt + ␦4 INVTSLt + ␦5 TLCFLt + ␧Lt

(13a)

T F = ␦0 + ␦1 FIXEDFt + ␦2 NDTSFt + ␦3 LATSFt + ␦4 INVTSFt + ␦5 TLCFFt + ␧Ft

(13b)

where: TL or TF = the logarithmic value of total taxes for LIFO or FIFO firms, respectively; FIXED = net property, plant, and equipment divided by the market value of equity;11 NDTS = non-debt tax shield derived as the sum of depreciation and investment tax credits divided by earnings before interest, taxes, and depreciation; LATS = available tax savings measured as the logarithmic value of tax loss carryforwards times cost of goods sold; INVTS = inventory turnover measured as inventories to net sales; and TLCF = net operating loss carryforward to net income before interest, taxes, and depreciation. The variables used in tax functions (13a) and (13b) are based on the assumption that a firm’s taxes depend upon net fixed assets, non-debt tax shield, tax savings available from adopting LIFO, efficiency of inventory management, and the amount of the tax loss carryforward. It is important to note that firms do trade-off or substitute various tax shields to minimize the marginal tax rate. The model used in this study examines not only the effect of the individual coefficients in the tax function but also the joint effects of these coefficients. Thus, when high tax shields increase the possibility of tax exhaustion, the firm is likely to have a lower marginal tax rate which may decrease the likelihood of LIFO use. Ceteris paribus, we expect that firms with relatively high values of net property, plant, and equipment scaled by the market value of equity (FIXED) are likely to pay lower taxes. FIXED is a measure of debt securability (Dhaliwal et al., 1992; Trezevant, 1992). Firms with a larger proportion of their assets represented by fixed assets are likely to raise larger amounts of debt or lower the cost of financing (Titman & Wessels, 1988). Therefore, the variable FIXED provides a tax shield by

248


enhancing the possibility of increased debt financing, which increases the level of interest deductibility, and consequently the FIXED variable indirectly lowers taxes. Assuming no available substitution of tax shields, the higher proportion of nondebt tax shield (NDTS) would lower the marginal tax rate, and therefore lower will be the taxes.12 The variable LATS is a proxy measure for tax savings. It is computed by multiplying the cost of goods sold with tax loss carryforwards. This measure is based on the argument that firms with relatively higher cost of goods sold and tax loss carryforwards are likely to pay lower taxes (see Bowen et al., 1995). Therefore, the expected sign of the coefficient for the LATS variable is negative.13 The variable INVTS represents efficiency in inventory management (Lee & Hsieh, 1985). The INVTS coefficient is expected to be negatively associated with taxes. This relationship could be best explained by using the following illustration. Suppose, net sales increases from $150,000 to $175,000, and cost of goods sold and ending inventory remain unchanged at $80,000 and $20,000, respectively. This would cause inventory to sales ratio (INVTS) to decrease from 13.3% to 11.4%, increasing gross margin from $70,000 to $95,000 thereby increasing taxes assuming that the marginal tax rate is the same as in the previous year. Finally, the variable TLCF is expected to be negatively associated with taxes for LIFO firms. In other words, firms are less likely to use LIFO if they have tax loss carryforwards, which could be used to shield taxes.14 Auerbach and Porterba (1987) indicate that firms expecting persistent loss carryforwards are likely to experience lower marginal tax rates. On the other hand, firms are more likely to use FIFO or other income-increasing methods even if they have tax loss carryforwards in the event the alternative available tax shields are tax exhaustive. Thus, the expected sign of the TLCF coefficient cannot be predicted. Table 1 gives further description of the variables used in the tax functions. It also gives the data item numbers used for extracting financial statement values from Compustat tapes.

Sample and Data Collection The data for the variables used in this study were obtained from the back data Compustat files. The sample firms were obtained from 1973 to 1981 years, a period of historically high inflation rates in the United States during which firms adopting LIFO could obtain substantial tax savings. This yielded an initial sample of 10,777 observations for firms using either the LIFO or FIFO inventory method. Since firms use a combination of inventory methods for financial reporting, LIFO firms are those who use LIFO for most of their inventory accounting and FIFO


249

Table 2. Sample Selection and Distribution by Industries (1973–1981). Panel A: Sample selection Number of observations on the back data compustat files for which the inventory method is either LIFO or FIFO from 1973 to 1981 Number of missing observations Number of observations where the primary inventory method is either LIFO or FIFO Number of LIFO observations (number of firms = 247) Number of FIFO observations (number of firms = 1006) Panel B: Sample Distribution by Industries Industry

01–19 (Agricultural products) 20–39 (Manufacturing) 40–49 (Transportation & utilities) 50–59 (Wholesale & retail products) 60–69 (Finance, insurance & real estate) 70–89 (Services) 90–99 (Public administration) Total number of sample observations

10,777

4,687 6,090 1,050 5,040 Number of Observations LIFO

FIFO

17 872 – 225 16 20 –

204 3,255 127 699 110 563 82

1,050

5,040

firms are those who use FIFO as their predominant inventory valuation method. After eliminating missing observations, the total sample is made of 6,090 observations. Of this number 1,050 (247 firms) represents LIFO observations and 5,040 observations (1006 firms) are of FIFO firms. Table 2, Panel A lists the sample selection procedure.15 Table 2, Panel B shows the two-digit SIC code industry composition of the LIFO and FIFO samples. The LIFO group consists of 1,050 observations distributed over 5 different two-digit industries with the manufacturing industry being the largest followed by the wholesale and retail industries. The FIFO observations are distributed over seven different two-digit industries where the largest concentration is also in the manufacturing industries followed by the wholesales and retail industries. Overall, the sample distribution for the two groups of firms appears to be concentrated in the manufacturing, wholesale, and retail industries. Table 3 presents descriptive statistics for the variables used for multivariate analyses, classified by LIFO and FIFO groups. We corrected for price inflation whenever a variable entered as a dollar value, using 1982 as the base-year and

250


Table 3. Sample Descriptive Statistics. Variable

Lntxpay Lgtasst Invvar Lev CI Relasst Invm Cprice Incvar Fixed Ndts Lats Invts Tlcf Tasst Sales Invent

LIFO

FIFO

Mean

Std. Dev.

Med.

1.294 4.837 0.134 0.148 0.243 0.005 0.277 6.828 0.138 0.883 0.292 2.408 0.166 −0.012 1177.25 1509.14 170.04

3.926 1.954 0.121 0.115 0.185 0.017 0.141 4.194 3.095 3.143 1.918 3.408 0.086 0.722 4362.33 6518.58 529.41

2.004 4.550 0.096 0.144 0.195 0.000 0.269 6.300 0.183 0.140 0.186 1.954 0.155 0.000 94.597 173.926 26.485

t-Value

Mean

Std. Dev.

Med.

−0.961 3.184 0.486 0.183 0.320 0.003 0.270 5.587 0.303 0.774 0.152 5.590 0.186 −0.229 209.597 293.798 50.935

4.461 1.921 0.737 0.228 0.973 0.018 0.164 3.921 5.398 3.336 1.545 3.694 0.135 12.343 928.597 1439.72 247.80

0.260 3.081 0.253 0.150 0.171 0.000 0.278 5.000 0.338 0.080 0.142 4.716 0.176 0.000 21.777 33.049 4.789

15.2∗∗∗ 25.28∗∗∗ 15.40∗∗∗ 4.24∗∗∗ 2.51∗∗∗ −3.85∗∗∗ 1.26 5.86∗∗∗ 0.95 −0.97 2.54∗∗∗ 5.91∗∗∗ 4.62∗∗∗ −0.57 14.27∗∗∗ 11.92∗∗∗ 11.14∗∗∗

Note: Tasst, sales, and inventories represents total assets, sales, and inventories in millions of dollars. All other variables are defined in Table 1. The t-value tests differences in means between LIFO and FIFO samples. ∗∗∗ , ∗∗ significant at 0.01, and 0.05 levels, respectively. Dollar values adjusted for price inflation using 1982 as the base year.

CPI as the index. Table 3 shows the mean, median, and standard deviation values for each of the two groups. Also given is the t-value for testing significant differences in mean values for each of the variables. The t-test shows that with the exception of INVM, INCVAR, FIXED, and TLCF, the remaining variables used in the probit and tax functions are significantly different between LIFO and FIFO groups thereby suggesting that the two groups differ from each other on several dimensions.16 Table 3 also provides statistics on selected financial statement variables for each of the two groups of firms. The total assets (TASST) of LIFO firms (median value of $94.6 million) are about four times the value of total assets for FIFO firms (median value of $21.8 million). Similarly, the size of LIFO inventory (median value of $26.5 million) is nearly five times the size of inventory for FIFO firms (median value of $4.8 million). The results of the Kolmogrov-Smirnov and Shapiro-Wilks tests show that the observed distribution of individual variables is not significantly different from a normal distribution.


251

EMPIRICAL RESULTS Estimates of Reduced-Form Probit Equations Table 4 shows the estimated coefficients for the reduced-form probit Eq. (12) in which the dependent variable is a dichotomous dummy variable defined to be unity if the firm is LIFO and zero if FIFO. We present the results without the coefficient for the industry variable. Columns 1 and 2 provide estimated regression coefficients and t-values. Table 4, Columns 1 and 2 shows that sample firms are more likely to be LIFO firms because of price-level increases. The inflation (CPRICE) variable is statistically significant at the 1% level. Columns 1 and 2 also shows that firms may not use the LIFO inventory method because of high variability of inventories (INVVAR), high leverage (LEV), high capital intensity (CI), and relatively high inventory as a component of total assets (INVM). The table shows that with the Table 4. Estimated Coefficients for Reduced-Form Probit Equations. I = ␥0 + ␥1 LGTASSTit + ␥2 INVVARit + ␥3 LEVit + ␥4 RELASSTit + ␥5 CIit + ␥6 INVMit + ␥7 CPRICEit + ␥8 INCVARit + ␧it Variable

Coefficient

t-Value

1

2

INTERCEPT LGTASST (+) INVVAR (?) LEV (−) RELASST (+) CI (+) INVM (−) CPRICE (+) INCVAR (−)

0.001 0.025 –2.824 –0.787 –0.663 –0.198 –1.113 0.025 –0.008

0.008 1.194 –11.446*** –4.795*** –0.563 –1.786* –5.932** 4.555*** –1.519

No. of observations Chi-square Log-likelihood Estimated R2 Percent correct classification

6090 906.4*** –2346.3 16.0% 83.0%

Note: Expected sign of the coefficients is in parentheses. CPRICE is the relative frequency of price increases in each industry during the 1973–1981 period. See Table 1 for definition of variables. ∗ Significant at 0.10 level. ∗∗ Significant at 0.05 level. ∗∗∗ Significant at 0.01 level.

252


exception of the CI variable, which is significant at 10%, the coefficients for INVVAR, LEV, INVM variables are statistically significant at less than 5% and the sign is generally in the expected direction.17

Estimates of Income Tax Equations Table 5 presents the estimates of selectivity-adjusted income tax equations in Columns 1 and 2. The dependent variable in both equations is the natural logarithmic value of the total income taxes for the year.18 All regressors are as defined in Table 5. Estimated Coefficients for Income Tax Equations. T = ␥0 + ␥1 FIXEDt + ␥2 NDTSt + ␥3 LATSt + ␥4 INVTSt + ␥5 TLCFt + ␧t Variable (Expected Sign)

Intercept FIXED (−) NDTS (−) LATS (−) INVTS (−) TLCF (?) LAMBDA Num. of obs. Adj. R-square F-value

Selectivity-Adjusted

OLS

LIFO

FIFO

LIFO

FIFO

1

2

3

4

6.666 (18.07)*** −0.145 (−4.04)*** −0.294 (−5.37)*** −0.518 (−6.55)*** −4.603 (−3.67)*** 0.558 (4.04)*** −3.612 (−13.09)***

−2.152 (−12.65)*** −0.103 (−4.39)*** 0.027 (0.55) −0.233 (−5.96)*** −5.158 (−8.43)*** 0.024 (3.44)*** −9.614 (−29.43)***

3.205 (11.13)*** −0.071 (−1.32) −0.270 (−2.64)** −0.513 (−3.29)*** −9.255 (−5.49)*** 0.565 (1.27)

0.761 (6.04)*** −0.130 (−4.47)*** 0.047 (0.46) −0.196 (−3.99)*** −3.863 (−6.47)*** 0.029 (2.47)**

1050 0.243 57.17***

5040 0.243 270.33***

1050 0.123 27.72***

5040 0.042 34.74***

Note: See Table 1 for definition of variables. Lambda is the inverse Mills’ ratio derived as: ␭ = [−␸(u)/(u)], where u is the predicted value of the reduced form probit, ␸ is the standard normal probability density function for u, and is its cumulative density function. a Figures in parentheses are t-values based on heteroskedasticity-consistent variance-covariance matrix derived in Heckman (1979), Columns 1 and 2, and White (1980), Columns 3 and 4. ∗∗ Significant at 0.05 level. ∗∗∗ Significant at 0.01 level.


253

Table 1. As stated in Section 3, the LAMBDA terms are the inverse Mills ratio.19 The statistical significance of the LAMBDA term in Columns 1 and 2 indicate a non-zero covariance between the error term in the inventory choice and the LIFO (FIFO) tax equations. In short, the self-selection of firms into the LIFO (FIFO) category is confirmed.20 The negative sign of LAMBDA terms show that firms, which expect to pay higher than average taxes in the LIFO (FIFO) category are less likely to be LIFO (FIFO) firms. Thus, the average firm’s choice of the LIFO or the FIFO method for inventory valuation in this sample is consistent with rational decision-making. The typical LIFO or FIFO firm in the sample would have been worse off had it chosen the alternate method. Note also that the sign pattern on the LAMBDA terms satisfy the condition for model consistency: ␴L␮ > ␴F␮ . The lambda coefficient for LIFO firm is −3.612 and for FIFO firms it is −9.614 and both these values are statistically significant. Other results shown in Table 5, Column 1 are noteworthy. In Column 1, all of the remaining regressors are significant at the 1% level and the sign of the regression coefficients are generally in the expected direction for LIFO firms with the exception of TLCF. We find that LIFO firms’ taxes are inversely related to tax shield provided by fixed assets (FIXED), non-debt tax shield (NDTS), available tax savings (LATS), inventory turnover (INVTS) and positively related to tax loss carryforwards (TLCF). In Table 5, Column 2, the results for FIFO firms show that the sign of the coefficients are in the expected direction for each of the regressors, with the exception of TLCF. Similar to LIFO firms, the positive sign of the TLCF coefficients suggest that firms paying higher taxes also have higher values of TLCF on their books. Taken together, these reported sign patterns are largely consistent with the theoretical predictions listed in Section IV.21 For comparison, Columns 3 and 4 in Table 5 show the corresponding estimates for the specifications without correcting for self-selection. The results are generally consistent with those of the selectively adjusted regressions reported in Columns 1 and 2. The extent of self-selection bias is assessed by comparing the corresponding coefficients in Columns 1 and 3 vs. Columns 2 and 4 of Table 5. The implied elasticity of the firms’ tax payments with respect to each of the independent variables (measured by multiplying the estimated coefficients by their respective means) is used to estimate the differences in the size of the coefficients between the selectivity-adjusted and OLS estimates.22 Our calculations reveal that for LIFO firms the implied elasticity for variables, FIXED and INVTS, is sizably different between selectivity-adjusted and OLS regressions. For instance, a 10% increase in inventory to sales (INVTS) is expected to decrease taxes by 7.64% (−4.603 × 0.166) using selectivity approach and 15.36% (−9.255 × 0.166) decrease using OLS. Large differences in implied elasticity between selectivityadjusted and OLS regressions are also found for FIFO firms. A 10% increase in

254


available tax savings (LATS) will decrease taxes by 13.02% (−0.233 × 5.59) under selectivity approach and by 10.96% (−0.196 × 5.59) using the OLS approach. Also, about 24% difference in implied elasticity is found for the INVTS variable between the selectivity-adjusted and OLS categories. In addition, we also performed the Wald test for the overall differences in the selectivity-adjusted and OLS regressions in each of the LIFO and FIFO categories. Our results show significant differences (p < 0.001) across selectivity-adjusted and OLS estimates.23

Estimates of the Structural Probit and Sensitivity Tests The estimates of structural-probit (not reported) model are based on the results of the reduced-form probit. Thus, using reduced-form probit, we derive the estimated value for predicted tax savings under LIFO and FIFO approaches. These econometrically computed predicted tax savings are introduced as an added variable in the structural probit equations. The results of the estimated structural probit equations based on Eq. (12) show that the sign and significance of the coefficients are largely similar to those for the reduced-form probit except the size variable (LGTASST), which is significant at the 1% level, and the inflation variable (CPRICE) which is negatively and significantly associated with LIFO use. As previously noted, for the reduced-form probit the LGTASST was not significant and the CPRICE variable was significantly positive. The results of the structural probit (not reported) also show the coefficient for the predicted tax savings, PRTXSAV, calculated as the difference between predicted less actual taxes under the two inventory valuation methods for each firm in the sample. For LIFO firms, it is calculated as the difference between predicted FIFO taxes and actual LIFO taxes. For FIFO firms, it is the difference between predicted LIFO taxes and actual FIFO taxes. For identifiability reasons, the structured probit equations exclude the INCVAR variable. The estimation results (not reported) reveal that the predicted tax savings variable is the most significant and a positive coefficient in explaining the use of LIFO, indicating that increases in the tax benefit of adopting LIFO increase the likelihood that firms will choose the LIFO method. Since the results of structural probit may depend on the choice of the variable excluded from the inventory choice equation, we conducted sensitivity tests by re-estimating the structured probit model with a different variable excluded in turn. The results of these tests are reported in Table 6. Each row in the Table represents a separate estimation of the structured model. The excluded variable in each case is listed in the first column. Clearly, all the predicted tax savings


255

Table 6. Sensitivity Test of the Structural Probit Equation. Variable INCVAR INVVAR RELASST INVM CI LGTASST LEV CPRICE

PRTXSAV

t-Value

0.296 0.294 0.295 0.296 0.294 0.286 0.292 0.296

35.764*** 35.427*** 38.850*** 35.885*** 35.917*** 36.243*** 36.363*** 35.780***

Note: Dependent variable equals 1 if the firm is LIFO, 0 if FIFO. Each coefficient reported above represents predicted tax savings computed as the difference between predicted taxes less actual taxes in separate estimates of the probit equation. The variable removed from the structural probit equation is named in the first column. For example, the coefficient on predicted taxes in Table 6 is reported as the first entry labeled INCVAR when INCVAR is removed. ∗∗∗ Significant at the 0.01 level.

in Table 6 are positive and significant at less than the 1% level and qualitatively similar in magnitudes in each case. Thus, the positive effect of the predicted tax savings variable appears to be robust with respect to differences in the specification of the structural probit equations.

Predicted Tax Savings Under Different Regimes The implications of self-selection in an inventory method selection study cannot be easily dismissed. The presence of a significant lambda term in the selectivityadjusted model suggests that the tax savings or tax savings foregone could be potentially large. Thus, the model could then be used to estimate what the tax savings or tax savings foregone would have been had a firm chosen the alternate method. As an illustration, we calculate the differences between the actual tax payments for each LIFO (FIFO) firm and what they would have paid had they been in the other category, with and without adjustment for self-selection. For consistency, we define the tax saving as actual LIFO (FIFO) taxes less predicted FIFO (LIFO) taxes. The calculated differences so obtained are reported in Table 7. The table provides the answer to the question: how much more or less taxes would LIFO (FIFO) firms had paid had it been in the other category? The predicted tax savings or tax savings foregone for LIFO and FIFO firms under the selectivity-adjusted and the OLS methods are presented in Table 7. Columns 1

256


Table 7. Predicted Dollar Tax Savings Under Different Regimes for LIFO and FIFO Firms. Selectivity-Adjusted

Mean Median Std Dev Maximum Minimum InterQuar-tile range No. of obs.

OLS

LIFO

FIFO

LIFO

FIFO

1

2

3

4

−282.20 −297.44 190.35 709.31 −723.67 188.59

12.300 1.242 43.000 449.45 −0.315 5.975

40.563 2.259 128.4 999.12 −18.320 22.589

11.301 4.994 43.023 498.70 −13.591 5.630

1032

5031

1032

5031

Note: Tax savings for the LIFO (FIFO) firms is the difference between the actual LIFO (FIFO) taxes and predicted FIFO (LIFO) taxes. The negative sign indicate tax savings and positive sign indicates tax savings foregone. Selectivity-adjusted values are based on Heckman-Lee (1979, 1978) procedure. OLS estimates are ordinary least squares regression-based estimates. To eliminate the effect of extreme values 18 observations were dropped from LIFO sample and 9 observations from the FIFO sample. Our results remain qualitatively the same with or without these outliers.

and 2 shows selectivity-adjusted tax savings/tax savings foregone while Columns 3 and 4 show the corresponding tax savings/tax savings foregone when OLS estimate is used. Selectivity-adjusted values (Columns 1 and 2) show that for LIFO firms the mean tax savings are $282.2 million (median $297.4 million) and mean tax savings for FIFO firms is $12.3 million (median $1.2 million).24 These results suggest that FIFO firms would have had sizable tax savings had they been LIFO firms. Had the non-random nature of the sample been ignored, the corresponding OLS estimated mean tax savings foregone would have been $40.6 million (median $2.3 million) for LIFO firms and $11.3 million (median $5.0 million) for FIFO firms (Columns 3 and 4). Compared to selectivity approach, the OLS results are consistent for FIFO firms but not so for LIFO firms.25 Compared to the OLS approach, we have greater confidence in the results of the selectivity-based approach because it is econometrically derived and it utilizes a set of explanatory variables in the regression models, which are well defined in the accounting and tax literature. We believe that large selectivity-based tax savings provide a partial explanation of why a number of firms adopted LIFO during the mid-1970s. Finally, the results in Table 7 for FIFO firms are fairly uniform. In each case, the selectivity adjusted or the OLS approach, the FIFO firms would have had tax savings had they used the LIFO inventory method.


257

SUMMARY AND CONCLUSIONS Managerial decision to use an inventory accounting method is based on a number of variables. Some of these variables may be linked to the tax strategy, compensation policy, debt covenants, stock prices, and working capital. Which ever inventory method is used (e.g. LIFO, FIFO), the firm needs to design its managerial accounting system to optimize its choice of the accounting method. For instance, the decision to adopt LIFO or FIFO suggests that a firm has the ability to forecast cash flows under either of these choices. In addition, a firm planning to use LIFO expects to remain profitable and that the investment in inventories is likely to increase due to expected inflation. Thus, management accounting needs to do careful profit planning for business units and segments and be able to examine the implications of the inventory method on firm taxes, managerial compensation, debt covenants, and inventory management. This discussion demonstrates that the choice of the inventory method affects a firm’s management accounting system in a wide variety of ways. Prior research examining a firm’s choice of either LIFO or FIFO as an inventory valuation method has concentrated on two major areas. One line of inquiry has measured the effects of this choice while the other has examined the determinants of the choice. This study extends LIFO-FIFO research by developing a model in which the choice of an inventory valuation method and the effects of this choice are jointly determined. Approaching the subject area in this way makes the self-selection bias arising from firms’ choice of LIFO and FIFO the central issue of our analysis. Self-selection occurs when the observed assignment of firms into these two accounting categories is due to a priori, unobserved decision processes. Ignoring the self-selection of firms and treating the LIFO-FIFO status as exogenous introduces bias into the empirical estimates. In dealing with this issue, we applied the two-stage regression procedure developed by Heckman (1976, 1979) and Lee (1978) to 1973–1981 data, a period during which the incentive to adopt LIFO was most pronounced. We estimated tax equations for LIFO and FIFO firms separately, simultaneously correcting for self-selection bias. The unbiased parameters from the tax functions are then used to create a measure of the tax benefits from adopting LIFO. Self-selection bias arises because sample items have been pre-selected. The researcher, as a result, has no opportunity to select the sample items randomly. In case of the selection of LIFO and FIFO firms, we could not select the sample firms randomly. We used the firms in our sample where the managers had already selected either LIFO or FIFO as the predominant inventory choice method. Hence, the researcher ends up using a non-random sample. The self-selection approach used here attempts to measure whether there is self-selection bias when

258


using pre-selected LIFO and FIFO firms. Knowing the self-selection bias, we are able to estimate the tax savings, which would have been had the firms not been pre-selected and the sample was randomly derived. The estimated Mills ratio (lambda term) based on the two-stage method identifies the extent of self-selection bias and then explicitly computes the best measure of the predicted tax savings in a structured probit estimate. Our analysis yields the following results. First, we find strong evidence that self-selection is present in our sample of LIFO and FIFO firms, consistent with the hypothesis that the managerial decision to choose the observed inventory accounting method is based on a rational cost-benefit calculation. Second, correcting for self-selection leads to the inference that LIFO firms would, on average, pay more taxes as FIFO firms, and FIFO firms could have had tax savings had they been LIFO firms. Our selectivity-adjusted calculation shows that the mean tax savings is $282.2 million for LIFO firms and that FIFO firms would, on the average, pay $12.3 million less in taxes if they were LIFO firms. Without correction for selectivity bias, the mean tax benefit foregone is $40.6 million for LIFO firms and $11.3 million for FIFO firms. Overall, the results suggest that the difference between the LIFO/FIFO tax savings could partly be the function of firm size. LIFO firms in our sample are on the average larger than the FIFO firms. In addition, the difference in tax savings may be related to specific industries. Finally, we believe that the inventory method (LIFO or FIFO) is reflective of the various economic constraints confronting the firm. Hence, the inventory method used by a firm is a rational economic decision. Two weaknesses may be related to our work. Despite controlling for firm size in the accounting choice function, one may argue that our results are affected by the size difference between the LIFO and FIFO firms. The LIFO firms in our sample are on the average larger than the FIFO firms. In addition, the difference in tax savings may also be specific to selected industries. We do not compute tax savings on industry-level because of insufficient data. Future studies could explore this issue further. Future work may also investigate the presence of self-selection bias on other management accounting issues. For instance, the effect of the selection of the depreciation method on managerial performance could be studied. An additional area of work could focus on the effect of the selection of the pooling vs. purchase method of accounting on the post-merger performance of merged firms.

NOTES 1. The LIFO reserve reported by LIFO firms since 1975 is also an as-if number (see Jennings et al., 1996).


259

2. Martin (1992) argues that FIFO method may be a logical tax minimizing strategy for some firms in the event sales and production grow faster than the inflation rate when idle capacity exists, and fixed manufacturing costs is relatively large. 3. Sunder (1976a, b) develops models to estimate the differences between the net present value of tax payments under LIFO vs. FIFO inventory valuation methods. He shows that the expected value of net cash flows depend on the future marginal tax rates, anticipated change in the price of inventories, cost of capital of the firm, pattern of changes in the year-end inventories, and the number of years for which the accounting change will remain effective. Caster and Simon (1985) and Cushing and LeClere (1992) found that tax loss carryforward and taxes are significant factors in the decision to use the LIFO method, respectively. 4. The accounting literature has yet to develop a unified theory explaining managers’ choice of an accounting method. However, an emerging body of accounting literature has advocated the concept of rational choice as a basis of managerial decision-making (see Watts & Zimmerman, 1986, 1990). 5. We recognize that a principal-agent problem can arise when self-interest of managers do not coincide with the interest of firms’ shareholders (Jensen & Meckling, 1976). We do not incorporate the agency issue into the theoretical framework for two reasons: (1) it is another source of self-selection in the data in addition to the self-selection caused by maximizing shareholders’ wealth; and (2) some empirical studies have found that managerial compensation and managerial ownership variables are not significant regressors in explaining inventory valuation choice (Adel-Khalik, 1985; Hunt, 1985). 6. The Heckman-Lee method has also been used in previous accounting studies. For instance: Adel-Khalik (1990a, b) applied the Heckman-Lee model to firms acquiring management advisory services from incumbent auditors vs. other sources, and to the endogenous partitioning of samples into good news and bad news portfolios of quarterly earnings announcements, Shehata (1991) uses the Heckman-Lee model to examine the effect of the Statement of Financial Accounting Standard No. 2 on R&D expenditures, and Hogan (1997) shows that the use of Big 6 vs. non-Big 6 auditor in an initial public offering depends upon a strategy which minimizes the total cost of under-pricing and auditor compensation. 7. The error terms in Eqs (10) and (11) are heteroskedastic and a correction must be made in calculating the correct standard errors of the estimates. In this study, this correction is achieved by using LIMDEP software in implementing the Heckman-Lee model. See Greene (1990) for a description of the correction process. 8. Following Dopuch and Pincus (1988), we also estimate using the “as-if” approach but we do not report the results in this paper. 9. Lindahl et al. (1988) characterize Lee and Hsieh’s (1985) probit model to be comprehensive which includes many of the variables used in previous studies. 10. The rationale for the use of the regressors in the probit function is covered extensively in Lee and Hsieh (1985). 11. Dhaliwal et al. (1992) compute the FIXED variable by including long-term debt in addition to the market value of equity in the denominator. 12. DeAngelo and Masulis (1980) demonstrated that a firm’s effective marginal tax rate on interest deductions is a function of the firm’s non-debt tax shields (e.g. tax loss carryforwards, investment tax credits). 13. Trezevant (1996) investigates the association of debt tax shield to changes in a noninvestment tax shield (cost of good sold) in the post LIFO adoption period.

260


14. A similar argument is made by Mackie-Mason (1990) when considering tax carryforward and debt financing. He argues that tax carryforward have a large effect on the expected marginal tax rate on interest expenses since each dollar of tax carryforward is likely to reduce a dollar of interest deduction. 15. We included firms with as few as one LIFO (FIFO) year, as long it has not been a FIFO (LIFO) firm in any other years in this period. There are several reasons for our choice: (1) specifying a minimum number of years for inclusion in the sample is arbitrary and introduces potential bias in estimates; (2) the time element is unimportant in the pooled cross-sectional analysis as each year of the data is treated as an independent observation; and (3) since firms obtain the benefits from choosing LIFO (FIFO) in the year of adoption, firms using LIFO (FIFO) for only one year will still provide us with as much relevant information on the inventory method choice as those using LIFO or FIFO for more than one year. 16. Aside from differences in inventory methods and substantial economic differences, the differences in t-values between the two groups may also be a function of possible violation of the assumptions of the t-test. 17. Probit results with industry dummies included are similar to the results reported in Table 4. In addition, the results indicate that the regression coefficients are positive and significant for the textile, chemicals, petroleum and coal, rubber, leather, primary metal and wholesale industries and are negative and significant for electronic and business services industries. 18. We also used the logarithmic value of income taxes paid (income taxes-total [Compustat item 16] minus deferred income taxes [Compustat item 126] as the dependent variable). The results are essentially similar to those reported in Table 5. In addition, we examined the possibility of using effective corporate tax rate as the dependent variable but decided against its use for lack of consistency in the definition of the effective tax rate measure in the literature (Omer et al., 1990). 19. The lambda term is based on the predicted value of the error term derived from the reduced form probit. Hence, the selectivity adjustment reported in Table 5 is based on the probit reported in Table 4. 20. The t-statistics shown in Table 5 are based on the correct asymptotic variancecovariance matrix derived in Heckman (1979). Also, the OLS regression reveals that there is no multi-collinearity among regressors (VIF values are no more than 2.0). 21. The results obtained using a different set of regressors for the tax function, are qualitatively similar. For instance, we developed a model containing the following variables: NDTS, TLCF, FIXED, INVVAR, RELASST, and INCVAR. The corresponding coefficients of the lambda terms are −1.815 (t = −5.292) for LIFO firms and −7.071 (t = −22.462) for FIFO firms. The sign and significance of other regressors in the tax equations are generally in the expected direction. Similar results are also found when we drop the INCVAR variable from the above tax function. 22. Differentiating the dependent variable with respect to the variable, FIXED, for example, yields (␦T/T␦F) where T is dollars of tax payments and F represents the FIXED variable. Multiplying by the mean value for FIXED gives us (F␦T/T␦F), the elasticity of taxes associated with fixed assets. Note that for the LATS regressor, both the dependent and independent variables are already in natural logarithmic value. Thus, the estimated coefficient is itself the elasticity figure. 23. We tested the normality assumption for each of the variables prior to adjusting for selectivity by either using the Kolmogrov-Smirnov or the Shapiro-Wilks statistic. The


261

results show that the observed distribution is not significantly different from the normal distribution. In addition, following Pagan and Vella (1989), we perform a moment-based test for normality in selectivity models. In this test, the predictions from the probit model are squared and cubed, weighted by the Mills ratio. Using the two-stage least squares results, the null hypothesis that squared and cubed terms are zero cannot be rejected. 24. The difference between actual LIFO (FIFO) taxes and predicted FIFO (LIFO) taxes for LIFO (FIFO) firms is either tax savings or tax savings foregone. The negative difference suggests tax savings and the positive sign difference indicates tax savings foregone. 25. We also computed tax savings/tax forgone using the “as-if” method described in Lindahl (1982), Morse and Richardson (1983), Dopuch and Pincus (1988), and Pincus and Wasley (1996). Our results (not reported show that the average “as-if” tax savings for LIFO firms is $10.9 million. The FIFO firms’ average tax savings foregone under the “as-if” approach is $60.0 million. Overall, our results show that the selectivity-adjusted approach has the highest tax savings for LIFO firms. The selectivity approach takes into consideration the joint decision of the inventory method choice and the tax effect of the decision.

ACKNOWLEDGMENTS An earlier version of this paper was presented at the 1998 Annual Meeting of the American Accounting Association and at the 2003 Management Accounting Research Conference. We are thankful to C. Brown, D. Booth, J. Ohde, M. Myring, S. Mastunaga, M. Pearson, M. Pincus, M. Qi, R. Rudesal, and S. Sunder for comments and suggestions and to D. Lewis and J. Winchell for computer programming assistance. The first author is thankful to the Kent State University’s Research Council for providing partial financial support for this project. A previous version of this paper is also available on ssrn.com.

REFERENCES Adel-Khalik, A. R. (1985). The effect of LIFO-switching and firm ownership on executives pay. Journal of Accounting Research (Autumn), 427–447. Adel-Khalik, A. R. (1990a). The jointness of audit fees and demand for MAS: A self-selection analysis. Contemporary Accounting Research (Spring), 295–322. Adel-Khalik, A. R. (1990b). Specification problems with information content of earning: Revisions and rationality of expectations, and self-selection bias. Contemporary Accounting Research (Fall), 142–172. Auerbach, A. J., & Porterba, J. M. (1987). Tax-loss carryforwards and corporate tax incentives. In: M. Feldstein (Ed.), The Effects of Taxation and Capital Accumulation (pp. 305–338). Chicago: University of Chicago Press. Ball, R. (1972). Changes in accounting techniques and stock prices. Journal of Accounting Research (Suppl.), 1–38.

262


Biddle, G. C. (1980). Accounting methods and management decisions: The case of inventory costing and inventory policy. Journal of Accounting Research (Suppl.), 235–280. Biddle, G. C., & Lindahl, R. W. (1982). Stock price reactions to LIFO adoptions: The association between excess returns and LIFO tax savings. Journal of Accounting Research (Autumn), 551–588. Biddle, G. C., & Martin, R. K. (1985). Inflation, taxes, and optimal inventory policy. Journal of Accounting Research (Spring), 57–83. Bowen, R., Ducharme, L., & Shores, D. (1995). Stakeholders’ implied claims and accounting method choice. Journal of Accounting and Economics (December), 255–295. Cushing, B. E., & LeClere, M. J. (1992). Evidence on the determinants of inventory accounting policy choice. Accounting Review (April), 355–366. DeAngelo, H., & Masulis, R. (1980). Optimal capital structure under corporate and personal taxation. Journal of Financial Economics, 8, 3–29. Dhaliwal, D., Trezevant, R., & Wang, S. (1992). Taxes, investment-related tax shields and capital structure. Journal of American Taxation Association, 1–21. Dopuch, N., & Pincus, M. (1988). Evidence on the choice of inventory accounting methods: LIFO versus FIFO. Journal of Accounting Research (Spring), 28–59. Duncan, G. (1986). Continuous/discrete econometric models with unspecified error distribution. Journal of Econometrics, 32, 139–153. Greene, W. H. (1990). Econometric analysis. Englewood, NJ: MacMillan. Hagerman, R. L., & Zmijewski, M. R. (1979). Some economic determinants of accounting policy choice. Journal of Accounting and Economics (January), 141–161. Hand, J. R. M. (1993). Resolving LIFO uncertainty: A theoretical and empirical reexamination of 1974–1975 LIFO adoptions and nonadoptions. Journal of Accounting Research (Spring), 21–49. Heckman, J. J. (1976). The common structure of statistical models of truncation, sample selection, and limited dependent variables and a simple estimation for such models. Annals of Social and Economic Measurement (Fall), 475–492. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica (January), 153–161. Hogan, C. E. (1997). Costs and benefits of audit quality in the IPO market: A self-selection analysis. Accounting Review (January), 67–86. Hunt, H. G., III (1985). Potential determinants of corporate inventory decisions. Journal of Accounting Research (Autumn), 448–467. Jennings, R., Mest, D., & Thompson, R., II. (1992). Investor reaction to disclosures of 1974–1975 LIFO adoption decision. Accounting Review (April), 337–354. Jennings, R., Simko, P., & Thompson, R., II (1996). Does LIFO inventory accounting improve the income statement at the expense of the balance sheet? Journal of Accounting Research (Spring), 85–109. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics (September), 305–360. Kang, S. K. (1993). A conceptual framework for the stock price effects of LIFO tax benefits. Journal of Accounting Research (Spring), 50–61. Lee, L. F. (1978). Unionism and wage rates: A simultaneous equations model with qualitative and limited dependent variables. International Economic Review (June), 415–433. Lee, J. C., & Hsieh, D. (1985). Choice of inventory accounting methods: Comparative analyses of alternative hypotheses. Journal of Accounting Research (Autumn), 468–485. Lindahl, F., Emby, C., & Ashton, R. (1988). Empirical research on LIFO: A review and analysis. Journal of Accounting Literature, 7, 310–333.


263

Mackie-Mason, J. K. (1990). Do taxes affect corporate financing decisions? Journal of Finance (December), 1471–1493. Maddala, G. S. (1991). A perspective on the use of limited-dependent and qualitative variables models in accounting research. Accounting Review (October), 788–807. Manski, C. (1989). Anatomy of the selection problem. Journal of Human Resource, 24, 343–360. Manski, C. (1990). Nonparametric bounds on treatment effects. American Economic Review, 24, 319–323. Martin, J. R. (1992). How the effect of company growth can reverse the LIFO/FIFO decision: A possible explanation for why many firms continue to use LIFO. Advances in Management Accounting (pp. 207–232). Greenwich, CT: JAI Press. Ming, X., & Vella, F. (1994). Semi-parametric estimation via synthetic fixed effects. Working Paper, Rice University (November). Morse, D., & Richardson, G. (1983). The LIFO/FIFO decision. Journal of Accounting Research (Spring), 106–127. Omer, T. C., Molloy, K. H., & Ziebart, D. A. (1990). Measurement of effective corporate tax rates using financial statement information. Journal of American Taxation Association (July), 57–72. Pagan, A., & Vella, F. (1989). Diagnostic tests for models based on individual data: A survey. Journal of Applied Econometrics, 4, S29–S59. Pincus, M. A., & Wasley, C. (1996). Stock price behavior associated with post-1974–1975 LIFO adoptions announced at alternative disclosure time. Journal of Accounting, Auditing and Finance, 535–564. Ricks, W. E. (1982). The market’s response to the 1974 LIFO adoptions. Journal of Accounting Research (Autumn), 367–387. Scholes, M. S., & Wolfson, M. A. (1992). Taxes and business strategy: A planning approach. Englewood, NJ: Prentice-Hall. Shehata, M. (1991). Self-selection bias and the economic consequences of accounting regulation: An application of two-stage switching regression to SFAS No. 2. Accounting Review (October), 768–787. Smith, C. W., & Warner, J. B. (1979). On financial contracting: An analysis of bond covenants. Journal of Financial Economics (June), 117–161. Sunder, S. (1973). Relationship between accounting changes and stock prices: Problems of measurement and some empirical evidence. Journal of Accounting Research (Suppl.), 1–45. Sunder, S. (1976). A note on estimating the economic impact of the LIFO method of inventory valuation. Journal of Accounting Research (April), 287–291. Sunder, S. (1976). Optimal choice between FIFO and LIFO. Journal of Accounting Research (Autumn), 277–300. Titman, S., & Wessels, R. (1988). The determinants of capital structure choice. Journal of Finance (March), 1–19. Trezevant, R. (1992). Debt financing and tax status: Tests of the substitution effect and tax exhaustion hypothesis using firms’ response to the Economic Recovery Act of 1981. Journal of Finance (September), 1557–1568. Trezevant, R. (1996). LIFO adoption and tax shield and substitution effect. Journal of American Taxation Association (Suppl.), 18–31. Trost, R. P. (1981). Interpretation of error covariances with nonrandom data: An empirical illustration of returns to college education. Atlantic Economic Journal (September), 85–90. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838.

CORPORATE ACQUISITION DECISIONS UNDER DIFFERENT STRATEGIC MOTIVATIONS Kwang-Hyun Chung ABSTRACT Acquisition is one of key corporate strategic decisions for firms’ growth and competitive advantage. Firms: (1) diversify through acquisition to balance cash flows and spread the business risks; and (2) eliminate their competitors through acquisition by acquiring new technology, new operating capabilities, process innovations, specialized managerial expertise, and market position. Thus, firms acquire either unrelated or related business based on their strategic motivations, such as diversifying their business lines or improving market power in the same business line. These different motivations may be related to their assessment of market growth, firms’ competitive position, and top management’s compensation. Thus, it is hypothesized that firms’ acquisition decisions may be related to their industry growth potential, post-acquisition firm growth, market share change, and CEO’s compensation composition between cash and equity. In addition, for the two alternative acquisition accounting methods allowed until recently, a test is made if the type of acquisition is related to the choice of accounting methods. This study classifies firms’ acquisitions as related or unrelated, based on the standard industrial classification (SIC) codes for both acquiring and target firms. The empirical tests are, first, based on all the acquisition cases regardless of


265

266

KWANG-HYUN CHUNG

the firm membership, and then, deal with the firms acquiring only related businesses or unrelated businesses exclusively. The type of acquisitions was more likely related to industry growth opportunities, indicating that the unrelated acquisition cases are more likely to be followed by higher industry growth rate than the related acquisition cases. While there were a substantially larger number of acquisition cases using the purchase method, the related acquisition cases used the pooling-of-interest method more frequently than in the unrelated acquisition cases. The firm-level analysis shows that the type of acquisition decisions was still related to acquiring firms’ industry growth rate. However, the post-acquisition performance measures, using firm’s growth and change in market share, could support prior studies in that the exclusive-related acquisitions helped firms grow more and get more market share than the exclusive-unrelated acquisitions. CEO’s compensation composition ratio was not related to the types of acquisition.

1. INTRODUCTION For the last three decades, mergers and acquisitions have been important corporate strategies involving corporate and business development as the capital markets rapidly expanded. Increased uncertainty about the economy has made it difficult for firms to resort only to internal growth strategies. Firms diversified through acquisition to balance cash flows and spread the business risks. They also tried to eliminate their competitors through acquisition by acquiring new technology, new operating capabilities, process innovations, specialized managerial expertise, and market position. Mergers and acquisitions (M&A), as an increasingly important part of corporate strategy, enable firms to grow at a considerable pace. Also, firms can quickly restructure themselves through M&A when they find it necessary to reposition. M&A provides firms with the creation and identification of competitive advantage. Porter (1987) identified the following four concepts of corporate strategy from the successful diversification records of 33 U.S. companies from 1950 through 1986: portfolio management, restructuring, transferring skills and sharing skills. Those concepts of corporate strategy explain the recent corporate takeovers as either related diversification or conglomeration. Also, Porter (1996) stressed that the essence of corporate strategy is choosing a unique and valuable position rooted in systems of activities that are much more difficult to match, while the operational techniques, such as, TQM, benchmarking, and re-engineering are easy to imitate. Thus, many companies take over the target firms with such positions in the context of operational effectiveness competition, instead of repositioning themselves based on products, services, and customers’ needs. Firms’ sustainable

Corporate Acquisition Decisions under Different Strategic Motivations

267

competitive advantage through buying out rivals could be another key motivation for corporate takeovers. According to Young (1989), it is necessary to identify how M&A will result in added value to the group, how quickly these benefits can be obtained and how the overall risk profile of the group will be affected when considering M&A as part of an overall corporate and business development strategy. The study suggests two key criteria for the successful corporate takeovers: the level of business affinity and the business attractiveness (i.e. market size, growth, profitability etc.) of the target company. In reality, firms acquire either unrelated or related business based on their strategic motivations: diversifying their business lines or improving market power in the same business line. These different M&A decisions are hypothesized to be influenced by the firms’ strategic motivations explained in the prior literature. This paper attempts to identify the different strategic motivations for acquisition activities of firms. More specifically, this study tests whether firms’ acquisition decisions are related to their industry growth potential, post-acquisition firm growth, market share change, and CEO’s stock compensation composition. It also tests if the type of acquisition is related to the choice of accounting method. In light of what prior studies (e.g. Scanlon, Trifts & Pettway, 1989) have done, this study classifies firms’ acquisitions as related or unrelated, based on the standard industrial classification (SIC) codes for both acquiring and target firms. This study utilizes Securities Data Company (SDC)’s Worldwide Mergers & Acquisition Database from 1996 to 1999. This database includes all transactions involving at least 5% of the ownership of a company where a transaction was valued at $1 million or more, and each firm may have many acquisition cases over the test period. Thus, the empirical testing is based on each case as well as on each firm. The rest of this paper is structured as follows: Section 2 explains how the hypotheses are developed with regards to the accounting methods, the motivations of firm’s acquisition decision including industry growth opportunities and CEO’s compensation as well as the consequences of the acquisition decision in terms of improved operating performance and increased market presence. The data used and test variables are explained in Section 3. Section 4 summarizes empirical results in both all-cases analysis and firm-level analysis. Finally, the concluding remarks are provided in Section 5.

2. HYPOTHESES DEVELOPMENT 2.1. Acquisition Trends and Industry Growth Opportunities If a firm’s motivation for acquisition is to balance cash flows and spread the business risks, it is more likely to acquire a different line of business. Also, when the industry

268

KWANG-HYUN CHUNG

is faced with limited growth potential, firms are less likely to enter new markets because it would be riskier for the management to manage new unrelated business. This hypothesis is supported by product-market theory, which suggests that risk increases as a firm moves into a new unfamiliar area. The conglomerate type of diversification was often used to fuel tremendous corporate growth as firms purchased many unrelated businesses during the 1960s, regardless of what good or service they sold. In the 1970s, managers began to emphasize diversification to balance cash flows individual businesses produced. Acquisition was regarded as a diversification to balance between businesses that produced excess cash flows and those that needed additional cash flows. It was known as portfolio management to reduce risk. During the 1980s there was a broad-based effort to restructure firms, scratching out unrelated businesses and focusing on a narrower range of operations. Expansion through acquisition was often limited to the vertical integration. In the 1990s, there was an increasing number of diversification into related businesses that focused on building dynamic capabilities as an enduring source of competitive advantage and growth. If firms’ motivation for M&A is the use of core competence in the acquired business and/or to increase the market strength, the new business should have enough similarity to existing business and this benefit can be augmented in the acquisition of the same line of business. The use of core competence and/or the increase of market strength should produce the competitive advantage that consequently increases market share. Thus, the following hypothesis is made in an alternate form with regards to each acquisition case: H1 . Firms’ acquisition type is related to their industry growth potential. More specifically, under limited growth potential, firms are more likely to seek for competitive advantage and growth through more related acquisition while firms under its higher growth potential in their industry are more likely to enter new markets to balance cash flows, which leads to unrelated acquisition.

2.2. Acquisition Decisions and Firms’ Post-acquisition Growth and Market Power The hypothesis above can be rephrased using firm’s own growth rate, instead of industry growth rate for the firm-level analysis. However, this study uses the ex-post growth rate after the acquisition, and thus, the post-acquisition growth rate would be interpreted as a firm’s post-acquisition performance instead of its own assessment of the growth potential. In fact, many firms acquired both related and unrelated businesses in each year. Therefore, this study selects two distinct


269

cases such as those who acquired exclusively the same business lines in each test period (exclusive-related acquisition) and those who acquired only the different line of business in each test period (exclusive-unrelated acquisition).1 Prior M&A studies suggested that changes in the opportunity to share resources and activities among business units have contributed to post-acquisition performance.2 Most studies find improved performance in the 1980s acquisition, compared to the earlier conglomerate acquisition wave of the 1960s, because of increased opportunities to share resources or activities in the acquired firm (i.e. more operating synergy effect).3 Diversifying character of the unrelated acquisition could be the reason for poorer performance,4 especially in the short period, compared to the related acquisition where we find a high opportunity for shared activities or resources. Thus, the following hypothesis is formed in an alternate form with regards to a firm’s post-acquisition growth rate: H2 . Firms’ acquisition decision, unrelated or related, would lead to the different post-acquisition growth. More specifically, firms with exclusive related acquisition cases are more likely to have higher growth rate after the acquisition than firms with exclusive unrelated acquisition cases. In addition to firms’ post-acquisition growth, I hypothesize that firm’s acquisition type can lead to the competitive position in its industry using the market share because one of the crucial motivations for the acquisition may be to increase market strength. Especially in a more competitive market, firms will be more likely to acquire the other competing firms to increase the market power, compared to the less competing market environment. Thus, the firms with exclusive related acquisition cases are more likely to be motivated by increasing market power than the firms with exclusive-unrelated acquisition cases. Thus, the following hypothesis in an alternate form is made with regards to the firm’s change in market share: H3 . Firms that acquired the same line of businesses are more likely to expect higher increase in market share in the industry because they get core competences and market power than the firms that acquired only the different lines of businesses.

2.3. Different Accounting Methods in Mergers and Acquisitions Before June 2002, there were two generally accepted methods of accounting for business combinations. One is referred to as the purchase method and the other is known as the pooling of interests method. These two methods are not alternative

270

KWANG-HYUN CHUNG

ways to account for the same business combination. The actual situation and the attributes of the business combination determine which of the two methods is applicable. The purchase method would be applicable in a situation where one company is buying out another. The pooling of interest method would be the case where the shareholders of one company surrender their stock for the stock of another of the combining companies. There has been a combining of ownership interests, which is not regarded as purchase transaction. Thus, firms can avoid recognizing goodwill. That’s why the pooling of interest method required certain criteria regarding the nature of consideration given and the circumstances of the exchange. However, the problem in the method is to determine the equivalent number of common shares acquired from the combining companies. If two companies in the business combination are similar each other, it is perhaps easier to determine the shares to be exchanged, compared to heterogeneous combinations.5 Thus, the following hypothesis in an alternate form is stated in terms of two different accounting methods in the acquisition. H4 . Firms with exclusive-related acquisition cases are more likely to use the pooling of interest method than those with exclusive-unrelated acquisition cases.

2.4. CEOs Compensation Plan and Firm’s Acquisition Decisions Previous literature regarding management compensation (e.g. Narayanan, 1996) suggested that the proportion of equity compensation is higher when the firm’s growth opportunities are greater. As future growth opportunities increase relative to assets in place, the value of the stock per unit of managerial ability increases. Therefore, the management receives fewer shares for the same perceived ability. To compensate, the proportion of the stock component must increase. If we know that the top management’s equity compensation ratio is related to the firm’s growth opportunities, then the ratio may be different between two different types of acquisitions where we see the different assessment of future growth and different post-acquisition firm growth and changed market power. Thus, the different type of acquisition decisions is hypothesized with the management’s compensation proportion as follows: H5 . The CEOs’ equity compensation ratio is different between the exclusiveunrelated acquisition and the exclusive-related acquisition.


271

3. DATA AND TEST VARIABLES From 1996 to 1999, there were 9,058 mergers and acquisition cases identified in the Securities Data Company’s (SDC) 2001 Worldwide Acquisitions and Mergers Database which includes all transactions involving at least 5% of the ownership of a company where a transaction was valued at 1 million or more. The SIC (standard industry classification) codes of the acquirer and the target were used to ascertain degree of relatedness. There are 5,142 cases where the acquiring and acquired firms had the same first two digit in their SIC codes, identified as related acquisitions, and there are 3,916 cases identified as unrelated acquisitions. Total sample cases are classified by the acquirers’ first two digit SIC codes and acquisition years from 1996 to 1999 in Table 1. It shows that year 1998 was the most active in the mergers and acquisition activities. The all-cases analysis compares two groups’ ex-post Table 1. Sample Distribution by Industry for Related Acquisition Cases and Unrelated Acquisition Cases (All Cases). Two-Digit SIC 10 13 14 15 16 17 18 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

Related Acquisition

Unrelated Acquisition

Total Cases

1996

1997

1998

1999

20 180 17 22 2 2 0 199 5 24 33 20 15 68 152 275 10 20 6 23 57 45 236 270 129

4 37 9 19 8 8 3 61 3 22 21 42 9 37 95 217 60 25 4 59 65 73 454 354 185

24 217 26 41 10 10 3 260 8 46 54 62 24 105 247 492 70 45 10 82 122 118 690 624 314

6 85 1 11 2 2 3 54 3 17 10 8 4 36 55 126 17 7 3 13 31 31 164 107 64

8 57 7 5 2 3 0 61 1 11 18 13 9 20 70 129 25 12 3 27 35 33 169 135 83

5 38 10 15 3 5 0 67 1 8 14 28 5 27 64 136 16 15 3 19 32 38 177 166 78

5 37 8 10 3 0 0 78 3 10 12 13 6 22 58 101 12 11 1 23 24 16 180 216 89

272

KWANG-HYUN CHUNG

Table 1. (Continued ) Two-Digit SIC

Related Acquisition

Unrelated Acquisition

Total Cases

1996

1997

1998

1999

38 39 40 41 42 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 67 70 72 73 75 76 78 79 80 81 82 86 87

202 15 13 7 11 5 13 1 18 197 177 128 74 2 31 27 12 23 12 57 82 466 29 35 151 31 20 33 30 1010 8 2 12 35 228 1 5 0 109

187 20 1 2 6 7 6 2 11 95 155 165 137 7 33 10 2 5 3 3 73 196 22 55 141 18 115 7 26 320 0 5 11 11 48 4 3 11 119

389 35 14 9 17 12 19 3 29 292 332 293 211 9 64 37 14 28 15 60 155 662 51 90 292 49 135 40 56 1330 8 7 23 46 276 5 8 11 228

122 1 8 5 3 6 1 0 8 46 57 61 68 2 11 14 1 5 4 18 43 179 10 17 74 11 33 11 16 250 0 1 12 9 110 2 2 5 44

100 8 1 2 3 2 5 1 3 59 80 87 55 3 19 8 0 8 3 14 43 159 9 24 81 12 33 10 8 323 5 3 6 14 75 0 1 6 60

92 12 3 1 6 1 7 1 10 78 102 90 45 1 20 8 9 9 5 11 30 196 16 24 81 17 38 13 21 381 2 3 3 12 62 1 3 0 68

75 14 2 1 5 3 6 1 8 109 93 55 43 3 14 7 4 6 3 17 39 128 16 25 56 9 31 6 11 376 1 0 2 11 29 2 2 0 56

Total Related cases Unrelated cases

5142

3916

9058 5142 3916

2130 1239 891

2269 1285 984

2452 1412 1040

2207 1206 1001

Two-Digit SIC

1997

1998

1999

Total Firms

Related

Unrelated

Related

Unrelated

Related

Unrelated

Related

Unrelated

Related

Unrelated

3 29 1 1 1 0 0 13 1 3 6 2 0 10 10 31 1 1 2 4 7 6 13 15 11

1 4 0 2 1 0 2 1 0 4 2 3 1 6 5 16 8 2 1 4 10 4 18 24 8

5 22 1 3 0 0 0 20 0 5 4 1 3 8 9 26 1 4 0 4 6 4 18 32 12

0 3 2 2 0 1 0 1 1 1 5 1 1 3 9 22 7 4 1 3 9 9 28 28 15

3 17 0 4 0 0 0 16 1 4 6 1 2 7 11 33 2 3 2 2 11 4 17 25 13

0 5 2 1 1 1 0 2 0 0 2 1 1 7 5 16 8 4 1 3 4 10 27 23 18

2 20 1 1 0 0 0 15 2 3 5 2 2 3 8 25 0 1 1 1 8 3 18 25 6

0 1 1 2 2 0 0 4 1 2 2 4 0 5 6 7 5 3 0 4 6 4 28 22 14

13 88 3 9 1 0 0 64 4 15 21 6 7 28 38 115 4 9 5 11 32 17 66 97 42

1 13 5 7 4 2 2 8 2 7 11 9 3 21 25 61 28 13 3 14 29 27 101 97 55

273

10 13 14 15 16 17 18 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

1996


Table 2. Sample Distribution by Industry for Exclusive Related Acquisition Firms and Exclusive Unrelated Acquisition Firms (Year/Total) (Firm-Level Cases).

274

Table 2. (Continued ) Two-Digit SIC

1997

1998

1999

Total Firms

Related

Unrelated

Related

Unrelated

Related

Unrelated

Related

Unrelated

Related

23 0 7 0 1 2 1 0 1 17 12 8 10 0 4 5 0 4 1 11 7 46 4 2 17 1

25 1 0 0 1 3 0 0 0 3 11 11 9 2 5 1 1 0 0 0 3 13 2 3 9 2

19 4 1 1 2 1 3 0 1 17 20 10 7 0 6 2 4 1 0 9 5 42 3 1 19 3

16 2 0 0 1 1 1 1 0 9 10 11 9 2 5 0 0 3 1 0 5 8 2 8 17 2

16 2 1 1 2 1 3 0 2 14 24 5 4 1 5 3 2 5 3 7 5 30 5 4 17 0

14 3 1 0 1 0 2 1 1 3 10 14 7 0 6 1 0 0 0 1 2 12 1 5 11 1

20 3 1 0 3 1 2 1 1 12 19 4 3 1 4 2 4 4 3 9 4 23 2 4 14 4

13 4 0 1 2 2 1 0 0 5 15 8 8 2 7 1 0 1 0 1 3 12 3 2 9 0

78 9 10 2 8 5 9 1 5 60 75 27 24 2 19 12 10 14 7 36 21 141 14 11 67 8

Unrelated 68 10 1 1 5 6 4 2 1 20 46 44 33 6 23 3 1 4 1 2 13 45 8 18 46 5

KWANG-HYUN CHUNG

38 39 40 41 42 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

1996

Total

1 2 2 47 0 0 3 2 23 0 1 0 6

7 1 2 17 0 1 1 1 3 1 1 1 7

3 2 5 59 2 0 2 3 17 0 1 0 8

11 2 1 7 0 0 2 2 3 0 0 1 5

1 7 3 65 2 0 1 3 21 0 0 0 7

7 0 2 16 0 0 1 1 5 1 0 0 10

0 2 5 63 1 1 0 3 12 1 2 0 6

10 0 1 10 0 0 1 3 3 1 0 0 4

5 13 15 234 5 1 6 11 73 1 4 0 27

35 3 6 50 0 1 5 7 14 3 1 2 26

442

275

471

304

456

282

396

256

1765

1117


67 70 72 73 75 76 78 79 80 81 82 86 87

275

276

KWANG-HYUN CHUNG

industry growth rates at least one year after the acquisition, which is the proxy for the industry growth potential. For further firm-level analysis, I sorted 9,058 cases by the acquiring firms, and identified two distinct firm groups to test the hypotheses developed in the previous section. The first group is where the firms acquired exclusively the related businesses (i.e. firms in the industry of the same two-digit SIC codes), and the other comparison group is where the firms acquired exclusively the unrelated businesses in each test period. Table 2 summarizes the 1765 exclusive related-acquisition firms and the 1117 exclusive unrelated-acquisition firms by year and the twodigit SIC codes. Every year saw more related acquisition firms than the unrelated acquisition firms. This firm-level analysis compares such test variables as industry growth rate as industry growth opportunity, firm’s growth rate as post-acquisition operating performance, firm’s change in market share as the acquisition motivation as increasing market power, the accounting method in mergers and acquisition, and the CEO’s equity compensation ratio as the proxy for firm’s growth potential. Both all-cases analysis and firm-level analysis have two-group parametric difference test (univariate t-test) between the related acquisition decision and the unrelated acquisition decision, and logistic regression using the test variables as independent variables. The all-cases analysis uses the logistic regression using the industry growth rates and the accounting method for each acquisition as the independent variables. The industry growth rate was measured one-year after from the acquisition to the fiscal year 2001, where the data are currently available in the Compustat tape, but the logistic regression includes the growth rate at least over the two years. The industry growth rate was calculated as the annual change in average net sales of the firms that belonged to the two-digit SIC codes. The firm’s growth rate and change in market share was also measured one-year after from the acquisition to the fiscal year 2001, and the logistic regression models includes them at least over two-year period. The CEO’s compensation ratio is measured from the Standard & Poor’s ExecuComp by dividing stocks granted and Black-Scholes value of options granted by his/her total annual compensation. The higher the variable is, the more dependent on equities their compensation is. The accounting method is a dichotomous variable where the purchase method is given 0, and the pooling-of-interest method is 1.

4. EMPIRICAL RESULTS 4.1. All-Cases Analysis A total 9,058 M&A cases were analyzed each year, and they are divided into related acquisition cases and unrelated acquisition cases. Table 3 shows the

1996 Related Acquistion Cases

Unrelated Acquistion Cases

1239

891

Observations

1997 t-Stat.

Related Acquistion Cases


1285

984

1998 t-Stat.



1412

1040

1999 t-Stat.



1206

1001

t-Stat.

Industry growth rate 5-years 4-years 3-years 2-years 1-year

0.5737 0.58 0.3623 0.237 0.1245

0.6654 0.6824 0.4072 0.2461 0.113

−3.41a −4.77a −2.81a −0.85 0.38

0.4085 0.4053 0.2158 0.0933

0.4772 0.4975 0.2644 0.1186

−3.51a −6.42a −4.55a −3.89a

0.2943 0.2843 0.1138

0.3257 0.3386 0.1359

−2.02b −5.21a −3.47a

0.1581 0.1546

0.1536 0.1753

0.46 −3.39a

Accounting method

0.0826

0.0619

1.55c

0.1121

0.0701

3.45a

0.1261

0.0702

4.53a

0.1119

0.0669

3.66a

Remarks: For year 1996 5-years: 4-years: 3-years: 2-years: 1-year:

year 1996–2001 year 1996–2000 year 1996–1999 year 1996–1998 year 1996–1997

For year 1997 4-years: 3-years: 2-years: 1-year:

year 1997–2001 year 1997–2000 year 1997–1999 year 1997–1998

For year 1998 3-years: year 1998–2001 2-years: year 1998–2000 1-year: year 1998–1999


Table 3. Comparison Between Related Acquisition Cases and Unrelated Acquisition Cases (All Cases Analysis). Year

For year 1999 2 years: year 1999–2001 1-year: year 1999–2000 a Significance c Significance

at the 1% level (two-tailed test). at the 5% level (two-tailed test). at the 10% level (two-tailed test).

277

b Significance

278

KWANG-HYUN CHUNG

comparison between the two groups from year 1996 to year 1999. Except in year 1999, the growth rates of the industry to which each acquisition case belongs show the significant difference between two groups in a hypothesized direction. The growth rates measured up to 2001 were even smaller in some industries than those measured up to 2000. This may contribute to the weaker result in the industry growth rates till year 2001, compared to those till year 2000. This study can’t expand the industry growth rate beyond 2001 for data unavailability. Also, the industry growth rate is ex-post instead of ex-ante to surrogate the industry growth potential. The accounting method supported the alternative hypothesis, indicating that while most acquisitions are accounted for by the purchase method, firms are more likely to adopt the pooling-of-interest method in the related acquisition cases, compared to the unrelated acquisition cases. The multivariate logistic regression results are provided by years in Table 4. In year 1996, all industry growth rates except from 1996 to 1998 are statistically significant in differentiating related and unrelated acquisition cases in a hypothesized direction. As consistent with the univariate comparison, the industry growth rates in both years 1997 and 1998 are all significant in explaining the firm’s acquisition decision. The unrelated acquisition decisions are more often made when the industry growth opportunities are foreseen. The accounting method is statistically significant all years.

4.2. Firm-Level Analysis Table 5 provides the comparison of five test variables between exclusive-related acquisition firms and exclusive-unrelated acquisition firms for the test periods. The growth rates of the industry to which the firms belong are in general consistent with the all-cases analysis in the previous section. In other words, the firms in the higher industry growth potential are more likely to acquire in the different line of business than those in the lower industry growth potential. However, the level of significance gets lower, compared to the all-cases analysis. Especially when the growth potential is measured after 1998, the level of difference gets insignificant. Firm’s growth rate as the post-acquisition performance is statistically significant only in year 1997. Also, firm’s change in market share after its acquisition is significantly different between two groups in years 1997 and 1998. The use of different accounting methods in each acquisition was evident after 1996. The firms in the similar business line fervently tried to use the pooling-of-interest method when the accounting rulemakers started to discuss the abolishment of the method in the late 1990s. CEO’s equity compensation rate is different between two groups in an expected direction, but none of the differences in the test periods is statistically significant.

1996

1997

1998

Observations

2122

2263

2452

Likelihood ratio Intercept Industry growth 5-years

14.945 0.4595 (51.07a )

25.937 0.577 (61.8a )

11.47 0.44 (44.87a )

−0.4204 (21.76a )

For year 1997 4-years: year 1997–2001 3-years: year 1997–2000 2-years: year 1997–1999 For year 1998

32.976 0.403 (44.29a )

−0.8002 (39.36a ) −0.1514 (0.7)

2-years

Remarks: For year 1996 5-years: year 1996–2001 4-years: year 1996–2000 3-years: year 1996–1999 2-years: year 1996–1998

54.109 0.5818 (64.61a )

24.579 0.3065 (31.37a )

2207 46.561 0.5013 (54.9a )

14.003 0.1185 (4.77b )

−0.8254 (24.36a )

−0.116 (0.36)

0.6261 (18.34a ) 54.5

0.567 (13.19a ) 48.8

−0.323 (12.12a ) −0.332 (7.57a )

3-years

Accounting method

25.788 0.3661 (36.73a )

−0.2377 (10.94a )

4-years

% Concordant

4.59 0.349 (30.05a )

1999

0.2979 (3.48c ) 45.5

0.3008 (3.54c ) 52

0.3013 (3.56c ) 50.5

0.3098 (3.78c ) 47.2

−0.2001 (3.43c ) −0.7437 (19.07a )

0.5593 (12.93a ) 51.1

0.565 (13.04a ) 56.4

0.5383 (11.95a ) 54.6

0.6392 (19.27a ) 45.6


Table 4. Logistic Regression Results for Related and Unrelated Acquisition Cases (All Cases Analysis). Year

3-years: year 1998–2001 2-years: year 1998–2000 For year 1999 2 years: year 1999–2001 a Significance

279

at the 1% level. at the 5% level. c Significance at the 10% level. b Significance

280

Table 5. Comparison Between Exclusive Related Acquisition Firms and Exclusive Unrelated Acquisition Firms (Firm-Level Analysis). Year Test Variables

1996

1997

Exclusive Unrelated Acquistion Firms

Industry growth rate Observations (firms) 5-years 4-years 3-years 2-years 1-year

441 0.6092 0.5943 0.3666 0.2263 0.1083

266 0.7041 0.6894 0.4127 0.2413 0.1096

Firm’s growth rate Observations (firms) 5-years 4-years 3-years 2-years 1-year

297–413 1.8036 1.3586 0.8967 0.5789 0.2588

180–250 1.4159 1.2122 0.8083 0.5712 0.2866

0.83 0.44 0.45 0.06 0.47

Change in market share Observations (firms) 5-years 4-years 3-years 2-years 1-year

297–413 0.788 0.553 0.4394 0.3237 0.1411

175–243 0.6124 0.454 0.3613 0.3098 0.1692

0.55 0.39 0.45 0.11 −0.51

Accounting method Observations (firms)

0.0973 442

0.08 275

t-Stat.

−1.79a −2.36b −1.55 −0.8 −0.18

0.78

1998

Exclusive Related Acquistion Firms


t-Stat.

466

297

0.4497 0.4391 0.2406 0.1096

0.4821 0.4959 0.2582 0.1197

310–427

219–276

1.0887 0.9319 0.5449 0.2724

0.8398 0.7535 0.4324 0.2218

310–426

215–272

0.5748 0.4335 0.295 0.1715

0.2541 0.1805 0.1691 0.1053

2.41b 2.56b 2.01b 1.99b

0.1158 475

0.0757 304

1.9a

−0.9 −2.19b −0.88 −0.83

1.23 1.25 1.71a 1.53

1999



t-Stat.

449

268

0.3275 0.3084 0.1221

0.326 0.3381 0.1366

323–389

193–229

0.6121 0.4496 0.1739

0.3941 0.3938 0.164

323–388

189–225

0.2479 0.1214 0.0589

0.0703 0.0476 0.0345

2.23b 1.41 0.95

0.1269 457

0.066 288

2.77c

0.05 −1.57 −1.3

1.92b 0.68 0.37



t-Stat.

392

252

0.1842 0.1691

0.1751 0.1801

312–335

195–214

0.2767 0.1893

0.2694 0.2348

312–335

192–211

0.1089 0.0311

0.0865 0.0494

0.5 −0.64

0.135 400

0.0703 256

2.8c

0.42 −0.86

0.13 −1.31

KWANG-HYUN CHUNG


0.3652 355

0.3431 216

Remarks: For year 1996 5-years: 4-years: 3-years: 2-years: 1-year:

year 1996–2001 year 1996–2000 year 1996–1999 year 1996–1998 year 1996–1997

For year 1997 4-years: 3-years: 2-years: 1-year:


For year 1998 3-years: year 1998–2001 2-years: year 1998–2000 1-year: year 1998–1999 For year 1999 2 years: year 1999–2001 1-year: year 1999–2000

0.94

0.3921 383

0.3946 253

−0.1

0.4065 409

0.4441 247

−1.55

0.4393 363

0.4374 228

0.07


Equity compensation ratio Observations (firms)

a Significance

at the 1% level (two-tailed test). at the 5% level (two-tailed test). c Significance at the 10% level (two-tailed test). b Significance

281

Table 6. Logistic Regression Results for Exclusive Related and Unrelated Acquisition Firms (Firm-Level Analysis). Year Observations Likelihood ratio intercept

1996 381 14.8395

418

453

504

14.5963

9.4373

5.0893

0.6431 0.7855 (8.81a ) (12.29a )

0.5715 (8.37a )

0.4891 (7.18)

Industry growth 5-years −0.4779 (7.86a ) 4-years −0.5957 (8.79a ) 3-years −0.6316 (5.71b ) 2-years −0.465 (1.41) Firm growth 5-years

1997 381

418

453

504

13.294

14.4448

10.432

0.6579 (9.39a )

0.7992 (5.62a )

0.5671 (8.23a )

452

5.2089 0.4908 (7.23a )

10.5679

498

552

11.6314

6.5834

0.4043 0.729 (4.61b ) (12.87a )

0.4503 (6.69a )

1998 452

498

14.212

552

14.0328

0.3922 0.7099 (4.32b ) (12.11a )

494

7.128

12.3096

528 11.9747

1999 494

528

12.46

12.01

0.45 0.6168 0.7605 0.6037 0.7517 (6.71a ) (10.77a ) (14.40a ) (10.38a ) (14.08)

492

492

5.9521

6.3144

0.4099 (5.07b )

0.4028 (4.87b )

−0.355 (4.58b ) −0.4297 (4.89b )

−0.3052 (2.25) −0.411 (2.47)

−0.1216 (0.35) −0.7261 (7.1a )

−0.1926 (0.24)

−0.581 (4.45b ) −0.229 (0.44)

−0.3418 (1.48) −0.057 (0.03)

−0.1626 (0.35) −0.9106 (5.26b )

−0.7725 −0.2747 −0.1884 (4.11b ) (0.61) (0.3)

0.1237 (1.24)

0.0677 (0.18)

0.174 (3.59c )

4-years

0.2199 (4.32b )

3-years

0.1691 (4.25b ) 0.2422 (3.28c )

2-years

0.1068 (2.03) 0.3052 (3.12c )

Change in market share

0.1884 (2.25) 0.241 c

CEO 0.239 compensation (0.34)

0.0596 (0.02)

0.2415 (0.44)

0.0646 (0.04)

0.1664 (2.42)

0.3358 b

0.3626 b

(3.38 )

(3.94 )

(3.87 )

0.2445

0.0683

0.2419

(0.36)

(0.03)

(0.44)

0.3612

0.3418

c

b

(3.19 )

(0.03)

(1.09)

(1.79)

(1.59)

0.2418

0.186

(2.69 )

(2.53)

(1.25)

−0.3984 −0.4567 −0.388

−0.2988 −0.1485 −0.297

−0.143

(1.4)

c

0.242 c

(6.14 )

0.0608 −0.3481 −0.4206 −0.375

0.2399 (3.56 )

(2.09)

(1.69)

(0.86)

(0.23)

(0.85)

(0.22)

0.1492 (0.53) 0.0844 (0.07)

0.0777 (0.06)

Accounting

0.0515 (0.01)

% Concordant 57

0.0271 (0.01)

0.0616 (0.03)

0.2475 (0.43)

0.0649 (0.02)

0.0245 (0.01)

0.0514 (0.02)

0.2568 (0.46)

0.5036 (2.3)

0.355 (1.25)

0.4724 (2.32)

0.4693 (1.98)

0.3395 (1.13)

0.468 (2.27)

0.906 (6.02b )

0.7754 (5.1b )

0.9136 (6.12b )

0.7778 (5.12b )

0.6667 (4.45b )

0.6584 (4.34b )

58.8

58.4

56

56.7

58.6

58.6

56.1

56.9

56.4

54.4

57.3

56.6

54.7

56.3

57

56.1

57.2

54.9

55.5

Remarks: For year 1996 5-years: 4-years: 3-years: 2-years:


For year 1997 4-years: year 1997–2001 3-years: year 1997–2000 2-years: year 1997–1999 For year 1998 3-years: year 1998–2001 2 years: year 1998–2000 For year 1999 2 years: year 1999–2001 a

Significance at the 1% level. Significance at the 5% level. c Significance at the 10% level. b

284

KWANG-HYUN CHUNG

The multivariate logistic regression models are formed to explain the firm’s acquisition decision using the industry growth potential, motivation of increasing market powers and improving operating performance, CEO’s wealth-increasing motivation, and the firm’s motivation of avoiding goodwill by using the poolingof-interest method. Because of the high correlation between the firm’s growth rate and change in market share, those variables are not in the model simultaneously for the potential multi-collinearity. Firm’s industry growth rates as proxy for industry growth potential are mostly significant in a hypothesized direction in explaining different types of acquisitions, except for year 1999, possibly because of the shorter period of measurement. The growth rate as post-acquisition performance measure was a less significant variable, compared to the industry growth rate. The post-acquisition performance is a good indicator to explain the firms’ exclusive-related acquisitions in years 1996 and 1997. Also, the change in market share can explain the types of different acquisition in years 1996 and 1997. On the contrary, the accounting method can explain the types of different acquisitions only in the later test periods. This result may be explained by the fervent use of the pooling-of-interest method in the late 1990s because of the imminent abolishment of the method. As found in Table 5, none of the CEO’s compensation ratios can explain the firms’ acquisition types in the models. The multivariate models explaining the firm’s acquisition decision strategy are not fit, especially in the late 1990s (Table 6).

5. CONCLUDING REMARKS Through M&A, firms can diversify to balance cash flows and spread the business risks, or firms can improve efficiency or effectiveness by reducing competition, and firms can foster their growth by creating more market power. Depending on their corporate strategic motivations, firms can acquire the unrelated businesses and/or the related businesses. Diversification motivation is more likely to lead to the unrelated acquisitions while the related acquisitions are more likely to result from the motivation of reducing competition and/or creating market power. Thus, this paper attempts to identify the firms’ different strategic motivations for their M&A activities by relating the corporate acquisition decisions to their assessment of industry growth potential, post-acquisition firm growth, market share change, choice of accounting methods, and CEO’s stock compensation composition. The empirical tests reveal that in all acquisition cases, the industry growth opportunities play a key role in choosing between unrelated acquisition cases and the related acquisition cases, regardless of firm membership. Both univariate comparison test and the multivariate logit regression show that we tend to have higher


285

industry growth rate for unrelated acquisition cases, compared to related acquisition cases, which is consistent with product-market theory indicating higher risk under unrelated diversification acquisitions. Also, the choice of accounting method was different between two types of acquisition cases. Firms under the related acquisition cases tend to favor the pooling-of-interest method to the unrelated acquisition cases, although most acquisitions were accounted for using the purchase method. Because this analysis includes all acquisition cases regardless of firm membership, a firm can have related and unrelated acquisitions in the same year. However, in the firm-level analysis, I excluded these firms that have both related and unrelated acquisitions in the same year so that only exclusive-related acquisition firms and exclusive-unrelated acquisition firms are shown in each year. This test also shows that the types of acquisition decisions were more related to acquiring firms’ industry growth rate. The post-acquisition performance measures, using firm’s growth and change in market share, were consistent in explaining the types of exclusive related and unrelated acquisitions in the earlier test periods (i.e. 1996 and 1997). The accounting choice in the firm-based analysis was compatible with the types of acquisition as in the industry-wide analysis except the earlier years. However, CEO’s compensation composition was not related to the different types of acquisition though previous studies suggested that firm’s growth opportunities affect CEO’s stock compensation composition ratio. This study has some limitations. First, there is potential misclassification between related acquisition and unrelated acquisition, because the two-digit SIC codes of both acquiring and acquired firms were mechanically used. Second, for the firms’ assessment of industry growth potential, the ex-post industry growth rate was used instead of the ex-ante variable. Because of the use of ex-post industry growth rate, the test variables for the recent test periods have a short measurement time span for the data availability. Last, this study used the compensation data confined to S&P 1,500 companies while the all acquisition cases cover most of public firms in the SDC’s Worldwide Acquisitions and Mergers Database.

NOTES 1. Thus, the exclusive related-acquisition firms in a test period could be classified as the exclusive unrelated-acquisition firms in the other test period. 2. Dess, Ireland and Hitt (1990), Hoskisson and Hitt (1990), Davis and Thomas (1993), and Brush (1996). 3. Walker (2000). 4. Berger and Ofek (1995). 5. In the late 1990s, the FASB indicated that the pooling-of-interest method is no longer appropriate accounting principle in business combinations. The potential accounting rule

286

KWANG-HYUN CHUNG

change seems to have pushed many firms in the business combination to use the method since the late 1990s.

ACKNOWLEDGMENTS This research was sponsored by Lubin Summer Research Grant (2002). I appreciate the comments and suggestions from J. Lee and M. Epstein (the editors), as well as participants at the 2003 AIMA Conference. I also thank Ryan Shin for excellent research assistance.

REFERENCES Berger, P. G., & Ofek, E. (1995). Diversification’s effect on firm value. Journal of Financial Economics, 37(January), 39–65. Brush, T. H. (1996). Predicted change in operational synergy and post-acquisition performance of acquired businesses. Strategic Management Journal, 17(January), 1–23. Davis, R., & Thomas, L. G. (1993). Direct estimation of synergy: A new approach to the diversityperformance debate. Management Science, 39(November), 1334–1346. Dess, G. G., Ireland, R. D., & Hitt, M. A. (1990). Industry effects and strategic management research. Journal of Management, 16(March), 7–27. Hoskisson, R. E., & Hitt, M. A. (1990). Antecedents and performance outcomes of diversification: A review and critique of theoretical perspectives. Journal of Management, 16(June), 461–509. Narayanan, M. P. (1996). Form of compensation and managerial decision horizon. The Journal of Financial and Quantitative Analysis, 31(December), 467–491. Porter, M. E. (1987). From competitive advantage to corporate strategy. Harvard Business Review, 65(May–June), 43–59. Porter, M. E. (1996). What is strategy? Harvard Business Review, 74(November–December), 61–78. Scanlon, K. P., Trifts, J. W., & Pettway, R. H. (1989). Impacts of relative size and industrial relatedness on returns to shareholders of acquiring firms. The Journal of Financial Research, 12(Summer), 103–112. Walker, M. M. (2000). Corporate takeovers, strategic objectives, and acquiring-firm shareholder wealth. Financial Management, 78(Spring), 53–66. Young, B. (1989). Acquisitions and corporate strategy. Financial Management, 67(September), 19–21.

THE BALANCED SCORECARD: ADOPTION AND APPLICATION Jeltje van der Meer-Kooistra and Ed G. J. Vosselman ABSTRACT Technological advances and increasing competition are forcing organisations to monitor their performance ever more closely. The concept of the balanced scorecard offers a systematic and coherent method of performance measurement that in particular concentrates on assessing present performance in the light of an organisation’s strategy and takes into account the importance of the various policy aspects. In this paper we study the extent to which the concept contributes to the desired improvement of performance. To this end, we examine the motives for adopting the concept and the decision-making process around this adoption. We study the functioning of the balanced scorecard as a means to control performance, assuming that its functioning is linked to an organisation’s problems and is influenced by other control instruments used. This is why we have done case research.

INTRODUCTION In the management accounting discipline performance measurement and management has received a great deal of attention over the past few decades. Various players in this field have developed “new” performance and control systems. Balanced scorecards (Kaplan, 2001a, b; Kaplan & Norton, 1992, 1993, 1996a, b), Advances in Management Accounting Advances in Management Accounting, Volume 12, 287–310 Copyright © 2004 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1474-7871/doi:10.1016/S1474-7871(04)12013-3

287

288

JELTJE VAN DER MEER-KOOISTRA AND ED G. J. VOSSELMAN

performance pyramids (Judson, 1990; Lynch & Cross, 1991), integrated performance measurement systems (Nanni et al., 1992): these are only some examples out of many. Some even seem to compete with one another, like the balanced scorecard and the performance pyramid. In theory and practice especially the balanced scorecard has been at the centre of interest. Possibly this is partly because of the authors being rather well-known in the consulting profession. The balanced scorecard is an instrument with which the performance of organisations can be measured systematically and coherently. In recent years attention has shifted more and more from measuring performance towards managing it. Measuring is a means to achieve eventual performance management and control. Kaplan and Norton (1996a, b) claim that the aim is discovering cause and effect relations between the various areas of organisational activity and the organisational outcomes. Therefore the balanced scorecard concept defines critical success factors and performance indicators which reflect performance. The thinking behind this is that critical success factors determine the realisation of the strategic aims of the organisation and that the performance indicators are a more detailed concretisation. The indicators indicate which activities now and in the near future have to be carried out in order to successfully realise the aims.1 This paper is on the adoption and application of the balanced scorecard concept on the level of a specific organisation. For this organisation the adoption, implementation and use of the balanced scorecard has been examined. The paper is theoretically informed by institutional theory as concretised by Abrahamson (1991, 1996) and Abrahamson and Rosenkopf (1993). Using institutional theory we will examine the process of adopting, developing and using the balanced scorecard at NedTrain, a Dutch organisation in the field of public transport. Furthermore, the NedTrain study is informed by theoretical notions on control concepts underlying the use of the balanced scorecard concept. Particularly the paper will draw on control concepts distinguished by Simons (1994, 1995). By examining the adoption as well as the implementation and use of the balanced scorecard we hope to get an insight into highly relevant questions to professional practitioners who are confronted with decision-making connected with performance and control systems. Is the adoption indeed a consequence of deliberate decision-making by professional practitioners? If so, does the balanced scorecard live up to (either implicit or explicit) expectations in the adoption phase? How does it affect firms’ operations? And if not, what then drives the adoption? And what happens with the balanced scorecard after the adoption? The issues raised in this paper are not only relevant to academics, but also to practitioners. Our study in particular meets Lukka’s (1998) criticism of much current management accounting research: it is insufficiently aimed at accounting and control possibilities for it to be able to intervene in a firm’s operations. We

The Balanced Scorecard

289

would like to add that in current management accounting research the professional controller (or management accountant) has a very low profile. In this study, being the professional that is responsible for the economic rationality of all business processes (including those around the adoption, development and application of a balanced scorecard) he is given a high profile. The paper is organised as follows. In the following section we will briefly present the balanced scorecard’s origin and, drawing on Simons (1994, 1995), go into choices in the design of the control system around the scorecard, with particular attention to the role of the controller (or management accountant). In the next section, drawing on institutional theory, we will describe the motives underlying the adoption of the balanced scorecard, after which we will justify the case research method and procedures chosen. In the last section but one we will extensively report on our case research and describe the adoption and application processes of the balanced scorecard. Finally we will make some general remarks about the major findings of the case study.

BALANCED SCORECARD AND ORGANISATIONAL CONTEXT Control Concepts It is already about fifteen years ago that Kaplan and Johnson wrote the book Relevance lost: The rise and fall of management accounting. It proved to be an important milestone in what at present is often called the Relevance Lost Movement. In reality, this movement was already started before 1987 in two papers by Kaplan in the important journal The Accounting Review (1983, 1984). Kaplan asserts that systems and procedures of cost accounting and management control were originally developed for firms manufacturing mass-produced standard products. They are simple and direct cost information and responsibility accounting systems mainly aiming at minimising production costs. In his view they are not suitable for modern industry, which is especially characterised by client-specific production, short life-cycles, CAD/CAM technology and much “overhead.” The solution proposed consists of a number of elements, including refining costing and cost calculation techniques, prolonging the time horizon of control techniques and a shift away from a firm-centred orientation to a value chain approach. Another important element is the balanced scorecard, involving a widening of the scope of accounting reports with non-financial information from four perspectives: the customer, the firm’s internal processes, innovation and financial aspects. From each perspective a restricted number of critical success factors are formulated on the basis of

290


the organisational strategy, after which performance indicators and standards are determined. Bjørnenak and Olson (1999) hold that the contribution of the above-mentioned accounting innovations is that information can be given on more objects and that cause and effect relations can be described in greater depth. Traditionally there was only information available about production costs. New techniques, such as Activity Based Costing, Life Time Costing and Target Costing, also provide firms with information per buyer, per distribution channel and market segment, as well as an insight into the permitted costs of future products and the costs made during the life of products. Another development, called Strategic Management Accounting, aims at the firm itself also gathering systematic information on the costs of buyers, competitors and the like, in addition to cost information on the firm itself, thus looking beyond the firm’s boundaries. In order to acquire an insight into cause and effect relations financial information is supplemented by non-financial information, permitting links to be discerned between activities and financial results. The balanced scorecard is a performance measurement system which tries to respond to the growing complexity of the environment and the activities of the firms. Traditionally the performance measurement systems are finance-centred and examine realised performance after the event. Realised performance is compared with norms formulated in advance. An analysis of the difference may be a reason for taking action. In a stable environment such feedback systems may be useful, but they are not suitable when the activities of firms keep changing. An insight into the relation between (expected) activities and (expected) financial results provides companies with means to manage performance. In this process, the relations between participants outside the company, such as suppliers, business outlets and buyers are also considered. The Balanced Scorecard aims at giving this insight by translating into activities, by means of critical success factors and performance indicators, objectives with respect to all relevant policy aspects. The balanced scorecard’s perspective is forward-looking and long-term. With respect to the design of the balanced scorecard-based control system two views can be distilled from Kaplan and Norton’s material, without it becoming quite clear which view they themselves adhere to. A first view is that the control system, as is the case with “responsibility accounting,” is dominated by cybernetic “feedback-control” systems. Such systems will undoubtedly result in certain behaviour patterns. Judging from the findings of much “behavioural accounting” research dysfunctional behavioural effects and forms of information distortion should not at all be excluded (e.g. Argyris, 1990). To the extent that the balanced scorecard is seen as an addition to the traditional systems of “responsibility accounting” it is consequently also an instrument fitting a formulated strategy.


291

For, in Anthony’s (1988) reasoning, “responsibility accounting” is an instrument for the promotion of an effective implementation of the chosen strategy. In our opinion, many publications in this connection argue from a rather mechanistic point of view. The relative importance of the various dimensions of the strategy, and hence of the various performance indicators, should be defined beforehand in a consistent planning. In many cases this is a misjudgement of the complexity of the company’s activities and the uncertainty in the environment. Frequently, it is only gradually that the major strategic dimensions are beginning to be grasped. This takes us to the second view on the design of the balanced scorecard-based control system. This card might prove to be more effective as part of an interactive control system than as part of a diagnostic control system. According to Simons (1994, 1995) interactive systems are future-oriented. Their aim is a continuous explication of ideas and strategically important information in a (rapidly) changing environment. So in fact there is a “feed-forward” system. In this sense they have to be distinguished from diagnostic control systems. The latter consist for example of “responsibility accounting” systems, where revealing future-oriented strategic information is not required. On the contrary, there is a focus on accounting for activities a posteriori, albeit for that matter with the explicit aim to learn from them. It is, however, quite well imaginable that balanced scorecard-based control processes play a major role in the formulation and continual reassessment of strategy and a much less important role as a means to effectively implement already determined strategies. Put differently: they may rather be an instrument for strategy development than a strategic measuring instrument. Another characteristic of an interactive control system is that the strategic discussion takes place throughout the organisation. It is not only the privilege of the (senior) management, but the aim is the deployment of everybody’s knowledge and skills in the organisation. Sharing strategic and operational knowledge and skills is considered important, ensuring that the relations between views and operational consequences are discussed and made clear to everyone. The balanced scorecard is in such a case not an instrument for the management only, but the entire organisation. This requires planning a coherent whole of cards involving everybody in the organisation.

The Role of the Controller It is the controller’s core business to prepare and find arguments for particular choices of control systems. It may be assumed that, depending on the controller’s

292


position and conception of duty, the balanced scorecard-based control concepts will be planned differently. Jablonsky, Keating and Heian (1993) have done research into the changing role of the “financial executive” in internationally operating companies in the United States. In view of the job description of the position of “financial executive” the conclusions are also relevant to the position of controller in European companies. Interviews in six companies and a survey among over 800 companies, in which the opinions of financial as well as non-financial managers were elicited, result in a distinction between two profiles: that of the “corporate policeman” and that of the “business advocate.” The core values of the “corporate policeman” profile are: “oversight and surveillance, administration of rules and regulations, and impersonal procedures” (p. 2). The core values of the “business advocate” profile are: “service and involvement, knowledge of the business, and internal customer service” (p. 2). The controller as “corporate policeman” is an extension of the senior management and will introduce an instrument like the balanced scorecard in the organisation top-down. The most important part of the balanced scorecard will be an account of the delegated powers with an emphasis on realising performance standards. These standards are derived from the strategic policy drafted by the senior management. The controller as “business advocate” is a member of the businesses and supports the business managements. This controller is expected to know about the activities of the businesses, so that he/she can contribute to the discussions about the developments and changes in the business. As a team member he/she advances ideas about the financial organisation that most adequately fit the changes taking place in the business. The balanced scorecard will in the first place be an instrument to achieve a joint formulation of a strategy for the businesses and a more detailed elaboration of strategic policy. By planning and elaborating the balanced scorecard by mutual arrangement a strategic learning process develops. This brings about communication of the strategy and enables all participants to check which contributions they are making to realising the chosen strategy (also cf. de Haas & Kleingeld, 1999). The balanced scorecard contributes to the growth of a generally shared view of the organisation and the way this vision can be realised. By means of determining the standards everyone’s contribution to this becomes clear. Such a use of the balanced scorecard is aimed much less at accountability. The controller as an extension of the senior management fits into the traditional concept of “responsibility accounting.” The controller in the capacity of “business advocate” fits much more into the interactive control system discussed above. Design and functioning of the balanced scorecard will run closely parallel with the role of the controller.


293

DECISION-MAKING WITH RESPECT TO THE ADOPTION OF THE BALANCED SCORECARD In this section we will discuss the decision making with respect to the adoption of the balanced scorecard. We are interested in the motives leading to such a decision. In addition, we will examine to which extent there is a relation between these motives and the functioning of the balanced scorecard. Like March (1978), Abrahamson (1991, 1996) and Abrahamson and Rosenkopf (1993) we start from the assumption that organisations, like individuals, are not isolated and hence allow themselves to be influenced by the behaviour in other organisations. Abrahamson (1991) distinguishes four perspectives which can explain why organisations do, or do not, decide to adopt new accounting techniques. The first perspective, “the efficient choice,” is especially driven by internal incentives. The other three perspectives, “forced-selection,” “fashion” (because leading organisations use certain methods) and “fads,” are in particular inspired by organisation-external incentives. The efficient-choice perspective will be discussed in the next subsection “Internal incentives”; the other perspectives will be discussed in the subsection “External incentives.” The last subsection will examine the influence of the various motives for adoption on how the balanced scorecard functions. Internal Incentives It is questionable if decision-making processes regarding the adoption of the balanced scorecard can proceed along precisely drawn lines of technical efficiency. Technically efficient decision-making in the first place presupposes that decision-makers in organisations are relatively certain of their aims. In the case of the balanced scorecard this might be the case when enhancing the effectiveness and efficiency of the operations is the goal. But then the decision-maker has to be comparatively certain of the degree of effectiveness and efficiency with which the balanced scorecard can realise the objectives as formulated. For the decision-maker to judge the potential effectiveness and efficiency of the instrument he will have to have an insight into causal ways and means relations and to be able to quantify these insights. Thus, he will as a matter of fact have to know beforehand how precisely the balanced scorecard is going to be designed and used in the context of a control system, what exactly the effects will be on the behaviour patterns of managers and other members of the organisation and how such effects are translated into improving the performance of the organisation. Only when all this has become clear can a relatively reliable cost-benefit analysis of the balanced scorecard be made. It is no wild conjecture to suggest that this

294


certainty cannot be found. Put differently: owing to the uncertainty and complexity the professional decision-maker will have to sail into uncharted waters. The impossibility of making a reliable cost-benefit analysis beforehand does not mean that in practice there are no demonstrable reasons of technical efficiency to consider the adoption of a balanced scorecard. One of them being that the performance measurement systems in use are deemed to be insufficiently effective. Thus, traditional systems are on the whole strongly financially oriented. They assess the performance in the short term, and in doing so hardly link up with strategic policy. The balanced scorecard does link up with long-term policies and assesses performance in the light of these policies. By assessing the performance from various perspectives the emphasis is not exclusively placed on financial performance. Information about non-financial performance moreover provides an insight into the causes of the financial results. Moreover, an important reason for adopting the balanced scorecard may be found in changes in and in the environment of the organisation. It is those changes which organisation-internally necessitate a reconsideration of existing control systems. These changes may take place in market conditions, due to which for example existing products come under pressure. Changes may also be initiated by technological developments, due to which existing products become obsolete and new production methods become possible. Developments in information technology can also bring about changes, because the data gathering and processing possibilities increase and information becomes accessible at all levels and workplaces in the organisation. The growth of an organisation, in size as well as geographical extent, can have consequences for the way activities are organised and hence controlled. It is therefore at least plausible that Kaplan and Norton developed the balanced scorecard in reaction to changes in production and service organisations. It may be assumed that, on the level of the organisation, the effectiveness criterion completely supersedes the efficiency criterion when adoption is being considered. Put differently: professional decision-makers will not need precise cost-benefit analyses in connection with the adoption of the balanced scorecard. They are simply looking for effective systems. It is only gradually that the “costs” of the system will appear.

External Incentives In addition to internal incentives, decision-making in connection with the introduction of the balanced scorecard can especially be influenced by external incentives. An organisation may be forced to introduce a new instrument. In


295

particular governmental organisations can have so much (legal) power that they can impose the use of new instruments on other organisations. Mimetic behaviour can also lead to the adoption of new instruments. Sometimes organisations belong to the same organisational collective, that is, to the same group of competing organisations with respect to performance and/or raison d’être. Imitating this group can lead to a so-called “fad.” Then the decision to adopt the balanced scorecard is not based on effectiveness and efficiency considerations concerning the instrument, but on the simple fact that (many) other organisations have already done the same. In such processes two phases have been distinguished (DiMaggio & Powell, 1983; Tolbert & Zucker, 1983). In the first phase efficient choice behaviour is uppermost; especially the assessment of the technical effectiveness and efficiency is important. In a second phase the real fad starts to take off. As more and more organisations adopt the balanced scorecard the attractiveness for a non-adopter increases. This may be connected with the pressure from “stakeholders” like customers, suppliers and capital providers. A high incidence of the balanced scorecard can make this instrument rational for stakeholders: they associate adoption of the balanced scorecard with rational decision-making. Inversely, non-adoption can make them suspect that the organisation’s management is incapable of rational decision-making (also cf. Meyer & Rowan, 1977). This may result in their discontinuing their contributions to the organisation. Therefore many decision-makers may be expected to keep on the safe side and decide on adoption after all. For, as more organisations in a collective have chosen to adopt the balanced scorecard, the continuity of a specific organisation has an interest in adoption. Political factors that are not or not easy to calculate will then make the decision to adopt rational after the event. Decisions to adopt within a collective of organisations for that matter do not always have to be fad- or fashion-driven. Quite possibly, information from early adopters may enable late adopters to decide on grounds of technical efficiency. Such information, for this to happen, does have to be made available and actively influence decision-making by non-adopters. Mimetic behaviour may also create a fashion in response to actions by trendsetting organisations or networks. The latter include researchers at universities and business schools, and also organisation consultants. They disseminate their ideas by means of various media, like professional journals, books, seminars and personal communication and may be considered to be suppliers in a market for balanced scorecards (and other administrative innovations). They are economically active actors, whose self-interest is paramount. Processes of fashion setting, accompanied by standardisation of products, assist the suppliers. This improves the marketability of the products. The demand in the market is from professionals from the business world and governmental organisations. They do wish to take

296


their decisions to adopt (purchase decisions) on the grounds of efficiency and effectivity motives, but have to move into uncharted waters because there is uncertainty about aims, aims-means relations and future conditions of the environment. In short, there is ambiguity (March & Olson, 1976). Fashion setting processes can help the buyer, because they may give the decision to adopt a semblance of rationality. According to Abrahamson (1996) fashions and fads only have a chance when there arises not only the collective belief that adoption of the balanced scorecard is rational, but also when there is progress. This means that, in the view of the stakeholders, a clear improvement vis-à-vis the original situation must take place.

Adoption Motives and the Functioning of the Balanced Scorecard Fashions and fads are not necessarily good or bad for organisations. The adoption of the balanced scorecard can for example enhance a company’s political image. The attractiveness for (potential) stakeholders increases in such cases, with inherent positive economic consequences for the organisation. Adoption is then mainly a legitimising decision for the stakeholders. However, adoption can be more than an act of legitimising. It can kick-start a learning process, the fruits of which may be reaped by the decision-maker. Decisions to adopt which are mainly externally inspired and are by the participants hardly deemed to contribute to an increase of the effectiveness and efficiency of the activities, will lack broad internal support. As long as external legitimisation has hardly any consequences for what is being done internally there will be “loose coupling” (DiMaggio & Powell, 1983; Meyer & Rowan, 1977). The balanced scorecard will internally hardly be significant, because the concept will be elaborated superficially and the information it generates will play no significant role in decision-making and influencing behaviour. If the internal participants are of the opinion that performance has to be improved and are convinced that the balanced scorecard can make an important contribution, adoption may be expected to be followed by the actual implementation and use of the concept. The internal participants will be prepared to invest time in designing scorecards. They will also use information from the cards when taking decisions.

RESEARCH METHOD AND APPROACH The aim of the study is to acquire an insight into the decision-making process in connection with the adoption of the balanced scorecard and the way it functions.


297

We also aim to look at the role of the controller in the adoption process and the functioning of this instrument. Such an insight can only be gained to a sufficient degree if in-depth research is done into a real-life situation. Hence, the choice was made for a single case-study. In the last subsection we indicated that, when external legitimisation is the predominant adoption motive, the instrument will only have a symbolic function. We emphatically want to study the internal functioning of the instrument and want to know if applying it makes the expected contribution to improving performance. Therefore, we opted for a situation from practice where internal incentives also played a role in the decision to adopt. Moreover we are only interested in a practical case in which the balanced scorecard has been tested for some years. This means that the decision to adopt the balanced scorecard took place some years ago and that we, for information on this ex-ante phase, will have to rely on the memories of people closely involved with the decision. Hence, an interpretation of the decision a posteriori cannot be entirely excluded. Such a distortion of information can be neutralised as much as possible by interviewing more persons about this phase and by consulting written documents. People play various roles in organisations and will therefore interpret processes, activities and information differently. The balanced scorecard is a management accounting instrument. Such instruments are regarded as being among the tools of the controller. Nonetheless, the instrument is expected to be a means with which members of an organisation can control performance. The position of controller is a supportive one and differs from the line functions. Because of these different roles acquiring an insight into how the balanced scorecard functions requires the perspective of the controller as well as the perspective from the line. We did the case-study at the Dutch Railways. This company had a long tradition as a public company. In the early nineties the government decided to change the Dutch Railways into a private company. Within the Dutch Railways we did research in the service unit NedTrain, which in the context of the privatisation introduced the concept of the balanced scorecard. Here we conducted interviews with the central controller and line managers and controllers of the separate units. These interviews were conducted from October 1998 to February 2000. Moreover written material has been studied. In the following section we will report on the NedTrain case-study. We will start by describing the activities of NedTrain. Further, we will discuss the changing positioning of NedTrain due to the privatisation process Dutch Railways has been undergoing since 1995. Then we will describe the changes NedTrain have made in order to function as a business unit responsible for its own results. In this change process the Balanced Scorecard has played an important role.

298


ADOPTION AND APPLICATION OF THE BALANCED SCORECARD AT NEDTRAIN NedTrain Activities NedTrain is an independent operating unit within the Dutch Railways. NedTrain is an internal service unit, whose purpose is developing and designing, maintaining, cleaning, refurbishing and overhauling rolling stock. Its most important products are technical advice during procurement and construction, inspections and repairs, cleaning, refurbishment and overhaul. NedTrain has an engineering unit, NedTrain Consulting, two overhaul companies incorporated in the Refurbishment and Overhaul business unit, and five maintenance companies and forty service locations incorporated in the Maintenance and Services business unit. The overhaul companies provide long-term maintenance including overhaul and refurbishment of existing stock and equipment. The maintenance companies and the service locations provide the short-term maintenance. NedTrain mainly offers service to the Public Transport business unit, which is directed at transporting people. Until the end of 1999 NedTrain also offered internal services to Goods Traffic, the unit providing goods transport in the Netherlands. This service is still being provided, even though this unit left the Dutch Railways by the end of 1999 and has since been part of the Railion company. NedTrain’s staff comprises about 3800 ftes. From State-Run Organisation to Profit Organisation The Dutch Railways as a state-run organisation was bureaucratic, centralist and functionally organised. Customer-supplier relationships were absent or almost non-existent and marketing was given little attention. Attention was focussed entirely on technology and safety. The Dutch Railways were a budget-driven organisation and had to make sure that the costs remained within the budgets agreed with the government. For the Dutch Railways the government was the most important stakeholder. The government played a regulating and financing role. They determined the prices, the extent of the services and the budget. The Dutch Railways were the country’s only providers of rail transport. The Dutch Railways units, including the service unit NedTrain, were considered to be cost-centres. They too had to ensure that their costs did not exceed the annually agreed budgets. If, contrary to expectations, this did not work out adjustments were negotiable. Consequently, there was no or hardly any controlling by means of financial information. In the early nineties a commission of experts advised the government on the place of the Dutch Railways. In the framework of the tendency started in the eighties towards hiving off and privatising numerous state-run activities this


299

commission’s remit was to study a Dutch Railways hive-off. Further, to find out how the Dutch Railways’ monopoly could be transformed into a market position in which competition plays a role. According to its recommendations the Dutch Railways should develop into an independent, enterprising and customer-oriented company. It would eventually be listed on the stock exchange in order to cut the financial links with the government as well. This recommendation had far-reaching consequences for the position of the Dutch Railways and its internal functioning. It implied a radical change with respect to its protected position in the past. The proposal was to cut up the Dutch Railways into an operating part, to be privatised, and an infrastructural part, which was to remain state-owned. This split, which was also required by European regulations, has by now been put into effect. Afterwards, additional agreements were made in order to curb Dutch Railways’ monopoly position. The Dutch Railways were to concentrate on the core network. Regional lines were opened up to independent transport companies. After the departure of Goods Traffic by the end of 1999 the Dutch Railways has further restricted itself to passenger transport only. The Dutch Railways operating units were also confronted with these radical changes. The service unit NedTrain became an independently operating unit, which had to pay its own way. It had to become result-responsible, enterprising and customer-oriented. It was not only to receive internal customers, but would also quite clearly have to behave as a market party and try to attract customers in competition with others. Before this, NedTrain had owned all rolling stock and decided what maintenance had to be carried out. Public Transport and Goods Traffic, the users of stock and equipment, had hardly any influence on all this. This situation now changed. Public Transport now became the owner and was henceforth to be buyer of maintenance services. Thus, a customer-supplier relationship arose which was unthinkable before. The owners of the stock and equipment can now freely decide to buy external services. NedTrain now clearly has to take into account the wishes of the customers and be able to offer its services in conformity with the market. This turnaround had to be completed in a couple of years.

Changes in the Organisational Structure In 1993 NedTrain, on the eve of the hiving-off of the Dutch Railways, started to reconnoitre the changes to take place. To this end a business plan had been drafted, defining new markets. However, quantified objectives were all but absent. The process of creating a result-responsible unit was described and the necessity of developing and implementing a market was discussed. Result-responsibility requires knowledge of costs and benefits, the various buyers and their wishes and an insight into inter alia one’s market-position and competitors. The starting

300


point was that NedTrain did not possess the accounts with which the financial results could be related to operational activities. For this, the necessary basic accounts were lacking. There were no assets and debtors accounts, nor was there a profit and loss account. Nor was information available about buyers, markets and competitors. In the past, technical aspects used to be exclusively focussed upon. This was also reflected in the composition of the company’s management, where financial and commercial expertise did not feature. NedTrain’s management realised that without external assistance the changes would not be successful. Therefore, an external consultant was hired, who made an important contribution to the design of the changeover process. In addition, in the course of time, new kinds of expertise have been taken on board. NedTrain’s management was supplemented with a financial and commercial manager. In NedTrain’s business units, too, a number of new people were appointed on the managerial level. The structure of the organisation was adjusted. NedTrain used to be an internal service unit within the Dutch Railways. It had its own budget and was responsible for the costs, which should stay within the budget. Now NedTrain became a business unit with responsibility for its own results. Before the privatisation NedTrain delivered its services only internally and, as owner of the rolling stock, determined itself the quality and level of services. After the privatisation NedTrain was no longer the owner of the rolling stock. Now it delivered its services to both internal and external customers, who decided about the quality and level of services. In order to make all the participants aware of the changing position of NedTrain its management introduced own-result responsible units. This was a radical change, as NedTrain used to have a functional structure with many operating units. Thus, the business unit Overhaul and Service was created for short-term maintenance, the business unit Refurbishment and Overhaul for long-term maintenance and the business unit NedTrain Consulting for advice for the purchase of new, and the conversion of existing, stock and equipment. Each unit was given its own management, including a controller. Since a withdrawal to the core network, and hence a decrease in service to the internal buyers, had been anticipated, the new policy planned for an increase in service to third parties. Moreover, by the end of 1999 Goods Traffic left the Dutch Railways and became an external customer. Figure 1 gives an overview of the new organisational structure.

Adoption of the Balanced Scorecard NedTrain had to become a profit oriented unit with its own customers. In the past performance was expressed in technical terms and, as the unit determined


301

Fig. 1. Organisational Structure NedTrain.

the quality and level of services itself, it did not need to be aware of customers’ wishes. The new orientation required information about costs and revenues of the various services and per customer, and information about customers’ wishes and satisfaction. Because the level of internal services was expected to decrease NedTrain was also expected to focus on external customers. Therefore, information about market developments and competitors had to be collected. The external consultant suggested the use of the balanced scorecard as one of the vehicles of change for NedTrain. As discussed before, the entire complex of changes should bring about a decrease in technical domination and place financial and commercial consideration more at the centre. The aim was for the engineers to be made to realise as quickly as possible that NedTrain’s operations had to generate a profit and contribute to Dutch Railways’ profitability. To achieve this, it is no longer enough to only look at technical aspects and strive for technical perfection. The wishes of customers and the financial consequences of the technical proposals must also be taken into consideration. In addition to this an important objective was to force the internal orientation towards a more external one: customers, competitors and market developments should play an important role in decision-making. The balanced scorecard fits these objectives excellently. This concept is based on the thinking that performance is to be examined from various perspectives, in which process not only internal perspectives are relevant but also external ones as mentioned above. Further, the concept recognises the importance of learning and adjusting to new developments. This was an important feature as NedTrain’s management knew that the change processes would create an unstable situation for a longer period in which continuous adaptations to new insights and environmental changes would be at the fore. The eventual decision to adopt the balanced scorecard was taken in common consultation by the NedTrain management team, which includes the central

302


management and the management of the business units, the central controller and the consultant. At the moment of adoption there was no clear vision on the design of the control system connected with the balanced scorecard. Nor had the management team developed a clear strategic vision on NedTrain’s future position in the market and within the Dutch Railways. It was thought necessary for the organisation to become as quickly as possible fully aware of the fact that technology is also expensive and that there should be an external orientation. The balanced scorecard was considered a good instrument for putting these subjects on the agenda. Design, implementation and usage costs of the instrument played no role in the decision-making. According to the central controller: “the feeling was to start tentatively; then you cannot go wrong over that.”

Implementation The following process was followed for the implementation of the balanced scorecard. By the end of 1993 a “kick-off meeting” was organised, with the management team and all managers and controllers of the various units being present. In the spring of 1994 the concept of the balanced scorecard was presented unit by unit to all managers in workshops. Per unit so-called tandems were used: a local controller and an external consultant acted as pioneers of the balanced scorecard concept. Next, working groups were formed in each of the several companies to formulate critical success factors, which were drafted per perspective. So there was no integrated approach for drafting the critical success factors for the various perspectives. Occasionally, the sheets with critical success factors per perspective were simply stapled together. Dozens of critical success factors and their indicators were not unusual. The local management including the local controllers made a selection from all these factors. NedTrain’s managing director and the central controller, in cooperation with the managers of the various units, determined the critical success factors and their indicators per business unit, there being about 15 such indicators per business unit and between two and three applying across the board. In these discussions some important indicators were removed, being those providing an insight into the capacity utilisation of staff, the amount of service and logistic performance. NedTrain’s central controlling staff further elaborated the balanced scorecards, that is, the layout of reports and systems for drawing them up. The implementation process took place without any major problems. People knew that the structure, working methods, management and control of NedTrain would have to change. It was also known that the changes would have to take place over a relatively short period. It was realised that henceforth financial aspects would have to be paramount and that an external orientation is necessary.


303

Moreover, the analytical character of the balanced scorecard strongly appealed to technicians: “they understand the cockpit metaphor which stresses the importance of looking at various aspects” (central controller). Add to this that the non-financial assessment moments matched settled (technical) practices. This does not remove the fact that the management of some units had some trouble accepting the balanced scorecard. This resistance was especially due to the organisation becoming more transparent, which created a perception of one’s independence possibly diminishing and one’s own actions becoming the target of unwelcome attention.

Use of the Balanced Scorecard Because the introduction process was rather rapid it turned out afterwards that the design of the cards had not been clear on all points. Thus, there were problems with the nature of some of the performance indicators, with the uniformity of the concepts used and the accessibility of the reports. Some important performance indicators turned out to be absent: the ones referring to customer satisfaction, logistic performance and innovative capability. These problems have been addressed in the course of time, starting by defining the concepts unambiguously and improving the reports. It was also rather tricky to define performance standards. In fact, the discussion about NedTrain’s future and the objectives derived from this had not even started yet, nor were people used to working with clear performance standards. Therefore, initially attention was exclusively focussed on performance measurement, without feasibility and desirability being considered. The scorecards only comprised information about the existing situation without any linkages to targets, because they did not exist. The balanced scorecard was, at first, mainly used to acquire an insight into the relation between technical and financial aspects and to make people in the organisation ponder the critical success factors of their own units. This also supported the strategic discussion about the internal and external positioning of NedTrain as a whole, as well as also the strategic discussion within the business units themselves. Following this discussion the units’ management developed targets, which over time were included in the scorecards. After the appointment of the commercial manager the market perspective has received much more attention. With the aid of so-called customer dashboards for the various markets the relationship with the customers is systematically charted. This process is still going on. An insight into market position and characteristics of customers provide relevant information for the strategic discussion. In this discussion the positioning of NedTrain and the development of strategic alliances are addressed. Issues were discussed, such as what are the activities we should

304


carry out, which activities can be outsourced to external parties, what is the positioning of companies building rolling stock, were discussed. The last-mentioned companies are in the process of acquiring rolling stock maintenance companies, so that they not only can sell the rolling stock but are also able to take care of its maintenance. This enables them to conclude contracts with railway companies during the lifetime of the rolling stock. NedTrain’s key asset is that they possess knowledge about the relation between the rolling stock and the rail infrastructure, i.e. rail tracks and energy. The tendency is towards concluding long-term contracts with customers through which NedTrain guarantees the functioning of the customer’s rolling stock. This leads to another way of bidding and pricing, and subsequently other cost information. The role of the controllers is to support their line managers in developing these types of contracts and to deliver the relevant cost information. Further, the NedTrain management discusses the deployment of subcontractors for specific activities in a more structured way. Meanwhile, the internal process perspective has also been picked up. The various units have taken over the concept of the balanced scorecard and have started their own interpretation. The units argue that the central scorecard is too financially oriented and hence less suitable for their own management. Moreover, for a report to the top management underlying information is required. Starting from their own critical success factors the units scrutinise the operational activities. Doing so, some units descend to the shop-floor. Thus, the NedTrain Consulting business unit has developed cards on two levels within their own organisation, with which all processes and activities are assessed. The controller of this business unit has played a prominent role in developing these cards. He involved all the organisational units in the development of appropriate performance indicators and targets. All echelons contributed to the strategic discussion and the thinking about consequences for processes and activities. In this way everybody knows what is going on and responsibilities are shared. A benchmark and customer satisfaction study have revealed more about their own functioning and helped in formulating the objectives and targets. There are discussions about a further elaboration of the innovation perspective and people find it hard to concretise this. Beside the concept of the balanced scorecard the quality model EFQM is being used. The advantage of the EFQM model is that it is more complete and has development stages, allowing one to see where one stands at this moment and which steps have to be taken. The information derived from the scorecards and the quality model lays the foundation for the central scorecards. The central scorecards are the business unit management’s monthly means of reporting to the central management and largely determine the topics of the central management team’s meetings. Within the units the cards are viewed as an important management tool. They make action-oriented management possible,


305

with possible mutual monitoring of agreements. The information in the cards determines the agendas. Within the NedTrain Consulting business unit this management method has been generally accepted. In other units the balanced scorecard is used less generally; especially lower down in the organisation its influence on operational decisions is more restricted. There the concept functions more as a report and less as a management tool.

DISCUSSION AND CONCLUSIONS In the theoretical sections we discussed the adoption of new accounting practices as an answer to the growing complexity of the environment companies face. In addition to this efficiency motive institutional theory points at motives of fashion and fad which could also lead to the adoption of new accounting practices. Institutional theory claims that the motives for adopting accounting practices will influence their use. We argued that the balanced scorecard is an accounting practice which can handle complexity by giving both financial and non-financial information about the relevant aspects of companies’ performance. As this information is closely related to companies’ strategy, this concept encourages strategic discussions and fits within an interactive control concept. We argued that such a control concept requires controllers playing the role of business advocates. Subsequently we described the adoption and use of the balanced scorecard at NedTrain. In this section we intend to analyse these processes in the light of the theoretical claims made. NedTrain has used the balanced scorecard as a vehicle of change. It is one of the instruments deployed in the changeover process. Forced by the government NedTrain had to transform itself from a governmental organisation into a profit organisation: a huge turnabout with consequences for all aspects of business management. This reversal had to take place in a relatively short time. This decision of the government created a great deal of uncertainty. NedTrain had to change its technical orientation into a profit orientation, which required an insight into costs and revenues, customers’ wishes, competitors and market developments. As the company could not provide this information it needed new accounting practices which were capable of dealing with its new positioning in a changing environment. Moreover, changing an existing orientation can only be realised by making the participants throughout the organisation aware of the consequences of this change for their activities and decision-making. This is a complex process, which is why NedTrain recruited an external party for support in the changeover process. Like most consulting agencies this party was very experienced in introducing the concept of the balanced scorecard. The central controller defines the role of the consultant as follows:

306


At that time the balanced scorecard concept was a hype. We needed a new instrument in order to emphasise the financial consequences of our activities in the first place. As the consultant advised the use of the balanced scorecard, we accepted this advice without deliberate discussions.

Thus the adoption motive “fashion” as mentioned by Abrahamson (1991) certainly played its role here. But, as indicated by the central controller, the perspective of an efficient choice has emphatically played a role as well. Internally people were convinced that effectiveness and efficiency of performance would have to be scrutinised once again, from the point of view of a profit organisation dealing with customers and competitors. They were also convinced that the existing instruments were not able to provide the required information and that they needed new instruments for giving this insight. When the decision to adopt was taken it was not known if the concept of the balanced scorecard would be the most suitable one here. For this, the organisation relied on the external consultant’s advice, but a cost-benefit analysis was not made. More important was the expected contribution in the short term to realising a change in behaviour on the part of the technicians. In retrospect, it is admitted that another tool of performance assessment and control would also have been possible. The NedTrain management team were agreed that it was necessary to make the engineers financially aware and that the internal orientation would have to be transformed into an external one. There was, in advance, no clear idea of the consequences of the concept of the balanced scorecard for the activities and the way they were managed. A start was made in the expectation that gradually everything would become clearer. Vague ideas and expectations may be expected when an organisation does not know how its market position will develop, what the demands of the buyers are, which competitors it will have to fear and what internal position it will assume. Gradually, a clearer picture has emerged in connection with all this, with the concept of the balanced scorecard playing a role. The central controller commented: We introduced the concept in an evolutionary way. We asked the business units to describe the information they needed for managing their businesses. They proposed the performance indicators. At the beginning we stressed the importance of financial indicators. Over time we realised that the internal operations are the drivers of financial results. The units had to indicate what the most important operational processes were. We started from the actual situation because we did not know our strategy. We were not used to talking about strategy and responding to competitors’ actions and customers’ wishes. The balanced scorecard has helped us in making the actual situation more clear. We needed this information before starting the strategic discussion.

The design processes have encouraged people to ponder the critical success factors and the contributions to this from the internal units. The adoption of the concept by the several units and the drill-down processes to the shop floor have sparked off across-the-board discussions in these units. Thanks to these discussions the


307

operational processes have been charted more satisfactorily, providing a clearer insight into the mutual relationships. These discussions have also contributed to more clearly delineating NedTrain’s future plans and the roles of the several business units. Here, the information generated by the cards and the ensuing discussions within the organisation’s various echelons also make a contribution. The role of the central and decentral controllers is a supportive one here, in the sense of ensuring the availability of information and joining the discussions about the significance of the information for the decisions to be taken. The balanced scorecard is not employed as an accounting mechanism, but as an opportunity to hold each other to account and thinking about planned actions. Commenting on the use of the balanced scorecard concept the controller of the business unit NedTrain Consulting reflected: In consultation with the business units the central management has determined the central scorecard. Each month we report about a brief set of indicators to the central management. Quarterly we report about the whole set of indicators followed by an in-depth discussion with the central management. Using this information about the current situation we are able to discuss the environmental developments and their implications for the business units’ activities. I have advised using the concept also within our business unit. We were convinced that people throughout the unit should be informed about the operational processes and their financial consequences. Our scorecards are much more focussed on operational processes, in particular the scorecards used within our units. The scorecards have helped us to discuss our internal positioning, both within NedTrain and the Dutch Railways, and our external positioning. NedTrain Consulting used to be more externally focussed due to its advisory role about introducing new technological developments and determining the technical requirements of new rolling stock. As the demand of internal customers is decreasing we are widening our focus on the external market. Therefore, we have paid a lot of attention to customers’ satisfaction and its influence on the internal operations and processes. We have developed service specifications for each of the processes, which can be measured. Further, we have put much effort in measuring the performance of our Research & Development unit as a means to manage their activities. I have regular meetings with people of this unit in order to discuss the most appropriate performance indicators. We use the scorecards for discussing the current situation and whether we are on the right track to realise our strategic goals. We do not have a culture of “settling accounts” but of “talking to.” What is very important is that all the people are informed about what is going on and that there is a feeling of shared responsibility.

The central and decentral management are satisfied with the concept. The financial perspective is claimed to have been put on the agenda and to have contributed to a more external orientation, the two objectives deemed urgent at the beginning of the change process. At present the concept functions as an important management instrument. It determines the topics of discussion centrally as well as within the several units. The success of the implementation is ascribed to a number of factors. In the first place, the problems were widely acknowledged as was the necessity

308


of carrying through changes. Moreover, the concept had – and still has – the backing of the entire management and moreover very well matched the technical orientation of NedTrain’s personnel, making the concept easily accessible and widely supported. A lot of energy has also been lavished on the introduction and implementation of the concept and attempts have been made to involve as many people as was possible in organising this. In this process the central controller and the business unit controllers played a pivotal role. In the implementation process they functioned as intermediaries between the management and the people within the units. They supported them in developing their scorecards. They also paid attention to the linkages between the various performance perspectives and were not only focussed on the financial aspects. They participated in the monthly and quarterly discussions about the scorecards within the various management teams. We can conclude that the role controllers play is that of a business advocate. At the beginning the concept played a central role in changing the content of the agenda: it introduced financial, customer and market aspects. The information of the scorecards was very helpful for conducting the strategic discussion throughout the organisation. At present the concept plays a central role in managing performance and in the ongoing strategic discussion. This is, because the information from the cards determines to a large extent the agendas of the meetings of the central and decentral management teams. Nevertheless, now that the strategy has become much clearer the concept is changing to a means of reporting and discussing actions for the coming period. Although the scorecards are regularly discussed, this does not remove the fact that there are certainly differences between the various units with respect to the significance of the concept for the way things are done. In the NedTrain Consulting business unit the balanced scorecard has been accepted by the entire organisation. In the Refurbishment and Overhaul business unit this is less so. Here cards have been drafted on various levels, but they have less meaning the lower one descends in the organisation. The difference in acceptance is ascribed to the difference in knowledge and perhaps the larger number of personnel plays a role. We asked ourselves whether using the concept of the balanced scorecard produces the effects on performance anticipated during the adoption phase. Some remarks are called for here. In the specific case of NedTrain we are observing a radical change. It turns out that initially the management did not have any clear ideas of all consequences of such a change for the organisation. There were ideas about the direction of the changes and there were more well-defined thoughts about the changes that would at any rate have to take place as quickly as possible. Steps were taken without being able to survey beforehand their consequences. Gradually, things became clearer, without for that matter making the picture complete. We can also conclude that in a radical change process it is not one


309

instrument that is deployed but a whole complex of them. It is all interlinked forces within this complex which make a contribution to a change process, and this makes it impossible to isolate the contribution of one single instrument in such a complex.

NOTE 1. Some of the basic assumptions underlying the balanced scorecard concept have been criticized by Nörreklit (2000). Part of her criticism concerns the causality concept of the balanced scorecard. She concludes that there is not a causal but rather a logical relationship among the areas analysed (p. 82). Rather than viewing the relationship between non-financial measures as causal, the focus should be on coherence between measurements. Coherance focuses “on whether the relevant phenomena match or complement each other” (p. 83).

REFERENCES Abrahamson, E. (1991). Managerial fads and fashions: The diffusion and rejection of innovations. Academy of Management Review, 16, 586–612. Abrahamson, E. (1996). Management fashion. Academy of Management Review, 21, 254–285. Abrahamson, E., & Rosenkopf, L. (1993). Institutional and competitive bandwagons: Using mathematical modelling as a tool to explore innovation diffusion. Academy of Management Review, 18, 487–517. Anthony, R. N. (1989). The management control function. Boston: Harvard Business School Press. Argyris, C. (1990). The dilemma of implementing controls, the case of managerial accounting. Accounting, Organizations and Society, 6, 503–511. Bjørnenak, T., & Olson, O. (1999). Unbundling management accounting innovations. Management Accounting Research, 10, 325–338. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147–160. Haas, M., & de Kleingeld, A. (1999). Multilevel design of performance measurement systems: Enhancing strategic dialogue throughout the organization. Management Accounting Research, 10, 233–261. Jablonsky, S. F., Keating, P. J., & Heian, J. B. (1993). Business advocate or corporate policeman? Morristown, NJ: Ferf Research, Financial Executives Research Foundation. Judson, A. S. (1990). Making strategy happen, transforming plans into reality. London: Basil Blackwell. Kaplan, R. S. (1983). Measuring manufacturing performance: A new challenge for managerial accounting research. The Accounting Review, 686–705. Kaplan, R. S. (1984). The evolution of management accounting. The Accounting Review, 3, 390–418. Kaplan, R. S. (2001a). Transforming the balanced scorecard from performance measurement to strategic management. Part 1. Accounting Horizons, 15(1), 87–105. Kaplan, R. S. (2001b). Transforming the balanced scorecard from performance measurement to strategic management. Part 2. Accounting Horizons, 15(2), 147–161.

310


Kaplan, R. S., & Norton, D. P. (1992). The balanced scorecard-measures that drive performance. Harvard Business Review (January–February), 71–79. Kaplan, R. S., & Norton, D. P. (1993). Putting the balanced scorecard to work. Harvard Business Review (September–October), 134–147. Kaplan, R. S., & Norton, D. P. (1996a). Using the balanced scorecard as a strategic management system. Harvard Business Review (January–February), 75–85. Kaplan, R. S., & Norton, D. P. (1996b). Strategic learning & the balanced scorecard. Strategy & Leadership (September–October), 18–24. Lukka, K. (1998). Total accounting in action: Reflections on Sten Jönsson’s accounting for improvement. Accounting, Organizations and Society, 3, 333–342. Lynch, R. L., & Cross, K. F. (1991). Measure up! Yardsticks for continuous improvement. London: Blackwell. March, J. G., & Olson, J. (1976). Ambiguity and choice in organizations. Bergen, Norway: Universitetsforlaget. March, J. G. (1978). Bounded rationality, ambiguity and the engineering of choice. Bell Journal of Economics, 587–608. Meyer, J., & Rowan, B. (1977). Institutional organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83, 340–363. Nanni, A. J., Dixon, J. R., & Vollmann, T. E. (1992). Integrated performance measurement: Management accounting to support the new manufacturing realities. Journal of Management Accounting Research (Fall), 1–19. Nörreklit, H. (2000). The balance on the balanced scorecard – A critical analysis of some of its assumptions. Management Accounting Research, 11, 65–88. Simons, R. (1994). Control in an age of empowerment. Harvard Business Review (March–April), 80–88. Simons, R. (1995). Levers of control: How managers use innovative control systems to drive strategic renewal. Boston: Harvard Business School Press. Tolbert, P. S., & Zucker, L. (1983). Institutional sources of change in the formal structure of organizations. Administrative Science Quarterly, 28, 22–39.

Advances in Management Accounting, Volume 12 (Advances in Management Accounting)

Advances in Management Accounting, Volume 11 (Advances in Management Accounting)

Advances in Management Accounting, Volume 9 (Advances in Management Accounting)

Advances in Management Accounting, Volume 13 (Advances in Management Accounting)