Tunis Mldllgcn State University This is to certify that the dissertation entitled A MODEL OF NYSE FIRM MANAGER POSITION AND PARTICIPATION CHOICE ON THE MARCH 1985 FASB EXPOSURE DRAFT: EMPLOYERS' ACCOUNTING FOR PENSIONS presented by Georgia Pierce Saemann has been accepted towards fulfillment of the requirements for Ph.D. Accounting degree in ‘ I Major professor I Date I ' lq % "(H-'- .- ALJ A - n- Ill ‘ n ’ ' 0-12771 ‘MSU LIBRARIES .—:—- RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. A MODEL OF NYSE FIRM MANAGER POSITION AND PARTICIPATION CHOICE ON THE MARCH 1985 FASB EXPOSURE DRAFT: EMPLOYERS' ACCOUNTING FOR PENSIONS BY Georgia Pierce Saemann A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Accounting 1987 ABSTRACT A MODEL OF NYSE FIRM MANAGER POSITION AND PARTICIPATION CHOICE ON THE MARCH 1985 FASB EXPOSURE DRAFT: EMPLOYERS' ACCOUNTING FOR PENSIONS BY Georgia R. Saemann The Financial Accounting Standards Board (FASB) must anticipate the concerns of corporate constituents over the economic consequences of mandated accounting changes if it is to build support for these changes. The present research examines the relationship hypothesized between the economic consequences associated with a proposed change in pension accounting (the ED) and the behavior (position and participation choice) of New York Stock Exchange (NYSE) firm managers in the accounting standard-setting process. A replication of prior research is provided by examining the relationship between corporate attributes and manager position and participation choices. Additionally, the relationship between these choices and a manager's stated expectations of the corporate costs associated with the ED are examined. Finally, the relationship between manager perceptions of the FASB and participation choice are examined. The study expands the sample frame used in prior studies to include a random subset of NYSE firm managers who did not lobby the FASB on the pension accounting issue. A questionnaire is used to obtain measures of manager position choice, cost expectations and perceptions of the FASB. Separate analyses of position and participation choice are run for participating and nonparticipating Georgia Pierce Saemann managers, and for opposing and supporting managers, respectively. Probit analyses and Mann-Whitney U tests are used to analyze these relationships. The principal findings suggest that, on the pension accounting issue (1) manager position and participation choices are not independent. Comment letter filers tend to oppose the accounting change. (2) Corporate attributes are related to the participation and position choices of NYSE firm managers. Firm size was significant in both the position and participation choice models and firm leverage was significant in the position choice model of nonparticipating managers. (3) Corporate attributes - firm size and pension plan status - are related to stated manager cost expectations of political, labor and administrative compliance costs, and debt costs, respectively. (4) Stated manager cost expectations (political, labor and administrative compliance costs) and perceptions of the FASB are related to manager participation choice. However, a significant relationship was not found between manager cost expectations and position choice. ACKNOWLEDGMENTS I would first like to thank Dr. David Lasater for chairing my dissertation committee and for serving as a mentor to me throughout the last two years. Thank you also to the other members of my committee, Dr. Ronald Marshall and Dr. Dale Duhan, for their suggestions and recommendations. A special thanks to Dr. Marshall for introducing me to the "politics" of accounting. I dedicate this dissertation to my husband, Bob, and our son, Eric, for their patience and support during my doctoral work. I am especially grateful to my husband who spent many hours helping me with the computer applications and with the preparation of the manuscript as well as providing much appreciated moral support. iv TABLE OF CONTENTS Page List of Tables ................................................... viii List of Figures ..................................................... x Chapter 1. INTRODUCTION .............................................. l 1.1 The Research Objective ............................... 1 1.2 Contributions of this Research ....................... 5 1.3 Organization of the Dissertation ..................... 7 2. SIGNIFICANT PRIOR RESEARCH ................................ 9 2.1 Introduction ......................................... 9 2.2 SFAS 33: General Price Level Accounting (GPLA) ...... 11 2.3 SFAS 34: Interest Capitalization .................... 21 2.4 SFAS 8: Foreign Currency Translation ................ 22 2.5 SFAS 52: Foreign Currency Translation ............... 27 2.6 Summary ............................................. 29 3. THE RESEARCH CONTEXT AND QUESTIONS ....................... 32 3.1 Introduction ........................................ 32 3.2 The Model of Position Choice ........................ 32 3.3 The Model of Participation Choice ................... 35 3.4 Employers' Accounting for Pensions .................. 37 3.5 Contracting and Monitoring Costs Associated with the ED ....................................... 48 3.6 The Hypotheses ...................................... 55 3.7 Summary ............................................. 66 METHODOLOGY .............................................. 68 4.1 Introduction ........................................ 68 4.2 Population Definition and Sample Selection .......... 69 4.3 Variable Measurement ................................ 70 4.4 The Questionnaire Design ............................ 84 4.5 Assessment of Reliability and Nonresponse Bias ...... 89 4.6 Construct Validity .................................. 94 4.7 The Statistical Analyses ............................ 96 4.8 Summary ............................................. 101 ANALYSIS OF DATA RELIABILITY AND VALIDITY ................ 104 5.1 Introduction ........................................ 104 5.2 Measurement Reliability ............................. 104 5.3 Construct Validity .................................. 107 5.4 Reliability and Validity of Dependent Variables.... 118 5.5 Nonresponse Bias ................................... 122 5.6 Summary ............................................ 133 ANALYSIS OF RESULTS ..................................... 137 6.1 Introduction ....................................... 137 6.2 Corporate Attribute Models ......................... 143 6.3 Manager Expectation Models .......................... 162 6.4 Manager Perception Models .......................... 174 6.5 Summary ............................................ 192 7. CONCLUSION AND LIMITATIONS .............................. 197 7.1 Introduction ....................................... 197 7.2 Conclusions and Limitations ........................ 199 7.3 Summary ............................................ 205 Appendices A. A COMPARISON OF THE FINANCIAL STATEMENT REQUIREMENTS UNDER THE MARCH 1985 EXPOSURE DRAFT AND FASB STATEMENT 36 AND APB OPINION 8 ........................ 206 B. THE COVER LETTERS AND QUESTIONNAIRE ..................... 209 C. PROBIT ANALYSES: CORPORATE ATTRIBUTES AND POSITION CHOICE ....................................... 219 D. PROBIT ANALYSES: CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE .................................. 220 E. PROBIT ANALYSES: MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE .................................. 221 F. PROBIT ANALYSES: CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE ...... 224 G. PROBIT ANALYSES: MANAGER COST EXPECTATIONS, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE ...... 234 Bibliography ..................................................... 238 Table 2A. 3A. 4A. 4B. 4C. 4D. 5A. SB. 5C. SD. 5E. 5F. 56. SH. SI. 5J. 6A. TABLES FINDINGS OF THE WATTS AND ZIMMERMAN STUDY ON GPLA (UNREGULATED FIRMS) .............................. SUMMARY OF THE ED'S EFFECT ON REPORTED PENSION EXPENSE.. CORPORATE COSTS ASSOCIATED WITH THE PENSION EXPOSURE DRAFT ................................ CORPORATE ATTRIBUTES .................................... MEASURES OF MANAGER PERCEPTIONS OF THE FASB ............. SUMMARY OF TESTS ON THE NORMALITY OF UNDERLYING DATA DISTRIBUTIONS ......................... CONSTRUCT MEASURES ..................................... SIMPLE CORRELATION MATRIX .............................. MULTIMETHOD-MULTITRAIT CORRELATION MATRIX .............. MCNEMAR TEST ON POSITION CHOICE ........................ THE PARTICIPATION CHOICE OF QUESTIONNAIRE RESPONDENTS AND NONRESPONDENTS ....................... THE STATED POSITIONS OF FILERS AND NONFILERS ........... A COMPARISON OF DESCRIPTIVE STATISTICS ON CORPORATE ATTRIBUTES FOR QUESTIONNAIRE RESPONDENTS AND NONRESPONDENTS ................................... THE PARTICIPATION AND POSITION CHOICE OF FIRST AND SECOND QUESTIONNAIRE RESPONDENTS ..................... A COMPARISON OF DESCRIPTIVE STATISTICS FOR FIRST AND SECOND REQUEST QUESTIONNAIRE RESPONDENTS ......... SUMMARY OF RESULTS ON RELIABILITY AND VALIDITY ......... PARTITIONING OF CORPORATE MANAGER POSITION AND PARTICIPATION CHOICES ................................ viii Page 15 40 76 80 83 99 109 111 115 121 124 125 127 129 130 134 139 6B. 6C. 6D. 6E. 6F. 6G. 6H. 6I. 6J. 6K. 6L. 6M. 6N. 60. 6P. 6Q. 6R. 68. SUMMARY OF SAMPLE REDUCTIONS ........................... PROBIT ANALYSES - CORPORATE ATTRIBUTES AND POSITION CHOICE ...................................... MANN-WHITNEY U TESTS - CORPORATE ATTRIBUTES AND POSITION CHOICE (ED FILERS) ....................... MANN—WHITNEY U TESTS - CORPORATE ATTRIBUTES AND POSITION CHOICE (ED NONFILERS) .................... PROBIT ANALYSES - CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE .............................. MANN-WHITNEY U TESTS ‘ CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE (TOTAL SAMPLE) ............... MANN-WHITNEY U TESTS - CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE (ED OPPOSERS) ................ MANN-WHITNEY U TESTS - CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE (ED SUPPORTERS) .............. PROBIT ANALYSIS - MANAGER COST EXPECTATIONS AND POSITION CHOICE ................................... MANN-WHITNEY U TESTS - MANAGER COST EXPECTATIONS AND POSITION CHOICE ...................... PROBIT ANALYSIS - MANAGER COST EXPECTATIONS AND PARTICIPATION CHOICE .............................. MANN-WHITNEY U TESTS - MANAGER COST EXPECTATIONS AND PARTICIPATION CHOICE ................. PROBIT ANALYSIS - MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE ......................... MANN-WHITNEY U TESTS - MANAGER PERCEPTIONS AND PARTICIPATION CHOICE .............................. AN OVERALL COMPARISON OF THE CORPORATE ATTRIBUTES, MANAGER EXPECTATIONS AND MANAGER PERCEPTIONS MODELS OF PARTICIPATION CHOICE ........................ PROBIT ANALYSIS - CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE ...... PROBIT ANALYSIS - MANAGER COST EXPECTATIONS, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE .................................. SUMMARY OF FINDINGS ON PROBIT ANALYSES .................. ix 140 147 150 151 157 158 159 160 166 167 172 173 178 179 181 183 189 193 Figure 2A. 3A. 3B. 3C. FIGURES THE RELATIONSHIP BETWEEN CORPORATE ATTRIBUTES AND CORPORATE MANAGER UTILITY ......................... THE RELATIONSHIP BETWEEN CHANGES IN CORPORATE COSTS AND A MANAGER'S POSITION CHOICE ................. THE RELATIONSHIP BETWEEN CORPORATE ATTRIBUTES, CORPORATE COSTS AND A MANAGER’S POSITION CHOICE ....... THE RELATIONSHIPS BETWEEN CORPORATE ATTRIBUTES, MANAGER COST EXPECTATIONS, PERCEPTIONS OF THE FASB, AND A MANAGER'S PARTICIPATION CHOICE .................. Page 12 57 59 63 CHAPTER 1 INTRODUCTION 1.1 The Research Objective The Financial Accounting Standards Board (FASB) has been in existence for over ten years and since its inception over eighty standards have been issued. A due process procedure has been followed wherein the views of constituents have been actively solicited by the Board. Donald Kirk, chairman of the FASB, has stated that the objective of this due process is to build consensus for financial accounting standards.1 The need to build consensus arises because the Board functions in a political setting.2 The Accounting Principles Board (APB) and its predecessor, The Committee on Accounting Procedure (CAP), were displaced, in part, because of a lack of support by their constituents.3 The continuance of the FASB may depend on its ability to build and maintain this support. 1The search for generally accepted accounting principles is important but as Kirk (1981) has noted, the objective of the due process procedures is not to search for and find consensus but rather to build it. 2Hinckley (1981, pp. 4,5) notes that there are three components to political activity: "(1) an application of power (in the sense of applying resources to determine outcomes), (2) a combination of conflict and coordination, and (3) a collective activity...." 3Lack of authority, inconsistency of standards, and inattention to normative criteria have all been cited as factors leading to the downfall of the CAP and the APB (Previts and Marino, 1979). These factors led to criticism and a lack of consensus by Board constituents. In order to build consensus and support, the FASB must be aware of the economic impact of each issue it addresses before the issue is placed on the formal agenda.4 By anticipating the claimed consequences which might arise from a proposed standard, the FASB might identify and include in the research task force, representatives from the interest groups who are most likely to be concerned with the proposal.5 Wyatt (1977) argued that the FASB needs to be aware of the economic and social consequences of a proposed change in accounting so that it can avoid surprises in the process of resolving technical problems, and so that it can be prepared to counter opposition. Finally, by anticipating the level of concern over economic consequences, the Board might assess the political propriety of placing the issue on its agenda at a particular point in time.6 The objective of this study is to add evidence to the existing body of knowledge about the behavior of corporate managers in the accounting standard-setting process. This body of knowledge may help policymakers forecast preferences and actions of individuals on proposed accounting changes. These forecasts, in turn, may be used by 4Rappaport (1977) uses the controversy over the FASB's publication of its Discussion Memorandum on "Accounting by Debtors and Creditors When Debt is Restructured" to illustrate the importance of the "preliminary assessment" of an accounting issue before it is placed on the formal agenda. 5For example, Kirk (1983, pp. 88—92) noted that, "the Board has to become expert on the impact of accounting changes on stock prices and behavior, [but] whether these two factors should affect the Board is a serious question." 61f the FASB is to survive, Gerboth (1973) argues that it must "supplement technical competence with political skill". A delayed action on an accounting issue may be necessary to promote a friendlier political atmosphere. policymakers to build consensus among their constituents on accounting standards. Watts and Zimmerman (1978) provided the first study on the relationship between certain corporate attributes of a company, and that manager's preference on a proposed accounting change. They used the microeconomic theory of self—interested individuals to predict the stated preferences of corporate managers on FASB Statement 33: General Price Level Accounting. Corporate attributes were used to proxy for the expected economic consequences associated with the proposed accounting change. Since 1978, improvements in research methodology and variable measurement have been made. The Watts and Zimmerman model has been used to explain and predict both the stated preferences (position choice) and the decisions to lobby (participation choice) of corporate managers on other accounting issues.7 The findings of these studies can be interpreted to provide support for the theory that there is a relationship between the expected economic consequences of an accounting proposal and the position and participation choices of a corporate manager.8 However, these studies are subject to some limitations. First, the findings may not be generalizable to all accounting issues. Only two proposed standards which affect the primary financial statements 7For example, see Zmijewski and Hagerman (1981), Griffin (1982, 1983), Kelly (1982, 1985), Lasater (1982), McKee (1982), and McKee, et a1 (1984). 8Economic consequences are defined to include the changes in contracting, monitoring and other corporate costs (e.g., political, administrative compliance, debt covenant, labor negotiation and management compensation costs) associated with a change in financial statement accounting and reporting. (FASB Statement 34: Interest Capitalization, and FASB Statement 8: Foreign Currency Translation) have been addressed in prior studies. Second, the positions of corporate managers who did not file comment letters on the accounting proposals have not been obtained in prior studies. Therefore, the studies of position choice have been limited to the population of comment letter filers. Further, the studies of participation choice have not controlled for a manager's stated position on an accounting proposal. Third, the construct (theoretical) validity of the proxies used in these studies for expected economic consequences has not been tested. The validity of a proxy may be assessed in terms of its predictive and its construct validity. Prior researchers have used only one method (corporate attributes) to measure expected economic consequences. The assessment of construct validity requires the use of two or more measurement methods for the same construct.9 Therefore, the construct validity of the proxies used in these prior research efforts could not be assessed. Finally, the models of participation choice used in prior studies have not included variables for manager perceptions of the financial accounting standard—setting process. In a political setting, these perceptions may be important determinants of a manager's decision to participate in the due process. Therefore, the inclusion of these variables may significantly improve the predictive ability of the participation choice model. 9See Campbell and Fiske (1959). 1.2 Contributions of this Research This study provides further tests of the relationship between economic consequences and corporate manager position and participation choice. The March 1985 FASB Exposure Draft: Employers’ Accounting for Pensions is used to address two research questions: 1) What is the association between the expected economic consequences of the proposed change in pension accounting and the stated position of a corporate manager? 2) What is the association between the expected economic consequences of the proposed change in pension accounting, manager perceptions of the FASB, and a coporate manager's decision to participate? The approach used in this study to answer these research questions reflects an attempt to address the limitations of prior research cited in the previous section. Two sets of proxies are used for the economic consequences associated with the proposed change - corporate attributes and managers' stated expectations of corporate costs. Three types of analyses are run using these proxies - correlation analyses between the proxies, and multivariate and univariate models of position and participation choice. First, a replication of prior research on the relationship between hypothesized economic consequences and the position and participation choice of a corporate manager is provided on the pension accounting issue. Corporate attributes offered in prior research efforts, are used to proxy for the economic consequences of the proposed change in pension accounting. Evidence is provided to support the theory that a manager's position and participation choices on the pension accounting issue are related to the expected economic consequences of that proposed change. These findings increase the generalizability of existing theories on corporate manager behavior in the accounting standard-setting process. Second, the positions of corporate managers who did n9; file comment letters on pension accounting are obtained through the use of a questionnaire. The inclusion of this group of managers produces an expanded sample for this study which is more representative of the population of FASB corporate constituents than the samples used in prior research efforts on position choice. Prior researchers have found that most corporate managers who file comment letters take an opposing position. Those findings suggest that the position and participation choices of a corporate manager may not be independent. The sample used in this study, which includes nonfilers, provides further evidence that those choices are not independent. Additionally, specific evidence about the relationship between economic consequences and manager position choice is provided for nonfiling as well as filing corporate managers. Finally, evidence is provided about the relationship between economic consequences and the participation choice of a manager given his position choice. Third, corporate managers' stated expectations of the costs associated with the proposed change in pension accounting are obtained through the use of a questionnaire. These cost expectations are used as alternative proxies for the economic consequences of a proposed change in accounting. An analysis of the correlations between these proxies and corporate attribute proxies provide evidence to support the theoretical validity of certain corporate attributes as proxies for economic consequenceslo. However, the evidence provided by this study does not indicate that stated manager expectations of economic consequences provide better proxies than corporate attributes for explaining and predicting manager behavior. Finally, additional variables not included in prior research efforts are described and tested in the model used to explain the corporate manager’s participation choice on the pension accounting exposure draft. These variables, which reflect the manager's perceptions of the financial accounting standard-setting process, are found to increase the predictive ability of the participation choice model.11 1.3 Organization of the Dissertation The remainder of the dissertation is organized as follows. Chapter 2 reviews the previous empirical tests of corporate manager lobby behavior in the financial accounting standard-setting process. Chapter 3 continues that line of research by hypothesizing that certain economic consequences are related to the position and 10For example, there is a high correlation between manager expectations of political, labor and administrative compliance costs, and firm size, the corporate attribute proxy for these costs. 11These variables include the corporate manager's desire to support the FASB and the manager’s perceptions of the costs of lobbying, the probability of influencing the policy outcome and the probability of affecting the FASB's continuance. participation choices of a corporate manager on the pension accounting issue. The FASB's March 1985 Exposure Draft: Pension Accounting by Employers, its expected effect on corporate financial statements, and the economic consequences of these effects are described. Finally, models of manager position and participation choice are offered for testing. The sample design, data collection, and statistical methodology are described in Chapter 4, including a discussion of methods used to assess the reliability and validity of variable measures. The findings on nonresponse bias, and the reliability and construct validity of the data are discussed in Chapter 5. Chapter 6 contains a discussion of the results of the empirical analyses using this data. Finally, Chapter 7 addresses the implications of the findings on prior and future research efforts and the limitations of the present study. CHAPTER 2 SIGNIFICANT PRIOR RESEARCH 2.1 Introduction The objective of this study is to add evidence to the existing body of knowledge about the behavior of corporate managers in the accounting standard-setting process. The particular context of this study is the pension accounting issue. Although the pension issue itself has not been specifically addressed in previous studies of manager behavior, a number of researchers have tested models of manager position and participation choice on other accounting issues. These models have been based on economic theories which assume that corporate managers are self-interested utility maximizers.1 A manager's position and participation choice on an accounting proposal are said to be driven by the proposal’s effect on the manager's expected utility. The corporate manager is hypothesized to support an accounting proposal if the expected utility derived from its adoption is greater than the expected utility from any possible alternative.2 Furthermore, the manager is hypothesized to lobby on the proposal, regardless of his or her position, only if the proposal’s expected effect on his or her utility is large. lUtility may be derived from wealth and effort. Downs assumes that an individual prefers more wealth and less effort. 2Alternatively, if all the possible alternatives have an expected negative effect on a manager’s utility, that manager may support a proposal if it has a lower negative effect than any other possible alternative. 10 According to agency and property rights theory, the manager’s utility is affected by a change in the value of the firm due to the existence of monitoring contracts between the managers and the owners and lenders of a corporation.3 According to this theory, shareholders and lenders of the corporation use accounting data to monitor the manager for value—reducing actions. Therefore, a manager's wealth and other utility variables are affected by a change in accounting standards when that change results in a change in the reported financial statements.4 Finally, the prior research on corporate manager lobby behavior has implicitly assumed an efficient capital market which supports a "no effects" hypothesis on the relationship between accounting changes and stock prices. Therefore, the only effects of an accounting change on firm value are assumed to be derived from contracting costs associated with debt covenants, management and other employee compensation contracts, and monitoring costs associated with consumer groups, competitors and government regulatory agencies. The magnitude of the effect of an accounting change on contracting and monitoring costs is expected to be determined in part by certain attributes of the corporation (i.e., firm size, leverage position...). The models of corporate manager lobby behavior used in prior studies are based on the theoretical link between corporate attributes and manager utility, 3See Jensen and Meckling (1976) for discussion of property rights theory and Ross (1973) for discussion of agency theory. 4Fama (1980) suggests that the labor market will eventually become aware of managers who do not act to maximize firm value and the future compensation for such managers will be reduced. 11 described above and summarized in Figure 2A. A manager’s position and participation choices are represented in the functional form, Di = fn(cl, c2, ..., Ck), (2.1.1) where, Di denotes an individual manager's binary position or participation choice on an accounting proposal and Ck denotes the corporate attribute used to proxy for the hypothesized contracting or monitoring costs. A review of the prior studies of corporate manager behavior in the accounting standard-setting process is provided in the present chapter. The models of position and participation choice, which have not been formally developed in prior published studies, are more fully developed in Chapter 3 using a decision theory framework.5 2.2 SFAS 33: General Price Level Accounting (GPLA) 2.2.1 Position Choice Watts and Zimmerman (1978) published the first empirical study using microeconomic theory to model the position choice of corporate managers in the accounting standard—setting process. They hypothesized that the position of a corporate manager (Di) on a proposed accounting standard is related to the size of the firm (c1) and the effect of the standard on the firm’s reported earnings (c2) such that, Di = fn(c1, C2) . (2.2.1) 5Lasater (1982) provides a formal development of a logit model of corporate manager position and participation choice. 12 Change in Corporate Accounting Attributes Contracting and Monitoring Costs Associated with the Change in Accounting Change in Manager Utility (Wealth and Effort) Position and Participation Choice on the Accounting Proposal FIGURE 2A THE RELATIONSHIP BETWEEN CORPORATE ATTRIBUTES AND CORPORATE MANAGER UTILITY 13 As firm size increases, the firm‘s political visibility increases and the potential effect of a mandated change in accounting standards on taxes and regulation (political sector costs) increases in relation to the effect on management compensation (private sector costs).6 Therefore, Watts and Zimmerman argued that the manager of a large firm is more likely to support standards that decrease reported earnings, resulting in decreased political sector costs, and the manager of a small firm is more likely to support standards that increase reported earnings, resulting in lower private sector costs. The February 15, 1974, Discussion Memorandum on General Price Level Changes (GPLA) was used as the basis for the Watts and Zimmerman study.7 Fifty-three corporate firms, which were represented by comment letters filed with the FASB, were included in their sample. The positions of the firm managers were determined from position papers filed with the FASB. Thirty-four papers were classified as "in opposition to" the proposal, eighteen as "in support of". One of the 34 papers could not be classified. The effects of GPLA on each firm's reported earnings were not measured for the discriminant analysis. Instead, the sizes of the depreciable and net monetary asset balances were used as proxies for the direction of the expected effect on the firms' reported earnings. Separate analyses were run for regulated and unregulated firms. 6Watts and Zimmerman (1978) do not address the effects of debt covenant or employee compensation costs. 7Their use of the GPLA Discussion Memorandum, which presents a neutral statement of the issue, as opposed to the Exposure Draft, should reduce the potential effects of strategic voting or game playing. See Brown (1981). 14 The Watts and Zimmerman model correctly classified 32 of the 34 positions taken by unregulated firms in comment letters filed with the FASB. Furthermore, the findings of the study, as shown in Table 2A, supported the hypothesis that the manager of a large firm would support the change and the manager of a small firm would oppose the change when GPLA income was expected to be lower than historical cost income. However, the findings did not support the hypothesis when GPLA income was expected to be higher than historical cost income. In this case, the managers of both large and small firms tended to oppose the change.8 The Watts and Zimmerman study had several statistical and theoretical limitations: (1) Inferences from the statistical results were limited by the small sample which "precluded using a hold-out sample and, furthermore, did not allow for more sophisticated econometric techniques to control for the multicollinearity" of the underlying factors. (2) The use of dummy variables as measures of the existence of management compensation plans, the effect on reported income, and the manager lobby position omitted potentially useful information, thereby reducing the power of the study. (3) The use of the GPLA issue as a means of testing may have weakened the results of the study since GPLA does not affect the primary financial statements. 8Watts and Zimmerman (1978) argued that the manager of a small firm who supported the GPLA Discussion Memorandum would not lobby because the marginal costs of lobbying are high for small firms. Therefore, they argued that all companies reporting higher GPLA income would tend to be opposers regardless of firm size. 15 TABLE 2A FINDINGS OF THE WATTS AND ZIMMERMAN STUDY ON GPLA (UNREGULATED FIRMS) GPLA Income Lower than Historical Cost Income Support Oppose Large firms 7 2 Small firms 1 16 Fisher Exact Probabilitya .0004 GPLA Income Higher than Historical Cost Income Support Oppose Large firms 0 3 Small firms 1 4 Fisher Exact Probabilitya .6250 aThe Fisher exact probability indicates the probability that two groups are from the same sample. See Siegel (1956, p. 96). 16 (4) Theoretically, using the size of the firm as a proxy for political visibility may not be valid. McKee (1982) addressed two limitations of the Watts and Zimmerman study. First, he used a jackknife hold-out procedure to test the predictability of the model.9 Second, he included an estimate of the magnitude of the GPLA's expected effect on restated earnings.lo Third, he included a variable for the volatility of the GPLA's expected effect on mean earnings. Watts and Zimmerman's study of position choice on the GPLA issue was replicated, and the fit of the original Watts and Zimmerman model (OM) was compared with McKee's extended model (GM). The magnitude and volatility of the financial statement effect for the GPLA proposal was determined using the Parker procedure.11 The discriminant functions were formulated based on two sets of data relating to the GPLA issue - the Discussion Memorandum (N = 30) and the Exposure Draft (N = 80). Using a jackknife holdout procedure to test the predictability of the two models, McKee found that the GM and OM misclassified approximately 20% and 27% respectively, of the companies lobbying on the GPLA Discussion Memoranda. On the GPLA 9Lachenbruck and Mickey (1968) recommend the following jackknife approach for small sample studies. One observation is held out in the estimate of the discriminant function which is then applied to predict the classification of the hold out. This process is repeated for each observation. 10Watts and Zimmerman (1978) included only the direction of income restatement effect in the form of a dummy variable. 11The Parker procedure was developed to estimate the effect of the GPLA Exposure Draft on restated earnings. Watts and Zimmerman used the Davidson and Weil estimation procedure, which did not comply strictly with the Exposure Draft. In four of the 30 cases included in McKee's study, the direction of change in restated earnings differed between the two procedures. l7 Exposure Draft, however, the GM misclassified approximately 59% of the lobbying companies versus a 39% misclassification rate for the OM. McKee’s findings contradict Watts and Zimmerman’s findings. They indicate that neither the OM nor the GM provides a good description of corporate manager position choice on the GPLA. Furthermore, the findings indicate that the manager position choice models may not be consistent between the Discussion Memorandum and the Exposure Draft on the same issue. McKee, Bell and Boatsman (1984) provided further improvements on the Watts and Zimmerman methodology arguing that the significance and stability of the model may be affected by problems with variable measurement and the use of discriminant analysis which assumes the explanatory variables are multivariate normal distributed.12 McKee e; _1 used a nonlinear logistic regression technique to analyze the data.13 They also used a different measure for the reported income effect of GPLA. Instead of a dummy variable for the direction of change in reported income, McKee et al used the estimated percentage change in 1961 to 1978 linear earnings trend. This was an improvement to the Watts and Zimmerman study because it incorporated the magnitude of change in reported income and a longer term effect for the variable measurement. Although the performance of the model was improved by these revisions, it remained unstable across the Discussion Memorandum and Exposure Draft samples. 12Like McKee (1982), McKee et a1 (1984) replicated the Watts and Zimmerman model on the GPLA Discussion Memorandum and Exposure Draft using a jacknife holdout procedure. 13Press and Wilson (1978) found that the logistic regression outperformed discriminant analysis in studies with nonnormal dependent variable data. 18 The McKee, Bell and Boatsman study provided some evidence that a logistic regression technique may provide a better statistical model for predicting position choice than the discriminant analysis. However, even when the stronger statistical tool was utilized, the findings indicated that the Watts and Zimmerman model did not provide a good predictor of the position choice taken by a corporate manager on the GPLA issue. This may be due to the use of the GPLA issue which does not affect the primary financial statements, the use of firm size as a proxy for political visibility, or a misspecification of Watts and Zimmerman's self—interest model.14 To summarize, when corrected for weaknesses in the statistical analyses, the findings from the studies on the GPLA issue do not support the Watts and Zimmerman model. This lack of support may reflect limitations of the studies. For example, the economic consequences hypothesized by Watts and Zimmerman may not be associated with the GPLA issue, which did not effect the primary financial statements. Further, there may be weaknesses in the proxies used for the economic consequences associated with the proposed change. Finally, the small sample size may have reduced the power of the studies. 14For example, a self-interest model may be descriptive of corporate firm manager behavior in the financial accounting standard setting process, but the Watts and Zimmerman model may not include the factors which drive that behavior. 19 2.2.2 Participation Choice Zmijewski and Hagerman (1981) examined the effect of a firm's political visibility (c1) and income strategy (c2) on the corporate manager’s decision to participate (Di) in the financial accounting standard-setting process.15 This may be represented notationally as, Di = fn(c1, c2). (2.2.2) They used four proxies, including firm size, systematic risk, industry concentration and capital intensity, to measure a firm’s political visibility. Income strategy was determined based on the methods used by the company to account for depreciation, inventory, pension costs and investment tax credit. Extreme income strategies were defined for those companies using gll income increasing or decreasing accounting methods and mixed income strategies were defined for those companies using some income increasing and some income decreasing accounting methods.16 Zmijewski and Hagerman tested the hypothesis that the manager of a highly visible firm or a firm which was following an extreme income strategy would be more likely to file a comment letter than the manager of a less visible firm who was following a mixed income 15Zmijewski and Hagerman(1981) argued that a firm manager follows an income strategy approach in selecting accounting techniques. A firm that utilizes all income increasing or decreasing accounting techniques is said to follow an "extreme income strategy". A firm that utilizes a combination of income increasing and decreasing accounting techniques is said to follow a "mixed income strategy". 16Levels of extremity were measured by weighting each accounting method. Three different scores were obtained using the different weightings and separate analyses on participation choice were run using each score. The findings from these separate analyses were consistent. 20 strategy. They tested this hypothesis by comparing the Watts and Zimmerman (1978) sample of lobbying, non—regulated firms with a random sample of non-regulated industrial firms listed on the CRSP tape. A comparison of the four political visibility factors and income strategies for the two populations yielded the following results: 1) A binomial test was performed using the Clopper-Pearson methodology. This test provided evidence to support the hypothesis that a firm manager following an extreme income strategy was likely to file a comment letter on the GPLA issue. Alternatively, a firm manager following a mixed income strategy was not likely to file. 2) A Mann—Whitney U test provided evidence to support the hypothesis that a corporation represented by a comment letter on the GPLA issue tended to be larger than a corporation not represented by a comment letter. 3) Finally, the Mann-Whitney U test did ggt provide evidence to support the hypothesis that there was a relationship between firm risk, industry concentration or capital intensity, and a manager’s decision to file a comment letter on the GPLA issue. To summarize, the findings from the study by Zmijewski and Hagerman support the hypothesis that there is a relationship between a manager’s participation choice and economic consequences of a proposed accounting change. Further, the findings suggest that corporate attributes may proxy for these consequences. 21 2.3 SFAS 34: Interest Capitalization Lasater (1982) used a logit probabilistic discrete choice model to examine the relationship between the position and participation choices (Di) made by a corporate manager and five wealth factors on the interest capitalization issue (SFAS 34). Financial statement proxies were used for debt covenant costs (c1), income tax effects (c2), administrative compliance (C3) and political costs (c4), and management compensation effects (c5) such that, Di = fn(c1, c2, C3, c4, c5) (2.3.1) A multinomial logit probabilistic choice model was used for a three-alternative case (favored, did not lobby, opposed). Binary logit models were used to separately predict position and participation choices. The models were found to be good predictors of the binary participation and position choices, correctly predicting the choices of a manager in more than 74 percent of the cases. However, Lasater argued that the three-alternative model is probably an incorrect characterization of the decision process since the participation and position choices are likely to be dependent, sequential events.17 The debt/equity ratio, a proxy for debt covenant costs, was the only individually significant factor (.005) in the pOsition choice model. Earnings before taxes (EBT), a proxy for political costs, was the only individually significant factor (.005) in the participation 17For example, a Downsian voter would choose a position and then decide whether or not to participate. 22 choice model. Based on these findings, Lasater concluded that the determinants of manager participation and position choices differ.18 Lasater's study, which included corporate attributes to proxy for the economic consequences of the interest capitalization issue, provided support for Watts and Zimmerman's model. These findings suggest that the insignificant relationships between manager position choice and corporate attributes on the GPLA issue may be issue specific. Stated otherwise, Watts and Zimmerman's model may not be generalizable to all accounting proposals. However, the model may be used to explain and predict manager position choices on accounting proposals which affect the primary financial statements. Finally, the findings suggest that corporate attributes may be used to proxy for expected economic consequences. 2.4 SFAS 8: Foreign Currency Translation Kelly (1982) and Griffin (1982) examined the participation choice of corporate managers on SFAS 8, "Accounting for the Translation of Foreign Currency Financial Statements".19 Kelly used t-tests and a probit analysis to test the relationship between a manager’s decision 18Lasater noted that the ex post explanation of the finding for EBT was not consistent with the effect he originally hypothesized for that variable. The relationship between EBT and a manager’s participation choice was derived from the hypothesized relationship between EBT and position choice. Therefore, because EBT was not significant in the position choice model, it was not hypothesized to be significant in the participation choice model. 19Kelly read the comment letters submitted to the FASB and concluded that none of the filers supported the proposed standard. Therefore, she argued that she was actually testing position agg participation choice. 23 to lobby (Di) against the foreign currency translation proposal, and five corporate attributes including foreign assets (Cl), leverage (c2), asset size (C3), percentage of management stock ownership (c4), and incentive compensation remuneration (C5). These relationships may be represented in the following functional form, Di = fn(c1, C2, C3, C4, c5). (2.4.1) In addition, Kelly examined the correlation between a manager’s participation Choice, the five corporate attributes, and subsequent Changes in financing or operating activities (Fi) when SFAS 8 was adopted. Kelly hypothesized that a manager who had filed a comment letter on SFAS 8 was likely to make subsequent Changes in financing or operating activities in order to avoid the impact of that statement on reported financial statements. Further, she hypothesized that there would be a similar relationship between the five corporate attributes and these subsequent Changes. This relationship may be represented in the following functional form. Pi = fn(Di, c1, C2, C3, C4, C5). (2.4.2) Kelly's sample included 52 companies responding to a Peat, Marwick, Mitchell and Company (1977) survey regarding the Changes made by managers in financing or operating activities subsequent to the adoption of SFAS 8.20 The results suggested that the hypothesized corporate attributes, except foreign assets (the proxy for the effect of SFAS 8 on financial statements), were associated with the manager’s decision to lobby on the foreign currency translation Exposure Draft. However, the manager's decision to lobby was not indicative of 20The nonresponse bias and measurement reliability limitations inherent in survey research were present in Kelly's study but were not addressed. 24 subsequent Changes in financing or operating activities by the company. Furthermore, the corporate attributes were pg; associated with subsequent financing or operating Changes. Based on these findings, Kelly concluded that reliance should not be placed on comment letters to assess the potential economic impact of an accounting Change. However, this conclusion is not supported by the findings. Kelly did not examine the positions taken by managers on individual aspects of the proposed accounting Change. She addressed only the participation choice of corporate managers. Therefore, it can only be concluded, from this study, that a manager's decision to lobby on an FASB pronouncement may not be indicative of the proposal’s impact on the financing or operating activities of the company he represents. Griffin (1982) used t-tests and logit analyses to assess the relationship between corporate attributes and a manager's decision to lobby on the foreign currency translation issue. The corporate attributes he used included the firm’s foreign exchange ratio (c1), leverage position (C2), size (C3), profitability (C4), and risk (C5).21 He hypothesized that, Di = fn(c1, c2, c3, c4, c5). (2.4.3) Griffin’s sample included 119 of the 147 firms represented by comment letters on SFAS 8 and 479 firms which had been included in a study by Dukes (1978) on the security price effects of that Statement. Like Kelly, Griffin found that leverage position and firm size (as measured by market share value), were significantly related to the 21The ratio of reported foreign exchange gain or loss to pretax net income was used to proxy for the effect of SFAS 8 on reported net income. 25 manager’s decision to file a comment letter on SFAS 8. Further, like Kelly, a significant relationship was not found between the foreign exchange ratio (a proxy for the effect of SFAS 8 on financial statements) and participation Choice.22 Finally, Griffin found that SFAS 8 did not have a major impact (as measured by the foreign exchange ratio) on corporate annual or interim financial statements. Although Kelly and Griffin used different samples and different proxies for firm size and the proposal’s financial statement effects, their results were consistent. Their findings suggest that a manager’s decision to file a comment letter on a specific accounting proposal is associated with corporate attribute proxies for leverage position and firm size. However, an association was not found between the manager’s decision to lobby and corporate attribute proxies for a proposal’s expected effect on reported financial statements. Kelly (1985) extended her study on SFAS 8 by using an expanded sample which included all firms that were forced to comply with the new standard. The proportion of foreign operations was used instead of foreign assets to proxy for the effect of SFAS 8 on reported financial statements. Finally, a matched pairs design was used to control for firm size. Kelly hypothesized that a manager's decision to lobby on the foreign currency translation accounting issue would be related to the firm’s proportion of foreign operations (C1), leverage (c2), asset size (C3), and management’s ownership percentage (c4). She omitted 22Firm profitability and risk were also insignificant in the model of participation choice. 26 the incentive compensation remuneration variable included in her earlier study. The group of comment letter filers was divided into two groups: (1) those concerned about the practical and implementation difficulties of the standard and (2) those concerned about income statement effects. Therefore, the hypothesis could be tested based on two binomial choice functions represented as, Di = fn(c1, C2, C3, C4) (2.4.4) In this function, Di denoted three sets of binary Choices - lobby due to implementation difficulties/do not lobby, lobby due to income statement effects/do not lobby, or lobby due to implementation difficulties/lobby due to income statement effects. The results from Kelly’s univariate analyses indicated that managers who lobbied due to implementation issues represented large firms with a high percentage of foreign operations. Managers who lobbied due to income statement effects had a high management ownership percentage and represented large, highly leveraged firms.23 Finally, managers who lobbied due to implementation issues represented firms with a higher percentage of foreign operations than managers who lobbied due to income statement effects. The results from these analyses were not as significant after controlling for firm size. 23Kelly had hypothesized the opposite relationship between the percentage of manager ownership and the manager's decision to file a comment letter. However, she noted that this finding supports Fama's (1980) argument that managers owning less of the firm are concerned with the influence of accounting reporting on their human capital values. 27 Only the management ownership variable remained significant in the income statement effects/lobby Choice analysis.24 The findings from Kelly and Griffin’s studies on SFAS 8 may be added to the findings from the study by Zmijewski and Hagerman on the GPLA issue. These findings increase the generalizability for the theory that there is a relationship between economic consequences and manager participation Choice. However, the findings suggest that the immediate effect of a proposed accounting Change on financial statements may not be a factor in the participation choice model. 2.5 SFAS 52 Foreign Currency Translation Griffin (1983) extended his study on the foreign currency translation issue by examining a model of corporate manager participation Choice on SFAS 52, Foreign Currency Translation. He tested a participation Choice model which included the five corporate attribute variables used in his study of SFAS 8. These variables included the firm's foreign exchange ratio (c1), leverage (C2), size (C3), risk (C4), and profitability (c5). In addition, a variable for the impact of the accounting Change on reported income (C6) was included in the analyses where, Di = fn(c1, C2, C3, C4, C5). (2.5.1) Univariate discriminant analyses were used to test the relationships between participation Choice and each separate 24The results of the logit analyses were not interpretable due to the multicollinearity between the explanatory variables. 28 independent variable. A multivariate discriminant analysis was used to test the predictive power of the model. Griffin’s sample included 156 of the 174 firms represented by comment letters on SFAS 52 and 296 firms which had been included in a study by Dukes (1978) on the security price effects of SFAS 8. The results from Griffin’s study suggest that managers who lobbied on SFAS 52 represented large firms which were not highly profitable. Further, managers who lobbied on SFAS 8 were more likely to lobby on SFAS 52. Statistically significant relationships were not found between participation choice and firm leverage or the impact of foreign exchange gains and losses on reported income. Finally, the models of participation Choice provided an average improvement of 8.71 percent over a naive method of prediction.25 The findings from this study differ from Griffin’s previous study on SFAS 8. The manager who lobbied on SFAS 8 tended to represent a firm which was more highly leveraged than the manager who did not lobby. However, on SFAS 52, Griffin did not find a significant relationship between leverage position and a manager’s participation choice. In fact, the leverage position of a company represented by a comment letter on SFAS 52 tended to be lower (insignificantly) than that of a company that was not represented by a comment letter. Further, firm profitability, which was significant in the manager’s decision to lobby on SFAS 52, was not significant in his decision to lobby on SFAS 8. These differences suggest that a manager's 25Griffin did not report the statistical significance of the overall model. 29 perceptions of the costs associated with the two accounting issues were not the same.26 The findings from Griffin’s study on SFAS 52 provide additional support for the theory that there is a relationship between manager participation Choice and economic consequences associated with a proposed accounting change. However, they suggest that the relationships may not be consistent between accounting issues or over time (if SFAS 8 and 52 are considered to be the same issue). 2.6 Summary Prior research on corporate manager behavior in the financial accounting standard-setting process has addressed the position and/or participation choices of managers on the GPLA, interest capitalization and foreign Currency translation issues. Proxies for political, debt, management compensation and administrative compliance costs have been tested. Improvements in research methodology and variable measurements have been made but the findings of these studies have not been consistent. The research on position Choice has been limited to the GPLA and the interest capitalization issues. The results on the GPLA issue, in terms of predictability were not significant after adjustments were made for weaknesses in the research methodology. However, weak support was provided for the firm size hypothesis using a Mann-Whitney U test. 26This finding is similar to the results reported by McKee (1982) and McKee et al (1984) on the GPLA issue. They found that manager position Choices on the GPLA were not the same on the Discussion Memorandum and Exposure Draft. 5‘. ._~ ‘AQ 30 On the interest capitalization issue, a model which included a debt cost proxy (the only significant factor in the model) was found to be a good predictor of a manager’s position Choice. The research on participation choice has addressed the GPLA, interest capitalization and foreign currency translation issues. Zmijewski and Hagerman, Lasater, Kelly and Griffin found that firm size was a significant factor in the model of participation Choice on each of these issues. Income strategy was also found to be a significant factor in the manager’s participation Choice on the GPLA issue.27 Firm leverage, profitability and percentage of manager ownership were significant factors in the participation Choices on the foreign currency translation issues.28 Only Kelly (1985) could find a significant relationship between an accounting proposal’s effect on corporate financial statements, and a manager's behavior in the accounting standard—setting process. She found that a proxy for the effect of SFAS 8 on corporate financial statements was significantly related to the manager's primary concern about that standard, as stated in a comment letter. The findings of these studies can be interpreted to provide support for the theory that there is a relationship between the economic consequences (contracting and monitoring costs) of an accounting proposal and the position and participation Choices of a 27Income strategy was not included in the participation Choice models on the interest capitalization or foreign currency translation issues. 28A significant relationship was not found between participation Choice and leverage position on the interest capitalization issue. Firm profitability and percentage of manager ownership were not included in the participation Choice models on the GPLA or interest capitalization issues. 31 corporate manager. The present study continues that line of research in the context of the March 1985 Exposure Draft, Employers' Accounting for Pensions. The position and participation choice models and the specific hypotheses tested in this study are developed in Chapter 3. CHAPTER 3 THE RESEARCH CONTEXT AND QUESTIONS 3.1 Introduction A review of prior studies on the position and participation choices of corporate managers in the financial accounting standard- setting process, including a brief overview of the theory underlying those studies, was provided in Chapter 2. The present Chapter includes a more complete discussion of the theoretical underpinnings of that line of research and provides a formal development of the models of manager position and participation Choices using a decision theory framework. An extension of the participation Choice model used in prior studies is offered based on the Downsian theory of political choice. A brief history of pension accounting standards is provided, and the contracting and monitoring costs specifically associated with the FASB Exposure Draft on Employers' Accounting for Pensions are discussed. Finally, hypotheses are set forth for the relationship between contracting and monitoring costs, and corporate manager position and participation Choice. 3.2 The Model of Position Choice It has been asserted that the utility of a corporate manager is affected by the cash flows of the company he or she represents.1 1For example, see Jensen and Meckling (1976) and Fama (1980). 32 33 Further, a reduction in manager utility is generally associated with an increase in corporate cash outflows.2 Given this assertion, a manager attempting to maximize utility is expected to apply accounting methods which minimize the cash outflows associated with explicit and implicit contracts (contracting and monitoring costs).3 Therefore, when a mandated Change in accounting standards restricts or Changes the set of generally accepted accounting methods, a manager will support the proposed Change if the accounting method he is currently applying is among the set of acceptable alternatives. Alternatively, a manager who would be required to change to a less favorable accounting method (in terms of corporate cash flow) will oppose the Change, unless he believes the current proposal represents the best attainable alternative. In this case, the manager believes the FASB is commited to a restrictive accounting Change and his preferred method is not among the FASB's set of acceptable methods. Given these assertions, a manager’s position choice on a proposed accounting standard is a function of the set of acceptable accounting methods. His preferred method (a*) is the one which will maximize his 2An exception may be an increase in management compensation costs which represents a direct cash transfer to the manager. 3Explicit contracts exist in the form of written debt covenants, management compensation contracts and union agreements. Implicit contracts exist in the form of unwritten agreements. For example, shareholders of a corporation expect managers to maximize firm value. Alternatively, governmental regulatory agencies, corporate competitors, and consumers are alert to companies (and managers) who attempt to monopolize an industry or earn excess profits. Published financial statements may be used by creditors, employers, shareholders and others to monitor manager activities for compliance with explicit and implicit agreements. 34 utility. Notationally, this may be represented as, E(U|a*) = max E(Ula), (3.2.1) aEA where A denotes the set of acceptable accounting methods and E(Ula) is the expected utility from selecting method aeA.4 The manager's position on a proposed accounting standard (a’) may be given by, 1 (support), if E(U|a’) = max E(u|a) 0 (oppose), otherwise. (3.2.2) The manager's expected utility from selecting an accounting method (E(Ula)) is derived from that method's effect on the cash outflows of the corporation he represents. Notationally, E(U|a) = E[£U(c,a)], (3.2.3) ceC where U(C,a) represents the utility assessment of the corporate cash outflows (costs) that will result if standard aeA is adopted. The magnitude of these costs, which result from a Change in the reported financial statements, may be affected by certain attributes of the firm. For example, the financial statements of a large corporation may be monitored more Closely by governmental regulatory agencies, competitors and consumers than the financial statements of a small corporation. Therefore, the manager of the large corporation may face higher costs in avoiding government interference due to mandated Changes in financial statement accounting. In this case, the manager’s utility assessment will be a function of certain corporate attributes feF. Notationally, then E(U|a) = E[ZU(C,a,f)]. (3.2.4) CEC 4Notation taken from Demski (1980, p. 25). 35 Here, U(C,a,f) represents the utility assessment of the corporate cash outflows (C) that a corporation which has certain attributes (f) will incur if standard (a) is selected. A manager's position on a proposed accounting standard may be given by, 1 (support), if E(U|a') = max E[ZU(C,a,f) D- = CEC 0 (oppose), otherwise. (3.2.5) 3.3 The Model of Participation Choice A manager who has taken a position on a proposed accounting standard may or may not file a comment letter with the FASB. The comment letters filed by corporate managers and others serve as lobbying mechanisms in a political rulemaking process. Downs' model of political behavior may be used to describe the manager’s decision to file a comment letter (the participation choice). Downs (1957) developed a positive economic model to explain political decision making behavior. The model is based on the premise that every individual is a self-interested, utility maximizer.5 Based on Downs’ model, a manager, having taken a position on an accounting policy, will base his participation Choice on three factors in a single period setting. First, a manager may consider the expected marginal effect of the proposed accounting change on his expected utility E[ZU(C,a',f)]. In this case, the contracting and monitoring costs (C), and the corporate 5Utility may be derived from wealth and effort. Downs assumed that an individual prefers more wealth and less effort. 36 attributes (f), which affected the manager’s position Choice on the accounting proposal (a’) may also affect his participation Choice.6 Second, the expected benefits to be obtained by lobbying for or against a proposed accounting standard may be affected by the manager’s perception of his ability to influence the policy outcome (prl). If a manager is not sure that the policy outcome can be influenced by lobbying, he may weight the expected effect on his utility by prl. Finally, a manager may consider the costs of lobbying (E(Ull)) in making his participation Choice. These costs may include the time and effort required to research the accounting proposal and to determine its effect on corporate financial statements, as well as the actual costs of lobbying.7 In a single period setting, the expected utility associated with the lobbying on an accounting policy (E(U|s)) may be represented notationally as follows. E(U|s) = pr1(a’)E(U|a') — E(U|l). (3.3.1) The manager will lobby on the proposal if and only if E(U|s) is greater than zero. Therefore, the manager’s participation Choice (Vi) may be represented as, 1 (lobby), if E(Uls) > 0 (3.3.2) 0 (abstain), otherwise. 6The participation choice model is an additive function of the position Choice model. Therefore, contracting and monitoring costs included in the participation Choice model are hypothesized to be significant only if they are significant in the position Choice model. 7Downs identifies these research and evaluation costs as the costs of becoming well-enough informed to participate in the policymaking process. 37 In extending the model to a multi-period setting, Downs argued that an individual may lobby even if the short-term benefits from lobbying do not exceed the costs. This action is attributed to "long- run participation value" (LRPV). A manager may prefer the existing financial accounting standard-setting process (indexed as j) to an unknown alternative if the costs, in terms of his individual wealth and effort, are expected to be greater under a possible alternative system.8 Further, the effect of LRPV (j) on the utility of a corporate manager (E(U|j) may be weighted by the perceived probability of affecting the continuance of the FASB by lobbying (pr2). In this case, the utility associated with lobbying may be represented as, E(Uls) = pr1E(U|a’) + pr2E(Ulj) - E(U|1). (3.3.3) Finally, the manager's participation Choice may be given by, 1 (lobby), if pr1E(U|a') + pr2E(Ulj) - E(U|l) > 0 0 (abstain), otherwise. (3.3.4) 3.4 Employers' Accounting for Pensions The models developed in Sections 3.2 and 3.3 may be used to describe the position and participation choices of corporate managers on the FASB’s proposed change in pension accounting. Changes in financial statements are asserted to affect corporate contracting and monitoring costs. Therefore, the expected effect of the pension accounting proposal on corporate financial statements is discussed in this section beginning with a brief history of pension accounting 8For example, constituents may fear governmental regulation if the FASB does not continue. 38 standards. The hypothesized relationship between Changes in financial statements and contracting and monitoring costs are discussed in Section 3.5. 3.4.1 A Brief History of Pension Accounting Standards In 1974, the FASB added two projects to its agenda: Accounting and Reporting by Employee Benefit Plans, and Employers’ Accounting for Pensions. These additions were precipitated primarily by the emergence of The Employee Retirement Income Security Act of 1974 (ERISA), which established standards for participation, vesting, and funding of pension plans leading to new, more expansive reporting requirements.9 From 1975 to 1980, the FASB focused primarily on the Employee Benefit Plans project. FASB Statement No. 35: Accounting and Reporting by Defined Benefit Pension Plans, was adopted in March 1980. SFAS 35 addressed financial reporting by the plans rather than the sponsoring employers. Based on the outcome of the benefit plans project, the FASB issued Statement No. 36: Disclosure of Pension Information, in May 1980 as an interim measure on employers' accounting, pending completion of that project. The FASB adopted SFAS 36 to supercede paragraph 46 of APB Opinion 8. The new statement required that an employer disclose the actuarial present value of vested and nonvested accumulated plan benefits, and the plans’ net assets available for benefits, in accordance with SFAS 9See FASB Discussion Memorandum, Employer's Accounting for Pensions and Other Postemployment Benefits (February 19, 1981, p. 118). 39 35.10 Additionally, disclosures were required for the assumed rates of return used in determining the present value of accumulated plan benefits, and the date as of which the benefit information was determined. Since the adoption of SFAS 36, the FASB has issued two Discussion Memoranda (February 1981 and April 1983), a Statement of Preliminary Views (November 1982) and an Exposure Draft on Employers’ Accounting for Pensions (March 1985). The position and participation Choices of corporate managers on the March 1985 Exposure Draft, Employers’ Accounting for Pensions is the focal point of this study.11 3.4.2 Income Statement Effects of the ED APB Opinion 8 allowed a wide range of computational methods to be used in determining pension expense. The March 1985 Exposure Draft (ED) substantially reduced this range of acceptable alternatives in an effort to improve comparability across corporate financial statements. The effect of these required changes was expected to reduce reported pension expense for most companies as shown in Table 3A. This table identifies and summarizes the effect of five components of pension expense, which were individually addressed by the FASB in formulating 10SFAS 35 specifies how the actuarial present value of accumulated benefits and the net assets available for benefits are to be determined, thereby providing a comparable basis for preparing the disclosures required by SFAS 36. 11SFAS 87 was adopted in late December 1985, superceding SFAS 36. The questionnaire, which is discussed in Chapter 4, was mailed in September 1985, before SFAS 87 was adopted. 40 TABLE 3A SUMMARY OF THE ED’S EFFECT ON REPORTED PENSION EXPENSE Component of Effect on Pension Expense Pension Expense Explanation Amortization Decreasea Use of projected unit of normal cost credit method Amortization of Decreaseb Amortization of prior service cost difference between assets available and benefits at time of ED’s adoption Interest cost Increase Use of higher interest rate to compute interest costC Return on Decrease Use of higher interest plan assets rate to compute return on plan assetsd Amortization of Increase in Determination of gains gains and losses Volatility and losses on pension on pension plan plan aFor most companies, since the majority of alternative actuarial methods result in earlier recognition of normal pension costs. bFor most companies since most are expected to report an excess of available plan assets over plan benefits at the time of adoption. CUse of higher interest rate to determine the present value of plan benefits would result in a further reduction in the amortization of normal and prior service costs. dNet effect of the interest cost and return on plan assets is a decrease in pension expense for most companies. 41 the ED.12 Those five components are discussed in this section. (1) The service (normal) cost component of a pension plan includes the actuarial present value of benefits earned by employees for the current year of service. The ED required that normal cost be measured using the projected unit credit method, whereas APB Opinion 8 allowed the use of “any acceptable actuarial cost method".l3 Most actuarial cost alternatives to the projected unit Credit method attribute a greater portion of pension benefits to the earlier years of service, resulting in an earlier recognition of pension expense.14 Therefore, in most cases, a change to the projected unit credit method would result in a reduction in the normal cost component of pension expense. (2) Prior service costs, which arise from amendments to a pension plan, may be amortized using a "systematic and rational method" according to APB Opinion 8. The ED required that unrecognized prior service costs be amortized egually over the future period of service for those plan participants affected by the amendment. In current practice, the interest method, which allocates the smallest amount of 12In fact, the March 1985 Exposure Draft would require separate disclosure of each component. These components may be combined under APB Opinion 8. 13The projected unit credit cost allocation method is based on an estimate of future retirement benefits that must be paid by the employer to each employee. The FASB's projected unit credit method requires that the estimate include each employee's expected future salary increases. See FASB Discussion Memorandum, Employers’ Accounting for Pensions and Other Postemployment Benefits, dated February 19, 1981 (Chapter 4 and Appendix E) for further discussion of actuarial cost methods. 14Many companies currently use the entry age normal actuarial cost method, which results in early recognition of pension expense. (See Accounting for Pension Costs, 1977) 42 the prior service costs to the years immediately following an amendment, predominates.15 Therefore, the proposed Change would require an increase in the amortization of existing unrecognized prior service costs for most companies. However, the more significant effect of the ED on prior service costs would result from the amortization of unrecognized net obligations or assets at the date of the ED's adoption.16 Companies reporting an excess of available plan assets over projected pension benefits (overfunded plans) would experience a reduction in pension expense due to the required amortization of net assets while companies with underfunded plans would experience an increase in pension expense. Since most companies, in their 1985 year-end financial statements, reported overfunded plans, the overall effect of the ED on prior service costs would be to decrease reported pension expense.17 (3) The interest cost component represents the increase in the projected benefit obligation due to passage of time. Because the pension component of their compensation is deferred until after retirement, the employees have, in effect, provided a loan to their employer. The interest on that "loan" is treated as a separate expense to the company. 15See March 1985 Exposure Draft, Employers' Accounting for Pensions (Summary of Major Changes). 161m a letter to Clients regarding FASB Statement 87, Arthur Andersen & Co. noted that the funding status of pension plans at the time of adoption will significantly affect reported pension expense due to the required amortization of unfunded benefits or excess assets. 17Only 4% of the 300 randomly selected NYSE firms included in this study reported an underfunded pension plan status. 43 Interest cost is computed by multiplying the assumed discount (settlement) rate by the plan obligation at the beginning of the year.18 According to the ED, the discount rate would be estimated by reference to rates implicit in current prices of annuity contracts, or rates of return on high-quality, fixed-income investments. This rate may be much different from the interest rates commonly used now. In early April 1985, for example, the long-term yield on U.S. Treasury Securities was approximately 11.75 percent, versus an average 8 to 9 percent rate used by companies for pension disclosure purposes. For most companies, the Change in discount rates would result in a decrease in the normal and prior service cost components of pension expense and an increase in the interest cost component. (4) The return on plan assets represents an offsetting revenue component in the determination of pension expense. The ED required that this component be based on assumed rates of return equal to the discount rates used to measure the interest cost component.19 The net effect of the interest cost and return on plan assets components is to require that a company record interest on the excess of plan assets overgaccumulated benefits and vice versa. APB Opinion 8 requires only that a net interest component be included in the determination of pension expense to the extent that annual pension expense provisions and fundings are not equal. For companies that currently record pension expense equal to the amount funded, this Change would result 18The discount rate used to determine interest cost is the same rate used to determine the present value of normal and prior service costs. 19Under SFAS 87, the rates of return on plan assets may not equal the discount rate used to record interest cost and the present value of pension benefits. 44 in increased pension expense when the pension plan is underfunded and decreased pension expense when the plan is overfunded.20 Again, since most companies report overfunded pension plans, the net effect of the interest cost and return on plan assets component should be to reduce reported pension expense. (5) A gain or loss on the pension plan may be realized due to a Change in value of either the projected benefit obligation or the plan assets. Either of these Changes can result from experience different from that assumed, or from a change in an actuarial assumption. The ED required that unrecognized gain or loss be amortized over the average remaining service life of the plan participants to the extent that the gain or loss exceeds ten percent of the greater of the projected benefit obligation or the fair value of plan assets. APB Opinion 8 required only that unrecognized gains or losses be amortized using an "acceptable systematic method" over a suggested period of 10 to 20 years. In practice, most companies use a spreading or averaging method which results in a straight-line amortization of the actuarial gain or 1033.21 The method required by the ED would approximate a straight- line amortization if the average remaining service life of plan participants remains relatively constant. The effect of the ED on pension expense would vary among the firms. 20A company reporting an excess of available plan assets (at current fair market value) over accumulated plan benefits (computed using the unit credit method) is overfunded. A company reporting an excess of accumulate plan benefits over available plan assets is underfunded. 21See March 1985 Exposure Draft, Employers' Accounting for Pensions (Summary of Major Changes). 45 However, comment letter filers on the ED expressed greater concern over the determination of pension plan gains and losses than over the method of amortization. Comment letter filers argued that, by requiring the use of a current market rate to value pension plan assets and to discount pension obligations, the FASB was introducing unnecessary volatility in reported pension expense. Although a ten percent "corridor" approach was proposed to reduce this volatility, most companies may be expected to experience increased volatility in the pension plan gain/loss component of reported pension expense. 3.4.3 Balance Sheet Effects As under APB Opinion 8, the ED required that accrued or prepaid pension cost be recognized on the balance sheet to the extent that actual funding does not equal reported pension expense. However, the ED further required that a liability be recorded when accumulated benefits (determined without the impact of future compensation levels) exceed the fair market value of plan assets as of the date of the financial statements.22 Alternatively, an asset would not be recognized when the fair market value of plan assets exceeds the accumulated benefit obligations. Therefore, the ED would require a larger liability for corporations which have underfunded pension plans, but would have no effect on the liabilities of most corporations which have overfunded pension plans. 22An offsetting intangible asset would be recorded equal to prior service costs. 46 3.4.4 Disclosure Requirements The ED required an employer sponsoring a defined benefit pension plan to disclose in footnotes: 1) a description of the pension plan including type of benefit formula, funding policy, and other significant matters; 2) the components of pension expense and ratio of that expense to covered payroll; 3) a statement of Changes in the fair value of plan assets showing contributions, benefits paid and actual return on assets; 4) a schedule showing plan assets, projected benefit obligations, unrecognized prior service costs, and unrecognized actuarial gains and losses; and 5) the assumed discount rate and salary assumptions, and the effect of a one-percentage-point Change in each on the projected benefit obligations and reported pension expense. These footnote disclosures required by the ED reflect a significant increase over the existing requirements of SFAS 36.23 SFAS 36 requires disclosure only of accumulated pension benefits, the fair market value of assets, and the discount rate used to determine the present value of accumulated pension benefits. 23Approximately 94% of the managers filing comments on the ED indicated that the footnote disclosures required by the ED were excessive. 47 3.4.5 Summary APB Opinion 8 and SFAS No. 36 provided the guidelines on employer's accounting for pensions at the time that the March 1985 Exposure Draft (ED) was issued. The ED reduced the set of acceptable methods for measuring and reporting pension expenses and liabilities, and increased the pension disclosure requirements. Under the ED, a company with an overfunded pension plan(s) may be expected to report an overall decrease in pension expense. Alternatively, a company with an underfunded pension plan(s) may be expected to report an overall increase in pension expense and would be required to report a pension liability to the extent that accumulated pension benefits exceed the fair market value of available plan assets. The expected changes in pension expense would reflect a decrease in the required amortization of normal and prior service costs, and in the net interest cost/return on assets components of pension expense. Finally, g1; companies may be expected to experience greater volatility in reported pension expense and would be required to increase their disclosures. The volatility in pension expense would reflect the required method of amortizing gains and losses on pension plans. (See Appendix A for an illustration of the Income Statement, Balance Sheet and Disclosure Requirements under APB Opinion 8 and SFAS No. 36 and under the ED.) 48 3.5 Contracting and Monitoring Costs Associated with the ED The potential effect of the ED on corporate financial statements was discussed in Section 3.4. These effects may, in turn, have economic consequences for companies that must change their accounting for pensions to comply with the new standard. Changes in reported pension expense, pension liability, and footnote disclosures may result in real cash outflows (costs) for an individual company. These costs (hereafter, contracting and monitoring costs) may be incurred as a result of explicit and implicit contractual obligations and public interference. Specifically, costs may be incurred due to the existence of debt covenant, management compensation and union agreements. In addition, political lobbying and legal costs may be incurred to avoid the interference of government regulatory agencies and corporate competitors and consumers who use published financial statements to monitor manager activities. Finally, internal recordkeeping, actuarial and auditing costs may be incurred in order to comply with the accounting Change. These potential costs are described in Sections 3.5.1 through 3.5.6 for companies reporting Changes in pension expense, liabilities and footnote disclosures. 3.5.1 Reporting Reduced Pension Expense A company reporting reduced pension expense may expect to incur certain contracting and monitoring costs or benefits. The relative magnitude of these costs and benefits may depend on the labor 49 intensity, size, and leverage position (corporate attributes) of the company. For example, a company that is in a labor intensive industry may experience increased labor costs as a consequence of reporting reduced pension expense. These increases in labor costs may come in two forms. First, the company may incur education and labor negotiation costs in explaining to employees that the reduction in pension expense reflects only a Change in accounting, and not a real change in "pension cost". Secondly, a reduction in reported pension expense may encourage labor representatives to demand greater benefits. Ultimately, the company may have to comply with labor demands for increased pension benefits. Second, Watts and Zimmerman (1978) argued that the manager of a large, profitable company will have incentives to avoid reporting increased net income due to concern over government interference, increased competition, and anti-trust suits. A company required to report decreased pension expense and increased net income may incur public relations and political lobbying costs to avoid government interference. Further, a company reporting increased net income may experience a real decrease in profits due to the entry of additional competitors into the industry. Finally, that company may incur legal costs to fight anti-trust suits.24 Third, a decrease in pension expense and the concurrent increase in net income may have a favorable impact on a company's accumulated earnings and, in turn, its debt/equity ratio.25 Debt covenants 24See Watts and Zimmerman (1978) for further discussion. 25Accumulated earnings will increase only if an amount equal to the increase in reported net income is not paid out to shareholders in the form of dividends. 50 generally include restrictions on a company's debt/equity ratio.26 Typically, these covenants require a company to maintain a debt/equity ratio which is below some maximum value. A highly leveraged company is more likely to violate a debt/equity restriction than a company that is not. Therefore, a highly leveraged company may be more likely to benefit by reporting decreased pension expense. Such a company may be less likely than otherwise to incur debt renegotiation costs on existing debt agreements due to future, unrelated decreases in reported income. Further, a decreased debt/equity ratio associated with the decrease in reported pension expense may result in lower interest rates on future debt agreements. Finally, if management compensation is based, in part, on reported net income, a company may experience increased management compensation costs when reported pension expense is reduced. The effect of the proposed Change in pension accounting on management compensation is determined by (1) the significance of the income related bonus plan to the total management compensation package, (2) flexibility of the bonus plan formula to Changes in accounting methods, and (3) upper and lower income bounds in the bonus plan formula.27 Therefore, the effect of the ED on an individual company's management compensation costs will depend on the detail and flexibility of the bonus plan.28 26See Fogelson (1978). 27See Healy (1985) for further discussion of management compensation plans. 28The details of most companies’ management bonus plans are published in proxy materials filed with the SEC. However, the cost of obtaining this information is high. Therefore, a corporate attribute was not hypothesized to proxy for management compensation costs. 51 To summarize, as a consequence of the ED, a company reporting decreased pension expense may incur increased labor, political (lobby and legal), and management compensation costs. However, this same company may experience decreased debt costs. The magnitude of these Changes may depend on the company’s labor intensity, size and leverage position. 3.5.2 Reporting Increased Pension Expense A company reporting increased pension expense may incur costs and benefits opposite to those incurred by a company reporting decreased pension expense. For example, a labor intensive company may incur reduced labor costs consequent to an increase in reported pension expense. In this case, the company may not have incentive to educate employees regarding the accounting Change because the increase in reported pension expense could be used to negotiate a more favorable labor contract for the company at a future date. A large, profitable company reporting increased pension expense and lower income may also benefit from reduced political costs in the long-run. By reporting reduced income, such a company may reduce the threat of anti-trust suits, government interference and increased competition. All other things equal, an increase in reported pension expense may negatively affect both future and existing debt costs because of a reduction in accumulated earnings and total equity (and a concurrent increase in the debt/equity ratio). Future interest rates may be adversely affected due to an increase in debt/equity ratios. 52 Furthermore, existing debt covenants of already highly leveraged companies may be violated and lenders may take advantage of technical violations to raise interest rates on Current debt agreements.29 Finally, a company may experience a decrease in management compensation costs due to an increase in reported pension expense. These costs may decrease if management compensation is based, in part, on reported net income. To summarize, a company reporting increased pension expense due to the ED's adoption may experience a reduction in labor, political and management compensation costs. However, that same company may also incur additional debt costs. 3.5.3 Recording a Liability for Unfunded Pension Obligations Under the ED, a company with an underfunded pension plan would be required to record a liability reflecting the excess of accumulated pension obligations over pension plan assets. The recording of this liability may or may not increase corporate debt costs depending on how creditors calculate debt/equity ratios. The disclosure of accumulated pension benefits and plan assets is already required by SFAS 36. Therefore, the potential effect of the ED is only to move net unfunded pension obligations from a footnote to the balance sheet. If lenders consider footnote information when monitoring borrowers, the movement of excess pension obligations from a footnote to the balance sheet will not affect existing or future debt agreements. 29See Aristar (1976). 53 However, Aristar (1976) provided examples of cases where companies were forced into technical default under long-term debt agreements because of mandated changes in accounting principles. In these cases, the lender used the situation to renegotiate the loan with a more favorable, higher interest rate.30 Given Aristar’s findings, the manager of a firm which is near restrictive debt covenant margins, may be concerned with technical default on those agreements. Therefore, that manager may expect to incur increased debt costs if required to record a liability for the excess of accumulated pension benefits over pension assets. 3.5.4 Volatility in Reported Pension Expense and Liabilities As discussed earlier, Changes in reported income and liabilities may affect a company’s debt/equity ratio. Therefore, a manager may be concerned with future fluctuations in reported pension expense and liabilities which are associated with the ED. Specifically, a manager may expect to incur increased debt costs due to volatility in reported net income and liabilities. The initial effect of the ED on the level of reported pension expense may be positive for most companies in terms of debt costs.31 However, future Changes in capital market conditions may lead to 30The inclusion of an offsetting intangible asset in the balance sheet may negate the technical default argument provided by Aristor. However Smith and Warner (1979) found that lenders, in computing ckflnjequity ratios, automatically reduce equity by the amount recorded as intangible assets. 31As discussed in Section 3.4.2, most companies may be expected to report reduced pension expense which is associated with reduced debt costs. 54 volatility in reported pension expense and liabilities, as discussed in Section 3.4.2. In this case, a manager may expect to incur an increase in debt costs due to future increases in reported pension expense and/or liabilities caused by fluctuations in discount rates or return on assets . 3.5.5 Increased Disclosure Requirements A company that must change accounting and actuarial cost methods, or increase disclosures, may incur increased administrative compliance costs to comply with the ED. These costs may include information acquisition (actuarial and internal recordkeeping costs), and financial reporting costs (auditing and financial statement preparation costs). In the complex area of pension accounting, these costs (administrative compliance costs, hereafter) may be large. A large company which has many employees, is likely to have more than one pension plan. Under APB 8 and SFAS 36, the assets and obligations of these pension plans may be combined. By contrast, the ED would require that separate disclosures be made for each plan. Therefore, a large company which has more than one pension plan may be expected to incur higher administrative compliance costs than a smaller company due to the ED’s adoption. 3.5.6 Summary A company that would be required to Change its accounting for pensions to comply with the ED may experience Changes in contracting 55 and monitoring costs. In general, Changes in reported pension expense are expected to be inversely related to political and labor costs and positively related to debt and management compensation costs. A Change in reported pension liabilities or a Change in the volatility of reported pension expense is expected to be positively related to debt costs. Finally, the increased disclosure requirements and the restrictive requirements of the ED are associated with increased administrative compliance costs. The ED's impact on labor, political, debt, management compensation and administrative compliance costs may result from its effect on corporate financial statements. Furthermore, these costs may be higher for a labor intensive, large, or highly leveraged firm. The hypothesized relationships between these expected costs, corporate attributes, and manager position and participation Choice on the ED are described in the following section. 3.6 The Hypotheses 3.6.1 Position Choice A general model of manager position choice on a proposed accounting standard was offered in Section 3.2. This model was based on the assertion that a manager’s utility may be affected by a mandated Change in financial reporting. Specifically, a manager’s utility may decrease when corporate cash ouflows increase due to mandated Changes in financial reporting. The general model of manager 56 position Choice presented in equation 3.2.4 is repeated here. 1 (support), if E(Ula') = E[ZU(C,a,f]. D'= CEC 0 (oppose), otherwise. (3.2.4) The expected corporate costs associated with the ED were described in Section 3.5. These included political (C1), labor (C2), debt (C3), administrative compliance (C4), and management compensation costs (C5). The corporate attributes which were expected to affect the magnitude of these contracting and monitoring costs were also discussed in Section 3.5. Firm size (fl), labor intensity (f2), and leverage position (f3) were expected to affect the magnitude of political and administrative compliance, labor, and debt costs, respectively. The relationships between expected contracting and monitoring costs and a manager’s position choice are hypothesized below. One overall hypothesis (H1.0) on manager position choice, and eight subhypotheses are offered. H1.0‘ The model set forth in equation 3.2.4 may be used to describe a manager's position Choice on the ED such that Di = fn(Cl,C2,C3,C4,Cs,f1,f2,f3). Hypotheses 1.1 through 1.5, as illustrated in Figure 3A, address the individual relationships between expected changes in contracting and monitoring costs and a manager’s position Choice. A manager who expects to incur large increases in these costs as a result of the ED's adoption, is hypothesized to expect a reduction in his utility. 57 Expected Change in CorporateCosts Political H1.1 + Labor H1.2 + Effecton +_ Effecton 4_ Manager theFdnanCial ,r Debt Manager Position Statements ”1.3 Utility Choice 4- Administrative Compliance + “1.4 + Management Compensation H1.5 FIGURE 3A THE RELATIONSHIP BETWEEN CHANGES IN CORPORATE COSTS ANDZXMANAGER'SPOSITIONCHOICE 58 Further, a manager expecting a reduction in utility is expected to oppose the ED. Therefore it is hypothesized that: H1.1: A manager's position Choice (Di) is inversely related to the expected Change in political costs (Cl). H1.2: A manager’s position Choice (Di) is inversely related to the expected change in labor costs (C2). H1 3: A manager’s position choice (Di) is inversely related to the expected Change in debt costs (C3). H1 4: A manager's position Choice (Di) is inversely related to the expected change in administrative compliance costs (c4). Alternatively, a manager who anticipates higher corporate management compensation costs may expect to experience an increase in personal wealth which more than offsets the negative effect of corporate cash outflows on his utility. Further, a manager who expects a net increase in his utility is hypothesized to support the ED. H1.5: A manager’s position Choice (Di) is positively related to the expected Change in management compensation costs (c5). Hypotheses 1.6 through 1.8, as illustrated in Figure 3B, address the individual relationships between the ED’s effect on reported pension expense (f4), corporate attributes, and a manager's position 59 Effect of ED Corporate onFinanCial ExpectedChangein Attributes Statements CorporateCosts A Restrictions& 4' Administrative / Disclosures ComplianceCosts / I Firnmsize / Changein Manager Utility +_ Manager ) H1.6 ‘~ \ ‘\ \ ‘\ Changein ._ PensionExpense PoliticalCosts Changeifi Management PensionExpense CompCosts Labor Changein Intensity r‘-1 PensionExpense “ LaborCosts H1.7 Leverage Changeixi 4_ Position r~-1 PensionExpense Dethosts Hl.8 andldabilities FIGURE3B THE RELATIONSHIP BETWEEN CORPORATE ATTRIBUTES, CORPORATE COSTS AND A MANAGER' 8 POSITION CHOICE Position Choice 60 Choice. A company reporting a decrease in pension expense is expected to experience an increase in political and labor costs. The same company is expected to experience a decrease in debt costs. Conversely, a company reporting an increase in pension expense is expected to experience a decrease in political and labor costs and an increase in debt costs.32 Further, a company’s size, labor intensity, and leverage position may affect the magnitude of the corporate costs incurred due to the ED's adoption. The manager of a large company is expected to face high political and administrative compliance costs when a decrease in pension expense is reported. The manager of a labor intensive company is expected to face high labor costs when a decrease in pension expense is reported. Finally, the manager of a highly leveraged firm is expected to face high debt costs when an increase in pension expense is reported.33 Therefore, if the ED requires a decrease (-) in pension expense, the manager of a large or labor intensive firm may expect to incur increased (+) corporate costs. That manager is hypothesized to oppose (-) the ED. Alternatively, if the ED requires an increase (+) in pension expense, the manager of a highly leveraged firm may expect to incur increased (+) corporate costs. That manager is hypothesized to oppose (-) the ED. 32Increased debt costs are also associated with an increase in reported liabilities. A company reporting increased liabilities due to the ED's adoption, is also expected to report increased pension expense. 33A firm’s size, labor intensity, and leverage may also affect the probability of incurring these costs. 61 “1.6: A manager’s position Choice (Di) is positively related to the interactive effect of firm size (f1) and the ED's effect on reported pension expense (f4). “1.7: A manager’s position Choice (Di) is positively related to the interactive effect of labor intensity (f2) and the ED’s effect on reported pension expense (f4). “1.8: A manager's position choice (Di) is inversely related to the interactive effect of leverage position (f3) and the ED's effect on reported pension expense (f4). 3.6.2 Participation Choice A general model of manager participation choice in the accounting standard-setting process was set forth in Section 3.3. Like the position choice model, this model was based on the assertion that a manager's utility may be affected by a mandated change in financial reporting. Specifically, a manager's utility may decrease when corporate cash ouflows increase due to financial reporting changes. Therefore, a manager may expect to benefit by lobbying if he believes he can influence the outcome of a proposed standard (a') which affects corporate cash outflows. The manager's perception of the probability of influencing the policy outcome (prl) may affect his decision to participate in the standard-setting process. A manager's utility from lobbying may also be affected by his perceptions of the costs of that action (1). Although the actual 62 costs of lobbying may be low, costs of research and evaluation may be high. Finally, a manager may expect to obtain long—run utility benefits (pr2(j)) from lobbying. In this case, the manager may believe that the current standard-setting process is preferable to some unknown alternative in terms of corporate costs and manager effort (j). Further, he may believe that by lobbying the FASB, he may affect its continuance (pr2). The general model of manager position Choice presented in equation 3.3.2 is repeated here. 1 (lobby), if E(U|a')pr1(a') + E(Ulj)pr2(j) - E(U|1) > 0 0 (abstain), otherwise. (3.3.2) The relationships between contracting and monitoring costs, corporate attributes, manager perceptions of the FASB, and a manager's participation Choice are illustrated in Figure 3C. The hypotheses used to test these relationships are described below. One overall hypothesis (H2.0) and eight subhypotheses are offered. H2.0: The model described in equation (3.3.2) may be used to describe a manager's participation Choice on the ED. The manager of a company, whose expected utility is affected by the ED through Changes in corporate contracting and monitoring costs, may have incentive to lobby on the ED. As these costs increase, the manager’s incentive to lobby against the ED is hypothesized to increase. Alternatively, if the costs associated with the ED are low (relative to the costs associated with possible alternatives), a manager's incentive to lobby for the ED is hypothesized to increase. Corporate Attributes Firm Size * ED H2.2 Labor *ED Intensity H2.3 ExpectedChanges inContractingand MonitoringCosts H2.1 Leverage‘*ED H2.4 ManagerPerceptions Probabilityof InfluencingFASB H2.5 Desire to Support the FASB H2.6.2.7 Lobby Costs H2.8 63 FIGURE3C Expected Changein Manager Utility Manager's Participation Choice THERELATIONSHIPSBETWEENCORPORATEATTRIBUTES, MANAGERCOSTEXPECTATIONS,PERCEPTIONSCHPTHEFASB, AND A MANAGER' 8 PARTICIPATION CHOICEa aThe relationships between the magnitude of expected Changes in contracting and monitoring costs and the expected Change in manager utility are shown for an opposing manager. An opposite relationship is hypothesized for a supporting manager. 64 Hypotheses 2.1 through 2.4 pertain to a manager who opposes the ED. The opposite relationships are hypothesized for a manager who supports the ED. H2.1‘ An opposing manager’s participation Choice (Vi) is positively related to the expected Change in corporate contracting, monitoring and other costs associated with the ED (C1, c2, C3, C4, c5). As in the position Choice hypotheses, it is asserted that the ED’s impact on the financial statements will affect the magnitude of expected corporate costs. Further, firm size, labor intensity, and leverage are expected to directly affect the magnitude of political and administrative compliance, labor, and debt costs, respectively. Therefore, if the ED requires an increase (+) in pension expense, the manager of a large or labor intensive firm may expect to incur increased (+) corporate costs. It is hypothesized that such a manager will be more likely to lobby (+) as these corporate costs increase. Alternatively, if the ED requires a decrease (-) in pension expense, the manager of a highly leveraged firm, facing higher debt costs, will be more likely to lobby (+) as his expected debt costs increase. H2_2: An opposing manager’s participation Choice (Vi) is positively related to the interactive effect of firm size (f1) and the ED's effect on reported pension expense (f4). 65 H2 3: An opposing manager’s participation choice (Vi) is positively related to the interactive effect labor intensity (f2)and the ED's effect on reported pension expense (f4). H2 4: An opposing manager’s participation choice (Vi) is inversely related to the interactive effect of leverage position (f3) and the ED’s effect on reported pension expense (f4). A manager who does not believe that he can influence the outcome on the ED may not lobby even if he expects the ED to have a significant effect on his utility. Therefore, it is hypothesized that: H2 5: A manager's participation Choice (Vi) is positively related to his perception of the probability of influencing the policy outcome (prl). Alternatively, a manager who believes that the continuance of the FASB would be beneficial to him in the long run may lobby even if he does not expect the ED to have a short-run effect on his utility. A manager who supports the FASB and believes that its continuance is affected by filing a comment letter is more likely to file than a manager who either does not support the FASB or does not believe that filing a comment letter provides support for the FASB. Therefore, it is hypothesized that: 66 H2.6: A manager’s participation choice (Vi) is positively related to his expectations of the benefits obtained (costs avoided) from continuance of the existing standard-setting body (j). H2.7: A manager’s participation choice (Vi) is positively related to his or her perception of the probability of affecting the FASB’s continuance by filing a comment letter (pr2) Finally, it is hypothesized that a manager is less likely to lobby on the ED if he perceives the costs of filing a comment letter to be high. “2.8: A manager’s participation Choice (Vi) is inversely related to the costs of lobbying (l). 3.7 Summary A corporate manager's position and participation Choice models on the pension accounting issue are derived from general decision theory and Downs’ model of political behavior. The manager’s utility is hypothesized to be affected by Changes in corporate contracting and monitoring costs. These costs are associated with Changes in financial statement reporting required by the March 1985 Exposure Draft, Employer’s Accounting for Pensions. The manager's position Choice is hypothesized to be related to expected Changes in corporate political, labor, debt, administrative compliance and management compensation costs. These costs are 67 asserted to be affected by the ED's impact on corporate financial statements. Further, certain attributes of a corporation are expected to affect the magnitude of these costs. Therefore, corporate attributes are offered as proxies for the contracting and monitoring costs. Firm size is offered as a proxy for political and administrative compliance costs. Labor intensity is offered as a proxy for labor costs. Leverage position is offered as a proxy for debt costs. The manager's participation Choice is hypothesized to be related to: 1) the magnitude of the expected contracting and monitoring costs which affected his position choice; 2) his perception of the probability of influencing the ED’s outcome by lobbying; 3) his desire to support the FASB; 4) his perception of the probability of affecting the FASB’s continuance by lobbying; and 5) the costs of lobbying. The research methodology used in this study to test the models of manager position and participation Choice is discussed in the next chapter. The sample design, measurement methods, and choice of statistical methods are explained therein. CHAPTER 4 METHODOLOGY 4.1 Introduction The specifics of the March 1985 Exposure Draft, Employers’ Accounting for Pensions (ED), and the potential effect of that proposal on corporate financial statements and corporate manager utility were discussed in the previous Chapter. In addition, theoretical models of corporate manager position and participation choice on the pension accounting issue were offered. The research methodology used to test these models is discussed in the present Chapter. First, the population is defined and the sample design is explained. Second, the proxies chosen to measure the dependent and explanatory variables in the position and participation Choice models are discussed. The sources used to obtain these measures, which include a questionnaire, are described. Further, the benefits of using a questionnaire and its limitations are addressed. Finally, the Choice of statistical methods used to analyze the models of manager position and participation choice is discussed. The advantages of a probit analysis over discriminant analysis are discussed as they relate to the data used in this study. 68 69 4.2 Population Definition and Sample Selection Mandated Changes in financial statement accounting may affect all preparers of corporate financial statements who must comply with Generally Accepted Accounting Principles (GAAP). Therefore, an ideal study of manager behavior in the accounting standard-setting process would include a random sample of all corporations required to comply with GAAP. However, only a small number of corporate managers participate in the accounting standard-setting process by filing a comment letter.1 In order to study both position and participation Choices, the requisite sample must include (1) a sufficient number of comment filers so that statistical inferences may be made on the position choice of filers, and (2) a sufficient number of both comment filers gpg nonfilers so that statistical inferences may be made on participation Choice. Therefore, the sample used in this study should include a representative number of corporate managers who filed comment letters. There are at least two ways in which a sample could be drawn which would include a representative number of filers. In prior studies of participation Choice, researchers have selected two separate samples.2 The first sample included all comment letter filers on the accounting issue which was being addressed. The second 1For example, only approximately 1% of the companies listed on the NYSE, AMEX and OTC filed comment letters on the pension Exposure Draft. 2See Zmijewski and Hagerman (1981), Lasater (1982), Griffin (1982, 1983), Kelly (1982, 1985). 70 sample included a random selection of companies listed with some report service (i.e., COMPUSTAT). Because these studies used two samples, the statistical inferences from analyses of participation and position Choices were not generalizable to any general population. An alternative approach would be to select one random sample from a subpopulation which includes a greater proportion of comment letter filers than the total population includes. The single sample approach would provide generalizable evidence on the participation and position choices of the subpopulation taken as a whole. This study used a single sample of 300 firms selected randomly from the population of firms listed on the NYSE. Fifteen percent of the managers of NYSE firms filed comment letters on the ED. By restricting the sample frame to NYSE firms, a random sample could be selected which included a representative number of comment letter filers and nonfilers. A disadvantage of the restricted sample frame is that no conclusions can be drawn regarding the behavior of managers of American Stock Exchange, Over-the-Counter, or privately held companies. However, there is enough variability in the attributes of the NYSE firms and in the behavior of NYSE firm managers, to make statistical statements about the models of position and participation Choice. 4.3 Variable Measurement A manager’s position and participation Choice (the dependent variables) on the ED were hypothesized in Chapter 3 to be related to 71 three sets of explanatory variables. Manager expectations of conppgcting and monitoring costs incurred due to the ED’s adoption were hypothesized to affect the manager’s utility and, in turn, his position and participation Choices. Further, corporate attributes were asserted to be related to the magnitude of these expected contracting and monitoring costs. Therefore, it was hypothesized that corporate attributes would affect a manager’s position and participation Choices on the ED. Finally, it was hypothesized that a mgpgger's perceptions of the FASB would affect his participation Choice. 4.3.1 Dependent Variables The dependent variables in this study are the corporate manager's position and participation choices on the ED. The methods used to measure these variables are described in this section. The participation choice of a manager is a strictly binary Choice. A manager either files a comment letter or he does not. Further, a list of firms represented by comment letters on an accounting pronouncement is publicly available through the FASB. Therefore, a true measure of manager participation Choice was obtained for this study by examining the FASB's list of comment filers on the ED. Ideally, a manager's position Choice on an accounting pronouncement would be measured on a continuous scale. The degree to which the manager opposes or supports a pronouncement would be determined. The best available measure of manager position Choice on 72 an accounting pronouncement is provided by comment letters filed with the FASB. However, as discussed in Section 4.2, only a small percentage of corporate managers actually file comment letters. Further, the comment letters are subject to varied interpretations. Prior researchers have relied on comment letters to measure manager position choices on accounting pronouncements. To reduce the difficulties of interpretation, manager position Choice has been measured on a binary (oppose/support) scale. The binary scale was used in this study to measure manager position choice. Comment letters were examined to determine the position Choice of filing managers. Further, a questionnaire (see Appendix B) was used to obtain information about the position Choices of corporate managers who did not file comment letters on the ED. Managers were asked about five subissues of the ED. These subissues included standardization of actuarial cost methods, balance sheet presentation of pension liabilities, determination of discount rates, valuation of plan assets, and footnote disclosures. The manager's overall position on the ED was coded as "support" if the manager supported at least three of these issues, and "oppose", otherwise.3 By determining the position choices of nonfilers, this study included a portion of the corporate manager population which was not included in prior research efforts. However, manager position Choices obtained from a questionnaire may be subject to weaknesses in reliability. If questions are not Clearly stated, the measures obtained may not represent the managers' true positions. Further, 3This method of scoring resulted in an equal weighting of each subissue. 73 managers may not have incentives to state their true positions on the ED in the questionnaire.4 Validating measures of manager position Choices stated on the questionnaire were obtained by examining comment letters filed with the FASB. A manager's overall position on the ED was determined from general statements made in the comment letters. Additionally, the comment letters were coded using the questionnaire's format. The McNemar test for the significance of Changes in related samples was used to assess the validity of the position Choices obtained from the questionnaire.5 The test was used to compare overall position Choices stated in comment letters, with the positions computed from equal weightings of the five subissues addressed in the questionnaire. In addition, positions stated in the questionnaire on those five subissues were compared with subissue positions stated in the comment letters. (See discussion p. 120 and Table 5D, p. 121.) 4.3.2 Independent Vapiables: Manager Cost Expectations Manager position and participation choices were hypothesized to be related to manager expectations of Changes in contracting and monitoring costs (manager cost expectations). Political, labor, debt, administrative compliance and management compensation costs were identified as costs which may be incurred due to the ED’s adoption. 4Managers may have stronger incentive to state their true position in a comment letter filed with the FASB. 5The McNemar test for the significance of Changes may be used to COmpare related samples when the measurement is at the nominal level. See Siegel (1956, p. 63). 74 Measures of manager cost expectations have not been obtained in prior research efforts. Instead, corporate attributes have been used to proxy for expected contracting, monitoring and other costs. The use of corporate attribute proxies may reflect the difficulty associated with obtaining a true measure of manager cost expectations. Corporate managers have not addressed the costs associated with an accounting change in the comment letters they file with the FASB. In fact, it may be argued that managers are unable or unwilling to provide reliable estimates of these costs. In either of these cases, it would pp; be worthwhile to ask the manager to provide estimates of the corporate costs he would expect to incur if an accounting change were adopted. Alternatively, managers may have estimated these costs, but they may not have been provided with a forum in which to state them.6 In this case, a questionnaire may provide managers with that forum. Therefore, a questionnaire was used in this study to obtain measures for manager expectations of the corporate costs associated with the ED’s adoption. Managers were asked to estimate both the out-of—pocket and opportunity costs which may be incurred to initiate compliance with the ED.7 The development of the questionnaire is discussed below. The measures used for each of the five costs identified in Chapter 3 (political, labor, debt, administrative compliance, and 6This assumes that the comment letter is not seen by managers as a forum by which to state the costs associated with a proposed accounting Change. 7Long-run costs were not addressed due to measurement difficulties associated with estimation and discounting. 75 management compensation) are summarized in Table 4A, and discussed in this section. Political and labor costs were combined as public relations costs on the questionnaire. Managers were asked to estimate the costs they would incur to explain the effects of the proposed Change in pension accounting to outside investors, and others (i.e., government regulatory bodies), as well as to employees. In the pretest, these costs were categorized separately on the questionnaire in terms of investors, creditors, employees, and others.8 However, one respondent to the pretest indicated that his company could not differentiate between the costs associated with the different groups. He noted that these costs may pertain to general publications and seminars which would not be directed at any specific group. Therefore, separate measures were not obtained, and the measure for public relations costs included a combination of hypothesized political (investors and others) and labor (employees) costs. Debt costs relating to changes in interest rates on existing and future debt agreements were included in the questionnaire. Managers were asked to estimate Changes in interest expense that may be incurred due to violations of existing indenture requirements. Additionally, managers were asked to estimate the marginal effect of the ED on interest expense incurred on future debt issuances. Administrative costs of possible debt restructuring were not specifically addressed. Respondents to the pretest indicated that 8A discussion of the questionnaire development is provided in Section 4.4. 76 TABLE 4A CORPORATE COSTS ASSOCIATED WITH THE PENSION EXPOSURE DRAFT Public relations costs Political and labor costs associated with monitoring activities of governmental agencies, investors, employees and others. Debt costs Costs associated with monitoring activities of creditors, including Changes in interest rates on existing and future debt agreements. Administrative compliance costs Costs associated with Changes in internal reporting, and with increases in actuarial and audit fees, as well as the costs associated with explaining the new reporting requirements to internal management. Management compensation COStS Costs associated with contractual compensation agreements with managers. er- ,a. (I) "C we .. “Are we. “A. V34. o #5,... VVA.. AA”, Vsl .4 . “A! v“ C \-CO‘ d a“; .‘A.-“ 77 these costs would be negligible when compared to the interest expense effect. Administrative compliance costs were separated into two categories. Administrative costs were described as actuarial and audit fees, the costs of computer installations, and other information processing costs. Internal management costs were added at the suggestion of the pretest respondents who were asked, "What are the most significant concerns that your company has regarding the current pension accounting proposal?" Both respondents to the pretest indicated that upper managements’ understanding of the Changes and the potential breakdowns in internal communications were their greatest concerns. They were not as concerned about the bookkeeping costs of compliance or the ED’s effect on the financial statements. Their concern with internal management costs was based on the complexity of pension accounting and the general lack of understanding by most people, including top management, of the actuarial methods and assumptions underlying the determination of pension obligations and funding. Finally, management compensation costs were measured in terms of percentage Change. Managers were asked if the compensation of upper level management would decrease if reported income decreased "as a result of a change in pension accounting" and, if so, by how much.9 9Managers were not asked if their compensation would increase if reported income increased. Therefore, thisrepresents a weakness in the measure. This limitation is discussed further in Chapter 7. 78 4.3.3 Independent Variables: Corporate Attributes In Chapter 3, it was asserted that the ED's effect on corporate financial statements would affect the nature of the contracting and monitoring costs incurred by a firm due to the ED's adoption. For example, a firm reporting lower pension expense may expect an increase in political, labor and management compensation costs. At the same time, that firm may expect to incur lower debt costs. Further, it was asserted that corporate attributes may affect the magnitude of these costs. Finally, it was argued that the ED's effect on corporate financial statements may depend on the funding status of the firm's pension plan. For example, a company with an overfunded pension plan was expected to report a decrease in pension expense. (A company with an underfunded pension plan was expected to report an increase in pension expense.) Therefore, pension plan status may be used to proxy for the ED’s effect on financial statements. Further, the corporate attributes — firm size, labor intensity, and leverage position - may be used to proxy for the magnitude of the costs incurred, and the probability of incurring them. Several measures for firm size, labor intensity, and leverage position are reported in audited financial statements and are published and summarized by data services. The Directory of Corporate Affiliations (1986) was used in the present study to obtain data on each company’s size, labor intensity and leverage position. This Directory provided data for fiscal years ending 1985. Additionally, the NAARS library of 1985 annual reports and the Standard and Poor's 79 library of 10-K reports were used to obtain information from footnotes on pension plan status. One corollary objective of this study was to provide a replication of work done by prior researchers on other accounting issues. Therefore, many of the measures used for corporate attributes were drawn from prior research efforts. The measures used in this study for firm size, labor intensity, leverage position, and pension plan status are summarized in Table 4B and are discussed here. Firm size was measured using two different corporate attributes - total sales and the book value of assets. Watts and Zimmerman (1978), Zmijewski and Hagerman (1981) and Kelly (1982, 1985) used the book value of assets to proxy for firm size. McKee, Bell and Boatsman (1984) used sales.10 Lasater (1982) used earnings before taxes (EBT) to proxy for firm size. EBT was not used as a proxy for firm size in this study due to the year—to-year volatility found in that measure. Labor intensity was measured using the number of employees per sales dollag. A corporate attribute proxy for labor intensity has not been included in prior research efforts. At least two alternative measures for labor intensity are available: (1) employees per sales dollar or (2) employees per asset dollar. Employees per sales dollar was used in this study because every company is reporting sales in current, Comparable dollars. The book value of assets, which are stated in historical dollars, reflect the age of a firm’s assets. Stated otherwise, the book value of assets may not be comparable across firms. 10An advantage of using sales to proxy for firm size is that the measure is in comparable current dollars rather than historical dollars. 80 TABLE 48 CORPORATE ATTRIBUTES Firm size - total sales; book value of assets Firm size is expected to affect the magnitude of political and administrative compliance costs associated with the pension ED. Labor intensity - number of employees per sales dollar Labor intensity is expected to affect the magnitude of labor costs associated with the ED. Leverage position - total debt to equity ratio Leverage position is expected to affect the magnitude of debt costs associated with the ED. Pension plan status - pension obligations to pension assets ratio Pension plan status is offered as a proxy for the ED’s effect on corporate financial statements. It is expected to affect the magnitude and direction of the Changes in political, labor, debt, administrative compliance, and management compensation costs associated with the ED. '1‘.“ ‘l- .4. an: .o ~ AL u 81 Leverage position was measured using a total debt/total eguity ratio as used by Zmijewski and Hagerman (1981) and Kelly (1982, 1985). Zmijewski and Hagerman found that this measure was highly correlated with other measures such as total long-term debt to total equity. Finally, pension plan status was measured using the ratio of accumulated pension obligations to pension plan assets as disclosed in the SFAS 36 footnote.ll This ratio provided a measure for the effect of the ED on corporate financial statements. A company with a ratio greater than one (an underfunded plan) was expected to experience an increase in reported pension expense and pension liabilities. A company with a ratio less than one (an overfunded plan) was expected to experience a decrease in reported pension expense and no Change in pension liabilities. 4.3.4 Independent Variables: Manager Perceptions of the FASB It was hypothesized in Chapter 3 that a manager’s perceptions of the FASB would affect his decision to lobby on the ED. Four specific factors were identified, including (1) the probability of influencing the FASB by filing a comment letter, (2) the costs associated with a Change in the accounting standard-setting body (LRPV), (3) the probability of affecting the FASB's continuance by filing a comment llAccumulated pension benefits, as disclosed in the SFAS 36 footnote, do not include the effects of projected increases in compensation. The ED would require that pension expense be determined based on projected pension benefits, which include these effects. Therefore, the pension obligations/assets ratio used to proxy for the ED’s effect on corporate financial statements is biased towards finding an overfunded pension plan. This bias may reduce the power of this study. 82 letter, and (4) the costs associated with filing a comment letter. A questionnaire was used to obtain measures for each of these factors as summarized in Table 4C and discussed below. (1) The probability of influencing tha FASB by filing a comment letter, and (4) the costs associared with that filing were measured in two ways. First, the managers were asked how they generally attempted to influence the FASB in the accounting standard-setting process. Additionally, they were asked what factors most affected their specific decision to file or not file a comment letter on the ED. Several alternative responses were offered for each question. These responses included the probability of influencing the FASB and the costs associated with filing. The managers were asked to allocate 100 points to the set of alternative responses (including an "other" category which could be added at their discretion). (2) The costs associated withra Change in the accounting standard-setting body (LRPV) were measured in three ways. Two of these measures addressed the managers' general perceptions, and one measure was specifically directed at the ED. First, the managers were asked if their companies supported the FASB through financial contributions to the Financial Accounting Foundation. This binary yes/no (1/0) measure was used on the premise that a manager who associated increased costs with a Change in the present accounting standard—setting body would support the FASB financially. A second measure of LRPV was obtained by asking managers about the effect of a shift in financial accounting standard-setting to the governmental sector. Managers were asked if they would expect such a 83 TABLE 4C MEASURES OF MANAGER PERCEPTIONS OF THE FASB (See Appendix B for Complete Questionnaire) Probability of influencing the policy outcome "We do not attempt to influence the FASB because we do not believe the Board is responsive to corporate constituents."a "We did not believe that our participation would have any effect on the standard issued by the FASB."b Cost of filing comment letters "We do not attempt to influence the FASB because it is too costly."a "The cost and effort incurred to research the proposal and prepare a comment letter" affected our decision to file a comment letter on the ED.b Long run participation value "Does your company support the FASB through financial contributions to the Financial Accounting Foundation? "If financial accounting standard setting shifted entirely to the governmental sector, what would be the effect on your company’s total annual cost of financial accounting, research and external reporting activities? "Our company’s desire to support the FASB in its rulemaking efforts affected our decision to file a comment letter on the ED.b Probability of affecting the FASB’s continuance "Do you believe that by filing a comment letter you increase the probability of the FASB’s continuance?" aManagers were asked to allocate up to 100 points to this response to the question, "How does your company attempt to influence the FASB?" bManagers were asked to allocate up to 100 points to this desponse to the question, "...which factors most affected your decision to file or not file a comment letter on the March 22, 1985 Exposure Draft on pension accounting?" 84 shift to result in an increase or a decrease in the costs of financial accounting, research and external reporting activities. Further, they were asked to estimate how great that increase or decrease might be in terms of a percentage Change. This percentage measure was used based on arguments that the downfall of the FASB may be followed by a shift to governmental accounting standard-setting.12 The third measure of LRPV was obtained by asking managers if their decision to file a comment letter on the ED was affected by a desire to support the FASB (based on a 100 point allocation). This measure was used on the assertion that a manager would not spend time and effort to support the FASB by filing a comment letter unless he believed that a Change in the accounting standard-setting body would be costly. Further, it may be argued that the third measure of LRPV may also be measuring the (3)4probapility of affecting the FASB's continuapce. A manager will file a comment letter to support the FASB only if he believes that the FASB’s continuance can be affected by filing. A second measure for the probability of affecting the FASB's continuance by filing a comment letter was measured by asking managers directly to estimate this probability. 4.4 Qpestionnaire Desiga As discussed above a questionnaire was used in this study to obtain measures for the dependent variable - position Choice. Additionally, the questionnaire was used to measure two sets of 12See Previts and Merino (1979). 5% ant. le ‘v 1» vs 4: 85 explanatory variables - manager expectations of corporate contracting and monitoring costs, and manager perceptions of the FASB. The questionnaire provided information which could not be obtained from any other available source. By determining the position Choices of nonfilers, this study could examine the behavior of a group of corporate managers which had not been examined in prior research efforts. Further, by measuring the managers’ expectations of corporate costs associated with the ED, the validity of the corporate attributes used in prior research to proxy for these expectations could be assessed. Finally, by measuring manager perceptions of the FASB, this study could examine the relationships between these perceptions and other factors in the manager’s participation Choice. However, the use of a questionnaire to obtain variable measures introduces potential limitations to this study. These limitations may result from weaknesses in measurement reliability or from questionnaire nonresponse bias. Measures obtained from a questionnaire may not be reliable if the respondents do not Clearly understand what is being asked. Furthermore, even if they understand the questions being asked, managers may be unwilling to respond truthfully or unable to provide reliable estimates. In either of these cases, the measures obtained from the questionnaire would also be unreliable. Further, nonresponse bias may occur if every manager who receives a questionnaire does not respond. In this study, questionnaires were nailed to a random sample of NYSE firm managers. A random sampling znethod was used to obtain a sample which would provide an unbiased 86 representation of the population.13 If less than 100 percent of these managers respond, the sample available for analysis would be self- selected rather than random. A self—selected sample is more likely to be biased than a randomly selected sample. Therefore, unless a 100 percent response rate is achieved, the results of this study may be subject to nonresponse bias. Several techniques were used in designing the questionnaire to increase the reliability of the measures obtained from the questionnaire and to increase manager response rates. In addition, statistical techniques were performed to assess measurement reliability and nonresponse bias. The questionnaire design is discussed in this section, and the methods used to assess measurement reliability and nonresponse bias are discussed in Section 4.5. 4.4.1 IncreasingyReliability of the Questionnaire The reliability of measures obtained from a questionnaire is a function of the clarity of the quesionnaire and the respondents’ ability and willingness to respond truthfully. To assess these factors, the questionnaire used in this study was pretested before it was mailed to the sample of 300 NYSE firms. The questionnaire used in this study was intended to be completed by a Chief financial officer (e.g., Vice-President of Finance or Controller). Therefore, the questionnaire was pretested by mailing it 13See Kerlinger (1964, p. 52): "Random sampling is that method of drawing a portion (or sample) of a population or universe so that each member of the population or universe has an equal Chance of being selected.... [Therefore], a random sample can be said to be representative of the population from which is was drawn." 87 to top level managers of four companies in the Detroit/Toledo area. Three NYSE firms that had filed comment letters on the pension ED and one firm that had not filed were selected for the pretest. A cover letter which indicated that the recipient was being asked to participate in a pretest accompanied the questionnaires. Two of the four managers returned the questionnaire within one week and follow-up interviews were conducted. One of the managers responding was the director of financial accounting research at a large public utility that had filed several comment letters on the pension accounting issue. The assistant controller of a large manufacturer that had not filed a comment letter also responded. The managers of the other two companies expressed interest in the questionnaire when a follow—up phone call was made, but ultimately did not respond. A revised questionnaire was sent to the randomly selected sample of 300 NYSE firm managers. This questionnaire was accompanied by a cover letter which explained the purpose and importance of the study. The cover letter also assured respondents that their names and that of the companies they represented would be kept strictly confidential. (See Appendix B for Questionnaire and Cover Letter.) 4.4.2 Reducing Nonresponse Bias The response rate to a questionnaire may be a function of its appearance and length, the timing of its receipt, and the importance of the subject it addresses to those who receive it. Several 88 techniques were used to increase the response rate on the questionnaire used in this study. The questionnaire and cover letter were prepared on high quality paper, and each cover letter was individually typed and addressed to the appropriate manager. Further, the length of the questionnaire was limited so that it would require approximately 30 minutes to complete. The first questionnaires were mailed on September 9, 1985, shortly after Labor Day. For most companies, this represents a relatively slow period when management is more likely to be available to respond. Further, the September mailing was also Close to the comment deadline (June 21, 1985) of the Pension Exposure Draft. Therefore, the topic of the questionnaire was likely to be of interest to corporate managers who were aware of the pronouncement. In addition, position Choices on individual aspects of the ED, assessments of costs, and reasons for filing or not filing on the ED were likely to be ’fresh’ in the minds of the managers. Second requests were mailed on October 9, 1985 to managers who had not responded as of that date. This mailing included a second copy of the questionnaire in case the first copy had been misdirected or lost. The questionnaires included in the second mailing were identifiable as second requests. Therefore, comparisons could be made between first and second request respondents as discussed in Section 4.5.2. 89 4.5 Assessment of Reliability and Nonresponse Bias 4.5.1 Reliability The reliability of a measurement method may be assessed in terms of its ability to measure a variable consistently across subjects. For example, a reliable measure of total sales was obtained for this study from the Directory of Corporate Affiliations which summarizes data reported in companies’ audited financial statements. Sales for each company were reported in these statements based on the same set of generally accepted accounting principles. Therefore, the sales reported by each company were consistently defined and measured.14 Based on this line of reasoning, corporate attribute measures obtained from audited financial statements are assumed to be reliable. However, measures obtained from a questionnaire are subject to a high degree of individual interpretation. Therefore, measures obtained from a questionnaire may not be consistently defined between subjects. Efforts were made to increase the reliability of the quesionnaire used in this study by careful construction and pretesting. The effectiveness of these efforts to increase reliability was assessed using Cronbach's coefficient alpha. Cronbach’s coefficient alpha was developed to assess the overall reliability of a questionnaire. It would be preferable to assess the reliability of individual measures obtained from the questionnaire. However, this cannot be done without data on the true variable 14Reliability does not require that every company use the same revenue recognition method. It only requires that the Choice of Inethods be based on the same criteria (GAAP). 90 measures. For example, in the questionnaire, managers were asked to provide an estimate of the debt costs their companies would incur if the pension ED were adopted. Managers who were unwilling or unable to provide an estimate of these costs may have responded by guessing, or by reporting an amount they believed the researcher wanted to see.15 In either of these cases, the reported cost expectations would be unreliable, but because the rrpa cost is unknown, the unreliability of the reported amount could not be assessed. Cronbach’s coefficient alpha requires the use of a symmetric questionnaire (one which includes many questions on the same overall concept). Therefore, the questionnaire used in this study was partitioned and treated as three separate instruments. For example, the first part of the questionnaire used in this study addressed several aspects of one overall concept - manageryperceptions of the EASE. That part of the questionnaire was divided in half (split- halves) and a total score was computed for each half. If that part of the questionnaire has overall reliability, the total scores obtained from each half should be highly correlated. Further, the degree of correlation between these half—scores may be used to assess the reliability of the questionnaire.16 A problem with using the split-halves approach is that different correlation coefficients may be obtained depending on how the questionnaire’s items are allocated to the "halves". To overcome this problem, Cronbach (1951) developed a formula to determine the mean 15See Payne (1951). l6Nunnally (1967, p. 9) suggests that, "In the early stages of research on predictor tests or hypothesized measures of a construct,... [a correlation] of .60 or .50 will suffice." 91 reliability coefficient for all possible ways of allocating the questions. Cronbach’s coefficient alpha was computed for the questionnaire used in this study from a covariance matrix of scores using the following algorithm. k 2 O'iz k i=1 a = k-l 1 - GtZ ,where (4.5.1) k = number of items, Giz = variance of item i, and Ot2 = total covariance. Separate coefficient alphas were computed for the measures obtained on manager perceptions of the FASB, and for the measures obtained on manager expectations of the corporate costs associated with the ED. (See Section 5.2 for discussion of results.) The reliability of manager position choice was not tested using Cronbach’s coefficient alpha. Rather, correlations were run between manager responses to the questionnaire and comment letters filed with the FASB. (The results of these analyses are discussed in Section 5.4.) 4.5.2 Nonresponse Bias Ideally, nonresponse bias would be assessed by comparing measures from the group of questionnaire respondents with measures from the group of nonrespondents. These comparisons would be made on the dependent and explanatory variables included in the study. However, data is not available for nonrespondents in most questionnaire 92 studies. In these cases, researchers have argued that late respondents to a questionnaire may be representative of the sample of nonrespondents. Therefore, comparisons made between first and second (or third or fourth...) request respondents may provide evidence of nonresponse bias. The present study is focused on the position and participation Choices of corporate managers. Inferences about these Choices are drawn from a self-selected sample. The potential existence of non- reponse bias in the test results causes the following four questions to be raised: (1) In tests where questionnaire data is used, can inferences about the position choices of managers who responded to the questionnaire be generalized to the population of managers who did not respond? (2) In tests where questionnaire dara is used, can inferences about the parricipation choices of managers who responded to the questionnaire be generalized to the population of managers who did not respond? (3) In tests where corporate attributes are used, can inferences about managers' position Choices be generalized from a self-selected subsample (i.e., the sample of questionnaire respondents) to the population of managers from which the subsample emerged? (4) In tests where corporate attributes are used, can inferences about managers’ participation choices be generalized from a self- selected subsample (i.e., the sample of questionnaire respondents) to the population of managers from which the subsample emerged? Data was available for both questionnaire respondents and nonrespondents on the participation Choices of managers and on 93 corporate attributes (question 4 above). Therefore, comparisons were made between respondents and nonrespondents for those variables. A Chi-square test was used to determine if the two samples differed on participation choice.17 In addition, Mann-Whitney U tests were used to determine if the corporate attributes - size, labor intensity, leverage position, and pension plan status - of the two samples differed.18 Data was not available for questionnaire nonrespondents on manager position choices, cost expectations or perceptions of the FASB (questions 1-3 above). Therefore, comparisons were made between respondents to the first and second mailings for these variables. A Chi-square test was used to determine if the two samples differed on position Choice. In addition, Mann-Whitney U tests were used to determine if the cost expectations and perceptions of the FASB stated by first and second request respondents differed. (The results from the tests on nonresponse bias are discussed in Section 5.5.) l7See Siegel (1956, p. 104): "When the data of research consist of frequencies in discrete categories, the Chi-square test may be used to determine the significance of differences between two independent groups." 18The parametric t-test is more powerful in finding differences between samples than the Mann-Whitney U test when all conditions of its statistical model (i.e., normal distributions, homogeneous variance, etc.)are met. However, when these conditions do not hold, the non-parametric Mann Whitney U test of rank orderings is more reliable and may be more powerful. See Siegel (1956, p. 126). Also, see Table 4D for summary of tests on the normality of underlying data distributions in this study. 94 4.6 Construct Validity The measures of corporate attributes obtained for this study are assumed to be reliable as discussed above. However, the relationship between these attributes and expected contracting and monitoring costs cannot be assumed. Prior researchers have repeatedly used corporate attributes to proxy for the contracting and monitoring costs associated with a mandated accounting change. Yet, no studies have addressed the empirical relationship between these corporate attributes (proxies) and contracting and monitoring costs (constructs).19 The third corollary objective of this study is to assess that relationship. The construct validity of a proxy may be assessed in terms of its ability to measure the construct it is intended to measure (convergent validity), and only that construct (discriminant validity). Ideally, construct validity would be assessed by examining the correlations between the proxies and true measures of the constructs. However, proxies would not be used in studies if true measures of the constructs were economically obtainable. Practically speaking, it is impossible to directly examine the relationship between the corporate attribute proxies and their contracting and monitoring cost constructs. To overcome this problem, Campbell and Fiske (1959) developed a method to assess the validity of proxies (methods of measurement) when at least two different proxies for the same construct are obtained. 19A construct is a concept which has been deliberately and consciously adopted for a scientific purpose. A proxy provides an operational measure for that construct. See Kerlinger (1964, p32). 95 Campbell and Fiske argue that two proxies for the same construct should be highly correlated if they both measure the construct they are intended to measure. Therefore, when at least two proxies for the same construct are highly correlated, evidence is provided to support their convergent validity.20 To assess the ability of the proxies to measure pply the construct they are intended to measure, it is necessary to obtain proxies for other constructs. Two proxies for different constructs should not be highly correlated. Therefore, when proxies used to measure different constructs are highly correlated, evidence is provided to indicate that they do not have discriminant validity. As an extension of that logic, Campbell and Fiske argue that a proxy has discriminant validity if it is more highly correlated with proxies for the same construct than it is with proxies for different constructs. Campbell and Fiske (1959) developed a method to assess the construct validity of proxies utilizing a multimethod-multitrait correlation matrix. Multimethod refers to the multiple proxies obtained for the same construct. Multitrait refers to the multiple constructs examined. A multimethod-multitrait correlation matrix was constructed for this study. As discussed in Section 5.3, the matrix included two proxies for each explanatory variable included in the manager position and participation choice models. The correlations between proxies for the same construct were examined to assess convergent validity. In 20Therefore, a high correlation between two proxies can only provide evidence of construct validity. The strength of the evidence on construct validity increases with the number of proxies offered. See Kerlinger (1964). 96 addition, to assess discriminant validity, these correlations were compared with the correlations between proxies for different constructs. 4.7 The Statistical Analyses of the Position and Participation Choice Models 4.7.1 Probit Analysis Probit analyses were used in this study to test the models of manager position and participation Choice and to determine the importance of each hypothesized explanatory variable. Probit analysis was specifically developed by Finney (1947, 1971) for qualitative choice models. Tobin (1958) and others have adapted the probit model to economic analyses. The probit model, which uses maximum likelihood estimation (MLE) has several advantages over ordinary least squares (OLS) regression models when relating a qualitative dependent variable to one or more explanatory variables.21 These advantages may be described as follows: (1) When the dependent variable is dichotomous, the statistical assumptions of the OLS model are violated. In this case, the error term possesses a discrete distribution, is not normally distributed, and will tend to depend on the values of the explanatory variables (heteroskedasticity). 21A Logit analysis provides the same advantages over OLS discriminant analysis. Amemiya’s (1981) survey of qualitative response models indicated that a probit and logit analysis produce equivalent predictions of the discrete Choice they are modeling. 97 (2) Furthermore, the linear probability OLS function does not constrain the predicted values to lie between the zero and one interval. Because the predicted values lie outside the zero-one interval, the significance of the beta coefficients on the individual explanatory variables are not consistent. To eliminate the problems associated with OLS, the discrete dependent variable must be converted to a continuous dependent variable. This may be done by substituting the probability of occurrence of a discrete event for the discrete event itself. Probit analysis uses a nonlinear probability function to make this conversion. The probit function accounts for the heteroskedasticity of the error terms and restricts the predicted values to the zero-one interval. Press and Wilson (1978) found that logistic regression with MLE (as used in probit analyses) outperformed linear discriminant analysis in empirical studies of discrete Choice problems, but not significantly. However, they found that when the explanatory variables are not normally distributed, the advantages of MLE are more pronounced. When the explanatory variables (in addition to the dependent variable) are not normally distributed, the probit analysis provides consistent estimators of the beta coefficients while discriminant analysis does not. Furthermore, discriminant analysis provides misleading results as to the significance of the regression coefficients. A coefficient which is really insignificant will tend 98 to be estimated as such by a probit analysis which is robust when the explanatory variables are not normally distributed.22 To summarize, probit MLE estimators from a discrete Choice model are unbiased and consistent even when the explanatory variables are not normally distributed. As shown in Table 4D, a preliminary analysis of the data used in this study indicates that the underlying distributions of the explanatory variables are significantly skewed. Because the dependent variables examined in this study are qualitative, and the explanatory variables are not normally distributed, the MLE probit analysis was used rather than the OLS discriminant analysis to examine the multivariate position and participation choice models. 4.7.2 Mann-Whitney U Tests Managers may differ significantly on the hypothesized explanatory variables even though the variables are not significant in the multivariate probit analysis. This may be due to multicollinearity of the explanatory variables which reduces the power of a multivariate analysis to find significance on the individual variables.23 Therefore, in addition to the probit analysis on the models of manager lobby behavior, Mann—Whitney U tests of rank order were computed on ZZMLE was found by Press and Wilson (1978) to be robust even when many of the explanatory variables were binary. 23A large number of explanatory variables in a multivariate analysis also reduces the power of significance tests on the individual variables when compared to a univariate analysis. 99 TABLE 4D SUMMARY OF TESTS ON THE NORMALITY OF UNDERLYING DATA DISTRIBUTIONS Kolmogorov- Variables Skewnessa Kurtosisb Smirnov TestC Manager Cost Expectations Political 7.154 53.043 10.083 and Labor Administrative 5.956 45.179 9.983 Debt 10.399 109.446 5.716 Mgmt Comp 4.045 19.291 6.675 Corporate Attributes Sales 7.724 70.065 8.616 Assets 6.558 50.681 8.720 Number of 2.928 12.924 10.258 Employees Employees .782 2.810 8.916 per sales Debt/Equity 4.125 20.517 8.720 Plan status 1.002 6.588 8.914 aA skewness of zero represents a symmetrical distribution. bA value of three for kurtosis indicates normal peakedness. CThe Kolmogorov—Smirnov test is used to assess the normality of the distribution underlying a set of data. A value over .014 indicates that the data is not normally distributed. (See Siegel, Table E) 100 TABLE 4D (CODt'd) Kolmogorov- Variables Skewnessa Kurtosisb Smirnov Testc Manager Perceptions Pr(influence-I) 2.487 7.8299 10.881 Pr(influence-II) 5.956 45.179 9.983 LRPV (I- FAF) 9.130 89.912 7.853 LRPV (I- shift) 8.618 81.599 6.717 LRPV (II) 3.907 19.259 10.118 Pr(LRPV) 2.246 7.164 7.534 Lobby cost (I) 4.714 34.659 9.631 Lobby cost (II) 2.122 5.988 10.401 aA skewness of zero represents a symmetrical distribution. bA value of three for kurtosis indicates normal peakedness. CThe Kolmogorov-Smirnov test is used to assess the normality of the distribution underlying a set of data. A value over .014 indicates that the data is not normally distributed. (See Siegel, Table E) 101 the individual explanatory variables. Univariate comparisons were made between ED supporters and opposers (position choice) and between filers and nonfilers (participation choice). The Mann-Whitney U test was used in this study to assess sample differences because the Mann-Whitney U test does not require normal data distributions or homogeneous variance of the independent variables. To apply the Mann-Whitney U test, the measures obtained for each variable are used to rank subjects in order of increasing size. The value of the Mann-Whitney U is based on a comparison of the rankings of the subjects in each group. Although information is lost by comparing rankings of subjects (rather than means), the Mann- Whitney U test has been found to be 95% as powerful as the t-test.24 Further, it is more powerful and more reliable than the t-test when normal distribution requirements are not met. 4.8 Summary Position and participation Choices for a sample of NYSE firm managers on the pension ED were examined in this study. Probit analyses and Mann-Whitney U tests were used to examine hypothesized relationships between these Choices and three sets of explanatory variables: (1) corporate attributes, (2) manager expectations of the corporate costs associated with the ED, and (3) manager perceptions of the FASB. Managers’ position Choices on the ED were obtained from a questionnaire which was mailed after the ED deadline but before the 24See Siegel (1956, p. 126). 102 adoption of the new pension accounting standard (SFAS Statement 87). Additionally, the questionnaire was used to obtain measures for manager expectations of the corporate costs associated the ED, and for manager perceptions of the FASB. Managers' participation Choices were determined by examining the list of ED comment letter filers received from the FASB. Finally, measures were obtained from published financial reports for corporate attributes which were used to proxy for corporate contracting and monitoring costs. The questionnaire was used in this study to provide data which was not obtained for prior studies. The position Choices of nonfiling managers on the ED were obtained, increasing the generalizability of this study. By obtaining measures for manager expectations of the costs associated with the ED, it was possible to assess the construct validity of the corporate attribute proxies used in prior studies. Finally, the relationships between manager perceptions of the FASB, expected contracting and monitoring costs, and participation Choice could be examined. However, the use of a questionnaire in this study introduced potential limitations in terms of measurement reliability and nonresponse bias. Unreliability in the measures of manager position Choice, cost expectations and FASB perceptions would reduce the power of the findings about these variables. Further, nonresponse bias would reduce the generalizability of the findings from the sample to the population of NYSE firm managers. Efforts were made in constructing the questionnaire to increase measurement reliability and to reduce nonresponse bias. Additionally, 103 the reliability of the overall questionnaire and nonresponse bias were assessed to determine the extent of these limitations. The statistical results of this study are presented in two Chapters. The limitations associated with the questionnaire data are evaluated in Chapter 5. Nonresponse bias and data reliability are discussed. Additionally, construct validity of the independent variable measures is addressed. The analyses of the position and participation choice models are presented in Chapter 6. The results of the probit analyses and the Mann-Whitney U tests are summarized and discussed therein. CHAPTER 5 ANALYSIS OF DATA RELIABILITY AND VALIDITY 5.1 Introduction The sample design and measurement methods used to obtain data for this study were discussed in Chapter 4. Further, the statistical methods used to assess measurement reliability and validity, and nonresponse bias were described. Finally, the statistical methods used in this study to estimate the models of manager position and participation Choice were discussed. The present chapter presents the results of tests used to assess measurement reliability and validity, and nonresponse bias. The reliability and construct validity of the independent variables used in this study are evaluated in Sections 5.2 and 5.3, respectively. The reliability and validity of the dependent variables are evaluated in Section 5.4. Finally, nonresponse bias is evaluated in Section 5.5. 5.2 Measurement Reliability of the Independent Variables A measure is said to be reliable if it is consistent between subjects and over time. The corporate attribute measures used in this study were derived from audited financial statements which require the use of consistently applied and well-defined generally accepted 104 105 accounting principles. Therefore, it may be assumed that these measures are reliable. However, measures obtained from the questionnaire were not subject to audit. Therefore, unobserved inconsistencies between managers in their interpretations of the questionnaire may exist. In this case, the measures obtained from the questionnaire may be subject to significant weaknesses in reliability. As discussed in Chapter 4, the reliability of an individual measure in a questionnaire cannot be determined unless the true measure is known. However, the reliability of a questionnaire, taken as a whole, may be assessed using Cronbach’s coefficient alpha if the questionnaire addresses a single general concept. The questionnaire used in this study addressed three general concepts - manager position choice on the pension Exposure Draft (ED), manager perceptions of the FASB, and manager expectations of the contracting and monitoring costs associated with pension accounting as proposed in the ED. Therefore, the questionnaire (see Appendix B) was divided into three parts and each part was treated as a single instrument. The reliability of the first part of the questionnaire, on manager position Choice (Part II, question 1, p. 215), was not tested using Cronbach’s coefficient alpha. This test was not needed because a true measure of that Choice was available for managers who had filed comment letters on the ED. Therefore, comparisons could be made between the positions stated by managers in the questionnaire and in comment letters (the true measures). The results of these comparisons are reported separately in Section 5.4. 106 A coefficient alpha of .82 was computed on the items in the second part of the questionnaire which pertained to manager perceptions of the FASB (Part I, pp. 211-214, and Part II, questions 2 and 3, p. 216). This alpha provides evidence that the measures obtained for these variables are reliable.1 Finally, an alpha of .14 was computed on the items in the third part of the questionnaire which pertained to manager expectations of the contracting and monitoring costs associated with the ED (Part II, question 4, p. 217). This low alpha suggests that the measures obtained for these variables are not reliable. However, it may also reflect the limited number of items (six) on manager cost expectations. Nunnally noted that the computed reliability of a questionnaire is a "direct function" of the number of items in that questionnaire.2 Further, the low alpha may reflect assymetry in the questionnaire on manager cost expectations. For example, managers who expect to incur high administrative compliance costs due to the adoption of a new accounting standard may not also expect to incur high debt costs. In this case, Cronbach's coefficient alpha, which relies on a symmetric instrument, may not be provide a good assessment of reliability. Unreliability in measures of manager cost expectations would reduce the power of this study. Significant relationships between variables in an analysis (multivariate or univariate) are less likely 1Nunnally (1967, p. 9) suggests that, "In the early stages of research on predictor tests or hypothesized measures of a construct,... reliability of .60 or .50 will suffice." 2See Nunnally (1967), p. 206. 107 to be found if the measures obtained for those variables are unreliable. Therefore, the power of analyses on manager lobby behavior and from tests of construct validity, which include manager cost expectations, may be weakened. 5.3 Construct Validity In Chapter 3, a manager’s position and participation Choice were hypothesized to be related to several explanatory variables. These variables included contracting and monitoring costs associated with a proposed Change in accounting, corporate attributes, and manager perceptions of the FASB. The inclusion of corporate attributes in the set of explanatory variables was predicated on the assertion that these attributes may be related to the magnitude of the costs associated with a Change in accounting. Five contracting and monitoring costs were hypothesized to be related to manager position and participation Choice. These costs included (1) political, (2) labor, (3) debt, (4) administrative compliance, and (5) management compensation costs. In addition, four manager perceptions of the FASB were hypothesized to affect manager participation Choice. These perceptions included (1) the probability of influencing the policy outcome, (2) the costs associated with a Change in the accounting standard-setting body (LRPV), (3) the probability of affecting the FASB’s continuance by filing a comment letter, and (4) the cost of filing a comment letter. Stated manager cost expectations and corporate attributes were offered as proxies for contracting and monitoring cost constructs. 108 Additionally, a manager’s general perceptions, and ED-specific perceptions were offered as proxies for constructs on manager perceptions of the FASB. The measures obtained for each of the constructs listed above are summarized in Table 5A. As shown in Table 5A, separate manager expectation proxies were not offered for political and labor costs. Further, separate corporate attribute proxies were not offered for political and administrative compliance costs, and no corporate attribute measure was offered for management compensation costs.3 Finally, separate ED-specific proxies were not offered for the cost of a change in the accounting standard- setting body and the probability of affecting the FASB’s continuance. However, two general proxies were offered for the cost of a Change in the accounting standard-setting body. A construct must be measured using at least two proxies if it is to be included in a multimethod-multitrait correlation matrix. Further, if the proxies for two different constructs are highly correlated, they should not be included in the matrix. A simple correlation matrix was constructed which included each of the proxies listed in Table 5A. This matrix was used to examine correlations between proxies and to select those proxies (and constructs) to be included in the multimethod—multitrait correlation matrix. 3This represents a limitation in the study. The empirical model used for testing is underspecified. See Chapter 7 for further discussion. 109 TABLE 5A CONSTRUCT MEASURES Contracting and Monitoring Costs: Construct Proxies Manager Expectations Corporate Attributesa Political costs Public relations costs Sales, assets Labor costs Public relations costs No. of employees, employees per sales Administrative Administrative compliance Sales, assets compliance costs and internal management Debt costs Debt costs Debt/equity ratio Management Management compensation ------------------ costs costs Manager Perceptions of the FASB: Construct Proxies General Perception Perception on ED Probability of "FASB nonresponsive to "Could not influence influence corporate constituents" outcome on ED" (Question 4, p. 212) (Question 2, p. 216) Cost of Change Contributions to FAF "Filed to support in standard- (Question 1, p. 211) the FASB" setting body Cost of shift to (Question 2, p. 216) government sector (Question 6, p. 213) Probability of "Probability of affecting "Filed on ED to affecting FASB’s the FASB’s continuance" support FASB" continuance (Question 2, p. 211) (Question 2, p. 216) Costs of filing "Too costly to file" "Costs of filing too a comment letter (Question 4, p. 212) high to file on ED" (Question 2, p. 216) aIn addition, pension plan status was offered as a proxy for the ED’s effect on corporate financial statements and on each of the contracting and monitoring costs. 110 5.3.1 The Simple Correlation Matrix The matrix in Table SB includes the simple correlations computed between the proxies offered for the each construct examined in this study. These correlations were computed on a sample which included only those subjects for whom all data was available. Data would have been better utilized by computing individual correlations separately for each combination of proxies. Using this latter technique, only data for the two proxies being correlated would be used in each computation. Therefore, the sample sizes would have been larger. However, the number of cases included in each computation would vary making comparisons between the correlations uninterpretable. The analysis of construct validity relies on comparisons between correlation coefficients. Therefore, the correlations were computed for a single sample of 70 managers. These correlations are shown in Table SB and are discussed here. Contracting and Monitoring Costs - A preliminary examination of the simple correlation matrix indicated that manager expectations of public relations (PREL), administrative compliance (ADMN), and internal management (INT) costs are highly correlated. The corporate attributes including sales (SALE), assets (ASSE) and number of employees (EMP), are also highly correlated. Furthermore, when compared to their relationships to other variables in the matrix, relatively higher correlations were computed between those manager cost expectation proxies (PREL, ADMN, INT) and corporate attribute proxies (SALE, ASSE, EMP). These findings suggest that these six 111 63:53: 93 £52: corn—“E8 :obgsEpoSoEo—no: 95 5 36305 on; 523 .2538 25m 95 8.. 8.35 93 .8033 956508 corn—280a 2.38 am osofiaos as $332.. mmé manages Co grape Enron Maroone— Co 330 058% am masons Co 38 1550 85298 ”.mmé “.58.: Co banana.— mmow 9 ”Ev. .«0 50b“ oom. mvo.) moo. boo. bNH.) mmo.) 93.: mmo.l mNH.I .mloln4 mmo.) mmo.l mac. 9:.) 0N0... Hmm HHUA HUA MAME >mmA ago 2.. $8 2835.8 5a "2a 9.8 8.3358 35355. 8.8%.“. so? :0..— =eaamfix§8 gauge :0 ouch... v0.89“... P202 930 95:32 £32m 303.6 4mm.— Uo ass 5% 383. some one 8138315 aim «.5: one as: 5325238 Sana 25a 38.. ion. mm? >5: 8: bromine m6 "8315 oo 35:2 aim >00 38 assuage 1:95 38am to 8.3 3o... 3% omopoeu moo.) moo.u ooo.- mmo.- owo. Noo.- ooo. moo.- moo.- Noo. ooo.u ~oo.- moo. Homo omo.) omo.- ooo.- mmo.u mo~.u boo.) ooo.- moo.) moo.) moo.) soo.n ooo.- omo.) omo ooo. omo. mom. mmo.u moo. moo.) omo.u ooo.- ooo.- mom. ooo.) ooo.) mmo.4 oooo moo. ooo.- moo. omo.u mmo. moo.u ooo. omo.- moo.) mom. ooo.u moo.u moo. ooo mmo.( omo.- omo. omo.u ooo.n ooo. «oo.u mmo. omo.- moo.u omo.( ooo.u ~oo.( roam ooo. ooo. moo.u omo.) ooo. Nmo. ooo.u ooo. moo. moo.) poo. mom. moo. >mmo mood omo.- mao.( ooo. omo. moo. moo. omo.- omo.u ooo.- boo.) omo.) >oo oao. moo. moo.u omo. moo. omo. moo. omo.- moo. oom. ooo. mam ooo.) oNo. omo.- ooo. moo.- ooo.) ooo. ooo.n moo.) moo.) axon ommi. ooo. omo.u ooo.- omo.- oom. ooo. moo. ooo.- ammo ohm. ooo. ova. mom. omm. omo. moo. omo. zoom mmo. moo. poo. moo. mom. moo. ooo. m\o ooo. mmo. ooo.u boo. mom. «on. azo ooo. moo.( ovm. wand moo. zzoa moo. moo. omm. mom. ommo ooo.» ~oo.: ~ao.- m\zm omm. ooh. moms omo. mam norm >oo man axon ammo zaoo m\o ozo zzoa ommo m\zm mmma ozm worm .mommo on. mmeedz ZOHB0 H0: x2=0 6However, as shown in Table 5C, pension plan status was highly correlated with manager expectations of debt costs. A high correlation was not found between leverage (debt/equity) and debt costs. 7The chi-square statistic used in the probit analyses is based on a market-share test wherein a constant is estimated for the null hypothesis that will replicate the Observed proportion of ED opposers/supporters. This is more stringent than a naive 50/50 proportion assumption. 146 Hl.1‘ A manager's position Choice on the ED (Di) is inversely related to firm size (f1). Bl < 0 H02 Bl = 0 31.3: A manager’s position Choice on the ED (Di) is inversely related to labor intensity (f2). 52 < 0 H0: 52 = 0 H1.3: A manager's position Choice on the ED (Di) is positively related to leverage position (f3). 33 > 0 H0: B3 = 0 H1 4: A manager’s position Choice on the ED (Di) is positively related to pension plan status (f4).8 54 > 0 Ho: 54 = 0 The sample used in this study to test these hypotheses on the position Choice Of a filing manager included 37 Opposers and 6 supporters. Further, the sample used to analyze the position choice of a nonfiling manager included 41 Opposers and 33 supporters. (See Table 6B.) The results Of the probit analyses which were performed for ED filers and nonfilers, are summarized in Table 6C and Appendix C. As discussed above, three analyses were run for each sample including 8Every company in the sample had an overfunded pension plan (pension obligations/pension assets < 1). Therefore, as that ratio gets Closer to one (larger), the ED's effect is smaller and the manager is more likely to support the ED. 147 TABLE 6C PROBIT ANALYSES CORPORATE ATTRIBUTES AND POSITION CHOICE Corporate Beta Coefficientab Attribute Filers Nonfilers Constant .5171 .7630 (.33) (.05) Firm size .0002 .0023 (Assets) (.17) (.05) Labor .1029 .0135 intensity (.11) (.27) Leverage .4397 .2025 (.09) (.05) Pension plan .6657 .3316 status (.30) (.21) No. of opposers 6 41 NO. Of supporters 37 33 Chi—square 6.47 10.10 (Significance) .1668) (.039) Correctly predicted 88% 57% aA positive relationship was hypothesized between manager position Choice and firm leverage and pension plan status. relationship was hypothesized for firm size, bOne-tailed probability Of t-statistic in parentheses. An inverse labor intensity. 148 each of three highly correlated corporate attributes used to proxy for firm size. The results from the models which included the book value of assets are presented in Table 6C. This analysis provided the best data fit. As shown in Table 6C, the position choice model presented in equation 6.2.1 provided a good fit of the data for the position Choice of an ED nonfiler (chi-square significant at the .04 level). Fifty- seven percent Of manager position Choices on the ED were correctly predicted using this model. As hypothesized (H1.1)I firm size was significantly and inversely related to manager position Choice. However, the coefficient on leverage position, which was statistically significant, was not in the hypothesized direction (H1 3). This finding suggests that the nonfiling manager of a highly leveraged firm tended to Oppose the ED even though it would require a reduction in reported pension expense. One explanation for this finding is that the manager of a highly leveraged firm may be concerned with the ED’s effect on volatility of corporate financial statements. As discussed in Chapter 3, an increase in income and balance sheet volatility is associated with the ED. Therefore, if the manager Of a highly leveraged firm is concerned with financial statement volatility, he may oppose the ED. The model presented in equation 6.2.1 did not provide a good fit for data on the position Choice of an ED iil§£ (chi-square significant at the .17 level). Further, none of the beta coefficients on the hypothesized explanatory variables were statistically significant. However, 88 percent of filing manager position Choices on the ED were correctly predicted using that model. 149 This prediction rate does not appear to be consistent with the reported Chi-square statistic. One explanation for this finding is that the model used to predict manager position Choice was derived from an analysis which included the individual manager whose position was being predicted.9 Therefore, because the sample used to derive that model was small (37 Opposers and 6 supporters), an individual manager’s behavior may have significantly affected the outcome Of the analysis. In this case, the prediction rate may be overstated. However, the Chi-square statistic will provide a better measure of the model’s goodness Of fit.10 The results Of the Mann-Whitney U tests are shown in Tables 6D and 6E for filers and nonfilers, respectively. These results are consistent with the findings from the probit analyses. As discussed in Chapter 2, only two accounting issues have been addressed in prior studies Of manager position Choice. On the interest capitalization issue, Lasater (1982) found that a logit model provided a good fit for the data. Further, Lasater found that firm leverage (debt/equity) was significant in that model. On the GPLA issue, as in this study, the model Of position choice for comment letter filers did not provide a good fit for the data. One explanation for the inconsistency of these findings is suggested based on a comparison of the sample sizes used in each 9A holdout sample was not used to estimate the model. 10Another explanation for the inconsistency of these findings is that the manager position Choice models are not consistent between comment letter filers and nonfilers. 150 TABLE 6D MANN-WHITNEY U TESTS CORPORATE ATTRIBUTES AND POSITION CHOICE ED FILERS Corporate Means Attributes Opposers Supporters U-testa Firm Size Sales (000’s) 3 7,264,070 $18,639,152 .191 Assets (000’s) $11,945,160 $15,011,871 .300 NO. of employees 47,824 18,455 .376 Labor Intensity Employees/sales 33.50 4.67 .071 (millions) Leverage Position Debt/Equity 2.48 1.28 .390 Effect of ED Plan status .72 .67 .242 Number of ED Opposers 37 Number of ED supporters 6 aOne-tailed probability Of Mann-Whitney U test. 151 TABLE 6E MANN-WHITNEY U TESTS CORPORATE ATTRIBUTES AND POSITION CHOICE ED NONFILERS Corporate Means Attributes Opposers Supporters U-testa Firm Size Sales (000’s) $1,691,871 $780,117 .030 Assets (000’s) $1,791,198 $681,759 .016 No of employees 12,168 7,832 .040 Labor Intensity Employees/sales 11.16 11.21 .475 (millions) Leverage Position Debt/equity 2.13 1.08 .038 Effect of ED Plan status .75 .65 .123 Number of ED supporters 41 Number of ED Opposers 33 aOne-tailed probability Of Mann-Whitney U test. 152 study.11 For example, the study on the interest capitalization issue examined a sample Of 129 firm managers. Samples Of 34 and 70 were used in the study on GPLA. Finally, the present study examined a sample Of 43 comment letter filers. Based on a comparison Of these sample sizes, it may be argued that the small samples used to analyze general price level and pension accounting reduced the power Of these studies. In this case, the models Of position Choice may have been correctly specified but the small samples precluded significant findings. The explanation Offered for the inconsistency of the findings on different accounting issues may also apply to the differences found within this study between filers and nonfilers. The sample of nonfilers included 74 managers (41 ED opposers and 33 supporters) as compared to the sample Of 43 filers. This larger sample size may have increased the power Of the analyses for nonfilers. Future research is needed to examine the position Choices Of a large sample of filing and nonfiling managers on a single accounting issue. This research would provide evidence about the differences and similarities of the decision models used by each group. The findings from the present study may pp; be used to draw inferences about these differences and similarities. 11An alternative explanation is that the position Choices made by managers are not consistent across different accounting issues. 153 6.2.2 Participation Choice The corporate attributes hypothesized in this study to affect manager position choice, were also hypothesized to affect participation Choice. It was hypothesized that a manager who Opposed the ED because Of the corporate costs associated with it, may file a comment letter if those costs are high. Further, a manager who supported the ED may file a comment letter when the corporate costs associated with it are low (as compared to foreseeable alternatives). The following statistical hypotheses on participation Choice were derived from equation 6.2.2. (See hypotheses 2.0, and 2.2 through 2.4 in Section 3.6.2 for further discussion.) Hypotheses 1.1 through 1.4 pertain to the participation Choice of a manager who Opposed the ED. Hypotheses 1.5 through 1.8 pertain to the participation Choice of a manager who supported the ED. Each Of these hypotheses are based on the assumption that the ED resulted in a decrease in in reported pension expense. H1.0: The model described in equation 6.2.2 may be used to predict a manager’s participation Choice on the ED. This predictive ability will be evidenced by a significant Chi-square on the probit analysis. x2 > 0 H0: x2 = 0 Hypotheses for ED Opposers H1.1: The participation choice (Vi) of an ED Opposer is positively 154 related to firm size (f1). Bl > 0 H0: Bl = 0 H1_2: The participation Choice (Vi) of an ED Opposer is positively related to labor intensity (f2). 52 > 0 H0: 82 = 0 H1 3: The participation choice (Vi) of an ED Opposer is inversely related to firm leverage (f3). 53 < 0 H0: 53 = 0 H1 4: The participation Choice (Vi) of an ED Opposer is inversely related to pension plan status (f4).12 54 < 0 H0: 53 = 0 Hypotheses for ED Supporters H1 5: The participation Choice (Vi) of an ED supporter is inversely related to firm size (f1). 51 < 0 H0: Bl = 0 “1.6: The participation Choice (Vi) Of an ED supporter is inversely related to labor intensity (f2). 32 < 0 H0: 52 = 0 12As the pension Obligation/pension asset ratio gets closer to one, the ED's effect on financial statements is smaller. Therefore, the manager with a higher ratio is less likely to file a comment letter. 155 H1 7: The participation choice (Vi) of an ED supporter is positively related to firm leverage (f3). 53 > 0 H0: 53 = 0 “1.8: The participation Choice (Vi) Of an ED supporter is inversely related to pension plan status (f4).13 54 < 0 H0: 54 = 0 Analyses Of manager participation choice were run on the total sample of 48 comment letter filers and 209 nonfilers to provide a replication of prior research.14 For the analyses of the total sample, it is asserted that a manager who files a comment letter on the ED is more likely to be an Opposer than a manager who does not file. Further, a manager who both Opposes and files a comment letter may expect to incur high corporate costs if the ED is adopted. Therefore, hypotheses 1.1 through 1.4 were applied to the analyses of the total sample Of 257 managers. The sample used in this study to test hypotheses 1.0 and 1.1 through 1.4 on the participation Choice Of an ED Opposer included 28 comment letter filers and 41 nonfilers. Finally, the sample used to test hypotheses 1.0 and 1.5 through 1.8 on the participation choice Of an ED supporter included 6 filers and 33 nonfilers. 13The relationship between manager participation Choice and pension plan status is based on the magnitude of the ED’s effect on financial statements. Therefore, the hypothesized relationship is not dependent upon the manager's position Choice. 14See Zmijewski and Hagerman (1981), Lasater (1982), Kelly (1982, 1985), and Griffin (1982, 1983). 156 The results Of the probit analyses on manager participation Choice are summarized in Table 6F and Appendix D. The results Of the Mann-Whitney U tests are summarized in Tables 66, 6H, and 61 for the total sample, ED opposers, and ED supporters, respectively. As discussed earlier, three probit analyses were run for each sample including each Of three highly correlated corporate attributes used to proxy for firm size. The results from the models which included the book value of assets are presented in Table 6F.15 As shown in Table 6F, the participation choice model presented in equation 6.2.2 provided a good fit Of the data in each analysis. Further, 84, 72 and 90 percent of manager participation Choices were correctly predicted for the total sample, the sample Of ED Opposers and the sample of ED supporters, respectively. Firm size was significantly and positively related to manager participation Choice in each Of the analyses. This finding provided support for Hypothesis 1.1 on the participation Choice Of ED Opposers. However, a negative relationship was hypothesized between firm size and participation choice for an ED supporter (H1.5). Significant relationships have been found in prior research efforts between manager participation Choice and corporate attributes — firm size and leverage.16 However, prior researchers have not controlled for the position Choice of a manager. The findings Of this study on the total sample are consistent with those Of prior studies. 15The model including assets provided the best fit for the data. 16See Zmijewski and Hagerman (1981), Lasater (1982), Kelly (1982, 1985), and Griffin (1982, 1983). 157 TABLE 6F PROBIT ANALYSES CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE Beta Coefficienta Corporate Totalb EDb EDC Attributes sample Opposers Supporters Constant —l.1601 -.3144 -.9280 (.00) (.29) (.23) Firm size .0013 .0022 .0050 (.00001) (.001) (.03) Labor -.0030 —.0378 -.0470 intensity (.15) (.10) (.29) Leverage -.0191 -.0734 -.4841 (.35) (.21) (.28) Plan status -.0320 -.1750 -.l693 (.47) (.38) (.44) Number of filers 48 37 6 NO. Of nonfilers 209 41 33 Chi-square 44.61 26.29 16.89 (Significance) (.0000) (.0000) (.0020) Predicted 84% 72% 90% aOne-tailed probability Of t-statistic in parentheses. bA positive relationship was hypothesized between participation Choice and firm size and labor intensity. An inverse relationship was hypothesized for leverage and pension plan status. CAn inverse relationship was hypothesized between participation Choice and firm size and labor intensity. A positive relationship was hypothesized for leverage and pension plan status. 158 TABLE 6G MANN-WHITNEY U TESTS CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE TOTAL SAMPLE Corporate Means Attributes Filers Nonfilers U-testa Firm Size Sales (000's) 3 1,254,560 $ 8,175,842 .0001 Assets (000’s) $ 1,194,829 $10,998,314 .0001 NO. of employees 12,387 42,872 .0001 Labor Intensity Employees/sales 16.00 28.16 .0001 Leverage Position Debt/equity 1.44 2.22 .0001 Effect of ED Plan status .81 .73 .0001 Number of nonfilers 209 Number of filers 48 aOne-tailed probability Of Mann-Whitney U test. 159 TABLE 6H MANN-WHITNEY U TESTS CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE ED OPPOSERS Corporate Means Attributes Nonfilers Filers U-testa Firm Size Sales (000’s) $1,691,872 $7,264,071 .0001 Assets (000's) $1,750,759 $11,622,192 .0001 NO Of employees 12,169 47,825 .0001 Labor Intensity Employees/sales 11.16 33.50 .086 (millions) Leverage Position Debt/equity 2.13 2.48 .173 Effect of ED Plan status .75 .72 .132 Number of nonfilers 41 Number Of filers 37 aOne-tailed probability Of Mann-Whitney U test. 160 TABLE 6I MANN-WHITNEY U TESTS CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE ED SUPPORTERS Corporate Means Attributes Nonfilers Filers U-testa Firm Size Sales (000’s) $ 780,117 $18,639,152 .0005 Assets (000’s) $ 681,759 $15,011,871 .0005 NO Of employees 7,832 18,455 .007 Labor Intensity Employees/sales 11.21 4.67 .022 (millions) Leverage Position Debt/equity 1.08 1.28 .044 Effect of ED Plan status .65 .67 .349 Number of nonfilers 33 Number of filers 6 aOne-tailed probability Of Mann-Whitney U test. 161 Further, the findings on the samples of ED Opposers and supporters do not differ from the findings on the total sample. The separate analyses on ED Opposers and supporters in this study provides evidence about the theoretical constructs which may underlie corporate attribute proxies. As noted above, a positive relationship was found between firm size and the participation Choice Of an ED supporter. Firm size was asserted to proxy for political and administrative compliance costs. Therefore, these findings suggest that a manager facing pagp costs due to the ED's adoption was more likely to file a supportive comment letter on the ED than a manager facing lower costs. This relationship is Opposite to that hypothesized. One explanation for this finding is that firm size does not proxy only for corporate costs in the models of manager behavior. The findings from the tests Of construct validity do not support this explanation. A significant correlation was found between firm size and manager eXpectations of political, labor, and administrative compliance costs, as discussed in Chapter 5 (Table 5C). Furthermore, significant correlations were app found between firm size and manager perceptions of the FASB which were hypothesized to affect participation choice. Stated otherwise, the evidence from the tests of construct validity support the use Of firm size as a proxy for corporate costs.17 An alternative explanation for the positive relationship between firm size and an ED supporter’s participation Choice may be 17It may be argued that the tests Of construct validity address only the ability Of two proxies to measure the same construct. They may not measure the true construct. 162 Offered. The magnitude of costs associated with an accounting proposal may be indicative of that issue’s general effect on the corporation. The manager of a company which would incur high costs due to a mandated Change in accounting may have incentive to participate in the standard-setting process on that issue whether he supports or opposes the specific proposal. Additional research on the participation Choices Of managers who support an accounting pronouncement is needed to provide evidence about these relationships. 6.3 Manager Expectations Models Prior researchers have relied on corporate attributes to proxy for the corporate contracting and monitoring costs associated with a mandated accounting change. These proxies have provided good predictors Of manager participation Choice, although they have not performed as well in models Of position Choice.18 In the present study, measures were Obtained for manager expectations of the corporate contracting and monitoring costs associated with the ED’s adoption. These measures were used in a correlation matrix to assess the validity Of the corporate attribute measures used in prior studies.19 The results Of these tests for construct validity provided support for the use Of certain corporate attributes as proxies for these costs. Further, as discussed above, 18Although Lasater (1982) found a significant relationship between firm leverage and manager position choice on the interest capitalization issue. 19The construct validity Of manager cost expectation proxies was also assessed. 163 the corporate attribute models provided a good fit on both position (for nonfilers) and participation Choices. The findings of this study provide support for the use Of corporate attributes in models Of manager position and participation Choice. However, manager cost expectations may provide a better model of these Choices. To test the relationship between manager cost expectations and manager position (Di) and participation choice (Vi), the following empirical manager expectations model was used. D- = a + 81(01 + CZ + C4) + 82C3+ B3C5 (6.3.1) V- = a + 81(C1 + CZ + C4) + 82C3+ 8305 (6.3.2) A manager’s position and participation Choice on the ED were hypothesized to be related to the manager's expectations of political (C1), labor (C2), debt (C3), administrative compliance (C4), and management compensation costs (C5). As shown in Table 5C, political, labor and administrative compliance costs were found to be highly correlated. Therefore, in the tests of construct validity, one proxy (administrative compliance costs) was used for all of these costs. An alternative approach would have been to sum the costs and treat them as one variable. This would be an acceptable alternative since the costs are each measured in the same units. To more fully utilize the data Obtained for the study, the latter approach was used in the probit analyses on manager position and participation Choice. 164 The results from the analyses of manager position and participation Choice are discussed in Sections 6.3.1 and 6.3.2, respectively. These results indicate that the model described in equation 6.3.1 did par provide a good predictor of position choice. However, the model described in equation 6.3.2 provided a good predictor Of manager participation Choice. 6.3.1 Position Choice The following statistical hypotheses on manager position Choice were derived from equation 6.3.1. (See hypotheses 1.0 and 1.1 through 1.5 in Section 3.6.1.) H2_0: The model described in equation 6.3.1 may be used to predict a manager’s position Choice on the ED as evidenced by a significant Chi- square on the probit analysis. x2 > 0 H0: x2 = 0 H2.1: A manager’s position Choice on the ED (Di) is inversely related to expected increases in political, labor, and administrative compliance costs (C1 + C2 + C4). 31 < 0 H0: Bl = 0 H2.2: A manager’s position choice on the ED (Di) is inversely related 165 to expected increases in debt costs (C3). 52 < 0 H0: 52 = 0 H2 3: A manager’s position Choice on the ED (Di) is inversely related to expected decreases in management compensation costs (c5).20 53 > 0 H0: 33 = 0 The sample used in this study to test these hypotheses on the position Choice of nonfiling managers included 42 Opposers and 36 supporters. (See Table 6B for detail on sample size reduction.) Analyses for the position Choice Of comment letter filers were not performed using manager cost expectations because a representative sample Of ED supporter/filers was not obtained. The results of the probit analyses and Mann—Whitney U tests which were run for ED nonfilers, are summarized in Tables 6J and 6K, respectively. As shown in those tables, the manager expectations model of position Choice did not provide a good fit for its data. Further, no significant relationships were found between individual manager cost expectations and position Choice. The lack of significance for the manager expectations model Of position Choice may reflect weaknesses in the reliability of these measures.21 As discussed in Chapter 5, individual managers may have been unwilling or unable to provide reliable estimates of the 20In the questionnaire, managers were asked if the ED would result in a decrease in management compensation (see Appendix B). 21A coefficient alpha of .17 was computed on the measures Of manager cost expectations. A coefficient of .50 or .60 is sufficiently reliable according to Nunnally (1967). 166 TABLE 6J PROBIT ANALYSIS MANAGER COST EXPECTATIONS AND POSITION CHOICE COMMENT LETTER NONFILERS Manager Cost Expectations Beta Coefficientsab Constant -.1244 (.22) Political, labor .0016 & admin comp costs (.22) (C1 + c2 + C4) Debt costs -.0002 (C3) (.26) Management comp -.0291 costs (c5) (.41) Number of supporters 35 Number Of Opposers 42 Chi-square statistic 1.69 (Significance) (.638) Correctly predicted 55% aAn inverse relationship was hypothesized between manager position choice and each contracting and monitoring cost. bOne-tailed probability of t-statistic in parentheses. 167 TABLE 6K MANN-WHITNEY U TESTS MANAGER EXPECTATIONS AND POSITION CHOICE COMMENT LETTER NONFILERS Manager Meansa Expectations Opposers Supporters U-testb Public relations costs $3,710 $6,000 .335 (C1 + 02) Debt costs $227,120 $1,940C .405 (C3) Administrative costs $23,770 $35,330 .352 (C4) Internal management $16,510 $15,750 .478 (C4) Management compensation .39% .35% .312 (05) Number Of Opposers 42 Number Of supporters 35 aMany corporate managers reported zeros for expectations Of contracting and monitoring costs. Therefore, the reported means may be misleading in terms Of differences. The Mann-Whitney U test is robust with respect to these nonnormal distributions. bOne-tailed probability Of Mann-Whitney U test. CReflects ommission Of one outlier. 168 corporate costs associated with the ED. Therefore, the large amount Of random error in the manager cost expectation measures may preclude significant findings even if a significant relationship does exist. The lack Of significance for the manager expectations model Of position Choice may also reflect an underspecification Of the empirical model used for testing. In Chapter 3, it was asserted that a manager would support an accounting pronouncement if it included the accounting method he was currently using. Otherwise, the manager would estimate the costs Of a set of alternative accounting policies which the FASB might find acceptable. If the costs associated with the current pronouncement were less than those Of the set Of possible alternatives, the manager would support that pronouncement. Alternatively, if lower costs were associated with any Of the set of possible alternatives, he would oppose the pronouncement. Manager expectations of the costs associated with each alternative accounting policy were not Obtained from the questionnaire. It was assumed that the greater the expected costs associated with the ED, the more likely these costs would exceed the costs associated with the set of possible alternatives. However, managers reporting high expected costs on the ED may have even higher cost expectations for the set of possible alternatives. In this case, the "greater cost" assumption may not hold, and significant relationships between manager cost expectations and position Choice may not be found. The manager expectations model Of position choice for 77 nonfilers did not provide as good a fit for the data as the corporate attributes model of 75 nonfilers in terms Of the Chi-square test. The 169 positive results Of the tests for construct validity, and the significance Of the corporate attribute position Choice model suggest that corporate attributes may be reliable and valid proxies for contracting and monitoring costs. Further, the results Of the tests for reliability, and the insignificance Of the manager expectations model indicate that reliable measures of manager cost expectations were not Obtained for this study. Additional research is needed to develop alternative, reliable methods for measuring expected contracting and monitoring costs. These measures could be used to provide additional evidence about the construct validity of corporate attribute proxies. 6.3.2 Participation Choice Manager expectations of the costs associated with the ED were hypothesized to affect manager participation Choice a; they were significant factors in the manager’s position choice.22 Since a significant relationship was not found between manager cost expectations and manager position Choice it may be argued that no relationship should be found for participation Choice. However, one explanation for the lack of significant findings on the model of manager position Choice was that manager cost expectations were indicative of the importance Of a mandated Change in pension accounting. If these stated cost expectations reflect the importance of the pension accounting issue to managers, they may affect their 221t was hypothesized that a manager who Opposed the ED because Of the costs associated with it may be more likely to file a comment letter if those costs are high. 170 decision to lobby. Therefore, the following statistical hypotheses on participation Choice were derived from equation 6.3.2. (See hypotheses 2.0 and 2.1 in Section 3.6.2 for further discussion.) H3 0: The model described in equation 6.3.2 may be used to predict a manager’s participation choice on the ED as evidenced by a significant chi-square on the probit analysis. x2 > 0 H0: 12 = 0 H3 1: The participation Choice (Vi) of an ED Opposer or supporter is positively related to the sum of the political, labor and administrative compliance costs associated with the ED (c1 + C2 + C4). 51 > 0 H0: Bl = 0 H4 2: The participation Choice (Vi) of an ED Opposer or supporter is positively related to the debt costs associated with the ED (C3). 02 > 0 H0: 52 = 0 H4 3: The participation Choice (Vi) of an ED Opposer or supporter is positively related to the management compensation costs associated with the ED (C5). B3 > 0 H0: B3 = 0 The sample used to test hypotheses 4.0 through 4.3 on the participation Choice of ED opposers included 28 filers and 42 nonfilers. Analyses for the participation Choice of ED supporters 171 were not performed using manager cost expectations because a representative sample of ED supporter/filers was not Obtained. The results of the probit analyses and Mann—Whitney U tests are summarized in Tables 6L and 6M, respectively. As shown in Table 6L the participation Choice model presented in equation 6.3.2 provided a good fit for the data. The Chi-square statistic on the probit analysis was significant at the .0135 level, and seventy-three percent Of the managers’ participation Choices were correctly predicted. As hypothesized, the total of expected public relations and administrative compliance costs were significantly and positively related to manager participation choice in the probit analyses. However, as shown in Table 6M, significant differences were not found between comment letter filers and nonfilers on the individual cost expectations. Only the expected administrative compliance costs of ED filers were greater than those Of nonfilers (at the .10 level). The findings from the Mann-Whitney U tests suggest that manager expectations of political and labor costs may not have been as significant as administrative compliance costs in the manager's decision to file. A review of the comment letters filed with the FASB supported this finding. The most consistent point of disagreement on the ED was the magnitude of disclosure requirements.23 The manager cost expectations model Of participation Choice provided as good a fit for its sample of 70 ED Opposers as the corporate attributes model provided for its sample of 69 ED Opposers. Further, political, labor and administrative compliance costs were 23Ninety-four percent Of the managers included in this study, who had filed comment letters, specifically mentioned the "excessive disclosure requirements" of the ED. 172 TABLE 6L PROBIT ANALYSIS MANAGER COST EXPECTATIONS AND PARTICIPATION CHOICE ED OPPOSERS Manager Cost Expectations Beta Coefficientsab Constant -.4955 (.01) Political, labor and .0070 administrative compliance (.02) costs (C1 + C2 + C4) Debt costs (C3) -.0002 (.31) Management -.2317 compensation costs (c5) (.14) Number of filers 28 Number of nonfilers 42 Chi-square statistic 10.69 (Significance) (.0135) Correctly predicted 73% aA positive relationship was hypothesized between manager participation Choice and manager cost expectations. bOne-tailed probability Of t-statistic is in parentheses. 173 TABLE 6M MANN-WHITNEY U TESTS MANAGER COST EXPECTATIONS AND PARTICIPATION CHOICE ED OPPOSERS Manager Meansa Expectations Nonfilers Filers U-Testb Public relations costs 3,710 76,250 .281 (C1 + C2) Debt costs 227,120 360 .460 (C3) Administrative costs 23,770 77,680 .077 (04) Internal management 16,510 398,390 .464 costs (C4) Management compensation .39% .19% .488 costs (C5) Number of nonfilers 42 Number Of filers 28 aMany corporate managers reported zeros for expectations of contracting and monitoring costs. Therefore, the reported means may be misleading in terms of differences. The Mann-Whitney U test is robust with respect to these nonnormal distributions. bOne-tailed probability of Mann-Whitney U test. 174 significant in the manager expectations model just as firm size was significant in the corporate attributes model. These findings suggest that a manager’s stated expectations of the costs associated with the ED may be indicative of the general importance of the accounting issue to the manager. In this case, these costs could be significant in the participation but not the position Choice model. 6.4 Manager Perception Models In Chapter 3, a manager’s participation choice on the ED was hypothesized to be a function of corporate contracting and monitoring costs apg manager perceptions Of the FASB. The relationships between contracting and monitoring costs, manager perceptions Of the FASB, and manager participation choice were examined in this study in three parts. First, a model Of participation choice which included only manager perceptions of the FASB was tested. This model was analyzed to provide a comparison with the corporate attributes and manager expectations models of participation choice. In addition, a model which included both corporate attributes and manager perceptions, and a model which included both manager cost expectations and manager perceptions were tested. These models reflect the total participation Choice model presented in Chapter 3. The results of these analyses, which address only the ED opposer, suggest that manager perceptions Of the FASB may be significant factors in an Opposing manager’s decision to file a comment letter. A 175 discussion of these results are provided in Sections 6.4.1 through 6.4.3. 6.4.1 Manager Perceptions Of the FASB TO test the relationship between manager perceptions of the FASB and manager participation Choice (Vi), the following empirical model was used. vi = a + blprl + sz + b3pr2 + 341 (6.4.1) Four manager perceptions were hypothesized to affect a manager’s participation Choice. These perceptions included the probability of influencing the policy outcome (prl), the costs Of a Change in the accounting standard-setting body (j), the probability Of affecting the FASB’s continuance (pr2), and the costs Of filing a comment letter (1). The following statistical hypotheses on manager participation Choice were derived from equation 6.4.1. (See hypotheses 2.5 through 2.8 in Section 3.6.2 for further discussion.) H5.0: The model described in equation 6.4.1 may be used to predict a manager’s participation choice on the ED as be evidenced by a significant Chi-square on the probit analysis. 12 > 0 H0: x2 = 0 176 H5.1: A manager’s participation choice (Vi) is positively related to his perception Of the probability of influencing the policy influence (Pr1)- [31>0 H0: 151:0 H5 2: A manager's participation Choice (V1) is positively related to his expectations Of the costs associated with a Change in the standard-setting process (j). 52 > 0 H0: 52 = 0 H5_3: A manager's participation Choice (Vi) is positively related to his perception of the probability Of affecting the continuance Of the FASB by filing a comment letter (pr2) B3 > 0 Ho: B3 = 0 H5 4: A manager’s participation Choice (Vi) is inversely related to his perception Of the costs Of filing a comment letter (1). B7 < 0 H0: b7 = 0 The measures used for manager perceptions Of the FASB were described in Table 4C. Two measures were Obtained for the probability Of influence and the costs of filing. Additionally, three measures were Obtained for the costs Of a Change in the standard-setting body, and one measure for the probability Of affecting the FASB's continuance. Twelve probit analyses were performed using different 177 combinations of the measures for each manager perception.24 The results of those analyses are reported in Table 6N and Appendix E. Each analysis provided a good fit for the data. The least significant Chi-square was significant at the .0038 level. Further, the models from these analyses correctly predicted from 60 tO 73 percent Of manager participation choices. The results Of the manager perceptions analysis which provided the best fit Of the data are shown in Table 6N. This analysis included the following measures: (1) the manager's perception Of his ability to influence the ED outcome, (2) his general perception of the costs of filing, (3) his decision tO contribute to the Financial Accounting Foundation, and (4) his perception of the probability Of affecting the FASB’s continuance by filing a comment letter. As shown in Table 6N, significant relationships were not found between individual manager perceptions and participation Choice. In fact, no significant beta coefficients were reported in any Of the analyses. However, as reported in Table 60, a significant difference between filers and nonfilers was found on three variables using a Mann-Whitney U test - filed to support the FASB (j), did not file since it would not help (prl), and did not file due to lobby costs (1). The contradictory findings on individual manager perceptions may be indicative Of an interaction between these factors in the decision process. 24There were three measures Obtained for LRPV, one measure for probability of affecting the FASB’s continuance, and two measures each for lobby costs and probability of influence. Therefore, there were twelve (3*1*2*2 = 12) separate analyses run. 178 TABLE 6N PROBIT ANALYSIS MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE ED OPPOSERSa Manager Perceptions Beta CoefficientbC Constant .4301 (.16) Probability Of influencing -.4847 the ED’s outcome (prl) (.49) Decision to contribute to FAF (j) -.6191 (.09) Probability Of influencing .0132 the FASB’s continuance (pr2) (.12) TOO costly to influence the -.3569 FASB on any issue (1) (.49) Number of filers 24 Number Of nonfilers 41 Chi-square 25.32 (Significance) (.0000) Correctly predicted 73% aThis Table includes the results of one of twelve models - the one which provided the best data fit. See Appendix E for the other eleven models. bAn inverse relationship was hypothesized between manager participation Choice and costs of filing. A positive relationship was hypothesized for LRPV and probability Of influence. cOne-tailed probability Of t-statistic in parentheses. 179 TABLE 60 MANN-WHITNEY U TESTS MANAGER PERCEPTIONS AND PARTICIPATION CHOICE ED OPPOSERS Manager Perceptions Means Nonfilers Filers U Testab LRPV (E122 ( 1.)) Probability of LRPV 8.88 15.00 .094 Part I - FAF .67 .67 .500 Part I - Shift to Gov’t 12.52 41.93 .357 Part II - Support 3.35 9.97 .028 Probability of Influence (Prl) Part I — Not responsive 10.54 3.44 .311 Part II - Wouldn’t help 17.61 0.00 .000 Costs of Filing (l) Part I - TOO costly 25.80 0.00 .000 Part II — Due to cost 18.17 .16 .017 Number Of nonfilers 41 Number of filers 24 aAn inverse relationship was hypothesized between manager participation choice and costs of filing. A positive relationship was hypothesized for LRPV and probability of influence. bOne-tailed probability of Mann—Whitney U Test in parentheses. CNO Mann-Whitney U statistic could not be computed because zeros reported by all filers. 180 An examination of the beta coefficients on the variables in the analyses reveals that these coefficients were in the hypothesized direction except for the coefficient on the manager's decision to contribute to FAF. The negative coefficient on FAF, which was significant at the .10 level, suggests that the manager Of a company making contributions to FAF was less likely to file a comment letter on the ED than the manager Of a company not making contributions. One explanation for this finding is that a manager who supports the FASB may not believe that its continuance will be positively affected by filing an opposing letter. In this case, a negative relationship between FAF contributions and the participation Choice of an ED Opposer may follow. The manager perceptions model Of participation Choice provided a good fit for data on Opposing managers even though no significant relationships were found for individual variables. In fact, the analysis presented in Table 6N performed as well as the corporate attributes and manager expectations models of participation Choice, as shown in Table 6P. These findings suggest that manager perceptions Of the FASB may be as important in a manager’s decision to file a comment letter as the contracting and monitoring costs associated with that particular accounting issue. 6.4.2 Corporate Attributes and Manager Perceptions To test the relationships between corporate attributes, manager perceptions, and manager participation Choice for ED Opposers, the 181 TABLE 6P AN OVERALL COMPARISON OF THE CORPORATE ATTRIBUTES, MANAGER EXPECTATIONS AND MANAGER PERCEPTIONS MODELS OF PARTICIPATION CHOICE ED OPPOSERS Overall Corporate Manager Manager Statistics Attributes Expectations Perceptions NO. of filers 37 28 24 NO. Of nonfilers 41 41 41 Total samplea 14 14 a; Chi-square 26.29 10.69 25.32 (Significance) (.0000) (.0135) (.0000) Correctly predicted 72% 73% 73% aSee Table 6B for detail on sample sizes. 182 following empirical model was used. Vi = a + Blfl + Bzfz + B3f3 + 54f4 + 85(pr1) + B6(j) + 87(pr2) + B8(l). (6.4.2) The statistical hypotheses on manager participation choice derived from equation 6.4.2, were presented above in hypotheses 2.1 through 2.4 (Section 6.2.2) on corporate attributes, and in hypotheses 5.1 through 5.4 (Section 6.4.1) on manager perceptions. Additionally, the following overall statistical hypothesis is tested. H6.0: The model described in equation 6.4.2 may be used to predict a manager's participation choice on the ED as evidenced by a significant chi-square on the probit analysis. X2>0 Hozx2=0 As discussed above, two measures were Obtained for the probability Of influence and the costs Of filing. Additionally, three measures were Obtained for the costs of a Change in the standard- setting body and firm size. Finally, one measure was Obtained for the probability of affecting the FASB's continuance, firm leverage, and labor intensity. Thirty-six probit analyses were performed using different combinations Of the measures for each variable.25 The results Of those analyses are reported in Table 60 and Appendix F. 25There were three measures Obtained for firm size. Additionally, there were three measures Obtained for LRPV, one measure for probability Of affecting the FASB’s continuance, and two measures each for lobby costs and probability Of influence. Therefore, thirty- six (3*12 = 36) separate analyses were run. 183 TABLE 6Q PROBIT ANALYSIS CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE ED OPPOSERSa Corporate Attributes and Manager Perceptions Beta CoefficientbC Constant -.4477 (.31) Firm size (f1) .0030 (assets) (.001) Labor intensity (f2) -.0503 (employees/sales) (.14) Firm Leverage (f3) .3601 (debt/equity) (.08) Pension plan status (f4) -.1818 (pension benefits/pension assets) (.41) Probability of influencing -.1162 the ED’s outcome (prl) (.50) Decision to contribute to FAF (j) -l.3523 (.02) Probability of influencing .0278 the FASB’s continuance (pr2) (.03) TOO costly to influence the -.4444 FASB on the ED (1) (.50) aThis table includes the results Of one of 36 models - the one which provided the best data fit. See Appendix F for other 35 models. bAn inverse relationship was hypothesized between manager participation Choice and firm leverage and between participation Choice and costs of filing. A positive relationship was hypothesized for firm size, labor intensity, pension plan status, FAF, and probability of influence. cOne-tailed probability of t-statistic in parentheses. 184 TABLE 6Q (Continued) Number Of filers Number of nonfilers Chi-square (Significance) Correctly predicted 20 35 39.62 (.0000 89 ) % 185 Each analysis provided a good fit for the data. The least significant Chi-square was significant at the .0068 level. Further, the models from these analyses correctly predicted from 72 to 89 percent Of manager participation Choices. The results Of the analysis which provided the best fit Of the data, and the highest prediction rate, are reported in Table 6Q. This analysis included corporate attributes - assets, debt/equity, employees/sales, and pension benefits/pension assets. In addition, that analysis included the following manager perceptions: 1) probability Of influencing the ED’s outcome (stated in negative terms in the questionnaire), 2) decision to contribute to the Financial Accounting Foundation 3) the probability Of affecting the FASB's continuance by filing a comment letter, and 2) the costs associated with filing a comment letter on the ED. The models Of participation Choice, which included both corporate attributes and manager perceptions (the total model), correctly predicted a higher percentage of manager participation Choices than the models which included only corporate attributes or manager perceptions. Further, the findings on the individual variables were relatively consistent between these models. A significant positive relationship was found between firm size (assets) and participation choice in the total model, as found in the corporate attributes model. Further, FAF contributions were found to be negatively related to manager participation Choice, as found in the manager perceptions model. However, statistically significant relationships were found in the total model between participation 186 Choice and two manager perceptions — FAF contributions, and the probability Of affecting the FASB’s continuance. These factors were only marginally significant in the manager perceptions model. These differences in significance on individual variables in the total and the manager perception models may reflect the different samples used to test each model. The results from the total model Of participation Choice provides evidence about the relationships between corporate attributes and manager perceptions in a manager's participation Choice. The inclusion Of corporate attributes and manager perceptions in the same model increased the correct prediction rate by an average of 23 percent. Correct prediction rates Of 72 and 73 percent were reported for the corporate attribute and manager perception models, respectively. The correct prediction rate on the total model Of 89 percent represents an increase of 24 and 22 percent over these rates, respectively.26 These findings suggest that corporate attributes and manager perceptions provide differentiable information about manager participation Choice (for an ED opposer). Further research is needed to develop a stronger theory about the relationships between manager perceptions and participation choice and to develop stronger measures. The results of that research could be used to provide better predictions Of manager participation Choice. Further, evidence may be provided about relationships between manager perceptions and corporate attributes. Such evidence could be used to assess the discriminant 26The 24 percent increase in predictability was computed as follows: (89%-72%)/72% = 24%. 187 validity Of corporate attributes as proxies for contracting and monitoring costs. 6.4.3 Manager Cost Expectations and Perceptions The relationships between corporate attributes, manager perceptions, and manager participation choice for ED Opposers, were tested and the results were reported in Section 6.4.2. The inclusion of both sets of explanatory variables provided an increase in correct prediction rates of about 23 percent. TO test the relationships between manager cost expectations, perceptions and participation Choice for ED Opposers, the following empirical model was used. Vi = a + 61(C1 + C2 + C4) + 32C3 + B3C5 +B4(pr1) + 35(j) + 86(pr2) + 37(1) (6.4.3) The statistical hypotheses on manager participation Choice derived from equation 6.4.3, were presented above in hypotheses 4.1 through 4.4 (Section 6.3.2) on manager cost expectations, and in hypotheses 5.1 through 5.4 (Section 6.4.1) on manager perceptions. Additionally, the following overall statistical hypothesis is tested. H7.0: The model described in equation 6.4.3 may be used to predict a manager’s participation Choice on the ED as evidenced by a significant 188 chi-square on the probit analysis. X2 > 0 Ho: x2 = 0 Twelve probit analyses were performed using different combinations Of the measures for each manager perception variable.27 The results Of those analyses are reported in Table 6R and Appendix G. Each analysis provided a good fit for the data. The least significant Chi—square was significant at the .0015 level. Further, the models from these analyses correctly predicted from 69 to 84 percent of manager participation Choices. The results of the analysis which provided the best fit of the data, and the highest prediction rate, are reported in Table 6R. This analysis included the following manager perceptions: 1) probability Of influencing the ED's outcome, 2) decision to contribute to the Financial Accounting Foundation (FAF), 4) the probability of affecting the FASB’s continuance by filing a comment letter, and 2) the general costs associated with filing a comment letter. The models Of participation Choice which included both manager cost expectations and manager perceptions (the total model) correctly predicted a higher percentage Of manager participation Choices than the models which included only manager cost expectations or manager 27Only one measure was Obtained for each manager cost expectation. There were three measures Obtained for LRPV, one measure for probability of affecting the FASB’s continuance, and two measures each for lobby costs and probability of influence. Therefore, there were twelve (1*1*1*3*1*2*2 = 12) separate analyses run. 189 TABLE 6R PROBIT ANALYSIS MANAGER COST EXPECTATIONS, PERCEPTIONS OF THE FASB, AND PARTICIPATION CHOICE ED OPPOSERSa Manager Cost Expectations and Manager Perceptions Beta Coefficientbc Constant Public relations and administrative compliance costs (C1 + C2 + c4) Debt costs (C3) (employees/sales) Management compensation costs (C5) Probability Of influencing the ED’s outcome (prl) Decision to contribute to FAF (j) Probability of influencing the FASB’s continuance (pr2) TOO costly to influence the FASB on any issue (1) .2779 (.47) .0077 (.02) .0002 (.34) .3546 (.34) .4495 (.50) .7980 (.06) .0158 (.10) .3338 (.49) aThis table includes the results of one of 12 models which provided the best data fit. bAn inverse relationship was hypothesized between manager participation Choice and costs of filing. hypothesized for expected costs, FAF, and probability of influence. A positive relationship was cOne-tailed probability of t-statistic in parentheses. aThis table includes the results of one Of 12 models provided the best data fit. - the one which See Appendix E for other 11 models. - the one See Appendix G for other 11 models. 190 TABLE 6R (Continued) Number Of filers Number Of nonfilers Chi—square (Significance) Correctly predicted 20 37 32.37 (.0000) 84% 191 perceptions. Further, the findings on the individual variables were consistent between these models. A significant positive relationship was found in the total model between manager participation Choice and the manager's expectations of public relations and administrative compliance costs. This same relationship was found in the manager expectations model. Further, FAF contributions were found to be negatively related tO manager participation Choice, as found in the manager perceptions model. The results from the total model Of participation choice provide evidence about the relationships between manager cost expectations and perceptions in a manager’s participation Choice. The inclusion of manager cost expectations and manager perceptions in the same model increased the correct prediction rate by approximately 13 percent. A correct prediction rate of 73 percent was reported for both the manager expectations and the manager perception models as compared to an 84 percent rate for the total model. These findings suggest that manager cost expectations and perceptions provide differentiable information about manager participation Choice (for an ED Opposer). However, the 13 percent increase in correct predictions for the total manager cost expectations/perceptions model is relatively smaller than the 22 percent increase in the correct predictions for the total corporate attribute/perception model. These findings suggest that corporate attributes may provide more differentiable information than manager cost expectations with respect to manager perceptions of the FASB. Nevertheless, the 13 percent increase in predictive ability for the total manager expectations/perceptions model provides further evidence 192 that manager perceptions Of the FASB may be important, predictive factors in a manager’s participation Choice. 6.5 Summary The overall findings Of the probit analyses on manager position and participation Choice are presented in Table 68. A discussion Of these findings is also summarized below. A replication Of prior research efforts on manager position and participation Choice was provided on the pension accounting issue. A corporate attribute model of position Choice was tested on the sample Of managers who had filed comment letters with the FASB. Additionally, a corporate attribute model of lobby Choice was tested on a random sample of 257 NYSE firms. The results Of these tests were consistent with the results reported by prior researchers on the GPLA, interest capitalization, and foreign currency translation issues. Significant relationships were found only between manager participation Choice and firm size. Further, only the manager participation choice model provided a good fit for the data, as evidenced by a statistically significant chi—square. As an extension Of prior research, models of position Choice for nonfiling managers, and separate models Of participation choice for ED supporters and Opposers were tested. A significant, positive relationship was found between the position Choice Of a nonfiling manager, and the size and leverage of the firm he represented. Additionally, significant, positive relationships were found between 193 TABLE 68 SUMMARY OF FINDINGS ON PROBIT ANALYSES Corporate Attributes Filers Nonfilers Total sample Opposers only Supporters only Manager Cost Expectations Nonfilers Opposers Manager Perceptions Opposers SIGNIFICANT FACTORS IN ANALYSES OF MANAGER POSITION CHOICE PARTICIPATION CHOICE Nonea Firm Size (-) Leverage (-) Firm size (+) Firm size (+) Firm size (+) Political, labor & admin costs (-) Political, labor & admin costs (-) Noneb aThe overall Chi-square statistic was not significant and there were no significant relationships found between manager position Choice and the individual hypothesized variables. bThe overall Chi-square statistic was significant but there were no significant relationships found between manager participation Choice and the individual hypothesized variables. 194 firm size and the participation choices Of both ED Opposers and supporters. These findings may indicate that the position Choice Of a filing manager differs from that Of a nonfiler. However, it may also reflect the power of the analyses used to test the models. The sample used to test the model for comment letter filers included only 6 ED supporters and 37 Opposers. Whereas, the sample used to test the model for nonfilers included 33 ED supporters and 41 opposers. The small sample Of filing managers may have reduced the power of that test. The consistency Of the findings on firm size in the participation Choice models Of ED Opposers and supporters suggests that there may be no need to differentiate between the two groups when modeling participation choice. However, these findings do not support the hypothesized relationship between firm size and participation choice. They suggest that a manager who faced high political and administrative compliance costs due to the ED’s adoption was more likely to file a supportive letter than a manager who faced low costs. One explanation for this finding, which is consistent with the theory underlying the firm size proxy, is that the costs associated with the ED may be indicative of the general importance of pension accounting to the manager. In this case, a manager who faces high costs due to any Change in pension accounting may be more likely to file a comment, regardless of his position, than a manager who faces lower costs. A further extension of prior research was provided by Obtaining measures of manager cost expectations. Models Of position Choice for nonfiling managers and models of participation choice for ED Opposers were tested. These models did not provide a better fit for the data 195 than those models which included corporate attributes as proxies for contracting and monitoring costs. This may reflect weaknesses in the reliability Of the measures Obtained for manager cost expectations. If the primary purpose of research on manager behavior in the accounting standard-setting process is to predict manager behavior, future efforts to Obtain manager cost expectations may not be useful. The evidence from this study suggests reliable measures of these expectations may not be obtainable. However, if the purpose Of this line of research is to explain manager behavior, further research on manager expectations of the costs associated with proposed accounting Changes is needed. The theory underlying the use Of corporate attributes in models of manager position and participation choice is that managers assess the expected costs of an accounting Change. Further, that assessment is used to make a position and participation decision. If a link cannot be established between corporate attributes and manager cost expectations, the models derived may be used only to predict manager behavior. Finally, manager perceptions of the FASB were included in models Of participation Choice for ED Opposers. A model was tested which included only manager perceptions. Additionally, models were tested which included both manager perceptions and expected contracting and monitoring costs (total models). The manager perceptions model Of participation Choice provided a good fit for the data. Further, the total models provided improvements in predictability over the individual manager perceptions, manager cost expectations, and corporate attributes models of participation Choice. These findings 196 suggest that manager perceptions Of the FASB are important factors in a manager’s decision to file a comment letter. Chapter 7 contains a discussion of the implications of this study. Additionally, limitations of the study are addressed. CHAPTER 7 CONCLUSION AND LIMITATIONS 7.1 Introduction The objective of this study was to add evidence to the existing body of knowledge about the behavior of corporate managers in the accounting standard-setting process. Prior researchers have addressed the relationship between manager position and participation Choices and corporate attributes asserted to proxy for economic consequences (contracting and monitoring costs) associated with a mandated Change in accounting. However, these researchers have not provided generalizable evidence about the corporate population taken as a whole. They have examined position choices only for comment letter filers. Further, on participation choice, they have performed analyses on samples which included the population of filers and a sample of nonfilers without controlling for position Choice. This study provided a replication of prior research efforts by applying the same research methodologies to the pension accounting issue (the ED). Corporate attributes were used to model the position Choice of a filing manager and the participation Choice of a randomly selected sample of managers which included both ED Opposers and supporters. Additionally, several extensions Of prior research efforts were offered. The position Choice of a nonfiling manager and the participation Choice of an ED Opposer and an ED supporter were examined separately. Finally, manager expectations Of the costs 197 198 associated with the ED and manager perceptions Of the financial accounting standard-setting process were measured and included as independent variables in the models of position and participation choice. The results from these analyses are summarized in the current section. Limitations of the study and their effect on the findings are discussed in Sections 7.2 and 7.3. 7.1.1 Position Choice The findings from this study on manager position Choice may be summarized as follows: (1) The model for filing managers, which included corporate attributes as explanatory variables was pp; significant. (A model of position Choice for filing managers was not analyzed using manager cost expectations due to the small sample of ED supporter/filers for whom data was available.) (2) The model for nonfiling managars, which included corporate attributes was significant. Firm size and leverage were negatively related to manager position choice. (3) The model for nonfiling mana ers, which included manager cost expectations was not significant. 7.1.2 Participarion Choice The findings on manager participation Choice may be summarized as follows: 199 (1) The models for Opposing and supporting managers, which included corporate attributes as explanatory variables, were each significant. Furthermore, firm size was positively related to a manager's participation choice regardless Of his position choice. (2) The model for opposing managers which included manager cost expectations was also significant. The sum Of expected political, labor and administrative compliance costs was positively related to manager participation choice. (A model of participation Choice for supporting managers was not analyzed using maaager cost expectations due to the small sample of ED supporter/filers for whom data was not available.) 7.2 Conclusions and Limitations 7.2.1 Position Choice and Corporate Attributes The insignificance Of the model of position Choice for comment letter filers, which included corporate attributes, may reflect weaknesses in the measures obtained for the dependent and/or explanatory variables. It may also reflect the small sample size used in the analysis (six ED supporters and 37 Opposers). A manager’s stated position on an accounting proposal may not be his true position. A participant in a political setting, such as the accounting standard—setting process, may have incentives to state a position which is not his true position. In this case, the power Of a study which uses stated positions may be weakened by unreliability or bias in the measurement Of the dependent variable. 200 Further, the use of corporate attributes to proxy for the economic consequences (contracting and monitoring costs) associated with the ED may not be valid. Evidence is provided by this study to suggest that the corporate attributes - firm size and leverage position - have construct (theoretical) validity. However, as discussed in Chapters 4 and 5, the strength of that evidence, as derived from Campbell and Fiske’s multimethOd-multitrait correlation matrix, is dependent upon the number Of proxies examined. The evidence provided by the present study on the construct validity Of corporate attribute proxies is weak since only one alternative proxy was Obtained for each construct. Further, the high correlation between firm size and manager expectations of political, labor and administrative compliance costs may par reflect the firm size hypothesis as set forth in Chapter 3. The correlation may reflect the fact that larger firms face higher costs simply because they are bigger.1 Finally, a small sample, or subsample reduces the power of a study by introducing bias and unreliability. The smaller the sample, the more likely it is to pp; be representative Of the population. When the sample is not representative, a relationship which exists in the population may not be reflected in the sample. Alternatively, a relationship found in the sample may not reflect the population. The position Choice model for comment letter nonfilers (using corporate attributes) utilized a much larger sample size than the analysis of comment letter filers (74 nonfilers versus 43 filers). 1On a relative cost per sales dollar basis, the correlation between firm size and expected political, labor and administrative compliance costs was insignificant. 201 This may account for the significance Of the nonfiler model as compared with the model of position Choice for comment letter filers. An alternative explanation for the difference in significance on the two groups is that the position Choice Of comment letter filers differs from that of nonfilers. The findings from Lasater’s study on the interest capitalization issue together with the results from the present study suggest that the latter explanation may hold. Lasater found a significant relationship between firm leverage and manager position Choice in his analysis of comment letter filers. This finding is consistent with the relationship found between firm leverage and manager position Choice for nonfilers on the pension accounting issue. 7.2.2 Position Choice and Manager Cost Expectations The insignificance Of the position Choice model for nonfilers, which included manager cost expectations, may reflect weaknesses in the measures Obtained for position Choice as discussed above.2 It may also reflect weaknesses in the measures Obtained for the explanatory variables. As discussed in Chapter 5, the manager cost expectation measures Obtained from the questionnaire did not demonstrate reliability. Biases may have also been introduced to the manager cost expectation measures through wording used in the questionnaire. For example, managers were asked if an increase in reported pension expense would result in a decrease in management compensation costs. A decrease in reported pension expense (which will be the ED’s effect 2The analysis included 42 ED opposers and 36 supporters. 202 for most companies) was not addressed. Further, for all other hypothesized economic consequences, managers were asked to estimate increases in costs associated with the ED. Decreases in these costs were not addressed. 7.2.3 Participation Choice, Corporate Attributes and Manager Cost Expectations The consistency Of the models Of participation Choice across ED Opposers and supporters suggests that a manager’s participation Choice may par be dependent upon that manager’s position Choice. Yet, as discussed in Chapter 5, the present study provides evidence that a manager’s position and participation Choice ara dependent. These combined findings suggest that a manager’s position Choica is dependent upon his participation Choice. Stated otherwise, a manager may first decide if he wants to file a comment letter on an accounting proposal. Then he may decide what position to take. The model Of participation choice used in this study was based on the Opposite assumption - that a manager first takes a position on an accounting proposal and then decides whether or not to file. A difference in ordering may not affect the hypothesized relationships between the dependent and explanatory variables in each model. However, it may have implications for prior and future research efforts. If a manager’s participation Choice is independent Of his position Choice, it would not be necessary to control for position Choice when examining participation choice models. Alternatively, if a manager’s position Choice is dependent upon his 203 participation Choice, it would be necessary to control for participation Choice when examining position Choice models.3 7.2.4 Participation Choice and Manager Perceptions Of the FASB The models which included para manager cost expectations or corporate attributes apg manager perceptions of the FASB provided better predictors of manager participation Choice than the individual partitioned models. These findings suggest that the economic consequences associated with the ED and manager perceptions of the FASB were each important factors in a manager’s decision to lobby on that proposal. The insignificance of the relationships between the individua1ymanaqeryperception factors and participation choice may reflect weaknesses in these measures. Although the questionnaire used in the study demonstrated overall reliability, construct validity Of the measures Obtained for manager perceptions of the FASB was not demonstrated. 7.2.5 Nonresponse Bias As discussed in Sections 7.2.1 through 7.2.4, the power of this study was reduced by limitations in the models used for testing and in the measures obtained for the dependent and explanatory variables. 3Prior studies of participation Choice have par controlled for position Choice, but studies of position Choice have controlled for participation Choice (by examining only the position choices Of comment letter filers). 204 The generalizability Of the results was also reduced by biases in the samples which were used for testing. First, inferences from the study are limited to the population of NYSE firm managers. The behavior Of AMEX and OTC managers was not addressed by the present study. Second, although a sample was drawn randomly from the population Of NYSE firms, data was Obtained for only 44% (131 of 300) Of the population. Further, these 131 firms were self-selected in that the managers of these firms Chose to respond to the questionnaire which was used to collect data. Therefore, the sample used may not be representative of the population Of NYSE firm managers. In fact, an assessment Of this nonresponse bias indicated that the sample Of 131 firms included a disproportionate number of comment letter filers (when compared to the total population Of NYSE firms). This bias limits the inferences made from the study to the sample. Stated otherwise, the findings cannot be generalized to the total population of NYSE firm managers. Finally, different samples (and sample sizes) were used for each analysis in the study.4 Therefore, inferences drawn from comparisons between the analyses using corporate attributes, manager cost expectations, and manager perceptions of the FASB as explanatory variables may be subject to bias. Inferences made from comparisons 4The samples used for each analysis included substantially the same firms. However, there was some variation as shown in Table 6B. For example, if a firm manager had filed a comment letter on the ED, that firm was included in the analysis of position Choice using corporate attributes as explanatory variables. However, if that manager had not also responded to the questionnaire, the firm was app included in the analysis of position Choice using manager cost expectations as explanatory variables. 205 between analyses Of position and participation Choice may also be subject to bias. 7.3 Summary The results from the present study are subject to the following limitations: 1) Reliability and validity Of the measures Obtained for the dependent variable - position choice 2) Reliability and validity of the measures Obtained for the explanatory variables - manager cost expectations and manager perceprions of the FASB 3) Validity of the measures Obtained for the explanatory variables - corporate attributes 4) Under and/or mispecification in the models used for testing 5) Nonresponse bias and varying samples and sample sizes. These limitations reduce the power of the study and reduce the generalizability of the inferences which may be drawn. Nevertheless, the findings from the study provide support for the theory that there is a relationship between manager position and participation Choice, and certain corporate attributes, manager cost expectations, and manager perceptions Of the FASB. However, the findings may not be generalized beyond the behavior Of the sample of NYSE firm managers included in the study. Further, relationships found between the dependent variables and the proxies used to measure the explanatory variable constructs may not be generalized to the constructs themselves. APPENDICES APPENDIX A A COMPARISON OF THE FINANCIAL STATEMENT REQUIREMENTS UNDER THE MARCH 1985 EXPOSURE DRAFT AND FASB STATEMENT 36 AND APB OPINION 8 206 APPENDIX A A COMPARISON OF THE FINANCIAL STATEMENT REQUIREMENTS UNDER THE MARCH 1985 EXPOSURE DRAFT AND FASB STATEMENT 36 AND APB OPINION 8 Disclosures Required by the March 1985 Exposure Draft Note 10 - Pension Plans The company has a defined benefit pension plan covering substantially all Of its employees. The plan is noncontributory and provides pension benefits that are based on the employee’s compensation during the three years before retirement. Future years Of service and salary increases are both considered for purposes of funding and pension expense accrual. Net pension expense for 1986, which represents 6.3% of annual compensation for covered employees, is comprised Of the following (in thousands): Normal cost $ 11,652 Amortization Of prior service costs 324 Amortization of excess plan assets (8,308) Amortization of unrecognized actuarial gains or losses 0 Interest cost on projected benefit 76,154 Obligations Return on investments (77,192) $ 2‘630 207 APPENDIX A (cont inued) The following table sets forth the plan’s funded status and amounts recognized in the company’s statement of financial postion at December 31, 1986 (in thousands): Actuarial present value Of benefit Obligations: Vested Nonvested Accumulated benefit Obligations Effect of projected future compensation levels Projected benefit Obligations Plan assets at fair value: $ 515,726 26,055 541,781 267,231 $ 809,012 Common stocks and other equity securities 3 892,092 Plan assets in excess Of projected benefit Obligations At December 31, 1986, $ 83,080 the weighted-average discount rate and rate of increase in future compensation levels used in determining the actuarial present value of the projected benefit Obligation were 9 percent and 6 percent, respectively. significant effect on the amounts reported. Those assumptions can have a To illustrate, increasing the discount rate assumption to 10 percent would have decreased the projected benefit obligation and $2,564,000 respectively, for the Increasing the rate of Change Of percent would have increased the penhsion expense by $183,600,000 year ended December 31, 1986. pension expense by $346,270,000 and year ended December 31, 1986. future compensation levels to 7 projected benefit Obligation and and $2,564,000, respectively, for the The following table presents information regarding the changes in the fair value Of the plan’s assets (in thousands): Plan assets at January 1 S 857,685 Add: Employer contributions 2,630 Income from investments including Change in fair value 93,117 Deduct: Pension benefits paid to participants Plan assets at December 31 61,340 $ 892,092 208 APPENDIX A (continued) The Disclosures Required by FASB Statement 36 Note 10 - Pension Plans The company has a defined benefit pension plan covering substantially all of its employees. Pension expense was $38,368,000 in 1986. A summary of estimated accumulated plan benefits and net assets available for plan benefits for the company’s plan is presented below (in thousands): Vested $ 515,726 Nonvested 26,055 Accumulated benefit Obligations $ 541,781 The discount rate used in determining the present value Of accumulated benefits was 9.0%. A Comparison Of Balance Sheet and Income Statement Effects Balance Sheet Effects: NO pension liability is recorded for the company under APB Opinion 8 or the March 1985 Exposure Draft because plan assets exceed projected benefit Obligations by $83,080. However, if the company had disclosed an excess Of projected benefit Obligations over available plan assets, a liability would have been recorded (in compliance with the March ED) for the amount by which accumulated plan benefits exceed available plan assets. Income Statement Effects (in thousands): Pension expense under APB Opinion 8: $ 38,368 Change in normal and prior service costs due to Change from entry age normal actuarial method to the projected unit credit method (26,392) Amortization of the excess plan assets (8,308) Interest cost on projected benefit Obligations 76,154 Return on investments (77,192) Pension expense under the March 1985 Exposure Draft $ 2,630 APPENDIX B THE COVER LETTERS AND QUESTIONNAIRE 209 APPENDIX B THE COVER LETTERS AND QUESTIONNAIRE FIRST REQUEST COVER LETTER September 9, 1985 Mr. Vice President, Finance Sample Corporation 25 Avenue New York, NY 00000 Dear Mr. Vice President: I am a Ph.D. candidate in accounting at Michigan State University, doing a dissertation on the financial accounting standard-setting process in general, and specifically, the pension accounting issue. Very little research has been undertaken to determine the costs experienced by companies as a consequence Of the financial accounting standard—setting process. I hope that my research will begin to fill that gap. I would appreciate your help in completing the attached questionnaire and returning it in the enclosed envelope by September 24. 1985. It should take approximately thirty minutes of your time. Please feel free to write in any comments or clarifications you may have as to your answers or the questionnaire itself. You may be assured that your responses will be kept strictly confidential. The results of the survey Of 400 executives will be reported in the aggregate and I will not associate your name or your company’s name with any results that are reported. Thank you for your helpful participation! Sincerely, Georgia R. Saemann, MBA, CPA Enclosure 210 APPENDIX B (continued) SECOND REQUEST COVER LETTER October 11, 1985 Mr. Vice President, Finance Sample Corporation 25 Avenue New York, NY 00000 Dear Mr. Vice President: I am concluding the data collection phase of my study on the Financial Accounting Standards Board (FASB) and the costs associated with mandated accounting changes. As Of this date, I have not received a completed questionnaire from you. I am certainly aware Of the value of your time, but I also believe that my study offers an important opportunity to express the views Of your company on the FASB and the current accounting standard-setting process (the pension issue in particular). I would appreciate it very much if you would complete those portions of the questionnaire which you believe are relevant. Also, do not feel restricted by the format of the questionnaire. Additional comments are welcome and will be incorporated into the final report. If you have any further questions or desire Clarification on any aspect Of this study, please call me at (517) 437-7341, extension 423. I will return you call if I’m not immediately available. Thank you again for your time and participation! Sincerely, Georgia R. Saemann, MBA, CPA Enclosure 211 APPENDIX B (continued) QUESTIONNAIRE A QUESTIONNAIRE ON YOUR COMPANY’S PARTICIPATION IN THE FINANCIAL ACCOUNTING STANDARD-SETTING PROCESS AND THE IMPACT OF THE PENSION ACCOUNTING ISSUE ON YOUR COMPANY PART I: YOUR COMPANY'S PARTICIPATION IN THE FINANCIAL ACCOUNTING STANDARD-SETTING PROCESS 1. Does your company* support the FASB through financial contributions to the Financial Accounting Foundation. Yes, we contribute approximately $ per year No 2. DO you believe that by filing a comment letter you increase the probability of the FASB’s continuance? Yes, by about % NO 3. Does your company participate in the activities of the FASB? Yes, we spend approximately man-days per year attending FASB hearings and industry association meetings related to financial accounting standard-setting. This costs our company approximately $ per year. NO *The term "company" means the group of companies whose results are consolidated for financial reporting purposes. 212 APPENDIX B (continued) 4. How does your company attempt to influence the FASB in the accounting standard-setting process? (allocate 100 points*) By direct involvement with the FASB (i.e., attending FASB hearings and industry association meetings) By filing comment letters We do not attempt to influence the FASB because it is too costly We do not attempt tO influence the FASB because we do not believe the Board is responsive to corporate constituents Other (please specify...) *A more comprehensive instruction would read: Please allocate 100 points to the responses below. The points represent the approximate weights or percentage points you would give to each reason for your decision. Example For a leisure weekend trip last January, I Chose to take the four—hour train ride from East Lansing, Michigan to Chicago because: __lQ__ I could do some reading en route The train schedule was good :EQ:EQ Regional airfare was high ._jfl2_ Other reasons (please specify...) J5 don'tJLIQt. 7‘6 {1% 100 Total 213 APPENDIX B (continued) Excluding independent auditor and annual report production costs, what does it cost your company per year for financial accounting, research and external reporting? Between $ and 3 Approximately how are these costs distributed? (allocate 100 points) Reviewing and evaluating the effect Of new FASB pronouncements on our company Preparing comment letters for the FASB and participating in FASB public hearings Other (please specify) 10 Total If financial accounting standard-setting shifted entirely to the governmental sector, what would be the effect on your company's total annual cost of financial accounting, research and external reporting activities? (please Check one answer and fill in the (percent) Increase by at least percent Decrease by at least percent Be about the same 214 APPENDIX B (continued) Before a response can be prepared, a general review or synopsis Of the FASB pronouncement and evaluation Of its impact on your company may be needed. On whom does your company rely to review, evaluate and respond to FASB pronouncements? (allocate 100 points for each process) Review Evaluation Reponse Our own technical staff Auditors Investment bankers Industry associations Other (please specify...) Total 10 10 10 Proposed accounting standards have the potential to affect your company's financial statements and costs of financial reporting. In evaluating the effect Of a proposed standard on your company, which Of the following factors are you likely to consider? (allocate 100 points) Effect on level of reported net income Effect on stability of reported net income Effect on leverage ratios Effect on audit and bookkeeping costs Effect on goodwill Of business colleagues (i.e., pressure from Clients or customers to file a comment letter, taking a prescribed position) Other (please specify...) 100 Total 215 APPENDIX B (continued) PART II: THE IMPACT OF THE PENSION ACCOUNTING ISSUE ON YOUR COMPANY 1. The following questions relate to the position of your company on several aspects of the pension accounting issue. Should pension cost measurement be standardized as to the allocation Of total costs? Should a liability for unfunded accumulated benefits be recognized on the balance sheet? For vested benefits For nonvested benefits Should an asset for overfunded accumulated plan benefits be recognized on the balance sheet? Should the discount rate used to calculate the present value Of pension liabilities be based on current prices for settling the employer’s Obligation? Should the discount rate used to calculate the present value Of pension assets be based on fair value? Should the following disclosures be required? The weighted-average assumed discount rate and rate of compensation increase used The effect on the projected benefit obligation and net periodic pension cost of a one percentage point Change in the discount rate or rate of compensation The components of net periodic pension cost Reconciliation of the funded status of the plan with amounts reported in financial statements 216 APPENDIX B (continued) With respect tO the issue of pension accounting, which factors most affected your decision to file or not file a comment letter on the March 22, 1985, exposure draft on pension accounting? (allocate 100 points) The magnitude of the costs which our company could incur if the March 22, 1985 exposure draft were adopted The potential effect of the March 22, 1985 exposure draft on future accounting issues Pressure from Clients or other business colleagues to file a comment letter taking a prescribed position We had already made our position Clear in previous comment letters Our company’s desire to support the FASB in its rulemaking efforts The cost and effort incurred to research the proposal and prepare a comment letter We did not believe that our participation would have any effect on the standard issued by the FASB We do not support the FASB’s rulemaking efforts in general Other (please specify...) 10 Total DO you believe there will be a change in the stock price (per share) of your company if the March 22, 1985 exposure draft is adopted? Yes, it could increase by as much as $ per share Yes, it could decrease by as much as $ per share NO 217 APPENDIX B (continued) 4. What specific costs do you believe your company may incur if the March 22, 1985 exposure draft on pension accounting is adopted? Please consider both the out-of-pocket and opportunity costs incurred to initiatarcompliance.* NOTE: You may believe that some or all of the following costs are insignificant. Please use a "O" in these cases. Administrative costs (including actuarial and $ audit fees, computer installations and other information processing costs) Public relations costs (including the time s spent to explain your pension plan to employees, investors, creditors or others) Internal management costs (including the time $ spent to reeducate management and to maintain separate records if a different method is to be used for funding than is used for accounting) Existing debt costs (including Changes in $ interest rates on existing debt caused by a violation of indenture requirements) Future debt costs (including Changes in interest $ rates on future debt as a result of a change in bond ratings) Other costs (please specify...) 5 5. Would a reduction in the reported income of your company as a result Of a change in pension accounting affect the compensation Of upper level management or any other profit sharing plan? Yes, total management compensation could decrease by as much as % NO *Out-Of-pocket costs include corporate expenditures for additional employees, materials, etc., and Opportunity costs include the costs of reduced attention to other problems or opportunities. 218 APPENDIX B (continued) IF YOU WOULD LIKE TO RECEIVE A SUMMARY OF THE STATISTICAL RESULTS, PLEASE INDICATE TO WHOM THEY SHOULD BE MAILED: (Name) (Title) THANK YOU FOR YOUR TIME AND CONSIDERATION IN COMPLETING THIS QUESTIONNAIRE. PLEASE RETURN IT IN THE ENCLOSED, SELF-ADDRESSED, STAMPED ENVELOPE. APPENDIX C PROBIT ANALYSES CORPORATE ATTRIBUTES AND POSITION CHOICE 219 APPENDIX C PROBIT ANALYSES CORPORATE ATTRIBUTES AND POSITION CHOICE COMMENT LETTER NONFILERS (41 Opposers and 33 supporters)a Beta CoefficientbC Corporate Analysis 1 Analysis 2 Analysis 3 Attribute (Sales) (Assets) (Employees) Constant .7482 .7629 .5506 (.050) (.045) (.098) Firm size -.00000 -.00000 —.00001 (.066) (.042) (.170) Leverage —.2103 -.2026 -.2084 (.044) (.045) (.043) Labor -.0111 —.0135 .0020 intensity (.302) (.266) (.462) Pension plan -.4100 -.3312 -.4579 benefit/assets (.158) (.212) (.131) Chi-square 8.79 10.10 7.00 (Significance) .0665 .0387 .1360 Predicted 55% 57% 54% aSee Table 6B for detail on sample size reduction. bA positive relationship was hypothesized between manager position choice, firm leverage, and pension plan status. An inverse relationship was hypothesized for firm size and labor intensity. COne-tailed probability Of t-statistic in parentheses. APPENDIX D PROBIT ANALYSES CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE 220 APPENDIX D PROBIT ANALYSES CORPORATE ATTRIBUTES AND PARTICIPATION CHOICE EXPOSURE DRAFT OPPOSERS (37 filers and 41 nonfilers)a Beta CoefficientsbC Analysis 1 Analysis 2 Analysis 3 Factor (Sales) (Assets) (Employees) Constant -.1103 -.3144 .2979 (.413) (.284) (.285) Firm size .00000 .00000 .00003 (.009) (.0005) (.003) Leverage .0009 -.0733 .0031 (.493) (.208) (.475) Labor -.0352 -.O378 -.0851 intensity (.093) (.096) (.003) Pension plan -.3326 -.1750 -.4233 benefit/assets (.265) (.374) (.238) Overall Statistics Chi-square 16.67 26.30 20.13 (Significance) .0022 .0000 .0005 Predicted 70% 72% 65% aSee Table 6B for detail on sample size reduction. bAn inverse relationship was hypothesized between manager participation Choice, firm leverage and pension plan status. A positive relationship was hypothesized for firm size and labor intensity. COne-tailed probability Of t-statistic in parentheses. APPENDIX E PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE 221 APPENDIX E PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (FAF and probability Of affecting the FASB's continuance) ED OPPOSERS (24 filers and 41 nonfilers)a Beta coefficientsbc Manager Analysis 1 Analysis 2 Analysis 3 Analysis 4 Perceptionsd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .2779 .4301 .0232 .1204 (.249) (.160) (.475) (.375) Pr (influence) -.0087 —.4847 -.0103 -.0677 (.134) (.494) (.111) (.500) FAF contribution -.5355 -.6191 -.l615 -.2918 (.107) (.092) (.343) (.245) Pr(LRPV) .0130 .1312 .0086 .0150 (.113) (.117) (.167) (.084) Lobby cost -.3330 -.3569 -.4777 -.4899 (.489) (.493) (.490) (.491) Log likelihood —32.76 -29.21 -33.98 -30.49 Chi-square 18.20 25.32 15.76 22.74 Significance .0011 .0000 .0034 .0001 Predicted 71% 73% 68% 70% 222 APPENDIX E (continued) PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Shift to government and probability Of affecting the FASB's continuance) ED OPPOSERS (26 filers and 36 nonfilers)a Beta coefficientsbc Manager Analysis 1 Analysis 2 Analysis 3 Analysis 4 perceptionsd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant —.0209 .0635 -.0209 -.0120 (.462) (.385) (.460) (.477) Pr (influence) -.0092 -.5297 -.0102 -.0625 (.121) (.494) (.109) (.497) Shift to .0010 .0017 .0018 .0019 government (.302) (.303) (.261) (.314) Pr(LRPV) .0097 .0097 .0057 .0115 (.188) (.189) (.267) (.149) Lobby cost -.3158 -.3346 -.4822 -.4651 (.491) (.499) (.492) (.489) Log likelihood -33.54 —29.72 -34.44 -31.14 Chi-square 17.24 24.89 15.46 22.05 Significance .0017 .0001 .0038 .0002 Predicted 60% 69% 61% 63% APPENDIX E 223 PROBIT ANALYSES (continued) MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Filed comment on ED to support the FASB) ED OPPOSERS (28 filers and 46 nonfilers)a Beta coefficientsbc Manager Analysis 1 Analysis 2 Analysis 3 Analysis 4 Perceptionsd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .0037 .1063 .0156 .0725 (.492) (.278) (.465) (.343) Pr (influence) -.0096 -.5075 -.0114 -.0528 (.109) (.494) (.083) (.489) Filed to .0079 .0060 .0079 .0068 support FASB (.186) (.248) (.185) (.220) Lobby cost -.3152 -.3350 -.1177 -.ll85 (.490) (.493) (.080) (.096) Log likelihood -42.99 —38.82 —43.56 -40.43 Chi-square 19.63 27.96 18.48 24.75 Significance .0002 .0000 .0003 .0000 Predicted 60% 68% 60% 65% aSee Table 6B for detail on sample size reduction. bAn inverse relationship was hypothesized between manager participation choice, Pr Of influence terms), and lobby cost. filed comment to support the FASB. (which was measured in negative A positive relationship was hypothesized for cOne-tailed probability Of t—statistic in parentheses. dMeasures for "pr(influence)/lobby costs". APPENDIX F PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE 224 APPENDIX F PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Sales, FAF and probability of affecting the FASB’s continuance) ED OPPOSERS (20 filers and 35 nonfilers)a Beta CoefficientsbC Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt 1) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .1922 -.0898 .0774 .0088 (.40) (.46) (.46) (.50) Sales .0015 .0017 .0015 .0017 (.02) (.02) (.02) (.02) Debt/equity .0448 .8629 .0272 .2968 ratio (.24) (.02) (.33) (.10) Employees/ .0078 -.0380 .0124 -.0422 sales ratio (.42) (.19) (.37) (.15) Pension .3278 -.9233 .1732 -.3306 benefits/assets (.32) (.16) (.39) (.33) Pr (influence) .0068 -.2890 .0081 -.1089 (.24) (.50) (.21) (.50) FAF contribution .9243 -l.4817 .5628 -1.0493 (.03) (.01) (.11) (.03) Pr(LRPV) .0152 .0241 .0080 .0238 (.12) (.05) (.21) (.05) Lobby cost .3200 -.3l93 .5137 -.5107 (.49) (.50) (.49) (.49) Chi-square 24.33 37.05 21.12 31.87 (Significance) (.0020) (.0000) (.0068) (.0001) Predicted 78% 87% 78% 82% (Sales, APPENDIX F CORPORATE ATTRIBUTES, 225 PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE shift and probability of affecting the FASB’s continuance) (continued) ED OPPOSERS (23 filers and 33 nonfilers)a Beta Coefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .2805 .4742 .3193 -.2829 (.33) (.29) (.30) (.35) Sales .0018 .0019 .0020 .0021 (.02) (.03) (.02) (.02) Debt/equity .0464 .5522 .0263 .1868 ratio (.29) (.06) (.34) (.15) Employees/ .0349 .0668 .0372 .0674 sales ratio (.18) (.06) (.17) (.06) Pension .1051 .4324 .0136 .0657 benefits/assets (.44) (.29) (.49) (.46) Pr (influence) .0057 .2417 .0083 .0867 (.28) (.50) (.21) (.50) Shift to .0114 .0111 .0104 .0114 government (.11) (.15) (.16) (.15) Pr(LRPV) .0184 .0220 .0121 .0221 (.07) (.04) (.11) (.05) Lobby cost .2860 .2772 .4928 .4997 (.50) (.50) (.49) (.50) Chi-square 24.96 35.78 22.53 32.92 (Significance) (.0016) (.0000) (.0040) (.0001) Correctly predicted 75% 80% 77% 77% 226 APPENDIX F (continued) PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Sales, and filed comment on ED to support the FASB) ED OPPOSERS (28 filers and 41 nonfilers)a Beta,Coefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .0804 .4117 -.0224 .2421 (.45) (.25) (.49) (.33) Sales .0012 .0012 .0012 .0012 (.04) (.05) (.03) (.05) Debt/equity .0158 .0804 .0042 .0625 ratio (.39) (.18) (.47) (.21) Employees/ .0248 .0436 .0254 .0483 sales ratio (.20) (.08) (.21) (.07) Pension .3895 .6169 .2562 .3757 benefits/assets (.26) (.17) (.32) (.26) Pr (influence) .0072 .4350 .0075 .0609 (.20) (.50) (.20) (.49) Filed to .0049 .0030 .0055 .0042 support FASB (.30) (.38) (.28) (.33) Lobby cost .3008 .3223 .0698 .0574 (.49) (.50) (.25) (.30) Chi-square 26.01 34.32 23.55 30.75 (Significance) (.0005) (.0000) (.0014) (.0001) Correctly predicted 74% 81% 74% 78% 227 APPENDIX F (continued) PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Assets, FAF and probability of affecting the FASB’s continuance) ED OPPOSERS (20 filers and 35 nonfilers)a Beta Coefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt 1) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant -.2333 -.4331 -.3084 -.4477 (.40) (.33) (.35) (.31) Assets .0031 .0029 .0030 .0030 (.001) (.003) (.001) (.001) Debt/equity -.0820 .8508 -.1224 .3601 ratio (.21) (.03) (.17) (.08) Employees/ -.0323 -.0463 —.0297 -.0503 sales ratio (.24) (.18) (.25) (.14) Pension .0778 -.8106 .0388 -.1818 benefits/assets (.46) (.20) (.48) (.41) Pr (influence) .0064 -.2016 .0033 -.1162 (.28) (.50) (.39) (.50) FAF contribution -1.3065 -1.6946 -.9044 -1.3523 (.01) (.005) (.04) (.02) Pr(LRPV) .0245 .0285 .0139 .0278 (.05) (.04) (.10) (.03) Lobby cost -.3052 —.2984 -.4401 -.4444 (.49) (.50) (.49) (.50) Chi-square 36.21 43.11 32.96 39.62 (Significance) (.0000) (.0000) (.0001) (.0000) Correctly predicted 85% 87% 85% 89% 228 APPENDIX F (continued) PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Assets, shift and probability of affecting the FASB's continuance) ED OPPOSERS (23 filers and 33 nonfilers)a Bera,CoefficientsbC Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant -.7752 —.7876 -.5966 -.6758 (.15) (.19) (.20) (.21) Assets .0032 .0027 .0031 .0030 (.001) (.005) (.001) (.002) Debt/equity -.0499 .4816 -.1277 .1863 ratio (.33) (.10) (.21) (.21) Employees/ -.0591 .0789 .0572 .0808 sales ratio (.11) (.06) (.11) (.06) Pension .2529 .2015 .2534 .1379 benefits/assets (.36) (.40) (.35) (.42) Pr (influence) .0049 .1835 .0018 .0813 (.33) (.50) (.44) (.50) Shift to -.0148 .0137 .0134 .0142 government (.08) (.10) (.11) (.10) Pr(LRPV) .0261 .0264 .0187 .0262 (.02) (.02) (.04) (.02) Lobby cost -.2666 .2830 .4165 .3999 (.50) (.50) (.50) (.50) Chi-square 34.84 40.68 32.88 39.00 (Significance) (.0000) (.0000) (.0001) (.0000) Correctly predicted 82% 79% 84% 82% 229 APPENDIX F (continued) PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Assets and filed comment on ED to support the FASB) ED OPPOSERS (28 filers and 41 nonfilers)a Beta Coefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant —.2829 .0691 -.2362 .0526 (.32) (.46) (.34) (.47) Assets .0022 .0020 .0019 .0017 (.003) (.01) (.002) (.01) Debt/equity -.0809 .0184 -.0905 .0224 ratio (.24) (.46) (.17) (.44) Employees/ -.0243 .0361 -.O311 .0472 sales ratio (.23) (.14) (.17) (.08) Pension -.1570 .3809 -.0883 .2388 benefits/assets (.40) (.28) (.44) (.34) Pr (influence) -.0018 .4271 —.0012 .0614 (.43) (.50) (.45) (.50) Filed to .0032 .0016 .0052 .0040 support FASB (.37) (.44) (.29) (.34) Lobby cost -.2909 .3157 —.0419 .0346 (.49) (.50) (.28) (.32) Chi-square 34.43 39.51 31.13 35.53 (Significance) (.0000) (.0000) (.0001) (.0000) Correctly predicted 81% 84% 75% 81% CORPORATE ATTRIBUTES, APPENDIX F PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Employees, FAF and probability of affecting the FASB’s continuance) 230 (continued) ED OPPOSERS (20 filers and 35 nonfilers)a Beta Coefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .8666 .9880 .5805 .9330 (.14) (.18) (.21) (.15) Number Of .00003 .00003 .00003 .00003 employees (.02) (.03) (.01) (.Ol) Debt/equity .0298 .6259 .0159 .2232 ratio (.31) (.05) (.39) (.14) Employees/ .0772 -.1112 .0842 -.1193 sales ratio (.04) (.02) (.03) (.01) Pension .5961 —1.2121 .3655 -.6646 benefits/assets (.22) (.12) (.30) (.22) Pr (influence) .0045 -.2435 .0061 -.0986 (.33) (.50) (.27) (.50) FAF contribution .9359 —l.4277 .6206 —l.0846 (.03) (.01) (.09) (.03) Pr(LRPV) .0131 .0206 .0062 .0202 (.15) (.07) (.27) (.08) Lobby cost .3232 -.3309 .5380 -.5348 (.49) (.49) (.49) (.49) Chi-square 25.20 36.17 23.15 33.19 (Significance) (.0014) (.0000) (.0032) (.0001) Predicted 76% 87% 76% 84% 231 APPENDIX F (continued) PROBIT ANALYSES CORPORATE ATTRIBUTES, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Employees, shift and probability Of affecting the FASB’s continuance) ED OPPOSERS (23 filers and 33 nonfilers)a Beta Coefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .2297 .2284 .2096 .3414 (.36) (.39) (.36) (.31) Number Of .00003 .00002 .00003 .00003 employees (.02) (.03) (.01) (.02) Debt/equity .0294 .3799 .0131 .1448 ratio (.32) (.13) (.42) (.16) Employees/ .0978 .1204 .1044 .1315 sales ratio (.02) (.01) (.01) (.01) Pension .1625 .4430 .0497 .1535 benefits/assets (.41) (.29) (.47) (.42) Pr (influence) .0038 .2921 .0060 .0817 (.36) (.59) (.28) (.50) Shift to .0059 .0044 .0045 .0045 government (.24) (.32) (.31) (.31) Pr(LRPV) .0159 .0185 .0100 .0183 (.09) (.07) (.15) (.07) Lobby cost .2871 .2948 .5032 .5056 (.49) (.50) (.49) (.49) Chi-square 25.46 35.42 24.20 34.06 (Significance) (.0013) (.0000) (.0021) (.0000) Predicted 77% 84% 75% 82% CORPORATE ATTRIBUTES, (Employees, APPENDIX F (continued) 232 PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE and filed comment on ED to support the FASB) ED OPPOSERS (20 filers and 35 nonfilers)a Beta CoefficientsbC Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .4362 .7537 .3416 .5870 (.22) (.11) (.26) (.15) Number of .00003 .00003 .00003 .00003 employees (.01) (.02) (.01) (.01) Debt/equity .0162 .0993 .0067 .0746 ratio (.39) (.17) (.45) (.19) Employees/ .0709 .0899 .0719 .0944 sales ratio (.02) (.01) (.02) (.01) Pension .4589 .7035 .3237 .4429 benefits/assets (.24) (.17) (.29) (.24) Pr (influence) .0062 .3995 .0067 .0632 (.25) (.50) (.24) (.49) Filed to .0029 .0050 .0025 .0043 support FASB (.39) (.31) (.41) (.34) Lobby cost .3007 .3215 .0791 .0707 (.49) (.49) (.23) (.26) Chi-square 22.23 36.74 25.92 33.73 (Significance) (.0002) (.0000) (.0005) (.0000) Predicted 72% 78% 72% 78% 233 APPENDIX F (continued) aSee Table 6B for detail on sample size reduction. bAn inverse relationship was hypothesized between manager participation Choice, firm leverage, pension plan status, Pr of influence (which was measured in negative terms), and lobby cost. A positive relationship was hypothesized for firm size, labor intensity, FAF, shift to government, and filed to support the FASB. COne-tailed probability of t-statistic in parentheses. dMeasures for "pr(influence)/lobby costs". APPENDIX G PROBIT ANALYSES MANAGER COST EXPECTATIONS, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE 234 APPENDIX G PROBIT ANALYSES MANAGER COST EXPECTATIONS, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (FAF and probability of affecting the FASB’s continuance) ED OPPOSERS (20 filers and 37 nonfilers)a Beta CoefficientsbC Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt 1) (Pt I/Pt II) (Pt II/Pt II) Constant .0454 .2779 .2989 .2993 (.46) (.47) (.26) (.25) Admin & public .0078 .0077 .0083 .0082 relation costs (.02) (.02) (.02) (.02) Debt costs .0002 .0002 .0002 .0002 (.34) (.34) (.34) (.34) Mgmt comp costs .1551 .3546 .1495 .5420 (.41) (.34) (.41) (.25) Pr (influence) .0112 .4495 .0130 .0660 (.11) (.50) (.09) (.50) FAF contribution .8732 .7980 .4882 .5307 (.04) (.06) (.15) (.13) Pr(LRPV) .0141 .0158 .0117 .0187 (.12) (.10) (.12) (.06) Lobby cost .3210 .3338 .4643 .4624 (.49) (.49) (.49) (.49) Chi-square 26.33 32.37 23.38 29.82 (Significance) (.0004) (.0000) (.0015) (.0001) Correctly predicted 80% 84% 80% 82% MANAGER COST EXPECTATIONS, 235 APPENDIX G (continued) PROBIT ANALYSES MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Shift to government and probability of affecting the FASB's continuance) ED OPPOSERS (22 filers and 35 nonfilers)a Beta CoefficientsbC Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt 1) (Pt I/Pt II) (Pt II/Pt II) Constant -.4690 -.4120 -.5083 -.5107 (.06) (.10) (.04) (05) Adm & public .0053 .0055 .0056 .0060 relation costs (.06) (.07) (.06) (.05) Debt costs -.0002 —.0002 .0002 -.0002 (.30) (.29) (.25) (.30) Mgmt comp costs .6671 .6925 .9202 .7317 (.14) (.17) (.09) (.16) Pr (influence) -.0117 -.5427 .0099 -.O631 (.09) (.49) (.14) (.50) Shift to .0016 .0068 .0025 .0081 government (.23) (.34) (.25) (.31) Pr(LRPV) .0101 .0103 .0068 .0119 (.19) (.19) (.25) (.16) Lobby cost -.2859 -.3004 .4350 -.4399 (.49) (.49) (.49) (.49) Chi-square 23.89 30.89 23.75 29.23 (Significance) (.0012) (.0001) (.0013) (.0001) Correctly predicted 77% 77% 79% 79% 236 APPENDIX G (continued) PROBIT ANALYSES MANAGER COST EXPECTATIONS, MANAGER PERCEPTIONS OF THE FASB AND PARTICIPATION CHOICE (Filed comment on ED to support the FASB) ED OPPOSERS (28 filers and 42 nonfilers)a BetajCoefficientsbc Variable Analysis 1 Analysis 2 Analysis 3 Analysis 4 Measuresd (Pt I/Pt I) (Pt II/Pt I) (Pt I/Pt II) (Pt II/Pt II) Constant .2271 .1243 .2881 .2435 (.17) (.31) (.11) (.15) Adm & public .0055 .0056 .0069 .0071 relation costs (.06) (.07) (.03) (.04) Debt costs .0002 .0002 .0002 .0002 (.27) (.26) (.28) (.28) Mgmt comp costs .0525 .1146 .2002 .3078 (.45) (.40) (.33) (.26) Pr (influence) .0130 .4993 .0143 .0536 (.07) (.49) (.07) (.49) Filed to .0019 .0001 .0011 .0001 support FASB (.42) (.50) (.45) (.49) Lobby cost .3021 .3217 .1337 .1495 (.49) (.49) (.10) (.09) Chi—square 24.76 32.43 23.84 29.55 (Significance) (.0004) (.0000) (.0006) (.0000) Correctly predicted 73% 69% 71% 73% 237 APPENDIX G (continued) aSee Table 6B for detail on sample size reduction. bAn inverse relationship was hypothesized between manager participation Choice, Pr Of influence (which was measured in negative terms), and lobby cost. A positive relationship was hypothesized for manager cost expectations, FAF, shift to government, and filed to support the FASB. cOne-tailed probability Of t-statistic in parentheses. dMeasures for "pr(influence)/lobby costs". 238 BIBLIOGRAPHY Accounting for Pension Costs: Current Practices in CanadaL the United Kingdom, and the United States. Accountants International Study Group. Brooklyn, N.Y.: Newport Press, 1977. A review Of pension accounting practices currently in use in several countries is provided. Amemiya, T. "Qualitative Response Models: A Survey." Journal of Economic Literarure. (December 1981): pp. 1483-1536. A critical evaluation Of a large number of multi- response and multivariate qualitative response models is presented and empirical examples are provided. Aristar, Inc. 10-K Report for the Year Ended June 30, 1976. In a note to its financial statements, Aristar Inc. states that it was in default under several long term debt agreements due to the adoption of FASB Statement NO. 12, Accounting for Certain Marketable Securities. Arthur Andersen & CO. "Summary of FASB Final Statements on Employers' Accounting for Pensions." Accounting News Briefs. (January 1986). A brief Overview of FASB Statement NO. 87, Employers' Accountingyfor Pensions, is presented with a discussion Of the potential effects on corporate financial statements. Brown, P. R. "A Descriptive Analysis of Select Input Bases Of the Financial Accounting Standards Board." Journal of Accounting Research. (Spring 1981): pp. 232-246. Brown uses multidimensional scaling in a study which reports results that suggest an FASB alignment on accounting issues with the Financial Analysts Federation. There is no consistent relationship found between the FASB and the Big Eight accounting firms. Campbell, D. and D. Fiske. "Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix." Psychological Bulletin. (1959): pp. 81-105. A correlation analysis is Offered to assess the construct validity Of variable measurement. Cronbach, L. J. "Coefficient Alpha and the Internal Structure Of Tests." Psychometrika. (September 1951): pp. 297-334. Cronbach demonstrates that a formula (alpha) may be used to compute the mean correlation coefficient for all split—halfs Of a measurement instrument. 239 Demski, J. S. Information Analysis. Addison-Wesley (1980). Demski provides a monograph on the economic aspects of information systems using utility theory. Directory Of Corporara Affiliates. "Who Owns Whom." National Register Publishing Company (November 1986.) A summary Of financial statement data is provided for companies listed with the NYSE, AMEX and OTC. Downs, A. Economic Theory of Democracy. Harper and Row (1957). A model is developed to describe government behavior in the context of self-interested individuals. The model is different from other economic models in that the government is not viewed as an exogenous variable with a strictly social function. In fact, Downs argues that the government considers social welfare only to the extent that it will be a factor in retaining power. Fama, E. F. "Agency Problems and the Theory Of the Firm." Journal Of Political Economy. (April 1980): pp. 288-307. Fama discusses the importance of the labor market in reducing the degree to which managers may consume perquisites in an agency relationship. Fama, E. F. and M. C. Jensen. "Separation Of Ownership and Control." Journal of Law and Economics. (June 1983): pp. 301-326. A discussion of "property rights theory" is provided wherein accounting serves as an integral part of contracts between the Owner and manager Of a company. Financial Accounting Standards Board. Employers’ Accounting for Pensions and Other Postemployment Benefits. (Discussion Memorandum). Stamford: Financial Accounting Standards Board (February 19, 1981). The historical development Of pension accounting is reviewed, actuarial methods and practices are described, and alternative accounting policies are discussed. ---. Employers’ Accounting for Pensions (Exposure Draft). Stamford: Financial Accounting Standards Board (March 22, 1985). ---. Statement Of Financial Accounting7Standards NO. 35: Accounting and Reporting By Defined Benefit Pension Plans. Stamford: Financial Accounting Standards Board (March 1980). ---. Statement of Financial Accounting Standards NO. 36: Disclosure of Pension Information. Stamford: Financial Accounting Standards Board (May 1980). Finney, D. J. Probit analysis. Cambridge University Press (1971). Finney presents an account Of the statistical techniques grouped under the heading of probit analysis. Basic on-hand methods Of computation are explained and problems are discussed in terms appropriate to computer applications. 240 Fogelson, J. H. "The Impact Of Change in Accounting Principles on Restrictive Covenants in Credit Agreements and Indentures." The Business Lawyer. (January 1978). Fogelson discusses the difficulties encountered by companies when they attempt to amend public debt instruments. Gerboth, D. L. "Research, Intuition, and Politics in Accounting Inquiry." The Accounting Review. (July 1973). Gerboth argues that "decisions made by authoritative accounting bodies must be based on explanations and predictions Of human behavior," and, therefore, they must consider intuition and experience as well as the results of systematic research. Griffin, P. A. "Foreign Exchange Gains and Losses: Impact on Reported Earnings." Abacus. (June 1982): pp. 50-69. Griffin examines the participation choice Of corporate managers on SFAS 8 using a multinomial logit analysis. Griffin, P. A. "Management’s Preferences for FASB Statement NO. 52: Predictive Ability Results" Abacus. (Vol. 19, NO. 2, 1983): pp. 130-138. Griffin replicates his prior study on the participation Choice of corporate managers (SFAS 8) on SFAS 52. He finds that the factors which were significant for SFAS 8 were not the same factors which were significant for SFAS 52. However, managers who had lobbied on SFAS 8 also lobbied on SFAS 52. Hagerman, R. L. and M. E. Zmijewski. "Some Economic Determinants of Accounting Policy Choice." Journal Of Accounting and Economics. (October 1978): pp. 141—161. Hagerman and Zmijewski provide evidence to show that firm size, systematic risk, capital intensity, industry concentration and management compensation plans do not affect manager accounting Choices consistently. A probit analysis is used to examine the effects of these factors on accounting choices made with regard to depreciation, inventory, investment tax credit and the amortization of past service costs. Healy, P. "The Impact of Bonus Schemes on Selection of Accounting Principles." Journal Of Accounting and Economics. (April 1985). The relationship between a firm’s annual net accruals and bounds in the firms’ bonus plan is examined. The results suggest that details of bonus plan bounds are important in the determination Of corporate accruals. Hinckley, B. Coalitions and Politics. Harcourt Brace Jovanovich, Inc. (1981). The study of coalitions is used to analyze political activity. Game theory, social psychological studies and empirical political studies are discussed. Variables of interest are derived from the three approaches and suggestions are made for future research. 241 Jensen, M. C., and W. H. Meckling. "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure." Journal of Financial Economics. (October 1976): pp. 305-360. A theory of the ownership structure of a firm is developed using agency, property rights and finance theory. Kelly, L. "Corporate Lobbying and Changes in Financing or Operating Activities in Reaction to FAS NO. 8." Journal of Accounting and Public Policy. (Winter 1982): pp. 153-173. The relationship between corporate lobbying and Changes in financing or operating activities in response to FASB Statement No. 8, is examined. A chi-square test indicates no significant relationship between the two reactions. However, certain corporate attributes are found to be related to the firm’s decision to lobby and to alter financing or operating activities. -—-. "Corporate Management Lobbying on FAS NO. 8: Some Further Evidence." Journal of Accounting Research. (Autumn 1985): pp. 619-632. Kelly examines the relationship between a manager’s decision to lobby on FASB Statement NO. 8 and the firm’s debt contracts and management's degree of ownership in the firm. With firm size controlled for, neither leverage nor management ownership was statistically significant, however, for filers opposed to the income effect, management’s ownership percentage was significantly lower than that of nonfilers. Kerlinger, F. N. Foundations of Behavioral Research. Holt, Rineholt and Winston, Inc. (1964). The nature of scientific research and methodological problems in research statistics and design are discussed with an emphasis on educational research. Kirk, D. J. "Concepts, consensus, compromise and consequences: Their roles in standard setting." Journal of Accountancy. (April 1981): pp. 83-86. Kirk notes that standard-setting by consensus, compromise, or consequences is not acceptable. The FASB’s primary goal must be conceptually sound standards. Kirk, D. J. "The FASB's Second Decade." Journal Of Accountancy. (November 1983): pp. 86-96. Key board and staff executives discuss the future of financial accounting standard-setting. Lachenbruch, P. A., and M. R. Mickey. "Estimation of Error Rates in Discriminant Analysis." Technometrics. (February 1968): pp. 1-11. Several methods of estimating error rates in discriminant analyses are evaluated on the basis Of a series of Monte Carlo experiments. 242 Lasater, D. B. Manager Incentives for Advocacy Efforts Before Accounting Standard Setting Bodies. Unpublished doctoral dissertation. The University of Texas at Austin (1982). Lasater uses the interest capitalization issue to provide evidence that the policy preference Of a corporate firm manager is significantly related to the corporation’s debt/equity ratio. The participation Choice is found to be significantly related to the corporation’s earnings before taxes, a factor not hypothesized in the model. McKee, A. J. Jr, et al. "Management Preferences Over Accounting Standards: A Replication and Additional Tests." The Accounting Review. (October 1984): pp. 647-659. The researchers replicate and expand the Watts and Zimmerman study using the position papers filed in response to the exposure draft on general price level Changes. Several biases are identified and corrected. The findings indicate that the Watts and Zimmerman model does not perform well when a jackknife holdout procedure is followed. When alternative measurements are used together with a logistic regression model, the theory’s performance improves, but not significantly. McKee, A. J. Preferences for Accounting Stapgards: The Use Of Discriminant Analysis in Forecasting Corporate Manager Lobbying Behavior. Unpublished doctoral dissertation. Oklahoma State University (1982). McKee replicates and expands Watts and Zimmerman's study on GPLA accounting, introducing a volatility factor into the analysis and using the Parker Algorithm to estimate the effect of FASB Statement NO. 33 on corporate financial statements. The findings indicate that the Watts and Zimmerman model, including the volatility factor, does not perform well when a jackknife holdout procedure is followed. Nunnally, J. C. Psychometric Theory. McGraw-Hill, Inc. (1967). This book is concerned with the theory Of psychological measurement. Tests of measurement reliability and validity are described and explained. Payne, S. L. The Art Of Asking Questions. Princeton University Press. Princeton, N.J. (1951) Payne provides a complete discussion on the wording Of single questions in order to avoid bias and misunderstanding. Peter, J. P. "Reliability: A Review of Psychometric Basics and Recent Marketing Practices." Journal of Marketing Research. (February 1979): pp. 6-17. The basic theories and measurement procedures for reliability and generalizability are reviewed, illustrated and evaluated. A critique is given on some recent marketing research studies in which reliability estimates were used and recommendations are made for future research. 243 Press, S. J., and S. Wilson. "Choosing Between Logistic Regression and Discriminant Analysis." Journal of the American Statistical Association. (December 1978): pp. 699-705. The Classificatory power Of discriminant analysis and logistic regression is examined for models which include at least one independent variable that is not normally distributed. Previts, G. J. and B. D. Merino. A History Of Accounting in America. John Wiley & Sons (1979). This history of American accounting begins in 1492 and continues through 1978. The controversies underlying the progress of accounting are discussed and the political, social, economic and personal elements are addressed. Rappaport, A. "The Economic Impact of Financial Accounting Standards - Implications for the FASB." Journal of Accountancy. (May 1977): pp. 89+. Rappaport proposes a process to be followed by the FASB in setting financial accounting standards. This process includes consideration Of economic consequences as well as the conceptual framework. Siegel, S. Nonparametric Statistics For the Behavioral Sciences. MCGraw Hill (1956). Siegel presents nonparametric tests for different levels Of measurement and research design. Smith, C. W., and J. B. Warner. "On Financial Contracting: Analysis Of Bond Covenants." Journal of Financial Economics. (June 1979): pp. 117-161. An analysis of the role of covenants using accounting numbers in debt agreements is presented. Watts, R. L., and J. L. Zimmerman. "Towards a Positive Theory of the Determination of Accounting Standards." The Accounting Review. (January 1978): pp. 112-134. Watts and Zimmerman use the position papers filed in response to the discussion memorandum on General Price Level Changes to provide evidence that the position Of management on a proposed standard will depend on the size of the firm and the effect Of the standard on the firm’s reported earnings. A linear discriminant analysis is used. Wyatt, A. R. "The Economic Impact of Financial Accounting Standards." Journal Of Accountancy. (October 1977): pp. 92-94. This article was adopted from a paper presented by Wyatt at an accounting symposium where he argued that the FASB must address the economic consequences of mandated accounting changes, but ultimately, should set standards which are consistent with the conceptual framework. 244 Zmijewski, M. E., and R. L. Hagerman. "An Income Strategy Approach to the Positive Theory of Accounting Standard Setting Choice." Journal of Accounting and Economics. (April 1981): pp. 129-149. Evidence is provided which indicates that an income strategy model is not generally applicable to all firms. Furthermore, managers Of firms that are politically visible (as measured by size and industry concentration) or have followed extreme income strategies (all income increasing or decreasing) are more likely to lobby than managers Of smaller firms or firms following more diverse income strategies.