. a V LR. i . . .L. .55?qu \ L, 3»; W ‘ .....afl1u.kn .. . ,, .w . amp?) . 3”! in.“ lo '03“...- An!" .q—‘hmfl‘é‘fi D. . 7‘ put lug , mugs? .5. u. .n i .ffiulnflg fluid... . «Incl ... :45. .3042“ . ll. 5:... , ..,.u..«... 9. zfi “.4.sz be?) 5 {(1. fitting} .7 . a ’ "2 ..!..».!.:...fl.... rut-«n ”TI: ... egfliuj}: .0... 3" i 2...“. glittnnng Six: ‘V- - inu'k V 1.13.. .. . t I P.\ . ,Itlliiriflflr £v‘t:£€.‘.le. .\ .5! g. 1. {£5 -‘w.\n.l.. .55 .45)... $1.11!!" .. 31¢ . is“. . =1}: :1. 19 4. 3)). 41...; 5...: {can .42.... i iIP<.-‘.Ifj‘. .1 .21.... .. g ineviigt}: ....l (air-:4?!) 7...".nvzwhw.’ . > I ,¢.'ALW.§A. . . . 5.5.3.225. .i t .. . . 4%.... . 5W“... Eu .. , . . . , I. ‘r . "34......rovh... -.;.u... 5&3 ~9. w .y .. y < . . ‘ " ‘ xi? A $3.». ru..1_...uiflgd1fi|udwia$ " tlll'. . . l .I U..."a.l.t}.:fl.§.fl.r“ AV! .317: i . ... . v) . $11». A. .H. title {Ills 2-“ GAN STATEU IHIHHIIHHIILIBWHWI 2a. ,- g " 3m Flumunr ‘ farms: dissertation entitled Determinants of the Choice Between Comprehensive and Partial Income Tax Allocation: The Case of the Domestic International Sales Corporation presented by Sanjay Gupta has been accepted towards fulfillment of the requirements for Ph.D. . Accounting degree in WW Major professor Date May 8, 1990 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU I. An Affirmative Action/Equal Opponunity Imitation chS-DJ ii ““ DETERMINANTS OF THE CHOICE BETWEEN COMPREHENSIVE AND PARTIAL INCOME TAX ALLOCATION: THE CASE OF THE DOMESTIC INTERNATIONAL SALES CORPORATION By Sanjay Gupta A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Accounting 1990 M45- 50%: ABSTRACT DETERMINANTS OF THE CHOICE BETWEEN COMPREHENSIVE AND PARTIAL INCOME TAX ALLOCATION: THE CASE OF THE DOMESTIC INTERNATIONAL SALES CORPORATION By Sanjay Gupta The purpose of this study was to examine firms’ motivations for choosing between comprehensive and partial allocation--alternative accounting methods for the extent of interperiod tax allocation. Based on the predictions of ‘positive accounting theory’, firms’ choice was hypothesized to be related to variables that surrogate for potential debt covenant violations and their political visibility. In addition, auditors’ preferences were hypothesized to affect the choice. Domestic international sales corporations, a special type of corporation legislated into the Internal Revenue Code in 1971 to stimulate exports, provided a unique data set to empirically examine the determinants of the accounting choice. The legislation allowed an indefinite deferral of income taxes on a portion of the DISC’s export earnings. However, for financial reporting, some firms provided taxes on those earnings (comprehensive allocators), whereas others did not (partial allocators). The empirical analysis was conducted on a sample of 320 firms that had a DISC operational in 1972, 1973, or 1974. Statistical tests were performed in both univariate and multiple regression frameworks. Overall, the results indicate support for the debt covenant and auditor preference hypotheses, but only partial support for the political cost hypotheses. Consistent with the theory, firms with higher leverage and lower interest coverage ratios are observed to adopt partial allocation, an income-increasing method. However, a similar hypothesis based on dividend restrictions is not supported. In addition, no support is found for the hypothesis that because public debt involves higher renegotiation costs, firms with public debt are more likely to use partial allocation. Also consistent with the theory, firms with higher political costs manifest through higher effective tax rates are observed to adopt comprehensive allocation, an income—decreasing alternative. However, support for this hypothesis is weak. Another political cost hypothesis based on firm size is not supported. Finally, strong support is found for the hypothesis that firms’ accounting method choice is related to the preferences of their auditors, a relatively new finding in the accounting method choice literature. The results generally were consistent across alternative definitions of the explanatory variables and sub- samples of firms by industry classification and the year when the DISC became operational. ACKNOWLEDGMENTS I thank my dissertation committee members, Professors Ed Outslay, Joe Anthony, and Steve Buzby for their valuable counsel and guidance in completing this thesis and influencing my thinking. I have also benefitted from many conversations with my doctoral colleagues, particularly Young-ho Nam. His companionship made the basement of Hubbard Hall almost bearable. I acknowledge the financial support from the Department of Accounting and the Patricia Roberts Harris Fellowship through the School of Urban Affairs without which it would have been impossible for me to complete the program. I also thank Mr. Michael Crooch, partner in the Chicago office of Arthur Andersen, for allowing me access to the comment letters to APB Opinions. I owe a special debt of gratitude to my family in India for the emotional support they provided me and to my daughters, Priyanka and Neha, for tolerating the narrow confines of Spartan Village. Finally, a very special thank you to my wife, Kiran, for reasons too numerous to mention. Because she made this whole endeavour worthwhile, it is to her that I dedicate this dissertation. TABLE OF CONTENTS List of Tables List of Figures Chapter 1. INTRODUCTION AND OVERVIEW 2. INSTITUTIONAL BACKGROUND 2.1 2.2 Interperiod tax allocation under GAAP 2.1.1 The problem 2.1.2 Historical development 2.1.3 Motivation for research on extent of interperiod tax allocation Domestic international sales corporations 2.2.1 Legislative intent and overview of I.R.C. provisions 2.2.2 Qualifications of a DISC 2.2.3 Portion of DISC income deferred from tax 2.2.4 Removal of DISCS 2.2.5 Permanent exemption of DISC deferred taxes 3. THEORETICAL FRAMEWORK AND PREVIOUS RESEARCH 3.1 3.2 3.3 Positive accounting theory 3.1.1 Its genesis 3.1.2 The theory and its implications for accounting method choice Previous research 3.2.1 Voluntary accounting method choice studies 3.2.2 The interperiod tax allocation literature Limitations of the theory and research 3.3.1 Measurement issues -- dependent variable 3.3.2 Measurement and specification issues —- independent variables 3.3.3 Omitted variables issues Page viii 11 11 11 17 22 24 25 28 30 33 36 39 39 39 42 47 48 52 54 55 58 59 TABLE OF CONTENTS (cont’d.) 4. HYPOTHESIS DEVELOPMENT 4.1 Financial statement effects 4.2 The debt covenant hypotheses 4.3 Political costs hypotheses 4.3.1 The firm size hypothesis 4.3.2 The effective tax rate hypothesis 4.4 The auditor hypothesis 5. EXPERIMENTAL DESIGN AND SAMPLE SELECTION 5.1 Experimental design 5.2 Sample selection and data 5.3 Variable definitions and measurement issues 5.3.1 Classification of DISC accounting method -- the dependent variable 5.3.2 The independent variables 5.4 Within sample profile analysis 5.4.1 Industry membership 5.4.2 Exchange listing 5.4.3 COMPUSTAT listing 5.4.4 Year when DISC operational (‘DISC Year’) 5.5 Statistical procedures 5.5.1 Univariate analysis 5.5.2 Multivariate analysis 6. ANALYSIS OF RESULTS 6.1 Univariate analysis 6.2 Multiple regression analysis 6.2.1 The logit model and its evaluation 6.2.2 The logit regression results 6.2.3 OLS regression results 6.3 Sample partitioned by industry 6.4 Sample partitioned by DISC-year 7. SUMMARY AND CONCLUSIONS 7.1 Summary of the study and the results 7.2 Suggestions for future research Appendix A EXTENT OF TAX ALLOCATION: The Conceptual Merits of the Alternative Approaches Appendix B SAMPLE FIRM INFORMATION vi 61 61 63 71 71 73 80 83 83 84 90 91 94 113 113 116 120 122 125 125 127 132 132 143 150 153 158 159 164 170 170 172 174 180 TABLE OF CONTENTS (cont’d.) Appendix C EXAMPLES OF DISC DISCLOSURES 189 Appendix B INDUSTRY CLASSIFICATION OF SAMPLE FIRMS 198 BIBLIOGRAPHY 202 vii 2.1 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 6.1 6.2 LIST OF TABLES DISC DEFERRALS SAMPLE SELECTION DESCRIPTIVE STATISTICS -- Financial Variables LEVERAGE MEASURES DESCRIPTIVE STATISTICS -- LEVERAGE MEASURES PEARSON PRODUCT -MOMENT CORRELATION COEFFICIENTS -- LEVERAGE VARIABLE PEARSON PRODUCT -MOMENT CORRELATION COEFFICIENTS -- SIZE VARIABLES AUDITOR ANALYSIS DESCRIPTIVE STATISTICS -- Independent Variables INDUSTRY ANALYSIS SAMPLE FIRMS BY EXCHANGE LISTING SAMPLE FIRMS BY COMPUSTAT LISTING SAMPLE FIRMS BY YEAR WHEN DISC OPERATIONAL UNIVARIATE STATISTICS PEARSON PRODUCT ~MOMENT CORRELATION COEFFICIENTS viii Page 37 85 89 97 99 101 109 112 114 117 119 121 124 133 145 6.3 6.4 6.5 6.6 LIST OF TABLES (cont’d.) LOGIT REGRESSION MODELS (All Firms -- 1971 Data) LOGIT REGRESSION MODELS (All Firms -- DISC-Year Data) UNIVARIATE STATISTICS -- BY INDUSTRY CLASSIFICATION UN IVARIATE STATISTICS -- BY DISC-YEAR 154 157 160 165 LIST OF FIGURES Page Figure 1 OVERVIEW OF THE INTERPERIOD TAX ALLOCATION PROBLEM 18 Figure 2 DISCS AND INTERPERIOD TAX ALLOCATION 27 Figure 3 FINANCIAL STATEMENT EFFECTS OF COMPREHENSIVE V. PARTIAL ALLOCATION 62 Chapter One INTRODUCTION AND OVERVIEW Accounting for income taxes has been one of the most debated and controversial areas in financial reporting since the 19403 [Rayburn, 1986]. Accounting Principles Board Opinion No. 11 (APB 11), Accounting for Income Taxes [AICPA, 1967], was issued in 1967 to resolve the controversies in this area. However, its issuance intensified the debate [Nair and Weygandt, 1981; Beresford et al., 1983], and several pronouncements subsequently were issued to amend, interpret, and/or supplement APB 11.‘ The primary criticisms of APB 11 and the related pronouncements were their underlying concepts, the complexity of their requirements, and the interpretation of the results generated under them. Because of this controversy, the Financial Accounting Standards Board (FASB) recently reconsidered the entire area of accounting for income taxes. In December 1987, Statement of Financial Accounting Standards No. 96 (SFAS 96), Accounting for Income Taxes [FASB, 1987], was issued, superseding APB 11 and most related pronouncements on the subject?“3 1For example, APB Opinions 16, 23, and 24, FASB Statements 31 and 37, FASB Interpretations 22, 25, 29, and 32, and FASB Technical Bulletins 81-2, 83- 1, 84-2, 84-3, and 86-1 are some of the many pronouncements issued after APB 11 to deal with various aspects of the accounting for income taxes controversy. zSee SFAS 96, Appendix D for an exhaustive list of amendments made to the existing pronouncements. 2 There are many contentious issues in the accounting for income taxes debate, such as interperiod tax allocation, discounting, accounting for net operating losses and interim periods, and the allocation of income taxes to members of a consolidated group. The focus of this study is on interperiod tax allocation because much of the controversy surrounding accounting for income taxes centers on this issue [Beresford, et al., 1983]. In addition, researchers (e.g., Beaver and Dukes [1972]) have noted that interperiod tax allocation is an ‘attractive’ research issue because it affects a large number of firms, unlike certain other industry-specific accounting controversies (e.g., full-cost v. successful efforts accounting). The interperiod tax allocation problem results primarily from differences in the timing of transactions that affect both pretax accounting income and taxable income which creates temporary differences. Two questions arise with respect to this problem: 1) how much to allocate (the ‘extent of allocation’--comprehensive or partial), and 2) how to allocate (the ‘method of recording the allocation’-- 3SFAS 96 has itself become the subject of much controversy for some of the very problems it sought to remedy, namely complexity and interpretation of the results (see FASB Status Report No. 203, dated July 14, 1989). These problems have caused the effective date for the statement, originally set for fiscal years beginning after December 15, 1988, to be postponed thrice: first to fiscal years beginning after December 15, 1989 (FASB Statement No. 100, Accounting for Income Taxes-Deferral of the Effective Date of FASB Statement No. 96, issued in December 1988), second to the fourth quarter of 1990 (see FASB Status Report No. 204, dated August 7, 1989), and again to fiscal years beginning after December 15, 1991 (see FASB Statement No. 103, Accounting for Income Taxes-Deferral of the Efiective Date of FASB Statement No. 96. See also Wolk, Martin, and Nichols [1989] and Chaney and Jeter [1989] for a discussion of the theoretical problems under SFAS 96. 3 deferred, liability, or net-of—tax). This study focuses on the extent of allocation question. Although both SFAS 96 and APB 11 generally require comprehensive allocation of deferred taxes, they provide an exception for transactions that satisfy the ‘indefinite reversal’ criteria described in APB Opinion No. 23 (APB 23), Accounting for Income Taxes-~Special Areas [AICPA, 1972]. Transactions to which these criteria apply are similar to other temporary differences except that the reversal of the initial differences between taxable income and pretax accounting income may be unpredictable, primarily because the tax consequences of the transactions are controlled by the enterprise and frequently require management to take specific actions before the differences will reverse. One such transaction is the undistributed earnings of subsidiaries. Firms can choose between providing deferred taxes on all of the indefinitely deferred income (comprehensive allocation) and providing deferred taxes on only a portion (including zero) of such income (partial allocation), depending on whether the transfer of the subsidiary’s earnings to the parent is imminent or relatively certain or it is management’s intention to reinvest those earnings indefinitely. However, this choice has no obvious cash flow implications for the firm.4 ‘A voluntary accounting method choice that can have an ‘obvious’ impact on a firm’s cash flow is the choice between LIFO and FIFO inventory costing methods through its impact on taxable earnings. It is possible that for such accounting method choices, the contracting theory motivations (discussed later) may be in the nature of second-order effects. 4 The purpose of this study is to empirically analyze whether, in the absence of explicit cash flow effects, firms’ choice between comprehensive and partial allocation is related to their constraints on contractual agreements (e.g., debt covenants), their political visibility, and the preferences of their auditors. Several hypotheses regarding the determinants of the accounting method choice are tested empirically. The hypotheses primarily are derived from the predictions of positive accounting theory that is driven by contracting cost (agency theory) and political cost arguments. These arguments have been used to investigate management’s motivation in choosing between different accounting methods permissible under generally accepted accounting principles (GAAP) for apparently similar transactions. Holthausen and Leftwich [1983, p. 77] summarize the implications of these arguments for firms’ accounting method choice as follows: Accounting choices have economic consequences if changes in the rules used to calculate accounting numbers alter the distribution of firms’ cash flows or the wealth of parties who use those numbers for contracting and decision making. In the past decade, several studies have examined the determinants of firms’ voluntary choice of accounting methods (and related questions).‘ These studies have uncovered empirical regularities consistent with the contracting and political cost arguments. The results indicate that firms’ debt/equity ratios (a proxy for potential debt covenant violations), existence of accounting earnings-based bonus plans, and firm size (a surrogate for political sensitivity) are related to their ’See Watts and Zimmerman [1986], Holthausen and Leftwich [1983], and Kelly [1983] for a review of these studies. 5 accounting procedure choice.‘ This study extends this line of research to the issue of the extent of interperiod tax allocation, thereby providing evidence on the robustness of the prior findings to another accounting method choice and contributing to the external validity of those results. Examination of an accounting method choice not studied before is motivated by Christie [1989]: In the early stages of a literature, one is typically more interested in regularities than anomalies. Given the reviews and methodological analyses by Ball and Foster (1982), Holthausen and Leftwich (1983), Watts and Zimmerman (1986), and Christie (1987) and the results in this paper, it seems reasonable to characterize the contracting and size literature as being in the ‘finding regularities’ phase. Regularities are currently more valuable than anomalies .. (emphasis supplied). Second, this study differs from most prior accounting choice studies that use only firm size to test the political cost hypothesis.’ Ball and Foster [1982], among others, have criticized that approach on the grounds that firm size may proxy for omitted variables, such as industry-membership. Empirical evidence (e.g., Bowen, Noreen, and Lacey [1981]; Zimmerman [1983]) indeed suggests that such aggregate measures as firm size may not adequately proxy for firms’ political costs. In this study, in addition to firm size, the firm’s effective tax rate is used as a surrogate for ‘Christie [1990] conducts formal tests of whether positive accounting theory can explain choice of accounting procedures. Based on results aggregated across several studies, he concludes that the data support the contracting and size hypotheses. 7Hagerman and Zmijewski [1979] also examined risk, capital intensity, and concentration ratios as possible surrogates of political costs. 6 political costs.8 This is based on evidence linking the politicization of US. firms’ effective tax rates and subsequent wealth transfers resulting therefrom. The use of the effective tax rate variable is particularly relevant in this study because the choice of the extent of interperiod tax allocation directly impacted the firm’s reported effective tax rate. Finally, the auditor’s role in firms’ accounting method choice is examined in this study. The low explanatory power of the typical accounting method choice models based solely on the contracting and political cost arguments suggests a potential omitted variables problem. One such variable is auditor preferences which virtually has been ignored in prior research.“ Domestic International Sales Corporations (DISCS), a Special type of corporation legislated into the Internal Revenue Code in 1971 to stimulate exports, provide a unique data set to examine the determinants of firmS’ extent of tax allocation choice. In general, a DISC is a nontaxable entity, the profits of which are taxed to its shareholders when distributed or deemed to be distributed. Each year a portion of the DISC’S export profits is deemed to be distributed to its shareholders, thereby subjecting that income to current US. taxation. However, US. income tax can generally be deferred indefinitely on the remainder of the DISC’S export profits. Since DISCS have been formed generally as subsidiaries of 8El-Gazzar, Lilien, and Pastena [1986] and Wong [1988] also used effective tax rates as a political cost variable. However, El-Gazzar, et al. were concerned with management’s choices in accounting for leases and Wong with New Zealand firms’ choice in accounting for export tax credits. ’Studies by Trombley [1989] and Thornton [1986] are exceptions. 7 existing corporations and their indefinitely deferred earnings satisfy the indefinite reversal criteria of APB 23, firms that set up DISCS could choose between comprehensive and partial allocation of taxes on that portion of DISC earnings. For financial reporting purposes, some firms did not provide taxes on the indefinitely deferred DISC earnings (hereafter the ‘partial allocators’), while other firms did record deferred taxes on those earnings (hereafter the ‘comprehensive allocators’). Thus, different financial reporting practices were followed by firms for a similar transaction, providing an opportunity to examine firms’ motivation for choosing among alternative methods. The empirical analysis is performed on a sample of 320 firms that had a DISC operational in fiscal years ending in 1972, 1973 or 1974, and for which tax allocation information with respect to the DISC’S earnings was available. Based on tax footnote disclosures in the financial statements, 82 firms were classified as comprehensive allocators and 238 as partial allocators. Because of the difficulties encountered in identifying the year in which the DISC was operational, two sets of analyses are performed: 1) on all firms assuming they had the DISC operational in fiscal year 1972, the first year DISCS became available, and 2) on all firms for the year the DISC actually became operational based on the firm’s footnote disclosures in its financial statements. In both analyses, data for the year prior to the year the DISC became (or was assumed to become) operational is used. For example, in analysis (1) above, fiscal 1971 data is used. In addition, analysis also 8 is performed on sub-samples of firms by industry classification and by year the DISC became operational. The empirical analysis employed Statistical tests in both univariate and multiple regression frameworks. The univariate analysis is conducted using both parametric and non-parametric tests and it indicates that firms’ DISC accounting method choice is related to variables that surrogate for potential debt covenant violations and political costs and their auditors’ preferences. Relative to comprehensive allocators, partial allocators have higher leverage and lower interest coverage ratios, which is consistent with the hypotheses that firms closer to potentially violating possible debt covenants will adopt an income-increasing accounting method. Also consistent with the debt covenant arguments, relative to comprehensive allocators, partial allocators pay out a greater proportion of their inventory of payable funds as dividends but this difference is not statistically Significant. Similarly, no support is found for the hypothesis that because public debt is costlier to renegotiate, firms with public debt are more likely to use an income-increasing accounting method. The results of the political cost hypotheses are mixed. Whereas support is found for the hypothesis that firms bearing higher political costs as manifested by higher effective tax rates are more likely to choose income-reducing accounting methods, no support is found for the firm size variable as a significant factor in firms’ DISC accounting method choice. Finally, strong support is found for an association between auditor preferences and firms’ adoption 9 of the DISC accounting method, which is a relatively new result in the accounting method choice literature. These results hold for both data sets (1971 data for all firms and data for the year when the DISC actually became operational) and across alternative definitions of the variables. The DISC accounting method choice is not independent of the sample firms’ industry membership based on one-digit SIC codes. Nearly 86 percent of the sample firms are either durable or nondurable goods manufacturers. The analysis is repeated on sub-sample by industry membership and the above results hold for the durable manufacturers but are weaker for the nondurable manufacturers. AS expected, some of the independent variables are correlated with each other (e.g., the debt covenant-based variables--leverage and interest coverage). Because the univariate analysis ignores these correlations, thereby suffering from the correlated omitted variables problem, those results must be interpreted with caution. To overcome this limitation, the data is also analyzed in a multiple regression framework using both logistic and the ordinary least squares procedures. The results of the regression analysis generally corroborate the univariate results discussed above. The rest of this study is organized as follows: the institutional background of interperiod tax allocation under GAAP and of DISCS is discussed in chapter two, a summary of the theoretical framework and a review of previous research is provided in chapter three, the hypotheses are developed in chapter four, the 10 experimental design and data used to test the hypotheses are described in chapter five, the results of the analysis are discussed in chapter six, and the study’s implications are summarized in chapter seven. Chapter Two INSTITUTIONAL BACKGROUND In this chapter the institutional details regarding interperiod tax allocation under GAAP and the entity DISC are provided in order to better appreciate the environment within which the accounting method choice is studied. First, the interperiod tax allocation problem is explained followed by a brief historical perspective on the evolution of the authoritative literature governing the problem. Next, the statutory provisions governing the formation and operation of DISCS are reviewed. 2.1 Interperiod Tax Allocation under GAAP 2.1.1 The Problem A primary objective in accounting for income taxes is to determine the tax expense1 amount on a firm’s financial Statement. The taxes currently payable by an entity are based on its taxable income, which is determined under the US. and foreign tax laws.2 A firm’s pretax financial accounting income, on the other hand, is determined under generally accepted accounting principles (GAAP), a different set of measurement rules than the tax laws. When a firm’s taxable income for a 1In this study, all references to taxes, such as in ‘tax expense’, ‘tax payable’, ‘tax liability’, etc., are to federal income taxes, unless otherwise mentioned. 2Specifically, the income taxes currently payable is determined by multiplying the taxable income by the statutory tax rate(s) and adjusting the resulting amount for tax surcharges and credits. 11 12 period is equal to its financial accounting income, the calculation of tax expense is relatively straightforward: tax expense Should equal taxes currently payable. However, financial reporting income usually differs from taxable income because of the vastly different objectives of the tax laws and GAAP, as underscored by the US. Supreme Court in its decision in Thor Power Tool Co. v. Commissioner of Income Tax3 The primary goal of financial accounting is to provide useful information to management, shareholders, creditors, and others properly interested; the major responsibility of the accountant is to protect these parties from being misled. The primary goal of the income tax system, in contrast, is the equitable collection of revenue; the major responsibility of the Internal Revenue Service is to protect the public fisc. Consistently with its goals and responsibilities, financial accounting has as its foundation the principle of conservatism, with its corollary that "possible errors in measurement [should] be in the direction of understatement rather than overstatement of net income and net assets." In view of the Treasury’s markedly different goals and responsibilities, understatement of income is not destined to be its guiding light. Financial accounting is hospitable to estimates, probabilities, and reasonable certainties; the tax law with its mandate to preserve the revenue, can give no quarter to uncertainty. ‘Generally accepted accounting principles’ tolerate a range of ‘reasonable’ treatments, leaving the choice among alternatives to management. [However], variances of this sort are questionable in a tax system. Because of the diversity in objectives, the Court concluded that "any presumptive equivalency between tax and financial accounting would be unacceptable." The Court pointed to the numerous differences in treatment accorded to expense and revenue recognition as evidence of the differing objectives between the 3439 US. 522, 79-1 USTC II 9139 (USSC, 1979). Footnotes and references to other court cases in the quoted material have been omitted. 13 tax laws and accounting principles. For example, the judicially created claim of tight doctrine has been invoked to require the immediate recognition of certain prepaid income for tax purposes, even though under GAAP such income would not be accrued until later so that revenues and expenses may be better matched.‘1 Similarly, the ‘wherewithal-to—pay’ rationale has been used to convert accrual-basis taxpayers into cash-basis taxpayers. The Court also observed that its prior decisions "demonstrated that divergence between the tax laws and accounting principles is especially common when a taxpayer seeks a current deduction for estimated future expenses or losses."5 The different objectives that result in different treatment of transactions generally give rise to different amounts of pretax accounting income and taxable income. These differences can be broadly categorized as temporary and permanent. A temporary difi’erence is the difference between the tax basis of an asset or a liability and its reported amount in the statement of financial position (balance sheet) that will result in taxable or deductible amounts in future years ‘The claim of right doctrine was developed by the Supreme Court in North American Oil Consolidated Co. v. Bumet, 286 US. 417, 3 USTC II 943 (USSC, 1932) in which it was held that an accrual-basis taxpayer must recognize contested income when received, even though its rights to the income have not been fixed. Under current law, prepaid interest, rent, and warranty incomes are always taxable in the year received. ‘For example, the Supreme Court has previously denied a current deduction for provisions against contingent liabilities for 1) non-performance of guaranties (Commissioner v. Hansen, 360 US. 446, 59-2 USTC ‘II 9533 (USSC, 1959)), 2) contested lawsuits (Lucas v. American Code Co., 280 US. 445, 2 USTC II 453 (USSC, 1930)), and unearned commissions on expected insurance policy cancellations (Brown v. Helven’ng, 291 US. 193, 4 USTC II 1223 (USSC, 1934)). 14 without regard to other future events (SFAS 96, para. 10). A subset of temporary differences is referred to as timing dijj‘erences.6 These differences originate in one period and reverse in a later period(s). For example, a timing difference occurs when for tax purposes income from the sale of goods on an installment basis is recognized when installment payments are received,’ whereas for financial accounting purposes the income is recognized when the goods are sold. In contrast, transactions or events that affect financial reporting income but not taxable income give rise to permanent differences.8 These differences arise solely as a matter of Statute and will never involve compensating differences in a later period(s), unless there is a change in the law; hence, these differences do not reverse. For example, interest earned on certain municipal obligations is not taxable [I.R.C., § 103], whereas fines and penalties are not deductible in computing taxable income [I.R.C., § 162(f)]. 6This was the term used in APB 11. SFAS 96 changed the terminology under the rationale that the term timing difference is too narrow in scope. It does not cover events such as increases in the tax basis of assets because of indexing for inflation or business combinations accounted for by the purchase method, which also create differences between the tax basis of an asset or liability and its reported amount in the financial Statements. The term ‘temporary difference’ is more comprehensive and covers these differences as well. However, for the purpose of this study, the distinction is not critical; hence the terms temporary and timing differences are used interchangeably. TUnder § 453 of the Internal Revenue Code of 1986 (I.R.C.). aSFAS 96 does not use the term ‘permanent difference’; instead such differences are referred to as "events that do not have tax consequences." These events "do not give rise to temporary differences" (see the definition of ‘temporary differences’ in SFAS 96, para. 206). 15 There is general agreement that in the case of permanent differences, the tax expense on financial statements should equal taxes payable. However, there is little consensus on what should be the tax expense amount in the presence of temporary differences. One alternative is to apply the same method as for permanent differences, i.e., report the taxes currently payable as the tax expense. This alternative is referred to as the flow through method,9 and under it the tax consequences of all temporary differences are ignored, i.e., no tax allocation is made.10 The alternative to using taxes currently payable as the tax expense amount is referred to as interperiod tax allocation.‘l Under this alternative, the tax effects of individual transactions or events are recognized in computing the tax expense of the period in which they affect financial reporting income. There are strong theoretical arguments for and against both the flow- through approach and interperiod tax allocation. These are elaborated in Appendix "Flow through’ is also used to describe one of the alternative methods for accounting for the investment tax credit (‘deferral’ being the other method)--see APB Opinions 2 and 4, Accounting for the "Investment Credit." However, references to the flow through method in this study do not apply to accounting for the investment tax credit. loSome proponents of this method would disclose the potential future tax consequences of the temporary differences in notes to the financial statements [Rosenfield and Dent, 1983]. 11A related argument, some contend, that must be settled before deciding on whether or not to allocate is whether income tax is an expense or a distribution of income (see Beresford, et al., [1983, pp. 24-27]). The authoritative position has been that income taxes are an expense (e.g., APB 11, para. 13(b)). The basic argument opposing that view is that, unlike the causal relationship that exists between incurring expenses and earning revenue through increased output of goods and/or services, there is no logical relationship between taxes and benefits received from government. 16 A. Briefly, most accountants believe that of the two alternatives, interperiod tax allocation is the more desirable given the accrual accounting model and its focus on transactions that have present and future cash consequences, as well as the definition of assets and liabilities under that model.12 This view concurs with that of the FASB, which believes that both current and deferred tax consequences must be recognized for events that have been recognized in the financial statements: In the Board’s view, an assumption inherent in an enterprise’s statement of financial position prepared in accordance with generally accepted accounting principles is that the reported amounts of assets and liabilities will be recovered and settled, respectively. The Board believes that assumption creates a requirement under accrual accounting to recognize the deferred tax consequences of temporary differences, (SFAS 96, para.79, emphasis supplied). In general, two questions arise with respect to the interperiod tax allocation problem: 1) how much to allocate (the ‘extent of allocation’), and 2) how to allocate (the ‘method of allocation’). Given that allocation is decided upon, the first question is concerned with whether all, or only some, temporary differences require allocation. The two alternative approaches to this question are comprehensive allocation and partial allocation (see Beresford, et al., [1983]). The ‘comprehensive’ method requires interperiod tax allocation for all temporary differences. ‘Partial allocation’ is an intermediate approach that requires 12Assets are defined as probable future economic benefits obtained or controlled by a particular entity as a result of past transactions or events. Liabilities are defined as probable future sacrifices of economic benefits arising from present obligations of a particular entity to transfer assets or provide services to other entities in the future as a result of past transactions or events. See FASB Concepts Statement No. 6, Elements of Financial Statements. 17 interperiod tax allocation for only some temporary differences. An important issue under partial allocation is to identify which differences require allocation. The second question is concerned with the method of allocation. If it is assumed that some interperiod tax allocation is required, the focus under this question is on issues of measurement and nature of the balance Sheet accounts that result from allocation. Three methods are mentioned in the literature (see Beresford et al., [1983]): deferred, liability, and net-of-tax." This research focuses on the first question--the extent of allocation. However, an overview of all components of the interperiod tax allocation problem is provided in Figure 1. 2.1.2 Historical Development The historical development of the authoritative pronouncements governing interperiod tax allocation under GAAP is discussed in detail by Rayburn [1986] and Beresford et al., [1983]. This section summarizes the major historical landmarks as well as recent developments in this area. 13Under the deferred method, tax effects of temporary differences are calculated using the tax rates and laws in effect when those differences originate. This method emphasizes the income statement. The balance sheet effect of a timing difference is viewed as a deferred credit or charge to be allocated to future periods, instead of as a receivable or payable. Under the liability method, tax effects of temporary differences are calculated using the tax rates and laws expected to be in effect when those differences reverse. This method emphasizes the balance Sheet and the resulting amounts are viewed as assets and liabilities in the usual sense. The net-of—tax method accounts for the tax effects of temporary differences as a reduction in related assets or liabilities. Under this method, the amount of the tax effect of the difference is calculated using either the deferred or the liability method. l8 9 COMPREHENSIVE ALLOCATION A EXTENT VARYING DEGREES 9 OF > OF PARTIAL ALLOCATION ALLOCATION l v INTERPERIOD 5 NO ALLOCATION TAX ALLOCATION (FLOW-THROUGH) PROBLEM T 9 DEFERRED METHOD 5 OF 9 NET-OF-TAX ALLOCATION 9 LIABILITY Figure 1 OVERVIEW OF THE INTERPERIOD TAX ALLOCATION PROBLEM 19 Prior to the 19408, income tax allocation was not an issue primarily because of low tax rates resulting in immaterial tax consequences of book-tax differences. Accounting Research Bulletin No. 18 (ARB 18), Unamortized Discount and Redemption Premium on Bonds Refunded, issued in 1942, was the first official pronouncement to introduce the concept of interperiod tax allocation, without specifically using that term, to be applied by either the net-of—tax or the liability method. ARB 23, Accounting for Income Taxes, issued in 1944, made two major contributions: 1) it introduced the idea of timing differences, and 2) it concluded that income taxes are an expense. ARB 23 required interperiod tax allocation for short-term timing differences (i.e., partial allocation) using the net-of-tax or liability method. In ARB 42, Emergency Facilities-~Depreciation, Amortization, and Income Taxes, issued in 1952, a clear preference for the liability method was indicated for the first time, but the net-of—tax approach also was permitted. With the issuance in 1954 of ARB 44, Declining-balance Depreciation, we have the beginning of the extent of allocation (partial or comprehensive) controversy. Although ARB 44 recommended partial allocation, the concept of comprehensive allocation gained support and ARB 44 (revised) was issued generally requiring comprehensive allocation for the first time. The net-of—tax presentation of timing differences expected to continue indefinitely was permitted. With ARB 51, Consolidated Financial Statements, issued in 1959, evolved the idea of the ‘indefinite reversal criteria’. The bulletin required a parent company 20 to provide deferred taxes on the undistributed earnings of its subsidiaries when those earnings were included in its consolidated income, except if those earnings would remain permanently invested in the subsidiary or distributed in a tax-free liquidation. Because of the growing controversy in this area, Accounting Research Study No. 9 (ARS 9), Interperiod Allocation of Corporate Income Taxes, was conducted in 1966 to evaluate the various issues.“ This was followed in 1967 by APB 11, Accounting for Income Taxes, the first extensive pronouncement on income taxes. Under APB 11 comprehensive allocation was required for all temporary differences (excluding five Special areas to be addressed later) using the deferred method. In 1972, APB Opinion No. 23, Accounting for Income Taxes-Special Areas, was issued to address three of the five items excluded by APB 11: 1.) undistributed earnings of subsidiaries and corporate joint ventures, 2) ‘general reserves’ of stock savings and loan associations, and 3) amounts designated as ‘policyholders’ surplus’ by stock life insurance companies. APB 23 officially introduced the ‘indefinite reversal criteria’ under which interperiod tax allocation was required for the above items unless management did not expect the differences to reverse, or the differences will not reverse for an indefinite future period. Hence, unlike the APB 11 mandate of comprehensive 1"ARS 9 recommended comprehensive allocation using the deferred method for deferred tax debits and the liability method for deferred tax credits. It also recommended discounting long-term deferred tax liabilities, which was proscribed by APB Opinion No. 10, Omnibus 0pinion--1966. Note, ARS’S are not official pronouncements; rather they are advisory in nature. 21 allocation, managerial discretion was permitted in deciding whether the differences were temporary or permanent, based on facts and circumstances. Managerial discretion was allowed because it was believed that "the tax consequences of these events are controlled by the taxpayer and frequently require that the taxpayer take Specific action before the initial difference reverses" (APB 23, para. 6). Several other pronouncements on accounting for income taxes subsequently were issued because the controversy continued to grow. For example, there were attempts to extend the indefinite reversal criteria to other types of temporary differences for which an enterprise may also have ‘control’ over the tax consequences. However, the FASB rejected those attempts stating that APB 23 "was not established to provide general criteria on the applicability of the indefinite reversal rule beyond the specific areas mentioned in it."15 The continuing debate on income taxes led the FASB to add to its technical agenda in January 1982 a project to reconsider all aspects of accounting for income taxes. As a result of that project, a FASB Research Report, Accounting for Income Taxes: A Review of Altematives [Beresford et al., 1983], was issued. This study contains an explanation of the major issues in the area and a discussion of the various accounting and reporting alternatives for each issue. This Study formed the basis for the FASB Statement No. 96 (SFAS 96), Accounting for Income Taxes, issued in December 1987. 1’See FASB Interpretation No. 22, Applicability of Indefinite Reversal Criteria to Timing Differences [F A88, 1978]. 22 SFAS 96 also requires comprehensive allocation for all temporary differences, but using the liability method. The Exposure Draft that led to SFAS 96 had proposed recognition of a deferred tax liability for APB 23 items as well. However, there was widespread disagreement with that proposal and the APB 23 items became the ‘principal focus’ of most comments received by the FASB [Liebtag, 1987]. In the final statement, the FASB decided to retain the exception to comprehensive allocation for the APB 23 items. However, the Board stated that "[i]t continues to believe that the deferred tax liability for those temporary differences should be recognized" (SFAS 96, para. 96). Based on the thrust of these comments and the FASB’S beliefs, it appears that the extent of allocation issue in particular, and accounting for income taxes generally, remain an unsettled area. 2.1.3 Motivation for Research on Extent of Interperiod Tax Allocation The selection of the extent of interperiod allocation as the research issue is motivated by several factors. First, there has been a phenomenal growth over time in firms’ deferred tax balances as documented in several studies.“ The findings in 1“For example, Davidson, Skelton, and Wei] [1977] examined the deferred tax credit account for 3,108 firms on COMPUSTAT for the period 1954 to 1973, and found that 79% of the changes in these accounts were increases. In dollars, the growth was even more dramatic--87% of the total dollar changes were increases. Lantz, Snyir, and Williams [1977] present evidence consistent with the findings of Davidson et al., [1977]. Wheeler and Outslay [1985], in a survey of regulated and non-regulated corporations, report that during 1977 to 1982 deferred tax credits as a percentage of total assets increased from 4.7 to 6.2% and as a percentage of owners’ equity rose from 12 to 16.8%. Growing deferred tax balances have attracted considerable attention and criticism in the popular press as well (e.g., Carmichael [1982]; McGoldrick [1984]). 23 these studies suggest that deferred taxes are simply rolling over instead of being drawn down. An important factor contributing to this development is the genera] requirement of comprehensive allocation,l7 which raises questions about its appropriateness. Second, different approaches to this issue have been taken in other countries.18 In contrast with the international trend, the US. has shifted from partial to comprehensive allocation.” This creates potential problems for multinational firms and has led to more calls for international harmonization of One would expect that this problem has only increased with the significant reduction in the depreciable lives of most assets under the Accelerated Cost Recovery System (ACRS) introduced by the Economic Recovery Tax Act of 1981. This is because most firms’ temporary differences are due to depreciation (e.g., of the 600 firms surveyed by Accounting Trends and Techniques during 1974-84, depreciation was the most frequently cited reason for temporary differences for each of those years--ranging from 76% to 84% of the companies surveyed). 17Wheeler and Outslay [1985] provide some evidence in support of this statement. In their survey, respondents were asked the question whether the comprehensive allocation method resulted in the creation of deferred tax accounts in which reversal was postponed indefinitely. Of the non-regulated companies, 76% agreed that it did whereas only 12% of the regulated companies agreed with the statement. Based on an examination of the written comments accompanying the responses, the authors find that the agreement rate of the regulated companies was low because they focused on ‘turn around’ (individual changes). However, the thrust of the growth-in-deferred-taxes argument is on ‘reversal’ (aggregate changes). 1”For example, in the UK. partial allocation is now required for all temporary differences [Accounting Standards Committee, 1985]. In addition, the International Accounting Standard No. 12 permits either partial or comprehensive allocation-- tax allocation is allowed but not required for temporary differences that will not reverse for at least three years [International Accounting Standards Committee, 1979] 1"See Rayburn [1986] for a chronological review of the authoritative literature on interperiod tax allocation showing this movement. 24 accounting standards.20 Third, a major criticism of APB 11 and related pronouncements was that the deferred tax amount measured and reported under them did not meet the definitions of liabilities and assets in Concepts Statement 6. Although SFAS 96 removes that inconsistency by requiring an asset and liability approach, other conflicts between SFAS 96 and the FASB’S conceptual framework remain.21 Finally, even though SFAS 96 was the result of a comprehensive reconsideration of APB 11 and related pronouncements, the controversy over the extent of allocation remains unresolved. 2.2 Domestic International Sales Corporations In this section the legislative intent and the statutory provisions governing the establishment and operation of Domestic International Sales Corporations (DISCS) 20See, for example, D. Beresford’s (Chairman, FASB) recent statement endorsing the internationalization of accounting standards (FASB Status Report No. 195, dated June 27, 1988, pp. 3-6). 21For example, FASB Concepts Statement No. 1, Objectives of Financial Reporting by Business Enterpn’ses, emphasizes that financial reporting should help users assess the amounts, timing, and uncertainty of an enterprise’s net cash flows. It is contended that by focusing more on amounts that are actually to be paid, partial allocation provides better information on prospective cash flows [Beresford, 1982]. However, SFAS 96 requires the continued use of comprehensive allocation, which emphasizes internal consistency but not prospective cash flows. Second, SFAS 96 requires that no future income or expense be assumed (paragraphs 14 and 15). However, this position is completely different from the FASB’S position in its recently issued pension accounting standard (SFAS 87) which relies extensively on assumptions about future events. In addition, as pointed out by the AICPA’S Accounting Standards Executive Committee (AcSEC), estimates and assumptions about the future are widely used in accounting and not allowing for their use would hurt the relevance and usefulness of the results under SFAS 96 [Liebtag, 1987]. 25 from their inception to the present are discussed in an attempt to fully understand the institutional setting within which the accounting method choice is Studied. 2.2.1 Legislative Intent and Overview of I.R.C. Provisions Under provisions originally established by the Revenue Act of 1971 [US Congress, 1971], a new type of tax entity, the DISC, could be formed for tax years beginning on or after January 1, 1972. The purpose of the DISC provisionszz was to stimulate US. exports and thereby help resolve the acute balance of payment problem being faced by the US. In addition, these provisions were aimed at removing two sources of discrimination against US. exporters. First, US. corporations engaged in export activities were subject to US. income tax currently on their foreign earnings regardless of whether these earnings were retained abroad or repatriated to the US. However, US. corporations producing and selling abroad could defer US. income tax on their foreign earnings as long as they were retained abroad. Second, most trading partners of the US. provided export incentives in the form of a refund of the value added taxes (VAT) paid by exporters and by not imposing income taxes on their foreign earnings. In contrast, the US. did not provide any direct subsidies to exporters. These objectives of stimulating exports and removing discrimination were accomplished by allowing an indefinite deferral of US. income tax on a portion of the export profits generated by DISCS. ”See the House and Senate Reports on the Revenue Act of 1971 [US Congress, House, 1971; US. Congress, Senate, 1971]. 26 The primary statutory provisions affecting DISCS are contained in I.R.C. sections 991 to 997. Briefly, given the Congressional intent to increase exports, the key requirement for DISC status is that substantially all of its gross receipts and assets must be export-related. If disqualification occurs, the DISC benefits are lost and prior years’ benefits must be recaptured. In general, the DISC itself is a nontaxable entity, but its profits are taxed to its stockholders when distributed or deemed to be distributed. Each year, a portion of the DISC’S export profits is deemed distributed to its shareholders, thereby subjecting that income to current US. taxation. However, US. income tax can generally be deferred indefinitely on the remaining portion of the DISC’S export profits unless one of the following events take place: the income is actually distributed, the corporation ceases to qualify as a DISC, its DISC election is terminated or revoked, or it is liquidated. The portion of DISC earnings deferred from tax has varied over time from 50 percent of its export income to 42.5 percent of incremental export income. Notwithstanding the indefinite tax deferral allowed by statute, for financial reporting purposes some firms provided for taxes on the indefinitely deferred DISC earnings, whereas others did not. The resulting interaction between the DISC provisions and the interperiod tax allocation problem explored earlier is summarized in Figure 2. Due to international disputes about the legality of DISCS, the Deficit Reduction Act of 198423 [US Congress, 1984] essentially eliminated the tax incentives to set up DISCS by imposing on DISC shareholders an interest charge 2"This law is also referred to as the Tax Reform Act of 1984. DISC EARNINGS 27 T approx. 50% I approx. 50% 1. v v Distributed Not or deemed deemed distributed distributed T T V V Currently Deferred recognized indefinitely T .1. — Some firms recorded deferred taxes T V COMPREHENSIVE ALLOCATION firms V Some firms did not record deferred taxes T V PARTIAL ALLOCATION firms V i.e., they treat the transaction as a V TEMPORARY DIFFERENCE Figure 2 V PERMANENT DIFFERENCE DISCS AND INTERPERIOD TAX ALLOCATION 28 on the deferred income. The 1984 Act also permanently exempted from tax the accumulated tax-deferred earnings of DISCS. This had important financial reporting implications for firms that had previously recorded deferred taxes on the indefinitely deferred DISC income (comprehensive allocation firms). The permanent tax exemption meant that those deferred taxes would never have to be paid. Hence, the deferred taxes were required to be reversed. In many cases, the, reversal had a Significant impact on the firms’ incomes. The important I.R.C. provisions affecting DISCS are discussed next. 2.2.2 Qualifications of a DISC A DISC is a domestic corporation created under the laws of any of the States or the District of Columbia. To qualify for the Special benefits of indefinite tax deferral, a DISC is required to: (i) satisfy the ‘gross receipts’ and ‘gross assets’ tests (described below), (ii) have only one class of stock with a par or stated value of at least $2,500, and (iii) elect to be treated as a DISC [I.R.C., § 992(a)(1)]. In addition, a DISC is required to have a bank account and maintain separate books and records [I.R.C., Reg. § 1.992-1]. However, not every domestic corporation qualifies for DISC status. For example, tax-exempt organizations, personal holding companies, banks and certain other financial institutions, insurance companies, and mutual funds are ineligible [I.R.C., § 992(d)]. 29 The purpose of the gross receipts and assets tests is to limit the DISC to export related activity. The gross receipts test requires that at least 95 percent of the DISC’S gross receipts consist of ‘qualified export receipts’ [I.R.C., § 992(a)(1)(A)]. In general, export receipts are considered qualified if they are derived from the sale or lease for use outside the United States of ‘export property’, or from providing services connected with the sale or lease of such export property, or from the sale of ‘qualified export assets’ [I.R.C., § 993(a)(1)]. To qualify as export property, the property must be: 1) manufactured, produced, grown, or extracted in the US. by someone other than the DISC, 2) held for sale or lease for use outside the US, and 3) made primarily (at least 50 percent of its fair market value) from US. components (the ‘foreign content test’) [I.R.C., § 993(c)(1)]. Thus, property manufactured, produced, grown, or extracted by a DISC and later sold or leased for use outside the United States does not qualify as export property. Also, generally, exports intended for ultimate use in the US. or exports subsidized by the US. government, intangibles such as patents, goodwill and trademarks, and property determined by the President (by Executive order) to be in short supply, do not qualify as export property [I.R.C., §§ 993(a)(2), (c)(2) and (c)(3)]. Further, the Tax Reduction Act of 1975 [US Congress, 1975] made energy resources such as oil, gas, and certain minerals ineligible for DISC benefits. The gross assets test requires that at least 95 percent of the DISC’S assets be ‘qualified export assets’ [I.R.C., § 992(a)(1)(B)]. Export assets that qualify include 30 export property (discussed above), necessary operational equipment and supplies, receivables from export sales, reasonable working capital requirements, producers’ loans, certain obligations of the Export-Import Bank or the Foreign Credit Insurance Association, and obligations of domestic corporations used solely to finance export sales under agreement with the Export-Import Bank [I.R.C., § 993(b)]. Failure to meet any of the above qualifications results in detrimental tax consequences for the corporation. Not only are DISC benefits lost for the year in which the disqualification occurred, but benefits received in all previous years must be recaptured [I.R.C., § 995(b)(2)(A)]. Recapture of the accumulated DISC benefits is spread over a period equal to two years for each year the DISC is in existence (up to a maximum of 10 years) [I.R.C., § 995(b)(2)(B)]. 2.2.3 Portion of DISC Income Deferred from Tax Since their enactment in 1972, DISC provisions underwent major modifications three times. The thrust of each change was to reduce the benefits allowed to DISC stockholders in the form of tax deferral. Hence, the portion of DISC income deferred indefinitely from tax has varied over time. Because this variation could potentially influence firms’ choice of accounting method for the DISC, a sample selection criteria was imposed such that the study period spanned only one tax regime. From 1972 to 1976, a DISC was deemed to have distributed to its stockholders 50 percent of its export profits and 100 percent of non-export profits, 31 thereby subjecting that income to current US. tax in the stockholders’ hands. Thus, during this period, the available tax deferral was limited to 50 percent of DISC export profits. The sample firms used in this study were required to have a DISC operational in fiscal years ending 1972, 1973, or 1974. Hence, the study period is contained within this period and the subsequent legislative changes only affected the DISC benefits outside the study period. The Tax Reduction Act of 1975 [US Congress, 1975] made the first change in DISC benefits. This Act eliminated DISC benefits for profits from the export of depletable energy products such as oil, gas, and certain minerals. However, the portion of DISC profits Sheltered from US. tax by exporting other products remained at a maximum of 50 percent. The most important change to the DISC provisions was brought about by the Tax Reform Act of 1976 [US Congress, 1976] under which tax deferral was allowed only for incremental export income. This change was motivated primarily by revenue considerations and because Congress believed that providing an export incentive for all exports made the DISC legislation less effective.” 2‘See US. Congress, Joint Committee on Taxation [1976]. It is ironic that when the DISC legislation was originally introduced in 1971, the House [US Congress, House, 1971] had recommended the incremental approach under the rationale that the DISC benefits should be given only to those firms that help correct the adverse balance of payments position by increasing their exports. However, the Senate [US Congress, Senate, 1971] rejected the House proposal on grounds of equity and administrative complexity, granting instead deferred tax treatment to one-half of the export profits of the DISC. The Senate proposal was enacted into law. 32 Under this incremental rule, the DISC’S adjusted taxable income” for the current year attributable to the non-incremental portion of the current year’s export receipts was deemed to be distributed to the DISC’s shareholders; hence not entitled to deferral. However, as before, deferral was permitted on 50 percent of the DISC’s remaining adjusted taxable income (i.e., the income attributable to the incremental export receipts)" The non-incremental portion was set at 67 percent of the average export gross receipts in a moving four-year base period consisting of the fourth, fifth, Sixth, and seventh years preceding the current year.27 For 2The adjusted taxable income of the DISC was defined as the current year’s taxable income reduced by producer’s loan interest, gain on the sale of certain property of the DISC, etc.--amounts considered ‘deemed distributions’ as before [I.R.C., § 995(e), prior to deletion by the Deficit Reduction Act of 1984]. 26Small DISCS-~those with adjusted taxable income of $100,000 or less in the current year--were exempted from the incremental rule. This exemption was subject to a phase-out such that DISCS with taxable incomes of $150,000 or more did not receive any exemption. See US. Congress, Joint Committee on Taxation [1976] 27The deemed distribution (dividend) due to non-incremental export receipts (DDN‘) in year t was computed thus: (0.67) x (EGR?) DD," = ATI, x EGR, where: ATI = adjusted taxable income EGR export gross receipts EGRB = base-period EGR = % (2.EGR), where i = t—4, ..., t-—7. To illustrate, suppose that in 1980 a DISC had EGR of $13,400 and ATI of $500. Assume also that the DISC was established in 1974 and had EGR’s of $0 in 1973, $2,000 in 1974, $6,000 in 1975, and $8,000 in 1976. Then: EGRB = In (0 + 2000 + 6000 + 8000) = $4,000 33 example, for taxable years beginning in 1983, the base period years were 1976 to 1979.” Thus, under the 1976 Act, DISC benefits were limited to 50 percent of the taxable income attributable to incremental export gross receipts. The Tax Equity and Fiscal Responsibility Act of 1982 [US Congress, 1982] further reduced DISC benefits by increasing the incremental portion of DISC income deemed distributed to 57.5 percent. This provision resulted in limiting deferral of US. tax on at most 42.5 percent of a DISC’S incremental export profits. 2.2.4 Removal of DISCS Ever Since DISCS were enacted into law, they had been a source of contention between the United States and several signatories to the General Agreement on Trade and Tariffs (GATT). These countries claimed that DISCS violated GATT by providing US. businesses an illegal export subsidy. In 1976, a GATT panel upheld this claim because no interest was charged on the taxes deferred by DISCS [US Congress, Joint Committee on Taxation, 1984]. (0.67) x (4000) DD“ = 500 x = $100 13400 DD' = 0.50 (500 — 100) = $200. Thus, the total deemed distribution of the DISC in 1980 is $300, and the income eligible for deferral is $200. (Example adapted from US. Congress, Joint Committee on Taxation [1976]). ”For taxable years beginning before January 1, 1980, the base period years were the DISC’S taxable years beginning in 1972, 1973, 1974, and 1975 [I.R.C., Reg. § 1.995-6]. 34 The primary complaint of the GATT members was that DISCS did not serve any economic purpose except to provide tax deferral to US. businesses. The US. defended DISCS on the grounds that they provided U.S. exporters with tax treatment that approximated its European trading partners’ territorial tax systems, which generally do not impose taxes on profits earned abroad. In contrast, the US. uses a worldwide or ‘global’ tax system which subjects US. taxpayers to US. income tax on their worldwide income, resulting in double taxation if the foreign source income is subject to foreign taxes as well. The primary mechanism in the I.R.C. that mitigates the hardship of double taxation is the foreign tax credit. In the case of DISCS, double taxation was partly avoided by allowing an indefinite tax deferral on a portion of export income (discussed earlier). The dispute over DISCS peaked in 1981 when the European Community requested from the GATT Council permission to retaliate against the US. The charge that DISCS were an illegal export subsidy and the threat of retaliatory action convinced the US. Congress that continuation of DISCS could potentially lead to the isolation of the US. in international trade. Thus, in 1982, while not conceding that DISCS violated GATT, the US. made a commitment to propose legislation addressing the concerns of the GATT members. The result of that commitment was to generally replace the system of DISCS by the system of Foreign Sales Corporations (FSCS) in the Deficit Reduction Act of 1984. FSCS differ from DISCS in several important ways: 1) FSCS are foreign corporations; DISCS were domestic, 2) FSCS must have a foreign presence, 35 economic substance, and must perform activities outside the US; DISCS were not required to meet any such requirements-~they were essentially ‘paper corporations’, and 3) FSCS provide for permanent exemption of a portion of their export income from US. taxation; DISCS provided for an indefinite deferral of US. taxes. These differences were designed to satisfy GATT. However, Congress was cognizant of the fact that the FSC requirements would impose undue burden on small exporters. Hence, the 1984 Act did not abolish DISCS per se; instead it continues the deferral treatment on income attributable to $10 million or less of qualified export receipts.” Deemed distributions relating to the incremental rule and to one-half of the DISC’S income are eliminated, i.e., substantially all of the DISC’S income attributable to the $10 million or less of qualified export receipts can now be deferred.” But an interest charge at a rate tied to the T-bill rate is now imposed on DISC shareholders on the deferred amounts [I.R.C., § 995(f)]. 2"I.R.C., § 995(b)(1)(E). If export receipts exceed $10 million, the DISC would not be disqualified; there would only be no deferral of income associated with the excess export receipts. 3"However, to make DISCS comparable with FSCS, one-seventeenth of the excess of DISC income over certain deemed distributions is deemed distributed [I.R.C., § 995(b)(1)(F)]. 36 2.2.5 Permanent Exemption of DISC Deferred Taxes The 1984 Act also provided that the accumulated tax-deferred income of DISCS would be permanently exempt from taxation." This resulted in a windfall for corporation with DISCS. As Table 2.1 Shows, over the 12-year period beginning in 1972, the tax-sheltered DISC earnings were estimated at about $28.15 billion, which translates into an estimated $13.19 billion in forgiven deferred taxes."2 Although the permanent exemption benefited all DISC firms, it had important financial reporting implications for the comprehensive allocators. The FASB required these firms to report the effects of the 1984 Act by reversing the previously provided deferred taxes and Showing that amount 1) as a reduction of income tax expense of the current year, 2) in one interim period only rather than as an adjustment of the estimated annual effective tax rate, and 3) not as an extraordinary item.” This resulted in significant income boosts for some firms.” 31Specifically, this was achieved by treating that income as ‘previously taxed income’ and, therefore, exempt from taxation (see § 805(b)(2) of the Deficit Reduction Act of 1984 [US Congress, 1984]). 32The deferred taxes forgiven is computed at the top marginal statutory tax rate--48 percent during 1972-78 and 46 percent during 1979-84. The Wall Street Journal [June 7, 1984], however, reported the accumulated tax-Sheltered DISC earnings to be estimated at $24.4 billion, resulting in $9.8 billion in taxes forgiven. 33See FASB Technical Bulletin No. 84-2, Accounting for the Effects of the Tax Reform Act of 1984 on Deferred Income Taxes Relating to Domestic International Sales Corporations [FASB, September 18, 1984]. 3‘For example, in fiscal 1985, Digital Equipment Corp. showed an increase of $63.25 million in its after-tax income due to the reversal of the DISC taxes. This benefit represented 14.68 percent of its 1985 income before-taxes. In addition, Digital showed a benefit from DISCS of $8.9 million in fiscal 1984. See also Business Week [August 27, 1984, pp. 77-78] for the effects on some other firms. 37 Table 2.1 DISC DEFERRALSa DISC Amount Tax Tax Number taxable deemed deferred Yearb of returns income distributed‘ incomed (1) (2) (3) (3) - (2) 1972 2,826 $ 1,566 $ 776 $ 800 1973 4,162 3,149 1,579 1570 1974 5,498 4,783 2,416 2367 1975 6,431 4,772 2,420 2352 1976 6,911 5,071 3,499 1572 1977 6,665 5,234 3,715 1519 1978 7,208 6,427 4,360 2067 1979 7,933 8,461 5,397 3064 1980 8,665 9,875 6,270 3605 1981 9,408 10,952 7,187 3765 1982 9,663 10,156 7,080 3076 1983 9,898 10,082 7,692 2390 28147 'All figures are estimates based on samples-~money amounts are in millions of dollars. I’Tax year refers to accounting periods ended between July of one year and June of the following year. However, for 1972, the effective date of the legislation was January 1, 1972; therefore, they include only part- year accounting periods for some corporations. cEstimates include small amounts oftdistributions considered received by stockholders from prior years' DISC taxable income. dTax deferred income amounts were inferred from the IRS data. Source: Internal Revenue Service, Statistics of Income Bulletin, Vol 6, No. 2 (Washington D.C.: Fall 1986). 38 On the other hand, without the tax forgiveness, the partial allocators would have seen their earnings fall because they would have to pay taxes on the accumulated DISC earnings for which previously they had not set aside reserves. For many firms the earnings drop would have been very significant.” While the 1984 tax forgiveness and the resulting reversals reported by some firms are not the events of interest in this study, they nevertheless highlight the importance of the financial reporting choice faced by firms, the length of time for which it was available, and the material dollar amounts involved. 3’For example, as of December 31, 1983, Boeing Co. and McDonnell Douglas Corp. had undistributed DISC earnings of $331 million and $323 million, respectively, for which federal income taxes had not been provided. This represented 69.7 percent and 74.2 percent, respectively, of their 1983 pretax earnings. Chapter Three THEORETICAL FRAMEWORK AND PREVIOUS RESEARCH In this chapter, the evolution of positive accounting theory--the basis for the hypotheses developed and tested in this thesis--is summarized. Next, a review of the relevant literature, especially studies empirically testing the theory’s propositions, is provided. The literature review is deliberately kept brief in this chapter with relevant research cited and discussed later as it applies. Both the theory and the literature outlined in this chapter have been extensively reviewed in Holthausen and Leftwich [1983] and Watts and Zimmerman [1986]. The following discussion draws heavily on their work. 3.1 Positive Accounting Theory 3.1.1 Its Genesis Up until the late 19705, the empirical financial accounting literature following Ball and Brown [1968] focused on accounting’s role in providing information for security valuation (the ‘information perspective’). Borrowing from developments in finance, this literature was founded on a belief in the efficient market hypothesis (EMH) and has demonstrated the potential usefulness of accounting earnings by documenting its association with stock prices (the ‘information content of earnings’ 39 40 Studies).1 However, this information perspective did not provide any predictions or explanations for accounting method choices that have no direct cash flow effects (i.e., accounting choices other than inventory methods). Instead, an extension of the information perspective to its logical extreme resulted in an accounting irrelevance proposition: in an efficient market, non-tax related accounting method choices are irrelevant because they do not have any cash flow effects. In contrast to the irrelevance hypothesis, Watts and Zimmerman [1978] introduced contracting and political cost arguments to offer a potential explanation for why firms’ accounting method choices have economic consequences. These arguments came to be referred as positive accounting theory2 and spawned another stream of empirical financial accounting literature that has focused on explaining variations in accounting practice across firms and industries. A critical assumption underlying these arguments is the existence of positive costs, broadly referred to as ‘contracting costs’, in markets and the political arena. Contracting costs include transaction costs, agency costs, bankruptcy costs, and information costs. These costs are borne by various parties to the firm and are tied to the firm’s accounting method choices. Thus, even tax-neutral choices of accounting procedures have a cash-flow effect and can cause wealth transfers to ’See Ball and Foster [1982], Lev and Ohlson [1982], and Watts and Zimmerman [1986, pp. 37-155] for a review of this literature. ”These arguments also have been referred to as ‘economic consequence theories’ [Holthausen and Leftwich, 1983], and ‘contracting and size theories’ [Christie, 1990]. 41 take place. Holthausen and Leftwich [1983, p. 81] summarize the importance of the positive contracting cost assumption as follows: A world with zero contracting and monitoring costs yields an accounting irrelevance proposition analogous to the Miller-Modigliani capital structure and dividend irrelevance propositions in finance. The value of a firm is invariant to the choice of accounting rules in such a world because users of accounting numbers can unbundle the accounting package offered to them by corporations (i.e., they can costlessly restate the financial statements using whatever measurement rules they choose). Strictly speaking, there is no role for accounting in a world with costless contracting and monitoring. The positive contracting cost assumption borrows from important developments in finance (regarding firms’ capital structure) and the industrial organization literature in economics. The capital structure debate in finance is rooted in Modigliani and Miller’s [1958] classical argument that in the absence of bankruptcy costs and corporate taxes, the value of the firm is independent of its capital structure. With the introduction of corporate taxation (which subsidizes interest), the value of the firm is maximized with 100 percent debt financing. However, observed systematic variations in firms’ capital structures (debt-equity ratios) across industries and over time led to alternative explanations. Among them are Jensen and Meckling’s [1976] contracting explanations based on positive bankruptcy and agency costs of debt. Because accounting numbers are used in the contracts aimed at reducing these costs, the contracting explanations include a role for accounting procedures in determining firm value. The industrial organization literature in economics is concerned, in part, with theories of the political process. For a long time, one view of the political process 42 was that politicians are motivated in improving social welfare (the ‘public interest hypothesis’). More recently, however, economists modeled politicians similar to other market participants as being motivated by self-interest (e.g., Stigler [1971]; Peltzman [1976]). Assumed in this self-interest theory is the existence of positive information, lobbying, and coalition costs, which are part of contracting costs. The existence of these costs and the resulting wealth transfers under the self-interest theory include a role for accounting method choice because accounting numbers are employed in the political process, thereby providing an incentive for managing reported numbers via the choice of accounting methods. 3.1.2 The Theory and its Implications for Accounting Method Choice The objective of positive accounting theory is to explain and predict accounting practice [Watts and Zimmerman, 1986, p. 2]. The agency theory arguments used in the capital structure debate in finance offer one potential explanation for variations in accounting procedures (practice). In this theory, firms are not viewed as having a separate existence but rather as being a ‘nexus of contracts’ between self-interested individuals (managers, stockholders, and bondholders), each seeking to maximize utility (welfare) dependent upon the firm’s performance and continued existence. The conflicting interests of these parties give rise to incentives to transfer wealth from other parties connected with the firm (agency costs), which, in turn, reduce firm value and hurt the firm’s chance of survival. This induces the parties to write contracts restricting their actions, thus reducing agency costs and increasing the firm’s value and its chance of survival. 43 The literature exploring these conflicts has developed along two dimensions: 1) the literature devoted to the ‘positive’ aspects of agency theory as discussed in Jensen and Meckling [1976]--also referred to as the property rights literature, and 2) the mathematical principal-agent literature dealing with the ‘normative’ aspects of agency theory as reviewed in Baiman [1982]. Both dimensions are concerned with the same phenomenon--the agent’s actions in the presence of moral hazard and asymmetric information’--and are highly complementary. However, they differ in that the former uses less formal models but has helped in generating empirically testable hypotheses (external validity), while the latter uses formal mathematical models and emphasizes analytical rigor (internal validity).‘ The discussion here is limited to the positive agency theory (property rights) literature since the proposed hypotheses are derived from that literature. Two well known sources of agency conflict are between 1) bondholders and stockholders, and 2) managers and suppliers of outside capital. Contracts between bondholders and stockholders contain a variety of restrictions on stockholders’ 3T he moral hazard problem arises because the agent’s (manager’s) action choices are not observed. Asymmetric information results from the agent possessing private information not known to the principal (stockholders). ‘Specifically, the positive agency theory perspective assumes the existing contracts between managers, stockholders, and bondholders are efficient and investigates the incentives faced by these parties. In contrast, the normative agency theory perspective is concerned with how to structure contracts between agents (managers) and principals (stockholders) that would induce the agent to expend the optimal amount of effort such that the principal’s welfare is maximized [Jensen and Meckling, 1976]. Raviv [1985, p. 245], however, notes that "there is only one theory, it is positive and devoted to the same issue. Both streams of the literature attempt to minimize agency costs or equivalently to design Pareto-efficient contracts." 44 ability to transfer wealth from bondholders to themselves. These restrictions are known as bond covenants and are analyzed by Smith and Warner [1979]. Management compensation contracts are written to align the divergent interests of managers and stockholders. The bond covenants and compensation agreements are important in accounting method choice studies because accounting typically plays a crucial role in these contracts. Restrictions are placed on parties’ actions in terms of accounting numbers (e.g., the use of debt-equity ratios in lending agreements; the use of accounting earnings in management incentive schemes). Violation of these restrictions imposes costs on the shareholders and managers (e.g., debt renegotiation costs). Because alternative accounting procedures can affect the calculation of the numbers on which these restrictions are based, the choice of accounting procedures can have a cash-flow and valuation effect. This linkage between a firm’s and firm managers’ cash flows and accounting procedures under costly contracting provides a potential motivation for managers not to be indifferent among tax-neutral accounting procedures that do not affect the firm’s political costs. A second potential explanation for variations in accounting procedures relies on political cost arguments that originate from the disclosure regulation debate in economics. Early rationales for corporate disclosure regulation (e.g., the ‘naive investor’ theory, the ‘functional fixation’ hypothesis,’ and the ‘market failures’ 5According to the ‘naive investor’ theory, investors untrained in accounting cannot interpret reported earnings of firms that use different accounting practices. Under the ‘functional fixation’ hypothesis, individual investors fixated on, say 45 rationale“) assumed that such regulation is aimed at improving social welfare (the ‘public interest hypothesis’). However, using implications of the efficient markets hypothesis and other arguments, these rationales reduce to claims about relative costs and benefits of private and governmentally regulated production of information, which are empirical issues. Observed inconsistencies between regulators’ seeming lack of concern with assessing the costs and benefits of regulation’ and the public interest hypothesis, led to the assumption that politicians and regulators, like individuals in general, act in their self-interest rather than being motivated by social welfare considerations. Under the self-interest assumption, the political process is viewed as "a competition among individuals for wealth transfers" [Watts and Zimmerman, 1986, p. 224]. Politicians and regulators are hypothesized to affect wealth transfers by "solving perceived or actual crises" [Watts and Zimmerman, 1986, p. 242]. Accounting numbers are used not only in identifying the crises, but also in subsequent earnings, interpret the earnings of firms the same way, regardless of the accounting procedures used to calculate them. Under both theories, protection of investors is used to justify regulating disclosure requirements. This inability of investors to discriminate among accounting numbers also is referred to as the ‘mechanistic’ view of accounting choice. ”The ‘market failure’ rationale is concerned with the socially non-optimal production of a good. An accounting market failure is said to exist when the information contained in accounting reports is non-optimal in a Pareto sense. Under this rationale, government regulation is posited to cure the non-optimality. ’See, for example, SEC’S Accounting Series Release (ASR) 190. In ASR 190, the SEC justified current replacement cost disclosures by large corporations on the ground that the "benefits of disclosure clearly outweigh the costs" without providing any evidence to support that claim. 46 regulation aimed at solving the crises. Specifically, firms’ ‘reported’ profits are suggested to influence the actions of Congress and regulatory agencies. For example, the 1980 windfall profit tax imposed on oil companies was largely motivated by the monopoly-like profits amassed by these firms during the 1970s; the transfer from individuals to corporations of $120 billion in tax burden over five years by the 1986 Tax Reform Act was partially prompted by reports of large profitable corporations not paying taxes (this is discussed in more detail later). Since accounting procedures affect the numbers used in the political process, politically sensitive firms are not indifferent between tax-neutral accounting methods. Specifically, positive accounting theory hypothesizes that firms subject to potential wealth transfers in the political process can reduce their political visibility by adopting earnings-reducing procedures. Summary. The positive accounting theory seeks to explain why firms choose different accounting procedures. The theory is based on the assumption that contracting in markets, within firms, and in the political arena is costly. In such a world, accounting numbers have economic consequences by affecting the firm’s and firm managers’ cash flows. This cash flow linkage provides a motivation for firms not to be indifferent to the choice of accounting procedures. The theory yields empirically testable predictions driven by the positive contracting costs assumption. Studies testing those predictions are thus of interest and are discussed next. 47 3.2 Previous Research The positive accounting theory outlined above has been used to provide explanations for various observed phenomenon: stock price effects of voluntary and mandatory accounting method changes,“ lobbying behavior in response to proposed accounting standards,” early versus late adoption of accounting standards,” and voluntary accounting method choices. Many researchers have addressed these phenomena.” 8For example, Holthausen [1981] looks at the stock price impact of voluntarily switching from accelerated to straight-line depreciation, whereas Leftwich [1981] and Lys [1984] investigate the stock price impact of mandatory accounting changes (business combinations and oil and gas accounting, respectively). While Holthausen [1981] finds no support for either the leverage or the bonus scheme hypotheses, Leftwich [1981] and Lys [1984] find limited support for the leverage hypothesis. Watts and Zimmerman [1986, pp. 284-311] critique these studies and conclude that 1) voluntary accounting change announcements are unlikely to have stock price effects because of market expectations, and 2) generally stock price based tests of positive accounting theory (relative to the choice studies) are unlikely to find support for the theory’s predictions because the magnitude of the cash flow effects of accounting changes is possibly small relative to the value of the firm. See also Lev and Ohlson [1982], and Holthausen and Leftwich [1983] for a review of these studies. ’’For example, Watts and Zimmerman [1978] examine the factors likely to affect corporate lobbying for accounting standards by investigating managers’ submissions on FASB’S Discussion Memorandum on General Price Level Adjustments (GPLA). They find support for the Size hypothesis, but their results appear to be driven by the large firms in the oil industry. See McKee et al. [1984] for a critique of the Watts and Zimmerman [1978] study. ”For example, Ayres [1986] studies the economic factors associated with the early v. late adoption of FASB Statement No. 52 regarding foreign currency translation. 1‘See Holthausen and Leftwich [1983], Kelly [1983], and Watts and Zimmerman [1986] for a review. 48 Since the question investigated in this study deals with the voluntary choice of an accounting technique, studies investigating this phenomenon are the focus of the following review. Other works dealing with methodological issues are considered later (in chapters 4 and 5). 3.2.1 Voluntary Accounting Method Choice Studies The voluntary accounting method choices analyzed in prior studies include interest capitalization v. expensing, research and development capitalization v. expensing, depreciation methods, inventory valuation methods, investment tax credit methods, amortization period of pension costs, and oil and gas accounting methods. Since none of these studies has looked at the extent of interperiod tax allocation choice, this research attempts to fill that void. Bowen, Noreen, and lacey [1981] examine the determinants of the corporate decision to capitalize interest costs for capital projects that increase current period’s reported earnings. They hypothesize that interest capitalizers are more likely to have 1) management compensation explicitly linked to accounting earnings (bonus scheme hypothesis), 2) more binding dividend constraints, 3) lower interest coverage ratios, and 4) higher leverage. They also hypothesize that larger firms (especially in the politically sensitive oil industry) are less likely to capitalize interest. They use a ‘matched-pairs’ design (firms matched on industry) and conduct both univariate and multivariate tests of their hypotheses. They find evidence consistent with the three debt variables (dividend constraint, interest coverage, and leverage), but no support for the bonus scheme hypothesis. Also for firms outside the oil 49 industry, the results of the political costs (Size) hypothesis are contrary to expectations. Daley and Vigeland [1983] examine why some firms, prior to 1974 when the FASB mandated the expensing of research and development (R&D) costs,12 capitalized R&D costs, thereby increasing current period reported earnings. In addition to the three leverage hypotheses used in Bowen et al. [1981], they hypothesize that R&D capitalizers will tend to have more public debt as a proportion of total debt in their capital structure. They conduct both univariate and multivariate analyses and find that coefficients of all independent variables have the predicted Sign and are statistically significant, except the interest coverage ratio. Further, contrary to expectations, the size variable is significant only for the small firms sub-sample. Also, in a matched-pairs analysis (firms matched on industry), only the public and non-public leverage coefficients were significant. Dhaliwal, Salamon, and Smith [1982] examine the determinants of the choice of depreciation methods adopted by firms. Apart from Size and leverage, they also use ‘firm control’ as an independent variable. They predict that a manager- controlled firm is more likely to use accelerated depreciation for tax purposes and straight-line for financial accounting purposes, thereby shifting reported earnings from later periods to earlier periods, whereas an owner-controlled firm is more likely to use accelerated depreciation for both tax and financial reporting 12See FASB Statement No. 2, Accounting for Research and Development Costs. 50 purposes.l3 They find the debt-equity and firm control variables are statistically significant at conventional levels; the size variable is significant only at the 0.15 level. Hagerman and Zmijewski [1979] examine the determinants of four accounting method choices: the accounting for inventory (LIFO v. FIFO), depreciation (accelerated v. straight-line), investment tax credit (flow-through v. deferred), and pension costs (short v. long amortization period). They dichotomize these choices as either income-increasing or -decreasing policies and test whether a firm’s income policy is a function of Size, risk, capital intensity, and concentration ratio (surrogates for political sensitivity), and the existence of a management incentive compensation scheme. Their results are not consistent across all accounting choices; Size and concentration ratio always have the predicted sign and are significant twice," but the other explanatory variables are not always Significant and even switch signs. Finally, there are three studies [Dhaliwal, 1980; Lilien and Pastena, 1982; Johnson and Ramanan, 1988] that deal with the choice between full cost (FC) and 13A firm was classified as management-controlled is no single block of stock greater than five percent was controlled by any party. A firm was classified as owner-controlled if one party owned 10 percent or more of the voting stock and exercised active control, or if one party owned 20 percent or more of the voting stock. ”Firm size is significant for the depreciation and investment tax credit accounting method choices, whereas concentration ratio is significant for the inventory and investment tax credit accounting method choices. 51 successful efforts (SE) for oil and gas accounting.” Because the results of these studies are generally similar, discussion is limited to Johnson and Ramanan [1988] Since it is the most recent. Johnson and Ramanan compare a sample of firms that changed from SE to FC between 1970 and 1976 with firms that retained the SE method throughout that period. They overcome a limitation of previous oil and gas accounting studieS--a ‘static ex-post’ research design--by conducting a logit analysis for the years -2, -1, 0, and +1, where year 0 corresponds to the year of FC adoption. Of the explanatory variables examined by them (size, leverage, interest coverage, ratio of dividends to inventory of funds available, and two surrogates for ‘drilling intensity’), only leverage and one of the drilling intensity surrogates were significant. The above voluntary accounting choice studies have tested the hypotheses emanating from positive accounting theory and have found some empirical regularities. Overall, the evidence indicates that there exists a relationship between firms’ accounting method choice on the one hand and surrogate measures of debt covenant violations, political costs (firm size), and management compensation (bonus plans) on the other. However, the evidence is not consistent across different accounting choices and time periods. This is particularly true of the firm 1’The two methods differ in their treatment of exploration costs for ‘dry’ wells (i.e., wells that do not produce commercially profitable deposits). Under the FC method, all exploration costs for ‘dry’ as well as ‘wet’ wells are capitalized and amortized against the cash flows generated from the wet wells. Under SE accounting, exploration costs of only the wet wells are capitalized and amortized; costs associated with the dry wells are immediately expensed. Usually, reported income in the current period is lower under the SE method. 52 size and bonus plan hypotheses. Further, the independent variables together explain only small proportions of the total cross-sectional variation in firms’ choices among alternative accounting procedures permissible under GAAP. The above shortcomings suggest two courses of action. First, in order to determine the robustness of prior findings, the hypotheses need to be tested with respect to other accounting procedure choices. This is consistent with Christie’s [1990] observation that because the positive accounting literature is in its early stages, ‘finding regularities’ is important. This study fills that need by investigating a voluntary choice not studied before. Second, in order to increase the proportion of the variability explained by these hypotheses more powerful tests are required. Recent research suggests that using other surrogates for political costs together with firm size [Wong, 1988], could potentially provide more powerful tests of the predictive ability of positive accounting theory. In this study, this is accomplished by including firms’ effective tax rate as an additional variable to capture firms’ political costs. Finally, to increase the cross-sectional variability explained by these models, increased emphasis must be paid to omitted variables. In this study, the role of auditors is included in addition to the debt covenant and political cost variables. 3.2.2 The Interperiod Tax Allocation Literature Although interperiod tax allocation has not been studied within the positive accounting theory framework, there exists a substantial body of literature on the 53 issue. In the interest of completeness, this literature is discussed here, but only briefly because it does not directly relate to the purpose of this study. The existing literature on interperiod tax allocation can be classified along two dimensions--theoretical and quasi-empirical. In the theoretical literature, arguments for and against nearly every facet of the interperiod tax allocation problem, including the extent of allocation, can be found.“ This literature consists mostly of ad hoc reasoning that is not backed by analytical models or empirical evidence and thus not useful in generating hypotheses about the variations in accounting practices. It is possible that the conceptual merits of these methods may drive the choice of firms. However, as in prior accounting method choice studies, it is hypothesized that more practical considerations (e.g., potential violation of debt covenants and political costs) govern firms’ selection of accounting procedures. For these reasons, the theoretical literature is not discussed here; instead the portion of this literature dealing with the extent of allocation, the issue relevant to this study, is included in Appendix A. l"See Beresford et al., [1983] for an overview of this literature. Detailed arguments for and against the interperiod tax allocation alternatives--flow-through, comprehensive allocation, and partial allocation-~can be found in Rosenfield and Dent [1983], Defliese [1983], and Beresford [1982], respectively, and are summarized in Appendix A. 54 The quasi-empirical" literature documents the growth of deferred tax balances (e.g., Davidson et al. [1977]).‘8 While these studies alert us to the problem of interpreting the deferred tax balances arising from the use of current accounting techniques, they do not explain or predict accounting practice. Beaver and Dukes [1972; 1973] are examples of early empirical research on interperiod tax allocation. Both studies used the relation between accounting earnings and stock prices to assess the ‘desirability’ of alternative accounting procedures. However, they did not attempt to explain or predict accounting practice with respect to interperiod tax allocation, which is the primary purpose of this research.” 3.3 Limitations of the Theory and Research The positive accounting theory outlined in this chapter and the studies (discussed above) empirically testing the theory’s propositions suffer from various limitations. The more general of these limitations are discussed here, while specific "These studies are labeled ‘quasi-empirical’ because they do not involve hypothesis formulation and testing, which is normally associated with ‘empirical’ accounting studies. Literally speaking, these studies are empirical. An alternative label for these studies could be ‘non-experimental’. ”See chapter two, section 2.1.3 for the findings in this and other related studies. ”It should be noted that desirability of alternative accounting procedures is a normative question that is designed to yield prescriptions. While positive accounting theory is concerned with explaining and predicting accounting practice, it does not render normative questions unimportant [Watts and Zimmerman, 1986, p. 9]. 55 limitations that apply to the hypotheses tested in this study are discussed in the next chapter. An important limitation is that the positive accounting theory is not fully formulated. The incomplete nature of the theory is partly manifest in the low explanatory power (R-squares) of models based solely on contracting and political cost arguments. One factor contributing to the low R-squares could be that of all the explicit and implicit contracts entered into by a firm, the literature to date has largely concentrated on debt contracts and bonus agreements because they are observable.” Including other contracts may increase the theory’s explanatory power. Another factor that may account for the low R-squares is that the theory "is a theory of extremes (e.g., ‘closeness’ to covenants) which therefore cannot explain the choices of the mass of firms not near the extremes" [Christie, 1990]. Other limitations can be broadly classified as measurement issues and omitted variables problems. 3.3.1 Measurement issues -- dependent variable A potential avenue for increasing the theory’s predictive ability is to use a portfolio of accounting choices instead of a single procedure choice as the dependent variable. In prior research, this has been attempted in two ways: 1) by using an income Strategy approach, and 2) by using net accounting accruals. ”Liberty and Zimmerman [1986] is an exception. They examine managers’ accounting method choices around labor union contract negotiations. These contracts differ from debt contracts and bonus agreements in that the accounting numbers are believed to affect the process only implicitly. 56 The income strategy argument. Studies using the choice of a single accounting procedure as the dependent variable implicitly assume that the selection of accounting procedures is independent of one another. Zmijewski and Hagerman [1981] (Z-H) hypothesize that it is unlikely that managers choose each accounting policy independently; rather, "management will adopt a multi-dimensional income strategy for the firm, with each policy being one dimension of that decision" (p. 133). Hence, Z-H use the firm’s ‘income strategy’ as the dependent variable in their model explaining firms’ choice of accounting procedures." If managers in fact choose accounting procedures on a portfolio basis, Z-H’S approach would provide more powerful tests of positive accounting theory [Watts and Zimmerman, 1986, p. 248]. There are two problems with the income strategy approach. First, it is impossible to measure the exact impact of the various accounting procedures used by the firm. Hence, "arbitrary assumptions of each accounting choice’s importance in determining [the firm’s] overall income strategy, as opposed to the actual dollar impact for each firm" become necessary [Holthausen and Leftwich, 1983, p. 93]. This may have led to the low explanatory power of Z-H’s cross-sectional model and its failure to predict Significantly better than the naive prediction that the firm will 21Z-H combine the effect of four accounting method choices (depreciation, inventory, pension costs and investment tax credit) to form strategies based on whether the effect of these choices is income-increasing or income-decreasing. 57 choose the most common strategy.22 Second, theoretically all accounting choices of the firm should be considered simultaneously. However, this makes the task extremely formidable. For these reasons, other accounting choice studies have focused on single procedures, and I do likewise in this research. The use of net accruals. An alternative to using the firm’s income strategy is to use its net accounting accruals as the dependent variable (e.g., Healy [1985]; Liberty and Zimmerman [1986]). Like the income Strategy approach, net accruals capture in one measure the aggregate net effect of all accounting procedure choices. Hence, they overcome the deficiency of single procedure choice studies and can potentially provide more powerful tests of the positive accounting theory. However, as Watts and Zimmerman [1989, p. 9] point out, "[the] use of accounting accruals as a summary measure of accounting method choice suffers from a lack of control of what accruals would have been in the absence of managerial accounting discretion." Without a model of accruals, this measure is, at best, a noisy surrogate for managerial opportunism, making it unclear whether the approach is indeed ‘better’ than single method choice studies. 2’Press and Weintrop [1990] replicate the Z-H study using 1985 data on a sample of 83 firms (Z-H used 1975 data for 300 firms) with marginally improved R-square and percentage of firms correctly classified. Relative to the naive policy of picking the most common strategy, Press and Weintrop’s models incorporating the leverage variable are significantly better at the .11 level (compared with .25 in Z-H). 58 3.3.2 Measurement and Specification issues -- independent variables The fundamental issue here is that although positive accounting theory is founded on the existence of contracting and monitoring costs in the market and political processes, the magnitude of these costs and their variation across firms are not taken into account in empirical tests of the theory. These costs are treated as unobservable and such proxy variables as leverage and firm size are employed instead. However, using proxy variables creates interpretation problems both when test results are statistically significant and when they are not [Leftwich, 1990]. Another problem is that generally the variables that proxy for the contracting and political costs are specified as a linear function of the accounting method choice studied. However, potential interactions among the independent variables which are thus ignored, call into question the linear specification. For example, avoiding costly debt contract violations motivates the choice of income- increasing accounting methods to relax the contract constraints, whereas the political process provides incentives to reduce reported earnings for some firms. These opposing incentives require managers to trade-off between contracting and political costs before selecting among accounting procedures. The final choice would depend upon the relative magnitude of the costs involved. But as Watts and Zimmerman [1986, p. 243] admit, 59 Little is known about the relative magnitude of political, regulatory, and contracting costs. Hence the parameters of managements’ decisions and how they vary across firms cannot, a priori, be specified. However, empirical evidence indicates that the selection of accounting procedures varies with variables that are likely to be related to political costs and contracting costs. Thus some statements and predictions can be made as to the trade-off of political costs and contracting costs. 3.3.3 Omitted Variables Issues Correlated omitted variables. As with any theory, it is possible that the contracting and size variables hypothesized to influence accounting method choice might surrogate for other omitted variables correlated with the included variables. For example, the firm’s production-investment opportunity set (of which leverage and size are characteristics) may drive its contracts of which the accounting system is a part. In this scenario, the choice of accounting procedures is endogenously determined with the firm’s production and investment decisions. This is also referred to as the ‘endogeneity argument’, and under it the focus shifts from ex- post managerial opportunism to ex-ante efficiency reasons (maximization of firm value) as the driving force behind accounting method choice (see Watts and Zimmerman [1989]). While positive accounting theory admits both rationales to co-exist, most of the empirical tests of the theory take the firm’s observed contracts as given and focus on the ex-post choice of accounting methods. This approach implicitly assumes that managers behave opportunistically and does not control for differences in the ex-ante set of acceptable accounting methods available to them. 60 Other omitted variables. As discussed before, variables based on debt contracts and bonus agreements have been included in the models. However, other explicit and implicit firm contracts (e.g., sales contracts, union contracts,23 etc.) as well as intra-firm transactions (e.g., transfer pricing; cost allocations [Ball, 1987]) also use accounting numbers and thus influence the choice of accounting methods, but are omitted in most studies. 2"Liberty and Zimmerman [1986], however, fail to find support for the hypothesis that managers manipulate reported earnings during labor union contract negotiations. Chapter Four HYPOTHESIS DEVELOPMENT The purpose of this chapter is to develop empirically testable hypotheses regarding the determinants of firms’ choice of methods in accounting for the extent of interperiod tax allocation. The hypotheses are formulated within the positive accounting theory framework presented in the previous chapter. Before the hypotheses are developed, it is necessary to understand the financial statement effects of comprehensive and partial allocation. 4.1 Financial Statement Effects Under comprehensive allocation, a firm with a DISC records deferred taxes on the indefinitely deferred DISC earnings. This increases the income tax expense and decreases net income after taxes because of the higher income tax expense. The balance sheet effects of this method are to decrease retained earnings as a result of the lower net income and increase the deferred tax balance. In contrast, the use of partial allocation results in a comparatively lower income tax expense, higher net income after taxes, higher retained earnings, and a lower deferred tax balance. This analysis assumes that the deferred tax account has a credit balance, which is usually the case. Figure 3 contains a summary of these effects. 61 62 Comprehensive Partial Allocation Allocation A. Income Statement Effects Income tax expense Higher Lower Net income (after tax) Lower Higher B. Balance Sheet Effects Retained earnings Lower Higher Deferred tax Higher Lower Figure 3 FINANCIAL STATEMENT EFFECTS OF COMPREHENSIVE VS. PARTIAL ALLOCATION 63 4.2 The Debt Covenant Hypotheses These hypotheses are based on predictions from the bondholder-stockholder conflict mentioned in the previous chapter. Recall that the bondholder-stockholder relationship entails conflict because decisions and actions beneficial to stockholders may be detrimental to bondholders. Specifically, the conflict arises because wealth can be transferred from bondholders to stockholders by paying dividends, diluting the claims of existing bondholders by issuing additional debt of equal or higher priority (claim dilution), pursuing a higher variance project than anticipated by the bondholders (asset substitution), and reducing planned investment (under- investment). With risky debt outstanding, the result of these actions is an overall reduction in firm value, i.e., the bondholders’ loss is greater than the stockholders’ gain. Under the assumption of non~zero contracting and monitoring costs, financial contracts (e.g., bond indentures or lending agreements) can increase firm value by controlling this conflict.1 Control is accomplished by inserting covenants in these 1A competing hypothesis regarding how the bondholder-stockholder conflict is controlled is that external markets (e.g., the market for corporate control) or the possibility of restructuring the firm’s claims provide stockholders with the necessary incentives to maximize the value of the firm (rather than just the value of equity), rendering the choice of financial contracts irrelevant to the value of the firm. However, Smith and Warner [1979] argue that Since restrictive covenants are persistently observed in loan agreements, even though they are costly to contract and monitor, the competing hypothesis should be rejected, i.e., external markets do not eliminate the bondholder-stockholder conflict. Leftwich [1983, p. 26, footnote 5] succinctly summarizes the argument as follows: "If external markets eliminated the potential conflict of interest, there would be no demand for costly bond covenants." 64 contracts restricting the different types of stockholder/management activities aimed at transferring wealth from bondholders to stockholders. Smith and Warner [1979] classify observed bond covenants into four categories:2 1. Production-investment covenants which restrict the firm from investing in other businesses, disposing of its assets, and engaging in mergers. These covenants are considered inefficient because they are costly to monitor; hence they are not frequently observed. Dividend covenants which are established by defining an ‘inventory of payable funds’, also referred to as unrestricted retained earnings (URE), and limiting the distribution of dividends to a maximum proportion (up to 100 percent) of this inventory. Financing policy-related covenants which restrict or prohibit the firm from issuing any additional debt and/or altering the priority of existing debt if it does not meet specified debt-equity and interest coverage ratios. These covenants prevent stockholders from diluting the claim of bondholders on the firm’s assets. 2See also Commentaries on Indentures (Commentaries) [American Bar Foundation, 1971] for a discussion of typical bond covenants. The Commentaries were written by leading experts in the field with the intent to standardize the non- negotiable provisions (‘boiler plates’) used in lending agreements. These boiler plates are highly representative of what is observed in actual practice and are used frequently in lending agreements. 65 4. Bonding covenants which require the firm to provide audited annual financial statements, the specification of accounting techniques, the purchase of liability insurance, and periodically a signed statement from the firm’s officers indicating compliance with all of its obligations under the agreement. These covenants lower the firm’s monitoring costs. Of the various types of bond covenants discussed above, financing—related constraints and/or restrictions on payment of dividends are most frequently observed.’ Evidence to this effect has been found in previous research (e.g., Shevlin [1987]). The constraints imposed by these restrictive covenants are written in terms of accounting numbers, specifically numbers reflecting the firm’s performance and its financial position, such as net income, working capital, current ratios, and tangible net worth. Hence, alternative accounting methods and procedures used to compute these numbers can potentially impact the level of these constraints. The constraint levels are a matter of concern to the firm because if the covenants are violated, the firm is considered to be in ‘technical default’. This entitles the bondholders to either accelerate the maturity of the debt or renegotiate the lending agreement. Alternatively, the firm can change its production, investment or financing activities so as to avoid covenant violations in the first place. Either course of action is costly. 3A potential reason for this occurrence could be that these covenants involve lower monitoring costs than, say, covenants regarding production-investment decisions of firms [Smith and Warner, 1979]. 66 The definitions of accounting numbers in covenants use generally accepted accounting principles (GAAP) as a benchmark [Fogelson, 1978; Leftwich, 1981]. Deviations from GAAP take the form of specific inclusions and exclusions, but generally are found in private debt agreements rather than public debt issues [Leftwich, 1983]. "For public debt issues, other than stating that they should be consistent with GAAP, covenants frequently do not specify how the accounting numbers will be computed" [Smith and Warner, 1979, p. 144]. Hence, if alternative methods to account for a transaction exist within GAAP, and the bond indentures do not specify the use of any one of them (or preclude the use of one or more of them), managers’ choice of accounting techniques can have an impact on the contractual constraints contained in the lending agreements. Based on an application of the agency theory arguments and given the financial statement effects of using comprehensive v. partial allocation methods, it can be argued that firms concerned with violating debt covenants may choose partial allocation to avoid the negative impact of comprehensive allocation on earnings, leverage, and interest coverage. Thus, under costly contracting, it is hypothesized that firms observed using partial allocation are closer to their covenants. Specifically, the formal debt-covenant related hypotheses are (all hypotheses are stated in the alternate form with the corresponding null hypotheses being the complement thereof): Hl : Ceteris paribus, firms that have higher leverage (debt- equity ratios) are more likely to use partial allocation. 67 H2 : Ceteris paribus, firms that have lower interest coverage ratios are more likely to use partial allocation. H3 : Ceteris paribus, firms that have a higher ratio of dividends paid to the inventory of funds available for the payment of dividends are more likely to use partial allocation. In prior research (e.g., Leftwich [1981]; Holthausen [1981]) it has been argued that renegotiation and default costs associated with public debt are usually higher than private debt because public debt is more widely-held.’ Hence, firms with relatively more public debt are more likely to be concerned about covenant violations that may necessitate costly renegotiation and are more likely to adopt accounting procedures that would avoid such violations. This suggests the following hypothesis for this study: H, : Ceteris paribus, firms with more public debt in their capital structure are more likely to use partial allocation. The above hypotheses are based on several assumptions which warrant further discussion. First, it is assumed that breaches of debt covenants are costly. Instead of defaulting on the loan, the firm may renegotiate or repurchase (call) the debt, or change its Operations (production, investment, financing activities). However, these alternatives also are costly. Leftwich [1981, p. 7] hypothesizes that ‘It should be noted that under the Trust Indenture Act of 1939, issues of debt to the public require the appointment of a trustee for the debtholders and an indenture agreement between the firm and trustee. The higher renegotiation costs with widely-held public debt arise because any alterations to the indenture agreement cannot be approved by the trustee without obtaining the concurrence of a majority (usually two-thirds) of the holders of the outstanding bonds. Private debt indentures, on the other hand, can be modified by mutual agreement between the firm and the (usually) small number of lenders. 68 these costs are the lower of renegotiating the agreement, redeeming the debt, defaulting on the loan, or changing the firm’s operations to avoid covenant violations. In the absence of specific knowledge or documentation of the magnitude of these costs, it is assumed for the purpose of this study that they are significant, i.e., it is in the firm’s interest to avoid covenant breaches. Another element of costs involved is information production costs. The thesis of this study is that if flexibility exists within GAAP, such as between partial and comprehensive allocation of taxes for indefinite deferrals, then accounting procedure choice affords the firm yet another possibility of avoiding covenant violations. This implicitly assumes that information production costs (e.g., bookkeeping) are not materially different between recording deferred taxes on a partial or comprehensive basis. Second, it is assumed that the debt covenant variables used in the study adequately proxy for the firm’s closeness to its constraints or the amount of ‘slack’ that exists in the covenants. This assumption is supported by Press and Weintrop [1990] who find measures of proximity to leverage constraints to be significantly correlated with leverage. To the extent this assumption is not met, however, the explanatory power of these hypotheses is reduced. In this context, the leverage hypothesis, which is concerned with the debt-equity ratio, requires Special mention. Equity, the denominator in that ratio, is certainly affected by the choice between partial and comprehensive allocation through the impact on retained earnings. However, it is not clear that debt, the numerator, will always be affected Since deferred taxes may not be considered as debt in the covenants. In examining 69 private lending agreements and the Commentaries [ABF, 1971], Leftwich [1983, p. 34] found that: Some definitions of liabilities include deferred taxes. Almost as frequently, others exclude "reserves for deferred income taxes and other reserves to the extent such reserves do not constitute an obligation." If deferred taxes are not defined as liabilities, differences in debt—equity ratios of firms using partial and comprehensive allocation are reduced which, in turn, reduces the explanatory power of this hypothesis. Third, it is assumed that there is no cross-sectional variation in the constraints proxied by the debt covenant variables. It is possible, however, for such variation to exist, for example, across industries. DeAngelo and Masulis’ [1980] analysis suggests that optimal leverage ratios are likely to differ across industries because of differences in non-debt tax shields (e.g., depreciation). To the extent cross-sectional variation in the constraints exists, measurement error is induced. Fourth, another assumption underlying the debt covenant hypotheses is that loan agreements do not specify which method will be used for the extent of interperiod tax allocation, and/or do not include rules to adjust the reported amounts for the extent of allocation method choice. This assumption is justified, at least for the public debt issues, since they generally rely on GAAP as a benchmark to reduce the costs of monitoring the firm’s covenants [Smith and Warner, 1979]. However, Leftwich [1983] provides evidence that the negotiated set of measurement rules in private lending agreements often differ from GAAP 70 and that the differences are aimed at reducing management’s flexibility to choose accounting rules favoring stockholders over bondholders. Finally, for the sake of simplicity, the debt covenant hypotheses also assume that there is no conflict of interest within bondholder groups when different classes of debt are issued by the firm (e.g., between junior and senior debtholders). Similarly, it is assumed that in the conflict between bondholders and stockholders the manager acts to maximize the stockholders’ wealth, i.e., he aligns his interest with that of the stockholders. This is a simplifying assumption, of course; in reality, assuming separation of ownership and control and that both parties are utility (wealth) maximizers, there is a conflict of interest between managers and stockholders [Jensen and Meckling, 1976].S ’Typically, management compensation contracts are used to minimize the agency costs resulting from the manager-stockholder conflict. Of the different elements observed in managers’ compensation packages, bonus awards often are explicitly based on accounting numbers. Hence, another hypothesis emanating from the contracting cost arguments is that the existence of accounting earnings-based bonus schemes creates incentives for managers to choose income-increasing accounting procedures. Most prior accounting method choice studies have tested this hypothesis by using a dichotomous dummy variable (0/1) for the existence or non-existence of accounting earnings-based bonus plans. The results have been mixed because Healy [1985] found that depending on how earnings are defined in specific plans, certain accounting decisions would not affect the managers’ bonus awards. Hence, it is not surprising that Hagerman and Zmijewski [1979] and Bowen et al. [1981] found no significant association between the existence of accounting earnings-based bonus plans and the accounting choices examined in those studies. Healy’s [1985] finding that 52.7 percent of the total observations in his study defined bonus awards on earnings before taxes has important implications for this study because the choice between comprehensive and partial allocation affects only income after taxes. To determine the impact of this problem in this study, the bonus plan disclosures in the financial statements and 10K’s of the sample firms were examined. Of the 35 firms that specifically mentioned the definition of 71 4.3 Political Costs Hypotheses 4.3.1 The Firm Size Hypothesis As discussed before, evidence suggests that the political process creates incentives for firms to choose between accounting methods. Specifically, firms which are more ‘politically sensitive’ are more likely to adopt accounting procedures that reduce reported earnings and/or defer them to later periods. Watts and Zimmerman [1978] argue that firms’ political sensitivity varies directly with their size. The notion that large firms are associated with monopoly power and concentration of wealth has existed for quite some time. This belief has resulted in large firms becoming the target of public criticism and action leading to added political costs in the form of higher levels of regulation (e.g., the anti-trust laws) and taxation (e.g., the excess profits tax).6 Following Watts and Zimmerman [1978], most prior accounting method choice studies use firm size to proxy for political sensitivity. Larger firms are hypothesized to adopt earnings-decreasing accounting procedures to reduce their visibility and the attendant political costs. Because earnings under comprehensive earnings used in the computation of bonus, 34 firms used a before-tax basis and only one firm an after-tax basis. Hence, the bonus hypothesis was not pursued further. ‘An example of corporate managers’ awareness and concern for these beliefs is borne out in the following remark of CA. Beech, Chairman of the Board, Beech Aircraft Corporation, in his letter to shareholders in the company’s 1975 annual report: "It has become common practice by certain groups to encourage greater regulation and tax liabilities on large corporations which is detrimental to our nation’s economic structure." 72 allocation are smaller than under partial allocation, it is hypothesized that larger firms are more likely to adopt comprehensive allocation. Formally, H, : Ceteris paribus, larger firms are more likely to use comprehensive allocation. Although some studies’ results support the size hypothesis,’ there are conceptual problems as well as empirical evidence that is inconsistent with this hypothesis. Conceptually, it can be argued that because of greater resources at their command, larger firms can lobby for more favorable treatment and thus can prove to be "powerful adversaries in the political process" [Watts and Zimmerman, 1986, p. 239]. The fact that large firms frequently are observed receiving wealth transfers, especially when in financial distress (e.g., the loan guarantee to Chrysler Corporation), provides some credence to that argument. Besides the conceptual problems, there is empirical evidence inconsistent with the size hypothesis. For example, Bowen et al.’s [1981] results for firms outside the oil industry are opposite of what is predicted by the size hypothesis-- larger firms in their sample capitalized interest, an income-increasing accounting procedure. Similarly, El-Gazzar et al. [1986] find that firms’ choice among lease accounting methods is the opposite of that predicted by the size hypothesis--larger firms used the operating method, an income-increasing approach. After partitioning their sample into large- and small-firm sub-samples, Daley and Vigeland [1983] found that the size variable was significant only in their small-firm sub-sample, a ’See Watts and Zimmerman [1986, Table 11.4, pp. 258-259] and Christie [1990, Table 1]. 73 result contrary to the size hypothesis. Finally, Johnson and Ramanan [1988] find no significance for the size variable, though it has the predicted Sign. This evidence suggests that firm size may not adequately proxy for political costs. For these reasons, Holthausen and Leftwich [1983] and Watts and Zimmerman [1986], among others, recommend that future research Should try to develop ‘better’ (more refined) proxies for political costs. 4.3.2 The Effective Tax Rate Hypothesis A direct way by which wealth can be transferred from corporations is via the tax system [Watts and Zimmerman, 1986, p. 235]. Because taxes are one element of the total political costs borne by firms, firms’ effective tax rates (ETRS) provide an alternate proxy for political costs. Assuming taxes are not systematically offset by the nontax components of political costs, such as regulation, quotas, and tariffs, higher ETRS would be indicative of higher political costs. Based on the firm size hypothesis discussed earlier, this would also imply that larger firms would have higher ETRS. Zimmerman [1983] empirically tested this implication for all firms on COMPUSTAT spanning the period 1946-1981 and found some supporting evidence. Although a monotonically increasing relationship between ETRS and firm size was not observed, Zimmerman found that the 50 largest firms in his sample had higher ETRS than the other firms (the ‘threshold effect’). However, this relationship held only for certain industries and over certain time periods.8 8T he overall results appear to have been driven by the oil and gas firms included in his sample. 74 Given the earlier discussion regarding higher political costs providing incentives for firms to adopt accounting procedures that reduce reported earnings and/or defer them to later periods, it can be argued that firms with higher ETRS are more likely to use income-reducing accounting methods. El-Gazzar et al. [1986] tested this hypothesis in the context of accounting for leases. Consistent with the hypothesis, they found that firms with higher ETRS were more likely to capitalize leases--an income-reducing accounting procedure.’ Because comprehensive allocation relative to partial allocation results in lower net income due to a higher income tax expense, the implication of the above discussion for this study is that firms with higher ETRS are more likely to choose comprehensive allocation, i.e., a positive relationship between ETRS and comprehensive allocation would be expected. There is another argument, however, that suggests a negative relationship between ETRS and comprehensive allocation is possible. This argument focuses on the impact the use of comprehensive v. partial allocation has on the computation of firms’ reported ETR on the financial statements and the important role played by ETRS during the 19705 and 19805 in affecting significant wealth transfers from US. corporations, making them nontrivially concerned about the computation of ETRS and the accounting methods affecting the computation. The following events 9Income is lower early in the lease life when leases are capitalized because the interest and depreciation deductions are greater than the rental expense under the operating method. 75 culminating with the Tax Reform Act of 1986 document the role of ETRS in affecting corporations’ tax burdens. Tax reform has been discussed in the US. since the 1930s [McIntyre, 1984]. Calls for tax reform have arisen from the public’s dissatisfaction with the tax system which is perceived as inequitable, inefficient, and replete with special interest loopholes. A factor cited as ‘evidence’ of the inequity and inefficiency of the tax code is the steady decline over three decades in the corporate share of the nation’s tax burden (from 25 percent in the 19503 to just over six percent in 1983 [US Congress, Joint Committee on Taxation, 1984, Table 6]). An important tool used to analyze this decline is corporate ETRS. ETRS have been used as instruments in tax policy debates for quite some time. For example, the minimum tax provisions introduced by the Tax Reform Act of 1969 were based on claims that profitable corporations were able to use tax preferences to lower their ETRS.” The TRA of 1976 made the minimum tax provisions much stiffer, once again based on declining corporate ETRS." Similarly, the relatively lower ETRS of mutual savings banks compared to commercial banks was the basis for the change in the bad debt reserve provisions in the TRA of ”See Senate Report 91-552 on the Tax Reform Act of 1969. "See the Joint Committee’s Explanation of the Tax Reform Act of 1976. 76 1969.12 ETRS also played a major role in the repeal of the statutory depletion allowance for oil companies." The 19805, in particular, witnessed a proliferation of studies on corporate ETRS followed by extensive debates on their proper use and computation. ETR studies have been conducted by Congress (Joint Committee on Taxation), the government (US. Department of Treasury), a labor-funded public interest group (Citizens for Tax Justice), and academics (both economists and accountants), among others. These studies consistently show that there is Significant cross-sectional and inter-temporal variation of ETRS between and within industries, with some firms seemingly paying little or nothing in income tax. These studies also Show that ETRS have declined over time. The collective evidence of these Studies provided much of the political stimulus for the largest corporate tax increase in history--an estimated $120 billion over five years--enacted by the Tax Reform Act of 1986.“ Most of these studies generally compute ETR as the ratio of the taxes currently payable to pretax income. Since 1973, the SEC’S Accounting Series Release No. 149 (ASR 149) [SEC, 1973] has required corporations to disclose in financial statements the ETR indicated by the income statement (hereafter, ‘reported ETR’) and reconcile the difference between their reported ETR and the applicable statutory federal income 12See the House Report 91-413 on the Tax Reform Act of 1969. 1’See the Wall Street Journal, December 3, 1974, p.4, col.2. “See, for example, Birnbaum and Murray [1987], Spooner [1986], and Fullerton [1986] for statements to this effect. 77 tax rate. Under ASR 149, the reported ETR is computed by dividing the income tax expense per books by the pretax income per books, where the income tax expense is the sum of the taxes currently payable and deferred taxes. AS discussed in chapter two, for most companies deferred taxes constitute a significant portion of each year’s income tax expense. By excluding deferred taxes from the numerator, ETRS computed in studies such as the Joint Committee on Taxation or the Citizens for Tax Justice are much lower than the reported ETRS for most companies. For example, Citizens for Tax Justice computed Boeing’s 1983 and 1984 ETR at -9.3 and -3.2 percent, respectively, whereas in its financial statements Boeing’s reported ETRS (which included deferred taxes in their computation) were 25 and 31 percent for 1983 and 1984. Given the publicity corporate ETRS have been receiving, firms were concerned about the exclusion of deferred taxes in their computation. There is some evidence of this concern in firms’ responses to the Citizens for Tax Justice’s study. For example, AT&T issued a statement saying "... this group’s studies are flawed. They misread financial statements and fail to take into account deferred taxes (emphasis supplied)” Union Pacific responded that "... (the) whole Study is a pile of bunk."” Finally, Egger [1985, pp. 956-958], an economist for the 1’The statement appeared in the Daily Tax Reporter (Washington D.C.: Bureau of National Affairs, July 18, 1986), No. 138, p. G-3. 1"Statement of John R. Mendenhall, vice president for taxes at Union Pacific, quoted in Birnbaum and Murray [1987, p. 12]. 78 corporate-backed Institute for Research on the Economics of Taxation, analyzed the Citizens for Tax Justice study as follows: The first problem is with the report’s measure of "effective tax rate." [T]here is nothing amusing about CT J’s ignoring rapidly accumulating deferred tax liabilities and its confusing the corporation’s current payment with its total tax burden. A deferred tax is a postponed tax, not an excused tax. It is a liability, an obligation which must be paid sometime in the future. By dismissing rapidly rising corporate liabilities to pay future taxes but relying on the income concept that gives rise to them, CT J produces a highly misleading, indeed meaningless, measure of corporate tax rates (emphasis in original). Given the preceding discussion of the politicization of ETRS and the resulting wealth transfer from corporations, it can be argued that firms with low ETRS have incentives to minimize public antagonism. One way to do so is by focusing on their reported ETRS and adopting accounting procedures that increase their reported ETR. The choice between comprehensive and partial tax allocation with respect to the indefinitely deferred portion of DISC earnings provided one such opportunity. Because comprehensive allocation relative to partial allocation results in higher income tax expense, thus a higher reported ETR, it can be argued that firms with lower ETRS are more likely to adopt comprehensive allocation, i.e., a negative relationship between ETRS and comprehensive allocation is possible. The above discussion suggests that firms’ accounting method choice is related to their ETR but because of the competing arguments presented above, the direction of this relationship is not specified. Formally, H6 : Ceteris paribus, firms’ choice between comprehensive and partial allocation is related to their effective tax rate. 79 The political cost hypotheses formulated above also warrant further discussion. First, as Holthausen and Leftwich [1983, p. 88] mention, "[t]he causal link between political visibility and accounting numbers is more tentative than any of the other causal links in the economic consequences literature." This is because the linkage is not based on explicit contracts, such as loan agreements and compensation contracts, between the firm and the other parties in the political process (politicians, voters, etc.). Second, as Ball and Foster [1982] point out, there are construct validity problems associated with the use of size as a proxy for political costs. Size has been used to operationalize many seemingly different and competing constructs, such as competitive advantage, information production costs, management ability and advice, and political costs. In view of this, they caution against making inferences if the size variable is found to be statistically significant, unless it can be shown that "an inference from firm size to political costs is a credible one" in the context of this study [Ball and Foster, 1982, p. 191]. Finally, size could be a proxy for omitted variables such as industry membership (e.g., the oil industry in the 19705). Third, the ability of effective tax rates to proxy for political costs has been shown to vary across time and with industry classification [Zimmerman, 1983]. It is possible that one of the two arguments put forth above may dominate over the other in the time period covered by the study. 80 4.4 The Auditor Hypothesis The preceding hypotheses are based on contracting and political cost arguments with a focus on how the economic consequences of those costs on managers, stockholders, and bondholders affect the firm’s choice of accounting methods. However, there are other parties, such as auditors, that also are involved in the firm’s financial reporting decisions and their preferences may affect the firm’s choice of accounting methods. Most prior accounting method choice studies have ignored the role of these other parties. Auditors frequently are observed lobbying on proposed standards. Their preference for certain accounting methods over others may exist for a variety of reasons. Watts and Zimmerman [1986, pp. 312-327] suggest two such reasons. First, assuming wealth maximization, auditors are likely to support standards that increase accounting complexity because that will increase their audit fees.” Second, as indicated previously, because accounting standards can impose costs on firms, the standard setting process provides an opportunity for auditors to provide another service to their clients, namely lobbying. Thus, auditors may lobby for standards that increase their clients’ wealth (and thereby the auditors’ wealth). 1’Simunic [1980] provides indirect empirical support for this proposition by observing a positive correlation between audit fees and accounting complexity. Complexity is measured in terms of the extent of the client’s decentralization (the number of subsidiaries) and diversification (number of 2-digit SIC codes the auditee operates in and the ratio of the auditee’s foreign assets to total assets). It is possible that this positive relationship may not necessarily hold in the 1980’s and later years because the intense competition for audit services has placed greater burden on audit firms to cut costs and reduce their fees. 81 Thornton [1986] advances the sunk cost argument according to which auditing firms support (lobby for) a particular accounting standard because of the expertise developed in that area and their desire to capitalize on the sunk costs incurred in developing the expertise. The preceding discussion only suggests why auditors may lobby for accounting standards. Although presently strong theoretical arguments do not exist for why preferences of auditors may influence their clients’ choice of accounting methods, several factors are likely. It is possible that when dealing with a complex standard, such as that of accounting for income taxes, firms may defer to their auditor’s expertise and advice. This may reduce the cost of implementation of the standard for the firm. It also is possible that where managers are indifferent to the choice of accounting methods, they may go along with the auditor’s stated preference for a particular accounting method [Trombley, 1989]. AS discussed before, the tax allocation issue was extremely controversial. Circumstantial evidence suggests that auditing firms had strong opinions on the extent of allocation they believed was appropriate. Opinion among the big—8 auditing firms appeared to be polarized between Price Waterhouse (PW) on the one hand and Arthur Andersen (AA) on the other. Whereas PW supported partial allocation, AA favored comprehensive allocation. PW’S opinion was based on a study of 100 of its large client-corporations. The study entitled Is Generally Accepted Accounting for Income Taxes Possibly 82 Misleading Investors? was widely quoted in the financial press” and in comment letters by other audit firms and corporations to the APB.” AA’s strong support for comprehensive allocation was well known. This support was expressed in journal articles, letters to the APB, and in the pivotal role played by AA on the APB.” In view of this evidence, the following hypotheses are proposed: H7,: Ceteris paribus, Arthur Andersen’s clients are more likely to follow comprehensive allocation. Hm: Ceteris paribus, Price Waterhouse’s clients are more likely to follow partial allocation. 1"For example, condensed versions of the study were published in The Wall Street Journal (July 21, 1967) and the Financial Executive (September 1967, pp. 70- 75). 1"Unlike the comment letters to the FASB, comment letters on exposure drafts of APB Opinions are not systematically housed anywhere. After making several personal contacts with the then members of the APB and others involved in the accounting for income taxes controversy, the comment letters on APB 11 were traced to the Chicago office of Arthur Andersen. I gratefully acknowledge their allowing me access to those files. ”Based on anecdotal evidence it appears that Mr. George Catlett, then partner of Arthur Andersen, was intensely involved in the accounting for income taxes debate. This was borne out by a selected reading of several thousand pages of letters and inter-office memoranda between AA personnel and Mr. Catlett that were found in the files of Arthur Andersen. Chapter Five EXPERIMENTAL DESIGN AND SAMPLE SELECTION The purpose of this chapter is to describe the empirical procedures employed to test the DISC accounting method choice hypotheses developed in the previous chapter. First, an overview of the experimental design used to conduct the study is provided; second, the sample selection procedures used to identify the DISC firms are described; third, the variables of interest in the study are defined and measurement issues affecting them are discussed; and finally, the Statistical tests used to analyze the data are presented. 5.1 Experimental Design The empirical analyses in this study are performed in a cross-sectional framework by comparing firms using different accounting methods for their indefinitely deferred DISC earnings. The type of experiment conducted here is generally referred to as ‘quaSi-experimental’, and the test design employed is a ‘passive observational method’ (see Cook and Campbell [1979]). In an ideal experiment, subjects (firms) would be randomly assigned to treatment and control groups (i.e., different accounting methods in an accounting method choice study). In this study, as in most empirical accounting research involving accounting method choice studies, firms self-select into either group, rendering random assignment impossible; hence the label ‘quasi-experimental’. Further, because observations are 83 84 taken and variables measured as they occur without any experimental intervention, the design is considered as one involving passive observation. Before the formal model and the statistical tests used in this study are presented, the sample selection procedures followed to identify the sample firms are described. 5.2 Sample Selection and Data Table 5.1 summarizes the sample selection procedure followed in this Study and described below. Firms with DISCS first were identified from the 1972, 1973, and 1974 annual report files of the National Automated Accounting Research System (NAARS) database.‘ Since DISCS first became available for tax years beginning January 1, 1972, the 1972 file was the appropriate starting point. The search was limited to three years so that the study period was contained in one tax regime’ and because of data collection costs.3 After eliminating companies with 1The NAARS database, jointly developed by the AICPA and Mead Data Central Corp, contains the complete financial portion of the annual reports of over 3,600 (3,300 in 1972) publicly traded companies including most Fortune 1,000 companies, those listed on the NYSE and AMEX, and O-T-C companies indicated by the Federal Reserve Board to be on the margin. The keywords used to identify the sample firms were ‘DISC’, ‘D.I.S.C.’, and ‘Domestic International Sales Corporation’. 2See chapter two, section 2.2.3 for an elaboration on this issue. ’NAARS only makes available online only the last five years’ annual report files. Special charges have to be paid for accessing each file not online for each day of use. 85 Table 5.1 SAMPLE SELECTION Firms Less: Less: Final with DISCS identified from NAARS .................. 491 firms Firms not on the Annual COMPUSTAT (Industrial or Research) files .................... (142) Firms for which DISC accounting method not determinablea ................................. ( 29) Sampleb 320 firms “Sample firms are categorized as comprehensive allocators and partial allocators based on the extent to which they provided taxes on their indefinitely deferred DISC earnings (see section 5.2 for details). bSee Appendix B for a list of the firms. 86 irrelevant references to DISCS (e.g., ‘computer discs’, ‘laser discs’, etc.)‘ and companies with DISCS in multiple years (i.e., firms that appeared on more than one NAARS file), 491 companies were identified as having a DISC(S) in at least one of the three years examined. A potential problem with any keyword search is the possibility that the keywords employed may not identify all qualifying firms on the database. This may seem especially likely in this study, given the number of income tax returns filed by DISCS during 1972-74 (see Table 2.1).5 One explanation for the large number of DISC tax returns relative to the sample size in this Study is that several of the sample firms established more than one DISC and each DISC was required to file a separate tax return.‘ Another explanation for the discrepancy is that a large number of tax returns were filed by DISCS, the majority stockholders of which were ‘Although an exact count of such references was not maintained, the number was very small. ’I.R.C., § 6011(c)(2) requires a DISC to file an income tax return even though it is a non-taxable entity. ‘For example, it is stated in Macmillan, Inc.’s fiscal 1973 financial statements that "[t]he Company has several Domestic International Sales Corporations (DISC) for the distribution of certain products to overseas markets" (Note 20; emphasis supplied). Other sample firms making similar disclosures include Great Northern Nekoosa Corp., Outboard Marine Corp., TRE Corp., Universal Corp., and VSI Corp. It is possible that these firms may have adopted a different accounting method (tax allocation policy) for each DISC. However, no statement to that effect was made by any of the sample firms; hence, the use of different accounting methods for each DISC is considered unlikely. 87 either small corporations or noncorporate entities.’ Given the sample selection procedure adopted in this study, it is unlikely that such DISCS would be included. However, in an attempt to assure that the keyword search did not result in the omission of significantly material DISCS, the 1972 annual reports and 10K’s of 30 companies randomly selected from the Fortune 100 corporations, not already included in the sample, were examined.8 None of the companies examined made any disclosure regarding a DISC. Because financial data was required for the study and initial identification of the sample itself was costly, firms not listed on Standard and Poor’s annual COMPUSTAT tapes were deleted to minimize further data collection costs. This requirement resulted in the deletion. of 142 firms. The use of current COMPUSTAT tapes to obtain early 1970’s financial data in conjunction with old NAARS files for identification of the initial sample could result in a failure to obtain a match, either if firms changed names or underwent reorganizations, bankruptcy, etc. To avoid these problems, two procedures were followed. First, ’Of the total number of DISC tax returns Shown in Table 2.1, the proportion filed by such entities was about 65% in 1972, 58% in 1973, and 66% in 1974 [IRS, 1980]. For this calculation, a corporation was considered ‘Small’ if it had assets under $5 million in 1972 or 1973, and under $10 million in 1974. These amounts roughly correspond to the minimum total assets of the sample firms in this study. 8’T he companies were selected from the 500 largest industrial corporations ranked on the basis of sales in 1972 [Fortune, May 1973]. Of the 30 companies, information was available on 24. This procedure was believed to be appropriate because Treasury studies Show that "large US. corporations with DISC subsidiaries were the primary beneficiaries of the DISC provisions" [Hartzok, 1980], which implies that larger firms are more likely to have a DISC than smaller firms. 88 MOODY’S Industrial manuals were consulted for name changes because COMPUSTAT does not systematically maintain such information on its firms.9 Second, the annual COMPUSTAT Research file was used in addition to the other Industrial files so as to include firms deleted from those other files due to acquisition/merger, bankruptcy, liquidation, etc.” Next, the annual reports of the remaining firms were analyzed to determine the extent to which they provided taxes on the indefinitely deferred DISC income. Twenty-nine firms were deleted because their disclosures were inadequate for determining their DISC accounting method. The final sample consisted of 320 firms. These are listed in Appendix B together with their 4-digit SIC code, exchange listing, COMPUSTAT file listing, CUSIP number, auditor, DISC-year (discussed in section 5.4.4), and the DISC accounting method (discussed in the next section). Descriptive data on the distribution of various financial variables for the sample firms is presented in Table 5.2. All figures are based on fiscal 1971 data ”The 1979 and 1988 editions of MOODY’S were examined. The two volumes cover all name changes that took place between 1969 and 1987 for companies listed in previous editions of MOODY’S. This procedure enabled the retention of 45 firms that would otherwise have been dropped. This procedure would not have captured name changes of firms included in COMPUSTAT but not covered by MOODY’S. However, the number of such occurrences should be small because MOODY’S is generally the more exhaustive of the two. 1"The fact that 141 firms in the study (about 44 percent of the total sample) are on the Research file also reduces a potential ‘survivorship bias’ in the sample. However, a problem that could arise is if firms’ DISC accounting method choice is systematically related to their COMPUSTAT listing. AS discussed later, that does not appear to be the case. 89 .Aemv uoofim oocmamn - moxmu couuomoo I who Am I A + av \ va mummma so sesame a aom ASN * mmv sconce mo asam> pasta: u m>z Ace \ Asa u wfivv sunset :0 assume a mom Aosv abuses coeaoo n omzoo Awe + va esooau umz u Hz Am + av some Hence n my Asa + as + mHv mmxmu meoocu muommn esoocH n eHmH Aav name aumu-wcog n can ANHV mmamm umz t un .mauwm Ham new comp Hmoa co comma .ommDCoouoa m we uo mumaaop mo mcoHHHHE CH. Non e~.oH as.o- oao.o oso.o Hom.o amo.o Amo.o Hmo.o Ham.o- «om mom HA.wm Am.n- ooa.o owo.o saw.o NNH.o amo.o omo.o mmm.u- mom mom em.em 4H.A aaa.sm AAA.HH cow.HAs smm.HH mom.m Hmm.o 0mm.mmu- Hz New mm.ow mo.~ emm.ms aam.- coH.NAA mAA.wH qu.m mHB.H HAm.Na- eHmH “cm Am.oe Se.e oaH.HAA soa.omm oom.mmao owm.amm AmH.Am mNH.om coo.o new mom Ho.m~H no.8 eeA.wH mam.m NmH.me~ mae.~ HHs.o ooo.o ooo.o xaa «on Ao.moH mA.w ame.maw mAN.HmN oom.mquH om~.mHN oma.ss «mo.H~ emm.m m>z AOM NN.qm mw.q oqa.eAN qwm.wMH oow.HomN o~o.osH aae.am AwA.NH mHH.N cmzoo AOM oe.mq Hm.m mmw.emm saN.qu ooa.mmsm eoH.moa er.m~ Hmw.a aoH.o as Aom em.mH AA.m ema.maH emw.me use.mwa omo.me emm.HH Hmo.~ ooo.o can Aom oq.Hs mm.m aau.eme Hwo.mam cow.Awwe aam.oqm ~m~.s~ som.mm eHm.q mo.eum cam: Esauxmz Nmn cases: Nmm assucuz_ swanmuum> umoHanum> Homo:MCHm .. moHHmHHHHmHmomma N.m oHan 90 and dollar amounts are in millions. The median sample firm has $74.3 in total assets, $39.4 in book value of equity, $66.1 in market capitalization, and $97.2 in sales. The median firm’s funded (long-term) debt is only $11.8 whereas its total indebtedness is $28.7. The median firm’s income before taxes iS $5.5 whereas its net income is a low $3.2. More than five percent of the sample firms have net losses. Overall, the sample appears to have a few very large firms (e.g., General Electric, Merck, Westinghouse Electric, RCA, Boeing, International Harvestor, and Polaroid) relative to the sample as the mean values of all variables mentioned above (except book value of equity) exceed that of even the firm in the 75th percentile (column 03 in Table 5.2). 5.3 Variable Definitions and Measurement Issues A model to empirically test the DISC accounting method choice hypotheses developed in the previous chapter may be expressed as follows: y = Xe + e (5.1) where y is a (n x 1) column vector of dependent variable observations (the DISC accounting method choice), X is a (n x k) matrix of independent variable observations, )6 is a (k x 1) column vector of unknown parameters, and e is a (n x 1) column vector of errors.“ The definition, measurement issues, and data 11The sources of error are 1) specification error arising from the omission of relevant variables in the model and 2) measurement error in the variables. 91 problems associated with the dependent variable are discussed first followed by the independent variables. 5.3.1 Classification of DISC Accounting Method -- The Dependent Variable AS discussed before, the DISC itself was a non-taxable entity, but its export- related profits were taxed to its stockholders when distributed or deemed distributed. DISC stockholders were typically other US. corporations (i.e., DISCS were formed as subsidiaries of these corporations, hereafter referred to as ‘DISC firms’) [IRS, 1986], that were statutorily permitted to indefinitely defer taxes on approximately 50 percent of the DISC’S export profits. For financial reporting purposes, three possibilities existed with respect to the indefinitely deferred DISC income:" 1) the DISC firm provided for deferred taxes on all of the indefinitely deferred DISC earnings, or 2) the DISC firm did not provide for deferred taxes on any of the indefinitely deferred DISC earnings, or 3) the DISC firm provided deferred taxes on a portion (less than 100 percent) of the indefinitely deferred DISC earnings. "Recall that these choices arise because the indefinite reversal criteria of APB 23 applies to the DISC transaction, thereby excluding it from the mandate of comprehensive allocation under APB 11 (see chapter two, section 2.1). 92 In general, firms followed either method 1) and are labeled comprehensive allocators or method 2) and are labeled partial allocators.l3 The primary source of information for classifying the sample firms into comprehensive and partial allocators was the disclosures made in the tax footnotes to their financial statements. Tax footnote disclosures are mainly governed by Accounting Series Release No. 149 (ASR 149) [SEC, 1973] that requires SEC registrants to disclose in their financial statements 1) components of the income tax expense, 2) reasons for timing differences between book and tax reporting resulting in deferred taxes, and 3) a reconciliation between the effective income tax rate (ETR) indicated by the income statement and the statutory Federal income tax rate [SEC, 1973].“ Under the ASR 149 requirements, the expected disclosure by comprehensive and partial allocators was as follows: "T his conjecture is consistent with Marocco [1985] and disclosures made by the sample firms in this study. Although none of the sample firms specifically indicated that they followed method 3) above, that possibility exists since all firms did not clearly state that they were following method 1) or 2). The label ‘partial allocation’ was preferred to ‘flow through’ simply because under APB 11 these firms had to provide deferred taxes on most other timing differences, i.e., they provided deferred taxes on some but not all timing differences which is the partial allocation approach, not flow-through. 1‘Note that ASR 149 was applicable to financial statements for periods ending on or after December 28, 1973. Hence, the detailed disclosures required by ASR 149 were not available for many of the sample firms for 1972, creating several problems that are discussed later. 93 Nature of Type of Firm Book-tax Difference Type of Disclosure COMPREHENSIVE Timing Source of ALLOCATOR Difference Deferred Tax PARTIAL Permanent Adjustment in ETR ALLOCATOR Difference Reconciliation Actual disclosures by the sample firms, however, varied considerably, ranging from mere statements in the footnotes (e.g., GCA Corp. and Great Northern Nekoosa Corp.) to detailed dollar effects of the DISC accounting choice (e.g., Gardner- Denver Co., Banner Industries, Inc., and Reed Tool Co.). For illustrative purposes, examples of some types of DISC disclosures (specifically, of companies mentioned as examples above and later in the text) are provided in Appendix C. One type of disclosure observed frequently deserves special mention because of its apparent contradiction with the expected disclosures. Several sample firms ‘ presented the DISC tax effect both as a timing difference (implying comprehensive allocation) and as a permanent difference (implying partial allocation). This apparent contradiction, however, disappears if the DISC had a different fiscal year than its parent corporation and the parent was a partial allocator.” In that case, ”The Treasury estimated that "approximately 60 percent of all DISCS owned by US. corporations have accounting periods lagging slightly behind their parent’s accounting period" [Hartzok, 1980]. Because the timing of the taxability of DISC income to the parent corporation depended on the DISC accounting period ending with or within the accounting period of the parent, the effect of lagging the DISC’S accounting period was to postpone by as much as one year the inclusion of the DISC income, such as dividends, in the parent’s gross income, thereby delaying the payment of taxes on that income. 94 deferred taxes would be provided on the currently taxable DISC earnings (the deemed distributions) and the parent’s ETR would be reduced by the indefinitely deferred DISC earnings. Some sample firms explicitly stated that their DISC had a different fiscal year (e.g., Ionics, Inc.), whereas a different fiscal year for the DISC was inferred for the others (e.g., Reed Tool Co.). All of these firms were classified as partial allocators. The DISC accounting method classification procedures discussed above resulted in 82 firms (about 26 percent of the sample) being classified as comprehensive allocators and 238 firms (about 74 percent of the sample) as partial allocators. 5.3.2 The Independent Variables The variables hypothesized to explain firms’ DISC accounting choice are the contracting cost variables (leverage, interest coverage, and dividend payout constraints, and extent of public debt), the political cost variables (effective tax rate, Size, and capital intensity), and firms’ auditor. Data to measure these variables were generally obtained from COMPUSTAT, annual reports, 10K’s, and/or MOODY’S. In this section the definition of these variables and the manner in which various measurement issues were resolved are discussed. Unless otherwise stated, all computations are based on fiscal 1971 data. Leverage Constraint (LEV). The LEV variable iS used as a surrogate for proximity to debt covenant restrictions (the Slack that exists in the firms’ leverage constraint) and/or the probability of default on debt agreements. Since the 95 constraints often are Specified in terms of net tangible assets or tangible net worth, two measures of LEV are used in this study and are computed as follows: ( LTD / COMEQ LEV = I [ TD / NTA where: LTD = long-term debt COMEQ = book value of equity TD = total debt NT A = net tangible assets. In addition to these two summary leverage measures, sensitivity of leverage to other definitions is examined. These other definitions and the results of the sensitivity analysis is discussed below. Ideally, capturing the proximity to debt covenant restrictions would require computing measures using the actual limitations specified in the debt contracts. However, in the absence of easy accessibility to such information, it is assumed that the higher the LEV variable, the closer the firm is to violating its covenants.“ Because the use of partial relative to comprehensive allocation results in higher net income thereby increasing the denominator of the LEV variable, firms concerned with violating their covenants are more likely to choose partial allocation (i.e., a positive relationship between LEV and partial allocation is hypothesized). ”Support for this assumption is found in Press and Weintrop [1990] who found other measures of proximity to leverage constraints based on an actual examination of debt contracts to be Significantly correlated with leverage. However, if LEV measures proximity to debt covenant constraints or default probability with error, parameter estimates are biased and inconsistent with the degree of bias and inconsistency related to the variance of the measurement error (the "errors-in- variables problem," see Pindyck and Rubinfeld [1981, p. 177]). 96 LEV has been defined in several different ways in prior accounting choice Studies (see Table 5.3). Although a ratio of debt to net tangible assets or net worth is always used, differences arise in the specification of the numerator and thedenominator of the ratio. Some studies use total debt in the numerator (e.g., Zmijewski and Hagerman [1981]), whereas others use only long-term debt (e.g., Dhaliwal [1980]). Following Smith and Warner [1979], some studies adjust the numerator for capitalized lease obligations (e.g., Bowen et al. [1981]) and others adjust the numerator for preferred stock on the premise that it is a form of ‘junior’ debt (e.g., Johnson and Ramanan [1988]). LEVl to LEV6 are based on previous studies and LEV7 is computed after adjusting for both capitalized lease obligations and preferred stock (see Panel A of Table 5.3). Another conceivable adjustment in the specification of LEV, although not made in any previous accounting method choice study, could be for deferred taxes. Foster [1986, pp. 75-80], among others, notes the considerable confusion that prevails over the nature of deferred taxes which have been variously viewed as debt, equity, and neither debt nor equity. Adjusting LEV for deferred taxes is of particular interest in this study because the DISC accounting method choice directly impacts a firm’s deferred tax balance. Following Leftwich’s [1983] finding that some debt contracts include deferred taxes in debt, LEV8 to LEVll also were computed for the sample firms (see Panel A of Table 5.3). Variations in the denominator of LEV are less frequent with the central issue being whether to use a book value-based measure (net tangible assets or 97 Table 5.3 LEVERAGE MEASURES Panel A: Book Value Measures Variable Definitiona Study Using the Definition LEVl LTD / BVEl Lilien and Pastena [1982]b LEVlA LTD / BVE2 Dhaliwal [1980] LEV2 (LTD+CLO) / BVEl El-Gazzar, et al. [1986]b LEV2A (LTD+CLO) / BVE2 LEV3 (LTD—CLO) /NTA Bowen, et a1. [1981]c LEV4 TD / NTA Zmijewski and Hagerman [1981] LEVS (LTD+PREFEQ) / NTA Johnson and Ramanan [1988] LEV6 (LTD-PUB-CLO) / NTA Daley and Vigeland [1983] LEV7 (LTD+PREFEQ—CLO) / NTA LEV8 (LTD+DTX) / NTA LEV9 (LTD+DTX—CLO) / NTA LEVIO (LTD+DTX+PREFEQ) / NTA LEVll (LTD+DTX+PREFEQ—CLO) / NTA Panel B: Market Value Measures Variable Definitiona Study Using the Definition LEVMl LTD / MVE Lys [1984]d LEVMlA LTD / (MVE+LTD) Chow [1982] LEVM2 (LTD+CLO) / MVE LEVM3 (LTD—CLO) / MVE LEVM4 TD / MVE LEVMS (LTD+PREFEQ) / MVE Collins, et a1. [1981] LEVM6 (LTD—PUB-CLO) / MVE LEVM7 (LTD+PREFEQ—CLO) / MVE LEVM8 (LTD+DTX) / MVE LEVM9 (LTD+DTX-CLO) / MVE LEVMlO (LTD+DTX+PREFEQ) / MVE LEVMll (LTD+DTX+PREFEQ-CLO) / MVE aVariables with COMPUSTAT item numbers in parentheses are: LTD = Long-term debt (9) BVEl a Common equity (60) BVE2 = BVEl + Pref stock (60+130) NTA - Net tangible assets (6—33) TD - Total debt (9+5) PREFEQ = Preferred stock (130) DTX - Deferred taxes - B/S (74) PUB - Public debt CLO = Capitalized lease obligations (84) MVE - Market value of equity (25*24) bIt is not clear whether they used common equity or (common equity + preferred stock) as their definition of ‘book value of equity'. cThey used the reciprocal of LEV3 in their study. For consistency with the other leverage measures, debt is included in the numerator. dIt is not clear whether he used total debt or long-term debt in the numerator. 98 tangible net worth per books) or the firm’s market capitalization. As Table 5.3 shows, market value of equity is used in the denominator primarily in stock market studies of mandated accounting procedures. This specification is based on modern finance theory that generally regards market values as more pertinent than book values. The various market value-based leverage measures (labeled LEVM) used in prior Studies and others constructed to mirror the book value-based measures are listed in Panel B of Table 5.3. No previous study documents the sensitivity of their results to alternative definitions of leverage." Descriptive statistics for the various LEV measures of the sample firms are presented in Table 5.4. LEVl to LEV2A are virtually alike with median (mean) values ranging between 0.31 to 0.32 (0.37 to 0.45). Note that the maximum values of these leverage measures can exceed 1.0 because they use BVE in the denominator. That is indeed the case for some sample firms (maximum values > 2.8) and results in the higher mean values as compared to measures that use NTA in the denominator. LEV3 to LEV11 (except LEV4) all have fairly similar magnitudes and range. Their median (mean) values lie between 0.11 (0.13) for LEV6 to 0.19 (0.20) for LEVlO. Note that LEV3 to LEV11 are comparable with each other since they all use net tangible assets as the denominator. Similarly the pairs LEVl and LEV2 and LEV1A and LEV2A are comparable. LEV4 is different from all of the other LEV measures in that it considers the total "Press and Weintrop [1990] is an exception but they also do not consider all the definitions of leverage included here, primarily the measures that include deferred taxes. 99 Table 5.4 DESCRIPTIVE STATISTICS -- LEVERAGE MEASURESa Panel A: Book Value Measures Variable” N Mean Std Dev Median Minimum Maximum LEVI 307 0.401 0.395 0.321 0.000 2.828 LEV1A 307 0.386 0.373 0.316 0.000 2.828 LEV2 296 0.424 0.455 0.319 0.000 3.756 LEV2A 296 0.407 0.418 0.312 0.000 3.027 LEV3 285 0.168 0.122 0.164 0.000 0.584 LEV4 296 0.432 0.169 0.432 0.010 0.955 LEV5 296 0.188 0.132 0.184 0.000 0.584 LEV6 272 0.128 0.105 0.107 0.000 0.481 LEV7 285 0.177 0.129 0.173 0.000 0.584 LEV8 295 0.194 0.132 0.189 0.000 0.609 LEV9 285 0.183 0.128 0.177 0.000 0 609 LEVlO 295 0.203 0.138 0.196 0.000 0.611 LEV11 285 0.193 0.134 0.187 0 000 0 611 Panel B: Market Value Measures (maximum not reset) Variable” N Mean Std Dev Median Minimum Maximum LEVMl 304 0.365 0.663 0.179 0.000 7.302 LEVMlA 304 0.200 0.178 0.152 0.000 0.880 LEVM2 293 0.381 0.687 0.178 0.000 7.302 LEVM3 293 0.346 0.667 0.169 0.000 7.302 LEVM4 304 0.824 1.180 0.501 0.017 11.025 LEVM5 304 0.391 0.736 0.179 0.000 8.679 LEVM6 282 0.269 0.520 0.135 0.000 6.388 LEVM7 293 0.372 0.741 0.169 0.000 8.679 LEVM8 303 0.392 0.702 0.193 0.000 8.029 LEVM9 293 0.374 0.705 0.180 0.000 8.029 LEVMlO 303 0.418 0.776 0.194 0.000 9.406 LEVM11 293 0.401 0.781 0.188 0.000 9.406 8Based on 1971 data for all firms. ”See Table 5.3 for the variable definitions. 100 indebtedness of the firm (current + long-term). Its distribution for the sample firms is remarkably well-behaved (median and mean = 0.43). Pearson product-moment correlations among all the book value-based measures were also computed and are presented in Panel A of Table 5.5. The correlations generally exceed 0.8 (except for LEV4 and LEV6) and they are all statistically Significant at less than the 0.001 level.” From the above analysis it appears that for this sample the alternative specifications of book value-based leverage measures (LEV) are essentially similar to each other and insensitive to whether capitalized lease obligations, preferred stock, and/or deferred taxes are considered debt. Hence, for subsequent analysis, only two book value-based measures, LEVl and LEV4, are retained. These measures were chosen because of their 1) simplicity, 2) ability to capture two quite different Specifications of leverage, namely the use of long-term debt v. total indebtedness in the numerator and the use of net worth (the debt-equity concept) v. net tangible assets in the denominator, and 3) relatively lower correlations with each other. Analysis of the market value-based leverage measures reveals generally higher mean values than the book value-based measures but the median values of the two groups are generally comparable (except LEV1, LEV1A, and LEV2). Note ”Spearman rank correlations were also computed and are of Similar magnitudes and highly significant statistically. In all subsequent correlation analysis as well, both Pearson and Spearman correlations were computed. However, only the Pearson correlation results will be presented unless the Spearman correlations are qualitatively different. :10]. new new new new new www new new new new new new new eee.w ewe.e eee.e wee.e eee.e eww.e eee.e wee.e ewe.e wen.e eww.e ewe.e nwe.e Hw>me new new new new www new new new new new new new eee.w ene.e nee.e wee.e enw.e wee.e ewe.e wee.e ewe.e ewe.e ene.e eee.e ew>ew new new new www new new new new new new new eee.w ewe.e awe.e wee.e wee.e nwe.e eee.e enw.e eww.e eee.e ewe.e e>me new new www new new new new new new new eee.w nee.e Hew.e ewe.e wee.e ene.e nee.e eee.e one.e ene.e e>mw new www new new new new new new new eee.w sew.e ewe.e eee.e wee.e Hnw.e sew.e eee.e nee.e n>mw www wAw www www www www www www eee.w enw.e een.e nwe.e eee.e eee.e eww.e eee.e e>mw new new new new new new eew eee.w wew.e wne.e eee.e ewe.e wee.e nwe.e n>mw eew new new new new new eeo.w eee.e eew.e eww.e new.e enw.e e>mw new new new new new eee.w oew.e eww.e ene.e wee.e e>ww new new eew eew eee.w wee.e ewe.e ewe.e ww eew eew new eee.w wee.e eee.e w>me wow wow eee.w eee.e ww wee uSew w>ww ww>ew ew>mw e>mw e>ee w>ew e>ew n>ww e>ww e>mw ww w>mw we w>ew newnenee> mounmooz oaam> xoom n< Hosea emm4m momq -- mHzmHlommoo ZOHHoa H000. on» an ammo some an voaoonou mw manowum> CESHoo as» use manmwuo> sou on» cmozuon sawueaoquo ones 00 mwmonuom5£ .mquwowuuooo scapuamquo on» one onwa unufiu on» an muonESZU .mcofiue>uomno uo amass: on» one ocwa vacuom as» aw .ezowewcweee ownewue> one new e.n enema eeme .eseww wee new eeee Hwew co eeeeee 00N 00N 00N 000 00N NON 00m 00m 00m 00m 00N 00N 000.a 000.0 000. 000. 000. 500.0 000.0 «00.0 000.0 000.0 0c5.0 000.0 Haz>mq 000 00m 000 00m NON 000 000 00N 00m 000 000 000.0 000. :00. N00. H00.0 000.0 000.0 N00.0 «00.0 «05.0 000.0 0HZ>mq 00N 00N 00N NON 00N 00N 00N 00N 00N 00m 000. 000. 000. 0N0.0 500.0 000.0 500.0 «50.0 H05.0 N00.0 0z>mq 000 00N NON 000 000 00m 00M 000 000 000. 500. 0H0.0 000.0 500.0 000.0 H00.0 N05.0 500.0 0Z>mq 00N NON 00m 00m 00m 00m 00m 00m 000. 000.0 «00.0 000.0 000.0 000.0 «Q5.0 500.0 52>mq NON NON NON NON NON NON NON 000.H 000.0 500.0 000.0 000.0 5H5.0 ¢N0.0 02>mq «00 «00 00N 00N e00 400 000.a 000.0 000.0 500.0 005.0 000.0 0Z>ma «00 00m 00m «00 «00 000.H 000.0 NH0.0 505.0 4H0.0 ¢Z>mq 00N 00N 00N 00N 000.H 050.0 005.0 000.0 02>mq 00m 00m 00m 000.H 005.0 000.0 Nz>mq :00 :00 000.H N05.0

MA :00 9000.H H2>mq aHZ>mq 0az>mq 0Z>mq 0Z>04 5Z>mq 0Z>mq 02>mq ¢£>ma 0£>mq Nz>mq CHZ>mA H2>mq noanuauo> Auomou no: moaae> aoawxoev mouammoz oaao> uoxumz A.U.ucoov 0.0 mafimw m .mflncwm 103 that because all the market value-based leverage measures (LEVM) were defined using a debt-equity concept, their values need not lie between 0 and 1. The LEVM measures (except LEVMlA) for some sample firms indeed exceed 1.0 causing them to be positively skewed. Pearson product-moment correlations among these measures generally exceed 0.9 (except for LEVMlA) and are statistically significant at less than the 0.001 level (see Panel B of Table 5.5).” As with the book value- based measures, it appears that for this sample that the various market value- based leverage measures are essentially Similar. Although, for comparative purposes, the subsequent analysis also was performed using LEVMlA, those results are not emphasized because the use of market value-based leverage measures in a voluntary accounting method choice study is generally questionable, given that debt targets are set in terms of book values rather than market values (e.g., Smith and Warner [1979]; Commentaries [American Bar Foundation, 1971]),3" 1"LEVM measures were also computed by resetting values greater than five to five. This reduced the skewness and the resulting means were lower but there was no change in the correlations. ”Myers [1977] explains theoretically why the use of book rather than market values in setting target debt ratios "makes sense." Following Miller and Modigliani, the market value of a firm can be viewed as consisting of 1) the present value of the assets in place and 2) the present value of future growth opportunities. Because a significant part of many firms’ market values are accounted for by the present value of future growth opportunities, Myers argues that a greater proportion of debt will be supported by assets in place. Hence the use of book values in lending agreements. 104 Interest Coverage (INTCOV). AS discussed before, some debt contracts prohibit firms from issuing additional debt unless a minimum prescribed ratio between income and interest charges is maintained. The interest coverage ratio (INTCOV) is used as a proxy for this ‘earnings test’ and is computed here as follows:” INTCOV = (IBEI + INT) / INT where: IBEI = income before extraordinary items and discontinued operations but after taxes and minority interest INT = interest expense. It is assumed that the lower the INTCOV, the more likely the firm is to violate its interest coverage constraint. Because the use of partial relative to comprehensive allocation results in higher income after taxes, firms concerned with violating their interest constraints are more likely to adopt partial allocation. That is, a negative relationship between INTCOV and partial allocation method choice is hypothesized. Firms with negative INTCOV (primarily because of negative IBEI) were reset to zero. Three firms with INTCOV greater than 1.00 were reset to 100. In addition, INTCOV of firms with zero INT (10 firms) was reset to 100 provided IBEI was positive (nine of the above 10 firms).22 The recoding procedures adopted 21The definition of INTCOV is Similar to that employed in prior research (e.g., Bowen et al. [1981]) and is the after-tax definition of interest coverage in COMPUSTAT. 22These procedures are Similar to those employed in prior accounting research (e.g., Daley and Vigeland [1983]; Shevlin [1987]; Wong [1988]). 105 here substantially reduced the skewness of the distribution (from 13.71 to 3.45), although it still remains positively skewed (median = 3.58; mean = 10.89). Sensitivity of the results to the recoding procedures was examined and is reported later. Dividend Constraint (DCOV). AS discussed before, restrictions on dividend payments are placed on firms in terms of an inventory of payable funds. Unlike the use of LEV as a proxy variable for closeness to debt covenant restrictions, 3 more direct test of the nearness to possible dividend covenant violations was motivated by Smith and Warner [1979, pp. 131-136] and is computed here as follows: DCOV = CDIV / URE where: CDIV common dividends URE = unrestricted retained earnings. URE represents the maximum amount of dividends that a company can pay its shareholders without violating its most restrictive (debt) covenants. As Bowen et al. [1981, p. 165] point out, "[i]ntuitively, the inverse of the dividend constraint measure (DCOV) is the number of years that the current level of dividends can be paid out of the current allowable dividend pool." In other words, the higher the DCOV ratio, the more likely the firm is to violate the dividend constraint. Because the use of partial relative to comprehensive allocation results in higher net income, thereby increasing the denominator of DCOV, a positive relationship between DCOV and partial allocation method choice is hypothesized. However, to increase 106 the power of the test, data on the existence of dividend constraints was also collected and appropriate coding procedures for DCOV was used as discussed below. Data for URE was hand-collected from MOODY’S and verified from annual reports and 10K’s where possible.” For the 15 firms with zero URE, DCOV was coded three, a value greater than the maximum DCOV value of any of the sample firms with positive URE (their DCOV was 2.89). For the 77 firms with no dividend restrictions mentioned, DCOV was coded zero.” If these firms are not included, the sample size drops to only 166 firms. For the 52 firms with a dividend restriction mentioned but a measure of URE not available, two procedures were followed: 1) the firms were omitted from the tests (DCOVl), and 2) the firms were included with their dividend constraint measure computed by substituting retained earnings (RE) for URE if RE was positive or coding it three if RE was zero or negative (DCOV2).” The median value of DCOVl (DCOV2) is a low 0.06 (0.05) whereas the mean is 0.39 (0.33). Since the distributions of DCOVl and DCOV2 are virtually similar, primarily DCOV2 is used in the subsequent tests as it results in a higher sample size. ”The earliest year for which URE data is available on COMPUSTAT is 1974. 2"The assumption here is that either there is no restriction or if one exists it is immaterial. In either case coding DCOV zero is appropriate. ”The recoding procedure used here is similar to prior research (e.g., Bowen et al. [1981]). 107 Public Debt (PLEV). As discussed before, because of differing renegotiation costs, a separate leverage constraint variable was computed for the extent of public debt in the firm’s capital structure as follows: PLEV = PUB / NTA where: PUB = public debt NTA = net tangible assets. Data for PUB was hand-collected from MOODY’S and verified from annual reports and 10K’s where possible. Over 68 percent of the sample firms had either zero or immaterial amounts of PUB. This motivated the use of a dummy variable (PLEVDUM) which was coded one if the firms had positive PUB and zero otherwise. Both PLEV and PLEVDUM were used in the subsequent analysis but the results were insensitive to the definition used. As with the LEV variable, a positive relationship between PLEV and partial allocation method choice is hypothesized because the larger the PLEV ratio, the higher are the renegotiation costs a firm is likely to incur if it is in violation of its covenants. Firm Size (SIZE). In most accounting research typically net sales or total assets are used as measures of firm size. However, Watts and Zimmerman [1986, p. 239] argue that "a better proxy for the negative/positive wealth transfers is a firm’s accounting earnings instead of firm size (total sales or assets)" In this Study, total assets, sales, net income as well as market and book value of equity were used as measures of firm Size. 108 Descriptive data for these SIZE measures reveals that they are all highly skewed (see Table 5.2). Logarithmic transformations were made resulting in substantially reduced skewness levels. AS expected, all of these SIZE measures are very similar, with the Pearson product-moment correlations exceeding 0.7 (0.8 when transformed, see Table 5.6). Hence, in the subsequent analysis only LSAL is used to proxy firm size. For sensitivity purposes, results using LTA and LNI were also obtained and are discussed later. Effective Tax Rate (ETR). As discussed before, choice of the DISC accounting method affects the firm’s reported ETR in its financial statements which is generally computed as follows: ETR TOTTX / IBIT where: TOTTX total income tax expense per books IBIT = net operating income before income taxes, minority interest, and extraordinary items (i.e., IBEI + MI + TOTTX).’6 Henceforth, this measure of effective tax rate is referred to as ETRl. In addition, two other ETR measures also were computed as follows: ETR2 = [TOTTX - (DTX,—DTX,_,)] / IBIT ETR3 = [TOTTX — (DTX,—DTX._I)] / [IBIT — (DTX,—DTX,_,)+0.48] where: DTXl = balance sheet deferred tax amount in year t. ETR2 corresponds to the effective tax rate measure used in studies such as the CT J and JCT. This measure removes the current portion of deferred taxes from 2“This is also the COMPUSTAT definition of pre-tax income. 109 Table 5.6 PEARSON PRODUCT-MOMENT CORRELATION COEFFICIENTS -- SIZE VARIABLESa Panel A: Raw Measures ygriable” TA COMEQ. MVE SAL NI TA 1.000c 307 COMEQ 0.970 1.000 307 307 MVE 0.814 0.841 1.000 304 304 304 SAL 0.966 0.933 0.815 1.000 307 307 304 307 N1 0.716 0.777 0.846 0.724 1.000 306 306 303 306 306 Panel B: Logarithmic Measures Variable” LTA LCOMEQ LMVE LSAL LNI LTA 1.000C 307 LCOMEQ 0.977 1.000 307 307 LMVE 0.848 0.892 1.000 304 304 304 LSAL 0.970 0.944 0.816 1.000 306 306 303 306 LNI 0.883 0.905 0.916 0.877 1.000 277 277 274 277 277 8Based on 1971 data for all firms. ”See Table 5.2 for the variable definitions. cNumbers in the first line are the correlation coefficients and in the second line are the number of observations. The null hypothesis of zero correlation between the row variable and the column variable is rejected in each case at the .0001 level. 110 the tax expense amount, thus leaving only taxes currently payable in the numerator. ETR3 is an attempt to derive the firm’s effective tax rate using its taxable income. However, given that we do not have access to individual firms’ tax returns, taxableincome is approximated from financial statements by making an adjustment for the change in the grossed up amount of deferred taxes in the denominator.” It should be pointed out that ETR2 and ETR3 are not affected by the DISC accounting method choice. Because IBIT is not available on the 135-item COMPUSTAT tapes, it was estimated as above. In addition to the 13 firms with missing IBEI data, 25 firms had missing data for M1. For these firms, MI was coded zero.” However, all three distributions are highly skeweduETRl and ETR2 are positively skewed, whereas ETR3 is negatively skewed. The various adjustments made to ‘improve’ the distributions and the sensitivity of the results to those adjustments is reported later. 2The following example is presented to clarify the rationale for the adjustments. Suppose for a given year a firm’s book income before taxes is $100, its taxable income is $60, and the maximum statutory tax rate is 48 percent. Then $48 ($100 x 0.48) is the income tax expense per books, whereas $29 ($60 x 0.48) is the tax currently payable per the tax return. The difference between the two, $19 ($48 - $29), is the amount of taxes deferred. Given the firm’s book income and tax expense, its taxable income may be approximated thus: [100 — (19+0.48)] = $60. Note that this is a highly simplified example. In reality, the firm’s actual marginal tax rate must be used. In addition, there are several other problems that arise when computing taxable income from financial Statements (e.g., Wheeler and Outslay [1986]). ”Excluding these firms made no difference in the distributions of the three ETR measures. This was not unexpected because over 95 percent of the sample firms have less than $0.5 million in minority interest, with over 75 percent having zero MI. 111 Auditor (AUD). AS discussed before, firms’ auditors are hypothesized to affect the DISC accounting choice. Because data on the AUD variable is qualitative, dummy variables are employed and were coded as follows: { 1 if auditor is AA AUDDUMl = I [ 0 otherwise f 1 if auditor is PW AUDDUM2 = I [ 0 otherwise where: AA = Arthur Andersen PW Price Waterhouse. Two dummy variables were used instead of one categorical variable because the predicted effect on the DISC accounting method choice is in opposite directions for these variables--AUDDUM1 is expected to be negatively correlated with partial allocation whereas AUDDUM2 is expected to have a positive correlation. AUDDUMl has a mean of 0.216 implying that 21.6 percent of the sample firms had Arthur Andersen as their auditor. Similarly, 9.7 percent of the sample firms were audited by Price Waterhouse. Data on the AUD variable was obtained from NAARS and MOODY’S and verified with their annual reports and 10K’s where possible. For descriptive purposes, a breakdown of the sample firms’ auditors by the DISC accounting method choice is provided in Table 5.7. Almost 90 percent of the sample firms had a ‘big-8’ auditor. The remaining firms (about 11 percent) were grouped together under ‘other’. 112 Table 5.7 AUDITOR ANALYSISa DISC Acct. Methodc Percent Auditor” C P Total Total AA 24 45 69 21.56 (34.78)d (65.22) AY 7 27 34 10.63 (20.59) (79.41) CL 8 22 30 9.38 (26.67) (73.33) DH 8 21 29 9.06 (27.59) (72.41) EB 13 32 45 14.06 (28 89) (71.11) PM 12 25 37 11.56 (32.43) (67.57) PW 2 29 31 9.69 (6.45) (93.55) TR 2 8 10 3.13 (20.00) (80.00) Other 6 29 35 10.94 (17.14) (82.86) Total 82 238 320 100.00 (25.63) (74.38) aBased on 1971 data for all firms. ”where: AA=Arthur Andersen & Co AY=Arthur Young CL=Coopers & Lybrand (also Lybrand, Ross Bros. & Montgomery) DH=Deloitte Haskins & Sells (also Haskins & Sells) EE=Ernst & Ernst PM=Peat, Marwick, Mitchell & Co PW=Price Waterhouse & Co TR=Touche Ross. ”C=comprehensive allocators, P=partia1 allocators. cRow percentages in parentheses. 113 Definitions of the independent variables and descriptive statistics on their distributions are summarized in Table 5.8. 5.4 Within Sample Profile Analysis Preliminary analysis was conducted to determine if the sample selection procedure resulted in different subsets of firms. Accordingly, the sample firms were classified by their 1) industry membership (SIC code), 2) stock exchange listing, 3) COMPUSTAT file listing, and 4) the year when the DISC was operational. The association between firms’ DISC accounting method choice and these classification criteria was analyzed by means of the Pearson chi-square statistic that tests the null hypothesis of independence between the row and column variables in a contingency table. For large n, this statistic is distributed approximately as X’- 5.4.1 Industry Membership Generally, the use of 4-digit SIC codes is considered desirable to analyze industry effects. However, the sample firms fell in 153 different 4-digit SIC classifications, making it virtually impossible to draw meaningful inferences.” However, for descriptive purposes, the sample firms’ 4-digit SIC class and their DISC accounting method are presented in Appendix D. ”Similarly, analysis by 2-digit SIC codes was not performed because the sample firms fell in 37 different 2-digit SIC categories. 114 0 u 000 05 0 00:05uCoE :05u05uumou 0co05>50 0c 05 0 00505ucmdv Doc 00: Dan 00:05ucoe co5DO5uumou 0co05>50 05 mm \ >500 u 5>000 we: \ >Hee 005A >00925 05 005 0 v >00925 55 0 n >00525 ezH \ 5525 + wawv <52 \ we I e>wu oezoo \ an; n 5>eq ”one mum5uaouaam ouoLB momoLDCouma c5 muonadc 8005 He .mEH50 550 you 0000 5505 so commme 050 00.0 05.5 005.0 500.0 000.5 000.0 000.0 000.0 000.0 520003< 050 50.0- 00.5 550.0 055.0 000.5 000.0 000.0 000.0 000.0 5ZDQQD< 055 50.005 50.05- 505.5 005.0 550.0 000.0 550 0 050.0 000.50- 0000 055 55.00 55.0 000.0 050.0 050.0 000.0 500.0 000.0 000.5- 5590 505 50.00 00.0 500.0 000.0 000.0 000.0 550.0 500.0 550.5- 5050 000 50.0- 05.0 500.5 050.0 505.0 005.0 000.0 550.0 550.5 5405 005 50.5 00.5 005.0 550.0 000.0 055.0 000.0 000 0 000.0 5>000 500 05.55 00.0 550.55 000.05 000.005 505.0 000.0 005.5 000 0 >00525 050 00.5- 05.0 500.0 050.0 000.5 000.5 000.0 000.0 000.0 230>050 000 50.00 00.0 000.0 500.0 000.0 000.0 055.0 000.0 000.0 52>05 005 00.0 05.0 005.0 500.0 000.0 500.0 500.0 550.0 050.0 0>05 500 50.0 00.5 000.0 500.0 050.5 050.0 550 0 005.0 000.0 5>05 z eese seen .>ee.eee see: seawxwz wn5 cmweez mnw amawcwz. Dennemmm5 «mo5nm5um> qufiCoaoUCH - 005H059580500000 0.0 @5008 115 om53uo£uo 0 eeeocweees eewewn see 05 5 om53uo£uo 0 ceeeeec< eseeum5m 1.16 To analyze the association of industry membership with the DISC accounting method choice, the sample firms were classified by one-digit SIC codes.30 As Panel A of Table 5.9 shows, industry #2 (non-durable manufacturing) and industry #3 (durable manufacturing) account for nearly 86 percent of the sample firms, which is not surprising given the nature of the DISC legislation (see chapter 2, section 2.2 for details). The null hypothesis of no association between sample firms’ DISC accounting method choice and industry membership is not rejected (Pearson chi- square=11.34; p=0.12), but the test is weak because of low expected frequencies in many cells.31 Hence, the analysis was repeated by combining industry #1, 4, 5, 6, 7, and 8 into one group (‘other’), and a strong relationship between accounting method and industry membership was observed (Pearson chi-square=9.36; p=0.009 -- see Table 5.9, Panel B). For this reason, a control for industry membership is exercised in the subsequent analysis. 5.4.2 Exchange Listing As Table 5.10 indicates, about 77 percent of the sample firms are listed on the NYSE, followed by 21.5 percent on the AMEX, with only 1.5 percent of the sample being OTC firms. The null hypothesis of no association between DISC accounting method and exchange listing was not rejected (Pearson chi-square: 1.17; 30Although this classification scheme is crude and unlikely to result in portfolios of entirely homogeneous firms, it is similar to that used in prior accounting research (e.g., Zimmerman [1983]). 31The X2 is a large sample test and its validity requires that expected frequency of each cell be at least five [Bhattacharyya and Johnson, 1977, pp. 424- 433]. In Panel A of Table 5.2, 56 percent of the cells have expected frequencies less than five. ll7 .AcNH.o I av Nem.HH n mumsvm-wzo Umusaeoo Amm.qNo Amo.mmu oo.oo~ omm mmm mm Hmuoy Aao.oov Amm.mmv Aumcuo .wsaummcwmcm .nuammnv «m.o m N H mmow>umm HmCOHmmmmoum w Acm.~ov Aom.~mV Aumnuo .wcfimfiuum>em .maouonv om.N w m m mmoH>Hmm mdomcmHHmomwz m noo.ooav Aoo.ov Aumcuo .mocmpsmcH .mxcmnv Hm.o H H o mmofl>umm Hmfiocmcum o AHN.¢oV Amm.mmV AHHmumu new mammofionsv Hm.m NH HH o manna m Aoo.omv Aoo.0mv Ammcwfiuam .wcflxosuu .memouaamuv mN.H c N N :oHumuuoamcmuu a mcoHumoHCSEEou q ANo.ouv Awm.oNv Amousm .Numcanoma .Hmmum a couwv «m.mo HHN mofi me wcflusuommscma magmusa m nom.Hov AoH.mmv AamoHEmno .umama .mmfiauxmu .eoomv oo.oH mm mm SN wcwusuommscma manmuse-coz N ANm.mNV quo.MNV AcoHuUSMumcoo .ESmHouumu .mchHev mo.q ma OH m mmousommu w>Huomuuxm H Amway HMMOH m o mama >uumsncH nonoo on ucmouom ovozumz .uoo< unHU-mco ommo wuvou QHm ufiwflv-a:o mp «scam «Haaum u< Hoemm .mHm>qmampozH m.m maan 118 .momoLquuma CH mmwmquoumm Bomu .mMOumooHHm Hmwuumanm .muoumooaam m>wmcm£mumaoonoo paw mmwumchou zcm mszQCH uo: op mauwm mamsmm msu .um>m30z mm pmwmammmao haummoua on zme LQHLB Axme mpoo onv mmwumcwmmu mum N¢ huumsch CH UmpSHocH .Auoumooaam Hmwuumm m mm Umflmwmmmao ..Quoo mummva com mpoo uHm LuHB Spam mco ma a§ zuumSch :H pmpdaocHn .nmwwfimmmHomu mum3 @605 cm .mauwm AmeMSOmoH w>wuomuuxmv H§ xuumSUcH .mfiuwm Haw you dump Hmma no woman» .Aooo.o a my aom.m a mumscm-fiso emusanu wm.¢n mo.mmu 00.03 on m8 3 H33. Amm.mov AHo.NmV mm.qH me am ma Am .5 .o .m .¢ .Hv umsuo Aom.HoV AoH.mmV cm.mo HHN on ma AmV mmHQmHSU wCHuSUUmwscmz ANw.mmv uAmm.ONV mo.¢a mm mm qN ANV moanmudv-coc wcHHSuomMSsz Mmuoe &MUOH m 0 nfimpoo on uwmwp-wcov mama >uumSUcH unmoumm opocumz .uoo< omHQ A.e.ucoov o.m magma umnuo .> wcauauowmacmz ”m Hmcmm SAMPLE FIRMS BY EXCHANGE LISTING 119 Table 5.10 DISC Acct. Methodb Percent Exchange‘I C P Total Total AMEX 15 54 69 21.56 (21.74)c (78.26) NYSE 65 181 246 76.88 (26.42) (73.58) OTC 2 3 S 1.56 (40.00) (60.00) Total 82 238 320 100.00 25.63 74.38 Computed chi-square = 1.171 (p = 0.557). aAMEX=American Stock Exchange, NYSE=New York Stock Exchange, OTC=Over-the-counter stocks. bC=comprehensive allocators, P=partial allocators. cRow ercenta es in arentheses. P 120 p=0.56). This could, in part, be due to the small frequency of AMEX and OTC firms in the sample which, in turn, is primarily due to the COMPUSTAT listing requirement. Most of the 142 firms deleted because of that requirement were AMEX or OTC firms. However, there is no a priori reason to expect the DISC accounting method choice to be related to exchange listing other than if exchange listing surrogates for firm size.32 As discussed before, firm size is hypothesized to influence the DISC accounting method choice, hence it is included as an explanatory variable in this study. Deletion of non-COMPUSTAT firms causes another inference problem: those firms’ DISC accounting method choice could be systematically different from the sample firms; hence, inferences from this study may not be extrapolated to non-COMPUSTAT firms.33 5.4.3 COMPUSTAT Listing As Table 5.11 indicates, the sample firms are almost evenly divided between the Industrial files (56 percent) and the Research file (44 percent). Firms on the Research file are those that did not ‘survive’ due to acquisition/merger, bankruptcy, liquidation, etc. However, there does not appear to be any relationship between 3‘ZPrior research (e.g., Grant [1980]) has documented differential information content of annual earnings announcements between samples of OTC and NYSE firms on the premise that reduced levels of information are available for smaller firms (i.e., OTC firms). 33Again, there is no a priori reason for such a difference to exists other than firm size--firms not on COMPUSTAT are likely to be much smaller. However, even the inclusion of firm size as an explanatory variable is unlikely to compensate for this selection bias. 121 Table 5.11 SAMPLE FIRMS BY COMPUSTAT LISTINGa DISC COMPUSTAT Acct. Methodc Percent Listing? C P Total Total CI 46 133 179 55.94 (25.70)d (74.30) CR 36 105 141 44.06 (25.53) (74.47) Total 82 238 320 100.00 =£5.63 74.38 100.00 Computed Chi-square = 0.001 (p = 0.973). Fisher's Exact Test (l-tail) p = 0.539 (2-tail) p = 1.000.e 8Based on 1971 data for all firms. bwhere: CI = Annual COMPUSTAT (Primary, Supplementary and Tertiary) Industrial files which cover approximately 2,400 companies. CR = Annual COMPUSTAT Research file which covers approximately 1700 companies deleted from the Industrial files due to acquisition/merger, bankruptcy, liquidation, delisting, or inconsistent reporting. cC=comprehensive allocators, P=partial allocators. dRow percentages in parentheses. eFisher's exact test is based on an evaluation of the conditional probabilities assuming the marginal totals are fixed and yields the probability of observing a table that gives at least as much evidence of association as the one actually observed, given the null hypothesis is true. 122 the sample firms’ DISC accounting method choice and their survivorship status (Pearson chi-square=0.001; p=0.97). 5.4.4 Year When DISC Operational (‘DISC Year’) Ideally, in an accounting method choice study, the factors hypothesized to affect that choice should be examined prior to when the choice is exercised by the sample firms. This approach precludes the reduction and/or elimination of differences between firms hypothesized to be related to the accounting method choice decision, thereby allowing for more powerful tests. The implication for this study would be to use 1971 data, given that DISCS first became available on January 1, 1972. However, over 55 percent of the sample firms were identified from the 1973 and 1974 files of NAARS, indicating that all sample firms did not set up DISCs in 1972.“ Alternatively it is possible that although some of these sample firms set up DISCS in 1972, disclosures in the annual reports were made later either because: 1) the DISC was not operational until later, or 2) although operational in 1972, the DISC was not disclosed in the annual report because ASR 149 was not in effect until December 1973, or because the ASR 149 requirements were not implemented, or 3) the DISC was not material until later. 3"The increase in the number of DISC tax returns from 1972 to 1973 and again from 1973 to 1974 (see Table 2.1) also provides support for this conjecture. 123 To investigate this issue further, annual reports and 10K’s for the five year period beginning with fiscal 1972 of all sample firms listed on the NYSE were examined.” Unfortunately, there was no consistency in the nature and location of the DISC disclosures. In some cases, while the disclosure indicated that the DISC was established before the year first listed on NAARS, it was not clear that the DISC was also operational in that earlier year (e.g., AAR Corp.). The DISC year for such firms was assumed to be the year listed on NAARS. On the other hand, disclosures by some firms clearly indicated that the DISC was both established and operational in an earlier year (e.g., Reed Tool Co.), in which case, the firm’s DISC year was recoded to that earlier year. Despite this additional check, 141 firms in the sample could have a DISC year earlier than the year indicated in Appendix B. Because of this uncertainty, the analysis for all firms was performed using: 1) fiscal 1971 data for all firms assuming that the DISC was set up and Operational in 1972, and 2) data for the fiscal year immediately preceding the year in which the DISC was found (or assumed) to be operational. As Table 5.12 shows, after recoding as above, 144 firms (45 percent of the sample) were determined to have their DISC year in 1972, 126 firms (about 39 percent) in 1973, and 50 firms (about 16 percent) in 1974. The null hypothesis of no association between the sample firms’ DISC year and their DISC accounting 3’The search was limited to NYSE firms because of data availability. Further, the NYSE firms constitute nearly 77 percent of the total sample (see Table 5.10). 124 Table 5.12 SAMPLE FIRMS BY YEAR WHEN DISC OPERATIONAL DISC DISC Acct. Methodb Percent Yeara C P Total Total 1972 20 124 144 45.00 (13.89)c (86.11) 1973 40 86 126 39.38 (31.75) (68.25) 1974 22 28 50 15.63 (40.00) (56.00) Total 82 238 320 100.00 (25.63) (74.38) Computed chi-square = 21.742 (p = 0.000). e'The year in which the DISC became (or was assumed to become) operational. This year does not always match the year when the firm was first listed on NAARS (see section 5.3 for details). bC=comprehensive allocators, P=partial allocators. cRow percentages in parentheses. 125 method is rejected at less than the 0.001 level (Pearson chi-square=21.47). The proportion of comprehensive allocators to partial allocators is about 1:6 in 1972, 1:2 in 1973, and almost 1:1 in 1974. This association motivates separate analysis by DISC year in addition to the analysis for the pooled sample. The dramatically increasing proportion of comprehensive allocators relative to partial allocators also raised the possibility of firms switching accounting methods. The five-year search of the annual reports and 10K’s of the sample firms on NYSE (discussed above), revealed that of the 246 NYSE firms examined, 10 switched accounting methods (nine from partial to comprehensive allocation and one from comprehensive to partial allocation). 5.5 Statistical Procedures 5.5.1 Univariate Analysis As an initial step, univariate tests of the relationship between the accounting method choice and contracting and political costs variables were conducted using both the parametric Student’s t-test and the non-parametric Wilcoxon rank-sum test.36 36The Wilcoxon rank-sum test is equivalent to the Mann-Whitney U test. Both these tests are identical to the Kruskal-Wallis one-way analysis of variance test for comparing k treatments if k=2. See Bhattacharyya and Johnson [1977] and Conover [1980] for a further discussion of these tests. 126 The Student’s t-statistic tests for the equality of means x1 and x2 between two independent normally distributed samples with 111 and 112 observations and is computed as follows: t = (Y, — x2) / ,/s§(1/n, + 1/n2) where sf, is the pooled variance and computed as follows: 3: = {(n,—1)s§ + (n2—1)s§} / (n,+n2-2) where sf and Si are the sample variances of the groups. Note that this t-statistic depends on the assumption that the population variances of the two groups, a? and oi are equal. A folded form of the F-statistic, F’, is used to test the assumption of equal variances, where: F’ = (larger of Si, Si) / (smaller of 5}, Si) If the null hypothesis of equal variances is rejected at a two-tail probability level of .10 or less, then the following t-statistic is used: t = (71 - Y2) / J(Si/n1 + Si/nz) However, both specifications of the t-statistic assume that the underlying population distribution is approximately normal. Nonparametric procedures overcome the normality assumption by being ‘distribution-free’. Because most variables used in the study are skewed (see Table 5.8) and the Kolmogorov- Smirnoff test for the null hypothesis of normality is rejected for most variables, the nonparametric Wilcoxon rank-sum test also is used. The test consists of ranking all observations in the combined sample and then determining the sum of ranks of the sample with the smaller size. The Wilcoxon rank-sum statistic tests the null 127 hypothesis that the two samples with nl and 112 observations are drawn from an identical population. Because both 11, and n2 are large, the normal approximations to the rank-sum statistic is used and is computed as follows: Wl — n,(nl + 112 + 1)/2 Z = \/n1n2(n1 + 112 + 1) / 12 where Wl denotes the rank-sum of the sample with n1 observations. However, the univariate analysis ignores correlations between the predictor (independent) variables. To the extent these correlations are significant, observed univariate differences must be interpreted with caution. This also motivates a multivariate regression approach which overcomes the above problem and enables examination of the joint explanatory power of the contracting and political costs variables. The main focus will be on the multivariate analysis. 5.5.2 Multivariate Analysis Recall from equation (5.1) and the subsequent discussion that the complete model to be estimated in this study is yi = bo + b,LEVi + bZINTCOVi + b3DCOVi + b,PLEV, + bSSIZE, (Predicted (+) (—) (+) (+) (—) sign) + b6ETR, + b7AUDDUM1, + bSAUDDUM2, + u, (3’) (-) (+) where: yi = 1 if the firm uses partial allocation, and zero otherwise, and LEV,, INT COV,, DCOV, PLEV, SIZE, ETR, and AUDDUMi are the predictor variables for firm i corresponding to the contracting costs, political costs, and auditor hypotheses. 128 The dependent variable y (the DISC accounting method) is not continuous but represents a discrete choice between comprehensive and partial allocation.37 The objective in models of this type is to predict the likelihood (probability) that a particular choice will be observed, given the factors hypothesized to affect that choice. The three common forms of the probability function most frequently used to estimate a model with a binary dependent variable are 1) the linear probability model, 2) the probit model, and 3) the logit model [Amemiya, 1981]. The linear probability model may be expressed in the usual regression framework as: y. = a + (5.3) with E(e,-) = 0. Here x, is the (1 x k) vector of independent variables for sample observation i. The conditional expectation E0), lg) = xfi and can be interpreted as the probability that the event will occur given the x,. The predicted value of y from the regression model, 0, = x? then yields the estimated probability that the event will occur given the x,. The major defect of the linear probability model is that 9,. is not constrained to lie between 0 and 1 as a probability should. Although this defect can be corrected by defining y, = 1 if y, > 1 and y, = 0 if y, < 0, "the procedure produces unrealistic kinks at the truncation points" [Amemiya, 1981, p. 1486]. In addition, because y, takes on only two values (0 or 1), the error term in equation (5.3), e,- is not normally distributed and Var(e,-) = E(y,-)[1 — E(y,-)]. The 37Models involving dependent variables specified in this manner are called ‘discrete choice models’, ‘qualitative response models’, or ‘categorical models’. 129 heteroscedastic error term causes the OLS estimates of )9 to be inefficient. The inefficiency problem can be remedied by using weighted least squares. However, the problem remains that E(y, I.) can lie outside the admissible (0,1) interval. The probit and logit are nonlinear models that overcome the major limitation of the linear probability model by using a cumulative probability function (an S- shaped curve) which is bounded in the interval (0,1) such that 0 5 E(y,|,) 5 1. The theoretical rationale for these two models is summarized below and is largely based on Maddala [1983, pp. 22-27]. Assume there is an underlying response variable y: defined by the following relationship y! = xa + e.- (5.4) Instead of observing yf, however, what is observed in practice is: f 1 if y,’ > 0 y = l l 0 otherwise (5.5) From (5.4) and (5.5) we get Prob(y,. = 1) = Prob(e.~ > x15) 1 - F(—x,fi) where F is the cumulative distribution function for 6. Observed values of y may be viewed as realizations from a binomial process with probabilities given by (5.5) and varying from trial to trial depending on x,. The likelihood function can then be expressed as L = ‘lTF(-x,e)'lT [1 — F(—x-,6)] (5.6) YF-o Yi=1 130 The functional form of F in (5.6) depends on the assumptions made about 6, and provides the point of departure between the probit and the logit models. In the probit model, 6,- are assumed to be independently normally distributed with mean 0 and variance 02 and r—xi-fi/a F(—x,.fi) = | {1 / (21r)"’} exp(—t2 /2)dt (5.7) J -00 In the logit model, the cumulative distribution of e, is the logistic and exp(—x.a) 1 F(—x,.fi) = = (5.8) 1 + exp(—x,,6) 1 + exp(x,.fi) Maximum likelihood estimates of I9 can be obtained by first substituting in (5.6) the assumed cumulative distribution of F, then simplifying the expression by taking logs, and then differentiating the log likelihood function with respect to fi- The resulting equations being nonlinear in 13 have to be solved using an iterative procedure such as Newton-Raphson or the scoring method. The choice between probit and logit should be based on theoretical grounds but "well-developed theory to determine the exact functional form [to be used] appears to be lacking" [Kmenta, 1986, p. 555]. However, in univariate dichotomous models, such as (5.2) estimated in this study, probit and logit usually give similar results except if the data is heavily concentrated in the tails [Amemiya, 1981]. The logit model is used in this study primarily because . it represents a close approximation to the cumulative normal and is simpler to work with" [Kmenta, 1986, p. 555]. 131 In addition to the logit analysis, OLS regression results are also presented despite the theoretical problems of OLS discussed before. This is primarily because Noreen [1988] provides simulation evidence demonstrating that OLS performs as well as probit for sample sizes of 50 or 100 when using accounting data typically used as explanatory variables in accounting choice studies. Chapter Six ANALYSIS OF RESULTS This chapter presents the results of the tests conducted to examine the possible motivations for firms to choose between comprehensive and partial allocation of taxes with respect to the indefinitely deferred DISC income. Recall there are two sets of data collected--1971 data for all firms and data for the year in which the DISC became (was assumed to become) operational (‘DISC-year’ data). First, the results of the univariate analysis for the complete sample on both data sets is presented. Second, the cross-correlations between the independent variables are discussed. Third, the results of the multiple regression analysis the logit and OLS procedures are evaluated. Finally, sensitivity of the results to partitioning the sample by industry membership and DISC-year are examined. 6.1 Univariate Analysis In this section univariate differences between the comprehensive and partial allocators for each of the explanatory variables hypothesized to affect firms’ DISC accounting method choice are analyzed. The significance tests for differences are performed using both the parametric Student’s t-test and the nonparametric Wilcoxon rank-sum test described earlier. The results of the analysis are summarized in Table 6.1. 132 133 Amn¢.ov Among: Ann¢.ov Amon.ov a.emm.~- ccgNo.nu moN.o «Nn.o ONm geamm.N- «ceNo.mu mon.o an.o 0Nn m ZDQMDQ Aamn.ov Awnc.ov AHNm.oV Amnc.ov :emn.N «¢om.N ¢m~.o mmN.o on acmn.N ecom.N «wa.o mmN.o omn N ZDQMDQZ “MNn.ov Anna.ov AmNm.ov Anna.ov «eem¢.Nt ee¢n¢.mu mHH.o «No.0 ONm :«cnn.Nu «eenn.mu NNH.o «No.0 ONn m v u N Eboo2< ANmm.ov Rama.ov ANmm.ov Aon¢.ov temH.N «emo.N mmH.o nom.o omm *cwm.H eanm.fl mwa.o mmN.o ONn m A U a ZDQQD< ANNH.oV Ammo.ov Amqa.ov “moa.ov emm.a mn.a mm¢.o onc.o oon eeeqm.N na.H acq.o mnq.o oon N m mam Amnfl.ov ANHH.OV Aana.ov AH¢H.ov Hm.a «4N.H oH¢.o mmc.o com 00mH.N Nn.a Noe.o nm¢.o oon N N mam Anaa.ov Anmo.0v ANmH.ov AoHH.ov eNm.H eano.N Hqc.o nm¢.o Nam ceeom.N HN.H wqa.o qu.o mom w H mam Amom.av Aunm.flv Aqom.Hv Annm.av nm.o No.0 HqN.q mmo.c Nam No.o nH.ou NNm.q wcm.¢ Non m A U Amgm Romm.ov Amom.ov Annm.ov AnHm.ov mm.o NN.ou Noa.o onm.o mom HN.H no.0- qu.o noN.o nmN m v U N >009 Ammo.mav ANNQ.NNV AmNN.ONV AnqN.nNV c.0HN.N ecmN.N Nwm.m mom.NH Non eeoo.N «on.H mun.m www.ca mmN m A u >OUHzH Amma.ov Amqa.ov AHNH.ov Anna.ov ecam.au «cco.Nn 044.0 moq.o NHm eenw.au «eNo.Nc ch.o mmm.o mom m v o c >ma Amoq.ov AcmN.ov ANHq.ov Ammm.ov 0mm.Hn «ema.Nu mH¢.o mNm.o Nam cqm.fl- cgnn.Hu NNc.o oqm.o Non m v U H >mg N a m U z N u m U z mam0nuoakm canoau0> unco0z c003u0n 00G0u0uuao A.>0Q .Uumv mca02 5003002 00C0u0uuwa A.>0o .vnmv uou owunduuum uou oaumauauw C00: anon umeuUhHD cusp HhmH moHHmHHHZD H.o 0Han 134 .AZDQMDQ cam MHm How 0000 H000-030 ”0m00 me0-0cov H0>0H H0. 050 00 0cmonchHm «ax .AZDQMDQZ cam mew How 0000 H000-030 ”0m00 HH00-0cov H0>0H m0. 050 00 0:00Hchme «« .AMHm How 0000 H000-030 ”0000 HH00-0GOV H0>0H OH. 030 00 0GO0HMHCwNm « .0000 Edm-xcmu coxooHHB 0:0 mo cow0mEonuaam H0500: 0:0 Co @0009 mg owumw0m0m -N 059 .H0>0H no.0 0:0 00 p00o0n00 09 :00 0Hnmwum> onHo0am 0 How masouw 030 C00300p 000cmwum> H0300 mo mHm0£0oa%£ 030 0030 000mowccH 0000-m cm MN m0oCMHum> CH m0uC0u0wva How c00msmcm ma 000mw0m0m-0 0:90 .000mooHHm Hmw0uma n m ”000000HH0 0>Hmc0£0uQEoo n on .maoauacamme magmaum> you m.m manna mam. A.e.ucoov 0.0 «Home 135 Leverage Constraint (LEV). Based on 1971 data, partial allocators have higher leverage than comprehensive allocators--the mean LEVI (LEV4) for the two groups was 0.42 (0.44) and 0.34 (0.40), respectively. This finding is consistent with the hypothesis that firms with higher leverage are nearer to possible debt covenant violations and, hence, more likely to choose income-increasing accounting procedures (i.e., partial allocation) in order to relax those constraints. The t-test indicates that the difference between the mean leverage of the two groups is statistically significant at less than the .05 level. The Wilcoxon Z-statistic is also significant at less than the .05 (.01) level for LEV] (LEV4).1 The mean market value-based leverage measure (LEVMl) is 0.31 for the comprehensive allocators and 0.38 for the partial allocators. While this difference is in the hypothesized direction, it is not statistically significant (t-statistic = —1.04; Wilcoxon Z-statistic = —0.96). Since LEVM] was affected by some extreme values (see Table 5.4), the tests were rerun after setting values of LEVMl greater than five to five, with no difference in the results. These results for LEVMl are not surprising in view of the fact that debt covenants are written in terms of book- values and not market-value based measures of leverage. Based on DISC-year data, results for LEV1, LEV4, and LEVMI are similar to those presented above. Consistent with the hypothesis, partial allocators have ‘There were 11 firms with missing data for intangible assets (IA) which affects the computation of LEV4. These firms were included for the univariate tests by coding IA as zero. The results are unaffected if these firms are instead omitted. 136 higher leverage in their capital structures than comprehensive allocators. The difference in the mean LEV] and LEV4 between the two groups is in the hypothesized direction and remains statistically significant. However, as before, the difference in the mean LEVMl of the two groups is not significant. Interest Coverage (INTCOV). Based on 1971 data, comprehensive allocators have a mean INTCOV ratio of 14.63, while that of the partial allocators is 9.6. This is consistent with the hypothesis that firms with lower interest coverage are more likely to violate their interest covenants and, hence, more likely to choose an income-increasing accounting method (i.e., partial allocation). The t-statistic for the difference between the means of the two groups is significant at less than the .10 level, whereas the Wilcoxon Z-statistic is significant at the .05 level. In the above computation, INTCOVs greater than 100 were reset to 100. Three firms were affected by this procedure. In addition, 10 firms had zero interest expense. However, nine of the 10 firms had positive IBEI and their INTCOV was reset to 100; if these nine firms are omitted instead, the difference between the two groups becomes more statistically significant (t-statistic=2.22 and Wilcoxon Z- statistic=2.17, both significant at less than .05).2 Based on DISC-year data, the mean INTCOV ratio of the comprehensive allocators is 17.3, whereas that of the partial allocators is 9.7. This difference again is in the hypothesized direction and more significant than using 1971 data--the t- 2Recall from chapter 5, section 5.2.2, that these recoding procedures were necessary to reduce the skewness in the distribution and are similar to those employed in prior research. 137 statistic (Wilcoxon Z-statistic) is 2.23 (2.71), which is significant at less than the .05 (.01) level. In these computations, resetting procedures similar to those described for the 1971 data were used.3 Dividend Constraint (DCOV). Based on 1971 data, relative to partial allocators, comprehensive allocators pay out a smaller proportion of their inventory of payable funds (unrestricted retained earnings, URE) as dividends--the mean DCOV ratio of the comprehensive allocators is 0.28 while that of the partial allocators is 0.34. This is consistent with the hypothesis that firms with higher DCOV ratio are closer to possible dividend covenant violations and, hence, more likely to adopt income-increasing accounting methods (i.e., partial allocation) to relax those covenants. However, the observed difference in the mean DCOV ratio of the two groups is not statistically significant. Recall that there were 52 firms with dividend restrictions mentioned but not quantified. The DCOV ratio of these firms was computed by substituting retained earnings (RE) for URE if RE was positive, or coding DCOV three if RE was zero or negative. There is virtually no change in the results if these firms are omitted instead--the difference in the mean DCOV ratio of the two groups remains in the hypothesized direction but is not significant.‘ ’1’ he skewness of the distribution was reduced substantially from 15.8 to 3.3. Without resetting, INTCOV ranged from —10.9 to 1676.7, resulting in large variances for the two groups and an insignificant t-statistic, although the Wilcoxon Z-statistic remained significant at less than the .01 level. ‘In addition, recall that for the 77 firms with no dividend constraints, the DCOV ratio was coded zero. If these firms are omitted instead, the mean DCOV for both groups is relatively higher (0.43 and 0.63, respectively), is in the 138 Based on DISC-year data, the mean DCOV ratio of both groups is relatively higher and the difference between the means is in the hypothesized direction but again is not statistically significant. Public Debt (PLEVDUM). A dummy variable was used for the existence of public debt and was coded one if the firms had public debt outstanding and zero otherwise. The means of a dummy variable offer a convenient interpretation because they simply represent the frequency with which the phenomenon occurs. Accordingly, based on 1971 data, 31.7 percent of the comprehensive allocators compared with 31.9 percent of the partial allocators had public debt outstanding. This difference is consistent with the hypothesis that because public debt is costlier to renegotiate, firms with public debt are more likely to be concerned about possible debt covenant violations and, hence, choose an income-increasing accounting method (i.e., partial allocation). However, the difference between the two groups is statistically insignificant. Because the use of a dummy variable ignores the dollar magnitude of public debt, PLEV also was computed as a ratio of the amount of public debt to 1) net tangible assets (NTA) and 2) total debt (TD). Under both measures, partial allocators were found to have a greater proportion of public debt in their capital structure than comprehensive allocators which is in the hypothesized direction but hypothesized direction, but not statistically significant (although the t-statistic for the difference is relatively higher than if these firms are included, —1.43 v. —-0.67). 139 the difference between the two groups remains statistically insignificant. Similar conclusions are reached with DISC-year data. Firm Size (SIZE). Firm size was measured as the log of net sales (ISAL) to reduce skewness. There is virtually no difference in the size of comprehensive and partial allocators. Based on 1971 data, comprehensive allocators are found to be marginally smaller than partial allocators, which is inconsistent with the political cost hypothesis that larger firms will choose an income-decreasing accounting method (i.e., comprehensive allocation). However, based on DISC-year data, comprehensive allocators are observed to be marginally larger than partial allocators, which is consistent with the political cost hypothesis. Neither of these differences is statistically significant. Sensitivity of the results to the measure of firm size was tested by using the log of total assets and the log of net income with no change in the results. Efi'ective Tax Rate (ETR). Based on 1971 data, the mean ETR1 for the comprehensive allocators is 46.5 percent compared with 44.6 percent for the partial allocators. The t-statistic does not indicate that this difference is statistically significant. However, the Wilcoxon Z-statistic is significant at less than the .01 level. Generally, similar conclusions can be reached when comparing ETR2 and ETR3 for the two groups. These results are consistent with the tax-based political cost hypothesis that higher effective tax rates are indicative of greater political costs 140 borne by firms which induces the selection of income-reducing accounting procedures (i.e., comprehensive allocation).S As Table 5.8 indicates, the distribution of all three ETR measures is highly skewed. Two observations in particular are strange and may be largely responsible for the skewness. AVC Corp. has the highest ETR1 and ETR2 (4.8 and 4.3, respectively) but the lowest ETR3 (—32.6). Control Data Corp.’s ETR1 and ETR2 are nearly —1.0 but its ETR3 is —12.3. Both firms are partial allocators and dropping them substantially increases the mean ETR3 of the partial allocators from 24.7 to 45.4 percent. Dropping the two firms also reduces much of the skewness of all three ETR measures, but the statistical significance is not affected. Because of the well-known difficulties that arise in interpreting effective tax rates of less than zero and greater than one, the analysis also was conducted by constraining each of the ETR measures to lie in the (0,1) interval. Relative to partial allocators, comprehensive allocators continue to have larger mean ETRs. The t-statistics for the difference between the two groups are marginally higher than before but remain insignificant. The Wilcoxon Z-statistics continue to be significant at the same levels. Based on DISC-year data, the overall conclusions reached earlier remain unaltered-comprehensive allocators have higher effective tax rates than partial allocators indicating support for the tax-based political cost hypothesis. However, 5Recall that there are 25 firms with missing minority interest (MI). These firms were included in the above analysis with their MI set to zero. However, there is no change in the results if these firms are omitted instead. 141 there are two exceptions: 1) the t-statistics for the differences between the mean effective tax rates for the two groups using ETR1 and ETR2 are now statistically significant at .05 and .10 level, respectively and 2) the Wilcoxon Z-statistic is significant at the .10 level for ETR1 and ETR3 but insignificant for ETR2. Auditor (AUDDUM). Two separate dummy variables for the auditor influence on the DISC accounting method choice were set up as follows: AUDDUMl was coded one if the sample firm’s auditor was Arthur Andersen and zero otherwise; AUDDUM2 was coded one if the sample firm’s auditor was Price Waterhouse and zero otherwise. As before, the means of AUDDUMl imply that, based on 1971 data, 29.3 percent of the comprehensive allocators compared with 18.9 percent of the partial allocators were audited by Arthur Andersen. Similarly, the means for AUDDUM2 imply that Price Waterhouse was the auditor for 12.2 percent of the partial allocators compared with only 2.4 percent of the comprehensive allocators. Both the t-statistic and the Z-statistic indicate that these differences are statistically significant at the .05 level for AUDDUMl and at the .01 level for AUDDUM2. The results are virtually identical when DISC-year data is used. These results strongly indicate the presence of an auditor influence on firms’ DISC accounting method choice, which is a relatively new finding in the accounting method choice literature.“ Consistent with Price Waterhouse’s position favoring ‘An alternative explanation for these results is that client preferences influence auditors’ positions. That explanation is not compelling in this case because the auditors’ positions on accounting for income taxes was known about 142 partial allocation, five times as many of its clients with DISCs used partial allocation as used comprehensive allocation. Similarly, consistent with Arthur Andersen’s stated preference for comprehensive allocation, nearly twice as many of its clients with DISCs used comprehensive allocation as used partial allocation. Industry Membership. As discussed before, the DISC accounting method choice of the sample firms is not independent of their industry membership based on one-digit SIC codes. This is further corroborated by the univariate tests. Whereas nearly 30 percent of all comprehensive allocators can be classified as non- durable goods manufacturers (industry #2), only 16 percent of the partial allocators fall in this industry group. This difference is statistically significant at the .05 level.7 However, just the opposite pattern exists for durable goods manufacturers--nearly 71 percent of all partial allocators fall in this industry group compared with just over 52 percent of all comprehensive allocators. This difference is statistically significant at the .01 level. These results are identical for 1971 data and the DISC- year data indicating that there was no change in the industry membership of firms which had 1973 or 1974 as their DISC-years. Summag. The univariate tests of differences between comprehensive and partial allocation provide some support for the costly contracting-based hypotheses emanating from positive accounting theory. Consistent with those hypotheses, firms 1965-66, long before DISCs became available in 1972. 7The significance levels for the industry differences are based on two-tail tests because a priori there are no directional expectations for firms to differ in their DISC accounting method choice based on their industry classification. 143 with higher leverage and lower interest coverage are found to adopt an income- increasing accounting method. However, a similar hypothesis based on dividend restrictions is not supported. In addition, no support is found for the hypothesis that firms with the presence of public debt or with a relatively higher proportion of public debt in their capital structure are more likely to use an income-increasing accounting method. The results of the political cost hypotheses are mixed--firm size is not found to be a significant factor in firms’ DISC accounting method choice. However, results of the effective tax rate variable support the hypothesis that firms bearing higher political costs as manifested by higher effective tax rates are more likely to choose income-reducing accounting methods. Finally, strong support is found for the auditor influence on firms’ DISC accounting method choice, which is a relatively new result in the accounting method choice literature. 6.2 Multiple Regression Analysis The univariate analysis discussed in the previous section ignores correlations among the independent variables. Based on the theoretical framework presented earlier, variables used to operationalize the contracting and political cost hypotheses are expected to be correlated. If these variables are indeed correlated, it becomes imperative to examine their joint effect on the DISC accounting method choice in 144 a multiple regression framework. Their inclusion in the regression model also is necessary to avoid a correlated omitted variables problem.B Pearson product-moment correlations among the independent variables for all sample firms using 1971 data and DISC-year data, respectively, are presented in panels A and B of Table 6.2."10 As the signs, magnitudes, and statistical significance of the cross-correlations in the two matrices are virtually similar, the discussion here focuses on panel A (all firms using 1971 data). The correlations among the contracting cost variables are consistent with the positive accounting theory predictions. As expected, all three leverage measures are negatively correlated with interest coverage but positively correlated with the dividend constraint variable and the public debt dummy. These correlations are statistically significant and range from 0.25 to 0.36. Given the magnitude of these correlations, it is unlikely that the simultaneous inclusion of the leverage, interest coverage, and dividend constraint variables in the regression model will result in a multicollinearity problem. Interestingly, the correlations between leverage and the 8Omission of correlated variables in a regression model results in biased parameter estimates. This bias does not disappear even as the sample size grows large making the parameter estimates inconsistent as well [Pindyck and Rubinfeld, 1981, pp. 128-130]. ”Spearman rank correlations were also computed but are not presented because they are qualitatively similar to the Pearson correlations. Mention is made in the text where the Spearman correlations differ substantially in magnitude or statistical significance. loCross-correlations with LTA (the third measure of firm size used in the study) are not presented because they are virtually identical to the cross-correlations with LSAL and the correlation between LTA and LSAL is 0.97. 1115 man ooo.fi man man man man man man mum mom mam won mam mom mom mom emmm.ou Hmo.o: aco.o omo.o wo«~.o cmm~.o wnNH.on «enH.ot ¢¢HN.01 @cHH.01 QHHH.0 owo.o: wmmfl.o: toma.ou Zbombo man man man man man man NAN non mam mom mam non mom mom ooo.H noo.o: Hmo.01 Hno.o: Nw0.01 wnHH.On ®o0H.o sm®H.o cmmH.o mmo.o omo.ou omo.o coo.o- Hmo.o ZDQmDQz man man man man man mum non mam mom mam mom mom mom coo.“ {NNH.ou nqo.o moo.ou Noo.ou mno.o Nmo.o Nmo.o Nqo.Ot Hmo.o: uno.o nmo.ou nco.ou NZDQQDC man man man man mum mom man mom mam mom mom won ooo.H mne.o: mmo.01 mno.ou «no.0 mmo.o qwo.o ono.o Nco.01 nqo.o: m~o.o wmo.o HZDQQD< man man man mum mom man mom mam mom can won ooo.u «nom.o comn.o c~sa.ou «no.0: nno.oa cm~n.ou «no.0 emwa.ou «mo.ou @mHH.ou nmhm man man nnN non mam con man mom mom mom ooo.H cmmn.o @mn~.on «sma.oo ennH.ou «NnN.ou @NHH.o @nma.01 mmo.on enma.on NmHu man nnN non man con man new mom won ooo.H mmNH.OI smna.01 Noo.on ann.ou @nNH.o wmo.on mac.01 Haa.ou Hmhm «mu mum NAN wmm hum «RN mum hum coo.H cumm.o «moc.o nmo.ou n~o.o: Nco.o 000.0 Hmo.o H24 nom nom «mm non Non non non ooo.H «mwn.o moo.o enma.ou cfinm.o anN.o {www.o Amqm com mon «mN nmN nmN ooo.H «mm~.o: :qu.c eNmH.o cmnm.o N>OUQ man mom mom mom ooo.a emmN.on cmw:.o: «wmm.o| >OUHzH mom mom non ooo.H camn.o «omn.o H2>mq mom mom ooo.H cmn~.o ¢>mq mom coo.“ t10>ma ZDQmDQ Ebomboz NZDQQD< HZDQoac mmhm thw HmHm H24 Amqm N>OUQ >OUHzH Hz>mq c>mq a>mq 90H90w00> mung Hama - «spam 044 "< Hosea umHzmHonmmoo ZOHHmqm 00m 00m mmN com com 000.H :NNH.01 cnnN.0 {MNN.0 c00~.0 N>OUQ 0N0 Nam Nam Nam 000.H «mmN.01 amon.01 «000.01 >OUHzH Ham Ham Ham 000.H «wan.0 enNN.0 Hz>mq Nam Nan 000.H «545.0 4>w4 Nam 000.a fl>mq Zbambo SDQmDQz NZDQQD< HZbQQ=< mmhm NmHm Hmhm H24 Amqm N>OUQ >OUHzH Hz>mq 4>m4 H>mq 90H90000> 0000 HmoquMHn . mfiuwh H.344 "m H0C0m A.e.ucoov «.0 manme 147 .0cofl0wcww00 0H£mHum> pom w.m 0HQMH 00ma .H0>0H Am0.v Ho. 0:0 00 00000000 00 0H00000> CESHOU 050 0:0 0Han00> 300 050 0003009 Cow0ma0uuoo 000m mo 00005000»: HHSC 050 0050 00000H0CH A®v .0. < .0CoH00>000£o mo 0008:: 0:0 000 0CHH 050000 050 00 0am 0000000mw0oo coHO0H0uuoo 050 000 0CMH 00000 050 00 0000552. A.e.ucoov ~.o magma 148 two industry dummies are lower than 0.15 and not statistically significant for the non-durable goods manufacturers. For this reason, no control for cross-sectional variations in leverage for firms in different industries was exercised. Firm size is significantly negatively correlated with the effective tax rate measures (except the correlation between LSAL and ETR3) although the magnitude of these correlations generally is less than 0.20.11 The negative correlation implies that for this sample as a whole larger firms have lower ETRS, which is inconsistent with the tax-based political cost hypothesis. However, this finding should be interpreted with caution because the ability of the effective tax rate to proxy for political costs is known to vary with time and across industry groups [Zimmerman, 1983]. Firm size also is significantly positively correlated with leverage and the public debt dummy. The latter correlation is consistent with the claim that accessing capital markets and borrowing from the public is associated with higher costs and requires greater resources, which larger firms may be better able to afford. The correlation between firm size and industry classification also is of interest because of the frequently mentioned argument that firm size could be a proxy for industry membership (e.g., Ball and Foster [1982]). In this sample, firm size is positively correlated with non-durable goods manufacturers but negatively correlated with durable goods manufacturers. Given these correlations and the income effect of the DISC accounting method choice, the finding presented earlier 11Spearman rank correlations are greater than 0.20 but less than 0.30. 149 that a greater proportion of the non-durable goods manufacturers (which are comparatively larger) chose comprehensive allocation (which results in lower income) is consistent with the political cost argument that larger firms have greater incentive to adopt income-decreasing accounting procedures. Similarly, the finding that a greater proportion of durable goods manufacturers (which are comparatively smaller) are partial allocators would be consistent with that argument. The observed correlations between firm size and industry membership, as well as the association between sample firms’ DISC accounting method choice and their industry membership, warrant the inclusion of industry as a control variable in the multiple regression model. Turning to the auditor variable, the cross-correlations of particular interest are those between the auditor dummies and 1) the contracting and political cost variables, and 2) the industry dummies. A finding of significant correlations between the auditor dummies and the contracting and/or political cost variables would complicate the analysis because that finding would be consistent with the argument that all these variables are proxies for some common construct. However, that problem is not present in this sample because both auditor dummies have statistically insignificant correlations with all of the contracting and political cost variables, with the magnitude of the cross-correlations being less than 0.10. The low correlations also suggest that the auditor may indeed be an important and relevant omitted variable in an accounting method choice model. 150 The absence of an association between auditor and industry classification (correlations are lower than 0.05) also is important in interpreting the results of this study. Generally, it is believed that the big-8 auditors, in particular, tend to specialize in certain industry groups in order to distinguish themselves from the other auditing firms. Hence, it could be argued that the earlier finding of a strong auditor influence on firms’ DISC accounting method choice could simply be because of an industry effect, which, incidentally, also is strongly present in this sample. However, the cross-correlations between the auditor variables and industry classification are less than 0.05, suggesting that it is unlikely that the auditor variable is a proxy for an industry effect.12 6.2.1 The Logit Model and Its Evaluation The multiple regression analysis is performed using the logistic model described earlier. The dependent variable is binary and takes on the value one if the sample firm is a partial allocator and zero otherwise. The independent variables--leverage, interest coverage, dividend constraint, public debt dummy, firm size, effective tax rate, and auditor dummies--are as defined before. In addition, industry dummies are used as control variables. The results of the logit analysis are evaluated in three ways. First, the significance of each independent variable (parameter) is evaluated by testing the null hypothesis that the parameter is zero. The test is performed using the 12Note, that the lack of a significant correlation also could be due to the restrictive definition of industry used in this study. 151 asymptotic t-statistic calculated by dividing the maximum likelihood estimate of the parameter by its standard error. Second, the joint significance of all variables in the logit models is examined via the likelihood ratio test (see Kmenta [1986, pp. 550—556]). The test statistic for this asymptotic test is LR = --2[L(¢3) — L(fi)] which is approximately distributed as chi-square with k degrees of freedom, where, in general, k is the number of explanatory variables in the model. L( x2“). Third, the classificatory success of the logit models is evaluated against the proportional chance criterion (CW0) suggested by Morrison [1969].13 Formally, Cpro = 0:2 + (1 — a‘) where a is the proportion of observations in Group 1 and (1 — a) is the proportion of observations in Group 2. The term (22 is the product of the conditional 13Another technique that also could be used for this purpose is discriminant analysis, which is based on the assumption that the underlying variables are jointly normally distributed with equal covariance matrices. Because some of the explanatory variables in this study are qualitative (i.e., the data is not multivariate normal), the logistic model is preferred over linear discriminant analysis. See Press and Wilson [1978] for further discussion on this issue. 152 probability of correct assignment given that the observation was classified in Group 1 and the probability that the classification of the observation was into Group 1. Similarly, (1 — 0:2) is the product of the two probabilities related to Group 2. An alternative to Cpro is to use the maximum chance criterion (Cm) under which all sample observations are naively assigned to the larger of the two groups. Formally, Cm, = max (02, 1 — a) with a as defined before. The choice between the two classificatory criteria depends on the objective of the research. Morrison [1969, p. 158] advocates that Cm, is the appropriate benchmark to use if the sole objective is to maximize the percentage correctly classified. However, he points out that: Usually a discriminant analysis is run because someone wishes to correctly identify members of both groups. [However] the discriminant function defies the odds by classifying an individual in the smaller group. The chance criterion should take this into account. Therefore, in most situations CPm should be used. Because in this study the classification of both comprehensive and partial allocators is of interest, Cpro is used as the benchmark for evaluating the performance of the logit models. Finally, to evaluate the "goodness-of-fit" of the logit models, the likelihood ratio index (LRI) is computed as follows: LRI = 1 — [um/15(0)] 153 where L(fi) and L(0) are defined as before. The LRI is a scalar measure similar to the R—square in the standard regression model and provides an indication of the logit model’s explanatory power. It can be shown that OSLRIgl and better the fit of the model, the closer the LR! will be to one [Kmenta, 1986, pp. SSS-556].“ 6.2.2 The Logit Regression Results The logit results for the full sample are presented separately using 1971 data in Table 6.3 and using DISC-year data in Table 6.4. Models 1, 2, and 3, respectively, correspond to the three effective tax rate measures (ETR1, ETR2, and ETR3 defined earlier). Panel A in each table presents the results of the three models based on LEV1 as the leverage measure, whereas the panel B results are based on LEV4 as the leverage measure. In the first row, the maximum likelihood parameter estimates of the variables are presented followed by the related asymptotic t-statistic in the second row. Also presented in the table are 1) the chi- square statistic (based on the likelihood ratio test discussed earlier) which tests the null hypothesis that all coefficients (except the intercept) are equal to zero, and 2) the percentage of firms correctly classified by the logit models. As Table 6.3 indicates, based on 1971 data, if leverage is defined as LEV1 both leverage and interest coverage are significant in the hypothesized direction at the .10 level for all models. However, if leverage is defined as LEV4, the leverage variable becomes significant at .05 level, whereas interest coverage becomes “See also Amemiya [1981, pp. 1502-1507] and Maddala [1983, pp. 37-41] for several other goodness-of-fit measures. 154 .153... 01.3. .2: we :0... ma 000.5oom: 3.: no 33.0—40.0... s... 091.340; 0:... .003... ..omamv..m0.u 00603000 0.... :. 9.00.0000 «x 0... 00 0.3.5.0 a. UH .—ACIQ uCqu-UWCOO 0:9 80.43: 163.:nflv—u— MO— \ QUCOMHQ>COO HQ QOOp-amflv—u— ”ORV l —_ mm. UQC‘VQU m— UCQ ”93:“ Oman...“ U00£m~04m~ flip a. —~H—a .0>o:= 0.000! ..a new .0>0. .0. 0:0 :3:0 002. 00 .:0u...:m.0 0. 0. .000N 0.: ..300000:. 0:0 0=0ux0v 00:0.u.uu060 0;. ..a 00:0 0.00:000x5 ..3: 0:0 00000 50.23 0000 0.00. 090:..0x.. 0:0 :0 00000 0. 0.00.0000 000:00-.:u 0:90 .».0>.0000000 .0090 0:0 .szm ..000 .00030003 0000 000 0>.000000 00050 050 cu 0:00000000 n 0:0 .N .. 0.0002o ..0>0. .0. 0:0 00 00:00.u.:w.0 00009.03. «4! ..0>0. n0. 0:0 00 00:00.0.cw.0 00000.0z. «« ..0>0. 0.. 0:0 00 00:00.u.:w.0 00000.0s. a .00.00.0000.0 0.0000ax00 0000.06000 0:0 0:.. 0:9000 0:0 0:0 000:5.000 0:0.u.um0ou 000 0000000 0:.. 000.0 02h .40N I :. «««.0.N 00.0. «eNN.N «00..N- n...- N..0 00.0- N..0 .0.0- «ano.N 4..0- «0.. n.mm n..wn 00... .co.0- nNn.. 0Nm.0- .0...- 0N0.o mNc.o- nmc.0 000.0- mmn.N 0m. 0. m «s400.N 03.0. «a...N «am..N- 0...- c..o 0..0. NN.0 «am0.0- «aoo.N 4..o- 00.. n.4N 0N.nm as... 4.0.0- NNO.. 4NN.0- 0N0.0- m.c.o 440 0. 440.0 000.0- 00N.. 0m..o- N «4000.N 00.0 «0N..N «aNN.N- N4..- 4..0 00.0. N..o 4N.0. «a...N 4..0 50.. n.mm N0.nn an.. nN0.0 NMO.. n54.0- 440..- 0.0.0 4N0-o- mmo.0 m00.0- 0N4.N N4..o . 4.0.0 Ewdwwdfiu examsaw £05.20 0:20:02 Neg—00:0 2.2.5.3 EN. .00.. .303. .m NDCUQ >0...th >M. ...mmum-AHZ. ..000: a.uo0uuoo u ..50 <02\oag 1 >00 ”a .ocam «««n~.N 0N.0- «s0..N «a...N- so..- NN.0 4..0- 0..0 «4m... «04.. m0.o 000. 0.45 0N.am 000.. 0N..o- 000.. 0.5.0. m..... «No.0 040 0- o~o.0 0.0.0- N4N.o 000.0 m «««n~.N Nm.o. «as..N «am..N- N...- ...0 N..o- 4..o «an... «M4,. mn.o 000. 0.4m Nm.nm 400.. .4..0- 0.0.. 0.N.o. N.o.—- 0.0.0 N00.o. 0N0.o 0.0.0- 0.5.0 040.0 N «««.0.N 4N.0- «a...N «<0N.N. 0N... 0N.o o..o. 0..O «04... 4.0.. no.. 00.. N.ms mm.nn 0N... n...0- 0.0.. 005.0- Nmo..- NN0.0 m40.o- .No.0 0.0.0- 045.0 .00.0 . .2: .3: .aa: .0 .3530 225...: 0:25;... £522 22.22.... 5.... .20.. 22.30,... 3000 >032. >0. Hmuoxuwz. £0.50 x.uu0uuoo N -.;o omxooxaa. 1 >0. "4 .ocam ..a.ao .xm. -- met.e ..c. 0.000: zo.mmmmeum 0.0oa n.0 0.009 155 insignificant in all models. Variables for the dividend constraint measure, the public debt dummy, and firm size are insignificant in all models, regardless of whether LEV1 or LEV4 is used. Both auditor dummies are significant at less than the .05 level and in the hypothesized direction. The industry dummy for durable goods manufacturers is highly significant at less than the 0.01 level. However, the dummy for nondurable goods manufacturers is not significant at conventional levels even though it was significant in the univariate analysis. The high correlation between the two industry dummies (—0.68) and the large standard errors of NDURDUM indicates that multicollinearity may be causing NDURDUM to be insignificant in the multiple regression model. To investigate this possibility, the models also were estimated by dropping the industry dummies one at a time. When NDURDUM is dropped, DURDUM becomes more significant than before (t-statistics exceed 3.7 compared with about 2.8 when both variables are included in the model). However, the chi-square test statistic for the overall logit model and the percent of firms correctly classified remains largely unaffected. The joint significance of the two auditor dummies and the durables industry dummy indicates that both variables provide incremental explanatory power over each other and is further evidence of a lack of auditor—industry association in this sample. The chi-square statistic indicates that the null hypothesis that all coefficients (except the intercept) in the model are zero is rejected in each model at less than the .001 level. The proportion of firms correctly classified by the models varies from 74.8 to 75.5 percent compared with the proportional chance criterion, Cpm, of 156 61.7 percent.15 The percent correctly predicted by the model significantly exceeds Cm, at the .001 level.“ The likelihood ratio index is around .10 for the various models. Turning to the logit results for the full sample based on DISC-year data (see Table 6.4), the results generally remain similar to those based on 1971 data. However, when LEV1 is used as the definition of leverage, interest coverage is significant at less than the .01 level for all models (compared with significance at the .10 level using 1971 data) and leverage itself is now significant at .05 (instead of .10) in models 1 and 3. When LEV4 is used as the definition of leverage, interest coverage becomes significant at the .05 level for all models (compared with no significance using 1971 data) although leverage itself is now significant at .10 level compared with the .05 level earlier. Both auditor dummies are significant at the .05 level as before and DURDUM is significant at the .01 level for all models. The dividend constraint measure, public debt dummy, firm size, and the nondurable industry dummy remain insignificant as before. The chi-square statistic 15Note, under the most successful naive model, Cm, all firms would be classified as partial allocators (the larger of the two groups) and thus achieve a success rate of 74.1 percent. Although the classification success achieved by the logit models is marginally higher, it is not significantly different from CM. 16’1‘ he Z-statistic for proportions was used for the significance test and it ranges from 4.62 (when 74.8 percent is used as the correct classification rate) to 4.86 (when 75.5 percent is used as the correct classification rate). The test statistic transforms proportions into standard normal measures and is computed as follows: Z = (p - Ci) / ¢[C,(1 - C,)]/N where p is the observed correct classification rate and C, is the chance criterion employed as the benchmark. 157 .~aaos ufiwod ozu u: nu—u mo amozfloam: 93¢ «a :o_.:u_fi:~ :a mofi~>0ua 33m dotoa :o.mnuuwuu aunucnum use 3. uuuauanum ~x 0;“ a. nudaaam m. u. ..onq u:_auum:cu a:u have: zoo:.—ax_— mc— \ aucoauo>coo us =00;.~ax—— mo—V t __ mu vacuumv mu can xovcu ouuau noosunox.~ as“ m_ _&_o .a>a:a udaaa: uda may ~0>u~ do. uzu :uzu mmu— ua u:su_um:m_a m_ an .ouun out Auguuumu:_ 9:4 uaouxov auza.u.uuoou a;u ~.q unzu mamasuoax: -=c 0:9 mummy sous: umou 0.46» uoo:.~ox«~ ago :0 awman m. uuuuuuaum mum37m._zu use“ .h~o>_uouamou .nabu sea .Nth ._¢hu .mOuamuae uuau xmu o>uuumuuo ouuzu uzu cu vcoammuuou M van .m .‘ m~uaoza I .~m>0‘ Ac. use um ou:mu~u_:mmm muumqu:_ «ax .~m>o. no. as» an oucau«u_:m_n magnumfizu ea .~u>m_ o_. 9:“ on oucaouuq:m_m maamuqfism a .mugumuuaum-u u_uau:a>mu fiuueuuoama 9;; o:.~ fizoooa any 1cm mauc3_umo usumoumumou oz“ muuoaou ocu~ umuwu ozh .ocm I :- «««~a.~ o_.o. «noo.~ ««-.m- 33... no.9- on.o c~.o. ««n¢.~. «qo.~ mo.o o... c.n~ a¢.~m mm... ono.o- «an._ ~c~.o- cam... omc.o- m“..o en:.o. q_o.o- o~o.. ono.o » «««_~.~ o_.o. .«no.~ ««._.~. mo.o. ao.:. mm.o o~.o. «amm.d. «mo._ 00.: . o... m.n~ -.~n a~_.~ _no.o- mam.. mmo.o- mmm.c- .mo.c- mm_.o ano_o- q~o.o- coo.~ can.o w «««oa.~ ~_.o- «ano.m «amc.~- No... n~.c- mm.o m~.o. «amm._- «mo._ mm.— n~_. ~.n~ o~.mn an... asc.o. mom.~ «so.o- mq~.~. sac.o- o~_.o ”no.0- m_o.o- m_o._ o~q.~ ~ .~%_ =u.ufiwmu~o .wewzaw magnum 2::mzaz, w:::::¢ .zzcc:¢ aha 4cm; 2:c>m_m m>ooe >oeez~ >md Hmmummez_. “Huts: xnuuouuoo u ..50 cezxaea u >ua an docs“ «gaoa.~ mn.o- «anc.~ ««o~.~- no..- ~o.e. o~.o o«.o- «namq.~- «afio.— nq.~ a... m.om ~a.~n n.~.~ ~n_.o. .on.~ mqm.o- om_._. _~o o- «mo.o mmo.o. e~o.o- mom.o -m.~ m «««NN.N mm.c- ««~o.~ ««n~.~- o~.o. mo.c- o~.c “m.o- «xamq.~- *No._ ~m.~ o... ~.m~ o¢.~n m~¢.~ _n~.o- mqm._ _m~.o. m-.o. «Kc.o- «no.0 mmo.o- o.o.c- mom.o coo_~ N «a«m~.~ oq.o- «swo.~ «¢s_.~. afl._- ao.o- oo.o _q.o- *«*~q.~- «aeo.~ mm._ n_.. o.on co.an ~__.~ ~m_.o. .nn.~ c—~.o- nqo.m- was.o- «No.c Nmo.o- o_o.o- mom.o m~o.~ A ._w4 :nuuuawadu .uuazaw u:cm:g 2:2zzcz Nz::=:¢ .::cc:< mew. ..«mfl ::c>m4u ~>ccm >oub2m >m~ HmMUMmhzn edges: >~uoouuou u -‘so om:OU\aaa n >ua n< .mcsm .Aauan nau>.om_a -- «Sham —~mqm Aann.ov Aacn.ov Amen.ov ANmn.ov no.0 Ho.ou mmm.o Nmm.o Nma nN.o mc.o- mmn.o w-.o nma m v 0 N >ouo nuao.NNv Anm~.~mv Acmo.m~v AnNm.va «eemo.m s:«s¢.~ om~.HH Non.cN mma c;cnm.~ g¢m~.a «mm.oa «Hm.mH mmH m A U >OUHzH Amnfl.ov Anaq.ov Amma.ov Amca.ov ¢:H5.au ccmm.H- mmc.o nmm.o mma e«nw.Hn «:oo.~- nmc.o m~m.o mmH m v U c >mq Amoq.ov ANm~.ov Amoq.ov ANm~.ov *Hq.au «em.au nwn.o mm~.o mmH cen.au ¢¢AA.H- mmm.o mmN.o mmfl m v U a >mg N u m o z N a m U z mfimonuonmm wanmfium> oncooz smoZan accououuuo A.>on .vumv undo: Cockpon mucoumuufio A.>oo .cumv uou oaumfiunuw now: now ufiumduopm coo: nuanucoVflpmHn cumD,H~mH sane Axxxm 0coo onv «manmuaa waausuoamaam: n< Assam ZOHHHZD m.o macaw 2L6]. HnHm.ov H¢m~.ov Hmmm.ov qu~.ov mH.o- H~.o- moH.o Hmo.o an o~.o- m~.o- sHH.o Hmo.o An A v u N zsooa< HHH¢.ov HHnm.ov Hooq.ov HHnn.ov «A.o- nH.o- mH~.o cmH.o mm oo.o- Hm.o- oo~.o mmH.o AA A A o H zoooa< HmmH.ov Humo.ov ANOH.OV Hmmo.ov A.m~.~ .An¢.~ mmm.o ona.o mm .A.HA.~ ..mn.~ AH¢.o nw¢.o an N n mHm HomH.ov Hemo.oc HmHH.ov HnNH.oV No.0 no.o ~mm.o «Hq.o on mo.o m~.o 0mm.o Amm.o an A N «Hm Hnmo.ov HNno.ov Ammo.ov HNHH.oV A.mm.~ A.H~.~ sH¢.o mn¢.o an *.A~.~ HH.o qu.o onq.o An N H mam HHHn.HV H0q~.HV Han.Hc Hmo~.Hc OH.o mH.o om~.n eqm.n an on.o- nm.o- ~A~.n AMH.n An m A u HmHm HmoA.oV Humm.ov qum.ov Hqu.oV H¢.H o¢.o- Hmm.o NHn.o «A mm.o ~¢.o- n0q.o «Hm.o mm m v u N >009 HoHo.mV Hoqm.¢~U Hns~.~Hv Hmnn.H~V No.0 no.H «sm.m oHH.~H on Nm.H- mm.o Hmo.w nmm.m An A A u >OUHzH quH.cV HnsH.oV HHnH.ov quH.ov ~¢.o- Hm.o- mnq.o A~¢.o mm 00.0- on.o- mmq.o AHs.o An m v u s >mH HoH¢.oc HHo~.oV Hmoq.ov HHw~.ov Nm.o- .An.H- omn.o nmm.o on H¢.o- mm.o- mm¢.o Hoq.o An m v u H >mH N a m o z N a m o z mHnosuoaAm onmmua> guano: ~30:qu oucoucuuwo “(So .vumv 2.302 cooxuon nonououuao A.>ao 6an now oHunHuaum coo: uou oHuuHuaum ago: some uan-WMHo mama HAmH haze Axxxm ovoo OHmv moanmunvcoz wcwuauoumacoz Hm Hocmm H.e.ucoov m.o mHnme 162 .stompo cam mew you ammo HHmu-osu ”ammo HHmu-mcov Hm>mH Ho. and um osmonchHm AAA .Azpomsoz van mew you ammo HHmu-oAu ”ammo HHmu-msov Hm>mH mo. ago um unmonchHm AA .Amem wow wow» meu-03u ”umou meu-mcov Ho>oH 0H. mnu um ucmowmacwwm A .umou Edm-xcmu coxooHH3 mfiu mo coHumEonumam Hmauoc onu co bmmmn ma owumwumum -N m£H .Ho>ma no.0 mzu um Umuomnmu on :mo manmwum> omeommm w you museum ozu Cmmzumn mmocmwum> Hmsvm mo mwmonuoaxn ocu umLu mmumowvcw ummu-m cm MM mmocmwum> CH mmoCmummmHv How Umumsmum ow owumwumum-u make .uoumooaam Hmfluuma n m ”Houmooaam o>HmCm£ouano t on .mcoHuHcHuoo mHanum> How m.m mHan mam. A.e.ucoov m.o mHan 163 year data is used.18 In addition, of the manufacturing durable firms audited byArthur Andersen, a significantly greater proportion used comprehensive allocation than partial allocation. In contrast, all of the manufacturing durable firms audited by Price Waterhouse followed partial allocation. These differences are in the hypothesized direction and statistically significant at the .05 level or better. Further, consistent with the tax-based political cost hypothesis, comprehensive allocators have a higher effective tax rate than partial allocators but the statistical significance of these differences is mixed.19 Partial allocators also are found to pay out a greater proportion of their inventory of payable funds in dividends and more often have publicly-placed debt than comprehensive allocators. Although these differences are consistent with the covenant-based and renegotiation cost hypotheses, they are not statistically significant. Finally, contrary to expectations, comprehensive allocators are smaller than partial allocators although the mean difference in LSAL for the two groups is not statistically significant. As panel B of Table 6.5 shows, the results for the manufacturing nondurable firms are generally consistent with the debt covenant-based, tax—based political cost, and the auditor explanations. However, with the exception of the effective tax rate measures which are generally statistically significant, the results are not as strong 18LEV1 (LEV4) is generally significant at the .10 (.05) level, whereas INTCOV is generally significant at the .01 level. 19Based on 1971 data, the difference between the mean ETR1 (ETR2) of the two groups is significant at the .10 (.05) level, but ETR3 is not significantly different for the two groups. Based on DISC-year data, none of the three ETR measures are significantly different for the two groups of firms. 164 as those for the manufacturing durable firms. A potential explanation for the weak results for the manufacturing nondurables could be the small sample size (n = 63) of firms in this industry class.” Overall, the industry analysis suggests that the observed associations between DISC accounting method choice and the contracting, political cost and auditor explanations are fairly robust. 6.4 Sample Partitioned by DISC-Year Because the sample firms’ DISC accounting method choice is not independent of the year in which the DISC was operational (‘DISC-year’--see chapter 5, section 5.4.4), the analysis was repeated separately for each of the three DISC-years (1972, 1973, and 1974) included in this study. The results are presented in Table 6.6. In general, the results for the three years individually are similar to the results for the entire sample and indicate support for the debt covenant, political cost, and auditor influence hypotheses regarding firms’ choice of accounting method for DISCs. The variations in the results across the three years are described below. The DISC-years 1972 and 1973 are generally comparable with two exceptions. First, consistent with the hypothesis, comprehensive allocators have a higher INTCOV ratio than partial allocators in both 1972 and 1973 and the ”The number of firms on which the univariate statistics (Table 6.5, panel B) are based varies from 53 to 59 instead of 63 because of missing data for some of the variables. 165 Table 6.6 UNIVARIATE STATISTICS — BY DISC-YEAR Panel A: 1972 DISC-Year Firms Only Mean Statistic for (Std. Dev.) Difference between Means Variable Hypothesis C P t Z LEV 1 C < P 0.309 0.397 —1.01 -1.24 (0.331) (0.344) LEV 4 C < P 0.363 0.444 -2.10** -1.87** (0 164) (0.151) INTCOV C > P 25.606 8.507 2.00** 2.66*** (35.638) (16.800) DCOV 2 C < P 0.419 0.303 0.61 1.07 (0.751) (0.740) PLEVDUM C < P 0.278 0.345 -0.56 —0.55 (0.461) (0.477) LSAL C > P 4.664 4.722 —0.15 -0.19 (1.399) (1.586) ETR 1 ? 0.472 0.435 2.35** 1.73* (0.041) (0.136) ETR 2 ? 0.415 0.380 0.76 1.53 (0.139) (0.187) ETR 3 ? 0.443 0.418 0.60 1.58 (0.124) (0.173) AUDDUM 1 C > P 0.278 0.216 0.59 0.58 (0.461) (0.413) AUDDUM 2 C < P 0.000 0.147 -1.75** —1.73** (0.000) (0.355) NDURDUM ? 0.278 0.216 0.59 0.58 (0 461) (0 413) DURDUM ? 0.667 0.655 0.09 0.09 (0.485) (0.477) 166 Table 6.6 (cont'd.) Panel B: 1973 DISC-Year Firms Only Mean Statistic for (Std. Dev) Difference between Means yariable Hypothesis C P t Z LEV 1 C < 0.333 0.464 —1.86** —0.99 (0.276) (0.498) LEV 4 C < 0.405 0.467 —2.13** —l.79>\’7‘r (0.130) (0.182) INTCOV C > 14.281 11.731 0.56 1.08 (22.520) (23.480) DCOV 2 C < 0.308 0.429 —0 72 0.59 (0.729) (0.889) PLEVDUM C < 0.289 0.325 —-O.39 -O.39 (0.460) (0.471) LSAL C > 4.685 4.751 —0 23 —0.12 (1.295) (1.551) ETR 1 ? 0.474 0.449 1.69* 1.85* (0.077) (0.072) ETR 2 ? 0.458 0.419 1.39 1.30 (0.128) (0.154) ETR 3 ? 0.469 0.433 2.18** 2.12** (0.070) (0.108) AUDDUM 1 C > 0.421 0.145 3.07*** 3.33*** (0.500) (0.354) AUDDUM 2 C < 0.026 0.084 —1.44* —1 18 (0.162) (0.280) NDURDUM ? 0.237 0.145 1.24 1.23 (0.431) (0.354) DURDUM ? 0.553 0.735 -2.01** —1.98** (0.504) (0.444) 167 Table 6.6 (cont'd.) Panel C: 1974 DISC-Year Firms Only Mean Statistic for (Std. Dev.) Difference between Means Variable Hypothesis C P t Z LEV l C < P 0.356 0.467 —1.18 —1.17 (0.306) (0.336) LEV 4 C < P 0.455 0 469 —0.37 -0.59 (0.143) (0.111) INTCOV C > P 15.676 8.422 1.08 0.48 (29.025) (11.794) DCOV 2 C < P 0.112 0.314 —1.52* -0.62 (0.143) (0.644) PLEVDUM C < P 0.381 0.346 0.24 0.23 (0.498) (0.485) LSAL C > P 5.382 5.317 0.15 0.37 (1.342) (1.633) ETR 1 ? 0 435 0.455 —0 77 —0 89 (0.060) (0.109) ETR 2 ? 0.361 0 395 -0.74 —1.12 (0.147) (0.158) ETR 3 ? 0.384 0.432 —1.17 -1.32 (0.144) (0.140) AUDDUM 1 C > P 0.143 0.192 —0 44 —0.43 (0.359) (0.402) AUDDUM 2 C < P 0.048 0.115 -0.85 —0.80 (0.218) (0.326) NDURDUM ? 0.381 0.000 3.91*** 3.40*** (0 498) (0.000) DURDUM ? 0.333 0.808 —3.68*** -3.25*** (0.483) (0.402) 168 Table 6.6 (cont'd.) 8See Table 5.8 for variable definitions. bC — comprehensive allocator; P = partial allocator. cThe t-statistic is adjusted for differences in variances if an F- test indicates that the hypothesis of equal variances between two groups for a specific variable can be rejected at the 0.05 level. The 2- statistic is based on the normal approximation of the Wilcoxon rank-sum test. * Significant at the .10 level (one-tail test; two-tail test for ETR). ** Significant at the .05 level (one-tail test; two-tail test for ETR and NDURDUM). Significant at the .01 level (one-tail test; two-tail test for ETR and DURDUM). >r >1- x- 169 difference between the two groups is statistically significant at the .05 level or better in 1972 but is not significant in 1973. Second, consistent with the tax-based political cost hypothesis, comprehensive allocators have higher effective tax rates than partial allocators in both 1972 and 1973, regardless of the effective tax rate measure used. However, whereas ETR3 is statistically significant in 1973, it is not significant in 1972. Third, consistent with the auditor influence hypothesis, a greater proportion of Arthur Andersen’s clients are comprehensive allocators than partial allocators. This difference is statistically significant at the .01 level in l973 but is not significant in 1972. Finally, whereas the dummy for manufacturing durables is significant in 1973, it is not significant in 1972. The results for the DISC-year 1974 are similar to the other years for the debt covenant-based hypotheses but not statistically significant. The results for the political cost hypotheses are in the opposite direction-~comprehensive allocators are observed to be larger and have a lower effective tax rate than the partial allocators. However, the differences are not statistically significant. Chapter Seven SUMMARY AND CONCLUSIONS In this chapter, the purpose of the study, the research hypotheses examined, the data used for the purpose, and the results of the empirical analyses are briefly summarized. Next, suggestions for future research are made. 7.1 Summary of the Study and the Results The purpose of this study was to examine firms’ motivations for choosing between comprehensive and partial allocation--alternative methods for the extent of interperiod tax allocation. In the absence of explicit cash flow effects, firms’ accounting method choice was hypothesized to be related to variables that surrogate for potential debt covenant violations and their political visibility, and the preferences of their auditors. These hypotheses (except the auditor preference hypothesis) were derived from the predictions of positive accounting theory that is based on contracting cost and political cost arguments. Because accounting numbers are used in debt covenants and the political process, these arguments suggest that the choice of accounting methods have economic consequences for firms and their managers and, hence, firms are not indifferent to their accounting procedure choice. Domestic international sales corporations (DISC), a special type of corporation legislated into the Internal Revenue Code in 1971 to stimulate exports, 170 171 provided a unique data set to empirically examine the determinants of the extent of tax allocation choice. The legislation allowed an indefinite deferral of income taxes on a portion of the DISC’s export earnings. However, for financial reporting purposes, some firms provided taxes on those earnings (comprehensive allocators), whereas others did not (partial allocators). The empirical analysis was conducted on a sample of 320 firms that had a DISC operational in 1972, 1973, or 1974. Statistical tests were conducted in both univariate and multiple regression frameworks. Overall, the results indicate support for the debt covenant and auditor preference hypotheses but only partial support for the political cost hypotheses. Consistent with the theory, firms with higher leverage and lower interest coverage ratios are observed to adopt partial allocation, an income-increasing accounting choice. However, a similar hypothesis based on dividend restrictions is not supported. In addition, no support is found for the hypothesis that because public debt involves higher renegotiation costs, firms with public debt (or a greater proportion of public debt in their capital structure) are more likely to use an income-increasing alternative. Also consistent with the theory, firms with higher political costs manifest through higher effective tax rates are observed to adopt comprehensive allocation, an income-decreasing alternative. However, support for this hypothesis is weak. Another political cost hypothesis based on firm size is not supported. Finally, strong support is found for the hypothesis that firms’ accounting method choice is related to the preferences of their auditors, a relatively new finding in the accounting method choice literature. 172 The results generally were consistent across different statistical tests, alternative definitions of the explanatory variables, and sub-samples of firms by industry classification and the year when the DISC became operational. 7.2 Suggestions for Future Research 7.2.1 The DISC tax forgiveness An extremely important aspect of the DISC legislation was the permanent forgiveness in 1984 of the indefinitely deferred DISC taxes. The revenue loss to the Treasury as a result of the forgiveness was estimated to be between $10 billion and $13 billion and was characterized by the financial press as a ‘windfall’ for US. exporters. It should be noted that the tax forgiveness benefitted all firms with indefinitely deferred DISC earnings, regardless of whether they provided taxes on those earnings in their financial statements. However, comprehensive allocators got a one-time earnings boost from the reversal of those taxes now permanently exempt. An extension of this research would be to examine the stock market reaction to the DISC tax forgiveness. This examination is potentially interesting for two reasons. First, it would provide an insight into the market’s perception of deferred taxes to help resolve an old debate: are deferred taxes a liability, stockholders’ equity or something else? Even though under SFAS 96 the presumption would be that deferred taxes are a liability, it is an empirical question because we do not know that they will be perceived as such just because they are labelled a liability 173 under GAAP. Second, the comparison of the market reactions of comprehensive and partial allocators to the tax forgiveness can provide additional insight into the functional fixation hypothesis that appears to have gained renewed interest (e.g., Hand [1990]). 7.2.2 Accounting Method Changes and Auditor Switches This study was concerned with firms’ motivations for adopting an accounting method in the presence of alternatives. A strong influence of auditor preference was found to exist. A further examination of this relationship would be to find out whether firms with DISCs subsequently changed accounting methods or switched auditors in order to align the accounting method followed by them with the stated preferences of their auditors. 7.2.3 Early v. Late Adopters of DISCs The data in this study shows that firms that could benefit from the DISC legislation did not set up a DISC or make it operational as soon as that option became available. Several questions arise in this regard. First, why did some firms set up DISCs early and others late? Second, were the early adopters somehow different than the late adopters? APPENDICES Appendix A EXTENT OF TAX ALLOCATION: The Conceptual Merits of the Alternative Approaches Tax allocation is really a two-part controversy: 1) whether or not to allocate, and 2) how much to allocate if allocation is agreed upon. The first part of the controversy is labeled the choice between the flow- through approach (no allocation) and interperiod tax allocation. The second part of the controversy deals with the two approaches to interperiod tax allocation--comprehensive allocation (considers all book- tax differences) and partial allocation (considers some, not all, book- tax differences). For purposes of the fellowing discussion, the two parts of the controversy are collapsed into one with flow-through, comprehensive, and partial allocation being viewed as three alternatives to the extent of tax allocation question. Adl three approaches have strong theoretical merits. Detailed arguments are contained in Beresford et al., [1983] from which the following discussion draws heavily. A.1 Flow-through Method The flow-through method eliminates deferred taxes entirely by not requiring interperiod tax allocation for any temporary difference. Under flow-through, reported tax expense for the period equals taxes currently payable or refundable as determined by the tax return. However, supporters of this method would require disclosure of the differences in 174 175 Appendix A (cont'd.) the tax bases of assets and liabilities and the amounts at which they appear in the balance sheet [Rosenfield and Dent, 1983]. At the core of the arguments supporting this method is the view that the income tax expense shown on the income statement is functionally related only to taxable income and not accounting income because this expense arises only when, and if, an entity has taxable income. In addition, this expense arises from the aggregation of all taxable income and tax deductible items. However, “[i]nterperiod tax allocation necessarily assumes the taxation of individual transactions and events, even though no transaction or event by itself gives rise to tax assessment” [Beresford et al., 1983, p. 20]. Supporters of flow—through also find simplicity and accuracy to be its major appeal [McGoldrick, 1984]. They argue that investors would not have to guess a company's earnings and tax expense for a given period. Further, footnote disclosure of relevant information regarding the future (e.g., temporary differences and changes in the tax laws and regulations) would enable investors to assess future cash flows. A major flaw with the flow-through approach, as Defliese [1983, p. 95] points out, is that it essentially results in cash-basis accounting and “[t]o revert to cash basis accounting would be a retrogression and a complete abdication of financial reporting objectives by the profession.” Second, the tax consequences of certain transactions (e.g., installment sales) are clearly identifiable and measurable, and the difference in the timing, of the tax consequences of" those transactions for financial reporting and taxation is generally of short duration. Hence, in these 176 Appendix A (cont'd.) cases, interperiod tax allocation may provide a better indication of future cash flows and. the resulting deferred tax account could ‘be meaningfully interpreted. For these reasons, the official position in the U.S., starting with ARB 23 issued in 1944, has required interperiod tax allocation. The theoretical merits and demerits of the two alternative approaches to interperiod tax allocation--comprehensive or partial-~are discussed next. A.2 Comprehensive Allocation Under the comprehensive approach, interperiod tax allocation must be applied to all temporary differences. This method is the complete opposite of flow-through. Both APB 11 and SFAS 96 require comprehensive allocation with the exception for transactions satisfying the indefinite reversal criteria under APB 23. The underlying theory for this allocation procedure is that income tax expense recognized in a given period for financial reporting purposes should be based on current or past transactions. Hence, offsetting temporary differences contingent on the happening of some future event (e.g., future asset purchases that may help offset any reversing difference) are ignored. The rationale for ignoring recurring temporary differences is not because the tax effects of those differences cannot be measured, but because the tax effects of each transaction or event should be recognized separately. Supporters of comprehensive allocation question the economic assumptions on which partial allocation is based. They contend that 177 Appendix A (cont'd.) partial allocation assumes continued economic prosperity and growth. It also assumes that repeal of favorable tax regulations that created the differences in the first place will not occur and other contingencies will not arise. These assumptions may not be met. In any case, they are subjective. Proponents of comprehensive allocation also argue that deferred taxes arising from temporary differences are similar to other balance sheet accounts (e.g., accounts payable) that roll over. Since there is no debate whether or not accounts payable should be recorded on the balance sheet, then why the disagreement over deferred taxes? Other arguments advanced to support comprehensive allocation include: 1) it results in more reliable information and consistent accounting by eliminating the need for judgement, thereby improving comparability of financial reports, 2) it conforms with the principle of matching revenues and expenses more thoroughly and consistently, 3) it reports more fairLy a company's total capitalization because deferred taxes are viewed as a source of capital, and 4) it ‘softens' otherwise inflated profits caused by the use of the historical cost convention. A.3 Partial Allocation Partial allocation is an intermediate position between flow through and. comprehensive allocation. Under this approach, interperiod tax allocation applies to some temporary differences but not others. An important issue with partial allocation is to identify which temporary differences require interperiod allocation. _ ,—-— 178 Appendix A (cont'd.) Proponents of this approach would allocate taxes on those differences that are reasonably expected to result in income tax payments within a relatively short time period, say five years (APB 11, para. 26). They would exclude temporary differences that give rise to an indefinite postponement of taxes or continuing tax reductions since these transactions involve contingencies that are remote at best and mere disclosure is sufficient. Advocates of partial allocation also suggest that deferred taxes are quite different from other roll-over accounts (e.g., Nair and Weygandt [1981]). For example, accounts payable balances roll over, but individual items within that account are affected by specific transactions--goods are received, new obligations incurred, and cash paid. In contrast, deferred taxes do not involve any transactions--no cash is paid, the amounts are not owed to anyone, and there is no specific date on which they will become payable. Deferred taxes are perpetual interest-free loans owed to no one. Other arguments advanced against comprehensive allocation and in support: of ‘partial allocation. include: 1) comprehensive allocation contradicts economic reality as witnessed by the unrealistic magnitude of deferred tax balances on corporate balance sheets, and 2) comprehensive allocation results in misleading comparisons between companies because it conceals the economic reality of tax deferral that varies significantly among companies. 179 Appendix A (cont'd.) A.4 Summary Although theoretically the flow—through approaCh is an available alternative to interperiod tax allocation, realistically it appears that some allocation will be required. Hence, the choice is reduced to the two alternatives under interperiod tax allocation—-comprehensive and partial allocation” .Both have strong conceptual merits” Comprehensive allocation stresses internal consistency and adherence to time accrual accounting model by focusing on individual transactions. Partial allocation stresses economic reality and representational faithfulness by focusing on the nature and aggregate amounts of deferred taxes. Appendix B SAMPLE FIRM INFORMATION SIC Exch- COMPU- Audi- DISC Acct Firm name (old name)‘ codeb ange° STATd CUSP‘ torf Yrg Methh AAR CORP 5080 NYSE CI 000361 AA 73 P AVC CORP 3714 AMEX CR 002280 PW 73 P ACME-CLEVELAND CORP 3541 NYSE CI 004626 EW 73 P AFFILIATED HOSPITAL PRDS 3069 AMEX CR 008230 EW 73 P AIKEN INDUSTRIES INC 3664 AMEX CR 008788 DH 73 C AIRPAX ELECTRONICS INC 3679 AMEX CR 009446 CL 73 P AJAX MAGNETHERMIC CORP 3560 AMEX CR 009644 DH 72 P ALCON LABORATORIES INC 2830 NYSE CR 013896 AA 72 P ALLEN GROUP 3714 NYSE CI 017634 CL 73 P ALLERGAN PHARMACEUTICALS INC 2830 NYSE CR 018492 CL 72 P ALLIED PRODUCTS 3523 NYSE CI 019411 AA 73 C AMERACE CORP 3041 NYSE CR 023519 EW 72 C AMERICAN BRANDS INC-DEL 2111 NYSE CI 024703 CL 72 P AMERICAN HOIST & DERRICK CO 5070 NYSE CI 026573 TR 72 P AMERICAN HOSPITAL SUPPLY 3841 NYSE CR 026681 TR 72 P AMETEK INC 3621 NYSE CI 031105 AY 73 C ANALOG DEVICES 3674 NYSE CI 032654 AY 72 P ANDERSON, GREENWOOD & CO 3494 NYSE CR 033849 AA 72 P ANGELICA CORP 7200 NYSE CI 034663 AA 73 C ANCLO ENERGY INC 1381 AMEX CI 035053 PW 74 P (Anglo Co Ltd) ANSUL CO 3560 NYSE CR 036627 PW 72 P ARO CORP 3540 NYSE CR 042627 AY 73 P ARROW ELECTRONICS INC 5065 NYSE CI 042735 OR 72 P AUGAT INC 3679 NYSE CI 051042 DH 72 P AUTOMATED BLDG COMPONENTS 3490 AMEX CR 052807 DH 74 C BAV LIQUIDATING CORP 1211 NYSE CR 055900 PM 73 P (Bates Mfg Co) BALLY MFG CORP 7990 NYSE CI 058732 AY 72 P BANNER INDUSTRIES INC 3560 NYSE CI 066545 AA 73 P BARD (C.R.) INC 3841 NYSE CI 067383 AA 72 P BECOR WESTERN INC 3530 NYSE CR 075873 TR 72 P (Bucyrus-Erie Co) BEECH AIRCRAFT CORP 3721 NYSE CR 076635 EW 72 C BEKER INDUSTRIES 2870 NYSE CR 077266 AA 72 P BELL & HOWELL CO 3688 NYSE CI 077851 AA 72 C BERTEA CORP 3728 AMEX CR 085815 AY 72 P BETZ LABORATORIES INC 2890 OTC CI 087779 EW 73 C BOEING CO 3721 NYSE CI 097023 TR 72 P BROOKS & PERKINS INC 3443 AMEX CR 114331 EW 72 P BROWN & SHARPE MFG CO 3540 NYSE CI 115223 CL 74 C BROWNING-FERRIS INDS 4953 NYSE CI 115885 AA 74 P BRUNSWICK CORP 3510 NYSE CI 117043 AA 72 C 180 181 Appendix B (cont'd.) SIC Exch- COMPU- Audi- DISC Acct Firm name (old name). codeb angec STATd CUSP‘ tor' Yr° Methh CRS SIRRINE INC 8911 NYSE CI 126270 TR 73 P (CRS Design Associates Inc) CAMERON IRON WORKS 3533 NYSE CI 133429 AA 72 P CAMPBELL INDS 3730 AMEX CR 134357 PM 73 P CARLISLE COS INC 3069 NYSE CI 142339 PM 74 P (Carlisle Corp) CARPENTER TECHNOLOGY 3312 NYSE CI 144285 CL 73 P CARRIER CORP 3580 NYSE CR 144465 AA 72 P CATERPILLAR INC 3531 NYSE CI 149123 PW 72 P (Caterpillar Tractor Co) CENTRAL SOYA CO 2048 NYSE CR 155177 AY 73 C CERTIFIED CORP 3990 AMEX CR 156897 AY 72 P CETEC CORP 3651 AMEX CR 157186 AA 73 P (Computer Equipment Corp) CHESAPEAKE CORP 2621 NYSE CI 165159 CL 72 C (Chesapeake Corp of Virginia) CHROMALLOY AMERICAN CORP 3470 NYSE CR 171106 PM 73 C CINCINNATI MILACRON INC 3541 NYSE CI 172172 EW 72 P CLARK EQUIPMENT CO 3537 NYSE CI 181396 PW 72 P COFFEE-MAT CORP 3580 AMEX CR 192180 AY 73 P COHU INC 3825 AMEX CI 192576 AY 74 P COLECO INDS 3942 NYSE CI 193378 PW 72 P COLEMAN CO INC 3949 NYSE CI 193558 EW 72 P COLT INDUSTRIES INC-DEL 3728 NYSE CI 196864 AA 73 P COMMERCIAL METALS CO 5051 NYSE CI 201723 TR 73 C COMMERCIAL SOLVENTS CORP 2890 NYSE CR 202381 AY 72 P COMPO INDS 2200 AMEX CR 204525 TR 72 C COMPUGRAPHIC CORP 3555 NYSE CI 204900 AA 72 P CONCORD FABRICS INC 2200 AMEX CI 206219 OR 72 P CONDEC CORP-OLD 3711 AMEX CR 206741 AY 73 P CONGOLEUM CORP 3079 NYSE CR 207192 AA 72 P (Bath Industries Inc) CONSYNE CORP 3843 AMEX CR 210729 PW 72 P CONTROL DATA CORP 3680 NYSE CI 212363 PM 72 P COOK INTERNATIONAL INC 7399 AMEX CR 216174 PM 73 P (Cook Industries Inc) CORENCO CORP 2070 AMEX CR 218687 AY 72 P CORNELIUS CO 3580 AMEX CR 219093 EW 72 P CORNING GLASS WORKS 3220 NYSE CI 219327 PW 72 P CRAIG CORP 5411 NYSE CI 224174 PM 73 P CROSS (A.T.) & CO —CL A 3950 AMEX CI 227478 EW 72 P CUMMINS ENGINE 3510 NYSE CI 231021 AA 72 P CURTIS NOLL CORP 5012 NYSE CR 231507 EW 72 C CURTISS-WRIGHT CORP 3390 NYSE CI 231561 CL 72 P CUTTER LABORATORIES INC-CL A 3841 AMEX CR 232219 PW 72 P DAN RIVER INC 2200 NYSE CR 235773 PM 74 C DANIEL INDUSTRIES 3823 NYSE CI 236235 PW 72 P 182 Appendix B (cont'd.) SIC Exch- COMPU- Audi- DISC Acct Firm name (oldgname)‘ codebygangec STATd CUSP‘ tor' Yr° Methh DATAPRODUCTS CORP 3688 AMEX CI 238107 AY 73 P DEN-TAL-EZ INC 3843 AMEX CR 248209 OR 72 P DENTSPLY INTERNATIONAL INC 3843 NYSE CR 249028 PW 73 P DEXTER CORP 2800 NYSE CI 252165 CL 72 C DICK (A.B.) CO 3570 NYSE CR 253034 AA 72 P DIGITAL EQUIPMENT 3680 NYSE CI 253849 CL 74 C DISNEY (WALT) COMPANY 7990 NYSE CI 254687 PW 73 P DISSTON INC 3420 NYSE CR 254700 AA 73 P DONALDSON CO INC 3564 NYSE CI 257651 EW 73 P DOVER CORP 3530 NYSE CI 260003 PM 74 C DRESSER INDUSTRIES INC 3530 NYSE CI 261597 AA 72 P DYNEER CORP 3714 AMEX CR 268163 EW 74 C (Aspro Inc) EECO INC 3689 AMEX CI 268420 AY 72 P (Electronic Engg Co of CA) EG&G INC 8911 NYSE CI 268457 AA 73 C E-SYSTEMS INC 3664 NYSE CI 269157 EW 73 P EAGLE-PICHER INDS 3714 NYSE CI 269803 PM 73 P EDO CORP 3664 NYSE CI 281347 PM 73 P EGAN MACHINERY CO 3550 AMEX CR 282255 CL 73 P ELCO CORP 3679 AMEX CR 284371 OR 73 P EMERSON ELECTRIC CO 3621 NYSE CI 291011 PM 74 P ENGELHARD CORP 3330 NYSE CI 292845 PM 72 C (Engelhard Mineral & Chem) ENSTAR CORP-DEL 1311 NYSE CR 293582 PM 72 P (Alaska Interstate Co) ENVIROTECH CORP 3558 NYSE CR 294098 DH 73 P EX-CELL-O CORP 3714 NYSE CR 300587 PM 73 C FMC CORP 2800 NYSE CI 302491 PM 72 P FEDERAL-MOGUL CORP 3714 NYSE CI 313549 EW 74 P FEDERAL SIGNAL CORP 3711 NYSE CI 313855 AA 73 P (Federal Sign & Signal Corp) FIELDCREST CANNON 2211 NYSE CI 316549 AY 73 P (Fieldcrest Mills Inc) FLUKE (JOHN) MFG CO 3825 AMEX CI 343856 OR 73 C FOSTER WHEELER CORP 1600 NYSE CI 350244 OR 72 P FOXBORO CO 3823 NYSE CI 351604 TR 72 P GAF CORP 2860 NYSE CI 361428 DH 72 P GCA CORP 3550 NYSE CI 361556 AY 73 C GALAXY CARPET MILLS 2272 AMEX CI 363171 DH 73 P GARDNER-DENVER CO 3560 NYSE CR 365550 AA 73 C GEARHART INDUSTRIES INC 1389 NYSE CI 368298 OR 74 P (Gearhart-Owen Inds Inc) GENERAL DYNAMICS CORP 3721 NYSE CI 369550 AA 73 P GENERAL ELECTRIC CO 3600 NYSE CI 369604 PM 74 P GEORGIA-PACIFIC CORP 2400 NYSE CI 373298 AA 72 C Appendix B (cont'd.) 183 SIC Exch- COMPU- Audi- DISC Acct Firm name (old name)' codg“ angec STAT“ CUSP‘ tor' Yrg Methh GLASROCK MEDICAL SERVICES 7394 AMEX CR 377118 AA 73 C (Glasrock Products) GLEASON CORP 3541 NYSE CI 377339 EW 72 P (Gleason Works) GLOUCESTER ENGINEERING INC 3550 AMEX CR 379702 OR 73 P GORMAN-RUPP C0 3561 AMEX CI 383082 EW 72 C GRANITEVILLE CO 2200 NYSE CR 387478 OR 73 C GREAT LAKES CHEMICAL CORP 2800 NYSE CI 390568 EW 72 P GREAT NORTHERN NEKOOSA CORP 2621 NYSE CI 391090 AA 72 P GROLIER INC 2731 NYSE CI 398784 PM 72 P GROW GROUP INC 2851 NYSE CI 399820 EW 73 P (Grow Chemical Corp) GUARDIAN INDUSTRIES 3211 NYSE CR 401370 CL 73 P GULTON INDUSTRIES INC 3651 NYSE CR 402784 OR 73 P HALLIBURTON CO 1389 NYSE CI 406216 AA 74 C HAMMERMILL PAPER C0 5110 NYSE CR 408306 PW 73 P HAMMOND CORP 3931 NYSE CR 408360 CL 73 P HARNISCHFEGER INDUSTRIES INC 3550 NYSE CI 413345 PW 72 P HARRIS CORP 3663 NYSE CI 413875 EW 72 P (Harris-Intertype Corp) HAZELTINE CORP 3663 NYSE CR 421596 OR 72 P HEWLETT-PACKARD CO 3680 NYSE CI 428236 OR 72 C HEXCEL CORP 3460 NYSE CI 428290 AA 73 P HIPOTRONICS INC 3825 AMEX CI 433515 PM 73 P HOFFMAN ELECTRONICS CORP 3664 NYSE CR 434434 AA 72 C HOMESTAKE MINING 1040 NYSE CI 437614 DH 73 C HOST INTERNATIONAL INC 5812 NYSE CR 441074 DH 73 C HUGHES TOOL CO 3533 NYSE CR 444492 DH 73 P HUNT (PHILIP A.) CHEMICAL 3861 NYSE CR 445582 OR 72 P I-T-E IMPERIAL CORP 3610 NYSE CR 450420 AA 72 P ILLINOIS TOOL WORKS 3714 NYSE CI 452308 AA 73 C INMONT CORP 2890 NYSE CR 457641 PW 73 C INOLEX CORP 2830 AMEX CR 457648 CL 73 P (Wilson Pharmaceutical & Chem) INSTRON CORP 3829 AMEX CI 457776 AY 72 P INTL FOODSERVICE CORP 5140 AMEX CR 459528 OR 72 P INTL PAPER CO 2631 NYSE CI 460146 AA 74 C IONICS INC 2086 AMEX CI 462218 TR 72 P ITEK CORP 3550 NYSE CR 465632 AA 72 P JEANNETTE CORP 3231 AMEX CR 472214 CL 74 P JOHNSTON IND-DEL 2221 NYSE CI 479368 AA 72 P (Geon Industries Inc) JOY MFG CO 3530 NYSE CR 481196 PW 72 P KANE-MILLER CORP 2010 NYSE CR 484098 OR 72 P KAWECKI BERYLCO INDS INC 1090 NYSE CR 486386 DH 73 C KING RADIO CORP 3664 AMEX CR 495620 AA 72 C KIRSCH CO 2510 NYSE CR 497656 AA 72 P 184 Appendix B (cont'd.) SIC Exch- COMPU- Audi- DISC Acct Firm name (old name)“ codeb angec STAT“ CUSP‘ tor' Yrg Methh KOLLMORGEN CORP 3621 NYSE CI 500440 OR 72 P LA POINTE INDUSTRIES 3663 AMEX CI 503840 OR 72 P LEE PHARMACEUTICALS 2844 AMEX CI 524038 DH 73 C LEEDS & NORTHRUP CO 3823 NYSE CR 524192 CL 74 P LEESONA CORP 3550 NYSE CR 524462 EW 72 P LEVI STRAUSS & CO 2300 NYSE CR 527364 OR 72 P LIONEL CORP 5945 AMEX CI 536257 PW 72 P LOCKHEED CORP 3760 NYSE CI 539821 AY 73 P (Lockheed Aircraft Corp) LOEWS CORP 6199 NYSE CI 540424 DH 73 P LOUISIANA-PACIFIC CORP 2421 NYSE CI 546347 AA 73 P LOWENSTEIN (M.) CORP 2200 NYSE CR 547779 DH 73 C LYKES CORP-DEL 3310 NYSE CR 550890 PW 74 P (Lykes-Youngstown Corp) M/A-COM INC 3674 NYSE CI 552618 PW 72 P (Microwave Associates Inc) MACMILLAN INC 2731 NYSE CI 554790 DH 72 P MALLINCKRODT INC 2890 OTC CR 561229 PW 72 P (Mallinckrodt Chemical Works) MANPOWER INC 7360 NYSE CR 564182 AA 74 G (Parker Pen Co) MANVILLE CORP 3290 NYSE CI 565020 CL 72 P (Johns-Manville Corp) MARATHON MFG CO 3533 NYSE CR 565821 AY 72 P MAREMONT CORP 3714 NYSE CR 566472 AA 73 P MARK CONTROLS CORP-OLD 3822 NYSE CR 570385 AA 72 P MASS MERCHANDISERS INC 5122 NYSE CR 575418 OR 72 P (Napco Industries Inc) MCDONNELL DOUGLAS CORP 3721 NYSE CI 580169 EW 72 P MCGRAW-EDISON C0 3610 NYSE CR 580628 AA 72 P MCLOUTH STEEL CORP 3310 OTC CR 582273 EW 74 P MEASUREX CORP 3823 NYSE CI 583432 CL 72 P MERCK & CO 2834 NYSE CI 589331 AA 72 P MIDLAND-ROSS CORP 3560 NYSE CR 597715 EW 73 P MILGO ELECTRONIC CORP 3661 NYSE CR 599734 AA 72 P MILLIPORE CORP 3811 NYSE CI 601073 CL 72 C MILTON ROY CO 3561 NYSE CI 602108 OR 73 P MISSOURI BEEF PACKERS INC 2010 AMEX CR 606011 AY 73 P MOHAWK RUBBER CO 3011 NYSE CR 608302 EW 74 P MONARCH MACHINE TOOL CO 3541 NYSE CI 609150 CL 73 P MONSANTO CO 2800 NYSE CI 611662 DH 73 P MOOG INC -CL A 3728 AMEX CI 615394 PM 73 P NARCO SCIENTIFIC INC 3841 NYSE CR 630854 PW 72 P NARDA MICROWAVE CORP 3679 AMEX CR 630871 OR 73 P NASHUA CORP 5081 NYSE CI 631226 PW 72 P NATIONAL-STANDARD CO 3310 NYSE CI 637742 PM 73 P 185 Appendix B (cont'd.) DISC Acct Methh SIC Exch- COMPU- Audi- Firm name (old name)“ codeb angec STAT“ CUSP‘ tor’ Yrg NAVISTAR INTERNATIONAL 3711 NYSE CI 638901 DH 73 P (International Harvestor Co) NEPTUNE INTERNATIONAL CORP 4950 NYSE CR 640745 PM 74 C NEW ENGLAND NUCLEAR CORP 2810 NYSE CR 644171 OR 74 C NORRIS INDUSTRIES INC 3714 NYSE CR 656389 DH 74 P NORTHROP CORP 3721 NYSE CI 666807 EW 72 C NORTON CO 3290 NYSE CI 668605 AY 73 P OAK INDUSTRIES INC 3822 NYSE CI 671400 AA 73 P OCCIDENTAL PETROLEUM CORP 1311 NYSE CI 674599 AA 74 P OGDEN CORP 7340 NYSE CI 676346 DH 73 P OLIN CORP 2800 NYSE CI 680665 PM 72 P OUTBOARD MARINE CORP 3510 NYSE CI 690020 AA 73 C PACCAR INC 3711 OTC CI 693718 EW 74 C PALL CORP 3590 AMEX CI 696429 OR 73 P PARK ELECTROCHEMICAL CORP 3679 NYSE CI 700416 TR 74 P PEABODY INTERNATIONAL CORP 8911 NYSE CR 704562 PM 73 P (Peabody-Galion Corp) PEAVEY CO 2040 NYSE CR 705041 PM 74 C PENNWALT CORP 2800 NYSE CI 709317 AA 73 P PERTEC COMPUTER CORP 3684 NYSE CR 715361 AA 72 P (Pertec Corp) PITTSTON CO 4513 NYSE CI 725701 PM 72 P PITTWAY CORP 3666 AMEX CI 725786 PW 74 P PNEUMATIC SCALE CORP 3550 AMEX CI 730162 EW 72 C PNEUMO CORP 5411 NYSE CR 730196 EW 74 C POLAROID CORP 3861 NYSE CI 731095 PM 73 P POPE & TALBOT INC 2421 NYSE CI 732827 AY 73 C PORTEC INC 3531 NYSE CI 736202 OR 72 P PROLER INTERNATIONAL CORP 3312 NYSE CI 743396 CL 73 P (Proler Steel Corp) PULLMAN INC 1600 NYSE CR 745791 AY 72 P PUREX INDUSTRIES INC 2840 NYSE CR 746252 PW 73 P (Purex Corp Ltd) QUANEX CORP 3312 NYSE CI 747620 PM 73' C (Michigan Seamless Tube Co) RCA CORP 3600 NYSE CR 749285 AY 74 P RANCO INC 3822 NYSE CR 752159 CL 73 P RAYCHEM CORP 3640 NYSE CI 754603 AY 73 P RAYTHEON C0 3664 NYSE CI 755111 CL 73 P REED TOOL CO 3533 NYSE CR 758260 AY 73 P REICHHOLD CHEMICALS INC 2820 NYSE CR 759200 PM 74 C RELIANCE ELECTRIC CO 3610 NYSE CR 759457 EW 72 P REXNORD INC 3560 NYSE CR 761688 AA 74 P RIEGEL TEXTILE CORP 2200 NYSE CR 766481 AA 73 C ROCKWELL INTERNATIONAL CORP 3721 NYSE CI 774347 DH 73 P ROGERS CORP 3679 AMEX CI 775133 CL 72 P ROHM & HAAS CO 2821 NYSE CI 775371 PM 72 P 186 Appendix B (cont'd.) SIC Exch- COMPU- Audi- DISC Acct Firm name (old name)“ code“ aaga“ STAT“ CUSP“ tor' Yrg Meth“ ST JOE MINERALS CORP 1211 NYSE CR 790155 DH 72 P SCIENTIFIC-ATLANTA INC 3663 NYSE CI 808655 AA 73 C SCOTT & FETZER CO 3630 NYSE CR 809367 CL 73 P SCOTT PAPER CO 2621 NYSE CI 809877 PW 72 P SEQUA CORP -CL A 3490 NYSE CI 817320 AA 72 P (Sun Chemical Corp) SERVO CORP OF AMERICA 3743 AMEX CI 817698 AA 72 P SETON CO 3100 AMEX CR 817814 OR 73 P SIGMA INSTRUMENTS 3679 AMEX CR 826588 EW 72 P SIGNODE CORP 3490 NYSE CR 826690 AA 73 C SIMMONDS PRECISION PRODS INC 3823 NYSE CR 828675 PM 74 P SINGER CO 3664 NYSE CI 829302 PM 73 P SMITH INTERNATIONAL INC 3533 NYSE CI 832110 AA 72 P SNAP-ON TOOLS CORP 3420 NYSE CI 833034 OR 74 P SOLA BASIC INDUSTRIES INC 3679 NYSE CR 834086 TR 74 P SPECTRA-PHYSICS 3811 NYSE CR 847567 AY 72 P SPEED-O-PRINT BUS MACHINES 5081 AMEX CI 847660 AA 73 C SPENCER FOODS INC 2010 AMEX CR 847889 DH 73 P STANDARD PRODUCTS CO 3714 NYSE CI 853836 AA 73 C STANRAY CORP 3743 NYSE CR 854701 AA 72 P STAUFFER CHEMICAL CO 2800 NYSE CR 857721 DH 72 P STEELMET INC 5093 AMEX CR 858263 OR 72 P STONE CONTAINER CORP 2631 NYSE CI 861589 OR 74 C STORAGE TECHNOLOGY CORP 3684 NYSE CI 862111 AA 72 P SULLAIR CORP 3560 NYSE CR 865112 AA 73 P SUNDSTRAND CORP 3728 NYSE CI 867323 OR 72 P SYSTRON-DONNER CORP 3820 NYSE CR 872056 PM 73 P TRE CORP 3720 NYSE CR 872628 PM 73 P TRW INC 3760 NYSE CI 872649 EW 73 P TAB PRODUCTS 2522 AMEX CI 873197 PW 73 P TECUMSEH PRODUCTS CO 3585 OTC CI 878895 OR 73 P TEKTRONIX INC 3825 NYSE CI 879131 DH 72 P TELEDYNE INC 3724 NYSE CI 879335 AA 73 C TERADYNE INC 3825 NYSE CI 880770 CL 73 C TEXSTAR CORP 3714 AMEX CR 883118 PM 73 C TEXTRON INC 3720 NYSE CI 883203 AY 72 P THERMO ELECTRON CORP 3560 NYSE CI 883556 AA 73 C THIOKOL CORP 2820 NYSE CR 884102 AY 73 C THOMAS INDUSTRIES INC 3640 NYSE CI 884425 EW 74 P TIMES MIRROR CO-DEL 2711 NYSE CI 887360 EW 72 C TORO CO 3520 NYSE CI 891092 PM 72 P TRAFALGAR INDUSTRIES INC 1211 AMEX CR 892711 PM 74 P (Flagstaff Corp) TROPICANA PRODUCTS INC 2030 NYSE CR 897090 EW 73 C TUFTCO CORP 3550 AMEX CR 899041 EW 73 P UOP INC 3350 NYSE CR 903200 AA 72 P (Universal Oil Products Co) 187 Appendix B (cont'd.) SIC Exch- COMPU- Audi- DISC Acct Firm name (old name)“ code“ angec STAT“ CUSP“ tor' Yrg Methh UNION CAMP CORP 2621 NYSE CI 905530 DH 72 P UNITED AIRCRAFT PRODUCTS INC 3720 AMEX CR 909313 CL 73 C UNITED INDUSTRIAL CORP 3665 NYSE CI 910671 EW 72 P UNITED MERCHANTS & MFRS INC 2211 NYSE CI 910858 OR 74 C UNITED TECHNOLOGIES CORP 3724 NYSE CI 913017 PW 72 P (United Aircraft Corp) UNITEK CORP 3843 AMEX CR 913249 AY 72 P UNITRODE CORP 3674 NYSE CI 913283 CL 74 C UNIVERSAL CORP-VA 5150 NYSE CI 913456 AY 74 C (Universal Leaf Tobacco Co) VLN CORP 3690 AMEX CR 918254 PM 72 C VSI CORP 3728 NYSE CR 918314 AA 73 C VALMAC INDUSTRIES INC 2016 AMEX CR 920228 EW 74 C VARIAN ASSOCIATES INC 3670 NYSE CI 922204 CL 72 P VETCO INC 3533 NYSE CR 925496 DH 73 P (Vetco Offshore Inds Inc) VULCAN CORP 3069 AMEX CI 929092 OR 72 P WAGNER ELECTRIC CORP 3714 AMEX CR 930455 CL 72 P WALLACE-MURRAY CORP 3430 NYSE CR 932355 AA 73 C WANG LABORATORIES -CL B 3680 AMEX CI 933696 EW 74 P WARNER & SWASEY CO 3540 NYSE CR 934408 EW 74 P WARNER COMMUNICATIONS INC 7810 NYSE CI 934436 AY 72 P WATKINS-JOHNSON 3664 NYSE CI 942486 DH 72 C WEATHERHEAD CO 3714 NYSE CR 947151 EW 73 P WESTERN PUBLISHING INC 2731 NYSE CR 959265 CL 72 P WESTINGHOUSE ELECTRIC CORP 3664 NYSE CI 960402 PW 73 P WESTVACO CORP 2621 NYSE CI 961548 PW 74 C WEYERHAEUSER CO 2421 NYSE CI 962166 AA 72 P WHITE CONSOLIDATED INDS INC 3630 NYSE CR 963626 EW 72 P WHITEHALL CORP 3811 NYSE CI 965010 EW 72 P (Whitehall Electronics Corp) WILL ROSS INC 2830 NYSE CR 969088 EW 73 P WILLCOX & GIBBS INC 5063 NYSE CI 969207 DH 72 P WILLIAMS COS INC 4922 NYSE CI 969457 AY 74 C WINNEBAGO INDUSTRIES 3716 NYSE CI 974637 OR 73 P WOLVERINE WORLD WIDE 3140 NYSE CI 978097 EW 72 P WOOD INDUSTRIES INC 3550 AMEX CR 978403 PW 74 P WYNN’S INTERNATIONAL INC 3585 NYSE CI 983195 AY 72 P ZAPATA CORP 900 NYSE CI 989070 AA 74 P ZENITH ELECTRONICS CORP 3651 NYSE CI 989349 AA 73 C (Zenith Radio Corp) “The "firm name" is the name by which the sample firm is listed on COMPUSTAT. The "old name" is the name by which the firm was listed on NAARS. Name changes were determined from MOODY's Industrial manuals. 188 Appendix B (cont'd.) “The four-digit primary industry classification numbers assigned to the firms by COMPUSTAT. These conform as nearly as possible to the Office of Management and Budget's Industry Classification (SIC) Codes. cAMEX-American Stock Exchange, NYSE=New York Stock Exchange, OTC=Over-the-counter stocks. “CI=Annual COMPUSTAT Industrial files, CR=Annual COMPUSTAT Research file. “Unique six-digit company identification number conforming to the CUSIP numbering system. 'where: AA-Arthur Andersen & Co AY-Arthur Young CL=Coopers & Lybrand (also Lybrand, Ross Bros. & Montgomery) DH=Deloitte Haskins & Sells (also Haskins & Sells) EE=Ernst & Ernst PM=Peat, Marwick, Mitchell & Co PW-Price Waterhouse & Co TR=Touche Ross OR=Other. “The year in which the DISC became (or was assumed to become) operational. “C = comprehensive allocator, P = partial allocator. Appendix C EXAMPLES OF DISC DISCLOSURES Presented in this Appendix are examples of disclosures made by some sample firms regarding their domestic international sales corporation(s) (DISC). Disclosures were primarily made in the tax footnote and/or the accounting policies footnote to the financial statements. Relevant portions of the footnotes are in italics and bolded for emphasis. In some cases, the entire footnote is not presented if considered unnecessary. It should be noted that these examples are merely illustrative and by no means exhaustive of the variety of disclosures encountered. The disclosures have been divided as follows: C.1 Examples of Disclosures by Comprehensive Allocators a. Mere footnote disclosure GCA CORP b. Dollar effect of DISC deferred tax GARDNER-DENVER CORP. C.2 Examples of Disclosres by Partial Allocators a. Mere footnote disclosure GREAT NORTHERN NEKOOSA CORP b. Dollar effect of DISC in ETR reconciliation BANNER INDUSTRIES, INC. C.3 Examples of DISCc with FYE Different from Parent Corporation a. Different fiscal year mentioned IONICS, INC b. Different fiscal year inferred REED TOOL CO. 0.4 Examples of Year When DISC Operational ("DISC Year") a. Not clear that DISC operational when DISC established AAR CORP b. Clear that DISC operational earlier REED TOOL CO. (see C.3) 189 190 Appendix C (cont'd.) 0.1 Examples of Disclosures by Comprehensive Allocators GCA CORPORATION Notes to Consolidated Financial Statements FYE September 30, 1973 1: Major accounting policies Income taxes: Prepaid and deferred tax accounting is used to recognize timing differences between tax and financial reporting for certain expense items. In addition, deferred taxes have been provided fOr the portion of income of the company’s domestic international sales corporation (DISC) that is eligible fer tax deferral. GARDNER-DENVER CO. Notes to Consolidated Financial Statements FYE December 31, 1973 Note (4) Income Taxes Income tax expense for 1973 is composed of the following Current Deferred U.S. Federal $15,721,000 $2,242,000 Foreign 4,502,000 -- State & local 1.235.000 -- $21,458,000 §2l242l000 Deferred taxes above result from timing differences in the recognition of revenue and expense for tax and financial statement purposes. sources of these differences in 1973 and the related tax effect of each is as follows: elements: Total $17,963,000 4,502,000 1.235.000 22317001000 Tax Effect Income on installment sales deferred for tax purposes . . . . . . . $ 260,000 Tax on portion of income from Domestic International Sales Corporation (DISC) deferred for tax purposes . . . . . . . . . . . . . 1,887,000 Other net . . . . . 95.000 2 242 000 The major 191 Appendix C (cont'd.) Gardner-Denver Co. (FYE 1973) Page 2 The total effective income tax, rate on consolidated pre-tax income is 46.7%, which differs from the expected U.S. Federal income tax rate of 482 for the following reasons: Z of Pre- Amount tax income Expected income tax at at statutory rate . . . . . . . . . . $24,346,000 48.0% Increase (reduction) in taxes resulting from 0 Benefits attributable to lower statutory tax rate of Western Hemisphere Corporation . . . 0 Foreign income, subject to foreign taxes, but not expected to be subject to additional U.S. tax . . . . . . (294,000) (.6) 0 State and local income taxes, net of Federal (802,000) (1.6) income tax benefit . . . . . . 642,000 1.3 o Amortization of prior years' investment credits . . . (160,000) (.3) 0 Misc. items, net . . . . . . . (32.000) (.1) b 05 \J N $23,700,000 It is expected that the cash outlay of income taxes with respect to the years 1974 through 1976 will be less than income tax expense for these years. At December 31, 1973, unremitted income for which U.S. taxes have not been provided totaled $24,700,000. Under existing U.S. income tax laws foreign income tax credits would be available to substantially offset any U.S. taxes. 192 Appendix C (cont'd.) 0.2 Examples of Disclosures by Partial Allocators GREAT NORTHERN NEKOOSA CORPORATION Accounting Policies FYE December 31, 1972 Deferred Taxes on Income: Income taxes are provided on net income as reported in the statement of income regardless of when such taxes are payable. That portion of the annual tax provision not currently payable, primarily resulting from the use of accelerated depreciation, is deferred. The Company has formed two domestic international sales corporation (DISC) subsidiaries to take advantage of legislation permitting federal income taxes on one-half of the income of qualified export sales to be deferred indefinitely, under certain conditions. Since the company meets these conditions, no provision was made in 1972 for $363,000 in taxes otherwise payable. BANNER INDUSTRIES, INC. Notes to consolidated Financial Statements FYE June 30, 1974 Note 8. Federal Income Taxes: The consolidated Federal income tax provision in the accompanying consolidated statement of income differs from the statutory rate as follows: 1974 1973 Income before taxes on income $5,725,741 $4,896,731 Less - State income taxes (255,000) (176,000) Income before Federal income taxes $5,470,741 $4,720,731 Tax provision, at Statutory rates $2,602,000 $2,283,000 Tax effect of foreign operating included above not deductible 176,000 34,000 Investment tax credit (138,000) (95,000) Tax effect of Domestic International Sales Cbrporation (104,000) (82,000) Other, net 9,000 (43,000) Provision for Federal income taxes $2,545,000 $2,097,000 193 Appendix C (cont'd.) Banner Industries, Inc. (FYE 1974) Page 2 Deferred tax expense results from timing differences in the recognition of income and expenses for tax and financial statement purposes. The sources of these items are summarized as follows: 1974 1973 Depreciation $275,000 $180,000 Accrued expenses (329,000) (37,000) Other, net (50,000) 19,000 (104,000) 162,000 The Company' does not currently' provide for Federal income taxes on undistributed earnings of its foreign subsidiary since it is the Company's intention. to ‘have the subsidiary' utilize such earnings for capital expansion and/or debt payment. 194 Appendix C (cont'd.) C.3 Examples of Disclosures by DISCs with FYE Different from Parent Corporation IONICS, INC. Notes to Consolidated Financial Statements FYE December 31, 1973 Note e. Income taxes: The Company and its subsidiaries file separate income tax returns. Deferred taxes arise from timing differences between tax and financial statement reporting as follows: Year Ended December 31 1973 1972 Undistributed earnings of foreign investees $ 60,900 $ 21,900 Long-term contracts 234,900 (32,200) Gain on retirement of debt (6,400) 55,300 Utilization of loss carryforward of subsidiary 41,800 44,000 Different year-end of Ionics Disc. Inc. (9,800) $35,600 Exercise of stock options 19,600 Miscellaneous (12,300) (10,600) $328,700 $114,000 Total income tax expense was $515,900 for 1973 and $350,400 for 1972. These amounts were less than those which would be obtained by applying the United States federal income tax rate of 48% to income before taxes, as follows: Year Ended December 31 1973 1972 Indefinite deferral of Disc income $28,400 $35,600 Investment credit (flow-through method) 10,000 11,000 State taxes (net of federal tax effects) (39,100) (26,800) Other 6,400 4,000 $5,700 $23,800 195 Appendix C (cont'd.) REED TOOL CO. Notes to Consolidated Statement of Operations FYE December 31, 1974 Note D. The provisions for federal income taxes are based on the Company and its eligible subsidiaries filing,a consolidated federal income tax return. Income taxes on continuing operations included in the consolidated statement of operations are as follows: Year Ended December 31, (in thousands of dollaga) 1972 1973 1974 Current $1,564 $2,925 $3,247 Deferred (164) (740) 39 $1,400 $2,185 $3,286 Deferred income taxes represent federal income taxes and current income taxes include state and foreign income taxes (Hf $164.000 III 1973 and $580,000 in 1974. Deferred income taxes result from timing differences in the recognition of revenue and expense for tax and financial statement purposes. The sources of these differences in 1973 and 1974 from continuing operations and the tax effect of each were as follows: Year Ended December 31, (In thousands of dollars) 1973 1974 Excess of tax over book depreciation $ 335 $ 308 Income from change in inventory method for tax purposes -- (222) Deferred compensation not deductible for tax purposes until paid (71) (180) 196 Appendix C (cont'd.) REED TOOL CO. (FYE 1974) Page 2 Deferred income of a Domestic International Sales corporation $ 45 $ 156 Accrued expenses not deductible 27 (90) Deferred charges expensed currently for tax purposes 74 75 Proceeds from settlement of Company's suit against the former principal stockholder recognized for financial statement purposes in prior years (963) -- Inventory reserve deducted for financial statement purposes in years different than for tax purposes (150) '- Other __1311 §__L§l $(Z40) $ 39 Differences in 1973 and 1974 between the Company's effective income tax rate reflected in the provision for income taxes on continuing operations and the amount resulting from application of the statutory federal income tax rate of 48% to income from continuing operations before income taxes are shown below: Year Ended December 31, (In thousands of dollara) 1973 1974 Provision for income taxes at statutory rate $2,294 $4,138 Increase (reductions) resulting from: Investment tax credits (95) (507) Nontaxable income of a Domestic International Sales corporation (45) (141) Excess of allowable tax depletion over book depletion on mineral deposits (119) (118) Other __15_0_ ___1é_ Provision for income taxes $2,185 $3,286 197 Appendix C (cont'd.) 0.4 Examples of Disclosures for Year when DISC Established/Operational AAR CORP Notes to Consolidated Financial Statements FYE May 31, 1973 Income Taxes In accordance with the provisions of the Revenue Act of 1971, the Company established a Domestic International Sales Corporation ("DISC") in 1972. Under the provisions of the Revenue Act, 50% of the DISC's earnings are not subject to taxation until distributed to the parent company. As it is the Company's intention to permanently reinvest such earnings to finance international expansion, no Federal income taxes have been provided on the earnings to be retained by the DISC. Benefits of the investment tax credit Appendix D INDUSTRY CLASSIFICATION OF SAMPLE FIRMS DISC SIC- Acc. Meth.“ Code Industry name“ C P Total 900 Fishing, hunting, trapping 0 1 1 1040 Gold and silver ores 1 0 1 1090 Miscellaneous metal ores l 0 1 1211 Bituminous coal and lignite 0 3 3 1311 Crude petroleum and natural gas 0 2 2 1381 Drilling oil and gas wells 0 1 1 1389 Oil, gas field services, nec 1 1 2 1600 Construction - not bldg constr 0 2 2 2010 Meat products 0 3 3 2016 Poultry dressing plants 1 0 l 2030 Can, preserve fruit, vegetable l 0 1 2040 Grain mill products 1 0 1 2048 Prep feeds for animals, nec 1 0 1 2070 Fats and oils 0 1 1 2086 Bottled and canned soft drinks 0 1 1 2111 Cigarettes 0 1 1 2200 Textile mill products 5 1 6 2211 BRD woven fabric mill, cotton 1 1 2 2221 BRD woven fabrc man-made silk 0 1 1 2272 Tufted carpets and rugs 0 l 1 2300 Apparel and other finished pds 0 l 1 2400 Lumber and wood pds, ex furn 1 0 1 2421 Saw mills, planning mills, gen 1 2 3 2510 Household furniture 0 l 1 2522 Metal office furniture 0 1 1 2621 Paper mills, ex bldg paper 2 3 5 2631 Paper board mills 2 0 2 2711 Newspaper: pubg, pubg & print 1 0 1 2731 Books: pubg, pubg & printing 0 3 3 2800 Chemicals and allied products 1 6 7 2810 Indl inorganic chemicals 1 0 1 2820 Plastic matl, synthetic resin 2 0 2 2821 Plastics, resins, elastomers 0 l 1 2830 0 4 4 2834 Pharmaceutical preparation 0 l 1 2840 Soap, detergent, toilet preps 0 1 1 2844 Perfume, cosmetic, toilet prep 1 0 l 2851 Paints, varnishes, lacquers 0 1 1 2860 Industrial organic chemicals 0 1 1 2870 Agriculture chemicals 0 1 1 2890 Misc chemical products 2 2 4 3011 Tires and inner tubes 0 1 1 198 199 Appendix D (cont'd.) DISC SIC- Acc. Meth.“ Code Industry name“ C P Total 3041 1 O 1 3069 Fabricated rubber pds, nec 0 3 3 3079 Misc plastics products 0 1 1 3100 Leather and leather products 0 l 1 3140 Footwear, except rubber 0 l 1 3211 Flat glass 0 1 1 3220 0 1 1 3231 0 1 1 3290 Abrasive, asbestos, misc, mineral 0 2 2 3310 Blast furnaces and steel works 0 3 3 3312 Blast furnaces and rolling mills 1 2 3 3330 Prim smelt, refin nonfer metal 1 0 1 3350 Rolling and draw nonfer metal 0 1 1 3390 Misc primary metal products 0 1 l 3420 Cutlery, hand tools, gen hardware 0 2 2 3430 Heating eq, plumbing fixture 1 0 l 3443 Fabricated plate work 0 1 l 3460 Metal forgings and stampings 0 1 l 3470 Coating, engraving, allied services 1 0 1 3490 Misc fabricated metal products 2 1 3 3494 Valve, pipe fittings, ex brass 0 l 1 3510 Engines and turbines 2 1 3 3520 Farm and garden machinery and equip 0 1 1 3523 Farm machinery and equipment 1 0 1 3530 Constr, mining, matl handle equip 1 3 4 3531 Contruction machinery and equip 0 2 2 3533 Oil field machinery and equip 0 6 6 3537 Indl trucks, tractors, trailers 0 1 1 3540 Metal working machinery and equip 1 2 3 3541 Machine tools, metal cutting 0 4 4 3550 Special industry machinery 2 7 9 3555 Printing trades machy, equip 0 1 l 3558 Pollution control machinery 0 1 l 3560 General industrial mach and equip 2 6 8 3561 Pumps and pumping equipment 1 1 2 3564 Blowers, exhaust, ventilation fans 0 1 1 3570 Office, computing, accounting mach 0 1 1 3580 Refrig and service ind machine 0 3 3 3585 Air cond, heating, refrig equip 0 2 2 3590 Misc machinery, ex electrical 0 1 1 3600 Elec, electr mach, equip, supply 0 2 2 3610 Elec transmissions and distr equip 0 3 3 3621 Motors and generators 1 2 3 3630 Household appliances 0 2 2 3640 Electric lighting, wiring, equip 0 2 2 3651 Radio and TV receiving sets 1 2 3 3661 Tele and telegraph apparatus 0 1 1 200 Appendix D (cont'd.) DISC src- Acc. Meth.“ Coda Industry name“ C P Total 3663 Radio, TV comm equip, apparatus 1 3 4 3664 Search, navigate, guide sys, equip 4 5 9 3665 Training equip and simulators 0 1 1 3666 Alarm and signalling products 0 1 1 3670 Electronic comp, accessories 0 1 1 3674 Semi-conductor, related device 1 2 3 3679 Electronic components, nec 0 8 8 3680 Electronic computing equip 2 2 4 3684 Computer disk and tape drives 0 2 2 3688 Computer peripherals 1 1 2 3689 Computer equipment, nec 0 1 1 3690 Misc elec machy, equip, supplies 1 0 1 3711 Motor vehicles and car bodies 1 3 4 3714 Motor vehicle part, accessory 5 8 13 3716 Motor homes 0 l 1 3720 Aircraft and parts 1 2 3 3721 Aircraft 2 4 6 3724 Aircraft engine, engine parts 1 1 2 3728 Aircraft parts, aux equip, nec 1 4 5 3730 Ship and boat bldg and repairing 0 1 1 3743 Railroad equipment 0 2 2 3760 Guided missiles and space vehicles 0 2 2 3811 Engr, lab and research equipment 1 2 3 3820 Measuring, controlling instr 0 1 1 3822 Automatic regulating controls 0 3 3 3823 Industrial measurement instr 0 5 5 3825 Elec meas and test instruments 2 3 5 3829 Meas and controlling dev, nec 0 1 1 3841 Surgical, med instr, apparatus 0 4 4 3843 Dental equipment and supplies 0 4 4 3861 Photographic equip and supplies 0 2 2 3931 Musical instruments 0 l l 3942 Dolls 0 1 1 3949 Sporting and athletic goods, nec 0 1 1 3950 Pens, pencils, other office matl 0 1 1 3990 Misc manufacturing industries 0 1 1 4513 Air courier services 0 1 1 4922 Natural gas transmission 1 0 1 4950 Sanitary services 1 0 l 4953 Refuse systems 0 1 1 5012 1 0 1 5051 Metal service centers - whsl l 0 1 5063 Elec apparatus and equip - whsl 0 1 l 5065 Electronic parts and equip - whsl 0 l l 5070 Hardware, plum, heat equip - whsl 0 l 1 5080 Machinery and equipment - whsl 0 1 1 5081 Comml machines and equip - whsl 1 1 2 201 Appendix D (cont'd.) DISC src- Acc. Meth.“ Code Industry,name“ C P Total 5093 0 1 1 5110 Paper and paper products - whsl O 1 1 5122 Drugs and proprietary - whsl 0 1 1 5140 Groceries and related pds - whsl 0 l 1 5150 Farm-product raw matl - whsl l 0 1 5411 Grocery stores 1 1 2 5812 Eating places 1 0 1 5945 Hobby, toy, game shops 0 1 1 6199 Finance-services 0 1 l 7200 Personal services 1 0 1 7340 Svcs to dwellings, other buildings 0 1 1 7360 Personnel supply services 1 0 1 7394 Equip rental and leasing services 1 0 1 7399 Business services, nec 0 1 1 7810 Motion picture prodn, services 0 l 1 7990 Misc amusement and rec services 0 2 2 8911 Engr, architect, survey services 1 2 3 Total 82 238 320 “As per Appendix B of the Standard & Poor's COMPUSTAT manual (April 11, 1988). “C=comprehensive allocators, P=partial allocators. BIBLIOGRAPHY BIBLIOGRAPHY Accounting Standards Committee (ASC), Statement of Standard Accounting Practice No. 15, Accounting for Deferred Taxes (London, Eng: ASC, October 1978), and as revised in 1985. Amemiya, T., "Qualitative Response Models: A Survey," Journal of Economic Literature (December 1981), pp. 1483-1536. American Bar Foundation (ABF), Commentaries on Indentures (Chicago, IL: ABF, 1971) American Institute of Certified Public Accountants (AICPA), Accounting Principles Board Opinion No. 11, Accounting for Income Taxes (New York, NY: AICPA, December 1967). ------ , Accounting Principles Board Opinion No. 23, Accounting for Income Taxes- -Special Areas (New York, NY: AICPA, April 1972). Ayres, F., "Characteristics of Firms Electing Early Adoption of SFAS 52," Journal of Accounting and Economics (June 1986), pp. 143-158. Baiman, S., "Agency Research in Management Accounting: A Survey," Journal of Accounting Literature (1982), pp. 154-213. Ball, R., "Accounting, Auditing and the Nature of the Firm," Unpublished Working Paper (University of Rochester, June 1987). ------ , and P. Brown, "An Empirical Evaluation of Accounting Income Numbers," Joumal of Accounting Research (Autumn 1968), pp. 159-178. ------ , and G. Foster, "Corporate Financial Reporting: A Methodological Review of Empirical Research," Joumal of Accounting Research (Supplement, 1982), pp. 161-234. 202 203 Beaver, W., and R. Dukes, "Interperiod Tax Allocation, Earnings Expectations, and the Behavior of Security Prices," The Accounting Review (April 1972), pp. 320-332. ------ , "Tax Allocation and 6-Depreciation Methods," The Accounting Review (July 1973), pp. 549-559. Beresford, D., "Deferred Tax Accounting Should be Changed," The CPA Journal (June 1982), pp. 16—23. Beresford, D., 1... Best, P. Craig, and J. Weber, Accounting for Income Taxes: A Review of Alternatives (Stamford, CI‘: Financial Accounting Standards Board, 1983). Bhattacharyya, G., and R. Johnson, Statistical Concepts and Methods (J. Wiley & Sons, Inc., 1977). Birnbaum, J., and A. Murray, Showdown at Gucci Gulch: Lawmakers, Lobbyists, and the Unlikely Triumph of Tax Reform (New York, NY: Random House, 1987) Black, H., Accounting Research Study No. 9, Interperiod Allocation of Corporate Income Taxes (New York, NY: AICPA, 1966). Bowen, R., E. Noreen, and J. Lacey, "Determinants of the Corporate Decision to Capitalize Interest," Joumal of Accounting and Economics (August 1981), pp. 151-179. Carmichael, J., "Rollover," Forbes (January 18, 1982), pp. 75, 78. Chaney, P., and D. Jeter, "Accounting for Deferred Income Taxes: Simplicity? Usefulness?" Accounting Horizons (June 1989), pp. 6-13. Chow, C., "The Demand for External Auditing: Size, Debt and Ownership Influences," The Accounting Review (April 1982), pp. 272-291. Christie, A., "Aggregation of Test Statistics: An Evaluation of the Evidence on Contracting and Size Hypotheses," Journal of Accounting and Economics (1990, forthcoming). Citizens for Tax Justice (CT J ), Corporate Taxpayers and Corporate Freeloaders: Four Years of Continuing Legalized Tax Avoidance by America’s Largest Corporations, 1981-84 (CT J, 1985). 204 Collins, D., M. Rozeff, and D. Dhaliwal, "The Economic Determinants of the Market Reaction to Proposed Mandatory Accounting Changes in the Oil and Gas Industry: A Cross Sectional Analysis," Joumal of Accounting and Economics (March 1981), pp. 37-71. Conover, W., Practical Nonparametric Statistics (J. Wiley & Sons, Inc., 1980). Cook, T., and D. Campbell, Quasi-Experimentation: Design and Analysis Issues for Field Settings (Boston, MA: Addison Wesley, 1977). Daley, L., and R. Vigeland, "The Effects of Debt Covenants and Political Costs on the Choice of Accounting Methods: The Case of Accounting for R & D Costs," Journal of Accounting and Economics (December 1983), pp. 195- 211. Davidson, S., L. Skelton, and R. Weil, "A Controversy Over the Expected Behavior of Taxes-~A Comment," The Journal of Accountancy (April 1977), pp. 53- 54. DeAngelo, H., and R. Masulis, "Optimal Capital Structure under Corporate and Personal Taxation," Joumal of Financial Economics (March 1980), pp. 3-29. Defliese, P., "Deferred Taxes--Forever," The Joumal of Accountancy (August 1983), pp. 94-98. Dhaliwal, D., "The Effect of the Firm’s Capital Structure on the Choice of Accounting Methods," The Accounting Review (January 1980), pp. 78-84. ------ , G. Salamon, and E. Smith, "The Effect of Owner Versus Management Control on the Choice of Accounting Methods," Journal of Accounting and Economics (July 1982), pp. 41-53. Egger, J., "Citizens for Tax Justice’s Latest Misrepresentation of Corporate Tax Burdens," Economic Report (Institute for Research on the Economics of Taxation [IRET], 1985) reprinted in Tax Notes (December 2, 1985), pp. 956- 963. El-Gazzar, S., S. Lilien, and V. Pastena, "Accounting for Leases by Lessees," Journal of Accounting and Economics (October 1986), pp. 217-237. Financial Accounting Standards Board (FASB), Interpretation No. 22, Applicability of the Indefinite Reversal Criteria to Timing Differences, (Stamford, CI‘: FASB, April 1978). 205 ------ , Technical Bulletin No. 84-2, Accounting for the Efi'ects of the Tax Reform Act of 1984 on Deferred Taxes Relating to Domestic Intemational Sales Corporations (Stamford, CT: FASB, September 18, 1984). ------ , Accounting for Income Taxes--Exp0sure Draft (Stamford, CT: FASB, September 1986) ------ , Statement of Financial Accounting Standards No. 96, Accounting for Income Taxes (Stamford, CT: FASB, December 1987). Fogelson, J., "The Impact of Changes in Accounting Principles on Restrictive Covenants in Credit Agreements and Indentures," The Business Lawyer (January 1978), pp. 769-787. Foster, G., Financial Statement Analysis (Engelwood Cliffs, NJ: Prentice-Hall, 1986). ------ , "Accounting Policy Decisions and Capital Market Research," Journal of Accounting and Economics (March 1980), pp. 29-62. Fullerton, D., "The Use of Effective Tax Rates in Tax Policy," National Tax Joumal (September 1986), pp. 285-292. Grant, E., "Market Implications of Differential Amounts of Interim Information," Journal of Accounting Research (Spring 1980), pp. 255-268. Hagerman, R., and M. Zmijewski, "Some Economic Determinants of Accounting Policy Choice," Journal of Accounting and Economics (August 1979), pp. 141-161. Healy, P., "The Impact of Bonus Schemes on the Selection of Accounting Principles," Joumal of Accounting and Economics (April 1985), pp. 85-107. Holthausen, R., "Theory and Evidence of the Effect of Bond Covenants and Management Compensation Contracts on the Choice of Accounting Techniques: The Case of the Depreciation Switch-Back," Journal of Accounting and Economics (March 1981), pp. 73-109. ------ , and R. Leftwich, "The Economic Consequences of Accounting Choice: Implications of Costly Contracting and Monitoring," Journal of Accounting and Economics (August 1983), pp. 77-117. Internal Revenue Code of 1986 (I.R.C.), as amended. 206 Internal Revenue Service (IRS), Statistics of Income--1972—74, International Income and Taxes, Domestic International Sales Corporation Returns (Washington, DC, 1980). International Accounting Standards Committee (IASC), International Accounting Standard No. 12, Accounting for Taxes on Income (London, Eng: IASC, 1979). Jensen M., and W. Meckling, "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure," Journal of Financial Economics (October 1976), pp. 305-360. Johnson W., and R. Ramanan, "Discretionary Accounting Changes from ‘Successful Efforts’ to ‘Full Cost’ Methods: 1970-76," The Accounting Review (January 1988), pp. 96-110. Kelly, L., "The Development of a Positive Theory of Corporate Management’s Role in External Financial Reporting," Journal of Accounting Literature (Spring 1983), pp. 111-150. Kmenta, J., Elements of Econometrics (New York, NY: Macmillan Publishing Company, 1986). Lantz, K., A. Snyir, and J. Williams, "A Controversy Over the Behavior of Deferred Taxes--A Comment," The Journal of Accountancy (April 1977), pp. 56-58. Leftwich, R., "Evidence of the Impact of Mandatory Changes in Accounting Principles on Corporate Loan Agreements," Joumal of Accounting and Economics (March 1981), pp. 3-36. ------ , "Accounting Information in Private Markets: Evidence from Private Lending Agreements," The Accounting Review (January 1983), pp. 23-42. ------ , "Aggregation of Test Statistics: Statistics v. Economics," Journal of Accounting and Economics (1990, forthcoming). Lev, B., and J. Ohlson, "Market-based Empirical Research in Accounting: A Review, Interpretation, and Extension," Journal of Accounting Research (Supplement 1982), pp. 249-322. Liberty, S., and J. Zimmerman, "Labor Union Contract Negotiations and Accounting Choices," The Accounting Review (October 1986), pp. 692-712. 207 Liebtag, B., "FASB on Income Taxes," The Journal of Accountancy (March 1987), pp. 80-84. Lilien, S., and V. Pastena, "Determinants of Intramethod Choice in the Oil and Gas Industry," Journal of Accounting and Economics (December 1982), pp. 145-170. Lys, T., "Mandated Accounting Changes and Debt Covenants: The Case of Oil and Gas Accounting," Journal of Accounting and Economics (April 1984), pp. 39- 66. Maddala, G., Limited Dependent and Qualitative Variables in Econometrics (New York, NY: Cambridge University Press, 1983). Marocco, M., "DISC/FSC Legislation: The Case of the Phantom Profits," Journal of Accountancy (January 1983), pp. 83-96. McGoldrick, B., "The Controversy Over Deferred Taxes," Institutional Investor (September 1984), pp. 247, 250, 253. McIntyre, R., "Just Taxes and Other Options," in Less Taxing Altematives: An Essay and Colloquium (New York, NY: The Democracy Project, March 1984). McKee, A., T. Bell, and J. Boatsman, "Management Preferences Over Accounting Standards: A Replication and Additional Tests," The Accounting Review (October 1984), pp. 647-659. Modigliani, F., and M. Miller, "The Cost of Capital, Corporate Finance, and the Theory of Investment," American Economic Review (June 1958), pp. 261- 297. Morrison, D., "On the Interpretation of Discriminant Analysis," Joumal of Marketing Research (May 1969), pp. 156-163. Myers, S., "Determinants of Corporate Borrowing," Joumal of Financial Economics (November 1977), pp. 147-175. Nair, R., and J. Weygandt, "Let’s Fix Deferred Taxes," The Journal of Accountancy (November 1981), pp. 87, 90, 92-94, 96, 98, 100, 102. Noreen, E., "An Empirical Comparison of Probit and OLS Regression Hypothesis Tests," Journal of Accounting Research (Spring 1988), pp. 119-133. 208 Peltzman, S., "Toward a More General Theory of Regulation," Journal of Law and Economics (August 1976), pp. 211-240. Pindyck, R., and D. Rubinfeld, Econometric Models and Economic Forecasts (McGraw-Hill, 1981). Press, E., and J. Weintrop, "Accounting-based Constraints in Public and Private Debt Agreements: Their Association with Leverage and Impact on Accounting Choices," Journal of Accounting and Economics (1990, forthcoming). Press, 8., and S. Wilson, "Choosing Between Logistic Regression and Discriminant Analysis," Journal of the American Statistical Association (December 1978), pp. 699-705. Raviv, A., "Management Compensation and the Managerial Labor Market: An Overview," Journal of Accounting and Economics (April 1985), pp. 239-245. Rayburn, F., "A Chronological Review of the Authoritative Literature on Interperiod Tax Allocation: 1940-85," The Accounting Historians Joumal (Fall 1986), pp. 89-108. Rosenfield, P., and W. Dent, "No More Deferred Taxes," The Journal of Accountancy (February 1983), pp. 44-45, 47-48, 50, 52, 54—55. Securities and Exchange Commission (SEC), Accounting Series Release No. 149, Amendment to Regulation S-X to Provide for Improved Disclosure of Income Tax Expense (SEC, November 28, 1973). Shevlin, T., "Taxes and Off-Balance Sheet Financing: Research and Development Limited Partnerships," The Accounting Review (July 1987), pp. 480-509. Simunic, D., "The Pricing of Audit Services: Theory and Evidence," Journal of Accounting Research (Spring 1980), pp. 161-190. Smith, C., and J. Warner, "On Financial Contracting: An Analysis of Bond Covenants," Journal of Financial Economics (June 1979), pp. 117-161. Spooner, G., "Effective Tax Rates from Financial Statements," National Tax Journal (September 1986), pp. 293-306. Stigler, G., "The Theory of Economic Regulation," The Bell Joumal of Economics and Management Science (Spring 1971), pp. 3-21. 209 Thornton, D., "Current Cost Disclosers and Nondisclosers: Theory and Canadian Evidence," Contemporary Accounting Research (Fall 1986), pp. 1-34. Trombley, M., "Accounting Method Choice in the Software Industry: Characteristics of Firms Electing Early Adoption of SFAS No. 86," The Accounting Review (July 1989), pp. 529-538. U.S. Congress, Tax Reform Act of 1969, P.L. 91-172, H.R. 13270, 9lst Cong., 1st Sess. (December 30, 1969). ------ , House, House Rept. No. 92-533, 92nd Cong., 1st Sess. (September 29, 1971). ------ , Senate, Senate Rept. No. 92-437, 92nd Cong., lst Sess. (November 9, 1971). ------ , Revenue Act of 1971, P.L. 92-178, H.R. 10947, 92nd Cong, -1st Sess. (December 10, 1971). ------ , Tax Reduction Act of 1975, P.L. 94-12, HR. 2166, 94th Cong., 1st Sess. (March 29, 1975). ------ , Tax Reform Act of 1976, P.L. 94-455, H.R. 10612, 94th Cong, 2nd Sess. (October 4, 1976). ------ , Tax Equity and Fiscal Responsibility Act of 1982, P.L. 97-248, HR. 4961, 97th Cong., 2nd Sess. (September 3, 1982). ------ , Deficit Reduction Act of 1984, P.L. 98-369, HR. 4170, 98th Cong., 2nd Sess (July 18, 1984). ------ , Joint Committee on Taxation, Study of 1983 Effective Tax Rates of Selected Large U.S. Corporations (Joint Committee Print, JCS 40-84, November 28, 1984) ------ , Joint Committee on Taxation (JCT), General Explanation of the Revenue Provisions of the Deficit Reduction Act of 1984, P.L. 98-369, HR. 4170, 98th Cong. (December 31, 1984). Watts, R., and J. Zimmerman, "Towards a Positive Theory of the Determination of Accounting Standards," The Accounting Review (January 1978), pp. 112- 134. ------ , Positive Accounting Theory (Engelwood Cliffs, NJ: Prentice-Hall, 1986). ------ , "Positive Accounting Theory: A Ten Year Perspective," Unpublished working paper, University of Rochester (June 1989). 210 Wheeler, J., and E. Outslay, Accounting for Income Taxes: A Survey of Practice (Morristown, NJ: Financial Executives Research Foundation [FERF], 1985). ------ , "The Phantom Federal Income Taxes of General Dynamics Corporation," The Accounting Review (October 1986), pp. 760-774. Wolk, H., D. Martin, and V. Nichols, "Statement of Financial Accounting Standards No. 96: Some Theoretical Problems," Accounting Horizons (June 1989), pp. 1-5. Wong, J., "Political Costs and an Intraperiod Accounting Choice for Export Tax Credits," Joumal of Accounting and Economics (January 1988), pp. 37-51. Zimmerman, J., "Taxes and Firm Size," Journal of Accounting and Economics (August 1983), pp. 119-149. Zmijewski, M., and R. Hagerman, "An Income Strategy Approach to the Positive Theory of Accounting Standard Setting/Choice," Journal of Accounting and Economics (August 1981), pp. 129-149.