ABSTRACT THE DISCLOSURE OF CAPITALIZED LEASE INFORMATION AND STOCK PRICE By Byung - Tak R0 The purpose of this research was to empirically examine whether or not the SEC's recent decision requiring the public disclosure of capitalized lease accounting data had any impact on the pricing of securities. In October l973 the SEC issued ASR No. l47 (effective as of November l973) calling for the disclosure of capitalized lease infor- mation which was not previously required by any accounting regulatory agency. The important items of disclosure are: (l) the present value (PV) of noncapitalized future financing leases and (2) the impact on net income (income effect: IE) if such leases were capitalized. The SEC believes that the capitalized lease data are considered "essential to investors" in evaluating the risk-return prospects of the firms which have long-term lease commitments. If the SEC's belief is true, the disclosure of the PV and IE numbers under ASR No. l47 will affect the investors' assessments of the risk-return prospects of the firms and hence the equilibrium pricing of securities. The present study assessed an effect of capitalized lease infor- mation upon equilibrium rates of security returns by comparing the re- turn distribution functions conditional upon the lease information with the unconditional return distributions. In making this assessment, it Byung - Tak R0 was assumed that a vector of returns (E) on all assets has a multivar— iate normal distribution. Under this assumption, the condition for lease information effect is F(3/§) # F(B) where §_= the capitalized lease information vector and F = the distribution functions of returns. The above condition is equivalent to E(E]§) f E(E) and/or Var(§]§) # Var(3). Either of these two conditions (or both) is sufficient for g to have information content. Since the two types of return data, conditional on §_and uncon- ditional, cannot be observed for a firm for the same time period, the present study employed control firms,the returns of which were used as unconditional returns, while the returns of treatment firms were utilized as conditional returns. There were two control groups in the present study: non-lease firms (Group l) and non-disclosure firms (Group 2). The treatment firms (Group 3) were those firms which dis- closed the capitalized lease data according to ASR No. 147 for 1973 and l974. All return data were obtained from a l975 edition of the CRSP tape. The names of both treatment and control firms were ob- tained by a reading of lO-K reports and annual reports for the three years, 1972 through l974. Depending upon types of the capitalized lease disclosure, the treatment firms were classified into two subgroups: the PV disclosure firms and the firms with both PV and IE disclosure. No single firm was found to disclose the IE alone. Then, each of these subgroups was again divided into two risk classes, high and low, according to its relative risks. The control firms of Group l (and Group 2) were indiv- idually paired with the treatment firms in Group 3 according to their Beta estimates. Thus, they were classified into the same subgroups and Byung - Tak Ro risk classes as were the treatment firms. All statistical tests were conducted by comparing the high and low risk firms of Group 3 with the high and low risk firms of Group l (and Group 2). The tests were aimed at seeing whether the two-com- ponent (high and low risk) mean return vector (and the 2 x 2 variance- covariance matrix of returns) of the treatment group was equal to the corresponding mean vector (and the variance-covariance matrix) of the control group. The equality of the two mean return vectors was statis- tically tested using Hotelling's T2 statistic, while the equality of the variance-covariance matrices was evaluated by employing Box's generalized test (Jf homoscedasticity of variances and covariances. The test period covered 21 months, January 1973 through Septem- ber 1974. In addition to a test for the 21-month period, this entire period was divided into five sub-periods for which separate tests were also conducted. The test results showed that the mean (expected) values of return distributions for the treatment firms changed signif— icantly as a result of the SEC lease decision. However, no evidence was found for significant changes in the variability of returns. Since a significant change in the mean of return distribution is suf- ficient, by definition, for the hypothesized lease information effect to exist, it was concluded that the SEC lease disclosure decision had an impact upon the pricing of securities. This conclusion is consis- tent with the SEC's belief that capitalized lease information is important to investors. Also, the empirical evidence found here sup- ports the traditional view that maintains the existence of information content in capitalized lease data. THE DISCLOSURE OF CAPITALIZED LEASE INFORMATION AND STOCK PRICE By Byung - Tak Ro A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Accounting and Financial Administration 1976 To my parents, wife, and two sons. ii ACKNOWLEDGMENTS A special note of gratitude is extended to Professor Ronald M. Marshall, chariman of my dissertation committee, and two other members, Professors Daniel W. Collins and Richard R. Simonds. They have been available for numerous discussions on an unlimited basis and have pro- vided invaluable guidance and suggestions without which the outcome of this research would not have been as successful as intended. No words will suffice to express my appreciation to them. This research was financially supported by the Haskins & Sells Foundation, New York, New York, and the Department of Accounting and Financial Administration, Michigan State University, East Lansing, Michigan. Without such support, a timely completion of this project would not have been possible. In addition, my personal thanks are due in particular to Profes- sors Gardner M. Jones and Alvin A. Arens, former chairmen of the department, for providing various departmental supports during the time of this dissertation endeavor. Most especially, I appreciate the encouragement from and the endurance of my wife, Eun, and my two sons, Young and Tad, to whom I am eternally indebted. TABLE OF CONTENTS Page LIST OF TABLES . ......... . . . . . . . . . . . . . . vi LIST OF FIGURES. . . . . . . . . . . . . ..... . . . . . . vii CHAPTER I INTRODUCTION . . . . . ................. 1 II THE NEW LEASE DISCLOSURE REQUIREMENTS AND EFFECTS ON ACCOUNTING VARIABLES . . . . . . ........ 10 Historical Review of Accounting Regulations for Reporting Leases ................. . . 10 The AICPA Pronouncements ............. . 10 The SEC Regulations Before ASR No. 147 . ..... . 13 SEC's Assessment of Previous Regulations . ..... 14 The Lease Disclosure Requirements in ASR No. 147 . . . 15 Effects on Financial Ratios ......... . . . . 17 III ANALYTICAL FRAMEWORK AND METHODOLOGY FOR EVALUATING THE EFFECT OF CAPITALIZED LEASE DISCLOSURE ..... . . . 23 Assessment of the Effects of Lease Information on Equilibrium Rates of Return . . . . . . . ..... . 23 Effects of Lease Information . . . . . . . . . . . . 23 Control by Matching ............... . 28 Two- Asset Assumption . . . .......... 30 Testing Hypotheses and Evaluation Methods ...... 34 Testing Hypotheses . ..... . . . . . . ..... 34 TestS(N1Mean Vectors ................ 36 Testscn1Covariance Matrices ........... . 39 Testing Procedures .................. 41 Test Periods . . . . . ...... . . ....... 41 Comparisons of Groups .......... . . . . . 44 IV SAMPLING DESIGN FOR DATA COLLECTION ........ . . 48 Sample Firms ......... . . . ......... 48 Disclosure Firms .................. 48 Non-Disclosure and Non-Lease Firms ........ . 52 Results of Sample Selection ............ 54 Security Return Data ................. 62 iv CHAPTER Page V ANALYSIS OF TEST RESULTS ................ 63 Test Results on Mean Return Vectors ......... 64 Test Results on Mean Vectors: Disclosure Firms (Group 3) vs. Non-Lease Firms (Group 1) ...... 64 PV Disclosure Firms: Group 3 vs. Group 1 . . . . 66 PV and IE Disclosure Firms: Group 3 vs. Group 1 . 72 Test Results on Mean Vectors: Disclosure Firms (Group 3) vs. Non-Disclosure Firms (Group 2) . . . . 78 PV Disclosure Firms: Group 3 vs. Group 2 . . . . 78 PV and IE Disclosure Firms: Group 3 vs. Group 2 . 84 Test Results on Covariance Matrices ......... 88 Test Results on Covariance Matrices: Disclosure Firms (Group 3) vs.Non-Lease Firms (Group 1) . . . . 90 PV Disclosure Firms: Group 3 vs. Group 1 . . PV and IE Disclosure Firms: Group 3 vs. Group 1 . 94 Test Results on Covariance Matrices: Disclosure Firms (Group 3) vs. Non~Disclosure Firms (Group 2) . 95 PV Disclosure Firms: Group 3 vs. Group 2 . . . . 96 PV and IE Disclosure Firms: Group 3 vs. Group 2 . 96 Assessment of the SEC Materiality Criteria for Lease Disclosure .................... . . 99 Some Remarks on Test Results ............. 101 IV SUMMARY AND CONCLUSIONS ................ 108 Summary of Test Results ............... 108 Conclusions ..................... 111 Study Contributions ................. 112 Study Limitations .................. 113 Suggestions for Future Research ........... 115 CHAPTER NOTES ......................... 116 APPENDICES APPENDIX A: THE FINANCIAL STATEMENTS FOR A MATERIALLY AFFECTED FIRM (BEFORE AND AFTER THE SEC'S T973 LEASE DISCLOSURE DECISION) ................. 129 APPENDIX B: KEY FINANCIAL RATIOS BEFORE AND AFTER THE SEC DECISION (1974) ............... 132 APPENDIX C: LIST OF SAMPLE COMPANIES BY GROUPS AND MATCHING OF FIRMS .................. 133 SELECTED BIBLIOGRAPHY ..................... 138 LIST OF TABLES Table Page 4-1 Firms by Ranges of the PV and IE Ratios ......... 50 4-2 Matching of Firms on Beta Estimates (B) ......... 55 4-3 Statistics of Beta Estimates Used for Matching ..... 58 4-4 Firms by the Number of Months Covered in Estimating Bj . 60 4-5 Sample Firms Matched by Industries ........... 61 5-1 Test Results on Mean Vectors: Group 3 vs. Group 1 . . . 65 5-2 Test Results on Mean Vectors: Group 3 vs. Group 2 . . . 79 5-3 Summary of Mean Vector Test Results . . . . ....... 87 5-4 Test Results on Covariance Matrices: Group 3 vs. Group 1 91 5-5 Test Results on Covariance Matrices: Group 3 vs. Group 2 97 5-6 Summary of Covariance Matrix Test Results ...... . . 98 5-7 Test Results on Mean Return Vectors for the Pre-Test Period (1972) ...................... 103 5-8 Test Results on Covariance Matrices for the Pre-Test Period (1972) ...................... 104 5-9 Observed Values of Test Statistics by Groups and by Types of Lease Disclosure ................ 105 5-10 Correlation Coefficients Between High- and Low-Risk Assets Within Groups .................. 107 vi Figure 3-1 3-2 3-3 3-4 5-1 5-2 5-3 5-4 LIST OF FIGURES A Possible Impact of Lease Capitalization on the Pricing of Securities ................. Lease Classification and Grouping of Firms Test Periods Based on Critical Events Grouping of Firms ................... Cumulative Average Return Cumulative Average Return Cumulative Average Return Cumulative Average Return Difference: Difference: Difference: Difference: vii PV Firms . . . . PV and IE Firms PV Firms . . . . PV and IE Firms Page 22 29 43 45 69 73 82 85 CHAPTER I INTRODUCTION The purpose of this research was to examine empirically whether or not the recent decision of the Securities and Exchange Commission (SEC) requiring disclosure of capitalized accounting data for noncap- italized financing leases had any impact on the pricing of securities. This research, then, is viewed as being important as well as timely because such information may be relevant to the recently formed Advis- ory Committee on Corporate Disclosure of the SEC and the Financial Accounting Standards Board (FASB) in their deliberations concerning the lease disclosure issue. Moreover, the results of the study may provide a basis for evaluating the continuing controversies for and against the public disclosure of capitalized lease information. In- deed, if captialized lease disclosure as required by the SEC does have an impact upon the pricing of securities, this would support the argument proposing disclosure of capitalized lease information. On the Other hand, if there is INT such impact, the counter-argument against capitalized lease disclosure will have a sounder foundation in that it is supported by empirical evidence. The philosophy of disclosure, from the standpoint of the SEC, is to provide information that will facilitate rational decision mak- ing by investors and speculators in a free and competitive capital 1 market. In light of this disclosure philosophy, the SEC believes that a: proper role of a regulatory agency is to monitor the accur- acy and availability of such information. In a move to further implement this philosophy, the SEC proposed in June 1973 to amend Rule 3-162 of Regulation S-X calling for improved disclosure of lease information by lessees in their lO-K reports filed with the Commission.3 After evaluating many letters of comment received in response to the proposal, the SEC eventually adopted the extended lease disclosure requirements in Accounting Series Release (ASR) No. 147 in October 1973.4 The Release was effective as of November 30, 1973. In general, ASR No. 147 requires increased disclosure of noncap- italized lease commitments by lessees in a footnote to the financial statements in the lO-K reports. Specifically, the Release calls for the disclosure of certain lease information which was not previously required by either the SEC or other accounting regulatory agencies. The new man- datory items of disclosure concern noncapitalized financing leases. For leases of this type, registrants are required to disclose: a) the present values (PV) of the minimum future lease commitments in the aggregate and by major categories of properties, b) the interest rate(s) used in computing the PV, and c) the impact on net income (income effect: IE) if such leases were capitalized.5 Two reasons for the extended lease disclosure are specifically pointed out in the Release.6 First, the disclosure of both the PV and IE is considered "essential to investors" since this information is felt to be "necessary to enable investors to compare meaningfully the capital and asset structures and the operating results of companies" that have such leases. Second, the existing lease disclosure requirements, in- cluding APB Opinions No. 5 and 31, are not sufficient to provide lease information "as needed by investors." Thus, these statements suggest that the lease data of the type called for in the Release will convey new information (important to investors) about the risk-return pros- pects of those firms which did not previously report such data through accounting sources. And this suggestion in turn implies that security priCes of these firms would be affected by the disclosure of such in- formation through its impact on investors' assessments of the risk- return prospects of the firms. The SEC's view on the importance (to investors) of capitalized lease information is not new. Rather, it can be regarded as a rein- forcement of the traditional arguments that insist on disclosure of such information in financial reports. For instance, Myers in Account- ing Research Study No. 4 recommends extended disclosure of capitalized lease information.7 He claims that, since lease commitments create property rights (assets) and liabilities on the part of the lessee, "the present value of contracted lease payments should be placed among the assets and liabilities on a balance sheet"8 and, therefore, that there should be "a periodic charge to income both for the inter- est element of each rental payment and for the periodic amortization of the asset value".9 It is implied in this argument that the exis- tence of property rights and obligation for those rights will affect the financial ratios which may be used by investors in assessing the risk-return prospects of the firms with lease commitments. Therefore, disclosure of capitalized lease data is important to investors. In addition to the above arguments, both APB Opinion No. 31 and the FASB Exposure Draft on accounting for leases agree that sufficient information regarding lease commitments can help users of the financial statements in assessing the financial conditions and operating re- sults of lessees.10 Similarly, the Subcommittee on Leases of the AAA Committee on Financial Accounting Standards states that disclosure of captialized lease information will improve "the information content "‘1 In addition, Vatter (1966), Buff of the financial statements. (1971), and Huefner (1970) all agree that a lease contract creates a liability so that reporting leases in the financial statements should be treated as such; that is, capitalized lease disclosure is a correct approach to accounting for leases.12 In contrast to the view that capitalized lease disclosure is needed to improve the information content of the financial statements, however, a counter-view asserts that disclosure of capitalized lease information would confuse investors and convey no new information about the value of a disclosing firm.13 According to this view, dis- closure of such information will not improve the information content of the financial statements. This counter-view can be divided into three broad argument categories. The first argument category focuses on the nature of lease com- mitments. According to Cook (1963) and Zises (1961 and 1973), for example, a lease commitment creates neither an asset nor a liability because ordinary financing leases are "executory" contracts which are in essence incomplete transactions and the reporting company has no 14 ownership right in the leased asset. It is also argued that, since the nature of lease commitments is so divergent, capitalization re- duces the amount of information conveyed and uniform disclosure will 15 mask the unique aspects of specific types of leases. Those who hold this view further maintain that the primary use by investors and creditors of information about lease commitments is in the projection of cash flows and in the analysis of a company's ability to meet its fixed charges. The only item of lease information that is useful for that purpose is the amount of gross rentals payable; the capitalized value of these rentals is meaningless. Therefore, any requirement to disclose both is unnecessary and misleading.16 Thus, these arguments suggest that the disclosure of capitalized lease data, such as the PV and IE, of noncapitalized financing leases, as if they did give rise to assets and related liabilities and without consideration of the diversity of lease commitments, would be poten- tially misleading to investors. Accordingly, to the proponents of these arguments the capitalized lease information of the type called for in ASR No. 147 is considered to be of little value to investors in evaluating the risk-return prospects of the firms. A second category of arguments that question the information content of lease disclosure focuses on the practical implementation problems associated with lease capitalization and disclosure. For example, what criteria should be used to identify leases that are to be capitalized? What interest rate(s) should be utilized in deter- mining the PV of lease commitments? Should the amount of rentals used in capitalization be net or gross? Should renewal or purchase options and contingent payments be considered? What information about capitalized leases Should be presented in the body of the financial statements and what information disclosed in the footnotes thereto? These are only partial examples of the implementation problems involved in lease capitalization and disclosure.17 Focusing on im- plementation problems of these types, Defliese warns "..., it would seem inadvisable to require the capitalization of finance-type leases by lessees" and "... it appears most inadvisable to require disclosure of an alternative net income amount on an 'as if' captialized basis" until the complexities of capitalizing leases are satisfactorily 18 resolved. For similar reasons, Donaldson (1962) and Axelson (1971) also suggest that a footnoting of the annual rental payments without their capitalization is adequate.19 Thus, it is implied in this second set of arguments that dis- closure of capitalized lease information may be meaningless until satisfactory solutions to these complex problems are found. Although the SEC provides guidelines on some of the implementation problems,20 the validity of these guidelines still remains untested. For example, the Release requires disclosure of the PV and IE of noncapitalized financing leases in the footnotes rather than in the body of financial statements. However, as the FASB Discussion Memorandum (1974) points out.21 this footnote disclosure requirement may not be meaningful in an efficient capital market in which all available information is fully reflected in security prices. Indeed, security price formation should be unaffected by lease disclosure in the body of financial statements as opposed to disclosure in footnotes. Furthermore, other potential implementation problems, such as how to present information about capitalized leases in the body of the financial statements, are not explicitly dealt with by the SEC. As a result, the SEC's failure to deal adequately with the practical implementation problems associated with disclosing the PV and IE numbers may severely restrict the informa- tion content of the capitalized lease data. A final argument category which questions the information content of capitalized lease data focuses on the possible consequences of disclosure. This argument maintains that the underlying economic realities (such as economic resources and real liabilities, cash flows, and the like) of a reporting company are not affected by the disclosure of capitalized lease data. Accordingly, the disclosure of such data in financial reports will not convey any new informa- tion about a firm's market value to users of the financial statements. Representing this view, Nurnberg argues: ...the underlying economic realities are unaffected [by lease disclosure]; firms still have the same resources and obliga- tions regardless of whether they [leases] are reported as assets and liabilities and regardless of their impact on financial ratios. ...the recognition of leases... as assets and liabil- ities...necessitates educating financial statements users to evaluate firms on the basis of the underlying economic realities, not on the basis of how these realities are reported in the financial statements.22 Given the above outlined conflicting views concerning the alleged information effect of capitalized lease data, it is considered appro- priate to evaluate empirically which of the two views is more consis- tent with observed empirical phenomena. If capitalized lease data convey useful information to investors in evaluating a firm's market value, as both the SEC and proponents of lease disclosure believe, then there should be an observable market reaction to the disclosure of capitalized lease data (the PV and IE) under ASR No. 147. And the extent to which such market reaction exists can be utilized as a means of assessing the information content of the capitalized lease data and the effect of the SEC lease disclosure decision on the pricing of securities. How- ever, if capitalized lease data convey no new information about a firm's risk-return prospects, as the counter-view asserts, then there should be no empirically observable market reaction to the event of 8 capitalized lease disclosure as required under ASR No. 147. No mar- ket reaction implies no information content of capitalized lease data and this, in turn, implies that the SEC decision had no impact upon the pricing of securities. As Gonedes and Dopuch (1974) note, an empirical assessment of the effects of accounting regulations such as the SEC lease dis- closure decision can be justified on the following grounds: Since assertions about effects are important parts of the justifications offered for recommendations and prescriptions, we can assess the strength of these justifications by evalua- ting the theoretical or empirical support for the assertions about effects....In short, assessments of the effects of al- ternative accounting procedures and regulations can be useful to accounting policy-making bodies in making their decisions and to their constituencies in evaluating those decisions.23 Nevertheless, very little empirical research has been done to assess the information effect (upon security pricing) of capitalized lease data or lease disclosure regulations by regulatory agencies, especially in the context of a market-based approach. This study consists of six chapters. A detailed examination of the SEC's extended lease disclosure requirements set forth in ASR No. 147 is provided in the first part of the next chapter. The latter part of the chapter discusses the possible effects on financial ratios of the capitalized lease data disclosed according to the Release and their potential relationship with security prices. Chapter III describes how the impact of capitalized lease data upon the pricing of securities can be evaluated both theoretically and empirically. The first section of this third chapter will develop an analytical framework for evaluating the information effect of lease disclosure, while the remaining sections will describe how the information effect can be investigated empirically. The investigation method employed in the present study wasa multivariate approach. Chapter IV provides descriptions of the sampling design used for selecting sample firms and for collecting stock price data, and an analysis of all test re- sults is presented in Chapter V. Finally, the concluding chapter will investigate the implications of the findings and suggest avenues of further research.24 CHAPTER II THE NEW LEASE DISCLOSURE REQUIREMENTS AND EFFECTS ON ACCOUNTING VARIABLES This chapter begins with a brief historical review of accounting regulations for disclosing leases in financial reports. HOpefully, this review will facilitate an understanding of the motivation under- lying the SEC'S new lease disclosure decision of 1973 (ASR No. 147). The second section describes the new extended lease disclosure require- ‘ments set forth in ASR No. 147 and notes in particular the incremental lease information (the PV and IE) over and above the disclosure called for in APB Opinion Nos. 5 and 31. The final section examines the pos- sible effect that the PV and IE disclosure will have on various finan- cial ratios used in traditional financial analysis and on the pricing of securities through its impact on the financial ratios. Historical Review of AccountinggRegulations for Reporting Leases The AICPA Pronouncements The first official pronouncement designed to improve disclosure by lessees of long-term leases in financial reports was initiated in 1949 when the Committee on Accounting Procedure of the American Insti- tute of CPAs issued Accounting Research Bulletin (ARE) NO. 38, "Dis- 1 closure of Long-Term Leases in Financial Statements of Lessees." A 10 ll primary recommendation in the Bulletin was that a lease which is in substance a purchase should be capitalized.2 However, a general consensus was that the relatively simple lease disclosure require- ments3 set forth in the Bulletin were considered inadequate to meet the need for more information concerning increasingly complex lease commitments. In 1962 the AICPA published Accounting Research Study (ARS) No. 4, "Reporting of Leases in Financial Statements," in which Myers recommended more lease capitalization and extensive disclosure, in- cluding those leases which may not be in substance purchases.4 Myers‘ recommendation was rejected by the Accounting Principles Board of the AICPA. In its Opinion No. 55 the Board stated that capitalization should be limited only to those leases which are in substance purchases6 and, thus, reinforced the basic position originally stated in ARE NO. 38. As a result, the Board's concern centered on specifying the criteria necessary to identify leases which are in effect installment purchases of property. Thus, a basic criterion suggested in the Opinion was whether or not the terms of a lease result in the creation of a mater- ial equity in the property.7 Despite the Board's effort to improve reporting of leases, a general consensus was that the contents of the Opinion were inadequate; the term "material equity" was ambiguous and did not lend itself to easy practical application. Furthermore, the Opinion failed to provide adequate information about leases that are not in substance purchases. Disclosure (in schedules or notes) of minimum annual rentals of such leases and the related lease periods were recommended.8 However, the 12 method of disclosing such supplemental lease information, as recom- mended in the Opinion, was criticized for being too general. For example, the Opinion stated that: "The specific details to be dis- closed and the method of disclosure will vary from one situation to another depending upon the circumstances."9 In June 1973 the Board issued its Opinion No. 31, “Disclosure of Lease Commitments By Lessees," which was designed to provide more complete information with respect to noncapitalized lease com- mitments. The new Opinion requires the following major disclosures with respect to a firm's lease commitments:10 (1) Total rental expense (reduced by rentals from subleases) with disclosure of such amounts for each period for which an income statement is presented and with contingent ren- tals disclosed separately. (2) The minimum rental commitments under all noncancelable leases,H in the aggregate and by major categories of pro- perties, for a) each of the five succeeding fiscal years, b) each of the next three five-year periods, and c) the remainder as a single amount. (3) Additional disclosures concerning a) a basis for calcula- ting rental payments if non-time factor is used as the basis, b) renewal and purchase options, escalation clauses, etc., c) related guarantees made or obligations assumed, d) restrictions on paying dividends, incurring additional debt, further leasing, etc., and e) any other necessary information. In addition to these mandatory disclosure requirements, the new Opinion recommended disclosure of the PV of future minimum noncancel- able lease commitments, reduced by the PV of rentals to be received from existing noncancelable subleases and with disclosure of the dis- count rate(s) used in computing the PV. It should be noted, however, that the PV disclosure is optional rather than mandatory. Because of this negligence in mandating disclosure of such information, APB 13 Opinion No. 31 has been criticized for failure to achieve its stated goal "to disclose sufficient information regarding noncapitalized 12 that investors and other users of financial lease commitments” statements believe to be important. hkn:surprisingly, there was little change in the general tendency of lease capitalization in practice even after the Opinion went into effect (as of December 31, 1973). According to the AICPA survey, only 7% of 600 sampled firms in 1973 and 8% in 1974 capitalized all of their long-term leases. For the same years, the firms which capitalized only part of their leases comprised 20% and 22%, respectively.13 The SEC Regulations Before ASR No. 147 The SEC has made several official pronouncements concerning lease disclosure in S-X Regulations or in the form of Accounting Series Releases. Basic rules for disclosure of long-term leases and commitments are found in Rule 3-16 of Regulation S-X. Regarding noncancelable leases which are not capitalized, Rule 3-l6(i),"Com- mitments and Contingent LiabilitiesJ'requires disclosure of the following information, if annual rentals under such leases are in excess of 1% of total sales and revenues of the most recent fiscal year: (1) the minimum annual rentals for the current and each of the five succeeding years; (2) the nature and effect of any provisions that would cause the annual rentals to vary from the minimum rentals; (3) a description of the types of property leased, important obligations assumed or guarantees made, and any other significant provisions of such leases. In addition, Rule 12-16 requires disclosure of rent expenses. As l4 seen above, however, the Rule does not call for disclosure of capit- alized information of noncapitalized leases. In November 1972 the SEC issued ASR No. 132,”Reporting of Leases in Financial Statements of Lessees." However, the Release was limited mainly to clarifying ambiguities in APB Opinion No. 5 so that it was in effect an interpretation of provisions in this Opinion. For example, much of the discussion in the Release was concerned with how to determine economic substance in order to iden- tify leases which should be capitalized and those which should not. SEC'S Assessment of Previous Regulations The preceding discussion provides a brief review of the offic- ial lease disclosure pronouncements which went into effect prior to the SEC's 1973 extended lease disclosure pronouncement (ASR No. 147). In light of the reasons for this new pronouncement, which were pointed out in the previous chapter, by implication the SEC took the position that the prior pronouncements on lease disclosure failed to achieve their stated objective--to provide sufficient information con- cerning lease commitments that users of financial statements believe to be important. For example, the SEC stated that: The Commission has carefully considered the contents of Opin- ion No. 31 to determine whether it provided for sufficient dis- closure to meet the needs of investors and has concluded that it does not,...the Commission believes that disclosure of the pre- sent value of financing leases and of the impact on net income of capitalization of such leases, neither of which is required by Opinion No. 31, are essential to investors.14 ~Thus, this statement can be interpreted to mean that disclosure of the capitalized lease data was not mandatory, but optional under those previous disclosure regulations, a situation which was not TS satisfactory from the standpoint of the SEC. Therefore, the general public had no access to capitalized information (such as the PV and IE) of noncapitalized future lease commitments through accounting sources. As a result, the SEC felt that there was a need for more ex- tensive and mandatory capitalized lease disclosure. Furthermore, it believed this need to be rather urgent stating that "---it is not in the interest of investors to delay additional disclosure requirements 15 any further." In response to this pressing need for extended lease disclosure, the SEC formally issued ASR No. 147 in October 1973, the contents of which are examined in the next section. However, the SEC has indicated that the disclosure requirements in the Release may be reconsidered if the FASB provides satisfactory guidelines for lease l6 disclosure in the future. Presently, accounting for leases is again 17 under comprehensive examination by the FASB, but no formal statement has been adopted at this time. The Lease Disclosure Requirements in ASR No. 147 The SEC's new lease disclosure requirements are described in section "C. Amendments to Regulation S-X" of ASR No. 147. Part of the disclosure requirements in the Release are reaffirmations of the three categories of diSclosure requirements in APB Opinion No. 31 (as noted previously) concerning total rental expense, minimum rental commitments under all noncancelable leases, and five additional dis- closures. Also, the Release defines a noncancelable lease as it is 18 defined in the new Opinion. As compared to APB Opinion No. 31, the Release provides more l6 refined guidelines regarding the first two disclosure items (total rental expense and minimum rental commitments). First, according to the Release, these two disclosures are required only when gross ren- tal expense exceeds 1% of consolidated revenues.19 However, no guide- line concerning the minimum amount of gross rental expense for dis- closure purposes is provided in the Opinion. Second, unlike the Opinion, the Release requires a separate indication of noncapitalized financing leases when the two items, total rental expense and minimum rental commitments, are disclosed.20 Third, the Release gives a defin- ition of a financing lease which is not found in the Opinion. In the Release a financing lease is defined as one that, during the noncan- celable lease period, either (1) covers 75% or more of the economic life Of the property or (2) has terms which assure the lessor a full recovery of the fair market value of the property at the inception of the lease plus a reasonable return on the investment in the pro- perty.2] In addition, ASR No. 147 calls for the disclosure of the follow- ing four items concerning noncapitalized financing leases22 which previously were not required under any official accounting pronounce- ment: (1) the PV of the minimum future noncapitalized financing lease commitments in the aggregate and by major categories of properties; ' (2) either the weighted average interest rate and range of rates or specific interest rates used in computing the PV of such leases; (3) the PV of rentals to be received from existing subleases; (4) the IE for each period for which an income statement is presented if such leases were capitalized, related assets were amortized on a straight-line basis and interest cost was accrued on the basis of the outstanding lease liablilty. 17 However, the Release states that, if the PV of the minimum lease commitments is less than 5% of the amount of long-term capital- 23 and if the IE is less than 3% of average net income for ization the most recent three years, then the above disclosures are not re- quired. That is, these two cut-off points, 5% for the PV disclosure and 3% for the IE disclosure, are specified by the SEC as criteria for judging the materiality of noncapitalized financing leases for disclosure purposes. The PV and IE information of noncapitalized financing leases is incremental in that its disclosure was not required until ASR No. 147 went into effect. Apparently, the SEC believed that this information is essential to investors in assessing a firm's market value and, therefore, that its public disclosure should be mandatory. Accordingly, the main question of concern in the present study is whether this incremental lease disclosure provides any new informa- tion which is pertinent to investors in evaluating the risk-return attributes of the disclosing firms. Effects on Financial Ratios Appendix A shows the financial statements of a hypothetical firm which is assumed to be materially24 affected by the SEC's new lease disclosure decision. Notice how the financial statements differ: those statements prepared according to ASR No. 147 compared to those not so prepared. If one assumes that the PV is part of liabilities and that the IE is an adjustment to reported net income, it is apparent that most financial ratios used in the traditional analysis of the firm will 18 be adversely affected. As shown in Appendix B, for example, the debt-equity ratio for 1974 of the hypothetical firm will be adversely changed from .52 to .66 if the PV of leases is included in long-term debt as a liability. Since the IE results in a reduction of reported net incone (which would be the typical case for most of the firms), the net income to stockholders' equity ratio will decrease from .05 to .04 and earnings per share will be reduced from $2.65 to $2.36. In fact, the extent to which the financial ratios are affected will be greater than this simple example suggests. For example, the amount of inventory may change due to depreciation adjustment for the leased property if it is used for a manufacturing purpose. Con- sequently, the magnitude of current assets as well as net income will change. Also, the amount of current liabilities as well as long-term debt will be affected if part of the lease payments is due by the end of the next fiscal year. A periodic charge to income both for the in- terest factor on the outstanding lease obligations and for the depre- ciation of the leased assets will change net income. Similarly, many other items in the financial statements will change, and a series of such changes will eventually lead to an associated change in almost every financial ratio. The effects on financial ratios illustrated in this simplified case are representative of the sort of impact that lease capitaliza- tion has upon firms' financial ratios. In a study of the effects of leeise_capitalization on financial ratios, Nelson (1963) found that Icapitalizationflof'leases generally affects financial ratios in an adverse manner. He examined fifteen financial ratios of eleven lessee saflflJle firms before and after lease capitalization. Those ratios that 19 he examined included current ratio, debt to equity, debt to total capital, return on total capital, fixed assets to tangible net worth, inventory to net working capital, net working capital to net plant, net plant to sales, and seven other ratios. In the case of certain sample firms, the change in the debt to equity ratio ranged from 8.32% (75.7% before lease capitalization to 82.0% after capital- ization) to 416.45% (15.8% to 65.8%). The change in the debt to total capital ratio varied from 0.99% (80.7% to 81.5%) to 289.78% (13.7% to 39.7%). AS compared to these two ratios, the current ratio was shown to be relatively less affected by lease capitalization with a range from 0.88% (1.14 times before capitalization to 1.15 times after capitalization) to 37.90% (3.43 times to 2.13 times). On the other hand, ten out of eleven sample firms experienced a favorable change in the ratio of net profit to net working capital ranging from .69% (14.5% to 14.6%) to 61.92% (32.3% to 52.3%). However, the re- turns on total capital of six sample firms declined from approximately 8% to 7% as a result of lease capitalization, while the returns for the remaining five firms increased from about 4% to 5%. The final result was that twelve financial ratios, including current ratio, debt to equity, and debt to total capital, were adversely affected as a result of capitalization, while two ratios (times inter- est charges earned and net profits to net working capital) improved. The effect of capitalization on the return on total capital was mixed. Thus, Nelson concluded that, in general, lease capitalization adversely affects a firm's financial position. He concluded that with capitalization the financial ratios become more meaningful because 25 the ratios meet their "objectives" more effectively. He argues that 20 "the financial analyst could easily have made faulty decisions--if he had based his analysis on ratios"26 which do not incorporate lease capitalization. This inference was based upon changes in the ranking of firms according to the magnitude of each of the financial ratios as a result of capitalization. According to Nelson, therefore, cap- italized lease data have information content since such data are be- lieved to improve the information content of the financial ratios. However, there is counter-evidence that lease capitalization does not improve the information content of financial ratios when the information content is measured by some other event. Elam (1975) examined the effect of lease capitalization upon the ability to pre- dict firms' bankruptcy using financial ratios computed after incor- porating capitalized lease data. He selected forty-eight bankrupt firms as a treatment group for the period 1966 through 1972 and matched them with non-bankrupt firms individually. Then, he compared twenty- eight financial ratios of the two types of firms over five years prior to bankruptcy and found that the inclusion of capitalized lease data in a firm's financial statements did not increase the power of finan- cial ratios to predict bankruptcy. Given the above evidence, the next question is whether the effect of lease capitalization Upon financial ratios will in fact affect investors' perceptions of the firm. As seen earlier in Chap- ter 1, a_prjgri_arguments by the proponents (including the SEC) of caPitalized lease disclosure suggest that investors' perceptions will be affected by disclosure of capitalized lease data such as the PV 3"C1 IE. However, the counter-arguments by those who question infor- mation content of capitalized lease data suggest that investors will 21 not reassess the value of the firm on the basis of the financial ratios affected by lease capitalization because the underlying econ- omic realities of the firm have not changed or because many unresolved implementation problems cause capitalization to be meaningless. If there is an association between the financial ratios and a neasure of the riskiness of a firm's security and if investors use these ratios as instrumental variables (i.e., as surrogates for the risk measure through which the riskiness of the firm can be assessed), then capitalized lease disclosure will affect investors' perceptions about the riskiness of the firm. The reason for this predicted reassessment of risk is obvious: the financial ratios after lease capitalization will differ from such ratios before capital- ization. A change in investors' perceptions will in turn lead to a readjustment of equilibrium security prices. Beaver, Kettler, and Scholes (1970) argue that accounting risk measures can be viewed as surrogates for the systematic (market) risk measures, and they found empirical evidence to support this argument. They selected 307 sample firms for the years 1947 through 1965 and examined an association between seven accounting risk measures (dividend payout, growth, leverage, liquidity, asset size, variabil- ity of earnings, and covariability of earnings) and a market risk measure (Beta). Among the seven accounting risk measures, earnings variability appeared to have the highest rank correlation (.90 for 1947-56 and .82 for 1957-65) with the market-determined risk measure. The next highest correlation was present between payout ratio and the same market risk measure (-.79 for 1947-56 and -.50 for 1957-65). Such ratios as growth, liquidity, and size had relatively low 22 correlations with the market risk measure. Regarding the direction of correlation, three ratios (growth, leverage, and earnings variability) were found to have positive correlations, while two ratios (payout and size) had negative correlations. A conclusion on the liquidity ratio was indeterminate because the sign of its correlation coeffic- ient was negative for the period 1947-56 and then positive for 1957- 65. As a result of these findings, they concluded that "accounting measures of risk are impounded in the market based risk measure"27 and, therefore, that there is a high degree of association between the accounting and market risk measures. A similar study was conducted by Hamada (1972) who examined the effects of changes‘in the firms' capital structures upon the system- atic risk of common stocks by the use of 304 sample firms selected for the years 1948-1967. He utilized leverage (debt-equity) ratio as a capital structure variable and found that approximately 21 to 24% of the observed systematic risk of common stocks can be explained by the leverage variable. As seen earlier in this section, lease capitalization will apparently affect those financial ratios which were used in the above two studies. Thus,the results of these studies suggest that capital- ized lease data, such as the PV and IE disclosed under ASR No. 147, may have a relationship with the systematic risk of a firm through the effect of such data on the financial ratios. The existence of this relationship further implies that there could be a relationship between the capitalized lease data and security prices. And to the extent that such a relationship exists, security prices can be used as a means to evaluate the information effect of the SEC lease dis- closure decision. CHAPTER III ANALYTICAL FRAMEWORK AND METHODOLOGY FOR EVALUATING THE EFFECT OF CAPITALIZED LEASE DISCLOSURE This Chapter consists of three sections. The first section describes an analytical framework for evaluating the effect of cap- italized lease disclosure upon the pricing of securities. The second section is devoted to discussions about the testing hypotheses and the multivariate test design used to investigate the hypothesized lease information effects. And in the final section, a description of the test procedures is provided. Assessment of the Effects of Lease Information on Equilibrium Rates of Return Effects of Lease Information The effect of new information upon equilibrium rates of secur- ity return can be assessed by comparing the return distribution functions conditional upon new information with the corresponding un- conditional return distribution functions. If new information has an effect upon equilibrium rates of return, the two distribution functions will not be equal. Other things being equal, the inequal- ity of these two functions can thenlxeattributed to the effect of new information. 23 24 In making the assessment of information effect on the pricing of securities, the present study assumed that: (l) the vector of rates of return on all assets has a multivariate normal distribution and (2) investors agree upon the multivariate normal distribution of returns (i.e., investors make their investment decisions on the basis of the prediction of the mean and variance of this distribution). With these assumptions made, the effect of new information was eval- uated within the context of a multivariate framework.1 The capitalized lease information (the PV and IE) disclosed under ASR No. 147 is viewed as one type of potentially new infor- mation available to investors. Thus, the effect of the capitalized lease information upon equilibrium pricing of securities can be eval- uated by a comparison of the return distribution functions conditional upon the lease information with the unconditional return distribution functions. Let fijt be the rate of return on asset j at time t where j = 1, 2, ..., n and t = 1, 2, ..., N. Further, let gjt be a random variable representing the capitalized lease information generated by a materially2 affected firm j at time t (affected by the SEC lease disclosure decision). Suppose that fit and E; are n-component column vectors as given by A (R N 9819...,{6 -t _ a 1 B ' (Blt’ 32H" J nt) Then, given some realization (g) of §_(with time subscript t omitted), R. ...,R )1 lt’ 2t"°" jt’ nt -t lease information §_is said to have information content if F(R/g) # F(B) where "F" stands for the distribution functions of 11;. Under the multivariate normality assumption, the above condition 25 for an information effect is equivalent to E(E/§) f E(E) and/or Var(E/§) f Var(B); that is, the mean and/or variance of the return distribution conditional upon the capitalized lease information dis- closed under the Release will not be equal to the mean and/or var- iance of the unconditional return distribution if lease information Q has information content. The reason for investigating a change in mean and/or variance is that a normal distribution can be fully described by these two parameters.3 Therefore, if one or both of these parameters have changed as a result of lease information effect, then either the means or variances (or both) for the two return dis- tribution functions will not be equal to each other. Any such in- equality is sufficient, by definition, for a lease information effect to exist. The mechanism by which lease information g_affects equilibrium rates of return can be explained by the two-factor capital asset pricing model (CAPM). This model was originally developed by Black (1972) on the basis of the works of Sharpe (1964), Lintner (1965), Mossin (1966), and Fama (1968). The descriptive validity of the model is well documented by Black, Jensen, and Scholes (1972), Miller and Scholes (1972), and Fama and MacBeth (1973). The form of the model is given by Edy E(1€z). EON). s) = 12022) I + (HAM) - E111 )) B (3-1) Z where E(RZ) = expected return on a "zero Beta” asset with COV(HZIHM) = O I = an n-component unit column vector E(RM) = expected return on the market portfolio which consists of all securities in the market 26 B_= an n-component column vector defined by Cov(E_RM)/Var(RM) which measure the relative risks of the assets. According to the model, equilibrium expected rates of return M), and g. If lease information §_has an effect on E(E) through its impact on either E(ED are determined by three factors: E(RZ), E(R §_or residual returns (E) (or both), Tidy HRZ). RAM). 6,. o) r 58/ £612). HIRM). B) (3-2) which is one sufficient condition for a lease information effect to exist. If E(E) is affected by Q, then Var(E) may also be affected be- cause, by definition, Var (R) = E(RER) - (E(E))2, and thus, lease information B may affect the variability of returns through its im- pact on E(:R'_'_R_) and/or (E(E))2. Then, the result will be an inequal- ity between the variance of E_Conditional upon §_and the unconditional variance: Var(i{/§_) f Var(:R_) where Var(R/_6_) = E(_R_' R/g) - (E(I_R_/§_))2. This inequality is another condition for the existence of a lease in- formation effect. Figure 3-1 explains how the disclosure of the PV and IE account- ing numbers in financial reports may have an effect on the distribu- tion of security returns through the impact of such disclosure on financial ratios and thus investors' assessments of the firms' risk- return prospects. Equilibrium rates of return (B/Q) are dependent upon investors' actions (a/g) which in turn depend upon their percep- tjons F(§/§) about the firms' risk-return prospects S, Again, these perceptions could be affected by the disclosure of the capitalized lease data (9) generated through accounting system (D) that is re- quired under ASR No. 147. 27 Accounting for Leases F | —l n Capitalization Noncapitalization 1 and disclosure and no disclosure F I F 1 Change in Change in net No change in Change in current and income due to current and net income long-term rent, depre- long-term due only to liabilities ciation, & in- liabilities rental and assets terest cost and assets expenses L I J l I l g The extent to which financial ratios are -— affected may differ, depending on whether or not the PV and IE numbers are disclosed. +-—F(§) l F(§/§,D) Investors' perceptions about the firms' risk- return prospects may change if the capital- ized lease information is disclosed. 17g Investors' actions may differ in the market. 1 3/9_ A possible result is different equilibrium security prices. Figure 3-1. A Possible Impact of Lease Capitalization on the Pricing of Securities Thus, if §_has an information effect, investors' predictions F(§) of the affected firms' §_prior to the release of §_will differ from the corresponding predictions of §_after the release of 9, That is, F(§/§) f F(§). . This inequality condition, then, allows for the possibility that investors' actions conditional on §(_/§) will differ from those actions a_that would otherwise have taken place. The SEC's assertion that §_is essential to investors implies, in fact, that a/§_f a, FInmthermore, if action differences do occur as a result of 9, then dif- ferences in equilibrium rates of return may also result in the sense that equilibrium returns prior to g_are not the same as equilibrium 28 rates after the release of 0, Control by Matching In conducting a test for the information effect of B, one would ideally like to have two types of return data for the same firm for the same time period: return data generated conditional on g_and un- conditional return data. However, it is impossible to obtain such data since only one return observation is available for a firm for a given time period. One approach in attacking this data problem is to use control firms. In the present study two types of control firms were used: (1) non-lease firms (Group 1) that had no long-term leases and there- fore were unaffected by the SEC lease disclosure decision and (2) non- disclosure firms (Group 2) that indicated they had noncapitalized financing leases but which did not meet the SEC's two materiality criteria (5% for the PV disclosure and 3% for the IE disclosure) and therefore did not disclose the PV and/or IE numbers in their financial reports. Thus, there were three different groups of firms used (See Figure 3-2), including the lease disclosure (treatment) firms (Group 3) that met the materiality criteria and therefore dis- closed the PV and/or IE numbers in their 10-K reports for 1973 and 1974. In order to see how the introduction of control firms operates in evaluating lease information effects, let the treatment and control firms be denoted by T and C, respectively. Suppose that the control firms in Group 1 and Group 2 are selected according to their relative risks such that, in the pre-test period, ET = BC at the individual firm level and BT = Beat the group level. The reason for equating 29 Capitalized Material Group 3 Financing Noncancelable 193595 leases / Noncapitalized Leases Operating and ( other leases Immaterial———Group 2 No noncancelable leases (but cancelable leases) Group 1 No leases Figure 3-2. Lease Classification and Grouping of Firms Betas is that, as described by the CAPM, the relative risk is the only factor that can be different in the pricing of securities between the treatment and control firms since E(RZ) and E(RM) are common to all firms. Then, the first moments of return distributions between the two types of firms will be the same before any effects of lease infor- mation are incorporated into equilibrium returns, if it is assumed that return observations are taken from the same time period so that R2 and RM are controlled for the two types of firms. Since 8; = BC by construction, all return-determining factors as defined in the CAPM are the same between the two groups of firms prior to the disclosure of 0,4 Thus, HBT/ Hill). HAM). 6T) = HBC/ HBZ). URN). BC) (3-3) Now, suppose that new lease information §_is disclosed to market agents and that this information is pertinent to a valuation of the exDected returns of the treatment firms. Then, the conditional ex- PeCted returns for these firms will become (3-4) and the relation between E(_1§T) and NBC) in (3-3) will hold as an inequality as shown 30 in (3-5) below if.g has an effect on the conditional expected returns through its impact on §# or residual returns (2%) (or both). HGT /_6_T) = HBT/ will. HAM). 6T, 9,) (3-4) “ET/2T) 7‘ E(R’C) (36) Likewise, one can expect to observe the following inequality be- tween the two conditional and unconditional variances of returns if the information effects exist: var (ET/9T) = RB“ 31,/9,) - (BET/9,112 a Var(§cl = Elsi: RIC) - (5(9an (3-6) The existence of the hypothesized lease information effect does not require the inequality conditions, (3-5) and (3-6), to hold at the same time; either one of them is sufficient. Two-Asset Assumption 0 0 6 5 some prev1ous studies assumed As Gonedes (1975) points out, that the equilibrium prices of all types of assets are affected by a 'given type of information in the same way. Thus, they treated all assets as one single sample (asset) and postulated that individual return observations drawn from this single sample were independent. And, under the assumption of a univariate return distribution effect, tests for an information effect were conducted on the returns on one selected asset. The test results on the selected single asset were then generalized to all assets. However, as Gonedes (1975) argues,7 the distributions of returns on different types of assets may be affected differently by the same source of information. Also, return observations may not be indepen- dent of each other. Under this situation, a univariate test based on 31 a single sample is inappropriate. The reason for this inappropriate- ness is that, if the sample is not independent, the return observations must be drawn from at least two different return distributions. Thus, conducting a univariate test under such circumstances is equivalent to trying to make inferences about a single return distribution on the basis of return observations from at least two different distributions. In addition, a single-sample test may cause the hypothesized informar tion effects on different assets to average out to no effect. There- fore, the results from a single-sample test could conceal information effects that are in fact Significant. Furthermore, the covariability of returns (if not independent) may mask the unique effect of new in- formation since a univariate test based on a single sample does not take into consideration the cross-sectional Correlations of individ- ual returns. Based on the assumptions that the SEC lease disclosure decision might affect firms in different risk classes differently and that returns on individual assets might be dependent, the present study dealt with two different types of assets. This is equivalent to say- ing that there are only two types of assets in the market and that all assets in the market can be classified into the two types. The main reason for this present study's assumption of only two types of assets was to gain degrees of freedom with a relatively small number of sample firms.8 Given that the CAPM holds, the relative risks (8) of the assets were used as an attribute on which the assets were classified into two types, high and low risk. The classification of all assets into the two risk classes is a simplification, but it was believed that 32 tests on the two assets are sufficient to detect any risk dependency of lease information effect.9 Moreover, if firms (or assets) are divided into two (or more) groups based on an attribute such as the relative risk, firms within a group are more homogeneous with respect to this attribute, while firms in different groups become more heter- ogeneous. This grouping of firms will improve the power of a test when the test is designed to detect a between-group difference to the extent that there is a risk dependency of lease information effect. Gonedes and Dopuch (1974) and Gonedes (1975) note that classify- ing firms according to their relative risks is appropriate in investi- gating the effect of new information on the values of firms because this attribute is generally determined by firms' production-invest- ]0 The theoretical discussion about how ment and financing decisions. the relative risk is related to firms' production-investment and fin- ancing decisions can be found in Fama and Miller (1972: Chapter 7) and Hamada (1972). For example, the relationship between a firm's production decisions and its relative risk can be explained by the facts that each of the firm's productive activities has some probability dis- tribution of future market value, and that the market value of the firm is simply the sum of the market values of its separate produc- tion activities. The existence of such a distribution implies that there exist the expected returns on individual activities and the related risks. In an equilibrium market, the risk is measured by its contribution to the dispersion in the market value of the market portfolio,and not to the firm itself. The risk of the firm is the weighted sum of the risks of the individual activities, the values 33 of which are to be maximized through the optimal production decision on the basis of the equilibrium relationship between their expected returns and risks. This optimal production will then lead to the max- imization of the market value of the firm as a whole. Based upon the two-asset market assumption, the PV (and IE) disclosure firms of Group 3 were ranked on the basis of their estimated Betas (E) and then equally divided into two subgroups, high and low risk. Firms of Group 1 and Group 2 were individually matched with the disclosure firms in Group 3 according to their Beta estimates and then divided into the same two risk groups (not on the basis of their own ranked §_but on the basis of their original matching with the disclosure firms). The Beta estimates were available from Security_Risk Evaluation (September 1973) published by Merrill Lynch, Pierce, Fenner and Smith, Inc. The Betas were estimated as of the end of August 197311 by using past sixty monthly returns with the single factor market model.12 After the two risk groups were constructed, a two-component re- turn (column) vector, (R: Rt) where H and L denote the high and low risk groups, was computed by taking the arithmetic (equally-weighted) average of the component individual returns in each of the two risk groups. Thus, each monthly average return was used as a sample unit. and it was assumed that monthly average returns are independent over time.13 Under the multivariate normality assumption, the return vec- tor is a normally distributed random vector so that the investigation of a change in the mean return vector and the variance-covariance matrix of returns is sufficient in order to evaluate the effect of lease information. 34 Testing Hypotheses and Evaluation Methods Testing Hypotheses Given the assumption of two-asset market, the return vector (RH RL) has a bivariate normal distribution with the two-component mean return (column) vector U and the 2 x 2 covariance matrix 2, each defined as Var(R”) Cowl?” BL) ‘ Cox/(RH AL) Var (N) Then, the effects of lease information can be assessed by a H u = (U UL) and z = comparison of the conditional mean return vector and covariance matrix for the treatment firms with the unconditional mean return vector and covariance matrix for the control firms. Thus,the conditions for lease information effects are UT # UC and/or 2T f EC’ The test for the equality of the two mean return vectors was conducted by the use of return differences defined as 3“ R” "R“ a: 2+ = RL " M T C since the treatment and control firms are individually matched and thus not independent of each other: that is, the firms are matched on the basis of §_and return observations are taken from the same time period with R2 and RM common to all firms. The multivariate normality assumption implies that, if R's are normally distributed, 3's are also normally distributed. Thus, the null hypothesis for the mean return difference vector test is given by 35 Testing for the equality of the two covariance matrices is equiv- alent to evaluating the ratio of two covariance matrices using two dependent samples.14 In this case it is assumed that each of the depen- dent samples is drawn from its own normal population with its own var- iance, 2T and 2c. The values of their means, UT and UC’ are not of interest. Then,the null hypothesis for the covariance matrix test can be written as: H02: 2T = ZC or Var(RH) Cov(RH RL) _ Var(RH) Cov(Rfl RA) Cov(RH RA) Var(RL) T Cov(RH RA) Var(RL) C As indicated earlier, the existence of lease information effect does not require the inequality of all component means of the two mean vectors and/or all component variances and covariances of the two co- variance matrices. The inequality of any one component of the two mean vectors and/or the two covariance matrices is sufficient. Given the null hypotheses stated in terms of the mean vectors and the covariance matrices, one may test the equality of each paired com- ponent separately using a univariate test method. However, one prob- lem with a univariate approach is that the joint level of significance (or confidence interval) which is desired cannot be determined exactly by the repeated use of a univariate test.15 Thus, to ensure a desired joint level of significance, one needs a multivariate test method which examines the equality of all the component means (or component variances and covariances) simultaneously. Moreover, a multivariate approach is consistent with the multivariate normality assumption which was intro- duced earlier in this chapter. As a multivariate test method, the present study used Hotelling's 36 2 T statistic for the mean vector test (H01) and Box's modified M statis- tic for the covariance matrix test (H02). Tests on Mean Vectors The objective of tests on mean return vectors is to see if the first moment (mean) of the security return distribution for the treat- ment firms has shifted as a result of the capitalized lease disclosure. With the introduction of a control group, testing for a shift in the first moment of return distribution is equivalent to testing for whether or not the mean return difference vector is equal to the null vector (with zero components) as stated in H0]. The null hypothesis for the mean vector test was tested by using Hotelling's T2 statistic16 which is essentially the squared standard t-statistic computed on the basis of a weight vector which maximizes the value of the squared t-statistic. In order to see how the process of maximizing the squared t-value leads to the derivation of the T2 statistic, let W_be a weight (nonull column)vector such that W_= [WH WL]' where WH and WL are weights of the high and low risk assets, respectively. Suppose that this weight vector is utilized in generating a weighted sum of two components of the return vector. Then, W] §_has a univariate normal distribution17 with mean W] §_and variance-covariance W} SdW_where Sd is a covariance matrix computed on §_(not on E). Since W_f Q_by construction, the null hypothesis H01 can be restated as H 99. = B'luT-U = 0 OA‘ c) For a given value of the weight vector, the null hypothesis H0A is a univariate hypothesis. The acceptance region for testing this 37 hypothesis is given by tzm where t2(W) IA th (3-7) the squared t-statistic conditional for W_and th = a critical value. This relation is satisfied for all W_f Q_if and only if max t2 ( W W) 3 th (3-8) Maximizing (3-8) subject to the constraint WdeW_= C where c is 2 18 some prespecified constant (usually unity) yields the T statistic that is called Hotelling's T2. In the computation of the value of 2 _the T statistic, all possible values of W_are checked and its one value which maximizes t2 is selected. Thus, except for the distinc- tion of the maximization process, the multivariate T2 test is similar to the standard univariate t-test.19 The form of Hotelling's T2 statistic is given by T -fixtW)~NQi CisdIC 5p - (3-9) usd + N (I, - BC) (I, - 5C) 1 / ISdl) - 1 where t(_W) = WWXT - X ) / (M1 5d E7 N which is the standard t-test statistic with a weight vector W Xi 1er X:= sample mean vector defined asiX = ( (.0 11 d the 2 x 2 covariance matrix defined as Var(UH) CovwH 3L) d Cow?!” 21L) Var(UL) "| |" = "determinant" (I) I It should be noted, however, that the null hypothesis H0] (or HOA) is an overall hypothesis expressed in terms of a vector and that the significance of the T2 statistic does not indicate which component of 38 this vector has led to the rejection of the overall hypothesis. In order to see which component mean return difference, high or low, has contributed to the rejection of the null hypothesis, the simul- taneous confidence intervals of the two-component mean return differ- ences can be evaluated. This evaluation is equivalent to making sim- ultaneous inferences about a finite number of linear functions Wfi at a given joint level of significance. Thus, for example, if the confidence interval of the high risk mean return difference does not contain zero, while that of the low risk mean return difference does, then it is meant that the rejection of the overall null hypothesis is caused by the significance of mean return difference for the high risk group.20 In order to evaluate which component mean return difference contributes to the rejection of thernfll hypothesis, the present study used three different weight vectors: the weights (WW) implicit in computing the value of T2 (or maximizing the value of t2), W2 = [l O]' for the high risk group, and W5 = [0 l]' for the low risk group. The weight vector W2 was employed to see if the high risk mean return difference is significantly different from zero, while the weight vector W3 was utilized to determine if the low risk mean return dif- ference is significant. According to Gonedes (1975), a value of the weight vector can be viewed as the proportions of investments in dif- ferent types of assets.21 For example, the weight vector W2 gives rise to the high risk assets under the hypothetical situation in which one places his entire investment in the high risk securities. Like- wise, W3 yields the low risk assets which is equivalent to an entire investment in the low risk securities. 39 When the null hypothesis is true, the 12 has the following F distribution with p (=2, H and L for the present study) and (N - p) degrees of freedom: = _H_:_E__. 2 _ P. N-P P(N - 1) T (310) F For a decision as to whether the null hypothesis is rejected or not, the computed F value is compared with the critical F value at a given significance level. If the computed F exceeds the critical value, then the null hypothesis is rejected and it can be inferred that the SEC lease disclosure decision had an impact upon the pricing of securities. Otherwise, the null hypothesis is accepted as stated. Tests on Covariance Matrices The purpose of the tests on the covariance matrices was to deter- mine if all paired components of the two covariance matrices compared between treatment and control firms were simultaneously equal. Any significant difference between one or more paired components is suf- ficient to reject the null hypothesis for the homogeneity of variances and covariances. The tests for the equality of covariance matrices were conducted by using one of the likelihood ratio criteria, known as the "general- ized" test of homoscedasticity of variances and covariances suggested 22 in Box (1949). Box modified Bartlett's M statistic so that it can be utilized in testing for the homogeneity of variance-covariance matrices. The form of Box's modified M statistic is given by23 G M = V 1n [Sal - i vk 1n |Sk| (3-11) 40 where V = Z vk where vk = degrees of freedom (nk - l) and k is a k group (G = 2 for T and C) Var(R”) Cov(R“ AL) k - COV(HH RL) Var(RL) which is the sample covariance matrix for group k Sa = (i vk Sk)/V which is the average of the sample covariance. Notice that Sk here is computed on raw return data, (RH and RL), while Sd in testing the equality of mean vectors is calculated on return differences (OH and BL). By introducing a scale factor C, Box showed that, if A2 is near to or greater than A?, 2 M / C = X (3-12) with degrees of freedom f = (G - 1) p (p + l) / 2, where C = 1 / (1 - A1) A1 = (2p2 + 3p - 1) (G + 1) / 6 (p + 1) G v (v1 = v2 = v) A2 = (p - 1) (p + 2) (G2 + G + 1) / 6 G2 v2. According to Box, if p and G s 5 and v 2 20, the chi-square approximation agrees well with the exact value of the chi-square dis- tribution. However, when p and G are large and v is small (less than 20), the F approximation must be used. In this case the forms of this F approximation differ depending upon the sign of (A2 - A?). If the sign is positive, M / b = F (3-13) with f1 = f and f2 = (f1 + 2) / (A2 - A?) as degrees of freedom. Here, b = f1 / (1 - A1 - (f.l / f2)). 0n the other hand, if the sign is negative, 41 f2 M / f] (B - M) = F (3-14) ° - = 2 _ = .. with f1 - f and f2 (f1 + 2) / (A1 A2). Here, 8 f2 / (1 A1) + (2 / f2)). The present study used (3-12) for a test period with v 2 20 and (3-13) or (3-14) for all other selected test periods.24 As a decision rule, the computed F or X2 values are compared with their corresponding critical values at a given level of signif- icance. If the former exceeds the latter at a given level of signif- icance, the null hypothesis is rejected. This rejection implies that at least one pair of corresponding components of the two covariance matrices are significantly different from each other. Hence, it can be concluded that the SEC lease disclosure decision had an impact upon the variability of security returns. TestingiProcedures Test Periods As mentioned in Chapter I, there were a series of events that took place during the course of the SEC lease disclosure decision. They were: (1) the publication of the ASR No. 147 proposal; (2) the adoption of the proposal; (3) the effective date of the Release; and (4) the first public disclosure of the actual PV and IE numbers for individual firms according to the Release. To investigate the effect, if any, of the SEC lease disclosure decision upon the pricing of securities, it is important to know when the market reaction to the SEC decision could have taken place. One could conjecture that the market would begin to anticipate the possible effect of the lease disclosure as early as the preliminary discussion 42 of the SEC decision. Alternatively, an intensive market reaction might have occurred around the time of the acutal release of the PV and IE numbers. The market reaction to the various events of the SEC decision could be anticipatory, immediate, or lagged (or some combination of these three types). In order to investigate all possible market reactions, a 21-month test period (January 1973 through September 1974) was selected. This period covers six months before the first event (the publication of the proposal) to six months after the first actual disclosure (March 1974)25 of the PV and IE accounting numbers in the lO-K reports as required in ASR No. 147. It was believed that includ- ing these six additional months would provide a period long enough to capture all possible market reactions to the events of the SEC decision. For purposes of trying to identify more precisely when the most intensive market reaction took place, the 21-month test period was divided into five subperiods on the basis of the critical events iden- tified above. Both the events and selected test periods are summarized 26 However, the effective date (November 30, 1973) in Figure 3-3 below. of ASR No. 147 was deleted from consideration for testing on the grounds that there is little to differentiate it from the adoption date in terms of potential significance to investors; that is, if there is to be any market reaction, it would seem logical to assume that it would occur around the time of the initial proposal or its adoption rather than the effective date. The actual effects on financial ratios could not be determined until sometime after the effective date when the capitalized lease figures were first disclosed. The first test period (TPl) covered the entire twenty-one months 43 beginning January 1973 through September 1974. Therefore, the test results should reflect the effects of all of the events surrounding the SEC lease disclosure requirement upon security pricing. If there existed any reaction to the SEC decision and/or the actual disclosure of the first financial statements containing the PV and IE numbers, it should be incorporated in the test results for this time period.27 Announcement Adoption of Effective First of the proposal the proposal date disclosure pan 1973 June 1973 Oct 1973 Nov 1973 Mar 1974 Sept | | | 1974 | | 21 mohths | | | TP 1: é;- A I I I I F 1 I I I I I I I | | 16 months I | | TP 2: < > I I I T I I I I I I TP 3 I l 12 months I I ‘ I ‘ T I 71 I I I I ° I ‘ I '1 TP 5 l I I & 6: LA 14 months \L 7 months \I Figure 3-3. Test Periods Based on Critical Events The second test (TP2) covered the period from the proposal date (June 1973) through September 1974. All possible effects, except an- ticipatory market reaction to the publication of the proposal, should be detected by this test. The third test (TP3) was for the period October 1973 through September 1974. This test was performed mainly to investigate the 44 effects of the -adoption of the proposal and of the first actual dis- closure. The fourth test (TP4) focused on only one special event, the first actual disclosure of the PV and IE numbers as required in ASR No. 147. The test period included two months before the first dis- closure month. Thus, the test results here should reflect both an- ticipatory and gx_pgst_market reaction to the event of the actual disclosure. ' The fifth test (TP5) covered the fourteen months from January 1973 through February 1974. This period was selected to detect the effects of all events prior to the first actual disclosure. The last test (TP6) was conducted to ascertain immediate and lagged market reactions to the event of the first actual disclosure alone. All previous test designs were not pertinent to this purpose. However, the test period ran only from March through September 1974 so that the sample size (= seven) was relatively small. Therefore, the power of the test may be questioned and the interpretation of the results will have to take this fact into account. Comparisons of Groups Depending upon the types of the capitalized lease information (PV or IE) that they reported, the firms in Group 3 were divided into two subgroups for conducting tests, the PV disclosure firms and the PV and IE disclosure firms. The final data collection shows that sixty firms belonged to the first subgroup, while there were thirty- nine firms in the second. No single firm was found which reported the IE numbers alone. Each of the two subgroups was subsequently 45 divided into the two risk classes, high and low risk, according to their Beta estimates. Likewise, the firms in Group 1 and Group 2 were classified consistently with the firms in Group 3. The final results are summarized in Figure 3-4. Types of Risk Groups Disclosure Level Size __, TT"H 30 PV ._.__ L 30 Group 3 (Disclosure) PV and IE” "H 20 T“‘*—~—-—-L 19 ___ -‘-H 30 PV..__________________L 30 Group 2 (No Disclosure) ' \PV and IE/H 20 ‘““**--——_L 19 _____ *“‘H 30 / PV \ L 30 Group 1 (No Leases) __—————-“’li 20 PV and IE"“‘*‘---L 19 Figure 3-4. Grouping of Firms . Based upon the groups formed, all statistical tests were con- ducted for each of the six different test periods by comparing the PV disclosure firms (and the PV and IE disclosure firms) of Group 3 with the corresponding paired firms of Group 1 and Group 2, respec- tively. A comparison of the firms in Group 2 with the firms in Group 1 was ignored since the former firms were treated as control firms rather than as treatment firms. Also, it was believed that a com- parison between Group 3 and Group 1 would be sufficient to evaluate an effect of the SEC lease disclosure decision on the pricing of securities. It was considered more relevant to compare Group 3 with Group 1 than to compare Group 2 with Group 1.28 46 Given the two null hypotheses (H01 and H02) and the design of test periods and groupings, all statistical tests were conducted with a special focus on the evaluation of the following questions: (1) Did the SEC lease disclosure requirement have any effect on the pricing of securities as measured by changes in the means and variances of re- turn distribution? (2) If an effect was present, did investors react differently to the PV disclosure alone than they did to the situation where both PV and IE numbers were disclosed? (3) At what point in time were the information effects, if any, of the SEC decision re- vealed in the stock price adjustments? And to which specific event(s) of the SEC decision was market reaction most intensive? (4) Were the lease information effects (if any) risk-dependent? (5) If the disclo- sure firms were affected by the SEC decision, were they affected in an adverse manner? In addressing the above questions, one would ideally like to make a comparative study of the effects of three kinds of disclosure: PV numbers only, IE numbers only, and both PV and IE numbers. How- ever, as indicated before, it was impossible to find a sufficient number of firms that reported the IE numbers alone. As a result, the present study concentrated on the two disclosure situations: the PV disclosure alone and the PV and IE disclosure. Nevertheless, a com- parative study of the effects of these two disclosure situations would enable one not only to address the above five questions directly, but also, indirectly, to evaluate the effects of the IE disclosure only. In addition, a comparative analysis of this sort would be quite useful in addressing the question of the relative information content of the balance sheet vs. income statement effects of lease commitments, 47 that is, whether the effects of the capitalized lease disclosure are on the firms' capital structures or on their earnings record. CHAPTER IV SAMPLING DESIGN FOR DATA COLLECTION This chapter consists of two sections. The first section describes the selection of sample firms and their pairwise matching between treatment and control groups. The second section explains how return (stock price) data were collected. Sample Firms The present study employed three different groups of firms, one treatment group (disclosure firms) and two control groups (non- lease and non-disclosure firms).1 Disclosure Firms The disclosure firms (Group 3) were those which reported the PV and/or IE numbers of noncapitalized financing leases in their lO-K reports according to ASR No. 147 for 1973 and 1974. The names of 324 firms meeting one and/or both of these conditions were in- itially obtained from Disclosure Journal (1973 and 1974).2 Then, the following additional criteria were used to further restrict the sample firms within this initial group: (1) firms must be registered on the New York Stock Exchange NYSE). (2) Stock price data must be available while they were traded on NYSE during the period from September 1968 through September 1974.3 48 49 (3) No capitalized lease data were reported on the lO-K or annual reports prior to ASR No. 147. (4) Fiscal year must end December 31. These criteria, except for (2), were checked by a reading of the lO-K and annual reports for the three years, 1972 through 1974. The final data collection showed that sixty firms met the above three conditions and disclosed the PV numbers for 1973 and 1974, while thirty-nine firms were found to meet the three conditions and report both PV and IE data for the same two years. It was found that no firm disclosed the IE numbers alone. Therefore, the primary treatment group (Group 3) consisted of ninety-nine firms in total, the sum of the sixty PV disclosure firms and the thirty-nine PV and IE disclosure firms. As Table 4-1 shows below, these treatment firms included some firms which did not meet the two materiality criteria, 5% for the PV disclosure and 3% for the IE disclosure. For example, there were four such firms in 1973 and seven in 1974 which disclosed their PV numbers, even if these numbers did not neet the 5% materiality criterion. The firms of this type were also found in the case of the PV and IE dis- closure group.4 Table 4-1 also provides various ranges of both PV and IE ratios and the number of sample firms in each range not only by the type of lease disclosure, but also by the relative risk level. For the sixty 5 to 61.4% for PV disclosure firms, the PV ratio ranges from -102.7% 1974 with a median value between .05 and .10. The range of this PV ratio for the thirty-nine PV and IE disclosure firms is from .83% to 61.4% with a much higher median value between .20 and .30. For the same year the IE ratio varies from -220% to 43.7%, and a median value 50 Table 4-1 Firms by Ranges of the PV and IE Ratios PV Firms PV and IE Firms Range of PV / L-Tca Range of PV / L-Tca Range of IE / ANIb Ratio Ratio Ratio 1973 1974 1973 1974 1973 1974 -1.027 - (-.001) 1 1 -1.027 - (-.001) 0 O -2.200 - (-.500) 2 2 .000 - .049 4 7 .000 - .049 5 4 -.499 - (-.200) 4 2 .050 - .099 35 32 .050 - .099 4 4 -.199 - (-.100) 8 9 .100 - .199 15 14 .100 - .199 8 9 -.O99 - (-.050) 8 9 .200 - .299 3 2 .200 - .299 10 B -.049 - (-.O30) 9 10 .300 - .399 1 4 .300 — .399 6 8 -.029 - .000 5 3 .400 - .499 1 0 .400 - .499 4 2 .001 - .029 0 1 .500 - .614 0 0 .500 - .614 2 4 .030 - .437 3 3 Total 60 60 Total 39 39 Total 39 39 Mean Ratio .091 .082 Mean Ratio 233 .245 Mean Ratio -.121 -.158 s. D.c .132 .165 s. of .155 .163 s. OF .195 .340 High 8 High 8 High 8 (30 Firms) (20 Firms) (20 Firms) Mean Ratio .091 .091 Mean Ratio .203 .210 Mean Ratio -.178 -.119 S. D. .051 .061 S. D. .131 .132 S. D. .265 .177 Low 8 Low 8 Low 6 (30 Firms) (19 Firms) (19 Firms) Mean Ratio .090 .074 Mean Ratio .264 .282 Mean Ratio -.062 -.158 S. 0. .180 .227 S. D. .176 .189 S. D. .121 .512 eamount of tthe flong- term capitalization (L—TC) is the sum of long-term debt. stockholders' equity, Tand the PVl bThe average net income (ANI) implies average of reported net income for the most recent three years. cStandard Deviation. 51 of this ratio falls within the range between -.05 and -.10. On the other hand, a mean of the PV ratio for the PV disclosure firms is .091 for 1973 and .082 for 1974, respectively, while means of this ratio for the PV and IE disclosure firms is .233 and .245 and the IE ratios have mean values of -.121 and -.158 for the same two years. Two interesting facts can be found in the table. First, the mean of the PV ratios in each year is much greater for the firms with both PV and IE disclosure than for the firms with PV disclosure alone (due to immaterial IE). Notice that the former firms also have larger IE ratios than the latter firms whose IE numbers were not material, which implies that the firms which disclosed both PV and IE numbers might be evaluated by investors as being potentially more risky than the firms which repor- ted only the PV numbers. Such an evaluation may be due to the fact that the greater PV and IE ratios mean a greater impact on the financial ratios which might be used by investors as instrumental variables in assessing the riskiness of individual securities. Second, for the PV and IE disclosure firms the mean and stan- dard deviations of the PV ratios are far greater for the low risk group than for the high risk group, although this is not true for the PV firms. This finding implies that, for those low risk PV and IE firms, the balance-sheet-based financial ratios (such as debt- equity ratio, current ratio, asset growth, and so forth) and also other financial ratios like accounting return on total capital might be more affected by the capitalized lease disclosure than the same ratios for the high risk PV and IE firms. As a result, there is a possibility that the security prices of the low risk PV and IE disclosure firms were affected more by the SEC lease 51 of this ratio falls within the range between -.05 and -.10. On the other hand, a mean of the PV ratio for the PV disclosure firms is .091 for 1973 and .082 for 1974, respectively, while means of this ratio for the PV and IE disclosure firms is .233 and .245 and the IE ratios have mean values of -.121 and -.l58 for the same two years. Two interesting facts can be found in the table. First, the mean of the PV ratios in each year is much greater for the firms with both PV and IE disclosure than for the firms with PV disclosure alone (due to immaterial IE). Notice that the former firms also have larger IE ratios than the latter firms whose IE numbers were not material, which implies that the firms which disclosed both PV and IE numbers might be evaluated by investors as being potentially more risky than the firms which repor- ted only the PV numbers. Such an evaluation may be due to the fact that the greater PV and IE ratios mean a greater impact on the financial ratios which might be used by investors as instrumental variables in assessing the riskiness of individual securities. Second, for the PV and IE disclosure firms the mean and stan- dard deviations of the PV ratios are far greater for the low risk group than for the high risk group, although this is not true for the PV firms. This finding implies that, for those low risk PV and IE firms, the balance-sheet-based financial ratios (such as debt- equity ratio, current ratio, asset growth, and so forth) and also other financial ratios like accounting return on total capital might be more affected by the capitalized lease disclosure than the same ratios for the high risk PV and IE firms. As a result, there is a possibility that the security prices of the low risk PV and IE disclosure firms were affected more by the SEC lease 52 disclosure decision for a certain time period than the security prices of the corresponding high risk firms. 0n the other hand, Table 4-1 also suggests that the directions of change in the two ratios, PV and IE, are highly consistent between the two types of firms; all the PV firms, except one, experienced a positive change in the PV ratios, while most of the IE firms revealed a negative change in the IE ratios. It should be noted, however, that such a consistent directional change in those ratios does not necessarily imply a similar directional change in the distribution of security returns. The reason is that, as Nelson (1963) found, some financial ratios (for example, net profits to net working cap- ital and return on total capital) were not negatively correlated with the magnitude of the capitalized amount of lease commitments. Fur- thermore, as Beaver, et. a1. (1970) noted, the financial ratios, such as earnings variability, leverage, and growth, vary in the same dir- ection as the relative risks, while other financial ratios, such as payout, size, and liquidity, change in the opposite direction. This two-way directional change impliesthat a larger PV (or IE) may have both larger positive and negative effects on the financial ratios at the same time. Therefore, the final direction of the net effect upon the distribution of returns may differ, depending upon which financial ratio(s) investors use in making their investment decisions. Non-Disclosure and Non-Lease Firms The names of potential firms for the two control groups were initially obtained from Disclosure Journal (1973 and 1974). Then, they were narrowed down according to the same four criteria that were 53 used in selecting the treatment firms. This restriction was accom- plished by a reading of the footnotes on leases and other background information, such as "Summary of Significant Accounting Policies" and "Supplementary Income Statement Information“ in the lO-K and annual reports of 1973 through 1974. In addition to the above criteria applied, each firm in Group 1 (non-lease firms) and Group 2 (non-disclosure firms) was paired with every firm in Group 3 (disclosure firms) on the basis of pub- lished Beta estimates and with a consideration of the standard indus- trial classification (SIC) code. The individual Beta estimates were available from the September 1973 issue of Security_Risk Evaluation published by Merrill Lynch, Pierce, Fenner and Smith, Inc. The Beta estimates were computed on the simple market model6 using monthly individual returns and the S 8 P 500 Index as a market index (RM). The estimation covered the 60-month period ending August 31, 1973.7 The Beta book provides two kinds of Beta estimates: the unadjusted (raw) Beta estimates and the adjusted ones. The present study used the adjusted Beta estimates. This adjusted Beta estimate8 is a rel- ative risk measure that takes account of the regression phenomenon reported by Black, Jensen, and Scholes (1972) and Fama and MacBeth (1973), the phenomenon of the overestimation of Bj of high risk stock and the underestimation of Bj of low risk stock. In the matching of two individual firms, the SIC code was taken into account as much as possible. The rationale for giving some consideration to the SIC code was based upon the fact that an industry factor explains a fairly significant portion of security return behavior. According to King (1966), about 10% of the variance 54 in returns could be explained by an industry factor. However, Fer- tuck (1975) found that King's finding was true only for certain in- dustries and that it could not be generalized to all industries. "In some industries, the industry effect is trivial and can be safely ignored. In others, it can be as large as a third of the market effect."9 The implication of this statement is that the industry effect exists in some industries but it is not as important as the market effect. Based upon this evidence, the present study sac- rificed the industry Classification in favor of the general market effect (measured by Bj) when there was a conflict between the match- ing of firms on the SIC code and the matching of firms on Bj. Such a sacrifice of industry classification was inevitable because it was difficult to find enough control firms so that every pair of firms could be in the same industry, while meeting the condition, éjT = éjC’ as well as the four selection criteria described before. Results of Sample Selection Because of the matching at the individual firm level, the num- ber of firms in each of the two control groups is equal to the number of firms in Group 3. Thus, there are 297 sample firms in total with ninety-nine firms in each group.10 The result of matching firms on Bj is given in Table 4-2. A careful review of this table shows that, for seventy-three matchings out of ninety-nine pairs between Group 3 and Group 1, differences in the magnitudes of two matched Betas fall within the range of less than 5% of éjT’ while the number of matchings with less than 5% of deviation in two Betas is seventy-four out of ninety-nine between 55 Table 4-2 Matching ofAFirms on Beta Estimates (8) No Group 3 Group 2 Group 1 1.40 1.40 1.40 2 1.41 1.42 1.36 3 1.48 1.47 1.46 4 1.78 1.77 1.65 5 .83 .83 .83 6 .87 .84 .87 7 1.76 1.71 1.62 8 1.06 1.09 1.07 9 1.48 1.46 1.47 10 1.39 1.39 1.37 11 1.40 1.40 1.41 12 1.12 1.12 1.08 13 1.04 1.00 1.04 14 1.29 1.29 1.30 15 2.06 2.01 1.92 16 1.40 1.35 1.34 17 1.86 1.83 1.74 18 1.11 1.12 1.10 19 1.03 1.03 1.03 20 1.22 1.24 1.23 21 1.01 .99 1.00 22 1.84 1.81 1.71 23 1.16 1.16 1.15 24 1.15 1.15 1.13 25 1.40 1.40 1.36 26 1.21 1.21 1.20 27 1.65 1.56 1.58 28 1.03 1.01 1.02 29 1.03 1.01 1.04 30 1.46 1.45 1.43 31 1.28 1.29 1.25 32 1.65 1.61 1.48 33 1.22 1.25 1.20 34 1.79 1.79 1.68 35 1.45 1.44 1.42 36 .93 .95 .93 37 1.37 1.32 1.30 38 1.33 1.33 1.31 39 1.03 1.00 1.04 40 1.30 1.30 1.31 41 1.10 1.11 1.09 42 1.56 1.49 1.54 43 1.47 1.45 1.47 44 1.41 1.41 1.40 45 1.03 1.01 1.04 46 1.06 1.05 1.04 47 1.16 1.15 1.16 56 Table 4-2 (continued) No. Group 3 Group 2 Group 1 48 .91 .93 .92 49 1.80 1.79 1.69 50 1.62 1.52 1.50 51 .99 1.00 .99 52 .98 .99 .98 53 1.94 1.95 1.88 54 .94 .94 .94 55 1.28 1.28 1.27 56 1.28 1.28 1.25 57 1.78 1.77 1.65 58 1.66 1.58 1.58 59 2.11 2.01 1.98 60 1.12 1.11 1.11 61 1.16 1.18 1.13 62 1.42 1.44 1.42 63 1.76 1.71 1.64 64 1.39 1.34 1.33 65 1.14 1.14 1.13 66 1.26 1.27 1.24 67 1.21 1.22 1.20 68 1.66 1.66 1.51 69 1.37 1.31 1.32 70 .76 .76 .75 71 1.50 1.50 1.46 72 1.17 1.20 1.16 73 1.36 1.31 1.31 74 1.03 1.02 1.04 75 1.24 1.25 1.20 76 .95 .95 .96 77 .99 .99 1.00 78 .87 .89 .87 79 1.71 1.58 1.59 80 1. 2 1.88 1.80 81 1.26 1.26 1.23 82 2.25 2.21 2.67 83 1.76 1.70 1.65 84 1.64 1.54 1.52 85 1.94 1.92 1.86 86 1.53 1.49 1.48 87 1.13 1.15 1.13 88 1.30 1.30 1.30 89 1.24 1.25 1.24 90 1.06 1.08 1.07 91 1.65 1.58 1.51 92 1.42 1.43 1.41 \O b ..a A ...a ...a b O _.I (A) 00 57 Table 4-2 (continued) No. Group 3 Group 2 Group 1 95 1.74 1.67 1.64 96 1.39 1.33 1.37 97 1.74 1.59 1.62 98 1.74 1.67 1.50 99 1.75 1.77 1.66 Mean of B = 1.366 1.348 1.328 Variance of B = .105 .095 .092 Group 3 and Group 2. In terms of magnitude the greatest absolute difference in two matched Betas is .42 which is found in the 82nd pair between Group 3 and Group 1. (See Table 4-2.) However, the relative ratio of this difference is only 18.7% ( = ( 2.67 - 2.25) / 2.25). Table 4-3 gives some summary statistics of individual Betas in the previous table. This summary table shows that, between the overall groups, the mean values of Beta estimates are close to each other. For example, the mean Beta (1.366) of Group 3 is fairly close to the mean Beta (1.348) of Group 2 and the mean Beta (1.328) of Group 1. The variances, associated with these mean values, are relatively small. The equality of mean Betas is shown to have been maintained even when each of the three overall groups was broken down into two subgroups by the types of lease disclosure. For the PV disclosure firms the greatest deviation in mean Beta is found when Group 3 is compared with Group 1. However, the absolute value of this deviation 58 Table 4-3 Statistics of Beta Estimates Used for Matching Overall Group 3 Group 2 Group 1 G3 G2 G1 PV PV-IE PV PV-IE Pv PV-IE Min. .76 .76 .75 .76 .83 .76 .83 .75 .83 (H) (1.28) (1.41) (1.28) (1.40) (1.27) (1.36) (L) (.76) (.83) (.76) (.83) (.75) (.83) Max. 2.25 2.21 2.67 2.06 2.25 2.01 2.21 1.92 2.67 (H) (2.06) (2.25) (2 01) (2.21) (1.92) (2.67) (L) (1.28) (1.40) (1.28) (1.40) (1.25) (1.37) Mean 1.366 1.348 1.328 1.309 1.453 1.294 1.432 1.276 1.409 (H) (1.535) (1.755) (1.501) (1.720) (1.477) (1.678) (L) (1.082) (1.135) (1.086) (1.128) (1.075) (1.126) Var .105 .095 .092 .077 .139 .067 .129 .059 .134 (H) (.035) (.049) (.030) (.046) (.022) (.083) (L) (.016) (.034) (.017) (.033) (.015) (.029) No. 99 99 99 60 39 60 39 6O 39 8. (30) 20) (30) (20) (30) (20) J (L) (30) 19) (30) (19) (30) (19) Note: H = High risk and L = Low risk. is on1y .033 (= 1.309 - 1.276). Likewise, the largest deviation in mean Beta in the case of the PV and IE disclosure firms is observed again when Group 3 is compared with Group 1 and its magnitude is no more than .044 (= 1.453 - 1.409). In the variances of the estimated Betas, there is also little difference between Group 3 and Group 1 (or Group 2), both when these two groups are compared at the aggre- gate group level and when each overall group is divided into two subgroups. For the PV disclosure firms, for example, difference in the variances of estimated Betas is only .018 (= .077 - .059) between Group 3 and Group 1 and .10 (=.077 - .067) between Group 3 and Group 2. The corresponding differences for the PV and IE disclosure firms 59 are even smaller .005 (= .139 - .134) and .010 (= .139 - .129). How- ever, the tendency is ohserved that the PV and IE disclosure firms have higher means and variances of Beta estimates, as compared to the PV disclosure firms. This tendency is consistent with the pre- vious fact that the PV and IE firms have relatively higher PV and IE ratios and, therefore, higher chances to be evaluated as being poten- tially more risky. The closeness of means and variances of Beta estimates between the treatment and control groups and a tendency for the PV and IE disclosure to have a relatively higher mean and variance of Beta esti- mates are also observed when each subgroup (by the types of lease disclosure) is again divided into two risk groups, high and low. The results of the matching suggest that the purpose of match- ing firms was reasonably satisfied in the sense that the firms in the treatment and control groups were in the same homogeneous risk class and had the same first moments of return dis- tributions during the pre-test (Beta estimation) period. In an effort to evaluate this fact,a statistical test was conducted for the period of one year (1972) selected as part of the pre-test period. The results from this test suggest that there was no statistically significant difference either in mean return vectors or in covar- iance matrices between the treatment and control groups.11 Thus, it is now believed that the conditions, UT = UC and 2T = 2C, were met for the 12-month period immediately before the entire 21-month test period. As mentioned before, the Betas were estimated over five years. About 93% of all firms' Betas were computed by using monthly return 6O observations over the 60-month period or the period close to it as shown in Table 4-4. However, there were some firms for which a relatively small number of return observations were available for estimating their Betas. For example, the numbers of the firms whose return observations available totaled less than forty-four were four in Group 3, one in Group 2, and three in Group 1. Nevertheless, the inclusion of these small numbers of firms in the sample was not con- sidered to be critical. Table 4-4 Firms by the Number of Months Covered in Estimating Bj No. of Months Group 3 Group 2 Group 1 15 - 44 4 l 3 45 - 54 3 5 4 55 - 60 92 93 92 (60)a Igllb I§§lP I§fllP Total 99 99 99 asixty months. bthe number of firms for which the sixty monthly return obser- vations were available in estimating Betas. Because of the difficulty of finding enough control firms so that every pair of firms could be in the same industry, the matching of firms on the basis of the SIC code was not as successful as inten- ded. Table 4-5 provides some insight into this fact. 61 Table 4-5 Sample Firms Matched by Industries No. of Pairsa SIC Code Industry G3 - G2 G3 — G1 1 Mining and construction - l 2 & 3 Manufacturing 25 19 211 Cigarettes 1 - 23 Apparrel and other finished products made from fabrics and similar mater- ials l - 27 Printing, publishing, and allied industries - l 28 Chemicals and allied products 281 Industrial inorganic chemicals 29 Petroleum refining and related in- dustries 291 Petroleum refining 324 Cement and hydraulic 33 Primary metal industries 331 Blast furnaces, steel works, and rolling and finishing mills 343 Heating equipment 35 Machinery except electrical 366 Communication equipment 367 Electronic components and access.; , 371 Motor vehicles and equipment 39 Miscellaneous manufacturing industrs. 394 Toys and amusement, sporting, and athletic goods - dl-bl b-f ...aN—Jdpd-J .—l-—l|—l —.a NI-HI-JII —J 4 Transportation and public utilities 4 401 Railroads 1 48 Communication - _I_.IN 5 Retail trade 0 2 6 Finance, insurance, and real estate - 602 Commercial and stock savings banks 1 Othersb 52 N NOU'I £0 Total 99 aGl, 02, and G3 stand for Group 1, 2, and 3. bThe consideration of the SIC code in matching firms was impossible for these numbers of firms. 62 Security Return Data Monthly returns, used as a stock price variable for this study, are continuously compounded rates of return computed as follows: (P + Dt) R t Pt-l = loge (4'1) t where 0 denotes dividends and t denotes a month. All return data were obtained from a 1975 edition of the CRSP tape12 available at Michigan State University. Since the test design of the present study was based upon the pairwise matching of firms, an unexpectedly high percentage of missing return data was observed. Even if only one of the matched firms missed return data for some month, the paired return observation (i.e., return difference) for this month was deleted as missing data. As a result, a few missing data were observed for some months of the test period, although the exact number of firms for which return data were missing varied from month to month. The main reason for the data missing was that there were some companies which had just stopped or started trading on the NYSE during the time period covered by this study. This fact could be identified by checking the names of such companies in the daily stock market news section of the Wall Street Journal, Despite the missing data,however, the sample size of returns did not change since a sample unit is a monthly average of component returns in each risk class, high and low. CHAPTER V ANALYSIS OF TEST RESULTS An analysis of the statistical test results is presented in this chapter which consists of four sections. The test results on mean return vectors and covariance matrices are analyzed in the first two sections. In the third section the relevance of the SEC's two materiality criteria1 for the PV and IE disclosure are empirically evaluated by comparing the security prices of firms that met the SEC'S materiality criteria 'U) those of firms whose lease commitments failed to meet these criteria. In the final section of this chapter general comments concerning the empirical findings are provided. Since the rejection of any or both of the null hypotheses (H01 and H02) concerning the mean vectors and covariance matrices is a sufficient condition for a capitalized lease information effect to be present, the only situation that leads to no information effect is the case where both of the null hypotheses are not rejected. The evidence from the test results suggests that the lease in- formation effects were present when the effects were measured by changes in expected returns. However, when the effects were measured by changes in variability of returns, no sign of information effects was observed. 63 64 Test Results on Mean Return Vectors The objective of conducting tests on mean return vectors is to see if the null hypothesis (H01) of no return difference is re- jected. That is, the goal is to see if the disclosure of capitalized lease data under ASR No. 147 caused investors to reassess the expected returns of the firms affected by the SEC disclosure requirement. Sta- tistically, this is equivalent to testing for a difference between the first moments of the conditional and unconditional return distri- butions. The rejection of the null hypothesis of no difference is consistent with the alternative hypothesis of information content of the capitalized lease data. According to the test results discussed in detail below, it appears that the information effects measured by changes in condition- al expected returns existed for the case of both the PV firms and the firms with both PV and IE disclosure when these disclosure firms were compared with non-lease firms (Group 1). However, when the same dis- closure firms were compared with non-disclosure firms (Group 2), no sign of information effects was observed. Test Results on Mean Vectors: Disclosure Firms (Group 3) vs. Non-Lease Firms (Group 1) Table 5-1 summarizes the test results on mean return vectors when the disclosure firms were compared with non-lease firms. Col- umns 2 through 7 reveal the test results for the firms differentiated on the basis of the PV numbers of noncapitalized future financing leases, while columns 8 through 13 provide summary results for the groups of firms differentiated on the basis of both PV and IE numbers. 65 Table 5-1 Test Resolts on Mean Vectors: Group 3 vs. GrOup l Pv DisclOSure Firms PV and IE Disclosure Firms Average . b Average . b "°"th5 Return Difference T2 Computed Weights Return Difference T2 Computed Weights F-Valuea F-Valuea High Low High Low High Low High Low ITPl: 21 -.0059 c -.0035 c 3.357 1.595 .767 .233 -.0180 c -.0065 C 6.378 3.030* .872 .128 (.0159) (.0222) (.0329) (.0316) . 21 2.847 1.352 1 O 6.268 2.978 1 O 21 .515 .245 O 1 .876 .416 O l TP2: 16 -.0058 -.0054 3.050 1.423 .728 .272 -.0186 .0011 7.081 3.304. 1.412 -.412 (.0155) (.0246) (.0294) (.0302) , 16 2.248 1.049 1 O 6.445 3.007 1 O 16 .765 .357 0 l .021 .010 0 l 1P3: 12 -.0102 -.0047 7.023 3.192" 1.013 - 013 -.0194 .0094 8.192 3.724. 2.356 -1.356 (.0133) (.0240) , (.0272) (.0265) 12 7.019 3.191 1 0 6.111 2.778 1 0 12 .466 .212 0 1 1.531 .696 O 1 1P4: 9 - 0095 -.0054 5.663 2.477 .801 .199 -.0177 .0142 5.173 2.263 10.704 -9.704 ~ (.0131) (.0213) (.0305) (.0285) 9 4.829 2.113 1 0 3.048 1.334 1 0 9 .572 .250 O 1 2.227 .975 0 1 1P5: 14 -.0029 - 0041 .852 .393 .600 .400 —.0130 - 0178 7.074 3.265. -.095 1.095 (.0164) (.0238) (.0362) (.0251) . 14 .449 .207 1 0 1.799 .830, l O 14 .422 .195 0 1 6.992 3.227 0 1 1P6: 7 - 0117 -.0022 4.848 2.020 .911 .089 -.0280 .0161 11.768 4.903"' 1.610 -.610 (.0142) (.0204) (.0244) (.0326) , 7 4.753 1.981 1 0 9.195 3.832 1 0 7 .080 .033 O 1 1.714 .714 0 1 Note (1) ': Significant at .10 level. . a: Degrees offreedom for the F-values are (2, 19) for TPl, (2, 14) for 1P2, (2. 10) for TP3. (2. 7) for TP4. (2, 12) for TPS, and (2. 5) for TP6. b: The first line of the weight vector column in each test period stands for the implicit weights (implicit in computing the value of T ). c: Standard deviation. Note (2) Selected fractiles of the F distribution and the exact F values are: d. f. Fractiles F-Value d. f. Fractiles F-Value (2. 19) .800 1.75 (2. 10) .800 1.90 .900 2.61 .900 2.92 .950 3.52 .950 4.10 .990 5.93 .990 7.56 (2. 14) .800 1.81 (2. 7) .800 2.04 .900 2.73 .900 3.26 .950 3.74 .950 4.74 .990 6.51 .990 9.55 (2, 12) .800 1.85 (2. 5) .800 2.26 .900 2.81 .900 3.78 .950 3.88 .950 5.79 .990 6.93 .990 13.27 66 PV Disclosure Firms: Group 3 vs. Group 1 For the PV firms the means of return differences (R? - R2), for example, between the two groups), computed over the number of months in each of the test periods, are shown in the second and third columns by the risk Classes, high and low. Along with the average return differ- ences, the associated standard deviations of return differences are indicated in parentheses. The values of T2 were computed according to Equation (3-9) by using different weight vectors shown in columns 6 and 7. The weights in the first line of the weight vector column for each test period are the weights that yield the maximum value of t2. The computed 12 values are given in the fourth column and the associated F-values are shown in column 5. The degrees of freedom and related exact F values for various fractiles are found in the footnote to the table. Thus, a final decision as to whether or not the null hy- pothesis on mean vectors should be accepted can be made by comparing the computed F values with the associated exact values at a given sig- nificance level. A glance at the F-value column reveals that the only test period yielding significant test results is the 12-month period (TP3) cover- ing October 1973 (the month in which the SEC lease disclosure decision was formally announced) to September 1974 (six months after the first public disclosure of the PV and IE numbers according to ASR No. 147). For this period the computed F value (3.192), based on the implicit weight vector2 [1.013 - .013], exceeds the critical value (2.92) at the .10 significance level.3 Therefore, the null hypothesis of 1K) re- turn difference is rejected for this test period. Furthermore, the evidence from the observed F values associated 67 with different risk groups indicates that this rejection of the null hypothesis is mainly attributable to a significant difference in mean returns of the high risk firms compared between Group 3 and Group 1. This is evidenced by the fact that for the 12-month period the observed F value (3.191) associated with the weight vector [1 O] for the high risk group is significant and far greater than the F value (.212) associated with the weight vector [0 l] for the low risk group. This fact implies that the conditional expected returns on the high risk assets were affected more by the SEC lease disclosure decision than were the conditional expected returns on the low risk assets. The rejection of the null hypothesis for the 12-month period and the greater contribution to this rejection by the high risk group can also be explained in terms of the magnitudes of return differences. For this period the return difference for the high risk group is -.0102 which is the largest difference next to -.0117 for the 7-month period. With the high and low risk groups combined, the 12-month period yields the largest difference, -.0149 (= -.0102 + (-.0047)). Despite the large values of return difference both for the 9-month period (the high and low groups combined) and for the high risk group for the 7-month period, the test results for those two periods did not appear to be significant, although the fairly large differences in returns for these two periods imply that there was some degree of market reaction to the disclosure of the PV and IE numbers. The reason for this lack of significance seems to be due to relatively small sample sizes (i.e., small degrees of freedom), only nine for the 9-month period and seven for the 7-month period. The test results for the other five periods, including the entire 21-month period, 68 did not lead to the rejection of the null hypothesis at the significance level of either .05 or .10. Thus, it can be maintained that either the event of the announcement of the ASR NO. 147 proposal or that of the first public disclosure of the PV and IE numbers in the financial reports by itself did not induce a significant market reaction. Rather the signif- icance of the test results for the 12-month period suggests that a signif- icant market reaction took place as a joint effect of the following two events: the announcement of the adoption of the proposal and the first public disclosure of the PV and IE numbers according to ASR No. 147. The above results are supported by the cumulative average re- turn difference (CARD) curve4 in Figure 5-1A and 1B. Notice that the CARD curves were constructed with cumulative average return "differen- ces" so that they should be compared with the horizontal zero return difference line for interpretation. As seen in Figure 5-1A, the over- all CARD curve begins to deviate from the horizontal zero return differ- ence line around September 1973 (just before the announcement of the adoption of the proposal), and this tendency of deviation continues up to September 1974. The pattern of this deviation over the time period is consistent with the significance of the test results for the 12-month test period. Moreover, Figure 5-lB suggests that for the same 12-month period the CARD curve of the high risk group experienced a greater change than did the CARD curve of the low risk group. Therefore, it appears that the high risk firms contributed more to the rejection of the null hypothesis for this test period. In fact, the evidence from the test results of the other five periods, except the lZ-month period, also suggests that the high risk firms were generally more sensitive to the various events of the SEC 69 0 50’0— CARD . 375'- .. a... ...,. w—wg l i 0250— ...,L O ‘ -4L - .,_....-_.._..- . ..__.- 1 1 . i 1 1 _ I All --250— t _ 1 I 1 '-500_ . : i ; LIIIII LII II IIIIII III A] Jan Jun Oct Mar Sept 1973 1973 1973 1974 1974 Figure 5-1A Cumulative Average Return Difference: PV Firms (Group 3 vs. Group 1) CARD , I -.250_ . High 11 III II III III III_LII 1_J I I r* r Jan Jun Oct Mar Sept 1973 1973 1973 1974 1974 Figure 5-lB Cumulative Average Return Difference: PV Firms (Group 3 vs. Group 1) Note: All = all firms, High = high risk firms and Low = low risk firms. 7O lease disclosure decision than were the low risk firms. (Compare the F values associated with the weight vector [l O] for the high risk group and [D l] for the low risk group in Table 5-l for each test period.) This finding is consistent with the fact that the high risk firms on the average had higher PV ratios (with a mean value of .091 for 1973, for example) than the low risk firms (.074 for l973).5 The higher PV ratios indicate the greater extent to which the financial ratios were affected by the disclosure of the capitalized lease data, which in turn suggests that the firms were likely to be evaluated by investors as being more risky. According to the test results, both the high and low risk groups of the PV firms were affected in the same way (adversely) by the SEC lease disclosure decision. The evidence for this effect is found in columns 2 and 3 of Table 5-l (and Figure 5-lB) where the signs of average return differences for all the six test periods are negative. This negativity implies that security prices of both the high and low risk PV firms, as compared with their counter-part non-lease firms, were adversely affected by the SEC lease disclosure decision. This finding seems to support not only the a_prjgrj_claims made by management and others who have tended to resist disclosing capitalized lease information, but also Nelson's finding that lease capitalization adversely affected the financial ratios of the firms. The adverse effect on security prices of both the high and low risk firms is also supported by Figure 5-lB wherein the CARD curves of the two risk groups remain below the horizontal zero return difference line over most of the entire 2l-month test period. The difference in the degree of the adverse effect between the two risk groups appears to 7l show up more remarkably in the last four months, June through September 1974. The above fact is consistent with the evidence that the absolute value of average return difference between the two risk groups is greater for the 7-month period (.0ll7 - .0022 = .0095) than for any other test period. (See columns 2 and 3 of Table S-l.) To summarize, the evidence obtained by comparing the PV dis- closure firms with the corresponding non-lease firms suggests that there was a lease information effect when the effect was measured by changes in expected returns. Therefore, it can be inferred that the SEC lease disclosure decision significantly affected the security prices of the PV disclosure firms. The existence of lease information effect is con- sistent with the a_prjgrj_arguments, including those of the SEC, that capitalized lease information is important to investors. Conversely, the counter-arguments which question the information content of capital- ized lease data are contradicted by the evidence presented here. In . addition, the evidence indicates that the degree to which the security prices of the PV disclosure firms were affected by the SEC decision was in general higher for the high risk firms than for the low risk firms. Furthermore, the security prices were affected in an adverse manner, which is consistent with Nelson's conclusion that lease capitalization gener- ally affected the financial ratios adversely. The observed lease infor- mation effect appears to have existed mainly during the lZ-month period which includes both the formal announcement of the SEC decision (i.e., the announcement of ASR No. 147) and the disclosure of the PV numbers in financial reports. The evidence (especially the average return differ- ence for the 7-month period) indicates that some degree of market reac- tion was present with respect to the disclosure of the PV numbers; but 72 this reaction was not sufficient to reject the null hypothesis. PV and IE Disclosure Firms: Group 3 vs. Group l The test results on mean vectors for the PV-IE disclosure firms are also summarized in Table 5-l, columns 8 through l3. Unlike the case of the PV firms, however, the test results for all periods, except TP4 (the 9-month period), were found to be significant at the .l0 level. (Compare the computed F-values in column ll with their corresponding critical values provided in footnotes.) This finding suggests that the hypothesized information effect on expected returns was not only present, but also spread throughout the entire time period from the initial proposal publication to the dis- closure of the actual PV and IE numbers in the lO-K reports. This find- ing is also supported by both the magnitudes of average return differences in columns 8 and 9 of Table 5-l and Figure 5-2A. The absolute values of average return differences for the PV-IE firms tend to be in general much higher than those for the firms with disclosure of PV numbers alone. This tendency is consistent with the fact that the mean PV ratios (for example, .210 for the high risk group for 1974 and .282 for the low risk group) for the PV-IE firms are much greater than the mean PV ratios (for example, .09l for the high risk group for 1974 and .074 for the low risk group) for the PV firms.6 Figure 5-2A also suggests that the overall CARD curve deviates considerably from the horizontal zero return difference line during al- most all of the entire Zl-month period. However, the major declines of this curve took place in approximately March 1973 and around the time of the PV and IE disclosure in the lO—K reports. Therefore, it is implied 73 .500“- .375— CARD .250... .125i- '- “\Rfiv ‘.250-— ”\‘E — :11111111111114111 Jan Jun Oct Mar Sept 1973 l973 l973 l973 l974 Figure 5-2A Cumulative Average Return Difference: PV and IE Firms (Group 3 vs. Group 1) o 375- CARD. 250— '125- 0° \/ Low -'250-— - High -0500.— }11L11_111111l1111]1 Jan Jun Oct Mar Sept l973 l973 l973 l974 l974 Figure 5-28 Cumulative Average Return Difference: PV and IE Firms (Group 3 vs. Group l) 74 that the major market reaction to the SEC lease disclosure decision took place three months prior to the initial publication of the ASR No. 147 proposal. .Moreover, there was a further market reaction upon the dis- closure of the PV and IE numbers. The above finding leads to the conclusion that the joint effect of both PV and IE disclosure on expected returns is far greater than the singular effect of PV numbers. This conclusion is based on the fact that in the previous case of the PV firms there was only one test period (the l2-month period) for which a lease information effect on mean returns was found to be present. The evidence from Table 5-l and Figure 5-23 indicates that the security prices of the firms in different risk classes were not affected to the same degree by the SEC lease disclosure decision. As seen in column ll of Table 5-l, the observed F values of the high risk group associated with the weight vector [l 0] for all test periods except the l4-month period tend to be greater than the observed F values of the low risk group associated with the weight vector [0 l]. This tendency is also found for the T2 values computed on the basis of the same weight vectors. Therefore, the interpretation is that the security prices of the high risk PV-IE disclosure firms were in general more sensitive to the events of the SEC lease disclosure decision than were the security prices of the low risk PV-IE disclosure firms. This fact is also sup— portedtn/Figure 5-28 wherein the directional deviation of the CARD curve for the high risk group from the zero line is observed during almost all of the Zl-month period, while such a directional deviation of the CARD curve is not present for the low risk group. However, it is interesting to note that, in contrast to the PV 75 disclosure firms analyzed previously, the contribution of the high risk firms here to the observed lease information effect of both PV and IE disclosure is not consistently greater for all test periods than that of the low risk firms. For the l4-month period (TP5), for example, the computed F value (3.227) associated with the weight vector [0 l] for the low risk group is significant and far larger than the F value (.830) associated with [l O] for the high risk group. This finding is consis- tent with the larger magnitude of average return difference for the low risk group. For the l4-month period the magnitude of average return dif- ference (-.Ol78) of the low risk group (with a relatively smaller stand- ard deviation of .025l) is larger than the difference (-.Ol30) of the high risk group (with a relatively larger standard deviation of .0362). But this is not true for the other five periods. The larger return difference for the low risk group is suppor- ted by Figure 5-28 wherein the CARD curve for the low risk group tends to show a larger deviation from the horizontal zero return difference line than does the CARD curve for the high risk group during the l4- month period (January l973 through February l974). However, this ten- dency was reversed after March 1974. This seems to suggest that the high risk PV-IE disclosure firms were more affected by the actual disclosure of both PV and IE numbers, while the low risk firms were more affected by the pre-disclosure events of the SEC lease disclosure decision. A possible reason for the latter case is that these low risk disclosure firms had a greater variance of Beta estimates (.034) than did their counter-part non-lease firms (.029), while the high risk disclosure firms had a smaller variance of Beta estimates (.049) than did their counter-part non—lease firms (.083).7 76 And a combination of this relatively high variability of the relative risks with investors' a_prjg§i_pessimistic expectations about the burden of lease commitments by the low risk disclosure firms might have resulted in the greater impact on the security prices of these firms during the l4-month pre-disclosure period. However, when the actual PV and IE num- bers (especially the latter) were revealed in March l974, the investors might have revised their a priori_pessimistic expectations about these low risk firms since they might have realized that the burden of lease commitments on these firms turned out to be much less than what they had previously assumed. For example, the mean IE ratio was only -.O62 for the low risk firms for l973, while the same ratio was -.l78 for the high risk firms.8 This interpretation is consistent with the signs of average return differences for the l4-month and 7-month periods in columns 9 and l0 of Table 5—l. Notice that there is a negative sign of average return difference for the l4-month period for the low risk firms, while a positive sign of average return difference for the 7-month period is observed. The positive sign of average return difference for this lat- ter period implies that the low risk disclosure firms experienced rather moderate evaluation, relative to non-lease firms, by investors or a re- version of previous adverse reaction once the true effects of leasing were made known. Finally, one interesting point is that the market reaction to the preliminary discussion of the SEC extended lease disclosure regula- tion appeared to begin as early as March l973, three months prior to the publication of the SEC's lease disclosure proposal. The evidence for this timing is found in Figures 5-2A and 28 from which it is clear 77 that the adverse price reaction began around March l973. This implies that the market reaction in the case of the PV and IE disclosure firms took place about six months earlier than in the case of the PV dis- closure firms. For the PV firms it was previously noted that the mar- ket reaction began around September l973. To summarize, the evidence obtained by comparing the PV and IE disclosure firms with non-lease firms suggests that the information effects of the various events related to the SEC lease disclosure decision were present when the effects were measured by the degree of changes in expected returns. This evidence is consistent with the traditional arguments which maintain the existence of information con- tent of capitalized lease data. The evidence for the information effects is much stronger here than in the previous case of the firms with disclosure of the PV numbers alone. The test results also sug- gest that the market reaction began as early as March l973, three months prior to the publication of the ASR No. 147 proposal. It has been further noted that, in general, the high risk disclosure firms appeared to be more adversely affected by the SEC lease disclosure requirement than were the low risk disclosure firms. The evidence also indicates that, over-all, capitalized lease disclosure did have an adverse effect on the valuation of the firms, although a slight upward readjustment of the security prices upon the disclosure of the capitalized lease data was observed for the low risk firms. This may indicate that negative effects of such lease data on various finan- cial ratios were not as bad as investors had originally anticipated. 78 Test Results on Mean Vectors: Disclosure Firms (Group 3) vs. Non-Disclosure Firms (Group 2) Tests for the equality of mean return vectors between disclo- sure firms (Group 3) and non-disclosure firms (Group 2) are inter- esting because firms in both groups had noncapitalized financing leases but the firms in the disclosure group disclosed the PV and/or IE numbers, while the firms in the non-disclosure group did not re- port such numbers presumably because the numbers did not meet the materiality criteria set forth by the SEC. Therefore, disclosure based on the materiality of noncapitalized financing leases is the main differentiating factor between the two groups. If the hypothesized lease information effects exist, the ex- pected returns of the disclosure firms should differ from the expec- ted returns of the non-disclosure firms provided that the SEC's materiality guidelines are meaningful. Alternatively, one might expect to observe similar price behavior for the two groups of firms (disclosure and non-disclosure) if investors perceive the SEC mater- iality criteria to be of little consequence.9 The evidence from the test results here suggests that when the two groups were compared, no effects on conditional expected returns were observed for either the PV firms or the PV-IE disclosure firms. This is different from the finding when the same disclosure firms were compared with the firms with no lease commitments. PV Disclosure Firms: Group 3 vs. Group 2 The test results for the PV disclosure firms, as compared with non—disclosure firms (Group 2), are given in columns 5 through 7 of Table 5-2. Since this table is parallel to Table 5-l, all the terms 79 .uumLLou ma ow Ucaom ucm twang—zooms 8m: Louuw> 2.6.83 93 mo mm3~o> mmme hmeamzc: mmmf. ”U ANV muoz .cowooowcomwo a moo co mopooooco owoqumm ooo .u .o .o co» _-m w_ooh co Amy ooo Apv moooz mom APV oooz _ o ooo. Npo. _ o o__. ooo. N o P ooo. mom. o F ooo. o~_.P h A__mo.v Amkmo.o Amk_o.v Aoomo.v . _No. oko. _omo. ooo. mPoo.- oo_o.- ompu.oo m_~.~o- moo. ooo.. mmoo. oN_o.- N .oak _ o ooo. moo. _ o ooo. moo. o_ o _ moo. moo. o F ooo. No_. o. A_omo.v Aoooo.v Aomoo.o Apomo.v . “No. moo. ooo. ooo. ouoo.- oooo.- _oo._ _oo.- oso. “No.— oooo. mmoo.- o. .oah _ o ooo. ooo. _ o moo. ooo. o o F ooo. o_o. o _ moo. oom.P o ANNNo.V Aoooo.v Aoopo.o Amomo.o . mom. RNA. ooo. mpo. Pooo.- opoo. om_.- om_._ ooo. mom.F mooo.- ___o.- o .oap _ o ooo. moo. _ o ooo. ooo. N_ o _ Foo. _oo. o _ Koo. ooo., NP Aommo.v Aoooo.v Amo_o.v Anmmo.v . ooo. moo. ooo. moo. N_oo.- oooo. ooo.o- ooo.“ ooo., ooF.m _ooo. “ooo.- N. .map F o ooo. _oo. F o ooo. ooo. o_ o . mop. oNN. o _ _oo.F ooN.N o_ AoNNo.o Ammoo.v Aoomo.o Aomoo.o . ooo.- mom.” mop. oNN. Nooo. omoo.- ooo.- moo.F oom._ ooh.m N_oo. No_o.- o_ .Nac _ o ooo. moo. F o moo. NN_._ Pm o F oo_. ooo. o _ ooo. ooo. _N ofloomo.v oAmooo.v vooNo.o oAmoNo.v mko. ooo. ouo. ooo._ oooo.- _ooo.- Noo.N Noo._- ooo._ mom.~ oooo. mooo.- Pm n_ah zoo oo_x zoo goo: 204 now: zoo oo_z mpzmpm: mmmymbmw » musmpmz om:_o>-d H mocmcmwopo ccoumm . mocmcmmmwo :Loumm . . n N moocm>< n nmuoaeou N mmocm>< mcucoz macro wgamopum_a NH nco >o mEL_u «Como—umwo >o N ozoco .m) m noocu “mcouum> com: :o mu_:mma umm» N-m mpaop 80 are as defined previously. When the F values in column 5 are compared with the correspond- ]0 it is found that none of the observed F values ing exact values, exceeds the associated critical value. This finding is consistent with the implications of the relatively small magnitudes of average return differences. As seen in columns 2 and 3 of the table, the average return differences for the high risk group are relatively small, ranging from -.OO3S to -.Ol20, while those of the low risk group vary from -.0005 to .0054. This fact is also supported by Figure 5-3A wherein the over-all CARD curve shows very little dev- iation from the horizontal zero return difference line. It follows that the various events of the SEC lease disclosure decision, either taken together or separately, did not significantly differentiate the PV firms (with material lease commitments) from the non-disclosure firms (with immaterial lease commitments). Ex- pected returns of the two groups of firms appeared to behave in essentially the same way over the entire 2l-month period as well as for the various sub-periods. It should be noted, too, that this result is somewhat different from the prior finding where the same PV firms were compared with non-lease firms (Group 1). There, the information effect was shown to be present for the l2-month test period. Also, unlike the previous case, the general tendency for the security prices of the high risk firms to be more affected by the events of the SEC decision is not observed here. In the case of the Zl-month (TPl) and l4-month (TP5) periods, for example, the low risk firms are shown to have been more affected than the high risk firms. 81 (Compare the T2 and F values in columns 4 and 5 of Table 5-2 associ- ated with the weight vectors, [l O] and [O l], respectively.) The evidence from Table 5-2 indicates that the high risk firms were adversely affected by the events of the SEC lease disclosure decision, while the low risk firms were generally not. This result is inconsistent with the previous case where both the high and low risk PV firms were shown to be adversely affected. The evidence for this fact is given in columns 2 and 3 of the table. As seen in the columns, the signs of average return differences are negative for the high risk group for each of the six test periods, while the signs of average return differences are positive for the low risk firms for all of the test periods except the 9-month period (TP4). These comparative behaviors of average return differences for the two risk groups are also supported by the CARD curves in Figure 5-38. The CARD curve of the high risk group lies below the horizon- tal zero return line during almost all of the entire 2l-month period, while the CARD curve of the low risk group lies above the zero line throughout the period. However, the computed F values associated with the weight vectors, [1 O] and [O l], are all far less than the corresponding critical values. Therefore, although a systematic pattern of deviation between the two CARD curves is present, this should not be intrepreted as evidence that the hypothesized lease information effect is significantly risk-dependent. In summary, no sign of lease information effect is present either for the various events taken together of the SEC lease dis- closure decision or for any single event evaluated separately, when the information effect is evaluated by comparing the conditional 82 -5004-— CARD 0375'- -—-a .o - W___-__.._* .250:- .125.. I 1 A11 1 I ‘-2501— I _ 1 1 'oSOO..__ 17L11Jo 111_1111111111141_j Jan Jun Oct Mar Sept l973 l973 l973 l974 l974 Figure 5-3A Cumulative Average Return Difference: PV Firms (Group 3 vs. Group 2) CARD .500—.ln“ A- N” '1 V ~ 1 0375-— g 1 .250- 1 I .125_W 1 1 1L111111L1L11111111111 I 1 1 ' 1 Jan Jun Oct Mar Sept 1973 1973 l973 l974 l974 - 05004- Figure 5-38 Cumulative Average Return Difference: PV Firms (Group 3 vs. Group 2) 83 expected returns of the PV disclosure firms with the unconditional expected returns of non-dosclosure firms. This finding is somewhat in contrast to the previous case where the same PV firms were compared with non-lease firms (Group 1). Therefore, it can be inferred that the materiality of lease commitments (as defined by the SEC) is not a critical factor on which the market differentiates the security prices of the firms with material lease commitments from the secur- ity prices of the firms with immaterial leases. Rather, a more im- portant factor seems to be whether or not the firms have noncapital- ized lease commitments. Prior to disclosure of actual numbers, in- vestors probably did not know which firm would be material in lease commitments (in Group 3) and which firm would not be material (in Group 2). Even after disclosure of such numbers, investors might also have suspected that a disclosure firm would be a member of Group 3 for one year, while this same firm could become a member of Group 2 for another year. Any (or both) of these situations could cause no difference between the two types of firms. In addition to the above, no evidence was found for the ten- dency that the high risk firms were more affected by the various events of the SEC lease disclosure decision than were the low risk firms. There may be two possible reasons for this finding: First, the PV ratios of the high risk firms were not very different from the PV ratios of the low risk firms so that the effects of the PV numbers on various financial ratios might not significantly differ between the two risk groups. The means of the PV ratios for the high risk firms were .091 for l973 and .09l for l974, while the mean PV ratios for the low risk firms were .090 for 1973 and .074 for 84 l974. Second, the means and variances of Beta estimates were fairly close between the high and low risk PV disclosure firms and their 1] However, there is respective counter-part non-disclosure firms. evidence that the high risk firms were in general affected in an adverse manner by the SEC lease disclosure decision, while the low risk firms were not.12 This finding is different from what was dis- covered when the same PV firms were compared with non-lease firms (Group l). PV and IE Disclosure Firms: Group 3 vs. Group 2 The statistical test results from comparing the PV-IE dis- closure firms with non-disclosure firms (Group 2) are summarized in columns 8 through l3 of Table 5-2. A glance at the F-values in column ll indicates that none of the test results for any of the six test periods appears to be sig- nificant at the significance level of either .05 or .lO. The re- sults are consistent with the implications of relatively small magni- tudes of average return differences (in columns 8 and 9) ranging from -.OlO4 U).OOl6 for the high risk group and from -.OO74 to .0002 for the low risk group. In general, the magnitudes of average return differences for the various test periods here are far less than those when the same PV-IE disclosure firms were previously computed with non- lease firms (Group l). (Compare columns 8 and 9 of Table 5-2 with the corresponding columns of Table 5-l.) The CARD curves in Figure 5-4A and 4B also reveal similar results. Therefore,it can be concluded that none of the events associated with the SEC decision had an effect upon the conditional expected returns for the PV-IE firms as compared 85 .5001— CARD -315— .250- 0125'- m r-j;_7‘\\V"\«/’E“v”'_7“-—"_‘S I All '.250"" 1 1 --500- l 1114LL F I 1111L111 111L1% Jan Jun Oct Mar Sept 1973 I973 l973 l974 1974 Figure 5-4A Cumulative Average Return Difference: PV and IE Firms (Group 3 vs. Group 2) .500..- - 375— CARD ~250— ~125- High --250- '1 111, 111 L111 .L111 11 T I Jan Jun Oct Mar Sept l973 l973 l973 l974 1974 Figure 5-48 Cumulative Average Return Difference: PV and IE Firms (Group 3 vs. Group 2) 86 with those firms which had noncapitalized financing leases but did not report the PV and IE numbers due to the immateriality of such leases. The tendency for the high-risk group to be more sensitive to the events of the SEC decision that the low-risk group is not re- vealed here. The evidence for this fact is given by the F-values associated with the two weight vectors, [l O] and [O l] for TPl, TP3, and TP5, in column ll of Table 5-2 and also the CARD curves in Figure 5-48. The test results suggest that both the high and low risk disclosure firms were adversely affected over-all by the SEC lease disclosure decision but not significantly so. The evidence for this finding is provided by the signs of average return differ- ences for various test periods in columns 8 and 9 of the table. An over-all summary of all the test results obtained from the mean vector tests is presented in Table 5-3. Based upon the evidence from the test results, it can be concluded that the SEC lease dis- closure decision had an information effect as revealed through changes in the equilibrium prices of those securities most affected by the disclosure requirements. The evidence for the information effects appears to be much stronger for the joint disclosure of the PV and IE numbers than for the PV disclosure alone, although the effects were also present in the case of the firms with disclosure of the PV numbers alone. Therefore, it seems that investors adjust their per- ceptions about return prospects of the firms mainly with respect to the joint disclosure of both PV and IE numbers rather than with respect to the PV disclosure alone. In general, the high risk firms were more adversely affected by the SEC decision than were the low 87 Table 5-3 Summary of Mean Vector Test Results Group 3 vs. Group l Group 3 vs. Group 2 Months PV Firms PV-IE Firms PV Firms PV-IE Firms * TPl: 2l NS S NS NS * H > L H > L H < L H < L (-H -L) (-H -L) (-H +L) (-H -L) TP2: l6 NS 5* NS NS * H > L H > L H > L H > L (-H -L) (-H +L) (-H +L) (-H +L) * * TP3: 12 S NS NS * > L H > L H > L H < L (-H -L) (-H +L) (-H +L) (+H -L) TP4: 9 NS NS NS NS H > L H > L H > L H > L (-H -L) (-H +L) (-H -L) (+H -L) * TPS: l4 NS S NS NS * H > L H < L H < L H < L (-H -L) (-H -L) (-H +L) (-H -L) TP6: 7 NS 5* NS NS * H > L H > L H > L H > L (-H -L) (-H +L) (-H +L) (-H -L) Note NS: Not significant at .lO level. 5: Significant. *: Significant at .lO level. H,L: High and low risk group. H > L: A greater contribution of high risk group to the * observed value of the test statistic. H > L: The contribution of high risk group is greater and significant at .l0 level. +H(L): Sign of the average return difference is positive for the high (low) risk group. -H(L): Sign of the average return difference is negative for the high (low) risk group. 88 risk firms, a finding which is not surprising since the high risk firms are by nature relatively more volatile. In addition, the evidence suggests that the market reaction began about three months before the initial publication of the ASR No. 147 (twelve months prior to the actual disclosure of the PV and IE numbers). Furthermore, the most intensive market reaction took place with respect to the pre-disclosure events of the SEC decision, although some degree of further reaction was observed upon the dis- closure of the PV and IE numbers in the lO-K reports. Thus, it can be interpreted that the information of the SEC lease disclosure de- cision was largely absorbed by the market before the actual numbers were released. The above conclusions were valid only when the disclosure firms (Group 3) were compared with non-lease firms (Group l). However, when the same disclosure firms were compared with non—disclosure firms (Group 2), no sign of information effects was observed either for the PV disclosure firms or for the firms with disclosure of both PV and IE numbers. This lack of information effect implies that Group 3 and Group 2 generated the same market reaction and, hence, that the materiality of lease commitments (as defined by the SEC) did not cause any significant difference in returns between the two groups. This fact is consistent with what was discovered when the same two groups were compared in terms of PV disclosure. Test Results on Covariance Matrices As indicated earlier, the purpose of conducting tests on the covariance matrices of total realized returns, RH and RL, was to 89 determine if the return variability of disclosure firms changed as a result of the SEC lease disclosure decision. This is equivalent to testing the null hypothesis (H02) of equal covariance matrices for the treatment and control firms. The rejection of this null hypothesis is another sufficient condition for concluding that the SEC lease disclosure decision did have information content in the sense that changes in return variability were associated with the SEC lease disclosure requirement. As discussed in Chapter II, disclosure of the PV and IE num- bers may cause investors to re-assess the riskiness of the securi- ties of the disclosing firms through an impact of such disclosure on the various financial ratios which may be used by investors. For example, including the PV in the long-term liability and treat- ing the IE as an adjustment to the reported net income will affect the financial ratios, such as leverage, payout, liquidity, and earn- ings variabliity, which Beaver, et. al (l970) found to be highly associated with the relative risk measure (Beta). As discussed earli- er, it was discovered that certain financial ratios were inversely related to Beta, while some other financial ratios were positively related. This two-way directional relationship implies that Betas can change in either direction, through the potential impact of capitalized lease disclosure on the financial ratios, since differ- ent investors may use different financial ratios for their invest- ment decisions. Changes in Betas in any direction could lead to the possibility of changes in the variability of returns. To the extent that such changes in the variability of returns exist, the informa- tion effects of the SEC lease disclosure decision can be evaluated 90 in terms of such changes. An inference from the test results in this section suggests that neither the return variability of the PV disclosure firms nor the firms with disclosure of both PV and IE numbers were signifi- cantly affected as a result of the SEC lease disclosure decision. Although the results of the covariance matrix test for some test periods revealed significance, an interpretation of these results leads to the conclusion that such significance was not a result of the SEC decision, but due to a random chance which could be expected under the null hypothesis at the .10 level. Test Results on Covariance Matrices: Disclosure Firms (Group 3) vs. Non-Lease Firms (Group l) The objective of conducting tests for the equality of covar- iance matrices was to see if the return variability of disclosure firms is different from the return variability of non—lease firms (Group l). If this difference is found, then it can be inferred that the disclosure of the PV numbers affected the variance-covariance structure of returns. The test results reported in this section suggest that hypothe- sized variance-covariance effect was not present either for the PV disclosure firms or for the PV-IE disclosure firms. PV Disclosure Firms: Group 3 vs. Group 1 The test results on covariance matrices for the PV disclosure firms, compared with non-lease firms, are given in columns 2 through 6 of Table 5-4. The values of the M statistic in column 4 were Table 5-4 Test Results on Covariance Matrices: 591 Group 3 vs .Group l PV Disclosure Firms PV and IE Disclosure Firns . Number 6 a b . a of Variance Computed Values Variance Computed Values Months M Statistic M Statistic Treat. Control x2 F Treat. Control X2 F it it TPl: 21 .0088 .0060 10.727 10 146 3.382 .0117 .0064 2.237 2.116 .705 .0044 .0045 .0073 - .0067 .0055 .0050 .0073 .0045 1P2: 16 .0093 .0074 5.464 NAC 1.670 .0114 .0076 .882 NAC .273 .0051 .0056 .0083 .0083 .0063 .0062 .0075 .0057 TP3: 12 .0083 .0073 I 4.490 NA 1.349 .0113 .0070 .996 NA .299 ’ .0050 .0057 .0081 .0074 .0060 .0063 .0067 .0052 TP4: 9 .0062 .0034 3.206 NA .924 .0094 .0043 3.856 NA 1.111 .0033 .0028 .0062 .0082 .0039 .0029 .0041 .0047 1P5: 14 .0104 .0076 7.919 NA 2.420. .0153 .0087 2.828 NA .864 .0051 .0058 .0088 .0061 .0066 .0065 .0104 .0049 TP6: 7 .0067 .0035 2.803 NA .766 .0061 .0027 6.142 NA 1.678 .0037 .0024 .0053 .0091 .0042 .0027 .0017 .0042 Note (1) 1"'z *: Note (2) Significant at .05 level. Significant at .10 level. a: Degrees of freedom for the chi-square value are 3 for TPl and the F-values have degrees of freedom of (3, 2888000) for TPl, (3. 162000) for TP2, (3, 87120) for TP3, (3, 46080) for TP4. and (3, 25920) for TP6. b: The first line of this column for each test period stands for the variances of the high risk treatment and control firms, the second line for the variances of the low risk treatment and control firms. and the third line for the covariance between the high and low risk groups within each of the treatment and control groups. c: NA implies "Not Applicable." Selected fractiles of the X2 and F distributions and the exact values are: Distribution Fractiles d. f. Values x2 .800 4.642 .900 6.251 .950 7.815 .990 11.345 3, 121680) for TP5, Distribution Fractiles d. f. Values F .800 (3 m) 1.55 .900 (3 on) 2.08 .950 (3 m) 2.60 .990 (3 o) 3.78 92 computed according to Eq. (3-11) by using the variances and covar- iances in columns 2 and 3. The observed chi-square (X2) and F values in columns 5 and 6 were calculated as defined in Eq. (3-12) through 2 or F statistic (3-14). As discussed in Chapter III, the choice of X depends upon the number of degrees of freedom. Box (1949) suggests the use of X2 if the number of observations is greater than or equal to twenty and the F statistic if it is less than twenty. Accordingly, the X2 statistic was chosen for the all-inclusive 21-month test period. In addition, the observed F value is also presented as supplementary information. For the other five test periods the use of the F sta- tistic is relevant due to their small sample sizes. The selected 2 and F distributions and the associated critical fractiles of both X values are given in the footnote to the table. An examination of the observed X2 and F values in columns 3 and 4 indicates that the observed X2 value (10.146) for the entire 21- month period far exceeds its critical value (7.815) at the signifi- cance level of .05. Also, the test result for the l4-month period (TP5) is shown to be significant at the .10 level. For all other periods the test results were not significant at the significance level of either .05 or .10. However, it is not clear why the observed variance-covariance effect for the PV disclosure firms is so high, as compared with the firms with disclosure of both PV and IE numbers for which no variance- covariance effect was found to be present. The observed effect for the PV disclosure firms is inconsistent with the implications of some other related evidence. First, as seen in Table 4-1 previously, the mean and standard deviation of the PV ratios for the PV disclosure 93 firms were far less than the mean and standard deviation of the PV ratios for the PV and IE disclosure firms, regardless of the risk class of these firms. Moreover, the IE ratios for the PV disclosure firms were all shown to be immaterial (by the definition of this group), while the IE ratios for the PV and IE disclosure firms were presented as being material. Conceptually, therefore, the financial ratios of the PV firms should be affected less by the disclosure of the capital- ized lease data than the financial ratios of the PV and IE disclosure firms. Consequently, the extent to which investors reassessed the riskiness of the PV firms' securities by using the financial ratios incorporating the PV numbers should be less than the degree to which investors re-evaluated the riskiness of the securities of the firms by using the financial ratios incorporating disclosure of both PV and IE numbers. Second, in terms of the mean and variance of the Beta estimates used in the matching of firms, there was little difference (in a rela- tive sense) between the PV firms and the PV and IE firms during the pre-test period when these disclosure firms were compared with non— lease firms. The evidence for this was given in Table 4-3 of Chapter IV. Little difference in the variance of the Beta estimates implies that a variance-covariance effect should not be present unless the SEC decision did cause a significant change in the variability of Betas during the test periods, a change which is not known. Third, for the 2l-month and l4-month periods, the magnitudes of variances and covariances for the treatment firms, as shown in column 2 of Table 5-4, are not very different from those for the treatment firms in column 3 for the case of the PV disclosure firms. This lack 94 of difference is even more clear when one comparatively looks at the magnitudes of variances and covariances in columns 7 and 8 for the same two periods in the case of the PV and IE disclosure firms. For these reasons, it would seem more appropriate to conclude that the observed variance-covariance effects for the two test per- iods were present by random chance rather than as a result of the SEC lease disclosure decision. This random chance could be expected under the null hypothesis with the probability of 10 out of 100. PV and IE Disclosure Firms: Group 3 vs. Group 1 The last five columns of Table 5-4 summarize the results of statistical tests for the firms with both PV and IE disclosure when these firms were compared with non-lease firms (Group 1). The values of the M statistic and the computed X2 and F values were all obtained in the same fashion as described earlier. Unlike the case of the PV disclosure firms, however, the com- parison of the observed X2 and F values with the associated exact values suggests that none of the test results for any test period was significant at .05 or .10 significance level. A glance at columns 7 and 8 of Table 5-4 implies that there exist some differences in the magnitudes of variances and covariances between the treatment and control firms. But such differences were not sufficient to reject the null hypothesis of equal variance and covariance matrices. Therefore, it can be concluded that the SEC lease disclosure decision had no effect on the variance-covariance structure of re- turns when the effect was measured by comparing the PV-IE disclosure firms with non-lease firms. As far as the variance-covariance effect 95 is concerned, this evidence is not suggestive of the traditional arguments, including those of the SEC, that capitalized lease data are important to investors in assessing the riskiness of securities. Rather, the evidence here is consistent with the counter-view that capitalized lease data convey no new information about the risk pros- pects of the firms. Test Results on Covariance Matrices: Disclosure Firms (Group 3) vs. Non-Disclosure Firms (Group 2) The tests on covariance matrices were also conducted by com- paring the lease disclosure firms with non-disclosure firms (Group 2). Recall that the non-disclosure firms, like the disclosure firms, had noncapitalized future financing leases but did not report the capitalized data of such leases because of the immateriality (as de- fined by the SEC) of their lease commitments. Therefore, disclosure based upon the materiality of the leases is a major difference be- tween the two groups of firms. The objective of the tests here was to see if the SEC lease disclosure decision had any differential effect on the return variabil- ity of the firms with "material" amounts of noncapitalized financing leases vis—a-vis the return variability of the firms with "immaterial" amounts of such leases. The covariance tests conducted by comparing the disclosure firms with the non-disclosure firms suggest that the SEC lease disclosure requirements had no effect on the variance-covariance structure of security returns. All observed values of the test statistic were far less than their corresponding critical values, not only for the entire 96 21-month period but also for the various sub-periods. PV Disclosure Firms: Group 3 vs. Group 2 The test results for the PV firms are given in columns 2 through 6 of Table 5-5. Since this table parallels Table 5-4, the definitions of all terms are exactly the same as before. Again, the evidence suggests that the various events associated with the SEC decision, including the actual disclosure of PV numbers, had no impact upon the dispersion of returns for the PV firms when the impact was measured by comparing the PV disclosure firms with non- disclosure firms. None of the observed X2 and F values exceeded its critical value at the .10 significance level. Therefore, the material- ity of non-capitalized future lease commitments (as defined by the SEC) was not a critical factor in differentiating the market reaction to the PV disclosure firms from the market reaction to the non-disclosure firms compared here. PV and IE Disclosure Firms: Group 3 vs. Group 2 The covariance test results for the firms with both PV and IE disclosure, as compared with non-disclosure firms, are summarized in the last five columns of Table 5-5. As in the previous case where the same disclosure firms were compared with non-lease firms (Group 1), there is no sign of a var- iance-covariance effect for any of the selected test periods. All the observed X2 and F values in columns 10 and 11 fail to reach their corresponding critical values at the .10 level. As a result, the null hypothesis of equal covariance matrices 97 .mmopo> uuoxm Lows» coo mcowuonwgumwu 2 new Nx mg“ yo mumuuoLm umuumme on» too .0 .n .o mo cowuocopoxm mgu poooo 21m mpno» mo ANV new APV mmuoz mom "muoz mmoo. oqoo. Nmoo. mmoo. Nwoo. Nmoo. Neoo. umuooEou mucopgo> mo:_o> uwuooEou mucopgo> mcucoz o n . o o . mo cmnsoz msgwu mcomopumwo m” new >2 magwu ocomopumwo >2 N aoogm.m> m ooogo "mou_guoz ouco+2o>ou co mayommm you» m-m apnop 98 between the treatment and control firms is not rejected. Accordingly, it appears that the SEC lease disclosure decision provided no new information about the risk attributes of the PV-IE disclosure firms. This conclusion is consistent with the conclusion reached previously when the same disclosure firms were compared with non-lease firms. There it was found that the variability of returns for the former firms was essentially the same as the variability of returns for the latter firms. An over-all summary of all the test results on covariance matrices is given in Table 5—6. As seen in the table, the only case Table 5-6 Summary of Covariance Matrix Test Results Group 3 vs. Group 1 Group 3 vs. Group 2 Months PV Firms PV-IE Firms PV Firms PV-IE Firms ** TPl: 21 ' S NS NS NS TP2: 16 NS NS NS NS TP3: 12 NS NS NS NS TP4: 9 NS NS NS NS TP5: 14 5* NS NS NS TP6: 7 NS NS NS NS ** Note (1) S : Significant at .05 level. But an interpretation of the test results for both the 21-month and the 14-month per- iod suggests that such a significance was a result of random chance rather than that of the SEC lease disclo- sure decision. (2) See the footnotes to Table 5-3 for the explanation of NS and S . 99 where the test results appeared to be significant was that involving the PV disclosure firms when these firms were compared with non-lease firms. As discussed before, however, this significance should not be interpreted as a result of the SEC lease disclosure decision. For all other cases no evidence was found for significant changes in the variance-covariance structure of returns for either the PV disclosure firms or the firms with both PV and IE disclosure. This finding im- plies that the SEC lease disclosure decision did not cause investors to reassess the risk prospects of the firms. As a result, the varia- bility of returns remained essentially unaffected by the SEC decision. Assessment of the SEC Materiality_Criteria for Lease Disclosure One interesting point is that the test results from the compar- ison of Group 3 (disclosure firms) with Group 2 (non-disclosure firms) can serve as a basis for evaluating the importance of the two material- ity criteria, 5% for the PV disclosure and 3% for the IE disclosure, which are suggested in ASR No. 147. As mentioned before, firms in both groups were similar to each other in that they all had noncapital- ized financing leases. Moreover, firms in both groups were matched with each other on the basis of their estimated Beta levels. Thus, it can be assumed that a main difference between the two groups of firms lies in the "materiality" of their lease commitments as defined by the SEC. In suggesting the two materiality criteria, the SEC seemed to believe that the two groups of firms have a different distribution of lease signals (PV and/or IE) and hence, that investors would assess those two types of firms differently. It was also believed that the 100 two cut-off points were not too "high" to eliminate lease signals that would be useful to investors, while they were not too "low” to cause some useless lease information to be made public. Furthermore, the SEC appeared to believe that the lease information which is impor- tant to investors is only those lease signals whose magnitudes meet the materiality crieria; the potential importance of immaterial lease signals is totally ignored for disclosure. In addition, accord- ing to the SEC, the lease information which meets the materiality criteria would not be available to the general public unless the dis- closure requirements such as those set forth in ASR No. 147 are en- forced. Under this situation and if the lease information effect exists, the behavior of returns for the disclosure firms will be different from the behavior of returns for the non-disclosure firms. This state- ment is equivalent to saying that mean return vectors and/or covar- iance matrices should not be equal between the two groups of firms if the two materiality criteria meaningfully differentiate the relative importance of lease commitments for these two groups of firms. The present test results provide no evidence for difference in return behavior between the two types of firms and thus no sign of the importance of the two materiality guidelines set forth by the SEC. As seen earlier in Table 5-2 and Table 5-5,which summarize the test results on mean return vectors and covariance matrices between Group 3 and Group 2, none of the observed values of the test statistics exceeded its corresponding critical values for either the PV firms or the PV-IE firms for any of the six test periods. A further con- densed summary of the two tables is provided in Table 5-3 and Table 5-6. 101 The CARD curves in Figure 5-3A and Figure 5-4A provide further evi- dence that the two materiality criteria were not important in distin- guishing the market adjustment to the extended lease disclosure requirements between the two types of firms. No evidence for the importance of the two materiality criteria can be explained with any one (or both) of the following two reasons: First, investors did not know prior to the disclosure of the actual lease numbers which particular firm was going to be in Group 3 and which firm would be in Group 2. Thus,they tended to react the same to these two groups up to the time when the financial reports were released. After the disclosure of actual numbers, investors might have suspected that a firm in Group 3 now could be a member of Group 2 in the future (or vice versa) so that the distinction between the two groups of firms is not important. Second, the two cut-off points (5% for the PV disclosure and 3% for the IE disclosure) were in fact so "high“ that the capitalized lease information which would be useful was eliminated from public disclosure. Alternatively, the SEC's cut-off points were too "low,'I which might cause some useless lease information to be made public. In either of these two possible situations, one could expect no differ- ence in the distribution of lease signals between the two groups of firms and,similarly, investors perceived in that way. Some Remarks on Test Results Several brief comments on the test results seem appropriate at this point. The first comment regards a question which may arise as to whether or not the test results obtained for the test periods 102 were also present during the pre-test period. If, for example, the evidence for significant changes in expected returns for the various test periods were also observed during the pre-test period, it would be difficult to attribute the changes in expected returns solely to the SEC lease disclosure decision. In order to evaluate this question, separate multivariate tests on mean return vectors and covariance matrices were conducted for a selected 12-month pre-test period (January through December 1972). The results of these tests are summarized in Tables 5-7and 5-8. The evidence from the results suggests that neither mean return vectors nor covariance matrices were significantly different between the treatment and control firms, either Group 3 vs. Group 1 or Group 3 vs. Group 2, during the selected pre-test period. This finding im- plies that the return distributions of the two types of firms were essentially the same immediately prior to the SEC lease disclosure decision. Thus, it appears that the significant difference in the expected values of their return distributions which was observed during the test periods was mainly contributed by the SEC decision. The second comment regards the fact that the SEC lease disclosure decision caused the security prices of the disclosure firms to differ more from the security prices of non-lease firms than from the secur- ity prices of non-disclosure firms. According to the test results, differences in the expected values and variance-covariance structure of returns between Group 3 and Group 1 were in general greater than differences in the expected values and variance-covariance structure of returns between Group 3 and Group 2. The evidence for this fact is found in Table 5-9 which was constructed by rearranging all of the .mpnou mos» uooao mco_uo:o_2xm Locuo 20$ uco comuoowcummu 2 02» mo mmpwuuogw umuompmm 02“ mo mmapo> Pou_awgu one cow Pum m_no> ow muocuoow mg» mom .Ao— .Nv ago mmo_o>12 mg» Low Eoummcw Co mmmcmmo "wuoz 103 F o .NN. ooo.. 2 o ooo. ooo. . N oooeo o _ _m2. ooo.> 2ommo.v 252oo.2 o _ om_. .oN. Amopo.v Amooo.o .m) ooo. oso. ooo.~ _oo.o oN_o.- 2m_o.- Neo.2 «No.- ooo. oom.> omoo. Pooo.- o oooco _ o oso._ “No.o _ o mp_. mom. . oooeo o _ moo.N eoo.o AomNo.v 2~mmo.v o _ oNN. _mm.P Amopo.o 2mm~o.v .m> moo. mom. ooo.~ oo~.o ompo.- PoNo.- ooo.- moo.> oeo._ oso.~ oNoo. oooo.- m oooeo zoo sow: zoo gov: 362 2o_: 362 sow: mapm>12 can: eoo e e uzmwmz N mucmcwmwwo cgopmm “zoom: N mucmcmmwwo cgoumm mnoogu wmogm>< wmogm>< mEL_2 mcomoFum_o NH uco >2 msgw2 «somopumwo >2 Amumpv UOwLwQ mehlmLa mcu Low mLouUm> CLsumm com: :0 mupzmmm umOP mum m—DMP 104 .mFoop mwcu poono mcomyocoFme gmguo msu 2o2 coo oowpoowgpmwo 2 62¢ 2o mmo>o> Foowuwco as“ Low o-m mpoo> op mpocuoom any mmm ”mpoz oooo. NNoo. oooo. PFoo. N oooco oooo. oNoo. opoo. mFoo. .m> oom._ ooo.o omoo. mmoo. ooo. ooN.N opoo. mpoo. o oooco opoo. Nooo. opoo. Fpoo. 2 2ooao N_oo. opoo. oooo. mPoo. .m> New. 22o. NNoo. nmoo. NNo. mmN.N oPoo. m_oo. m oooco _ogu:ou .uom2> Pogpcou .powgh a=_o>-2 aopo>-2 omooosoo 6_omwooom z oooooeoo geomwooom z muco22o> mucowgo> monogw m52>2 mcomopomwo 2H coo >2 msgw2 mgomopumwo >2 Amnmpv UOwLma Hmmhimsn— ms“ L02. mwumLumz mUCMwLm>OU :0 mppzmmm amp—v mum wpnmh. 105 computed values of the test statistics for both the mean vector and variance-covariance tests. In the case of the mean vectors of the PV-IE firms, for example, the computed F values in column 3 are all larger than the F values in column 5. This finding, which is in general true for the PV firms and also in the case of differences in variance-covariances, is not an unexpected one since, by construc- tion, firms in Group 1 were shown to have no long-term financing leases, while firms in Group 2 were those which had immaterial non- capitalized financing leases. Table 5-9 Observed Values of Test Statistics by Groups and by Types of Lease Disclosurea Mean Vectorsb Covariance Matrices Months Group 3 Group 3 Group 3 Group 3 vs. Group 1 vs. Group 2 vs. Group 1 vs. Group 1 PV PV+IE PV PV+IE PV PV+IE PV PV+IE * ** TPl: 21 1.595 3.030 1.070 .479 10.146x 2.116x 1.332x 1.486x sz; 15 1.423 3.304* 1.306 .105 1.670 .273 .597 .246 ‘k * TP3: 12 3.192 3.724 1.441 .029 1.349 .299 .114 .097 TP4: 9 2.477 2.263 .698 .006 .924 1.111 .719 .292 ‘k * TP5; 14 .393 3.265 .474 .394 2.420 .864 .724 1.508 * TP6: 7 2.020 4.903 .652 .356 .766 1.678 .762 .183 Note a: Prepared from Tables S-L 5-2, 5-4, and 5-5. b: Observed values of the F statistic associated with the im- plicit weight vectors only. x: Chi-square values. **: Significant at .05 level. *: Significant at .10 level. 106 The third comment here concerns the appropriateness of the mul- tivariate design for the covariance matrix test. In an effort to assess this appropriateness, correlation coefficients between the high- and low-risk assets within each primary group can be evaluated. If the correlation coefficients are significantly high, the covariability of returns of the two types of assets is also expected to be high. In this case, for example, the return variability of the high risk assets may change as a result of a change in return variability of the low risk assets or vice versa (because of covariability) even when there is in fact no information effect upon the variance structure of the high risk assets. This covariability problem is completely ignored in a univar- iate test design so that the unique effect, if any, of lease information upon the variability of returns of the high risk assets cannot be pro- perly detected. The multivariate covariance test method of the sort used in the present study can incorporate such a covariability problem in its test design by evaluating the equality of the off-diagonal ele- ments (covariances) as well as the diagonal elements (variances) of the two covariance matrices compared between the treatment and control groups. In a univariate test design, however, these off-diagonal elements are assumed to be zero. Table 5-10 shows the sample product-moment correlation coef- ficients of returns between the two risk groups of firms within each primary group for two selected test periods (as an example). A significance test indicates that these coefficients (except .530) are all significant at the .10 level. Since the within-group covar- iability of the two asset returns appears to be significantly high, this high covariability must be incorporated into statistical tests. Hence, the choice of the multivariate test design in the present 107 study is considered to be appropriate rather than the use of a univariate design, for example. Table 5-10 Correlation Coefficients Between High- and Low-Risk Assets Within Groups Disclosure Months Group 3a Group 2 Group 1 PV 21 .797 .890 .735 .965 9 .662 .870 .839 .934 PV and IE 21 .834 .785 .664 .686 9 .588 .530b .788 .784 Note a: The correlation coefficients in the left-hand column were computed by using return data when Group 3 was compared with Group 2 and those in the right-hand column were calculated by using return data when Group 3 was compared with Group 1. Because of mis- sing return data for some month(s) for certain compan- ies matched, the two related correlation coefficients for each period are somewhat different. Otherwise, both of them must be the same. Not significant at the .10 level. CHAPTER VI SUMMARY AND CONCLUSIONS Summary of Test Results The present study represents an attempt to determine if the SEC's 1973 decision to extend its lease disclosure requirements for non-financing type lease commitments had an effect on the pric- ing of securities. The SEC requires its registrants to disclose capitalized lease data in footnotes to their financial statements, with the belief that the capitalized lease disclosure of the sort required under ASR No. 147 is essential to investors in assessing the risk-return attributes of the firms with lease commitments. This research is considered as being important and timely in the sense that the results of this research may be relevant to the recently formed Advisory Committee on Corporate Disclosure of the SEC whose main charge is to determine what effect some of the SEC's disclosure requirements have had on the pricing of securities. Also, the results may be relevant to the Financial Accounting Standards Board in its deliberation concerning the lease disclosure issue. The study measured the hypothesized effects of capitalized lease information in terms of the degree of changes in both expected values (means) and variability (variance-covariance structure) of returns for the firms affected by the SEC decision. The significance of changes in expected returns was statistically tested in a 108 109 multivariate context by the use of Hotelling's T2 statistic, and the significance of changes in variability of returns was tested by using a generalized form of Bartlett's M statistic, proposed by Box (1949) for the multivariate test of homoscedasticity of variances and covar- iances. The reasons for looking into a change in measures of both central tendency and dispersion of return distribution and the ration- ale for conducting the multivariate tests were explained in Chapter III. Based upon the test results on measures of the central tendency of return distributions, the following tentative conclusions appear warranted: (l) The price effects of the SEC lease disclosure decision were found to exist when those effects were evaluated by compar- ing the disclosure firms (primary treatment group) that had "material" lease commitments (as defined by the SEC) with the non-lease firms (control group) which had no long—term lease commitments. However, when the disclosure firms were compared with the non-disclosure firms (control group) with "immaterial" lease commitments, no sign of lease infor- mation effect was observed. (2) The evidence suggests that the price effect of both PV and IE disclosure was relatively greater than the price effect of disclosing the PV numbers alone. For the firms which disclosed both PV and IE numbers of noncapitalized financing leases, there is strong evidence that both anticipatory and gx_p9§t market reaction took place with respect to almost every event of the SEC lease decision, including the 110 publication of the proposal of ASR No. 147 and the dis- closure of actual PV and IE numbers in financial reports. However, for the firms which disclosed the PV numbers alone, the differential market response was observed only for the (12-month) period covering the formal announcement of the SEC decision and the actual disclosure of the PV and IE numbers. There was some indication of further mar- ket reaction following the actual disclosure, but this reaction was not sufficient to reject the null hypothesis of no information effect. This failure to reject the null hypothesis was probably caused by the relatively low power of the test. (3) Evidence collected in this study suggests that the informa- tion effects of the various events of the SEC lease decision were risk-dependent in the sense that the magnitude of the effects varied according to the firms' riskiness. In gen- eral, the high risk firms were more sensitive to the events of the SEC decision than were the low risk firms. (4) Capitalized lease disclosure did in general have an adverse effect on the valuation of the firms. The high risk firms tended to be more adversely affected than the low risk firms. This finding is consistent with Nelson's results which suggest that lease capitalization generally has an adverse effect on financial ratios. Somewhat in contrast to the general results observed in the mean vector tests, evidence from the covariance tests suggests that there were no lease information effects on the variance-covariance structure 111 of returns for either the PV disclosure firms or the firms with both PV and IE disclosure. Although the test results on the covariance matrices for the PV firms appeared to be significant, this signifi- cance seemed to be a result of random chance rather than an outcome of the SEC decision. Regarding the importance of the SEC's two materiality criteria, 5% for the PV disclosure and 3% for the IE disclosure, the test re- sults of the present study suggest that neither of these criteria seems to be important to investors in differentiating the disclosure firms from the non-disclosure firms. All the statistical tests con- ducted here showed no significant differences in mean returns and variance—covariances between the disclosure firms and the non-dis- closure firms. Rather, the most important distinguishing factor seems to be whether or not a firm has lease commitments. Conclusions Based on the empirical findings of the present study, it can be concluded that the SEC extended lease disclosure decision of 1973 had an effect upon the pricing of securities for those firms which were materially affected by the decision. The information effect appeared to exist only with respect to expected values (means) of security returns, but not with respect to return variability. This implies that the SEC decision contributed to a significant shifting of the entire distribution of security returns of the affected firms without change in dispersion. Notice that a significant change in either central tendency or dispersion (or both) of return distribution is a sufficient condition for the hypothesized information effect to 112 exist. Furthermore, the information effect was present with respect to the earlier events of the SEC decision as well as the actual dis- closure of capitalized lease data in financial reports. These findings tend to support the SEC's contention that capital- ized lease information is important to investors. Moreover, the em- pirical evidence supports the traditional view which asserts that capitalized lease data have information content. Indeed, evidence from the present study indicates that the counter-view claiming no information content of capitalized lease data is rejected. The conclusion reached here in favor of the existence of the information content of capitalized lease data is consistent with the findings of Hamada (1972) and Beaver, et. a1 (1970). They discovered that accounting risk measures, including capital structure variables, have a high association with the market (relative)risk which is a determinant of the first moment of return distributions. However, the conclusion of the present study is inconsistent with the findings of Benston (1973) and Hagerman (1975) whose studies suggest that SEC disclosure requirements have had little or no observable effect upon security prices. Study Contributions The main contribution of this study lies in the fact that it is one of the first attempts to empirically evaluate the information con- tent of capitalized lease data via the assessment of the effects upon security prices of lease disclosure regulations by regulatory agencies. Hopefully, the findings of the present study will provide a useful basis for the evaluation of the lease disclosure issue by the SEC's 113 Advisory Committee on Corporate Disclosure and the Financial Account- ing Standards Board. In the area of research methodology the present study employed a multivariate testing procedure which was considered most appr0priate for evaluating the potential information effect of an accounting event. Another possible contribution of this study is its warning that the use of a single critical time point (like that in the before-after type of test design) at which an event is assumed to take place may not be sufficient to thoroughly detect its hypothesized information effect when that event is a final outcome of a series of related events. In the present study multiple time points were selected on the basis of the various events of the SEC decision, and separate tests for various sub-periods, in addition to an omnibus test for the entire test period, were conducted. The test results for the various sub-periods were found to reveal certain information which could not be obtained through the omnibus test for the entire test period. Study Limitations The conclusions for the present study may be qualified because of the following limitations: the small size of samples, the indepen- dent-group assumption in the variance-covariance test,1 the homogen- eity of control group firms, the independence of returns over time, and the sample firm selection criteria (the NYSE firms which have the same fiscal year ending December 31). The sample size in the present study was the total number of months for which returns (RH RL) of the two risk groups were computed. The largest total number of months for a test period was twenty-one, 114 while the smallest was only seven. This small sample problem could be alleviated somewhat by choosing a longer test period or using weekly (or daily) return data for the selected test periods for this study. The independent-group assumption is a difficult one to be met, given the grouping design of this study which employed pairwise match- ings of treatment and control firms. An attempt to find a multivariate variance-covariance matrix test technique which can be applied to the dependent-group case was unsuccessful. Also, it is not known what impact the violation of the independence assumption would have on the test results on covariance matrices. The homogeneity of control group firms employed in the present study was not tested. The study assumed that firms in each of the two control groups came from a homogeneous population which implies that the returns in a given control group were drawn from the same population. Regarding the independence of returns over time, the present study assumed that monthly return differences, (8” 8L) = (R? - HE R# - Rt), in the mean vector tests as well as raw returns, (RH RL), in the covariance matrix tests were identically distributed in every month. However, this assumption may be questionable since the behavior of returns in the assumed disclosure month, March 1974, might not have been the same as the behavior of returns in the non-disclosure months 2 If return behaviors in different months were during that same year. not identical, it would be difficult to explain what the summary sample statistics (means and variances) really mean. However, because of the feature of the pairwise matching of the firms and since the sample firms, both treatment and control, had the same fiscal year-end and 115 the same disclosure month, a problem (if any) arising from violating the assumption of independent and identically distributed returns over time was expected to be substantially alleviated in the present study. The present study used only the NYSE firms having December 31 as a fiscal year-end. Therefore, the findings may not be generalized to the NYSE firms having a different fiscal year-end nor to the firms which are registered on other stock markets. Suggestions for Future Research The above potential sources of limitation open avenues for future research in the subject area considered by this study. In addition,there are also other possible options for future research. For example an an- alysis similar to the present study could be conducted on the non- disclosure firms (in Group 2) with "immaterial" lease commitments after they begin disclosing the PV and/or IE numbers. Then it can be evaluated whether or not the market re-evaluates these firms upon the disclosure of such numbers. As another possibility fincresearch, grouping of firms by different risk classes may also be made on the basis of the ratio3 of the PV to the amount of long-term capitaliza- tion and, then,some sort of statistical significance tests can be conducted. Also, the research design of the present study can be refined through a better matching of firms on the estimates of error terms (residual returns), in addition to Beta, and/or on the SIC code, if it is assumed that industry factors can explain a significant portion of security price behavior. CHAPTER NOTES ll. 12. CHAPTER I NOTES The Disclosure Policy Committee of the Securities and Exchange Commission, Disclosure to Investors: A Reappraisal of Federal Administrative Policies Under the '33 and '34 Acts (The Wheat Report);Chapter II, "B. The Philosophy of Disclosure," Chicago, Illinois: Commerce Clearing House, Inc., 1969. Rule 3-16 contains provision about the disclosure of commitments and contingent liabilities including lease commitments. The Securities and Exchange Commission, "Securities Act of 1933 Release No. 5401" (June 6, 1972), in SEC Docket, June 19, 1973. The Securities and Exchange Commission, ASR No. 147 "Notice of Adoption of Amendments to Regulation S-X Requiring Improved Dis- closure of Leases" (October 5, 1973), in SEC Docket, October 23, 1973. See ”C. Amendments to Regulation S-X" of ASR No. 147. See "A. Introduction" of ASR No. 147. Myers, John H. Accounting Research Study No. 4 "Reporting of LeaseS'hiFinancial Statements," New York: AICPA, 1962, p. 3. Ibid., p. 38. Ibid., p. 49. APB Opinion No. 31 "Disclosure of Lease Commitments by Lessees,” AICPA, June 1973, Paragraph 7. A similar statement is also found in APB Opinion No. 5, Paragraph 16. FASB Exposure Draft "Account- ing for Leases," Stamford, Connecticut: Financial Accounting Standards Board, 1975, Paragraph 81. The Subcommittee on Leases of the AAA Committee on Financial Accounting Standards, "Response to Financial Accounting Standards Board Discussion Memorandum on Accounting for Leases," a supplement to The Accounting Review, 1976, p. 229. See Vatter (1966), p. 135, Buff (1971), pp. 21-3, and Heufner (1970), pp. 30-6. 116 117 (CHAPTER I NOTES, Cont'd.) 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 223. 24» The argument claiming that capitalized lease data provide no in- formation content is not necessarily the same as the argument against the disclosure of such data. For example, management may oppose the disclosure of capitalized lease information because the potential outcome of this disclosure would adversely affect the credit standing of their firm. In this case management assumes that capitalized lease data could have information content. But they oppose disclosure of such data because of a worry about its potential adverse effect. Also, see FASB Discussion Memorandum "Accounting for Leases," Stamford, Connecticut: Financial Accounting Standards Board, July 1974, pp. 22-4. Ibid., p. 32. Ibid., pp. 33-4. For further discussion in this context, see Section V of FASB Discussion Memorandum (l974). Defliese (1973), p. 24. Donaldson (1962), p. 123, and Axelson (1971), p. 52. See ”B. Interpretations and Comments" of ASR No. 147. FASB Discussion Memorandum (1974), pp. 18-9. Nurnberg (1973), p. 385. Gonedes and Dopuch (1974), p. 80. Before turning to the next chapter, it is appropriate for one to note that this study will restrict its scope only to the disclosure issues related to the accounting for leases and to the disclosure of leases by lessees. Thus, issues associated with leases from a valuation point of view or those from the lessors' point of view will not be a concern. 10. 11. CHAPTER II NOTES ——_——— The full text of this Bulletin was later restated in Chapter 14 of ARB No. 43 (1953). Paragraph 7 of Chapter 14 of ARB No. 43. The Bulletin requires disclosure, in financial statements or in notes thereto, of: (l) the amounts of annual rentals to be paid with some in- dication of the periods for which they are payable and (2) any other important obligations assumed or guarantees made in connection therewith. (See Paragraph 5 of Chapter 14 of ARB No. 43.) See "Summary and Conclusions: Lessee" of Chapter 1 and "Proposed Lease Presentation - Lessees" of Chapter 4 ASR No. 4. APB Opinion No. 5 "Reporting of Leases in Financial Statements of Lessee,” New York: AICPA, September 1964. Paragraph 5 of APB Opinion No. 5. Paragraphs 10, 11, and 12 of APB Opinion No. 5. The Board argued that leases covering merely the right to use property in exchange for future rental payments do not create an equity in the property and are thus nothing more than executory contracts. Accordingly, it was recommended that information about leases of this type should be disclosed in schedules or notes rather than in the body of the financial statements (Paragraph 14). Paragraphs l4 and 16 of APB Opinion No. 5. Paragraph 18 of APB Opinion No. 5. Paragraphs 8-10 of APB Opinion No. 31. The Opinion defines a noncancelable lease as one that has an in- itial or remaining term of more than one year and is not cancelable, or is cancelable only upon the occurrence of some remote contin- gency or upon the payment of a substantial penalty. (See Footnote 2, Paragraph 9 of the Opinion.) 118 119 (CHAPTER II NOTES, Cont'd.) 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. Paragraphs 1 and 7 of APB Opinion No. 31. Of the remaining firms, 50% in 1973 and 52% in 1974 were such firms that had long-term leases but did not capitalize them. And 23% of the 600 firms in 1973 and 18% in 1974 were shown to have no long- term leases. (See Accounting Trends and Techniques (1975), p. 188.) See "A. Introduction“ of ASR No. 147. Securities Act of 1933 Release No. 5401 "Notice of Proposed Amendment to Regulation S—X...," in SEC Docket, June 19, 1973, p. 1. See "A. Introduction" of ASR No. 147. As preliminary works, the FASB issued a discussion memorandum on accounting for leases in July 1974 and an exposure draft in August 1975. See Note 11 of this chapter for the definition of a noncancelable lease. See "Rule 3-16(q). Leased assets and lease commitments" in "C. Amendments to Regulation S-X” of ASR No. 147. See items (1) and (2) of "C. Amendments to Regulation S-X" of A§R_ No. 147. ———————__ See Note 19 above. See item (4) of "C. Amendments to Regulation S-X" of ASR No. 147. The Release defines the amount of long—term capitalization as the sum of long-term debt, stockholders' equity, and the PV of the minimum lease commitments. (See item (4) - (i) of "C. Amendments to Regulation S-X" of ASR No. 147. The notion of materiality used hereafter is the same as that defined by the SEC. By the term "objectives," Nelson means what the financial ratios try to measure. For example, the objective of the current ratio is to measure a firm's liquidity or ability to meet short term obligations and the objective of the return on total capital is to measure the rate of return on investment. Nelson (1963), p. 54. Beaver, et. al (1970), p. 679. Ii II 1 10. 11. CHAPTER III NOTES An example of using a multivariate analysis for the evaluation of information effect can be found in Gonedes (1975). See the second section of Chapter II about the definition of the SEC's two materiality criteria, 5% for the PV disclosure and 3% for the IE disclosure. For further discussion about the normality of return distribution, see Fama and Miller (1972), pp. 216-7 and 259-65. However, no difference in the first moments of return distributions between groups does not imply that their variancefi)= are Etge same. Since) the tvariance of return 15 defined as Var ')1( 2, conditio of E(RT) = (HCL )alon goes not neceSsarily imply) Var(_T = Var(l unless E( E(§' ), which is not known. Gonedes (1975), p. 222. As examples, see Ball and Brown (1968), Fama, Fisher, Jensen, and Roll (1969), and Gonedes (l974). Gonedes (1975), p. 223. The sample firms in the present study comprised sixty PV disclosure firms and thirty-nine firms with both PV and IE disclosure. Each of these two disclosure groups was again divided into two subgroups, high and low risk. (See Figure 3-4 later in this chapter.) Then, the average monthly return of the individual component returns in each of the two risk groups was used as a sample unit. By risk dependency is meant whether or not assets in different risk classes are affected differently in a systematic manner by the same source of information. Gonedes and Dopuch (l974),pp. 81-7 and Gonedes (1975), p. 223. August 1973 was about two months subsequent to the publication of the ASR No. 147 proposal. Therefore, there is a possibility that the Beta estimates used in matching firms might have already been contaminated by anticipatory market reaction to the proposal. 120 121 (CHAPTER III NOTES, Cont'd.) 12. 13. 14. 15. 16. 17. 18. However, it was thought that the inclusion of only two months in the 60-month Beta estimation period would not be very critical. Also, August 1973 was six months before the first disclosure month (March 1974) of the PV and IE numbers, according to ASR No. 147, by the December 31 fiscal year-end firms. It was believed that the period of six months would be long enough to avoid any contamination caused by the effect of the PV and IE disclosure. Also, going back further would pick up some additional noise fac- tors which could mask the unique effect of the capitalized lease disclosure. The reason for assuming March 1974 to be the first disclosure month is explained in Note 25 of this chapter. The single-factor market model is given by _ ’\I Nj-a+6j11M+ej where a M c. J This independence assumption implies that returns (RH and RL) are identically distributed in every month, although this assumption may be questioned since the behaviors of returns in different months (for example, March and September) may not be same. However, it is believed that, because of the feature of pairwise matching between the treatment and control firms which have the same fiscal year end with the same disclosure month, an impact of a possible violation against the independence assumption could be considerably reduced. Therefore, the test results of the present study would not be critically distorted. (See Chapter V1 for further discussion of this problem.) risk-free rate return on the market portfolio % m error term with E(cj) = O and Cov(ei cj) = O for i f j. The covariance test here is based on the same idea that inferences about the equality<1ftwo variances which use n pairs of observations between two dependent samples can be made within the context of a univariate test (for example, the standard t-test). For further ex- planations and an example of the test for the equality of two vari- ances using two dependent samples, see Lord (1963) and Glass and Stanley (1970). pp. 306-8. For further discussion about the joint level of significance in a multivariate context with an example, see Bolch and Huang (1974), pp. 76-77. This T2 statistic was originally suggested by Hotelling (1931) as a generalization of the standard t-test statistic. This is because weighted sums of normal variables are themselves normally distributed. See Morrison (1976), pp. 129-131 for the derivation of the 12 from the univariate t-statistic through a maximization process using a weight vector. 122 (CHAPTER III NOTES, Cont'd.) 15?. 22C)- 221 - 2222- 2223- 24- 255- 265 A so, as in the case of the univariate t-test, the multivariate T -test is known to be robust against violations of the normality assumption. (See Harris (1975), pp. 7 and 87.) See Morrison (1976), pp. 134-6 and Bolch and Huang (1974), pp. 86-7 for further discussions and an example of this idea. Gonedes (1975), p. 226. Bartlett (1937) developed the following statistic: G M = N ln S - Evk ln Sk where S = (Z vk Sk)/N, N = Z Vk’ and k denotes a group (k = l, k k 2,...,G) Harris (1975, p. 85) says that this M statistic implicitly assumes independence between groups compared, although neither Bartlett (1937) nor Box (1949) mentioned this assumption explicitly in his original paper. If the independence assumption is critical, the choice of the M statistic will be inappropriate for the present study since groups compared were not independent of each other. It is not exactly clear what impact a violation against the inde- pendence assumption would have on the test results obtained. Never- theless, since it was hard to find a multivariate test statistic that exactly fits the covariance matrix tests of this study, it was decided to use the M statistic as a possible alternative. The selected test periods will be explained in the next section of this chapter. The data collection for the present study showed that about 95% of the sample firms filed their lO-K reports (containing the PV and/or IE numbers) with the SEC during March every year from 1972 to 1974. Therefore, it seems appropriate to use March 1974 as the critical month in which the incremental lease information was first publicly disclosed under ASR No. 147. The choice of a critical time point for the evaluation of an information effect is not consistent in several studies. For ex- ample, Gonedes (1975) used the actual disclosure month (March for the firms with the December 31 fiscal year-end) as a critical month. On the other hand, Sunder (1973) utilized the end of the accounting period as a critical time point in assessing an effect of annual earnings announcement, rather than the date of the formal announcement of the earnings. One must be cautious in interpreting the test results to be obtained since some of these events also overlapped with the events related to APB Opinion No. 31. The Opinion was formally announced in June 123 (CHAPTER III NOTESoCont'd.) 27. 28. 1973 and effective as of December 31 of the year. However, three reasons can be pointed out for asserting that any information effect observed during the selected test periods could be ascribed to the SEC decision. First, the tests were conducted on the capitalized data of noncapitalized "financing" lease commitments, the data which are not available under the Opinion. (See ChapterII.) Second, although the Opinion recommends disclosing the PV of noncancelable leases in general, it entirely ignores the IE of such leases. There- fore, any observed information effect of both PV and IE disclosure can be exclusively ascribed to the SEC decision. Finally, the dis- tinction between materially-affected firms (Gourp 3) and immaterially- affected firms (Group 2) is possible only with the SEC's two mater- iality criteria, 5% for the PV disclosure and 3% for the IE dis- closure. The Opinion provides no single guideline for differentia- ting the materiality of lease commitments. Therefore, the implications of the test results from comparing Group 3 with Group 2 can be interpreted only with respect to the SEC decision. This test period and TP5 below are overlapped by eight months with the Beta estimation period (the Betas used in matching firms). However, the inclusion of eight months in the 60-month Beta estima- tion period was not considered to be so critical that the estimated Betas could be significantly contaminated by the lease information effect. Therefore, it is believed that the test results for these two periods are not significantly distorted. It should be noted, however, that some firms in Group 2 which did not meet the SEC's two materiality criteria for 1973 and 1974 may be qualified as the members of Group 3 in the future since they could meet any (or both) of the two materiality criteria. Like- wise, some firms in Group 3 may not be able to meet the materiality criteria in later years so that they may become the members of Group 2 in the future. In this sense the firms in both Group 3 and Group 2 can be considered homogeneous with respect to their long- term lease commitments so that a comparison of Group 2 with Group 1 can be a possibility for separate statistical evaluation. CHAPTER IV NOTES —— The definition of each group and the reason for using two control groups were provided in the previous chapter. Disclosure Journal, the editions of 1973 and 1974, Leasco, Inc. The reason for going back to September 1968 is that the Beta estimates used in matching firms were computed on the basis of sixty monthly return observations for the period beginning Septem- ber 1968 and ending August 1973. The validity of including these firms with the immaterial lease commitments in Group 3 may be questioned since Group 3 is by definition supposed to consist of those firms which reported the capitalized lease data by meeting the materiality criteria. The reason for including those firms in Group 3 is based on the fact that (1) "disclosure" was of a primary concern in the present study rather than the materiality of lease commitments and (2) the benefit derived from increasing the number of sample firms by in- cluding such firms would exceed the cost of excluding them from the samples. It was hoped that the inclusion of a small number of such firms, for example, four and seven for 1973 and 1974 in the total sample of PV firms, would not distort the test results critically. The same logic was also applied to the case of the firms with both PV and IE disclosure. Since the present study used monthly returns of the two different risk groups as sample units, the inclusion of such firms in Group 3 did not affect the sample size of return observations; only the number of securities in each risk group was affected. Of the seven PV firms in 1974, for example, three were included in the high risk group, while the remaining four were in the low risk group. For one firm the negative PV numbers were reported since the capit- alized amount of rental income from subleases exceeded the capital- ized amount of primary future rental payments. (See Appendix A regarding the way that the PV number is computed.) The term "neg- ative" is used in the sense that this PV of subleases in fact reduces the amount of liability which is different from typical cases wherein lease capitalization tends to increase the magnitude of a firm's liability. 124 125 (CHAPTER IV NOTES,C0nt'd.) 6. 10. 11. 12. See Note 12 of Chapter III regarding the form of the simple market model. . Some test periods (TPl, TP2, and TP5) which were explained in the third section of Chapter III were overlapped with the Beta estima- tion period. However, the overlapped periods were relatively short (eight months in TPl, three in TP2, and eight in TP5) so that the test results were assumed not to be critically affected. See Security Risk Evaluation (September 1973), Glossary. Fertuck (1975), p. 847. Because of the relatively strict definition of control firms (as seen in Figure 3-2) and their matching criteria, it was hard to meet the fourth condition regarding the fiscal year end for all control firms when these firms were selected. As a result, there were three firms in Group 2 which have non-December fiscal year- ends (one with October, one with November, and one with January), while five firms in Group 1 failed to meet the December fiscal year-end condition (two firms with February, another two with Novem- ber, and one with January). For details of this test, see the last section of Chapter V. The tape is edited by the Center for Research on Security Prices (CRSP) at the University of Chicago. There are two different types of returns stored in the tape: RETl and RET2. Returns of the first type are adjusted monthly returns, including all distri- butions, while returns of the latter type are adjusted monthly returns, including only price changes and non-taxable distribution. The present study used RETl and converted returns defined in RETl 2nto)continuously compounded rates of return on the basis of Eq. 4-1 . CHAPTER V 1911-25 See the second section of Chapter 11 about the definition of the materiality criteria. This implicit vector is one which maximizes the value of t2 and, thus, the observed value of F. All values of T2 and F associated with other weight vectors should be less than or equal to the values of T2 and F related to this implicit weight vector. Notice that this fact is true for all test periods as seen in Table 5-1 (and later in Table 5-2). See the second section of Chapter III for a theoretical discussion thereof. This is the critical value based on the two-tailed test. The present study used the two-tailed critical values of test statis- tics for a decision as to whether or not the null hypotheses are accepted. The reason for using the two-tailed critical values is that no assumption was initially made about a specific direction of lease information effect upon the pricing of securities. The assumption was made on the basis of an empirical finding that capitalized lease disclosure did not necessarily affect the firms (financial ratios) in one direction. (See Nelson (1963).) The overall CARD curve was constructed using Eq. (5-2) and the CARD curves by different risk groups using Eq. (5-4) below: 11 at :32] ('I‘zjtT - 11th) / n (5-1) _ 21 __ CARDt = 1:1 at (5-2) \Nhere n = total number of firms (sixty for the PV firms and thirty- nine for the PV-IE firms) t = a month,from January 1973 to September 1974 11 d = g (113?“ - 113.5th / n (5-3) 126 127 (CHAPTER v NOTES, Cont'd.) 1C).. 1 1 - 122 - 21 = -G 2 dt G t i=1 CARD where G = the high and low risk group For further details, see Table 4-1 in Chapter IV. See Table 4-1. See Table 4-3 about the means and variances of Beta estimates by the different risk groups. See Table 4-1. A separate evaluation of the importance of the SEC materiality guidelines on the basis of the empirical findings from the present study is given in the third section of this chapter. See Note (2) of Table 5-1 for the exact values at various selected fractiles of the F distribution. See Table 4-3. The reason why the low risk firms were not adversely affected is not clear. CHAPTER VI NOTES See Note 24 of Chapter 111 about this assumption. See Beaver (1968) regarding empirical evidence about the difference in return behavior between disclosure months and non-disclosure months. See Chapter II about the definition of this ratio and its impor- tance viewed from the SEC's standpoint. 128 APPENDICES APPENDIX A THE FINANCIAL STATEMENTS FOR A MATERIALLY AFFECTED FIRM (BEFORE AND AFTER THE SEC'S 1973 LEASE DISCLOSURE DECISION) APPENDIX A THE FINANCIAL STATEMENTS FOR A MATERIALLY AFFECTED FIRM (BEFORE AND AFTER THE SEC'S 1973 LEASE DISCLOSURE DECISION) A. Before the SEC Decision Consolidated Balance Sheet at Dec. 31, 1974 and 1973 1974 1973 Assets: Current assets 142,380 128,940 Noncurrent assets 214,620 174,960 TOTAL ASSETS 357,000 303.900 Liab. & SE: Current liab. 56,370 47,370 Long-term debt 65,290 45,470 Stockholders' equity 235,340 211,060 TOTAL LIAB. & SE 357,000 303,900 Note: Lease Commitments Consolidated Statement of Earnings for 1974 & 1973 1974 1973 Net sales 133,570 120,982 Cost of sales 87,420 82,355 Gross Profits on sales 46,150 38,627 Operating expenses including rentals 21,645 17,737 Operating income 24,505 20,890 Interest expense 3,280 2,650 Income before taxes 21,225 18,240 Incomelaxes(50%) 10,612 9,120 Net Income 10,613 9,120 Earnings per share 2.65 2.14 - Total rental expenses for all noncancelable leases amounted to: Financing leases Other leases Less: Rental income from subleases 1974 1973 4,687 4,972 842 490 5,529 5,462 250 235 5,279 5,227 APPENDIX A (Cont'd.) 130 -The future minimum rental commitments as of December 31, 1974 for all noncancelable leases were as follows: Type of Lease Type of Property Financing Other Real Equip- Total Leases Leases Properties ment Other 1975 3,099 2,637 462 2,750 225 124 1976 3,724 3,260 464 3,386 218 120 1977 3,717 3,251 466 3,384 216 117 1978 3,700 3,242 458 3,382 208 110 1979 3,696 3,240 456 3,381 207 108 1980-84 18,090 16,092 1,998 17,518 572 . 1985-89 17,216 15,699 1,517 16,763 453 1990-94 16,203 15,031 1,172 16,203 . After 21,110 18,775 2,335 21,110 1994 .__. TOTAL 90,555 81,227 9,328 87,877 2,099 579 B. After the SEC Decision The Consolidated Financial Statements: Same as before Note: Lease Commitments -Total rental expenses for all noncancelable leases: Same as before -The future minimum rental commitments existing as of December 31, 1974 under all noncancelable leases: Same as before 131 APPENDIX A (Cont'd.) -The present values of all noncapitalized financing leases at December 31, 1974 and 1973 were as follows: Interest rates used in PV compt'n Present Values Asset Category_ 1974 1973 1974 1973 Real property 8.5% 8.3% 34,257 33,672 Equipment 7.8% 7.5% 1,252 1,086 Others 7.2% 7.0% 265 193 Less: PV of rentals from subleases 6.7% 6.5% 2,565 2,424 Total 33,209_ 32,527 —If (1) all the above noncapitalized financing leases were capitalized, (2) the related property rights were amortized on a straight line basis, and (3) interest costs were accrued on the basis of the outstanding PV lease liabilities, net income for the two years ended December 31, 1974 and 1973 would have been reduced as follows: 1974 1973 Amortization of lease rights 3,025 2,988 Interest costs 3,252 3,024 6,277 6,012 Less: Rental expenses 4,063 3,924 2,214 2,08 Less: Income taxes (50%) 1,063 1,002 Decrease in net income 1,151 1,086 APPENDIX B KEY FINANCIAL RATIOS BEFORE AND AFTER THE SEC DECISION (1974) APPENDIX 8 KEY FINANCIAL RATIOS BEFORE AND AFTER THE SEC DECISION (1974) Before After Change (1) (2) ((2)-(1)) /(1) Liquidity ratios: Current assets to current liabilitiesa 2.52 2.40 -.05 Current assets to total assets .40 .36 -.10 Capital structure ratios: Total liabilities to total assets .34 .40 +.l8 Total liabilities to stockholders'equity .52 .66 +.27 Return on investment ratios: Net income to total assets .03 .02 -.33 Net income to stockholders' equity .05 .04 -.20 Earnings per share 2.65 2.36 -.11 aIn the computation of this liquidity ratio, it is assumed that the current portion of the lease PV is $3,000. 132 APPENDIX C LIST OF SAMPLE COMPANIES BY GROUPS AND MATCHINGS OF FIRMS .00 022021002020 .2200 0000020 00202 .00 20202 22030: 0202 .2200 m e 0 .2200 0020002 2002: .2200 20000002 .2200 2000023 200002 .2200 0200020 oo0eooe2ooeo ooooo< .2200 00000200 .020 ..02020 202000222 .00 200202 .2200 02200 .2200 02002 .oo oe0o0o2 0202 2020 .2200 0020>00m 0000202200 .00 0om2< NWE202 0000012020 0 20020 .00 20302 00200002 200020s< .2200 00200002 2020020 .020 .0020202 .00 000002 00002002 .020 .20>02 202 .020 ..0220 .m.: .00 00002022 m20>2 .60 220002023000 00200002 22000>000 .020 .002020 0200200m .2200 0202200002 .2200 000 002< .2200 00002 .020 ..0220 0030000 .020 ..0omm< 0020200000 .22oo 2ooo2 00023 .2200 2002202m 0202002 .00 002202 020 0302020002 A02202 022000000212020 N 2:020 mzm02 2o moz0xupm mm0z<2zou m02z 302 00002020 .2200 202005020 .2200 222200 .020 .22020202 2000200220 .2200 0020000220020 0002020 .020 .200200 .2200 200000 .2200 000001020 000200 .020 ..0020 20200 .00 000002002 0002000< .020 .00us< .020 .0005< .020 .22000202 200020E< .020 .0020020< 200020e< .2200 00005020 00000< .020 .0020< Nma202 02200000020 m 2:020 20 00 00 00 m0 N0 0 ,— 133 lr—NMQ'LOQDNCDON 0 Z 134 .00 200200202 .2 00202 .00 000202 .020 .020202 002000 .2200 02202 .2260 000020202 2 z o .00 0000 000022022 0000020 .020 .002002 0 0202 .2200 200x00 .00 20202 0020000220020 .2200 00002020 0020020 .2200 0000022 .00 22:0 .00 000 20022 22020200 .2200 00002 .00 20002 2202 .2200 020E2< .020 ..0220 0022< .00 202 202202 .020 .E0202<-20202 .2200 020E2000>00 0020200 2000020202 000020202 .020..0220 200202002 .00 22020000 .00 00002000: 20002> .00 00200002 0200200 .2200 20002025 200 002.0220 .00 0000020020 02000< .2200 2 < 0 .020 .0203 20203 02020>003 .00 2202-020201200200 .2200 020000: 0000200 .00 0202x02 .020 .020~2< .000 020000020020 200020E< .2200 00020000 .020 .200002202100000202 .00 200000220202 .2200 000 0020 .020 .2000000 .020 ..0220 020020< .020 .20030200202 .020 .20302020: 0202000 .2200 20022 .00 002022020 0020200 200020E< .020 .0_0200 .00 0 000212200000 .2200 000: 0020200 .020 ..0220 0:022 .020 .02002 202002 .2200 2200202002 .2200 0203020 02000002 .2200 0200022 20< 22022 .020 .002000 02020020 200020 .020 .00202 20< 2200000 .00 00002020 300 020 .2200 0002000 00 .2200 0003022 0 20022020 .00 02020 .00 200222 0 0200 202000 .00 000 00020200200 .020 .0020020< 00020200200 .020 ..0220 000000 .020 ..00 0 2200002 .000200 .2200 02020002 0 000 22000 .00 020000>20 2000 .2200 00230002 0000020 00 mm 20 00 mm 20 00 N0 0m 00 0N 0N 2N 0N mN 0N 0N NN 0N ON 00 00 0.0.02000 0 20222222 135 .2200 220002002 00000 .020 .0202 20002 .00 000 0 000 000N< .020 ..220 200 .020 ..0220 00000 0020200 .00 00200202 .2200 000002020 0 00020202 220200020 .2200 002202 .00 20302 020220200 .00 x0000: 0 2000200 .2200 20222000021002E< .00 0 2000 .020 .00202 .00 20002000 0 0002 0200< 00020220020 00000202 .2200 .02020 20300002 .2200 000212200202 .2260 02020202 00o 00000 202020 .2260 02026202 000022 .620 ..0020 02000262 02202 .00 0000 0202002 2020202 .00 00020200 200020E< .020 .22022000 200020E< .2200 20022 2200002 .00 00002202 .020 ..00 00200 .00 000000002 000000 0020 .2200 0000000200 .00 202222 230202 .020 ..00 220> 302 00 2200 .020 ..0220 000320202 .00 2002202002 .020 ..0220 0200 .020 ..0220 000220 200022 0022020000 22020200 .020 ..00 0 02202 .3 .0220 0202 00000 22002000 .020 .0200 0 2000020302 .2 .2200 0000202 .020 .22000202 0020200 .2260 2022200002 .020 .00020000102030 .00 220220 00200002 .00 20000002 2000 00200002 .020 .2202222 .020.000220002 200220002 0 02002 .60 2000020 260002 .2260 0620 0002>002 .2200 0.20020202 .00 .002 020500 00020202 .020 .02200 2200202 020202 .2200 0200002 .00 00000 0202002 .00 200022 .020 ..0220 2002 .020 ..0220 22020200 2000 000202 .2200 20000210202 .020 ..0220 0020200 00020000 .2200 0020000220020 2 0 .020 .0020000220020 000: .020.0220 202000: .020 .0.2002 .00 20022000 .2 .0 00 mm mm mm 00 mm 20 00 N0 00 om 02 02 22 02 02 22 02 N2 02 02 2.2.02002 0 x0ozm22< 136 .00 000020» .0 .2 .00 202500 .2200 0205000>20 0002 020000>20 00000202 00202 .2200 000020202 2002020 00202 .2200 2202002220 .2200 000220020 .2200 020002020 .2200 0020002u0002000 .2200 20002220 .00 0 20202 .2200 020250 .00 020000 .00 00200000 0 000 00002200200 .00 0202002 00002 .020..0220 0 2 2 .2200 020202 0205302 .00 020202 00002 .2200 205202 0 0200000 .2200 020202 20200000 02000000225500 000000 0000200 .2200 00002 200200 .020 ..0220 00220 000 .2200 0000 .2200 00002-05022 02202 00000 00202200 .2200 000020202 000220002020 .2200 2022000 .2200 020022-0000220 .2200 0002200 .2200 00002-2202 .020 .02002 2202 .020 .02022 0 0000000 .2200 2 > 2 00205220020 0 020500 200202002 .00 .2 020200 02202 2002000 020 .00 2000022 20000220000 .2200 000000002 0 020220000 0020200 .2200 202000200 02000 .020 .2000020 .020 .000320202 .2200 02002 00005020 0 52205202 200002 .20000 00 .00 000 22022000 00 .2200 220200 00 .020 ..0220 000202 000320200 00 .2200 250020200 020 02 .2200 00000: .0.00220020 0000200 02 .2200 0>020000 22 .00 .002 000>000 02 .020 .00222-2o->00 02 .020 .002000 2030000 22 .ucH ..0UCH 2202mm$mm mm .020 ..0220 02002202 .2 .2 02 .00 0000 2002 02 .020.0220 020502 02 .020 2050022 00 .00 020500 20002 002022 00 .020 .00300u202002 20 .00 220002002 02000022 00 .020 .002000 0222 0002002 00 .2200 00032202 20 .020 .0203202 20202 20002052 202 00 .020 .00200202 0020200 20202 00 .2200 0000 20002052 20202 00 2.0.05000 0 20020222 137 .000 .00302 0 00500000 02020202 0 02000050 20002052 .2200 2300002202 00222 .2200 02020022 .2200 005302 0002052 20 .2200 00020250 500020 2000000225500 20222 .020 .020520200 00020020 .2 .0 .22002022022203000 .2200 000020202 00002 00000 2022200022 05000023 .00 0202000222 2200002 0002020022 0 020>002 0020000220020 000000000 02000000225500 .2200 020002 0020200 .2200 00000 00022202 .00 20302 00020000 .20000 00 .2200 000020202 200002200 .020 .0202020302 020022 .2200 0002002220 200022 .2200 .000>20 00020200200 .020 .02030200022 200020000200 .2200 200302220 0002052 20 .2200 00002002 .2200 000 222222 .2200 00200 0020200 5002 x02-2220200 200002030 .00 .022 20002202 .2200 00022022 2000-2000200 2202 302 00 .00 02000000 0000-0000 .2200 000205002020 .020 .020020020 .00 000 20000 .2200 00002 0200 .2200 2022 000202500 05000002 020 .020 ..0220 0000002 2200002 .020 .00202 202 2200002 .00 0 00202 00002 .2200 2202 .0 000 .00 2002202u0202002 .00000 00 .00 000 20022 .020 ..0220 002022 .020 ..2 < 2 .020 .00200202 20202 02020 .020 .0500 .020 .000x00 .2200 200250220000 .020 .2 2 0 .00 000 220 0200220 00 .00 000 22022000 00 00 20 00 00 20 00 N0 00 00 00 mm 20 00 mm 20 mm 2.0.05000 0 20020222 SELECTED BIBLIOGRAPHY l0. BIBLIOGRAPHY Archibald, T. Ross. "The Return to Straight-Line Depreciation: An Analysis of a Change in Accounting Method." In Empirical Research in Accounting; Selected Studies (1967), a supplement to The JEur- nal of Accounting Research, pp. l64—80. Axelson,Kenneth S. "Needed: A Generally Accepted Method for Measuring Lease Commitments." Financial Executive, July l97l, pp.40-52. Ball, Raymond and Brown, Philip. "An Empirical Evaluation of Accounting Income Numbers." The Journal of Accounting Research, Autumn T968, pp. l59-77. Bartlett, M. S. "Properties of Sufficiency and Statistical Tests." In Proceedings of the Royal Society of London, Series A, Vol. CLX—IlGO), June 1937, PP. 268-82. Beaver, William H., Kettler, Paul, and Scholes, Myron. ”The Association Between Market Determined and Accounting Determined Risk Measures." The Accounting Review, October 1970, pp. 654-82. "The Information Content of Annual Earnings Announce- ments." In Empirical Research in Accounting: Selected Studies (1968), a supplement to the Journal of Accounting_Research. Benston, George J. "Required Disclosure and the Stock Market: An Evaluation of the Securities Exchange Act of T934." The American Economic Review, March 1973, pp. 132-55. Black, Fischer. "Capital Market Equilibrium with Restricted Borrowing.” The Journal of Business, July 1972, pp. 444-55. , Jensen, Michael C.,and Scholes, Myron. "The Capital Asset Pricing Model: Some Empirical Tests," and Miller, Metron H. and Scholes, Myron. "Rates of Return in Relation to Risk: A Re-Examination of Some Recent Findings." Both in Studies in the Theory of Capital Market, edited by Michael C. Jensen. New York: Praeger Publishers, l972. Bolch, Ben N. and Huang, Cliff J. Multivariate Statistical Methods for Business and Economics, Englewood Cliffs, New Jersey: Prentice- Hall, Inc., l974. I38 ll. 12. l3. l4. 15. l6. l7. l8. 19. 20. 21. 22. 23. 24. 25. 26. 139 Box, G. E. P. "A General Distribution Theory for a Class of Likelihood Criteria." Biometrika 36 (1949), pp. 3l7-46. Buff, Jerome H. "Reporting of Leases in Financial Statements-- The Investors Viewpoint.“ Retail Control, June-July l97l, pp. l9-24. Collins, Daniel w. "SEC Product-Line Reporting and Market Effic- iency." The Journal of Financial Economics 2, l975, pp l25-64. Cook, Donald C. "The Case Against Capitalizing Leases." Harvard Business Review, January-February 1963, pp. l45-62. Defliese, Philip L. Should Accountants Capitalize Leases?: The Economic Case Against It at Present. New York: Coopers & Lybrand, l973. Demski, Joel S. Information Analysis. Reading, Massachusetts: Addison-Wesley Publishing Co., l972. Donaldson, Gordon. "New Framework for Corporate Debt Policy." Harvard Business Review, March-April l962, pp. ll7-3l. Elam, Rick. "The Effect of Lease Data on the Predictive Ability of Financial Ratios." The Accounting Review, January 1975, pp. 25-43. Fama, Eugene F. ”Efficient Capital Markets: A Review of Theory and Empirical Work." The Journal of Finance, May l970, pp. 383-4l7. , Fisher, L., Jensen, M., and Roll, R. "The Adjustment of Stock Prices to New Information." International Economic Review, February l969, pp. l-2l. , "Risk, Return, and Equilibrium: Some Clarifying Com- ments." The Journal of Finance, March l968, pp. 29-40. , and Mac Beth, James D. "Risk, Return, and Equilibrium: Empirical Tests." The Journal of Political Economy, May-June 1973, pp. 607-36. , and Miller, Merton H. The Theory of Finance. Hinsdale, Illinois: Dryden Press, Inc., 1972. Feltham, Gerald A. "The Value of Information." The Accounting Review, October l968, pp. 684-96. Fertuck, Leonard. "A Test of Industry Indices Based on SIC Codes." The Journal of Financial and Quantitative Analysis, December 1975, pp. 837-48. Glass, Gene V. and Stanley,.LIL Statistical Methods in Education and Psychology. Englewood Cliffs, New Jersey: Prentice-Hall, Inc., l970. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 4D. 4l. l40 Gonedes, Nicholas J. "Capital Market Equilibrium and Annual Accounting Numbers: Empirical Evidence." The Journal of Account- ing Research, Spring 1974, pp. 26-62. , and Dopuch, Nicholas. "Capital Market Equilibrium, Information Production, and Selecting Accounting Techniques: Theoretical Framework and Review of Empirical Works." In Studies on Financial Accounting Objectives: l974, a supplement to The_ Journal of Accounting Research, pp. 48-]29. , "Risk, Information, and the Effects of Special Items on Capital Market Equilibrium." The Journal of AccountigggRe- search, Autumn l975, pp. 220-56. Hagerman, Robert L. "A Test of Government Regulation of Account- ing Principles." The Accounting Review, Octover l975, pp. 699-709. Hamada, Robert S. "The Effect of the Firm's Capital Structure on the Systematic Risk of Common Stocks." The Journal of Finance, May T972. Pp. 435-52. Harris, Richardtl. A Primer of Multivariate Statistics. New York: Academic Press, l975. Hotelling, H. "The Generalization of Student's Ratio." Annals of Mathematical Statistics, Vol. 2 (l93l), pp. 360-78. Huefner, Ronald J. "A Debt Approach to Lease Accounting." Financial Executive, March 1970, pp. 30-6. Jaffe, Jeffrey F. ”Special Information and Insider Trading." The Journal of Business, 1974, pp. 4lO-28. Kaplan, Robert S. and Roll, R. "Investor Evaluation of Accounting Information: Some Empirical Evidence." The Journal of Business, April 1972: PP. 225-57. King, B. F. "Market and Industry Factors in Stock Price Behavior." The Journal of Business, January 1966, pp. 139-89. Lintner, John. "The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets." Review of Economics and Statistics, February 1965, pp. 13-37. Lord,l=JW."Elementary Models for Measuring Change." In Problems in MeasuringAChange, edited by C. N. Harris. Madison, Wisconsin: The University of Wisconsin Press, 1963. PP. 21-38. Mandelker, Gershon. "Risk and Return: The Case of Merging Firms." Journal of Financial Economics 1, l974,.pp. 303-335. Marshall, Ronald M. "Interpreting the API." The Accounting Review, January l975, pp. 99-lll. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 141 May, Robert G. and Sundem, Gary L. "Cost of Information and Security Prices: Market Association Tests for Accounting Policy Decisions." The Accounting Review, January 1973, pp. 80-94. Miller, Metron H. and Scholes, Myron. "Rates of Return in Rela- tion to Risk: A Re-Examination of Some Recent Findings." In Studies in the Theory of Capital Markets, edited by Michael C. Jensen. New York: Praeger Publishers, 1972, pp. 47-77. Morrison, Donald F. Multivariate Statistical Methods. New York: McGraw-Hill, Inc., 1976. Mossin, Jan. "Equilibrium in a Capital Asset Market." Econo- metrica, October 1966, pp. 768-83. Nelson, A. Tom. "Capitalizing Leases--The Effect on Financial Ratios.” The Journal of Accountancy, July 1963, pp. 49-58. Nurnberg, Hugo. "Leases, Purchase Commitments, and Pensions Revisited." The CPA Journal, May l973, pp. 375-89. Sharpe, William F. "Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risks." The Journal of Finance, September l964, pp. 425-42. Sunder, Shyam. "Relationship Between Accounting Changes and Stock Prices: Problems of Measurement and Some Empirical Evidence." In Empirical Research in Accounting: Selected Studies (1973), a supplement to The Journal of Accounting Research, pp. l-45. Vatter, William J. "Accounting for Leases." The Journal of Accounting Research, Autumn l966, pp. l33-48. Zises, Alvin. "Disclosure of Long-Term Leases.” The Journal of Accountangy, February 1961, pp. 37-47. , "The Pseudo-Leases--Trap and Time Bomb." Financial Executive, August 1973, pp. 20-5.