A Comparison of Sfas 33 Disclosures and Historical Cost Information in Predicting Stock Prices (Inflation, Changing Prices).

Transcription

1 Louisiana State University LSU Digital Commons LSU Historical Dissertations and Theses Graduate School 1984 A Comparison of Sfas 33 Disclosures and Historical Cost Information in Predicting Stock Prices (Inflation, Changing Prices). Heibatollah Sami Louisiana State University and Agricultural & Mechanical College Follow this and additional works at: Recommended Citation Sami, Heibatollah, "A Comparison of Sfas 33 Disclosures and Historical Cost Information in Predicting Stock Prices (Inflation, Changing Prices)." (1984). LSU Historical Dissertations and Theses This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Historical Dissertations and Theses by an authorized administrator of LSU Digital Commons. For more information, please contact gradetd@lsu.edu.

2 INFORMATION TO USERS This reproduction was made from a copy of a docum ent sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this docum ent, the quality of the reproduction is heavily dependent upon the quality o f the m aterial subm itted. The following explanation o f techniques is provided to help clarify markings or notations which may appear on this reproduction. 1.The sign or target for pages apparently lacking from the docum ent photographed is Missing Page(s). If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure com plete continuity. 2. When an image on the film is obliterated with a round black mark, it is an indication o f either blurred copy because o f m ovem ent during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image o f the page can be found in the adjacent frame. If copyrighted materials were deleted, a target note will appear listing the pages in the adjacent fram e. 3. When a map, drawing or chart, etc., is part o f the material being photographed, a definite m ethod o f sectioning the material has been followed. It is custom ary to begin filming at the upper left hand corner o f a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again beginning below the first row and continuing on until com plete. 4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Custom er Services D epartm ent. 5. Some pages in any docum ent may have indistinct print. In all cases the best available copy has been film ed. University Micrcxilms International 300 N. Zeeb Road Ann Arbor, Ml 48106

3

4 Sami, Heibatollah A COMPARISON OF SFAS 33 DISCLOSURES AND HISTORICAL COST INFORMATION IN PREDICTING STOCK PRICES The Louisiana State University and Agricultural and Mechanical Col. Ph.D University Microfilms International 300 N. Zeeb Road, Ann Arbor, Ml Copyright 1985 by Sami, Heibatollah All Rights Reserved

5

6 A COMPARISON OF SFAS 33 DISCLOSURES AND HISTORICAL COST INFORMATION IN PREDICTING STOCK PRICES A Dissertation Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Accounting by Heibatollah Sami B.S., Iranian In stitute of Advanced Accounting, 1973 M.S., Central Michigan University, 1981 August 1984

7 1985 HEIBATOLLAH SAMI All Rights Reserved

8 ACKNOWLEDGEMENTS I wish to express my appreciation to members of my d iss e rta tio n committee, Professors Jerry E. Trapnell who served as the Chairman of the Committee, Steven M. Buco, Anthony P. Curatola, Bart P. Hartman, Arnold Saxton, and Richard A. White. I am grateful to all my Committee members for their valuable suggestions and guidance in writing this manuscript. Their encouragement and support inspired me throughout this study. I would like to thank Suzanne Gatipon for her patience and expertise in typing this manuscript. I would also like to thank my mother Shamsehdoulat, and my father-in-law Mojtaba Mehdian for their support, for their belief in my judgments and for their everlasting understanding. Special recognition is extended to my wife Manidjeh and my daughter Sepideh, for th e ir understanding and patience during the course of this study. I would like to dedicate this dissertation to them. i i

9 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS... LIST OF TABLES... ii v LIST OF FIGURES.... vii ABSTRACT... v iii Chapter 1. INTRODUCTION... 1 Historical Background... 1 Theoretical Controversies... 6 Statement of the Problem Need for the Study LITERATURE REVIEW Replacement Cost Studies Constant Dollar S tudies SFAS 33 Studies Summary RESEARCH METHODOLOGY Steps of the Study Time Periods Sample Selection Procedures Variables and Their Computation Data Layout and Model Estimation Computation of Prediction Errors Analysis of the Error Terms i i i

10 TABLE OF CONTENTS (con t) Chapter Page 4. ANALYSIS OF THE DATA Selected Sample and Industry Grouping Correlation Analysis Prediction Models and Their Predictive A bility Descriptive S ta tistic s SUMMARY AND CONCLUSIONS Summary Conclusions Scope and Lim itations Suggestions for Future Research REFERENCES APPENDIX A. List of the Companies Included in the Samples VITA iv

11 LIST OF TABLES Table Page 2.1 Summary of the Studies' Results Reconciliation of Sample Size and Statement 33 Data Bank Industry Grouping in the Sample Selected Correlations Between Independent Variables for Period by Industry Grouping Selected Correlations Between Independent Variables for Period by Industry Grouping Correlation Among Cumulative Weekly Security Returns Over the Eleven Week Test Period (Y) and Independent Variables (X's) of the Period by Industry Grouping Correlation Among Cumulative Weekly Security Returns Over the Eleven Week Test Period (Y) and Independent Variables (X's) of the Period by Industry Grouping Least Square Estimates of Parameters for Prediction Models of Period by Industry Grouping Least Square Estimates of Parameters for Prediction Models of Period by Industry Grouping AN0VA Summary with ARE as Dependent Variable ANOVA Summary with SRE as Dependent Variable Means and Standard Deviations of Prediction Errors for Hold-Out Samples by Model v

12 LIST OF TABLES (con't) Table Page 4.12 Means and Standard Deviations of Prediction Errors for Hold-Out Samples by In d u stry Means and Standard Deviations of Prediction Errors for Hold-Out Samples by Industry * Model v i

13 LIST OF FIGURES Figure Page 3.1 The Data Break-Down Data Layout Layout of ARE v ii

14 ABSTRACT The primary objective of this study was to determine the a b ility of d iffe re n t subsets of information under the Statement of Financial Accounting Standards No. 33 (SFAS 33), historical cost/constant dollar (HC/CD), current cost (CC), and current cost constant dollar (CC/CD), to predict the changes in stock prices as compared to that of HC information. The study was carried out using the data available for the years The test period was the eleven weeks surrounding the issuance dates of the firms' annual reports which was approximated by using the date of annual earnings announcement that appeared in the public media such as the Wall S tree t Journal. Two samples for two time periods ( and ) were selected from the companies lis te d on the nonfinancial file of the FASB Statement 33 Data Bank. The s e c u rity 's cumulative weekly return adjusted for dividends and stock sp lits was used as the dependent variable and percentage change in earnings per share and the related components were used as the independent variables. Each sample was assigned to different industry groups based upon the firs t two digits of the companies v i i i

15 SIC code. Each industry group was divided between two subsamples. Subsample (1) was used to estimate the prediction models through le a s t square regression. Subsample (2) (hold-out sample), which included ten observations in every case, was used to test and compare the predictive ability of the models. Prediction erro rs of the four models, HC, HC/CD, CC, and CC/CD models, were analyzed through a spit-plot design ANOVA. Duncan's multiple range test was carried out as a post ANOVA test wherever appropriate. The results indicated that, in predicting the changes in stock prices, HC information competes with three subsets of in fla tio n adjusted data available under SFAS 33 and none of these three can significantly and/or consistently outperform the HC data. Furthermore, the industry was no.t a sig n ific a n t factor in predicting the changes in stock prices. ix

16 CHAPTER 1 INTRODUCTION The growing in fla tio n in the past decade has been perceived to impair the usefulness of h isto ric a l cost information to decision makers such as management, investors, and creditors. This has prompted the authoritative bodies within d iffe re n t countries including the United States to look for a way of disclosing the effect of inflation and changing prices on the accounting data in financial reporting. After a long period of deliberation the Financial Accounting Standards Board (FASB) issued the Statement of Financial Accounting Standards No. 33 (SFAS33) [FASB, 1979] which is supposed to deal with the problem of inflation and changing prices in an experimental manner. As indicated in the Statement, the FASB w ill review the effect and usefulness of d iffe re n t disclosure requirements of the Statement and make necessary revisions. The purpose of this research is to provide some evidence with regard to the usefulness of the information content of SFAS 33 d isclosures as compared to that of h isto rical cost information. Historical Background In addition to going concern and realization, stable 1

17 monetary unit measure "was one of the three basic postulates of accounting" [Rosenfield, 1981, p The Study Group on Business Income [1952, p. 20] recognized the fact th at the postulate of stable monetary unit is the oldest of the above three postulates. However, during the 1920s, companies started to w rite up th eir assets "as sound values were considered to be higher than recent costs" [Hendriksen, 1977, p. 61]. Before the end of 1920s, the 1929 stock market crash followed by the "Great Depression" caused the companies to start writing down their assets and stop the write-ups quickly [Rosenfield, 1981, p. 95]. The accountants of that period, who were blamed in part for the crash because of th e ir involvement in the w rite ups, decided not to be exposed to such a charge again. "They insisted that assets be carried no higher than historical cost" [Rosenfield, 1981, p. 95]. In 1936 the American Accounting Association (AAA) endorsed the accountants' position in a "fundamental axiom" of accounting which says "accounting is... not e s se n tia lly a process of valuation, but the allocation of historical costs and revenues to the current and succeeding fiscal periods" [1936]. The American Institute of Accountants (AIA) - a predecessor of the American Institute of Certified Public Acountants (AICPA) - emphasized the same thought in In one of the six rules adopted by the AIA

18 3 membership in that year as indicated in the Report of Special Committee on Development of Accounting Principles: "profit is deemed to be realized when a sale in the ordinary course of business is effected, unless the circumstances are such that the co llectio n of the sale price is not reasonably assured" [1934, p. 11]. During and a fte r the second World War, the high rate of inflation caused pressure on the accounting profession to recognize the effe c ts of inflation on depreciation. However, the AIA outlawed the price-level depreciation in Accounting Research Bulletin no. 33 in 1947 and the Committee on Accounting Procedure resisted further pressure to react to the accounting problems of inflation by stating: Should in fla tio n proceed so far that original dollar costs lose their practical significance, it might become necessary to restate all assets in terms of the depreciated currency, as has been done in some countries. But it does not seem to the Committee that such action should be recommended now.... [1948] The AAA also noted the need for considering the effects of fluctuations in the purchasing power of money in its revised Accounting Concepts and Standards Underlying Corporate Financial Statements [1948]. The debate on the e ffe c ts of flu c tu a tio n s in the purchasing power of money years. and price level depreciation lasted more than ten The AICPA (AIA prior to this time [1958]) announced the results of its 1957 opinion survey of inflation adjusted depreciation. These results indicated the desirability

19 4 of making the stockholders aware of the effects of price level changes in some manner. The problem of in fla tio n or d eflatio n (changing p ric es) and i t s im plications for fin an cial reporting were recognized by the Accounting Principles Board (APB) as la te as In th is year the APB concluded that "the assumption in accounting that fluctuations in the value of the dollar may be ignored is unrealistic...." [Tierney, 1963, p. 56]. Sprouse and Moonitz agreed with th is statement in the Accounting Research Study (ARS) No. 3 [1962, pp ]. Consequently, the APB started a program which authorized research by the staff of the Accounting Research Division of the AICPA to investigate the ways to deal with th is issue. This investigation resulted in the publication of ARS No. 6, "Reporting the Financial E ffects of Price-Level Changes," [1963] which recommended presentation of supplementary financial statements that are adjusted for the changes in the general level of prices. However, an alternative to this approach that may be argued is restatem ent for the changes in the specific level of prices but ARS No. 6 mostly addressed the changes in the general le\ *h of prices. Later in 1969 the APB issued Statement No. 3 [APB, 1969] which also recommended price-level adjusted information in supplementary form. However since the rate of inflation was not that high actual disclosures per APB Statement

20 5 No. 3 were not widespread. In 1973 and 1974 the inflation rate, based on the Consumer Price Index, was 8.8 and 12.2 percent, respectively. The high rate of inflation renewed the debate regarding the usefulness of historical cost accounting in an inflationary economy. This renewed controversy prompted the FASB and the S ecu rities and Exchange Commission (SEC) to start investigating different ways of dealing with the issue which would re su lt in protecting the public interest. In February 1974, the FASB responded by issuing a discussion memorandum. In April 1974, the FASB held public hearings on the topic and several months later, in December 1974, issued an exposure draft which favored general-price-level (GPL) adjusted statements as supplemental information to the h is to ric a l cost statem ents. While the FASB was in v itin g comments on this exposure draft, the SEC issued Accounting Series Release (ASR) No. 190 in March 1976, which required replacement cost disclosures to be reported in the Form 10-K for fiscal years ending on or after December 25, This disclosure requirement indicated the SEC s belief that the public interest had not been met by e ith er h is to ric a l cost statem ents or the FASB's Exposure Draft [Bloom and Dehessay, 1981, p. 48]. Consequently, it imposed its own standards to protect the public interest. As the result of this action by the SEC, in November 1975, the FASB withdrew its 1974

21 6 Exposure Draft, pending further study. In December 1978, another Exposure Draft, after a few years, the FASB issued "Financial Reporting and Changing Prices." Later, the FASB [Griffin, 1979] held a conference in New York City to call attention to the need for more adequate disclosures which reflect the effects of changing prices and discuss its exposure draft. Most participants acknowledged th is need; however, they asked the FASB not to require complicated disclosures, such as those indicated by the Exposure Draft, which would not be understandable to the users. Finally, in September 1979, the FASB issued SFAS 33. Based on this statement, companies which have more than $125 m illion in inventories and gross property, plant, and equipment or more than $1 b illio n to ta l assets a fte r depreciation are required to disclose supplemental information regarding the general p rice-lev el adjustments, current cost adjustments, and current cost-constant d o llar adjustments. In October 1979, the SEC repealed its ASR 190 disclosure requirements in favor of the SFAS 33. The SFAS 33, to some extent, is a mixture of the 1974 Exposure Draft and the SEC's ASR 190 disclosure requirements. Theoretical Controversies In general, to deal with the effect of in flatio n and changing prices three different approaches have been proposed in the lite ra tu re. These are historical cost/

22 7 constant d o llar, current cost and current cost/constant dollar approaches. Sweeney's book "S tab ilized Accounting" [1964] is one of the early sources containing the concepts and p rocedures for gen eral purchasing power accounting. A part of his argument [p. 4] in favor of the general purchasing power adjustments (p rice-lev el adjustment) is that although it appears that men work for money rather than consumption goods: But even then money is desired, by the miser and the lover of power as well as by the average individual, only because it represents ability to buy d e s ir a b le commodities and services. But, i t may be argued, the things th a t men buy with money are not always consumption goods. Much of the time they are 'production goods,' such as factory machinery and buildings. When men buy pro d u ctio n goods, however, they do so only because in most cases they hope thereby to obtain eventually more of the consumption goods th at they want than they could obtain by buying them in the f ir s t place. In consequence of the foregoing, accounting data should ideally be measured with reference to the progress made in obtaining either more consumption goods or greater power over them. These consumption goods are those represented in the cost of living. William Paton, in an introductory section to the 1964 reissue of Sweeney's book stated that "stabilized accounting... s t i l l clearly points the way." Robert S terlin g [1970, pp. 337, 340] advanced the argument a step further and stated: We view 'e n te rp r is e ' as nothing more than a convenient name for the vehicle that humans

23 8 establish to maximize... their u tility. In view of the m otivation of the enterprise, it should be obvious that we think the Consumer Price Index is most appropriate. It is the clo sest su b stitu te for a u t i l i t y measurement that is currently available. Another argument has been th a t "the accountant's yardstick" is constantly changing which makes i t hard if not impossible to reach any conclusion with respect to the re la tiv e economic position of an enterprise from direct comparison of the conventional accounting statements [Paton, 1920]. Paton [1920, p. 31 states that "accountants deal with an unstable, untrustworthy index; and, accordingly, comparisons of unadjusted accounting statements prepared at different periods are always more or less unsatisfactory. " In a similar argument, Moonitz [1961, pp ] refers to the unstable measurement unit as a problem of inconsistency in the financial periods of time. statements of an entity over different He suggests the preparation of financial statements based on a stable unit of measurement to make them more in line with the consistency concept. Other benefits that are claimed to attract attention to the historical cost/constant dollar information include th e ir o b jec tiv ity and a u d ita b ility [Davidson and Weil, 1975]. F inally, the following benefits from historical co st/constant d o llar - "current d o llar" - are claimed by the Committee on Concepts and Standards Underlying Corporate Financial Statements of the AAA [1951, p. 470]:

24 9 1. The appraisal of managerial effectiveness in terms of the preservation of the current dollar equivalent of the capital invested in the b u sin ess and not merely in its in itia l dollar amount. 2. The an aly sis of earning power in terms of the current economic backdrop. 3. The d eterm in atio n and ju s tif ic a tio n of sound wage p o lic ie s; negotiations with labor unions. 4. The determination by government of long-range p o lic ie s w ith re sp e c t to "control" of the economy through monetary policy, price regulation, lim itation of profits, taxation, etc. 5. The creation of an informed public opinion with respect to p ro fits, prices, wages, e t c., and the e f f e c t of in fla tio n (or d eflatio n ) upon fin an cial relatio n sh ip s generally. 6. The determ ination of managerial policies with respect to pricing, credit, dividends, expansion, and the like. All aforementioned benefits of historical cost/constant dollar information have attracted considerable attention. This is evidenced by the work of several authors trying to provide short cut procedures to convert the historical cost information into reasonable approximations of historical co st/constant dollar data. These attempts include the work of Samuelson [1972], Petersen [1973], Davidson and Weil [19751 and Parker [1977]. From a theoretical point of view it has been argued th a t the h is to ric a l cost/constant dollar information neither is interpretable (does not have real world meaning) nor is relevant for decision making [S terlin g, 1975].

25 Stikney and Green [1974] argued that providing financial statem ents based on constant dollar would not make a 10 m aterial difference and the ranking of companies based on rate of return would not change. In opposition to p rice-lev el adjusted inform ation, Revsine and Weygandt [1974] argued th a t financial statement users would like to receive inform ation in terms of money rather than general purchasing power. A second approach dealing with inflation and changing prices is current cost accounting. One way to approximate the current input value is the use of replacement cost. It has been argued th at disaggregation of net income into holding gain and replacement cost income would result in measures of return generated by decisions relating to the holding and operating functions respectively [Edwards and Bell, ]. Burton [ 1975, p. 69] argued that the effective compliance with the matching concept requires the u tiliz a tio n of current cost and concluded: N e v e rth e le ss, in an in fla tio n a ry time when revenues are largely based on current market phenomena, costs must be sim ilarly based if the matching process is to produce a meaningful measure of results. This seems to argue strongly for a measurement system using current economic costs. Under such an approach, expenses would be based on the current cost of replacement of the particular assets sold or used. In this way the matching p ro cess would show a long-run average cash flow figure based on current costs at the time tr a n s a c tio n s occur. This may be described

26 s ' 11 as the earning power of the firm. Two other arguments are presented in favor of current cost accounting [Rosenfield, 1981, p. 116]. The firs t one is th at "the cost avoided by owning an asset is a satisfactory amount at which to present it in the balance sh eet." The second argument is that "the sacrifice that would be incurred by an en terp rise on being deprived of an asset... is a sa tisfa c to ry amount at which to present the cost of sellin g or using the asset in the income statement and the sacrifice is the current buying price of similar assets." On the other hand Sterling [1975] argued that under current value, the financial statement numbers are stated in terms of number of d o llars (NOD) and interpretable as the measures of NOD, but only the asset figures can be interpreted in terms of command over goods. Furthermore, there is a problem with the measurement unit which differs from one period to another resulting in the information not being adjusted for the changes in the general purchasing power of the monetary unit. Consequently the financial statements of the same company, prepared using the current cost, are not comparable over different periods of time. These d eficien cies of current cost have brought about the suggestion for a third approach called "current cost/ constant dollar". The c u rre n t co st/co n stan t d o llar approach would

27 achieve the dichotomization objective advocated by Edwards and Bell and, also, would result in financial statements 12 of the same company being stated in terms of the unit of measurement with the same purchasing power. This would also enhance the com parability of the financial statements of the same company over time. Current value/ constant dollar inform ation would provide the present and po ten tial investors, creditors, and other users of financial reports with the kind of information that has real world in terp retatio n, does not contain measurement unit error, over goods. over goods, and is interpretable as a measure of command Because of its interpretability as command S terlin g [1975] believes it to be the most re le v a n t in fo rm atio n for any decision-making model. Trapnell and Sami [Forthcoming] argue that current cost/ constant dollar information have representational faithfulness and the outcome of the FASB experiment with regard to the SFAS 33 is expected to support the reporting of this information. Beaver and Landsman [1983, ch. 2] argue that historical cost computations such as depreciation reflect the anticipated in f l a t i o n ; consequently, constant dollar and current cost accounting would result in double-counting at least a portion of in flatio n (the anticipated portion). This double-counting of in fla tio n would cause the rates of return reported under constant dollar and current cost

28 13 accounting not to be interpretable. They further argue th at the usefulness of SFAS 33 data depends upon the ability to provide information about the effects of unanticipated inflation and the magnitude of resulted measurement error from double-counting the inflation. The information provided under SFAS 33 is theoretically appealing and, based upon the analysis provided in this sectio n, should provide information not contained in the h is to ric a l cost data. However, the usefulness of this information in a real world situation is an empirical question which provides the basis for the problem that this study is designed to investiagate. Statement of the Problem This study concentrated on the ability of different sets of information under SFAS 33 (historical cost/constant do llar, current cost, and current cost/constant dollar) to predict the changes in stock prices as compared to that of historical cost information. Based on the semi-strong form of market efficien cy, stock prices fu lly reflect all publicly available information. Consequently, financial statement data, including the SFAS 33 information, should in theory be reflected in the stock prices in an unbiased manner. In addition the ability of any subset of information available in the financial statements to predict changes in stock prices with a lower error would provide evidence of its usefulness to the users as compared to other subsets.

29 14 Based on the Statement of Financial Accounting Concepts No. 2 (SFAC2), themost important qualitative characteristics for decision usefulness are relevance and r e lia b ility [FASB, 1980, Paragraphs 46, 47, and 591. As indicated in th is statement, the inform ation is relevant if it has predictive and/or feedback value, and if it is provided on a timely basis. The ability of any information set to predict the changes in the stock prices with lower error would indicate th at eith er the information set has caused the change in the stock prices or is highly correlated with some other event(s) that has caused the change. Consequently, th at information set would have more predictive value and as a result more decision usefulness. Need for the Study The FASB will evaluate the results of its experimentation with SFAS 33 to decide whether to continue with th is statement, revise it, or elim inate i t. Review of the lite r a tu r e revealed insufficient evidence regarding the usefulness of required information under SFAS 33. Most of the research re la tiv e to changing prices is related to the information content of replacement cost disclosures under ASR No. 190 and the usefulness of price-level adjusted data using some estimation technique. However, the results of these studies are not quite applicable to the information av ailab le under SFAS 33 for various reasons. F irst,

30 15 the available information set under SFAS 33 is more comprehensive than the information set under ASR 190 requirements. Second, financial statement users may have a better understanding of the inform ation set available under SFAS 33 as the re su lt of th e ir experience with the ASR 190 data. Third, studies th at dealt with the price-level adjusted data estimated the information using some short cut estim ation approaches which might have resulted in d iffe re n t numbers than those now available under SFAS 33. Several studies investigated the effects of supplementary information under SFAS 33 from different points of view. The overall re s u lts of these studies are mixed (some supported the SFAS 33 and some did not) and inconclusive. 1 The only two studies th at are most directly related to th is research are those by Beaver and Landsman [1983] and Soroosh Joo [1982]. The re su lts of the study by Beaver and Landsman are open to debate because of the following problems: 1. Security return (the dependent variable) is calculated over a one year period - extending from January 1 to December 31 of each year - for a ll the companies in their sample which had December 31 year-end. This would provide a s e rio u s bias in favor of h is to ric a l cost inform ation and against SFAS 33 data. This bias e x ists because usually by December 31, at l e a s t th re e q u arters of h isto ric a l cost information have been reported and there might be some other news releases by the firms regarding th e ir annual results. On the other hand, SFAS 33 inform ation would not be known until after the issuance of annual reports which always

31 takes place at le a st a month or so after the f is c a l year-end (December 31). An unbiased inform ation content study would have included the time period surrounding the date of the annual report. 2. The two most important v ariab les under SFAS 33 in fo rm atio n - the post holding gain net income (POST) and the post holding gain net income adjusted for the effe c ts of inflation (POSTP) - are not calculated properly. Different computation procedures used to calculate these two earnings v ariables could confound th e ir correlation with the returns. 3. In computation of different variables regarding the SFAS 33 information, Beaver and Landsman add (su b tract) items such as purchasing power gain ( lo s s e s ) and holding gain (losses) to the "income from continuing operations. This is only appropriate if the degree of association of th ese item s and "income from continuing o p eratio n s with the returns is the same and no i n t e r a c t i o n between these components of earnings ex ist which might not have been the case. The re su lts of the other study by Soroosh Joo is also, open to the following criticisms: 1. In the computation of resid u als, which were analyzed to d e te c t information e ffe c t, the researcher took out the effects of economy-wide ev en ts. According to Beaver [1981, p. 35] these events include inflation as well. 2. The s t a t i s t i c a l analyses are carried out for each of the eleven weeks of the te s t period separately while i t was more appropriate not to separate the weeks and determine the effect over the to ta l period. The market may have taken more than a week to respond to the new information. The above analysis indicates that the existing evidence s t i l l is not s u ffic ie n t to determine whether the SFAS 33 d isclosures provide information that is more useful 16 in predicting changes in the stock prices. Furthermore,

32 i t is not clear which set of supplementary information - historical cost/constant dollar, current cost, or current cost/constant dollar - is more useful to the users, specifically, in predicting the stock price changes. Finally, no other study investig ated the effect of industry on the usefulness of accounting information. This study has trie d to extend the prior research to investigate the above issues, to provide needed evidence to the accounting profession, and to correct the above mentioned deficiencies.

33 CHAPTER 2 LITERATURE REVIEW Review of the lite ra tu re has revealed several studies in the area of SFAS 33 and other related areas. This review mostly concentrated on the studies that were carried out in the past few years. The results of this review are presented in this chapter in four sections. First, those studies that concentrated on the effects of replacement cost disclosures under ASR 190 are presented. Second, the studies th at investig ated the inform ation content and e ffe c ts of constant dollar information through some approxim ation techniques are reviewed. Third, those studies th at attempted to determine d iffe re n t effects of supplementary information under SFAS 33 are discussed. Finally, a summary of the results of the studies in each of the above sections is presented and analyzed. Replacement Cost Studies The study by Abdel-Khalik and McKeown [ 1978] concentrated on testing the following three Broad Hypotheses (BH): BH(1) The s e c u ritie s markets have already impounded replacement cost information in security prices prior to its disclosure. BH(2) The d isclosure of replacement cost in f o r m a t i o n is expected to have 1 8

34 19 information content that would Induce security price revisions. BH(3) Replacem ent cost in fo rm a tio n has no information content. They selected a sample of firms which met the following c r i te r i a : (1) fis c a l year-end of December 31, 1976, (2) estim ates of th e ir replacement cost-based income (ERI) and h is to ric a l cost-based net income (ENI) were disclosed by Value Line, (3) th e ir stock prices were available on the CRSP tapes, and (4) th e ir fin an cial statem ents were av ailab le on the COMPUSTAT tape. One fu rth er refinement of the sample was th at only those firms with stationary (this was determined through analysis of covariance) beta were retained. In the analysis, they used two market indices, an equally weighted index which resulted in a sample of 203 firms (149 non-utilities; and 54 u tilitie s and airlines), and a value-weighted index which resulted in a sample of 211 firms (153 n o n - u tilitie s, and 58 u t i l i t ie s and a ir lin e s ). The firms in each group within each sample were assigned to two portfolios based on the ratio (high or low) of th e ir holding gain to their historical cost net income. This resu lted in four p o rtfo lio s w ithin each sample. A form of capital asset pricing model was used to determine the beta, financial risk, and unlevered beta and student ts ta tis tic s were used to test the quality of the means of each of the two portfolios within each

35 20 group. Furthermore they conducted a 2 x 2 analysis of variance (ANOVA) of cumulative average residuals as of December 1975 and The results of their analysis supported the f ir s t broad hypothesis but not broad hypotheses 2 and 3. Another study th at focused on the effect of the replacement cost disclosures as mandated by the Securities and Exchange Commission's (SEC) Accounting Series Release (ASR) 190 on the stock market was by Gheyara and Boatsman [1980]. They obtained a treatment and a control sample of firms from the COMPUSTAT tape which filed Form 10-K. C rite ria used to choose the treatment firms were: (1) su b je c tiv ity to the ASR 190 reporting requirements, (2) a v a ila b ility of common stock returns on the CRSP daily returns f ile, (3) having fis c a l year end of December 31, 1976, (4) responding to the request for a copy of th e ir 1976 Form 10-K and the date of i t s filin g with SEC, and (5) determining the holding gain component of th e ir replacement cost income was possible. The control sample consisted of those firms which were not subject to the ASR 190 reporting requirements and satisfied criteria 2 through 4. The samples included 106 treatment firms and 83 control firms that met the above criteria. Residuals and unadjusted returns were calculated, for each of the 50 day period surrounding the Form 10-K r e le a s e date, using an estim ate of the market model

36 21 = a + 3j_ rmt + Uit). Four testing procedures were applied. They were: (1) the residual variance inspection, (2) analysis of crosssectional return data, (3) analysis of paired time serie s return data, and (4) analysis of performance of high and low holding gain p o rtfo lio s, were applied. In a ll of these testing procedures, the null hypothesis of no significant effect of replacement cost disclosures could not be rejected. Consequently, they concluded th a t "the replacement cost disclosures mandated by Accounting Series Release 190 simply did not introduce information during the 50 trading days surrounding disclosure." Beaver, et. a l. [1980] examined the association between changes in the security prices and various aspects of disclosures under the SEC's Accounting Series Release (ASR) No They used a sample of 553 firms th at was drawn from a group of 627 firms on which replacement cost data were obtained from December 31, 1976 to November 30, The replacement cost data were obtained from the 10-K report filed with the SEC, historical cost data were obtained from the Annual Industrial COMPUSTAT tape, and daily security return data were obtained from the CRSP file. The sampled firms were partitioned into eight groups (portfolios) based upon replacement cost disclosure items that would reduce the within group variation (diss im ila rity ) and increase the between group v ariatio n

37 22 (dissim ilarity). The beta for each firm was calculated, using the CRSP monthly return f ile, through a time series linear regression. The firms within each of the eight portfolios were ranked according to beta and divided into two, high and 'low1 betas, subportfolios. The weight of each subp o rtfo lio was determined such that the beta was equal to one for each of the eight portfolios. The daily returns were determined for each of the eight p o rtfo lio s for th is th irty day period around three dates of interest: (1) ASR 190 proposal date, (2) ASR 190 adoption date, and (3) the latest possible date to file the 10-K reports with the SEC for firms with December 31 fiscal year-end. Also the returns were determined for the overall period of June 1, 1975 through April 22, 1977 which covered the above three dates. Most extreme portfolios were paired into four pairs and the null hypothesis of no difference, for the three dates and the overall period, was tested using the Hotelling T^ s t a t i s t i c the results of which were not significant. Further analysis of the returns by comparing them to nonreporting firms, also, did not reveal any significant difference. Consequently the researchers concluded no s ig n if ic a n t association between security returns and ASR 190 disclosures. In one study Ro [1980] investigated the effect of

38 the information content of replacement cost (RC) disclosure 23 under ASR 190 on the security returns. Two samples of firms, a control sample and one treatment sample, were selected. Firms were included in the treatment sample if they (1) reported RC data for 1976, (2) were listed on NYSE for the period under study, (3) had fiscal year-end of December 31, (4) were listed on the CRSP daily return file s, and (5) responded to the request for a copy of th e ir 1976 Form 10-K. Application of these c r ite r ia r e s u lte d in an i n i t i a l l i s t of 227 treatm ent firms. A l i s t of p o tential control firms was developed from the Value Line Data Base tape for 1976 by applying the $100 m illion m ate ria lity standard plus above c rite ria (1) through (4). Subsequently, control firms were paired with treatment firms based on (a) the sign of the change in earnings per share from 1975 to 1976, (b) the systematic risk (beta), (c) the week of 1976 Form 10-K report release, and (d) the f ir m 's industry membership if possible. This procedure resulted in 78 pairs which constituted the final sample. The Value Line RC forecasts were used as proxy for the m arket's expectations of ASR 190 RC data and the 78 pairs of firms were assigned to two groups (portfolios) based on the sign of the differential difference of actual RC and historical cost earnings and the Value Line forecast of the same numbers for treatment firms which resulted in 23 good-news and 55 bad-news pairs.

39 24 Average weekly portfolio return difference for good-news and bad-news p o rtfo lio s were computed for each of the 26 weeks, encompassing the Value Line forecasted RC and actual 10-K RC disclosures for 1976, by taking the average of individual component return differences in each portfolio; fin a lly, a difference between the two portfolio return differences was computed for each of the 26 weeks. These fin al differences were analyzed using a t - t e s t. The difference in returns was significant at the.10 level. However, the researcher claimed th at th is significant difference was due en tirely to the negative abnormal returns of the bad-news p o rtfo lio s and concluded that there is at best only very weak evidence of information content (effect) of ASR 190 RC disclosures. In another study Ro [1981] investigated the effect of replacement cost (RC) disclosure rules set forth in Accounting Series Release (ASR) 190 on the weekly transaction volumes of common equity securities. Potential treatment firms (those subject to RC disclosure) included 235 firms from 750 firms that responded positively to the request for a copy of th e ir Form 10-K. A firm was included in the fin al treatment sample if (1) it disclosed RC data for 1976, (2) i t was lis te d on NYSE for the 174-week period covered, (3) i t did not have any suspension of its share trading during the period, (4) its fiscal year-end was December 31, and (5) met the matching criteria stated

40 25 below. These These crite ria resulted in a sample of 73 firms. treatment firms were matched with control firms, th at met a ll the above c r ite r ia except (1), based on the following crite ria : (a) the beta (risk) of return, (b) the week of Form 10-K filing with the SEC, (c) the sign of the change in historical cost earnings from 1975 to 1976, and (4) industry membership if possible. The return beta and the volume beta were estimated via a market model through running a regression and using weekly returns and volumes for the 86-week period prior to the in itia l proposal of ASR 190. Based on the magnitude of each of betas and five other ratios treatment firms were assigned to high and low portfolios, 14 in total. Two of ratios attempted to measure the effects of holding gains (realized and unrealized), the other two measured the e ffe c ts of holding gain on net income and the last one measured the e ffec t of changing prices of assets on the firm's capital structure. The difference between each treatment firm and its matched control firm with regard to the developed volume variable was calculated for each of the nine event weeks (from proposal week to 10-K RC disclosure week). These differences were analyzed using a matched-pair t-te s t which did not reveal significant differences among treatment and control firms in most of the 14 portfolios in most of the event weeks. Consequently, the researcher concluded

41 26 th at, if the revision of common stock portfolios is a criterio n of the information content of replacement cost disclosures, then, the re su lts could have implied that the data did not contain information. Another study that investigated the impact of replacement cost disclosure of Accounting Series Release (ASR) 190, on security prices is the study by Lustgarten [1982]. A sample of companies, which were lis te d on the NYSE, was selected. Companies were included in the sample i f (1) they disclosed replacement cost information, (2) th e ir replacement cost inform ation were available on COMPUSTAT tape, (3) th e ir security price information, for 54 prior months and 4 subsequent months to fiscal year end, were available on CRSP tapes, and (4) all necessary information were available in COMPUSTAT files to calculate the variables described below. Three independent v ariab les included in the study were (1) the forecast error of the earnings for the firs t disclosure year which was assumed to represent the information content of the historical cost earnings number, (2) the difference between accumulated depreciation on a replacement cost basis and historical cost basis for the disclosure year as the signal of information from replacement cost, and (3) the logarithm of the sales amount for the disclosure year as the size variable. Earnings, total assets, outstanding shares, and market value of the firms were used

42 27 as d eflato rs to scale the independent variables. The dependent variable was the 10 month cumulative residual (6 months before and 4 months after the end of the fiscal year of disclosure) which was calculated using the capital asset pricing model that was estimated from 48 monthly (beginning 54 months prior to the end of the fiscal year of disclosure) data. The data were analyzed through ordinary least squares (OLS) and weighted least squares (WLS) regression. The results of the OLS analysis indicated that the coefficient of the replacement cost v ariab le was significant using the total assets and outstanding shares deflators. Also, the re s u lts of WLS analysis confirmed the significance of the event period regressions. However, further analysis of the data by p a rtitio n in g the firms into d iffe re n t p o rtfo lio s indicated th a t nearly a ll of the responses were observed in months one through eight which made the author unwilling to reach any strong conclusion. Diamond [1978] conducted a field experiment to investig a te the e ffe c ts of replacement cost disclosures on term loan decisions by loan officers, the extent to which replacement cost disclosures are utilized by loan officers in making th e ir decisions, on th e ir judgment models. and the effect of this data The experiment was carried out among 100 loan o ffic e rs from six C alifornia banks who were asked a series of 40 loan decisions based upon

43 28 five financial statement ratios. These five ratios were: (1) current ratio, (2) total debt to net worth, (3) times in te re s t earned, (4) rate of return on to ta l assets, and (5) rate of change of p ro fits over the la s t five years. The subjects were divided into two treatment groups. One treatm ent group received a set of ratios based on historical cost, while the other group received an additional se t of r a t i o s based upon the replacement cost data. Each loan o ffic e r was asked to indicate the in te re s t rate at which the loan would be granted which was used as the dependent variable. The data were analyzed through a repeated measures ANOVA to te s t the decision effects of replacement cost disclosures the results of which indicated that the replacement cost disclosures had no impact on the loan officers decisions. Furthermore, data for each loan officer was analyzed through a multiple linear regression. The results of th is analysis indicated th at the debt to net worth and the rate of return ratios had the highest explanatory power in the regression equations. In a different study Hayes [1978] tried to determine whether the disclosure of replacement cost data under ASR 190 had any systematic effect on the estimated risk parameters of the reporting firms. The underlying proposition of th is study was that if the replacement cost data do

44 have system atic effect on the estimated risk parameters of the sample firms, then the data most likely had information content for market p a rtic ip a n ts. The sample included 45 replacement cost reporting firms for which a time series of forecast erro rs of systematic risk estimates is determined by analyzing a time series of two weeks systematic risk estimates of each firm through a Box-Jenkins analysis. Averages of the time series of forecast errors are then analyzed using a sign test to determine if the signs of the average erro rs have a nonrandom pattern. A nonrandom p attern indicates the existence of a shift in the estimated risk parameters of the sample firms over the period under study. The data re la tin g to two d iffe re n t time periods, time period immediately preceding and following the disclosure date and the time period after the disclosure date, were analyzed. For the f ir s t time period the analysis included the data from all of the 45 firms and a second analysis included only decision model data from those firms th a t based upon a th e ir replacement cost disclosures might have led the investors to revise their estimate of the risk in ess of the firm. For the second time period, the analysis included only the data from the firms that based upon the decision model an upward revision in their riskiness might have occurred. None of the analysis indicated any shift in the estimated risk parameters of the firms.