». .. ... 1.. . . ........n \ u I I2? . .2. «1... I... ...I.v xiv... ‘ ‘nlfur Q- . ,. ¢ . . :.Jt~".vvlfs.lul...r . . . ‘ L .. ‘ . . . . .vl . ‘10 ! l| . . . . nu . :1.‘ w' :l....l?.. 2 A. I . InWHOM];13%|!!qu MSG“ mng I Michigan State University This is to certify that the dissertation entitled FINENESS OF INFLATION-ADJUSTED ACCOUNTING DISCLOSURES AND STOCK PRICE VARIABILITY presentedby Sung Soo Kwon has been accepted towards fulfillment of the requirements for Ph.D. degree in Accounting 153% A. 8M 43 Major professor U G:— Date fl/L/(Mfizf 9? / 9&7 MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES roturn on or bolero date due. I DATE DUE DATE DUE DATE DUE ——T=r If MSU Is An Atflrmdive Adlai/Equal Opportunity Institution FINENESS OF INFLATION-ADJUSTED ACCOUNTING DISCLOSURES AND STOCK PRICE VARIABILITY BY Sung Soo Kwon A DISSERTATION \ Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Accounting 1989 NW (D ‘T ABSTRACT o FINENESS OF INFLATION-ADJUSTED ACCOUNTING DISCLOSURES AND STOCK PRICE VARIABILITY BY Sung Soo Kwon The primary purpose of this study is to investigate the variability of stock prices at the release of inflation- adjusted information disclosures relating to ASR 190. Theory suggests that price variability at the release of information is associated with the "informativeness" of that information. The hypotheses are constructed to examine whether inflation disclosures led to certain hypothesized changes in the information environment. Several different metrics are used to measure the variability of stock returns. The empirical results, based on daily, cumulative, and ratio-based analyses using relative stock returns variability metrics, indicate that the SEC's mandated inflation-adjusted data are associated with increased stock returns variability. This finding is robust under several sensitivity analyses including the post-disclosure investigation and with the use of a comparison group. The finding of increased stock returns variability is applicable to medium and large firms but not for small firms. The different behavior between large/medium firms and small firms may partially explain why previous studies failed to detect information content for the inflation-adjusted data. Interindustry analysis of the information content of the inflation data is also conducted. The evidence suggests that the informativeness of inflation- adjusted information differs across industries, and that these differences are not solely attributed to varying rates of inflation. Several variables, along with the slope coefficients of the market model, are scrutinized to test for changes. in. systematic :risk: conditional on ‘the release of inflation-adjusted. data. The 'average systematic risk of firms exhibits a marginal change between the two different disclosure environments. Therefore, any conclusion based on the observed increase in stock returns variability should be qualified according to the degree of violation of the stability assumption. Finally, the evidence indicates that the phenomenon of increased stock returns variability is not unique to the lgroup of firms with. the highest level of disclosure. However, this indistinguishable stock returns behavior between firms with the lowest and highest level of disclosure may be due to the crude proxy used as the measure of fineness for the required disclosures. AQKEQELEDQEHEHI§ I would like to express my genuine appreciation to a number of people who made this work possible. I am very grateful to the members of my dissertation committee: Professors Stephen L. Buzby (Chair), John J. Wild, and Siva Swaminathan for their invaluable comments and constructive criticism on my work. Particularly, I would like to thank Professor Stephen L. Buzby for being the chairman of the committee and for his instrumental guidance throughout the period of my doctoral program at Michigan State University, and Professor John J. Wild for providing my initial interest in this area of research and rendering prescient guidance and continuous encouragement. 2[ am also indebted to Professors Jerry Han, Amitabh Dugar, Ron Marshall, Peter Schmidt (Economics), James Stapleton (Statistics), Kirt Butler (Finance),and Nick Dopuch (Washington University) for their helpful cements and to doctoral students in the .Accounting Department for their encouragement. I wish to express my appreciation to the Department of Accounting at Michigan State University for financial support during my doctoral program and for providing computing resources during the preparation of the dissertation. iv Finally, I would like to express my sincere appreciation to my wife HwaJung, daughter YoungJoo, son TaeGyu (David), and our parents in Korea. Without their endurance and assistance, it would have been almost impossible to complete the doctoral program. TABLE OF CONTENTS chooser Page 1. INTRODUCTION OOIOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 1 2. PRIORRESEARCH OOOCOCOOOOCOOOOOOOOCOOOOOOOOOOOOOOOOOOO 5 2.1 Accounting Series Release No. 190 ............... 5 2.2 Non-Directional Tests of Information Content .... 6 3. HYPOTHESIS FORMUIATION OOOOOOOOOOOOOOOO0.00.0.0...0.0.11 3.1 BaCkground 0.0.0....O00....0.0.0.000000000000000011 3.2 Hypotheses OOIOOOOOOOOOOOOOO00.0.000000000000000012 4. DATAANDMETHODOIIOGY OOOOOOOOOOOOOOOOOO0.0.00.0000000017 4.1 Sample and Firm Characteristics .................17 4.2 Research Methodology ............................25 4.2.1 Price Variability Measures ............ ..... ....25 Standardized Variance of Returns (SVR) .........25 Standardized Cumulative Variance of Returns (SCVR) OOOOOOOOOOOOOOOOOOOOO0.0... 26 Squared Abnormal Return Ratio (SARR) .......... 27 Standardized Variance of Returns Ratio (SVRR).. 28 4.2.2 Industry Analysis ............................. 28 4.2.3 Test of Changes in Risk ....................... 29 5. EMPIRICAL RESULTS AND IMPLICATIONS .................. 32 The Test of Hypothesis 1 (HA1) ................. 32 Prior Year vs. Initial Disclosure Year ........ 32 Sensitivity Analyses .......................... 42 1 Other Pronouncements around the Disclosure Year ...................... 42 2 Prior Year vs. Post-Disclosure Year ......... 44 3 Initial Disclosure Year vs. Post-Disclosure Year ..................... 48 . Results from the Control Group .............. 50 . The Effect of Event-Date Clustering ......... 54 The Test of Hypothesis 2 (HA2) ................. 60 The Test of Hypothesis 3 (HA3) ................. 70 The Test of Hypothesis 4 (HA4) ................. 74 The Test of Hypothesis 5 (HA5) ................. 79 5 5 5 5 wintnuwmtn vim 0 I I O O O Uihwbnah-H 01p vi 6. SUMMARY, CONCLUSIONS, AND LIMITATIONS ............... 84 6.1 Summary, Conclusions, and Limitations .......... 84 6.2 other Limitations 0.000000000000000000000..0000. 86 FOOTNOTES 000000000000000000.000000000000000...00000000000 88 AEEQDQLX Page APPENDIX A Replication of Table 10 in Beaver, Christie, and Griffin [1980] 0000.00.00.0000000000000000 92 APPENDIX B Replication of Fig. 1 in Gheyara and Boatsman [1980] 0.000.000.0000000000 000000 0000 93 APPENDIX C Patell's Measure .............. ...... ... ...... 95 APPENDIX D Truncated Variability Measure ................ 98 APPENDIX E Summary Statistics for SVR and smueasures 0.0.0.0000.000. 00000 99 APPENDIX F Daily Stock Returns Variability Tests (21-day period, 403 firms) ...............100 APPENDIX G Distribution of lo-K filing Dates in the Post Disclosure Year ..............102 APPENDIX H Characteristics of Comparison Group Sample ...104 APPENDIX I Distribution of lo-K filing Dates for the Comparison Group ....... .......... 105 APPENDIX J Size Effects -- Five Portfolios ..............107 BIBLIOGRAPHY ..0000.0.0.0....0000000000000 0000000000000000 109 vii LIST OF TABLES Table Ease 1(A). Characteristics for ASR 190 Sample ........... ..... 20 1(8). Distribution of lo-K filing date .................. 22 2. The COMPUSTAT-Definitions of Ratios . ..... ......... 30 3. Test of Mean Stability in Event and Estimation Peri-0d .0000000....00.....0.0.000000.0 34 4. Daily Stock Returns Variability Tests Across PRIOR and INITIAL DISCLOSURE Years .............. 35 5. Cumulative Stock Returns Variability Tests Across PRIOR and INITIAL DISCLOSURE Years .............. 36 6. Daily Stock Returns Variability Ratio Tests . ...... 37 7. Effective Dates of Accounting Pronouncements for1976-1977 00.0.00000000000000.000000000 000000 43 8. Daily Stock Returns Variability Tests Across PRIOR and POST Years 0.00.0.000000000000000....00 46 9. Cumulative Stock Returns Variability Tests Across PRIOR and FOST Years 0.0...0.0000000000000000...O 47 10. Daily Stock Returns Variability Tests Across Initial Disclosure and Post Years ............... 49 11. Daily Stock Returns Variability Tests for the Comparison Group .............. .......... 51 12. Cumulative Stock Returns Variability Tests for the Comparison Group .................. ..... 53 13. December Fiscal Year-End Vs. Non-December Fiscal Year-End (Standardized Variance of Returns) ..... 56 14. December Fiscal Year-End Vs. Non-December Fiscal Year-End (Standardized Cumulative Variance of Returns, 5-day) ..................... 57 viii 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. December Fiscal Year-End Vs. Non-December Fiscal Year-End (Standardized Cumulative Variance of Returns, 3-day) ..................... December Fiscal Year-End Vs. Non-December Fiscal Year-End (Comparison Group) ..................... MARCH30, MARCH31, OTHER filing-Date Firms (Standardized Variance of Returns) .............. MARCH30, MARCH31, OTHER filing-Date Firms (Standardized Cumulative Variance Of Returns, 5-daY) 00.000000000000000. 0000000000 MARCH30, MARCH31, OTHER filing-Date Firms (Standardized Cumulative Variance Of Returns, 3-daY) ooooooooooooooooooooooooooooo Size Effects -- Three Portfolios (SVR) ..... ....... Size Effects -- Three Portfolios (SCVR, 5-day) .... Size Effects -- Three Portfolios (SCVR, 3-day) .... Industry Analysis .................... ............. Characteristics of the Industries with Significant Reaction ............................ Tests for Changes in Financing, Production, and Investment DeCiSions 0.0.0.0....0000.00.00.00...0 Tests for Changes in Firms' Systematic Risks (fl) .. Tests Conditional on the Quantity of the Required Disc1°sures 0...00.000.00.00.000000.000.00.000... Tests Conditional on the Quantity of the Required Disclosures (Size-Controlled) ix 58 59 61 62 63 64 71 72 75 80 83 LIST OF FIGURES Figure Page 1. An Example of Sample Stratification ................ 16 2. Event and Estimation Periods .............. ......... 19 Chapter 1 INTRODUCTION A necessary condition for information (e.g., inflation- adjusted disclosures) to be in demand by users of accounting data is that the information be informative (Demski [1984]).1 The informativeness of information can be interpreted in terms of a change in a firm's information-disclosure environment (e.g. , the transition from a coarse environment to a less coarse environment). Ohlson [1979] derives a theoretical rationale for a link between information disclosures and stock prices. He shows that stock prices have a higher variance in a finer reporting environment than in a coarser environment. That is, more state descriptors are revealed in a finer reporting environment yielding a greater revision of previous predictions of probabilities and outcomes. The addition of an informative signal (e.g., inflation-adjusted data) to a vector of information variables is comparable to moving from a coarser to a potentially finer reporting environment, and consequently the movement should yield higher stock price variability.2 The primary objective of this study is to investigate the variability of stock prices associated with the release of inflation-adjusted information required by ASR 190. Premised on the predictions of Ohlson's model, I test for 2 increased variability of stock prices associated with the information required in ASR 190. That is, I test the hypothesis that inflation-adjusted disclosures yield a transition from a coarser to a finer information environment. In addition, I examine several sub-hypotheses based on differential firm reactions to variations in the information environment. The research examined in this paper is distinct from earlier studies in several respects. First, tests of stock price variability are based on a within-firm comparison (as suggested by the analytics of Ohlson's [1979] model) of the two sets of the lo-K reports for the predisclosure year and the initial disclosure year. Previous studies did not make a within-firm comparison in examining stock market behavior around inflation-adjusted information disclosures. Accord- ingly, my methodology avoids the need for external control groups and the potential for experimental error therein.3 Even though the time-series comparison based on the within- firm methodology makes unnecessary the use of such external control groups, it may be more convincing to examine a comparison group that was not required to disclose inflation- adjusted data. Therefore, I use a comparison group for such a purpose, as a part of sensitivity analyses, in Section 5.1.2.4. Second, this research uses the variability of abnormal stock returns as a test statistic (consistent with the model of Ohlson) to examine the information content of inflation-adjusted data. Relatively few studies (e.g., 3 Beaver, Christie and Griffin [1980] and Gheyara and Boatsman [1980]) have used the variability statistic to test for the marginal information content of inflation-adjusted data. The advantages of using the variance of the return test over a mean shifts test are (a) positive and negative responses will not cancel one another, and thus one can detect considerable price effects which are possibly obscured in the latter, (b) the former avoids the need for a market expectations model and the necessity of predicting the sign of any price changes,4 and (c) the use of the former is consistent with Ohlson's analytical model. Third, this study uses short event periods (ranging from one to five days), which makes this research more consistent with "typical" information content studies and will enable me to make stronger inferences regarding the informativeness of inflation-adjusted information. On the other hand, the use of short event periods results in a more stringent test regarding the time period necessary for firms' prices to fully impound the information, as compared to an association test, and hence may reduce the likelihood of finding a significant effect. Most previous studies used longer event periods: for example, 1 week in Ro [1980], 1 month in Lustgarten [1982], and 1 year in Bublitz, Frecka and McKeown [1985]. Therefore, even in the presence of significant results for ‘these. earlier’ studies, it is not clear' that inflation-adjusted information disslssgrss made at the filing date of the lo-K report are informative. Rather, the inflation-adjusted information may have been completely 4 impounded in stock prices prior to the actual release of the information in the lo-K report. Fourth, inter-industry comparison of price variability, conditional on the rate of inflation for each industry, is conducted. For example, specific prices within industries may rise or fall at excessive rates while the general level 5 Differential rates of of inflation remains constant. inflation across industries imply that the demand for inflation-adjusted information may vary from industry to in- dustry. Freeman [1983] and Bublitz, Frecka and. McKeown [1985] argue the merits of such an inter-industry analysis. Fifth, the stock price variability of large firms will be compared with that of small firms in order to investigate for any implications related to firm-size. Since no previous study of ASR 190 has made a size-related comparison, this investigation may provide additional insights on the market's reaction to inflation data. The remainder of this paper is organized as follows. Chapter 2 provides a brief review of the prior research that is most relevant to this study. Chapter 3 presents the analytical model and its associated hypotheses. The data and the research methodology are discussed in Chapter 4. Chapter 5 presents the empirical results and their implications. Chapter’ 6 summarizes the findings and discusses possible limitations of this study. Chapter 2 PRIOR RESEARCH 2.1 Asssunting Ssriss Relsase Ng.129 The Securities and Exchange Commission's Accounting Series Release No. 190 (ASR 190) issued in March 1976 mandated the public disclosure of certain replacement cost accounting data by its registrants.6 ASR 190 required replacement cost disclosures as a note to the lo-K financial statements or as part of a separate section of the lo-K 7 The following financial statements following the notes. disclosures were required of companies for fiscal years ending on or after December 25, 1976: (a) the current replacement cost of inventories measured at each fiscal year-end; (b) the estimated current cost of replacing the plant assets together with the current depreciated replacement cost of the plant assets on hand at the end of the fiscal year; (c) the approximate amount that cost of sales (cost of goods sold) would have been if it had been calculated. by estimating the current replacement cost of goods and services sold, for 'the two ‘most recent fiscal years; (d) the approximate amount of depreciation, depletion, and amortization which would have been recorded if it were estimated on the basis of average current replacement cost of plant assets, for the two most recent fiscal years: (e) 6 description of the methods used in estimating the amounts of the above items; and (f) any additional information which management believes is necessary to prevent the above information from being misleading. According to ASR 190, the replacement cost disclosures are not required if the firm's inventories and productive assets aggregate less than $100 million or comprise less than 10% of total assets at the beginning of the most recently completed fiscal year. The SEC rescinded its replacement cost disclosure requirements when FASB Statement No. 33 became operative in 1979. 2.2 Nsn-Directiongl Tests of Information Content 8 Non-directional tests are defined in this paper as those in which positive and negative stock reactions are not distinguished and, thus, will not cancel one another. In other words, the direction (or sign) of the stock price reaction can not be separately identified in non-directional tests. Typical examples of non-directional tests are stock price variability tests and absolute abnormal stock return tests. Beaver, Christie and Griffin [1980], hereafter BCG, studied the market's reaction 'to three events related to ASR 190; (l) the initial SEC proposal (August 21, 1975), (2) the actual adoption (March 23, 1976), and (3) the first ASR 190 filings (assumed as March 31, 1977 for lo-Ks).9 They examined the daily returns of 553 firms affected by ASR 190. Their tests focused on the 15 days before and after each event, as well as for the entire interval of over 22 months (June 2, 1975 - April 22, 1977). Their sample was drawn from .7 a group of firms on which replacement cost data were obtained for the fiscal year-ends from December 31, 1976 to November 30, 1977. BCG partitioned firms according to certain variables that were subject to the SEC-required replacement cost disclosure.10 They examined the relative volatility of stock price changes during the intervals of interest (the three event periods) relative to certain non-report periods. Two types of non-report periods were used: (1) the previous thirty-one days, and (ii) the overall period. The authors justified their use of this nondirectional test by arguing that since the direction of the price change can not be specified, some transformation of the residual return that abstracts from its sign is needed. They noted that the inability to correctly specify the direction of the price effect may lead to offsetting price effects within a given group, which can reduce between-group differences and increase within-group variation. As a volatility measure, BCG used Uit = “‘11:” / R—i’ . where Rit is the daily return on security i for day t during a report period, and R11 is the mean square stock return during a non-report period. As indicated by the authors [p.152 and p.155], two problems are related to the use of this measure of volatility. First, this is not a ratio of variances unless the daily mean return is zero. Second, and more importantly, the common movement of the ratio over time will undoubtedly be related to the arrival of market-wide 8 information, and thus the probability of detecting firm- specific (or portfolio-specific) information will be decreased.11 Consequently, it is not surprising that BCG do not find any significant difference between the high portfolio group and the low portfolio group in reacting to the inflation—adjusted data.12 In Appendix A, BCG's results of mean daily stock return volatility ratios are reproduced and replicated using my sam- ple for the period of March 24 through April 7 in the initial disclosure year. As I note in Section 4.1, BCG's sample is different from mine due to the difference in the data collection methods. However, the measures of stock volatility ratios are significantly less than one in both cases. In addition, as in the results of SVR, the most significant information effect was detected in day {-1}, that is, march 30 (see Panel B of Table 1(8)) in both cases. Moreover, in either case, the least effect was found in April 6. Gheyara and Boatsman [1980], hereafter GB, also tested whether ‘the SEC's :mandated replacement cost disclosures provided incremental information ‘to the [capital market.13 They used 106 reporting firms as a treatment sample and 83 exempted firms as a control sample to examine their hypothesis. The variability measure was used to obtain a graphic portrayal of stock price changes during the 50 day report period surrounding the lo-K release date. Regression parameters were estimated over the 200 day non-report period. Beaver's [1968] "U-statistic"_ was calculated for reporting 14 firms and control firms. The plots of stock prices 9 indicated that they fluctuate in what appeared to be a random manner about an expected value of approximately one. The following caveats are in order in interpreting the results in GB. First, few confounding variables (e.g., size and beta changes) are controlled for, and thus it is not so easy to interpret their results. Second, the incremental information content of 10-K reports for non-reporting firms may be higher than that for reporting firms because of the 'size effect' in that reporting firms are larger in size.15 Therefore, this may have obscured the difference between the two groups in price variability tests for replacement cost disclosures. Third, their analysis is across-firm and not within-firm as suggested by the analytics in Ohlson [1979] (and Lev [1988]). Specifically, GB investigated the stock price response before and after the 10-K filing date of the first mandatory disclosure year but did not compare within- firm stock price variability for the 10-K filing date of the first effective year with the 10-K filing date of the previous year. In Appendix B, GB's graph that displays average values of Beaver's U for the 106 treatment firms on trading days adjacent to the public filing date of the 10-K report was reproduced and replicated using my sample. As in the GB's case, Beaver'Us are clustered around one in my case. However, since much larger number of firms are used in my sample, it is difficult to compare abnormal behavior of Beaver's Us of GB with that of mine on certain event days. 10 In general, prior studies have failed to find evidence of information content in the inflation-adjusted disclosures mandated by ASR 190. However, the evidence to date leaves many issues unresolved (e.g., lack of a theory of how the market should react to inflation-adjusted information, grouping (matching) problem, and the size-effect problem). The importance of this topic, along with various methodological refinements that have subsequently appeared in the literature, provides the motivation for my re-examination of the information content of ASR 190's inflation-adjusted information disclosures. Chapter 3 HYPOTHESIS FORMULATION 3-1 W114 Ohlson [1979] provides an analytical model that links additional information disclosure and stock price variability under a setting of uncertainty. In his model, information is viewed as a state descriptor helpful in assessing the probability distribution of states in a capital asset valuation model. As information becomes publicly available ix: the financial. reports (e.g., inflation-adjusted. supple- mentary disclosures), investors can be expected to update their beliefs (i.e., their probability distributions about the changed states of nature). using a simple linear asset pricing model (which is assumed to specify the information dynamics), Ohlson shows that stock prices have a higher variance in a finer information environment than in a coarser environment (Ohlson [1979, p.214]). This is because, in a finer environment, investors have more state descriptors that lead to greater revisions in their prior probabilities of states. Ohlson's analytics are. dependent on. the firm's financing, production, and investment (FPI) decisions not being systematically affected by the information disclosed. This implies that risks and expected returns will be 11 12 essentially independent of the disclosure environment. An implication of the above assumption is that information disclosure cannot alter the fundamental value characteristics of a stock. From the point of view of empirical research, this result is important because the information content hypothesis cannot be tested unless the change in disclosure had no systematic impact on FPI (Ohlson [1979, p.213]). Ohlson's analytical model provides a framework for this study's hypotheses. Suppose that some supplementary information (inflation-adjusted information) is revealed for the first time in a 10-K or an annual report. If such information adds more state descriptors to the existing information vector, then the new information environment can be considered to be finer than the previous one (see footnote 2). Therefore, the variability of stock prices associated with the release of the new information should be greater in the current year (the new information environment) than in the prior year (the old information environment). 3.2 nypgtheses The intent of ASR 190 (see footnote 6) was to provide information to users that would assist them in better understanding the current costs of operating the business, along with information which can not be obtained from historical cost financial statements alone. Thus, if ASR 190 disclosures are informative, then a transition from a coarser to finer information environment is predicted. 13 HA1: The SEC's mandated inflation-adjusted data (ASR 190) increased the stock price variability around the release of the 10-K report for the first mandatory disclosure year relative to the prior non-disclosure year. Prior research by Atiase [1985], Grant [1980], and Collins, Kothari, and Rayburn [1987] provides empirical evidence consistent with the hypothesis that the amount of predisclosure information production and dissemination is an increasing function of the market value of the firm. According to these earlier studies, more traders and professional analysts are processing the available information about larger firms vis—a-vis smaller firms, and thus the stock prices of larger firms will be more informative. Thus, if inflation-adjusted data have information content, and if the size effect hypothesis is applicable to a within-firm comparison, then the price reaction will be more pronounced for smaller firms relative to larger firms at the release of this information. Accordingly, the following hypothesis is formulated: HA2: The difference in price variability between the initial disclosure year and the prior year is more pronounced for smaller firms than for larger firms. Freeman [1983] and Bublitz, Frecka and McKeown [1985] suggest that it would be fruitful to consider interindustry differences in information content studies of inflation- adjusted disclosures. The varying degrees of inflation rates 14 across industries provides additional motivation to examine the following alternative hypothesis: HA3: The degree of informativeness of inflation- adjusted information disclosures differs between industries, and these differences are associated with industry-specific inflation rates. Tests will also examine for FPI changes between information environments. Several ratios are used to proxy for firms' FPI changes (see Section 4.2.3). As noted by Ohlson [1979] and McNichols and Manegold [1983], the analysis of systematic risk does not imply that any ex post assessment of covariance risk will remain constant as the information environment changes. Market participants may, in fact, revise their expectations of an individual firm's beta as a result of additional disclosures, but, given the assumption that good news is as likely as bad news, these revisions may not be systematic. HA4: The beta (or a set of specific ratios that are representative of accounting determined risk measures) of firms which disclose inflation- adjusted information is not significantly different in the disclosure year vis-a-vis the previous year. The degree of expectation for a change in the information environment depends on the quantity (i.e., fineness) of the required disclosures made in the initial disclosure year, if such disclosures are informative.16 As shown in Figure 1, the sample of firms in this study will be 15 stratified according to the extent of their inflation related disclosures. In this scenario, a larger stock price reaction is predicted for the firms with the greatest level of disclosure, and conversely a smaller stock price reaction is predicted for the firms with the lowest level of disclosure. HA5: The difference in the stock price variability between the prior year and the initial disclosure year is more pronounced for firms with the greatest level of disclosure as compared to firms with the lowest level of disclosure. 16 FIGURE 1 An Example of Sample Stratification Effective Date lo-K Filing of Date ASR No. 190 [Mt] I I Dec. 25, 1976 1977 where Mt = Mandatory disclosure at day t in the initial disclosure year. W Degree 1 (lowest level of Mt) Degree 2 Degree K (highest level of Mt) If inflation-adjusted data are indeed informative, firms with degree K disclosure will show more significant stock price reaction than firms with degree 1 disclosure. This follows from Ohlson's [1979] analysis regarding the fineness of information. Chapter 4 DATA AND METHODOLOGY 4.1 W The sample of firms is collected from the Standard & Poor's COMPUSTAT file by applying the criteria specified in ASR 190. I attempted to duplicate BCG's sample in order to compare my price variability results with their price volatility results. However, since they obtained their sample directly from the SEC, and since the availability of the data in the COMPUSTAT tape, CRSP tape, and microfiches is limited due to missing values or illegible SEC dates, the size of my sample (407 firms) is smaller than that of BCG (553 firms, see Appendix A). Other sample selection criteria are described below. First, the dates on which the lo-K reports are received by the SEC must be stamped on the reports themselves. The objective of this criterion is to allow an identification of the official public release date of the 10-K. These dates are collected manually from microfiche copies of the lo-K reports for 1975-1977. The period of 1975-1977 is chosen because the data for the 10-K filing dates should be obtained for the fiscal year-ends from December 31, 1976 to November 30, 1977. Since the disclosures of inflation-adjusted data were required of my sample companies for fiscal years ending 17 18 on or after December 25, 1976, the initial disclosure year is 1976 (1977) and the prior year is 1975 (1976) for the December Fiscal Year-End firms (the Non-December Fiscal Year- End firms), respectively. Second, stock return data must be available on the University of Chicago's Center for Research in Security Prices (CRSP) Daily Returns File for each firm in the sample for the 210 day estimation period of the market model and the 5 day event period. Thus, the estimation period begins 212 days prior to the 10-K filing date (day {0}). I also examine the behavior of daily stock prices during the event period for purposes of making stronger inferences regarding the informativeness of inflation- adjusted information (see Figure 2). Among the 654 firms which met the ASR 190 criteria, 407 firms have available stock return data from the CRSP Daily Returns File and legible lo-K filing dates from microfiche copies. Panel A of Table 1(A) presents some information of the sample screening results. Panels of B and C of Table 1(A) present some characteristics of the sample. The sample is classified by fiscal year-end, exchange listing code, and 2-digit standard industrial classification code (SIC). The majority of the firms in the sample have a December 31 fiscal year-end (69.3%) and are listed on the New York Stock Exchange (NYSE, 72.3%). Almost 31% of the sample firms come from 3 industries: electrical, gas, and sanitary services industry (60 firms), chemicals and allied products industry(38 firms), 19 FIGURE 2 Event and Estimation Periods [1] - ev er 0 Estimation Period |--> (210 days) <-- T I I I I I -212 -3 -2 -1 0 +1 |-> <-| 3-day Event Period [2] 5-dgy even; pegiod Estimation Period |--> (210 days) <-- I I I I I I I -212 -3 -2 -1 0 +1 +2 |-> <-| 5-day Event Period where: Day {0} = The date of receipt of the 10-K by the SEC. 20 TABLE 1 (A) Characteristics for ASR 190 Sample Panel A: Sample selection criteria No. of firms Compustat firms: Primary industrial, Tertiary, Supplementary Industrial Primary Industrial and in the S&P Industrials Index.. ................ 2400 Less: Inventories and plant assets less than $100 million or less than 10% of total assets................(1746) Less: Stock returns unavaiable on the CRSP daily returns tape or unable to read the lo-K filing date from a microfiche or missing microfiches...........(247) Sample size 07 Panel B: Classification of Sample WM M__g____g_<;_chan e Listin ode January ........ 14 NYSE and in the S&P February........ 9 Industrials Index ... 105 March........... 7 NYSE and in the S&P April .......... 4 Utilities Index...... 13 May ........... 9 NYSE and in the S&P June ........... 22 _ Transportation Index .. 3 July ........... 12 NYSE and not in the August.......... 8 S&P 500 Index ....... 175 September....... 23 AMEX and not in the October......... 13 S&P 500 Index ........ 97 November........ 4 Over-the-counter and December........ 2&2 not in the S&P 500 Total 4 7 Index (CRSPDLY firms)..13 AMEX and in the S&P Industrials Index ... 1 Total 407 21 TABLE 1(A) (cont'd.) By Igg-Qig't gas: $19 Code 2M9 Name We ' Bergen; 10 Metal Mining ........................ 6 1.5% 12 Bituminous Coal and Lignite Mining .. 2 0.5 13 Oil and Gas Extraction .............. 14 3.4 15 General Building Contractors ........ 1 0.2 16 Heavy Construction Contractors ...... 4 1.0 20 Food and Kindred Products ........... 21 5.2 21 Tobacco Manufactures ................ 3 0.7 22 Textile Mill Products................ 4 1.0 23 Apparel and Other Textile Products .. 3 0.7 24 Lumber and Wood Products ............ 2 0.5 26 Paper and Allied Product ............ 16 3.9 27 Printing and Publishing ............. 9 2.2 28 Chemicals and Allied Products ....... 38 9.3 29 Petroleum and Coal Products ......... 11 2.7 30 Rubber and Misc. Plastics Products .. 4 1.0 31 Leather and Leather Products ........ 3 0.7 32 Stone, Clay, and Glass Products ..... 11 2.7 33 Primary Metal Industries ............ 24 5.9 34 Fabricated Metal Products ........... 9 2.2 35 Machinery, Except Electrical ........ 30 7.4 36 Electric and Electronic Equipment ... 20 4.9 37 Transportation Equipment ............ 20 4.9 38 Instruments and Related Products .... 11 2.7 39 Misc. Manufacturing Industries....... 1 0.2 40 Railroad Transportation ............. 3 0.7 42 Trucking and Warehousing ............ 1 0.2 44 Water Transportation ................ 3 0.7 45 Transportation by Air ............... 3 0.7 47 Transportation Services ............. l 0.2 48 Communication ....................... 7 1.7 49 Electric, Gas, and Sanitary Services. 60 14.7 50 Wholesale Trade-Durable Goods ....... 5 1.2 51 Wholesale Trade-Nondurable Goods .... 4 1.0 53 General Merchandise Stores .......... 11 2.7 54 Food Stores ......................... 8 2.0 57 Furniture and Home Furnishings Stores 1 0.2 58 Eating and Drinking Places .......... 1 0.2 59 Misc. Retail ........................ 4 1.0 62 Security, Commodity Brokers & Services 1 0.2 65 Insurance Agents, Brokers 8 Services .. 3 0.7 67 Holding and Other Investment Offices .. 8 2.0 70 Hotels and Other Lodging Places ....... 4 1.0 72 Personal Services ..................... 1 0.2 73 Business Services ..................... 2 0.5 75 Auto Repair, Services, and Garages .... 1 0.2 78 Motion Pictures ....................... 4 1.0 80 Health Services ....................... 4 1,9 Total ...............................407 100.0% 22 TABLE 1(8) Distribution of 10-K Filing Dates Panel A: All Firms (407 firms) No. of No. of Henth _firne Eereent _firne Reteent January 12 2.9% 14 3.4% February 8 2.0 6 1.5 March 256 62.9 260 63.9 April 34 8.4 33 8.1 May 11 2.7 8 2.0 June 8 2.0 10 2.5 July 3 0.7 4 1.0 August 10 2.5 8 2.0 September 22 5.4 23 5.7 October 12 2.9 8 2.0 November 8 2.0 12 2.9 December 2; 5,7 g; 5.2 Total =2; 100.0% 22; 100.0% Panel B: December Fiscal Year-End Firms (282 firms) [1] Breakdown by month .2eried _Dieeleente_xeer .2reeieeleenre_xeer No. of No. of _firne £ereent. _firne Eereent February: 2/1 - 2/10 0 0.0% 1 0.4% 2/11 - 2/20 3 1.1 0 0.0 2/21 - 2/28 0 0.0 0 0.0 March: 4/1 - 4/10 4/11 - 4/20 4/21 - 4/30 Total 3/1 - 3/10 1.4 5 1.8 3/11 — 3/20 1.8 5 1.8 3/21 - 3/31 87.6 250 88.6 April: 6 4 0Q 00 Ibd 19 6. 00 N N 00 N ab N PM ‘1th H O O O 0 d? ”L 00 N H H O O O 09 23 TABLE 1(B) (cont'd.) [2] Further breakdown of 3/21 - 3/31 period ' .2renieeleenre_1eer No. of No. of Day _firne firne 3/21 ............. a ................ 0 3/22 ............. 0 ................ 10 3/23 ............. 3 ................ 4 3/24 ........ ..... s ................ 1 3/25 ............. 10 ................ 12 3/26 ............. 0 ................ 23 3/27 ............. 0 ................ 0 3/28 ............. 27 ................ 1 3/29 ............. 19 ................ 81 3/30 ........ ..... 53 ................105 3/31 ........ ..... 122 ................ 1; Total ............. £21 ................;22 Panel C: Non-December Fiscal Year-End Firms (125 firms) D sc s Y Prsdisclosurs Ysa; No. of No. of Menth _firne Eereent _fitne Eereent January 12 9.6% 14 11.2% February 5 4.0 5 4.0 March 0 0.0 0 0.0 April 11 8.8 12 9.6 May 11 8.8 8 6.4 June 8 6.4 10 8.0 July 3 2.4 4 3.2 August 10 8.0 8 6.4 September 22 17.6 23 18.4 October 12 9.6 8 6.4 November 8 6.4 12 9.6 December 2; 18.4 31 16.8 Total 12 100.0% 125 100.0% 24 and machinery except electrical industry (30 firms). Panel A of Table 1(B) shows that a considerable number of firms (63.9% for the predisclosure year and 62.9% for the initial disclosure year) file their 10-K reports with the Securities and Exchange Commission (SEC) in March. Panels B and C provide further details of the distribution of 10-K filing dates for firms which are subdivided into December and non-December fiscal year-end firm groups. The majority of firms (88.6% for the predisclosure year and 87.6% for the initial disclosure year) in. the December fiscal year-end group filed their 10-Ks in the period 3/21 - 3/31. Further breakdown of this period shows a majority of firms filed their 10-K reports between 3/29 and 3/31 (see part 2 of panel B). This suggests the possibility of an "event-date clustering problem" with respect to the measurement of abnormal returns. Dyckman, Philbrick, and Stephan [1984, p.29] show that event-date clustering by time generally reduces the power of statistical tests in detecting abnormal stock price behavior. On the other hand, Brown and Warner [1985, p.15] conclude that the power of statistical tests is not significantly different when there is clustering in event dates. Therefore, while the issue of event-date clustering is still an empirical question, the evidence suggests that (if anything) the tests may be slightly "biased" in favor of the null hypothesis of "no effect" (see Section 5.1.2.5). For the Non-December fiscal year-end group (shown in panel B of Table 1(B)), the lo-K filing dates in both the initial disclosure year and the predisclosure year are far 25 more evenly distributed relative to the December fiscal year- end group. 4.2 Eeeeeren_netnedelegx The dependent variable of interest in tests of the hypotheses is the stock price variability around the release of the lo-K report. 4.2.1 Etiee_YerieeilitY_Eeeenree The market model of Sharpe [1964] and Lintner [1965] is adopted to remove market-wide factors from stock returns. The model describes a linear relationship between the stock return of firm i and the return on a market portfolio. The one-factor market model for firm i is characterized as: Rid = “i + 31 * Rmd + eid where, Rid = return on security i at time d, Rmd = return on market portfolio at time d, (CRSP equally-weighted index) 51 = C°v(Rid' Rmd)/var(Rmd)' 1 E(Rid) ' Bi * E(Rmd)' eid = error term 9 II I. Stangezdiaeg_!e£igneele§;BetB£n§11§VR To gauge stock price variability, a "U" statistic, initially proposed by Beaver [1968, p.81], is computed. This is the simplest form of my price variability measures and is defined as follows: SVR = “zit/512 <1> where “it is the abnormal stock return during the forecast period for stock i at time t; that is, “it = Rit ‘ (a: + bi * Rmt)' 26 where a1 and bi are the OLS estimates of the market model parameters (01, Hi), and Si is the standard deviation of the residuals during the estimation period. A similar metric was also introduced by Patell [1976, p.258]. Patell's U was applied to my data, however, it provides the same results as my SVR metric in all cases. Therefore, I chose to present only the results of the SVR metric in the tables of this paper to avoid redundancy in data presentation. However, a discussion of Patell's daily stock price variability measures is included in Appendix C. II - Cumul Retu_s___ If the sxsst public release date of the 10-K report does not match the lo-K receipt date by the SEC, the use of a cumulative return variance metric may reduce the potential error in determining the event day. In response to this concern, the following standardized cumulative metric, similar to McNichols and Manegold [1983, p.73] and Wild [1989, p.4] is computed: L SCVRiL = (l/Iot§g[u21t(Ti-4)]/[Citsi2(Ti-2)]} <2> —— T. — where: Cit = 1 + l/Ti + (amt - Rm)'/d§:(Rmd - Rm)“, __. ' T1 = (1/T ) 2' I Rm 1 d=1Rmd Cit is the increase in variance due to prediction outside the estimation period (see Patell [1976, p.256]), T- is the 1 number of observations used to estimate the market model for firm i, and L is the number of cumulation days. 27 Assuming that the u2 it terms in the numerator of the SCVR are serially independent (and thus the SCVRiL does not significantly deviate from the specified F-distribution), then the cumulative ZSCVR statistic, similar to McNichols and Manegold [1983, p.73] and Wild [1989, p.11], can be computed to conduct direct comparison tests of the SCVR values between the predisclosure environment (PE) and the disclosure environment (DE).17 N £§iSCVRiL,DE ‘ SCVRiL,PE)] zSCVR = <3) N [(2/L22§(T1,DE'3)/(Ti,DE'6) + ('I'i,pE-3)/013,131.34»n‘1 1: where the variables are defined in equation <1> and <2>. Due to the potential violation of the serial independence assumption (Footnote 17), less emphasis is placed on ZSCVR in interpreting the results in Chapter 5. III. S- aerd Ab_o__rmleturati__ SARR A squared abnormal return ratio metric is an undeflated variability measure computed as: SARR = “zit/”zit' <4> where “it is the abnormal stock return for stock 1 at time t in the initial disclosure year, and “it' is the abnormal stock return of stock 1 at time t' in the prior year.18 McNichols and Manegold [1983, p.66] discuss the merit of using undeflated variability measures by arguing that if the estimation-period variances are appreciably lower in the initial disclosure year, then the SVR in that year could be inflated, biasing the test against the null hypothesis. 28 IV - Htanaird ed Vanua_ mum-331- A standardized variance of returns ratio metric is computed as the ratio between SVR for the initial disclosure year and SVR for the prior year: SVRR = SVRt / SVRt. = (u’it/si‘) / (u’it./s'i’) <5> where “it and “it' are as defined in equation <2>, and Si (s'i) is the standard deviation of the residuals during the estimation period in the initial disclosure (prior) year, respectively.19 By the nature of the ratio form, its use reinforces the research design based on matched-pair tests as suggested by Ohlson [1979]. 4-2-2 IDQB§£IY_An§l¥§i§_iflAll To test HA3, the firms in my sample are first classified according to two-digit SIC codes. Next, the industries which experience the highest inflation and the. industries which experience the lowest inflation are identified. The differences in price variability measures are then investigated between these different categories of industries. The tests also consider firm-size which is measured by the market value of equity. 4.2.3 Isst 9f Changes in Risk (HA4) Beaver, Kettler and Scholes [1970] examined the association between market determined and accounting determined risk measures. They suggested that earnings variability, dividend payout, accounting beta (covariability of earnings), and leverage are significantly correlated with market determined risk measures. Also, McNichols and 29 Manegold [1983] used similar variables to test for possible FPI changes. Therefore, this study adopts these variables in examining changes in risk. The five ratios to be scruti- nized are i) debt/equity, ii) dividend payout, iii) sales/total-assets, iv) debt/total-assets, and v) earn- ings/total-assets (or return-on-total-assets) . Accounting data for computation of these variables are obtained from the Standard and Poor's Compustat files. Table 2 shows detailed information regarding the source elements of those ratios from the COMPUSTAT file. For each firm, the relevant ratios are computed for each of the two years (disclosure year and predisclosure year). The differences between the disclosure and predisclosure year ratios are analyzed. Several parametric and nonparametric tests are used to test for differences between these two years. In addition to these five ratios, the slope coefficients from the market model (Section 4.2.1) are also used to measure the firms' systematic risks. Specifically, the slope coefficients from the estimation period for the initial disclosure year SVRs are used as the prior year systematic risks and the slope coefficients from the estimation period for the post disclosure year SVRs are used as the disclosure year systematic risks. 30 TABLE 2 The COMPUSTAT-Definitions of Ratios Ratio Definition (COMPUSTAT Data Item No.) 1. Debt/Equity Debt = Assets(6)-Common Equity(60)-Preferred Stock(130) Equity = Common Equity(60) + Preferred Stock(130) Dividend Payout(Dividends per Share/Earnings per Share) Dividends per share(26) Earnings per Share(58) Sales/Total Assets Sales(12) Total Assets(6) Debt/Total Assets Debt = Assets(6)-Common Equity(60)-Preferred Stock(130) Total Assets(6) Earnings/Total Assets Earnings(20) Total Assets(6) Chapter 5 EMPIRICAL RESULTS AND IMPLICATIONS 5.1 Ins Isst ot Hypothesis 1 (HA1) 5.1.1 Etisr Yss: 1s, Initial Qisslosuts Xsa; The first hypothesis predicts that the SEC's mandated inflation-adjusted data increased the stock returns variability around the release of the lo-K report for the disclosure year relative to the prior year. Before I proceed with the relative stock price variability tests, I need. to know' whether' my standardized. variance of return metrics based on Beaver [1968, p.81] and Patell [1976, p.258] are direct test metrics for increased variance. According to Patell [1976, p. 256-257], the standardized variance of returns (SVR) metric cannot be considered a test solely of increased variance during the event period if the expected value of “it (the prediction error) is not equal to 0, since it measures the prediction errors as deviations from the hypothesized mean value of 0.. Only in the event of a zero value for E(uit) can we use the SVR as a direct test metric for increased variance. Patell [1976, p. 256-257] used the following test statistics to examine the above assumption: Vit = “it/ (8i Citk) -- a Student t statistic with N N Ti-z degrees of freedom, th = [2 Vit] / [ 2(Ti-2)/(Ti -4) 11 -- A normalized =1 i=1 sum of Vit' 31 32 where “it and Si are defined in equation <1>, Cit reflects the increase in variance due to prediction outside the estimation period (see Appendix C), and Ti is the number of days in the estimation period. By the Central Limit Theorem, the distributions of zvt are predicted to be unit Normal for large samples under the null hypothesis of zero Vit' As indicated by the values of th in Table 3, the mean standardized prediction error (vit) is not significantly different from zero in any event day. Therefore, Table 3 assures that the estimation and event-period means of the firm-specific stock returns (abnormal returns) are not significantly different from each other. This evidence suggests that one may proceed with the relative stock price variability tests. The empirical evidence for the first hypothesis is contained in Tables 4, 5, and. 6 ‘which show' stock price variability measures from four different metrics (SVR, SCVR, SARR, and SVRR) and their matched-pair test results for the five days around the event day (lo-K report filing date). Table 4 provides results based on the daily Standardized Variance of Returns (SVR) measures. The null hypothesis of no difference between the predisclosure year and the initial disclosure year in stock price variability is rejected at day {-2}, {-1}, and {+1} with the parametric t-test, and at days {-2}, (-1}, {0}, and {+2} with the both nonparametric tests (Sign test and Wilcoxon Signed-Rank test). Especially, the null hypothesis is rejected at days {-2} and {-1} under both 33 TABLE 3 Test of Mean Stability in Event and Estimation Perioda 2210: Yes; S 9 ear Ptior Yea; ASR 190 Year Event Day Vt Vt th th -2 0.0160 0.0167 0.3199 0.3350 -1 0.0280 -0.0036 0.5613 -0.0713 0 0.0482 -0.0096 0.9665 -0.1920 +1 0.0314 0.0237 0.6287 0.4753 +2 -0.0466 -0.0273 -0.9350 -0.5467 a The PRIOR and ASR 190 designations refer to inflation- adjusted information releases through the 10-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. 34 TABLE 4 INITIAL DISCLOSURE Yearsa Daily Stock Returns Variability Tests Across PRIOR and 21191.1231 A§B_129_X§§I ' tennezieen_1£:xelneib Wilcoxon Event SVR SVR Student Sign Test Signed-Rank Day Mean Mean t Z Z * *** *** -2 0.67 0.78 l.29(0.10) 2.63(0.00) 2.32(0.01) ** *** *** -1 0.64 0.80 1.67(0.05) 2.93(0.00) 3.04(0.00) ** *‘k 0 0.86 0.91 0.30(0.38) 1.64(0.05) 1.94(0.03) * +1 0.77 0.98 1.50(0.07) 0.05(0.48) 0.9l(0.18) * ** +2 0.91 0.90 -0.06(0.52) 1.44(0.08) 1.94(0.03) * *** *** Meanc 0.77 0.88 l.48(0.07) 2.53(0.01) 2.42(0.01) a SVR is defined in equation <1>. The PRIOR and ASR 190 designations refer to inflation—adjusted information releases through the 10-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. Event day {0} is the date of receipt of the 10-K by the SEC. b A paired comparison is made by taking the difference in SVRs between the disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. c Mean SVR is computed by taking the average of five day SVRs for each firm and then determining the average of all firms. A similar measure (the cumulative U) is used on p. 73 of McNichols and Manegold [1983]. By the nature of construction, this value should be equal to the SCVR value at day {2}. 35 TABLE 5 Cumulative Stock Returns Variability Tests Across PRIOR and INITIAL DISCLOSURE Yearsa Event SCVR Day Mean Erier_Yeer A§B_129_Yeer SCVR Mean anngtison (P-xalne)b Wilcoxon zSCVR Sign Test Signed-Rank Z Z Panel 1: Cumulation Period of Day {-2} through Day {+2} -2 -1 0 +1 +2 0.66 0.65 0.71 0.72 0.76 0.77 0.78 0.82 0.85 0.86 1.10(0.16) ** 1.87(0.03) ** 1.80(0.03) *** 2.58(0.01) *** 2.25(0.01) *** 2.53(0.01) *** 2.63(0.00) *** 2.93(0.00) *** 3.42(o.00) *** 2.48(0.01) *** 2.30(0.01) *** 2.81(0.00) *** 2.57(0.01) *** 2.63(0.00) *** 2.37(0.01) Panel 2: Cumulation Period of Day {-1} thr * ough Day {+1} *** *** -1 0.63 0.79 1.54(0.06) 2.83(0.00) 3.00(0.00) *** *** 0 0.74 0.84 1.00(0.16) 3.13(0.00) 2.52(0.01) * *** *** +1 0.75 0.88 1.35(0.09) 2.83(0.00) 2.63(0.00) SCVR and Z are defined in equation <2> and <3>, respectively. Tge PRIOR and ASR 190 designations refer to inflation-adjusted information releases through the lo-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. by the SEC. Event day {0} is the date of receipt of the mean rank and the number of cases where SCVR initial disclosure year is greater than SCVR1 The symbols of *, **, aha non-disclosure year. statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. I o b The results for the nonparametric tests are based on the for the for *** the prior indicate 10-K 36 TABLE 6 Daily Stock Returns Variability Ratio Testsa ane : ” Dny Msnn Student tC *** —2 3.05 10.76(0.00) *** -1 3.28 11.61(0.00) *** 0 2.99 10.43(0.00) *** +1 2.82 9.58(0.00) *** +2 3.37 11.54(0.00) *** Mean of {-2,+2}d 3.10 22.50(0.00) Panel B: SVRRb Dex Mean fitneent_t° *** -2 4.65 15.96(0.00) *** -1 4.76 16.36(0.00) ' *** 0 4.42 15.06(0.00) *** +1 4.05 13.52(0.00) *** +2 4.60 15.52(0.00) *** Mean of {-2,+2}d 4.50 32.30(0.00) a The squared abnormal return ratio metric (SARR) and the standardized variance of returns ratio metric (SVRR) are defined in equations <4> and <5>, respectively. b SARR or SVRR greater than 10.0 is truncated at 10.0. Similar results were obtained using a 5.0 truncation rule. c A paired comparison is made by taking the difference in SARRs (or SVRRs) between the initial disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one-tailed tests (SCVR or SVRR greater than 1.0). d Mean SARR (SVRR) is computed by taking the average of five day SARRs (SVRRs) for each firm and then determining the average of all firms. 37 the parametric test and the nonparametric test. Stated differently, the inflation-adjusted. disclosures are consistent with incremental information content relative to the no-disclosure prior year and with a reaction 1 to 2 days prior to the "official" 10-K filing date. This early reaction" phenomenon is common in information content studies (see Chambers and Penman [1984], Foster [1977], and Morse [1981]). Average SVRs of the five-day event period are less than one in both the prior year and the initial disclosure year in Table 4. This fact seems 'to be inconsistent with the evidence for the incremental information content of the 10-K, documented by Foster and Vickrey [1978]. However, this phenomenon disappears around the 10-I( filing date in the post-disclosure year (see Table 8). Mbre importantly, these less-than-one SVRs are consistent with the less-than-one stock volatility ratios of the prior research by Beaver, Christie, and Griffin [1980] (see Appendix A). Throughout this thesis, I use both Sign and Wilcoxon Signed-Rank statistics (except Table 23) as nonparametric tests to reduce measurement method bias. However, the Sign test considers only the signs of the differences and disregards their magnitudes, and thus more weight is naturally placed on the Wilcoxon Signed-Rank test. Similar results based on truncated values are presented in Appendix D to examine whether extreme values of the stock price variability measures played an important role in the 38 findings of Table 4.20 The rejection of the null hypothesis of no difference in stock. price variability between the predisclosure year and the initial disclosure year is persistent even in the absence of extreme values. The Standardized Cumulative Variance of Returns (SCVR) values for days {-2} through {+2} and days {-1} through {+1} are displayed in panels 1 and 2, respectively, in Table 5. Consistent with the results in Table 4, these findings in- dicate that stock returns variability at the release of the 10-K report is greater in the initial disclosure year relative to the predisclosure year. Specifically, in panel 1 of Table 5, both the direct test statistic (ZSCVR) and the nonparametric test statistics Show significant differences in SCVRs between the initial disclsoure year and the prior non- disclosure year in all days but day {-2} of the ZSCVR test. Similar results are evident for the three day cumulation period in panel 2 of Table 5, with the exception of day {0} of the zSCVR test. Appendix E provides summary statistics for the SVR and SCVR measures in Tables 4 and 5. Two aspects of these statistics are worthy of notice. First, the notable difference between mean and median values of the SVR and SCVR measures suggests the violation of the normality assumption. The higher mean value indicates that the distribution is skewed to the left. The Kblmogorov-Smirnov goodness of fit test (not shown) also suggests that the normality assumption is violated in almost all stock returns variability measures (e.g., Kolmogorov-Smirnov ZS and P-values are 5.63(0.00), 39 6.24(0.00), and 6.8l(0.00) at days {-1}, {0}, and {+1}, respectively for the initial disclosure year and 6.30(0.00), 7.47(0.00), and 6.33(0.00) for the predisclosure year). The evidence of nonnormality of the distribution of daily residuals is consistent with the evidence of Brown and Warner [1985]. Dyckman, Philbrick, and Stephan [1984, p.27] indicate that nonnormality tends to increase the likelihood of rejecting the null hypothesis when it is true. Given the large sample size, this lack of normality may not be a serious problem. However, greater emphasis will be given to the nonparametric tests in this study. Second, there is a slight tendency for extreme values to affect the results of the prior year more than those of the initial disclosure year, thus increasing the bias toward the null hypothesis. Table 6 provides evidence of increased stock price variability in the disclosure year relative to the prior year based on the use of ratio-based tests. The results for both the Standardized Abnormal Return Ratio (SARR) and the Standardized Variance of Returns Ratio (SVRR) are consistent with HA1. Specifically, all days show highly significant stock price reactions in the initial disclosure year vis-a- vis the prior year. More importantly, these results reinforce the daily SVR evidence by eliminating the possibility of the abnormally deflated estimation-period variances related to daily SVR measures in the initial disclosure year. Data are truncated at 10.0 because of some near zero values of abnormal' returns in the prior year.21 The results from the parametric t-test are only presented in 40 Table 6 because two different sets of data between the initial disclosure year and the prior year are combined into one set of data (ratios), and thus there is no need to compare one with the other as in the nonparametric tests. The empirical results in Tables 3-6 provide evidence of increased stock price variability relating to ASR 190 around the release of the lo-K report on the first mandatory disclosure year relative to the prior year. The results from daily, cumulative, and ratio metrics are all consistent with the increased stock price variability hypothesis, and rejects the null hypothesis of no information content for inflation- adjusted data. However, as indicated in Footnote 17, the serial independence assumption might have been violated. So, the validity of the results based on the direct test statistic (ZSCVR) should be qualified (see Section 4.2.1). Appendix F presents daily stock returns variability and its statistical comparison between the predisclosure year and the initial disclosure year for the extended 21-day event period. As one might suspect, the average SVR (0.90) for the first 7 days ({-10} through {-3}) in the predisclosure year is significantly higher than those of the three-day event period (0.75) and five-day event period (0.77). This may be due to the effects of other special events (e.g., earnings announcements, :management. and financial analysts' earnings forecasts, etc.) in the earlier period. However, this phenomenon is not pronounced in the initial disclosure year in which the average SVR for the same first 7 days is 0.85, 41 and the corresponding figures for the three-day and five-day event periods are 0.90 8 0.88, respectively. More importantly, the mean difference between the predisclosure year and the initial disclosure year shows the higher, positive deviation for the three-day event period (0.15) and the five-day event period (0.11) than for the first seven-day period (-0.05). Therefore, the mean difference can be considered consistent with the initial findings in Table 4 through Table 6. 5.1-2 W 5.1.2.1 Qtnet Eggnonncensnts Atonng tns Qisclosnte XQQI The increase in stock returns variability in the initial disclosure year has been observed in Section 5.1.1. However, we know that in the year prior to the lo-K disclosure of current cost data, several FASB standards also became effective. Thus, it is important to investigate whether the effects that I have observed are due to the current cost disclosures. and. not. other’ changes in. the reporting environment. Table 7 shows a listing of pronouncements made by the Financial Accounting Standards Board around the effective date of ASR 190 (December 25, 1976). In fact, no accounting series release of the SEC is relevant to the type of sample examined in this study in determining the potential confounding effect, and thus only the pronouncements of the FASB are presented in this table. As shown in Table 7, the two most probable events that could possibly have obscured the results of the increased stock returns variability in the previous tables are SFAS No. 42 TABLE 7 Effective Dates of FASB Pronouncements for 1976-1977 SFAS No. Title Effective Date 7. 13. 14. 15. 16. 18. Accounting and Reporting by Development Stage Enterprise ......... ..... . 1/1/1976 Accounting for the Translation of Foreign Currency Transactions and Foreign Currenecy Financial Statement ...... 1/1/1976 Accounting for Leases ............ ........... 1/1/1977 Financial Reporting for Segments of a Business Enterprise ..................... 12/15/1976 Accounting by Debtors and Creditors for Troubled Debt Restructurings ............... 12/31/1977 Prior Period Adjustments .... ..... ........... 10/15/1977 Financial Reporting for Segments of a Business Enterprise - Interim Financial Statements .. 12/1/1977 43 13 (Accounting for Leases) and SEAS No. 14 (Financial Reporting for Segments of a Business Enterprise). However, the substantial amount of inflation-adjusted information was first given in the lo-K reports, not in the annual reports, and the evidence from prior research by Foster and Vickrey [1978, p.925] shows that the lO-K filing date is preceded by the initial public release date of the annual report for their sample firms (96 firms). Thus, it is less likely that these events confounded the basic results contained in Tables 4-6. The earlier release in the annual report might very well mitigate the confounding problem because the incremental information contained in the annual report (i.e., incremental over earlier earnings announcements, etc.) could generate an information effect and could cause stock returns variability to rise at the time of release (Foster and Vickrey [1978, p.925]). 5.1.2.2 Eris; gsnr vs, Post-Qisslgsure stt To check whether the previous section's results (Section 5.1.1) for the information content of inflation-adjusted data are due to an artifact of the research design or other unknown confounding factors, another within-firm comparison is made between the prior non-disclosure year and the second year of disclosure (post-disclosure year). Although Ohlson's theoretical predictions for the change in information environments equally apply to any future filing dates of the lo-K report, the second year of disclosure is chosen because ASR 190 was repealed shortly afterwards (i.e., when SFAS No. 33 became effective in 1979). The distribution of the lo-K 44 filing dates for the post-disclosure year is shown in Appendix G. The results in Table 8 are consistent with and even more significant than the earlier test results, and provide further evidence of Changes in information environments for the prior vs. initial disclosure year comparison. Specifically, the mean SVR across event days for the post- disclosure year experienced a more than 43%, on average, increase when compared to the prior year mean SVR. Moreover, the pattern of distribution of paired t statistics across event. days is the same Ias that. of Table 4 (significant results at days {-2}, {-1}, and {+1}) while the pattern of the distribution of the nonparametric test statistics shows even more pronounced differences across two years relative to the one year in Table 4. That is, in all 5 event days, the null hypothesis of no change in the information environment is rejected at the 1% significance level in most instances. This finding between the prior non-disclosure year and the post-disclosure year is reinforced by the SCVR evidence in Table 9. In both the direct test statistic (ZSCVR) and the nonparametric statistics, the null hypothesis of no information content of the inflation-adjusted data is rejected at the 1% significance level for all 5 event days. For this and following sections, the results from the ratio- test metrics are omitted simply because their results are more significant, and I want to follow the conservatism convention in drawing conclusions. 45 TABLE 8 Daily Stock Returns Variability Tests Across PRIOR and POST Yearsa Erier_xeer Beet_Year tennerieen_12:xelnelb Wilcoxon Event SVR SVR Student Sign Test Signed-Rank Day Mean Mean t z z *** *** *** -2 0.66 1.13 3.46(0.00) 3.56(0.00) 4.06(0.00) 44* 44* 44* -1 0.61 1.01 3.64(0.00) 3.05(0.00) 4.39(0.00) *** *** 0 0.86 1.00 0.74(0.23) 2.34(0.01) 2.50(0.01) *** *** *** +1 0.75 1.19 2.83(0.00) 2.44(0.01) 2.85(0.00) ** *** +2 0.94 0.99 0.25(0.40) l.83(0.03) 2.35(0.01) *** *** *** MeanC 0.77 1.06 3.74(0.00) 2.85(0.00) 4.63(0.00) a SVR is defined in equation <1>. upon 387 instead of 407 firms due to missing or illegible dates on the microfiches of the 10-K reports. The results are based The PRIOR and POST designations refer to inflation-adjusted information releases through the 10-K report for the year before (prior year) and the year after (post-disclosure year) the issuance of ASR No. 190 by SEC. of the lo-K by the SEC. Event day {0} is the date of receipt b A paired comparison is made by taking the difference in SVRs between the post-disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. . 0 Mean SVR is computed by taking the average of five day SVRs for each firm and then determining the average of all firms. A similar measure (the cumulative U) is used on p. 73 of McNichols and Manegold [1983]. By the nature of construction, this value should be equal to the SCVR value at day {2}. 46 TABLE 9 Cumulative Stock Returns Variability Tests Across PRIOR and POST Yearsa Erier_Yeer Beet_xeer Qennn:ieen.l£:xeluelb Wilcoxon Event SCVR SCVR 2 Sign Test Signed-Rank SCVR Day Mean Mean z 2 Panel 1: Cumulation Period of Day {-2} through Day {+2} *** *** *** -2 0.66 1.13 4.53(0.00) 3.56(0.00) 4.06(0.00) A** *** *** -1 0.64 1.07 5.95(0.00) 4.37(0.00) 5.25(0.00) *** *** *** 0 0.71 1.04 5.61(0.00) 3.56(0.00) 4.27(0.00) *** *** *** +1 0.72 1.08 6.99(0.00) 4.88(0.00) 4.97(0.00) *** *** *** +2 0.77 1.06 6.47(0.00) 2.85(0.00) 4.63(0.00) Panel 2: Cumulation Period of Day {-1} through Day {+1} *** *** *** -1 0.61 1.01 3.88(0.00) 3.05(0.00) 4.39(0.00) *** *** *** 0 0.74 1.00 3.67(0.00) 2.75(0.00) 3.68(0.00) *** *** *** +1 0.74 1.06 5.46(0.00) 3.05(0.00) 4.16(0.00) SCVR and z are defined in equation <2> and <3>, respectively. TEXRresults are based upon 387 instead of 407 firms due to missing or illegible dates on the microfiches of the lO-K reports. The PRIOR and POST designations refer to inflation-adjusted information releases through the lo-K report for the year before (prior year) and the year after (post disclosure year) the issuance of ASR N0. 190 by SEC. Event day {0} is the date of receipt of the 10-K by the SEC. b The results for the nonparametric tests are based on the mean rank and the number of cases where SCVRi t for the initial disclosure year is greater than SCVRi' for the prior non-disclosure year. The symbols of *, **, anfi *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 47 5.1.2.3 n' ' 'sc o u e 0 -Di c os Unless other FASB Standards affect the stock returns variability in the post-disclosure year, there is no reason to believe that the disclosure of inflation-adjusted data in the lO-K report leads to a change in information environments between the initial disclosure year and the post-disclosure year. Table 10 presents evidence for the increased stock returns variability in the post-disclosure year relative to the initial disclosure year. The level of significance has become less pronounced in this comparison than in the comparison between the prior year and the post-disclosure year, but a little bit more stronger than in the comparison between the prior’ year and the initial disclosure year. However, this evidence will no longer be a threat to the previous findings if we know that three other FASB standards also became effective in the post-disclosure year (e.g., SFAS Nos. 15, 16, and 18 in Table 7). The usefulness of the comparison between the prior year and the post-disclosure year (Section 5.1.2.2) is, nonetheless, greatly reduced by the increased stock returns variability in the post- disclosure year relative to the initial disclosure year. 5.1-2-4 Wen Even though our within-firm comparison methodology makes unnecessary the use of a comparison group, it may be reassuring if no evidence of increased variability can be documented in the comparison group that was not required to disclose inflation-adjusted data. The failure to reject the null hypothesis of no changes in information environments for 48 TABLE 10 Daily Stock Returns Variability Tests Across INITIAL DISCLOSURE and POST Yearsa Initial Dieeleenre Beet_Yeer temperieen_lzzxelnelb Wilcoxon Event SVR SVR Student Sign Test Signed-Rank Day Mean Mean t z z 44* *4 -2 0.78 1.13 2.54(0.01) 1.22(0.11) 2.05(0.02) ** *‘k -1 0.81 1.01 1.78(0.04) 1.02(0.15) 2.04(0.02) 0 0.92 1.00 0.56(0.29) -0.51(0.73) 0.57(0.29) 4 * * +1 0.95 1.19 1.38(0.08) l.42(0.08) 1.45(0.07) +2 0.93 0.99 0.42(0.34) 0.41(0.34) 0.46(0.32) *** * *** Meanc 0.87 1.06 2.40(0.01) 1.42(0.08) 2.76(0.00) a SVR is defined in equation <1>. The results are based upon 387 instead of 407 firms due to missing or illegible dates on the microfiches of the lO-K reports. The INITIAL DISCLOSURE and POST designations refer to inflation-adjusted information releases through the lo-K report for the year of (initial disclosure year) and the year after (post-disclosure year) the issuance of ASR No. 190 by SEC. Event day {0} is the date of receipt of the 10-K by the SEC. b A paired comparison is made by taking the difference in SVRs between the post-disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. c Mean SVR is computed by taking the average of five day SVRs for each firm and then determining the average of all firms. A similar measure (the cumulative U) is used on p. 73 of McNichols and Manegold [1983]. By the nature of construction, this value should be equal to the SCVR value at day {2}. 49 the comparison group of firms should increase confidence in the previous section's results for the information content of the inflation-adjusted data. Therefore, the comparison between the results of the experimental group and those of the comparison group will provide us with an opportunity to examine whether the increased stock returns variability in the initial disclosure year is due to other Changes in the reporting environment (other FASB standards and SEC ASRs). For this purpose, one hundred firms whose inventories and productive assets aggregate less than $100 million but more than $50 million at the beginning of the initial disclosure year are selected from the industrial COMPUSTAT tape. The selection of the comparison group of firms is made bearing in mind the goal of the minimization of the possible confounding size effect. The summary characteristics of the comparison group of firms are presented in Appendices H and I; Table 11 indicates ‘that 'the increased stock returns variability is observed only at day {-2} under the nonparametric tests and days {—2} and {0} under the parametric t-test. Except for day {-2}, there is no common feature between Table 4 (experimental group) and Table 11 (comparison group). Specifically, the experimental group additionally showed significant reaction on days {-1}, {0}, and {+2} with the nonparametric tests and days {-1} and {+1} with the parametric t-test, and also the level of significance is much higher in the experimental group. The different behavior between the experimental group and the comparison group is further investigated in Table 12 in which 50 TABLE 11 Daily Stock Returns Variability Tests for the Comparison Groupa Erier_Year A§B_129_Yeer temperieen_lzzxelnelb Wilcoxon Event SVR SVR Student Sign Test Signed-Rank Day Mean Mean t Z 2 4* + ** -2 0.62 1.07 1.90(0.03) 1.50(0.07) 1.71(0.04) -1 0.69 0.73 0.21(0.42) -0.40(0.69) -0.16(0.57) ** 0 0.65 1.04 1.67(0.05) 0.50(0.31) 1.00(0.16) +1 0.83 0.72 -o.52(0.70) 0.00(0.54) 0.02(0.49) +2 0.65 0.90 1.19(0.12) 0.60(0.27) 1.20(0.12) ** * Meanc 0.69 0.89 1.81(0.04) 0.30(0.38) l.56(0.06) a SVR is defined in equation <1>. The PRIOR and ASR 190 designations refer to inflation-adjusted information releases through the lO-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. Event day {0} is the date of receipt of the lO-K by the SEC. b A paired comparison is made by taking the difference in SVRs between the disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. c Mean SVR is computed by taking the average of five day SVRs for each firm and then determining the average of all firms. A similar measure (the cumulative U) is used on p. 73 of McNichols and Manegold [1983]. By the nature of construction, this value should be equal to the SCVR value at day {2}. 51 the analysis of the SCVR across the 5-day and 3-day event periods. Panel 1 of Table 12 can be comparable to that of Table 5. Even though the comparison group showed significant stock price reaction on days {-2}, {0}, {+1} and {+2}, under the Wilcoxon Signed-Rank test, the levels of significance cannot be comparable to those of the experimental group. It seems to me that the significant stock price reaction on days other than day {-2} is driven by day {-2]. Also, unlike Panel 1 of Table 5, the Sign test results in Panel 1 of Table 12 are much weaker (i.e., marginally significant on day {-1} only). The results based on ZSCVR seem to be inconsistent with the null hypothesis of no change in information environments between the prior year and the initial disclosure year for the comparison group. The increased stock price variability has been observed in all event days at the five or ten percent significance level. However, as indicated in Section 4.2.1, less emphasis should be placed on this metric in interpreting the results due to the possible violation of the serial independence assumption. On the contrary, Panel 2 of Table 12 provides evidence consistent with the null hypothesis of no change in information environments between the prior year and the first ASR 190 year for the comparison group. Neither the ZSCVR statistic nor the nonparametric test results show significant differences in mean SCVRs in any of the three event days under the five percent significance level. Considering the results from the above sensitivity analyses, it is unlikely that the rejection of the null 52 TABLE 12 Cumulative Stock Returns Variability Tests for the Comparison Group Exist_1sg; ASR 19g Ysat annatison (P-valus)b Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean Z z Panel 1: Cumulation Period of Day {-2} through Day {+2} ** ** * -2 0.62 1.07 2.19(0.02) 1.50(0.07) 1.71(0.04) ** -1 0.66 0.90 1.69(0.05) 0.70(0.24) 0.98(0.16) *** *‘k 0 0.66 0.95 2.50(0.01) 0.70(0.24) 1.79(0.04) ** * +1 0.70 0.89 1.90(0.03) 0.30(0.38) 1.33(0.09) *** * +2 0.69 0.89 2.25(0.01) 0.30(0.38) 1.56(0.06) Panel 2: Cumulative Period of Day {-1} through Day {+1} -1 0.69 0.73 0.20(0.42) -0.30(0.69) -0.17(0.57) * 0 0.67 0.89 1.51(0.07) 0.10(0.46) 0.95(0.17) +1 0.73 0.83 0.92(0.18) 0.00(0.50) 0.73(0.23) SCVR and Z are defined in equation <2> and <3>, respectively. TEXRPRIOR and ASR 190 designations refer to inflation-adjusted information releases through the 10-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. Event day {0} is the date of receipt of the lo-K by the SEC. b The results for the nonparametric tests are based on the mean rank and the number of cases where SCVRi,t for the initial disclosure year is greater than SCVRi,t for the prior non-disclosure year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests.event days under the 5% significance level. 53 hypothesis of no information content of the inflation- adjusted data in the previous tables is due to an artifact of the research design or other changes in the reporting environment. 5-1-2-5 MW As indicated by Dyckman, Philbrick, and Stephan [1984, p. 29], event-date clustering by time may reduce the power of statistical tests in detecting abnormal stock price behavior. To check the validity of this argument, I divided my sample into two groups: December Fiscal Year-End and Non-December Fiscal Year-End. Since most of firms in the December Fiscal Year—End group filed their lO-K reports on 3/30 and 3/31 in the initial disclosure year, I also used another sample- division strategy in which the firms in my sample were divided into DEC30, DEC31, and OTHERS groups. If the event- date Clustering effect is pronounced, then we should observe more significant stock price effect in the Non-December Fiscal Year-End or Other Filing-Date group. Tables 13, 14, and 15 present the test results of difference in SVR, SCVR(S-day), and SCVR(3-day) between initial disclosure year and the prior year in both the December Fiscal Year-End group and the Non-December Fiscal Year-End group. In the SVR case of Table 13, the December Fiscal Year-End firms show significantly increased stock price reaction at days {-2}, {-1}, and {+2} of the initial disclosure year while the Non-December Fiscal Year-End firms show it only at day {0} under the nonparametric tests. In the analysis of the SCVR across 5-day (Table 14) and 3-day 54 TABLE 13 December Fiscal Year-End Vs.Non-December Fiscal Year-End Standardized Variance of Returnsa Erier_Year A§B_120_Yeer tennerieen_1£:xalnelb Wilcoxon Event SVR SVR Student Sign Test Signed-Rank Day Mean Mean t Z Z Panel 1: December Fiscal Year-End Firms (282 firms) * *** *** -2 0.67 0.83 1.48(0.07) 2.80(0.00) 2.57(0.01) * *** *** -1 0.64 0.80 1.39(0.08) 3.39(0.00) 3.50(0.00) 0 0.79 0.81 0.10(0.46) 0.54(0.30) 0.99(0.l6) ** +1 0.70 0.98 l.64(0.05) -0.06(0.57) 0.81(0.21) * ** +2 0.71 0.85 0.94(0.17) 1.37(0.09) 1.81(0.04) ** *** *** Mean 0.70 0.85 2.05(0.02) 2.80(0.00) 2.48(0.01) Panel 2: Non-December Fiscal Year-End Firms (124 firms) -2 0.67 0.68 0.08(0.47) 0.45(0.33) 0.32(0.38) -1 0.65 0.80 0.93(0.l8) 0.09(0.46) 0.34(0.37) ' ** ** 0 1.03 1.15 0.3l(0.38) 2.07(0.02) l.88(0.03) +1 0.94 1.00 0.24(0.41) 0.00(0.50) 0.41(0.34) +2 1.37 1.02 -0.68(0.75) 0.45(0.33) 0.75(0.23) Mean 0.93 0.93 0.00(0.50) 0.27(0.39) 0.68(0.25) 55 TABLE 13 (cont'd) a The mean (median) market value of December Fiscal Year- End firms is 819.3 (305.3) million dollars while the mean (median) market value of Non-December Fiscal Year-End firms is 475.6 (179.0) million dollars. b The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 56 TABLE 14 December Fiscal Year-End Vs.Non-December Fiscal Year-End Standardized Cumulative Variance of Returns (5-day)a Kim}: W: W'80 -va ueb Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean Z Z Panel 1: December Fiscal Year-End Firms (282 firms) * *** *** -2 0.67 0.83 1.33(0.09) 2.80(0.00) 2.57(0.01) ** *** *** -1 0.65 0.82 1.90(0.03) 2.92(0.00) 3.22(0.00) ** *** *** 0 0.70 0.81 1.62(0.05) 2.56(0.01) 2.36(0.01) *** *** *** +1 0.70 0.85 2.56(0.01) 3.28(0.00) 2.39(0.01) *** *** *** +2 0.70 0.85 2.81(0.00) 2.80(0.00) 2.48(0.01) Panel 2: Non-December Fiscal Year-End Firms (124 firms) -2 0.67 0.68 0.07(0.47) 0.45(0.33) 0.32(0.38) -1 0.66 0.74 0.65(0.26) 0.45(0.33) 0.35(0.36) 0 0.78 0.88 0.92(0.18) 1.17(0.12) 1.25(0.11) * * +1 0.82 0.91 0.96(0.17) 1.35(0.09) 1.28(0.10) +2 0.93 0.93 -0.01(0.50) 0.27(0.39) 0.68(0.25) a See Footnote a in Table 13. b See Footnote b in Table 13. 57 TABLE 15 December Fiscal Year—End Vs.Non-December Fiscal Year-End Standardized Cumulative Variance of Returns (3-day)a WW Was -v ueb Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean 2 Z Panel 1: December Fiscal Year-End Firms (282 firms) * *** *** -1 0.64 0.80 1.36(0.09) 3.45(0.00) 3.50(0.00) *** ** 0 0.72 0.81 1.05(0.15) 2.98(0.00) 2.18(0.02) *** *** *** +1 0.71 0.86 2.19(0.01) 2.74(0.00) 2.39(0.01) Panel 2: Non-December Fiscal Year-End Firms (124 firms) -1 0.65 0.80 0.85(0.20) 0.18(0.46) 0.34(0.37) * 0 0.84 0.98 l.08(0.l4) 1.26(0.12) 1.44(0.08) * +1 0.87 0.98 1.07(0.14) 0.90(0.21) 1.34(0.09) a See Footnote a in Table 13. b See Footnote b in Table 13. 58 (Table 15) event periods, the December Fiscal Year-End group shows the increased stock price reaction at all event days of the initial disclosure year, but the Non-December Fiscal Year-End group shows it at day {+1} in the 5-day event period and days {0} and {+1} in the 3-day event period under the nonparametric tests. These findings support the null hypothesis of no effect from. event date clustering. The results based on parametric tests are even more supportive of the trivial effect hypothesis of event date clustering because no significant stock price reaction is documented for Non-December Fiscal Year-End group of firms in, Table 13 through Table 15. Table 16 repeated the analysis of the SCVR across the 3- day event period for both the December Fiscal Year-End and the lNon-December' Fiscal Year-End firms in the comparison group. As we expected, neither group shows'any significantly different stock. price behavior in the initial disclosure year. The results from the classification of sample into DEC30, DEC31, and OTHERS filing date groups are presented in Tables 17 (SVR), 18 (SCVR,5-day) and 19 (SCVR, 3-day). DEC31 group shows the most significant Changes in stock returns variability in the initial disclosure year. This phenomenon becomes less pronounced when it comes to OTHERS group and even smaller for the DEC30 group. Again, these findings are inconsistent with the hypothesis of a significant effect from event date clustering. 59 . TABLE 16 December Fiscal Year-End Vs.Non-December Fiscal Year-End Comparison Groupa Brier_Yeer A§B_129_Year temperieen_12:xalnel Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank- SCVR Day Mean . Mean Z Z Panel 1: December Fiscal Year-End Firms (41 firms) -1 0.75 1.00 0.80(0.21) -0.31(0.73) -0.20(0.58) o 0.67 0.91 1.05(0.15) -0.62(0.83) 0.05(0.48) +1 0.81 0.88 0.38(0.35) -0.31(0.73) -0.30(O.62) Panel 2: Non-December Fiscal Year-End Firms (59 firms) -1 0.66 0.55 -0.4l(0.66) -0.13(0.60) -0.20(O.58) 0 0.67 0.87 1.09(0.14) 0.78(0.22) 1.20(0.12) +1 0.67 0.80 0.89(0.l9) 0.26(0.40) 1.19(0.12) SCVR and Z are defined in equation <2> and <3>, respectively. IggRPRIOR and ASR 190 designations refer to inflation-adjusted information releases through the lO-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. Event day {0} is the date of receipt of the lo-K by the SEC. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 60 TABLE 17 MARCH30, MARCH31, OTHER Filing-Date Firms Standardized Variance of Returns Erier_Year A§B_122_Year tentatieen_12:xnlnela Wilcoxon Event SVR SVR Student Sign Test Signed-Rank Day Mean Mean t Z Z Panel 1: March 30 Filing-Date Firms (53 firms) 44 -2 0.86 1.01 0.51(0.31) 1.92(0.03) 1.19(0.12) -1 0.53 0.78 1.26(0.ll) 0.82(0.20) 0.94(0.17) 0 0.50 0.83 0.85(0.20) 0.82(0.20) 0.58(0.28) +1 0.56 0.96 1.10(0.14) 0.55(0.29) 1.08(0.14) +2 0.49 0.57 0.30(0.38) 0.27(0.39) 0.85(0.20) 4 4 Mean 0.59 0.83 1.29(0.10) 0.82(0.20) 1.33(0.09) iiirms) * * -2 0.60 0.67 0.56(0.29) 1.36(0.09) 1.45(0.07) *** *** -1 0.64 0.75 O.58(0.28) 3.53(0.00) 3.49(0.00) ** 0 0.75 0.94 0.81(0.21) 0.81(0.21) 1.75(0.04) +1 0.81 0.96 0.47(0.32) -0.81(0.84) -0.75(0.77) +2 0.78 0.78 0.03(0.49) 0.45(0.33) 0.92(0.18) . ** * Mean 0.72 0.82 0.86(0.20) 2.08(0.02) 1.27(0.10) 'mirms) 4 4 -2 0.66 0.79 1.04(0.15) 1 45(0.07) 1.42(0.08) * -1 0.67 0.84 1.29(0.10) 0.79(0.21) 1.15(0.13) 0 1.01 0.92 -0.39(0.65) 1.05(0.15) 1.02(0.16) +1 0.80 1.00 1.23(0.11) 0.13(0.45) 1.22(0.11) * * +2 1.08 1.04 -0.14(0.56) 1.32(0.09) 1.46(0.07) 4 44 Mean 0.85 0.92 0.75(0.23) 1.32(0.09) 1.67(0.05) a The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 61 TABLE 18 MARCH30, MARCH31, OTHER Filing-Date Firms Standardized Cumulative Variance of Returns (5-day) EIIQI_X§§I A§B_122_X§§I QQEEQI1§2£_i£:¥§lE§1a Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean 2 Z Panel 1: March 30 Filing-Date Firms (53 firms) ** -2 0.86 1.01 0.53(0.30) 1.92(0.03) 1.19(0.12) 4 -1 0.69 0.89 1.02(0.15) 0.82(0.20) 1.40(0.08) 4 4 0 0.63 0.87 1.53(0.06) 1.10(0.14) 1.48(0.07) 44 4 +1 0.61 0.89 2.05(0.02) 1.10(0.14) 1.35(0.09) ** * +2 0.59 0.83 1.96(0.03) 0.82(0.20) 1.33(0.09) Panel 2: March 31 Filing-Date Firms (122 firms) * * -2 0.60 0.67 0.37(0.36) 1.36(0.09) 1.45(0.07) 444 444 -1 0.62 0.71 0.68(0.25) 2.81(0.00) 2.54(0.01) 444 444 0 0.66 0.78 1.15(0.13) 2.63(0.00) 2.52(0.01) 4 444 44 +1 0.70 0.83 l.40(0.08) 2.81(0.00) 1.94(0.03) 4 44 4 +2 0.72 0.82 1.27(0.10) 2.08(0.02) 1.27(0.10) Panel 3: Other Filing-Date Firms (231 firms) 4 4 -2 0.66 0.79 0.99(0.16) 1.45(0.07) 1.42(0.08) 4 4 -1 0.67 0.82 1.60(0.06) 1.05(0.15) 1.33(0.09) 0 0.78 0.85 0.90(0.18) 1.18(0.12) 1.04(0.15) 4 44 4 +1 0.79 0.89 1.53(0.06) 1.97(0.02) l.59(0.06) 4 44 +2 0.85 0.92 1.23(0.11) 1.32(0.09) 1.67(0.05) a See Footnote a in Table 17. 62 TABLE 19 MARCH30, MARCH31, OTHER Filing-Date Firms Standardized Cumulative variance of Returns (3-day) Erier_Yeer 183.129.1eer tennerieen.l£:xelnela Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean Z Z Panel 1: March 30 Filing-Date Firms (53 firms) -1 0.53 0.78 0.92(0.18) 0.82(0.20) 0.94(0.17) * 0 0.51 0.81 1.50(0.07) 0.00(0.50) 0.48(0.32) ** +1 0.53 0.86 2.06(0.02) 0.82(0.20) l.06(0.15) Panel 2: March 31 Filing-Date Firms (122 firms) *** *** -1 0.64 0.75 0.59(0.28) 3.53(0.00) 3.49(0.00) *** *** 0 0.69 0.84 1.15(0.13) 3.35(0.00) 2.85(0.00) * *** *** +1 0.73 0.88 l.41(0.08) 2.63(0.00) 2.22(0.01) Panel 2: March 31 Filing-Date Firms (231 firms) * -1 0.67 0.84 1.27(0.10) 0.79(0.21) 1.15(0.13) * 0 0.84 0.88 0.41(0.34) 1.54(0.06) 1.15(0.13) * +1 0.83 0.92 1.20(0.12) 1.18(0.12) l.59(0.06) a See Footnote a in Table 17. 63 5.2 st 0 H t es 2 In this section, the hypothesis of differential stock price reaction between small and large firms is tested. Previous research (Atiase [1985], Grant [1980], etc.) provides evidence of’ more "informative stock. prices" for large firms 'vis-a-vis small firms. 'The range of firms' dollar market values for the small firms in this study is $ 12.77 - $142.21 million, while that of Atiase's [1985] sample is $ 1.80 - $19.98 million. Grant [1980] used OTC firms as a proxy for small firms, and thus the mean market value of his small firms is even less than that of Atiase. This suggests that the size effect due to differential amounts of available information between large firms and small firms may not be as pronounced in this study's sample as in previous research. Daily stock returns variability is examined after firm- size is controlled in Table 20. The 390 out of 407 firms have a complete set of data for size and ratio analyses in the industrial COMPUSTAT ‘tape. .A firm is classified as ”small" if 12.77 < Market Value (MV) < 142.21, Mean MV = 70.79 in million $, and Median MV = 65.93 in million s, as “nedium” if 142.48 < MV < 516.70, Mean MV = 271.74 in million $, and Median MV = 234.34 in million $, and as "1stgs" if 524.00 < MV < 17,123, Mean MV = 1,798.0 in million $, and Median MV = 988.00 in million $. The results in Table 20 show that small firms' average SVR (0.92) is substantially larger than that of large firms (0.64) in the predisclosure year. This evidence is consistent with that in prior research. The results for the initial disclosure year are A /_u 64 TABLE 20 Size Effects -- Three Portfoliosa W W -v b Wilcoxon SVR SVR Student Signed-Rank Day Mean Mean t z Panel A: Small Firms (130 firms) -2 0.56 0.67 0.91(0.18) 0.92(0.18) 4 -1 0.72 0.67 -0.28(0.6l) 1.28(0.10) 0 1.32 0.77 -1.44(0.92) -0.94(0.83) +1 0.95 0.82 -0.65(0.74) -1.04(0.85) :2, 1.06 0.94 -0.26(0160) 019910.16) Mean 0.92 0.77 -1.05(0.85) 0.36(0.36) Panel B: Medium Firms (130 firms) 4 -2 0.81 0.90 0.46(0.32) 1.31(0.09) 44 444 -1 0.60 0.94 1.96(0.03) 2.61(0.00) 4 44 0 0.73 1.04 1.27(0.10) 1.96(0.03) +1 0.83 1.24 1.18(0.12) 1.25(0.11) :2 0.89 0.85 -0.16(0.56) 0.69(0.25) 44 Mean 0.77 0.99 1.7110.05) 1.25(0.11) Panel C: Large Firms (130 firms) 4 44 -2 0.60 0.79 1.49(0.07) l.86(0.03) -1 0.63 0.75 0.77(0.22) 0.78(0.22) 44 44 0 0.59 0.94 1.80(0.04) 2.00(0.02) ** ** +1 0.52 0.91 2.10(0.02) 1.85(0.03) 4 44 444 Mean 0.64 0.85 2.12(0.02) 2.46(0.01) 65 TABLE 20 (cont'd.) a The PRIOR and ASR 190 designations refer to inflation- adjusted information releases through the lO-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. SVR is defined in equation <1>. b A paired comparison is made by taking the difference in SVRs between the initial disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 66 mixed (i.e., average SVRs of small, medium, and large firms are 0.77, 0.99, and 0.85, respectively). When it comes to the within-firm comparison of SVRs between the initial disclosure year and the prior year, HA2 is not supported by the evidence in Table 20. The difference in price variability between the predisclosure year and the disclosure year is significant for large and medium firms, but not for small firms. For the medium and large firms, the 3-day average SVR for the ASR 190 year is significantly larger than that for the predisclosure year. For example, average SVRs between the disclosure year and the prior year are 0.99 and 0.77 for the medium size group, and 0.85 and 0.64 for the large size group, respectively. Tables 21 and 22 present results from the SCVR (5-day) and the SCVR (3-day) analyses based on firm-size, respec- tively. As in the SVR case of Table 20, only medium and large firms experienced significantly increased stock returns variability at all event days except day {-1} in the initial disclosure year under the nonparametric test. The test for a size effect is also conducted using 5 portfolios. These results are similar to the 3-portfolio-case, and thus, they are presented in Appendix J. The results from the within—firm comparison are consistent with several hypotheses. First, the "quantity" of the required disclosures made in the initial ASR 190 year by large firms may be significantly greater than that by small firms. This possibility is examined in conjunction with the test for HA5 (see Section 5.5). Second, market participants 67 TABLE 21 Size Effects -- Three Portfolios (SCVR, 5-day) Erier_Yeer A§B_129_Yeer tennerieen_1£:xelnela Wilcoxon Event SCVR SCVR ZSCVR Sign Test Signed-Rank Day Mean Mean Z Z Panel 1: Small Firms (130 firms) -2 0.56 0.67 0.61(0.27) 0.26(0.40) 0.92(0.18) * -1 0.64 0.67 0.23(0.41) 1.32(0.09) 1.24(0.11) 0 0.87 0.70 -l.60(0.95) 0.79(0.22) 0.24(0.41) +1 0.89 0.73 -l.74(0.96) 0.61(0.27) 0.40(0.35) +2 0.92 0.77 -1.88(0.97) 0.44(0.33) 0.36(0.36) Wing *** * —2 0.81 0.90 0.53(0.24) 2.19(0.01) 1.31(0.10) ** ** ** -1 0.70 0.92 1.73(0.04) 2.02(0.02) 2.08(0.02) *** ** ** 0 0.71 0.96 2.41(0.01) 1.84(0.03) 1.92(0.03) *** ** * +1 0.74 1.03 3.23(0.00) 1.67(0.05) l.41(0.08) *** +2 0.77 0.99 2.80(0.00) 0.96(0.17) 1.25(0.11) firms) 44 44 -2 0.60 0.79 1.07(0.14) 1.67(0.05) l.86(0.03) * -1 0.61 0.77 1.24(0.11) 0.97(0.17) 1.35(0.09) 44 44 44 0 0.60 0.83 2.16(0.02) 1.67(0.05) 2.02(0.02) *** *** *** +1 0.58 0.85 2.98(0.00) 3.59(0.00) 2.79(0.00) *** *** *** +2 0.64 0.85 2.64(0.00) 2.72(0.00) 2.46(0.01) a See Footnote b in Table 20. 68 TABLE 22 Size Effects -- Three Portfolios (SCVR, 3-day) Erier_Year ASB.129.Yeer temperieen_lzzxelnela Wilcoxon Event SCVR SCVR 2 Sign Test Signed-Rank SCVR Day Mean Mean Z Z Panel 1: Small Firms (130 firms) 44 4 -1 0.72 0.67 -0.29(0.61) 1.84(0.03) 1.28(0.10) 0 1.02 0.72 -2.40(0.99) 0.26(0.40) -0.03(0.51) +1 1.00 0.75 -2.36(0.99) 0.79(0.22) 0.27(0.40) Panel 2: Medium Firms (130 firms) 44 44 444 -1 0.60 0.94 1.92(0.03) 1.84(0.03) 2.61(0.01) 444 444 444 0 0.66 0.99 2.57(0.01) 2.54(0.01) 2.31(0.01) 444 4 44 +1 0.72 1.07 3.43(0.00) 1.32(0.09) 1.72(0.04) Panel 3: Large Firms (130 firms) -1 0.62 0.75 0.68(0.25) 0.79(0.22) 0.78(0.22) 44 44 4 0 0.61 0.84 1.90(0.03) 1.84(0.03) 1.53(0.06) 444 444 444 +1 0.58 0.86 2.82(0.00) 2.37(0.01) 2.50(0.01) a See Footnote b in Table 20. 69 may have less confidence in the validity and reliability of the current cost estimates for small firms, and thus they tend to ignore small firms' inflation-adjusted data in updating their probability distributions about the changed states of nature. Third, unlike the case of a between-firm (cross-sectional) comparison for the same period, the hypothesis of more ”informative stock prices" for large firms relative to small firms may not be applicable to the within- firm comparison between two different periods. Recall that the results of the between-firm comparison for each year are, in general, consistent with the evidence of prior earnings information content research. To summarize, the difference in price variability between the predisclosure year and the initial disclosure year is more pronounced for medium and large firms. This result is consistent. with. the inflation-adjusted data of medium size and large firms as informative to market participants. 5.3 Ins Isst of Hyngtnssis 3 Hypothesis 3 tests whether the degree of informativeness of inflation-adjusted data differs across industries accord- ing to their inflation rates. Table 23 provides the empirical results of price variability tests for each industry group. Only those industries that have at least 8 firms in each two-digit SIC code group were analyzed to reduce the small sample effect. Due to a small number of firms in some industries, the violation of normality may be a 70 TABLE 23 Industry Analysisa Erier_Yeer,A§B_129_Yeer tennarieen_12:xelnelb Wilcoxon SCVR SCVR Sign Test Signed-Rank SIC N Mean Mean P-value Z 44 4 13 14 0.67 1.17 0.03 1.35(0.09) 20 21 1.02 0.49 0.96 -1.20(0.89) 26 16 0.87 0.59 0.40 -0.36(0.65) 27 9 1.26 0.86 0.75 -l.01(0.86) 28 38 0.55 0.63 0.31 0.12(0.45) 4 29 11 0.28 1.44 0.11 1.51(0.07) 44 44 32 11 0.85 1.29 0.03 1.87(0.03) 33 24 0.98 0.84 0.15 0.34(0.37) 4 34 9 0.73 0.92 0.09 0.89(0.19) 44 35 30 0.56 0.90 0.05 1.20(0.12) * ** 36 20 0.47 1.15 0.06 l.68(0.05) 44 4 37 20 0.46 0.83 0.02 1.57(0.06) 38 11 0.88 1.09 0.73 -0.18(0.57) 49 60 1.02 0.77 0.82 0.04(0.48) 53 11 0.50 0.45 0.50 0.27(0.39) 44 54 8 0.30 0.77 0.14 1.82(0.04) 67 8 0.57 1.11 0.50 1.12(0.13) a Results are for the cumulation period of day {-1} through {0}. N is the sample size of each industry. The PRIOR and ASR 190 designations refer to inflation-adjusted information releases through the 10-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. SVR is defined in equation <1>. b The results for the nonparametric tests are based on the mean rank and/or the number of cases where SCVRi t for the initial disclosure year is greater than SCVRi for the prior non-disclosure year. The symbols of *, **, anfi *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 71 more serious problem than in earlier analyses. Thus only the results of nonparametric tests are presented in Table 23.22 According to the data presented in Table 23, six industries (their SIC codes are 13, 29, 32, 36, 37, and 54) showed significant stock price reaction to inflation-adjusted data under the Wilcoxon Signed-Rank test during the cumulative two-day-event-period, day {-1} through day {0}. Of the six industries, three industries are only marginally significant, and the other three are significant at five percent. Their results are consistent with the hypothesis of differential stock price reaction across industries. The two-day- cumulation-period is used because the results are significant for only two industries (SIC codes 54 and 67) in the three- day-cumulation-period, and thus we may lose information if we only show the results from that period. These seven significant industries are further investigated in Table 24 in terms of average market value, three-year-average- producer-price-index, and the cumulation period with a significant stock price reaction.23 The data for producer price indexes was obtained from a monthly report," Wholesale Prices and Price Indexes", published by Bureau of Labor Statistics. The 3-Year-Average Producer Price Index (PPI) is the average of the producer price indexes for 1975-1977, and the average of all commodities was 184 (175 for 1975, 183 for 1976, and 194 for 1977). The industries are classified as High (>184+30),Low (<184-30),and Middle (greater than 154 and less than 214). The range of the 3-Year-average- Producer-Price-indexes among commodities is from 91 (home 72 TABLE 24 Characteristics of the Industries with Significant Reaction Average 3-Year- Market Average- Cumulati n SIC Name Value Sizea PPI Type Period (M$) (1967=100) 13 Oil and Gas 556.0 Large 274.4 High {-1,0} Extraction 29 Petroleum and 2,183.0 Large 106.7 Low {-1,0} Coal Products 32 Stone,Clay, and 279.0 Medium 160.6 Middle {-1,0} Glass Products 36 Electric and 1,054.0 Large 132.9 Low {-1,0} Electronic Eq. 37 Transportation 513.0 Medium 112.5 Low {-1,0} Equipment 54 Food Stores 276.0 Medium *0 *C {-1,1} 67 Holding and 490.0 Medium *c *C {-1,1} Other Invest. a The size classification used in the three portfolios size-test is used (see Section 5.2). b The cumulation period in which the specific industry showed significant stock price reaction to inflation-adjusted data is indicated. The food stores industry (SIC code No. 54) showed significant stock price reaction in the cumulation period of both day {-1} through day {0} and day {-1} through day {1}. C Since Bureau of Labor Statistics started collecting producer price indexes by nil SIC codes only quite recently, data for this industry is not available. Other industries were among those selected by the Bureau at that time for publication. 73 electronic equipment) to 381 (coal). Consistent with the previous section (size analysis), the industries that show a significant difference in price variability between the predisclosure year and the initial disclosure year are comprised of medium or large firms. The anaysis using the producer price index does not seem to support the hypothesis that the difference in price variability between the prior year and the initial disclosure year is more pronounced for industries with high inflation rates than for industries with low inflation rates. Most industries in our analysis experienced low or average inflation rates during the test period (1975-1977). As an exception, only the oil and gas extraction industry experienced high inflation during the period. Thus, the degree of informativeness of inflation-adjusted information disclosures differs across industries but not simply by the difference in inflation rates. However, this conclusion is based on only seven industries and some marginal significance level, and should be accordingly qualified. 5-4 W This hypothesis examines whether or not the FPI decisions of firms which disclose inflation-adjusted information significantly Changes in the disclosure year relative to the prior year. The test results for changes in FPI decisions are presented in Table 25 for all firms combined (panel A), December fiscal year-end firms (panel B), and Non-December fiscal year-end firms (panel C). 74 TABLE 25 Tests for Changes in Financing, Production, and Investment Decisionsa Panel A: All Firms (390 firms) £rier.Yenr A§B_129_Year tennerieen_l£:xelnelb Wilcoxon Student Signed-Rank Ratio Mean Stdv.c Mean Stdv.c t Z Debt/Equity 1.31 0.95 1.32 0.93 -0.86(0.39) -0.62(0.54) Dividend *** Payout 0.32 0.34 0.36 0.69 -0.95(0.34) -3.50(0.00) Sales/Total * *** Assets 1.35 1.05 1.36 1.02 -1.7l(0.09) -4.81(0.00) Debt/Total Assets 0.52 0.13 0.52 0.13 -1.60(0.ll) -0.79(0.43) Earnings/Total ** Assets 0.06 0.04 0.06 0.04 0.61(0.55) -2.26(0.02) Panel B: December Fiscal Year-End Firms (270 firms) Debt/Equity 1.27 0.82 1.27 0.76 0.30(0.76) -0.31(0.76) Dividend *** Payout 0.37 0.22 0.41 0.81 -0.83(0.4l) —3.45(0.00) Sales/Total *** *** Assets 1.13 0.79 1.15 0.79 -3.59(0.00) -4.72(0.00) Debt/Total Assets 0.52 0.13 0.52 0.13 -0.6l(0.54) -0.37(0.71) Earnings/Total Assets 0.06 0.04 0.06 0.04 0.68(0.50) -1.53(0.12) Panel C: Non-December Fiscal Year-End Firms (120 firms) Debt/Equity 1.39 Dividend Payout 0.21 Sales/Total Assets 1.85 Debt/Total Assets 0.52 Earnings/Total Assets 0.06 1.19 0.50 1.36 0.15 0.04 1.45 0.23 1.22 0.27 -1.24(0.22) -0.45(0.65) -0.14(0.89) -1.64(0.ll) 0.15(0.88) -O.65(0.51) -1.14(0.25) * -1.72(0.09) -0.87(0.38) * -1.72(0.09) 75 TABLE 25 (cont'd.) a 390 out of 408 firms have a complete set of data for size and ratio analyses in the industrial COMPUSTAT tape [390:270(December Fiscal Year-End firms) + 120(NonDecember Fiscal Year-End firms)]. The PRIOR and ASR 190 designations refer to inflation-adjusted information releases through the 10-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. b A paired comparison is made by taking the difference in ratios between the initial disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1% ,respectively, in tgg tailedzteete- c Standard deviation 76 The majority of the evidence does not suggest that the firms' FPI decisions have changed with the introduction of inflation-adjusted information disclosures. In panel A, the null hypothesis of no change in FPI is rejected by both the parametric and the nonparametric test in only one case, i.e., for the sales/total assets ratio. The results for dividend payout and earnings/total assets reject the null hypothesis with the nonparametric test, but not with the parametric test. Panels B and C of Table 25 indicate that the statistically’ significant results for the Idividend payout ratio in panel A is due to . the December fiscal year-end group, while the earnings/total assets ratio is significant for the Non-December fiscal year-end group. The sales/total assets ratio rejects the null hypothesis for both groups. Thus, tests of the null hypothesis of no Change in firms' FPI decisions yield some mixed evidence. Panel 1 of the Table 26 shows the deciles of the distribution of £8 for both the prior non-disclosure year and the initial disclosure year. The median B-value of all 406 firms in the predisclosure year is 1.05 which is very close to that (1.08) in the initial disclosure year. In addition, the standard deviation of the fis in the predisclosure year is 0.56 which is comparable to the 0.54 in the initial disclosure year. However, the paired-test results in panel 2 of Table 26 presents some evidence (at the 10% significance level) of a Change in firms' systematic risks. The results in Table 26 are therefore consistent with the mixed evidence of 77 TABLE 26 Tests for Changes in Firms' Systematic Risks (B)a Panel 1: Deciles of the distribution of matched pair fls 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Imenizier_xeer 0.38 0.56 0.71 0.88 1.05 1.17 1.31 1.49 1.77 3.03 Ih§_A§B_122_!§QE 0.42 0.63 0.78 0.93 1.08 1.21 1.35 1.54 1.77 3.56 Panel 2: Tests for Changes in HS Erier_Yeer A§E_122_Yeer temperieen_l£:xeluelb 3 p Wilcoxon Mean Standard Mean Standard Student Sign Test Signed-Rank Deviation Deviation t 2 z * * 1.06 0.56 1.10 0.54 1.91(0.06) 1.37(0.19) l.68(0.09) a The slope coefficients (8) from the market model (section 4.2.1) are used to measure the firms' systematic risks. Specifically, the slope coefficients from the estimation period of the ASR 190 year are used as the prior year systematic risks and the slope coefficients from the estimation period of the post disclosure year are used as the ASR 190 year systematic risks. The PRIOR and ASR 190 designations refer to inflation-adjusted information releases through the lO-K report for the year before (prior year) and the year after (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. b A paired comparison is made by taking the difference in fis between the initial disclosure year (or ASR 190 year) and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, reSPectively, in tEe_teiled:tee_e- 78 Changes in firms' FPI decisions presented in Table 25. As indicated in Ohlson [1979, p.213], the disclosure policy change in itself cannot alter the fundamental value characteristics of a security. If the firms' production financing programs (and thus, prices) are indesg affected in some systematic fashion, then a Change in information environments evidenced by increased stock returns variability cannot be solely attributed to any supplementary disclosure (the disclosure policy change). Therefore, any conclusion based on the observed increase in stock returns variability should be qualified according to the degree of violation of this assumption. 5.5 W This hypothesis is based upon the premise that the degree of a change in the information environment depends on the "quantity" of the required disclosures - assuming, of course, that such disclosures are informative. So far, I have used the term "quantity" to mean more and finer information. That is, I used the term "quantity" interchangeably with the term "quality". In fact, quantity is only a subset of quality in a more accurate classification. However, because of difficulty in measuring the quality of disclosure, I examined only the quantity of disclosure in this section. The number of pages of inflation- adjusted disclosures is used as a crude proxy for the quantity of disclosure. Table 27 presents the mean SCVR values across the event days {-1} through {+1} for both the 79 TABLE 27 Tests Conditional on the Quantity of the Required Disclosuresa Erier_xear A§E_122_Yeer tennerieen_l£:xalnelb Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean Z Z Panel 1: One Page (44 firms) 44 -1 0.52 1.10 1.94(0.03) 1.06(0.15) 0.95(0.17) 44 0 0.57 1.01 2.07(0.02) 0.15(0.44) 0.49(0.31) *** * +1 0.57 1.22 3.67(0.00) 0.75(0.23) 1.61(0.06) Panel 2: Two Pages (154 firms) ** *** ** -1 0.69 0.86 2.01(0.02) 2.34(0.01) 2.15(0.02) ** ** 0 0.87 0.87 -0.03(0.49) 2.18(0.02) 1.63(0.05) 44 44 +1 0.85 0.89 0.82(0.21) 2.01(0.02) 1.69(0.05) Panel 3: Three Pages (110 firms) -1 0.71 0.64 -0.57(0.72) 0.29(0.39) 0.10(0.46) 0 0.71 0.66 -0.65(0.74) 0.10(0.46) 0.01(0.50) +1 0.76 0.66 -1.41(0.92) 0.48(0.32) -0.39(0.65) Panel 4: Four Pages (57 firms) -1 0.59 0.59 0.00(0.50) 0.26(0.40) 1.14(0.13) *** ** 0 0.59 0.73 0.88(0.19) 2.38(0.0l) 1.65(0.05) 4 +1 0.61 0.78 1.26(0.10) 0.79(0.21) 1.20(0.12) Panel 5: Five Pages and Over (31 firms) ** ** -1 0.42 0.62 0.47(0.32) 1.80(0.04) 2.04(0.02) 4 44 44 0 0.81 1.21 1.32(0.09) 1.80(0.04) 1.67(0.05) 44 4 4 +1 0.80 1.20 1.63(0.05) 1.44(0.07) 1.51(0.07) 80 TABLE 27 (cont'd.) Erier_1eer A§B_129_Yeer tennerieen_12:!elnelb Wilcoxon Event SCVR SCVR Z Sign Test Signed-Rank SCVR Day Mean Mean Z 2 Panel 6: All Firms (396 firms) 4 444 444 -1 0.64 0.77 1.28(0.10) 2.76(0.00) 2.76(0.00) 444 444 0 0.75 0.83 1.20(0.12) 3.07(0.00) 2.29(0.01) 44 444 444 +1 0.76 0.87 1.99(0.02) 2.66(0.00) 2.32(0.01) SCVR and Z are defined in equation <2> and <3>, respectively. IggRresults are based upon 396 instead of 407 firms due to missing microfiches of the 10-K reports. More than a half page but less than a full page disclosure in the microfiches of the 10-K report was treated as one full page disclosure. However, less than a half page disclosure was not counted. The PRIOR are ASR 190 designation refers to inflation-adjusted information releases through the 10-K report for the year before (prior year) and the year of (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. Event day {0} is the date of receipt of the lO-K by the SEC. b The results for the nonparametric tests are based on the mean rank and the number of cases where SCVRi t for the initial disclosure year is greater than SCVRi' for the prior non-disclosure year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 81 prior non-disclosure year and the initial disclosure year, along with their parametric and nonparametric test results. As expected, significant stock price reactions are evident for firms that had the highest level of disclosure (i.e., firms with 5 or more pages). However, this finding exists along with an anomaly. Specifically, firms categorized in the second lowest level of disclosure also show significant price changes across the information environments, and firms with the lowest level of disclosure yield significant differences in SCVRs between the prior non-disclosure year and the initial disclosure year under the parametric test. The failure to reject the null hypothesis of no difference in the stock price reaction between firms with highest level of disclosure and firms with the lowest level of disclosure is subject to at least two limitations. First, the use of the number of pages as a proxy for the measure of fineness of the required disclosure may be inadequate in that the two-page-disclosure can be regarded as the highest level of disclosure for small or medium firms but not for large firms. Second, more pages do not necessarily imply the highest level of disclosure due to data redundancy and the possibility of an inefficient presentation. Stated another way, quantity is not the same as quality. Table 28 presents the test results conditional on the quantity of the required disclosures after firm-size is controlled. In each group, we expect firms with a higher level of disclosure (3 pages and over) to show a more significant stock price effect in the initial disclosure year 82 than firms with a lower level of disclosure (2 pages and fewer). However, the results from Table 28 are mixed in that the above assumption is supported only in small firms, but neither medium nor large firms are supportive of the assumption. The results in Table 28 are consistent with those of Table 27, and they are subject to the same criticisms as those given in the analysis of Table 27. 83 TABLE 28 Tests Conditional on the Quantity of the Required Disclosures Size-Controlled Erier_Yeer A§R_129_Yeer tennerieen_1£:xelnela Wilcoxon Event SCVR SCVR ZSCVR Sign Test Signed-Rank Day Mean Mean 2 Z Panel 1: Small Firms [A].Leee1tnen_end_Eenel_te_2_ne9ee (58 firmS) —1 0.93 0.57 -1.34(0.09) 1.05(0.18) -0.l4(0.56) 0 1.29 0.67 -3.33(1.00) -1.31(0.93) -1.42(0.92) +1 1.24 0.70 -3.50(1.00) -1.05(0.88) -1.37(0.92) [B] Mszs thnn and Egnal to 3 pages (70 firms) . 4 4 -1 0.55 0.73 0.76(0.22) 1.43(0.09) 1.57(0.06) * 0 0.81 0.74 -0.39(0.65) 1.43(0.09) 1.13(0.13) ** * +1 0.81 0.79 -0.20(0.58) 1.91(0.04) 1.52(0.06) ane : e 1um irms [A] Leee_tnen_ene_tgnelnte_2_neeee (62 firmS) 444 44 444 -1 0.56 1.22 2.58(0.01) 2.29(0.02) 2.39(0.01) *** ** ** 0 0.69 1.17 2.66(0.00) 2.03(0.03) 1.97(0.03) *** ** *** +1 0.65 1.35 4.75(0.00) 2.03(0.03) 2.64(0.00) [B] uere_tnen_ann_Eeuel_te_3_ne9ee (64 firms) -1 0.66 0.64 -0.07(0.53) 0.50(0.35) 0.92(0.18) ** 0 0.65 0.81 0.90(0.18) 1.75(0.05) 1.24(0.11) +1 0.80 0.81 0.08(0.47) 0.00(0.50) -0.19(0.58) ane : rge irms [A] Lsss tnan nng Egnal to 2 pages (74 firms) 44 4 4 -1 0.55 0.66 1.67(0.05) 1.63(0.07) 1.31(0.10) ** 0 0.54 0.93 1.85(0.03) 1.16(0.15) 1.23(0.11) *** +1 0.62 0.84 2.22(0.01) 0.93(0.21) 1.26(0.11) [B] t a nd E al to 3 a es (55 firms) -1 0.74 0.51 -0.86(0.81) 0.13(0.50) -0.12(0.55) 0 0.65 0.80 0.79(0.22) 0.40(0.39) 0.19(0.43) * +1 0.57 0.82 1.57(0.06) 0.94(0.21) 0.58(0.28) 84 TABLE 28 (cont'd) a The results for the nonparametric tests are based on the mean rank and the number of cases where SCVR- for the initial disclosure year is greater than SCVR-' for the prior non-disclosure year. The symbols of *, **, an *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. Chapter 6 SUMMARY, CONCLUSIONS, AND LIMITATIONS 6.1 Snnnnty, Qonslusigns, nng Limitstisns Although most previous studies fail to show information content for Accounting Series Release No. 190 data (ASR 190), persuasive arguments can be made for the relevance of current cost data in various decisions.24 Moreover, the evidence to date leaves many issues unresolved. Therefore, the importance of this topic for accounting practice, along with the limitations in prior research, provided the motivation for a re-examination and extension of earlier research. The major results of this paper are summarized below with their interpretative limitations: First, daily stock returns variability measures, cumulative stock returns ‘variability' measures, and ratio- based stock price variability measures nil indicate that the SEC's mandated inflation—adjusted data are associated with increased stock returns variability at their release. Second, the likelihood of confounding factor explanations to the increased stock returns variability at the release of inflation-adjusted information is teducsd by (1) similar results for the post-disclosure analysis, and (2) the absence of significant changes in stock returns behavior for a comparison group of firms. However, the usefulness of the 85 86 comparison between the prior year and the post-disclosure year is greatly lessened by the increased stock returns variability in the post-disclosure year relative to the initial disclosure year. Third, the marked evidence for the potential effect of event-date clustering has nst been observed. Fourth, the evidence for increased stock returns variability between the predisclosure year and the disclosure year is greater for W and 1.41192 firms, but not for small firms. Fifth, the informativeness of inflation- adjusted information gifts]: across industries, and these differences are not solely attributed to varying rates of inflation. However, this conclusion is based on only seven out of seventeen industries and should be accordingly qualified. Sixth, tests of the null hypothesis of no change in firms' financing, production, and investment (FPI) decisions between the prior year and the initial disclosure year yield some mixed evidence. The results based on the direct test of the slope coefficients of the market model are also consistent with the mixed evidence about the Change in the average systematic risk. Unless the assumption of no Change in firms' FPI decisions or systematic risks is maintained, a change in information environments evidenced by increased stock returns variability cannot be solely attributed to any Change in the disclosure policy (e.g. , supplementary disclosure of inflation-adjusted data). Therefore, any conclusion based on the observed increase in stock returns variability should be qualified according to 87 the degree of violation of such an assumption. Seventh, the phenomenon of the increased stock returns variability is nst nnigns to the group of firms that had the highest level of disclosure. The similar evidence is also documented when firm size is controlled. However, this indistinguishable stock returns behavior between firms with the lowest and highest level of disclosure may be due to the crude proxy used as the measure of fineness for the required disclosures. Finally, the evidence of increased stock returns variability at the time of the disclosure of inflation- adjusted information is important in that it suggests that market participants reacted to policy changes relating to such information, and the required disclosures affected the information environment. Again, the above conclusion depends on how seriously one views the limitations of this research. 6.2 Oths; Limitntions Other potential limitations in interpreting the results of this research can be summarized as follows: First, the violation of the normality and serial independence assumption in the unexpected earnings leads us to place less emphasis on the results from the parametric t-test and the ZSCVR test. Second, as in any other empirical studies, this research is also subject to the criticisms related to the application. of an accounting' theory, such as potentially misleading application of unrealistic assumptions (e.g., constant risk-free rate, linear information dynamic, constant price of risk, etc.). Third, in the industry analysis, the 88 producer' price index rather than the nnsxnsstsg rate of inflation is used to examine nnsnnsstsd stock returns behavior of the firms. Fourth, as mentioned in Section 5.5, a more sophisticated proxy that can directly deal with the quality' of disclosure, rather’ than. the quantity of disclosure, could have been used as the measure of fineness for the required disclosures. 89 FOOTNOTES 1 If information is demanded, it should have information content, but not vice versa. This is because investors may not act according to new information. That is, the change in an investor's belief does not necessarily lead to a modification of his or her act. 2 According to Ohlson [1979], A finer information environment is naturally defined as an environment in which the set of nyniinn1s_stnts gsscrintgts is n snnersst as compared with some alternative (coarser) environment. For example, a firm may disclose the market value of the inventory in addition to its historical cost. Mote data is therefore disclosed and nv ro nt is ' .... The variability of the price is intget in tns tiner enrirennent- (pp. 214-215) 3 Treatment firms systematically differ in size from control firms because of the materiality threshold associated with ASR 190. In this study, the size problem may be exacerbated in that the inflation-adjusted disclosure standard applies to firms in a manner related to size. 4 Market expectations regarding inflation-adjusted data and the direction of the price effect given expectations are difficult to specify owing to lack of a theory in this area (see Beaver, Christie, Griffin [1980, p.131] for further discussion of this matter). 5 According to DeBerg and Shriver [1987], Specific prices within some industries continue to rise at excessive rates: for example, in 1986, prices within service-oriented industries such as medical care, transportation, and entertainment Climbed 7.8%, 6.0%, and 5.0%, respectively, and the education industry is also experiencing sharp increases in tuition prices. (Footnote 18 in p. 77) 90 FOOTNOTES (cont'd.) 6 The purported purpose of ASR 190 was to: Provide information to investors which will assist them in obtaining an understanding of the current costs of operating the business which can not be obtained from historical cost financial statements taken alone. Such information will necessarily include subjective estimates and it may be supplemented by additional disclosures to assist investors in understanding the meaning of the data in particular company situations. A secondary purpose is to provide information which will enable investors to determine the current cost of inventories and productive capacity as a measure of the current economic investment in these assets existing at the balance sheet date. (ASR 190, p. 462) 7 According to ASR 190, "the note or the separate section may be designated unaudited" [ASR 190, p. 457]. 8 Since the focus of this study is on stock price variability (or nondirectional) tests, the discussion of the prior research using directional tests of stock prices is omitted. See Deberg, C. and K. Shriver [1987] for such a discussion. 9 BCG used both directional and nondirectional tests. Their results of mean daily stock return volatility ratios (nondirectional test metric) are reproduced and replicated using my sample in Appendix A. 10 Depreciation, Cost of Goods Sold,and Net Monetary Position for the income statement and Net Productive Capacity, Inventory, and Net Monetary Position for the balance sheet are such variables used in the partitioning. 11 As an alternative measure, Beaver, Christie and Griffin [1980] suggested that one use abnormal stock returns and compute a volatility ratio using these returns. This measure would take out the common movement in the U ratio due to economy wide information. I will implement their suggestion later in the paper. 12 High and low are defined as "above" and "below" the median values of certain variables (e.g., the difference in depreciation between historical cost and replacement cost, which is deflated by historical cost-based net income) used in the partitioning. See Beaver, Christie, and Griffin [1980, p.135] for more details. 91 FOOTNOTES (cont'd) 13 GB used several directional tests in addition to the nondirectional test. Fig. 1 of GB is replicated in Appendix B of this paper. 14 Beaver's "U-statistic” is defined as a standardized variance of returns (SVR) metric: see equation <1> in Section 4.2.1. 15 Atiase [1985] and Grant [1980] report that the degree of stock price reaction to information is inversely related to the capitalized value (size) of the firm. 16 My initial plan was to investigate the extent of voluntary disclosure for inflation-adjusted information made in the predisclosure year in addition to the quantity of the required disclosure presented in the disclosure year. However, after examining 150 firms and finding that none had any voluntary disclosure in the 10-K report of the predisclosure year, I abandoned this aspect of the study. 17 The auto-correlation function test indicates that the violation of the serial correlation assumption occurred on days {-1}, {0}, and {+1} in the initial disclosure year and days {+1} and {+2} in the prior year at the 5% significance level. Specifically, auto-correlation coefficients of SVRs for lag 1 through lag 4 across the 5 days in the event period are 0.39, -0.37, -0.43, and -0.09 for the initial disclosure year and 0.03, 0.06, -0.33, and -0.26 for the prior year. 18 The first-order approximate mean and variance of SARR can be easily derived using a Taylor series as follows: SARR = 1.12 it/ua it'l E(SARR) z E(u21t)/E(u21t') = Citaiz/Cit'aliz z Siz/SIiZ VAR(SARR) z VAR(u’it) + VAR(U’itu)- 19 The first-order approximate mean and variance of SVRR can be easily derived using a Taylor series as follows: SVRR = (u=it/si=) / (u’itI/s'i') E(SVRR) ‘3 Ehl’it/Si’) / E(u‘it./s'i’) z 1, VAR(SVRR) z VAROJz iii/812) + VARUI2 itg/S'i’) z 4(Ti-3)/(Ti-6). 92 FOOTNOTES (cont'd) 20 Five percent of sample firms that experienced extreme changes in information environments are further investigated in the Wall Street Journal Index for potential effects from confounding events, such as dividends announcements, merger announcements, and other special events. The evidence shows that no significant announcements were made around the 10-K filing dates in the initial disclosure year for these firms. 21 McNichols and Manegold [1983, p.65] also use the 10.0 criterion in dealing with outliers. 22 See Section 5.1.1 for the discussion of nonnormality. 23 Only those industries which showed significant stock price reaction to inflation-adjusted information under the Wilcoxon Signed-Rank screening test are analyzed. 24 According to DeBerg and Shriver [1987], It should be noted that three of the seven Board members dissented to the adoption of SPAS 89 which made voluntary the supplementary disclosure of current cost/constant purchasing power information. In expressing their views, these members indicated that much of the due process (e.g., research, debate, deliberations, and decisions) and application experience will have to be repeated in periods of future substantial inflation. Further, they felt that the failute ts accgunt for snecitis snd gsnetgl ptics gnangss wQuig sstignsly tsgnss tns relevangs and tepresentational faithtniness of niststisai cost financiai ststsment . (p. 77) APPENDICES 93 APPENDIX A Replication of Table 10 in BCG [1980]a Mean nniiy sesntity tstntn vgistiiity tatisb B§§_Ll2figl ID1§_B§§§Q£QD 1. Interval (lo-K filing) 3/24/77 0.46 0.59 3/25/77 0.45 0.77 3/28/77 0.56 0.70 3/29/77 0.60 0.74 3/30/77 0.73 0.96 3/31/77 0.41 0.62 4/1/77 0.54 0.75 4/4/77 0.59 0.88 4/5/77 0.44 0.65 4/6/77 0.37 0.49 4/7/77 0.45 9.55 Mean 0.51 0.70 2. Sample size 553 firms 407 firms 3. Estimation period 654 days 654 days (6/2/75 - 12/31/77) 4. Fiscal Year End December firms 465 firms 282 firms NonDecember firms 88 firms 125 firms a Table 10 of Beaver, Christie, and Griffin [1980] is replicated for the reporting group using the sample of this study. The reporting group is the group of the firms which were required by ASR 190 of the Securities and Exchange Commission to disclose the inflation-adjusted data. b Uit = (Ru-)2 / (R12) where Rit = daily return on security i on day t, (Ra?) = mean square return during a 'non-report' period. Non-report period is 6/2/75 through 12/31/77 (654 days). 94 APPENDIX B Replication of Fig. 1 in GB [1980] PaneI A: Average Beaver‘s U for I06 EreaEmenE firms on trading days adjacent to the public filing date of the Form 10-K (day 0) (p.113 of GB [1980])3 [Beaver's Ut] 2.7- + — + + ++ + + + 1|llllltlilillllllitlIIIIl -30 -20 -10 0 +10 +20 [Trading Day] 95 APPENDIX B (cont'd.) a Figure 1 of Gheyara and Boatsman [1980] is reproduced in Panel A and replicated in Panel B for treatment firms using the sample of this study. The treatment group is the group of the firms which were required by ASR 190 of the SEC to disclose the inflation-adjusted data. See equation <1> in Section 4.2.1 for Beaver's U. 96 APPENDIX B (cont'd.) Panel B: Average Beaver's U for 407 firms on trading days adjacent to the public filing date of the Form 10-K (day o>b [Beaver's Ut] 2.2- + + 0.8- ++ ++ + + -30 -20 -10 0 +10 +20 [Trading Day] b Mean (1.03), Median(0.92), Min.(0.63), Max.(2.66), and Standard Deviation (0.34). 97 APPENDIX C Patell's Measure Lt__=s U Patell [1976, pp. 256-258] refined Beaver's U by taking into account the increase in variance due to prediction outside the estimation period and giving explicit recognition to the number of observations used to estimate the market model. Thus, Patell's U is defined as follows: Patell's U = Uit = ——————————————— where Ti Cit = 1 + 1/T1 + (Rm. - fiwdflmmd - Rn)“ where__ Ti Rm = (1/Ti) 3 Rmd d=l Cit is the increase in variance due to prediction outside the estimation period, T- = the number of observations used to estimate the market model for firm i. To test whether the mean of Patell's U is equal to 1, the additional test statistic (Patell's 2), which was also derived by Patell [1976, p.258], is needed. Patell's Z is constructed under the assumptions of the normal distribution 98 APPENDIX C (cont'd) and cross-sectional independence of Patell's U. By applying the central limit theorem, an approximately unit Normal . variate can be developed: N (U- - 1) i=1 1t 2 _ ....................... Ut — N 1 [:2 2(T1 - 3)/(Ti ' 6)];S 1=1 where E [ Uit] = 1 Var [Uit] = 2(Ti - 3)/(Ti - 6) since F(1,Ti-2) = t’Tiez = [N(0,1)]1/(X’Ti_2/Ti-2) = (“itz/aizcit) * (Ti'2)°iz/(Ti'2)siz = “ita/Citsi' and Var(uit/sicit%) E[uitz/Citsizl (Ti'2)/(Ti'4) > 1 E[Patell's U] = E[uit1/Citsi’*((Ti-4)/(Ti-2))] = 1 99 APPENDIX C (cont'd) Prion Xsnt A33 129 test Comnarisgn (P-vaiue) Wilcoxon _ _ Student Signed-Rank Patell's Ut Patell's Ut t 2 Day Mean zUt Mean ZUt (one-tail) (one-tail) *** *** * *** -2 0.67 -4.87 0.78 -3.32 1.27(0.10) 2.30(0.01) 444 444 44 444 -l 0.64 -5.27 0.80 -3.04 1.64(0.05) 3.00(0.00) ** ** 0 0.86 -2.18 0.91 -1.53 0.30(0.38) 1.92(0.03) 444 4 +1 0.77 -3.44 0.98 -0.54 1.49(0.07) 0.89(0.19) ** +2 0.91 -l.51 0.90 -1.63 -0.07(0.53) 1.91(0.03) *** ** * *** Mean 0.76 -3.45 0.86 -2.01 1.46(0.07) 2.37(0.01) a One firm (cusip no. 74100410) had more than 5 missing returns in the estimation period and thus was omitted from the analysis. Therefore, the results are based upon 406 firms in the 5-day-event- period-case. The symbols of *,**,and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests (two-tailed tests for ZUt). 100 APPENDIX D Truncated Variability Measurea Etior Yea; A§R igg Xsa; Qonnntison (g-valus) Wilcoxon SVR SVR Student Signed-Rank Day Mean Mean t Z * *** -2 0.64 0.73 1.28(0.10) 2.31(0.01) 44 444 -1 0.60 0.75 2.02(0.02) 3.03(0.00) ** 0 0.70 0.78 1.12(0.13) 1.96(0.02) 4 +1 0.69 0.80 1.40(0.08) 0.89(0.19) 4 444 +2 0.66 0.78 1.49(0.07) 1.99(0.02) *** *** Mean 0.66 0.77 2.93(0.00) 2.79(0.00) a SVR is defined in equation <1>. Any value of price variability which is greater than 5.0 is truncated at 5.0. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 101 APPENDIX E Summary Statistics for SVR and SCVR Measures Day Year Median Stdev Min Max Panel 1: SVR -2 Prior 0.24 1.17 0.00 7.88 ASR 190 0.32 1.35 0.00 14.05 -1 Prior 0.19 1.31 0.00 15.58 ASR 190 0.34 1.36 0.00 9.92 0 Prior 0.24 2.59 0.00 42.22 ASR 190 0.35 1.82 0.00 19.44 +1 Prior 0.24 1.58 0.00 20.60 ASR 190 0.31 2.36 0.00 32.30 +2 Prior 0.20 3.27 0.00 55.34 ASR 190 0.31 1.84 0.00 16.47 Panel 2: SCVR (5-day cumulation period) -2 Prior 0.24 1.15 0.00 7.73 ASR 190 0.31 1.33 0.00 13.84 -1 Prior 0.32 0.89 0.00 7.69 ASR 190 0.45 0.98 0.00 8.75 0 Prior 0.40 1.02 0.01 13.85 ASR 190 0.55 0.93 0.00 8.84 +1 Prior 0.48 0.92 0.01 11.59 ASR 190 0.60 1.06 0.01 11.30 +2 Prior 0.50 0.98 0.01 11.35 ASR 190 0.60 0.98 0.01 9.11 Panel 3: SCVR (3-day cumulation period) -1 Prior 0.19 1.29 0.00 15.35 ASR 190 0.33 1.34 0.00 9.77 0 Prior 0.38 1.41 0.00 20.71 ASR 190 0.46 1.14 0.00 10.61 +1 Prior 0.41 1.17 0.01 15.41 ASR 190 0.54 1.30 0.01 14.61 102 APPENDIX F Daily Stock Returns Variability Tests (21-day period, 403 firms)a Prior ASR I50 Mean 122: XQQE Differenee QQEE§£1§QD_LEZXQLB§1 _ Wilcoxon Event SVR SVR ASR 190 minus Student Signed-Rank Day Mean Mean Prior Year t Z -10 0.85 1.06 0.21 0.82(0.21) -0.05(0.52) - 8 0.94 0.85 -0.09 -0.61(0.73) -0.28(0.61) - 7 0.85 0.80 -0.05 -0.45(0.67) -l.04(0.85) - 6 0.91 0.80 -0.11 -0.91(0.82) -0.59(0.72) - 5 0.92 0.85 -0.07 -0.52(0.70) -0.17(0.57) - 4 0.96 0.70 -0.26 -2.45(0.99) -0.96(0.83) * - 3 0.85 1.06 0.21 1.35(0.09) 0.79(0.21) * *** - 2 0.67 0.78 0.11 1.28(0.10) 2.37(0.01) 44 444 - 1 0.64 0.82 0.18 1.75(0.04) 3.02(0.00) 44 0 0.85 0.91 0.06 0.36(0.36) l.98(0.02) * + 1 0.76 0.97 0.21 1.48(0.07) 0.98(0.l6) 44 + 2 0.92 0.90 -0.02 -0.14(0.56) 1.95(0.03) 44 + 3 0.79 0.91 0.12 0.77(0.22) 1.99(0.02) + 4 0.76 0.81 0.05 0.57(0.29) 1.22(0.11) + 5 0.84 0.84 0.00 0.04(0.48) 0.23(0.41) 4 44 + 6 0.76 0.93 0.17 1.46(0.07) 1.72(0.04) 444 + 7 0.79 0.94 0.15 1.25(0.11) 2.31(0.01) + 8 0.79 0.78 -0.01 -0.10(0.54) 1.13(0.13) 4 44 + 9 0.77 0.95 0.18 1.52(0.06) 2.01(0.02) +10 0.82 0.87 0.05 0.54(0.30) 0.26(0.40) 4 44 Mean 0.82 0.88 0.06 1.27(0.10) 1.96(0.03) 103 APPENDIX F (cont'd) a SVR is defined in equation <1>. The firms whose cusip numbers are 72605610, 74100410, 76688910, and 91360610 were omitted from the analysis due to insufficiency of return data during the estimation and event periods. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. 104 APPENDIX G Distribution of 10-K filing Dates in the Post Disclosure Year Panel A: All Firms (387 firms) c s ea No. of out _firne _% January 11 2.8 February 8 2.1 March 230 59.4 April 48 12.4 May 11 2.8 June 7 1.8 July 3 0.8 August 7 1.8 September 19 4.9 October 13 3.4 November 6 1.6 December 25 §,2 Total 387 100.0 Panel B: December Fiscal Year-End Firms (271 firms) [1] Breakdown by month Estiod ' t 's u ea No. of firms .1 February: 2/1 - 2/10 1 0.4 2/11 - 2/20 2 0.7 2/21 - 2/28 0 0.0 March: 3/1 - 3/10 1 0.4 3/11 - 3/20 9 3.3 3/21 - 3/31 220 81.2 April: 4/1 - 4/10 37 13.6 4/11 - 4/20 1 0.4 4/21 - 4/30 0 0.0 Total I.) m \l H O O O 105 APPENDIX G (cont'd.) ”1 period Esst Qisclosnte Yea; No. of Qny firms 3/21 ........................ 1 3/22 ........................ 1 3/23 ........................ 2 3/24 ........................ 9 3/25 . ........... ............ 0 3/26 ........................ 0 3/27 ........................ 22 3/28 ........................ 7 3/29 ........................ 24 3/30 ........................ 48 3/31 ........... ............. 125 Total ........................ £22 Panel C: Non-December Fiscal Year-End Firms (116 firms) W No. of mo t _fime 31 January 11 9.6 February 5 4.3 March 0 0.0 April 10 8.7 May 11 9.6 June 7 6.1 July 3 2.6 August 7 6.1 September 19 16.5 October 12 10.4 November 6 5.2 December 4 24 20.9 Total ii; 100.0 106 APPENDIX H Characteristics of the Comparison Group Sample Panel A: Sample selection criteria .......No. of firms Compustat firms: Primary industrial, Tertiary, Supplementary Industrial Primary Industrial and in the S&P Industrials Index...................2400 Less: Inventories and plant assets more than $100 million and more than 10% of total assets plus Inventories and plant assets less than $50 million .......... ....... . ...... (2238) Less: Stock returns unavaiable on the CRSP daily returns tape or unable to read the 10-K filing date from a microfiche or missing microfiches......... ..(éZ) Sample size 00 Panel B: Classification of Sample By Fiscal Year-End dus assification Number of January ........ 7 sic Codes titns February........ 1 March........... 7 0000-0999 ......... 0 April .......... 5 1000-1499 ......... 3 May ........... 5 1500-1799 ......... 1 June ........... 11 2000-3999 ......... 68 July ........... 0 4000-4999 ......... 2 August.......... 3 5000-5199 ......... 2 September....... 11 5200-5999 ......... 7 October......... 4 6000-6799 ......... 7 November........ 5 7000-8999 ......... is December........ _11 Total $22 Total 100 107 APPENDIX I Distribution of 10-K filing dates for the Comparison Group Panel A: All Firms (100 firms) _Dieeleeure_xear _Bredieeleenre_1eer No. of No. of Eenth _firne _3_. _firne .1 January 4 4.0 5 5.0 February 3 3.0 6 6.0 March 40 40.0 39 39.0 April 9 9.0 10 10.0 May 4 4.0 l 1.0 June 7 7.0 7 7.0 July 4 4.0 6 6.0 August 5 5.0 5 5.0 September 12 12.0 11 11.0 October 0 0.0 0 0.0 November 3 3.0 3 3.0 December __2 2,9 _1 1,9 Total $22 100.0 122 100.0 Panel B: December Fiscal Year-End Firms (282 firms) [1] Breakdown by month Estiod _Qieeleenre_xeer Etegisciosure Year No. of No. of titns _t_ firms _t February: 2/1 - 2/10 0 0.0 0 0.0 2/11 - 2/20 0 0.0 0 0.0 2/21 - 2/28 0 0.0 0 0.0 March: 3/1 - 3/10 1 2.4 2 4.8 3/11 - 3/20 2 4.8 2 4.8 3/21 - 3/31 33 80.6 33 80.6 April: 4/1 - 4/10 4 9.8 4 9.8 4/11 - 4/20 1 2.4 0 0.0 4/21 - 4/30 __g - g,g __9 9,9 Total _4_1 100.0 _4_1 100.0 108 APPENDIX I (cont'd.) [2] Further breakdown of 3/21 - 3/31 period _Dieeleeurexeer Wed ' sclo ea No. of No. of Dny firms 3/21 3/22 3/23 3/24 3/25 3/26 3/27 3/28 3/29 3/30 3/31 Total 0000000000 HIJ 0:» 0000000000000 0000000000000000 0 Hos .1 U OUU‘OONHI—‘HH 0000000000000 0000000000000000 33 Panel C: Non-December Fiscal Year-End Firms (59 firms) Jieelesnnefiar _Ene_1J_l__re_Ye_d ' s osu ar No. of No. of Month Jim January I" L. February March April May June July August September October November December Total l"‘|~o " \O uONmIthchubUh N H UIOOQOII-I'O‘O‘OIUIG HONU'IGIDQQQHQ H O O 0 “ml *‘ WNNOHUIOSQHGNGUI H H HIa H WOQQOPHOUOQ HOUIUINQQNUNUI H O O O 109 APPENDIX J Size Effects -- Five Portfoliosa gsnngtison (E-value)b Wilcoxon WW SVR SVR Student Signed-Rank Day Mean Mean t Z Panel A: Size I Firms (78 firms) -2 0.55 0.76 1.17(0.12) 1.00(0.16) -1 0.78 0.64 -0.54(0.70) 0.77(0.22) 0 1.08 0.69 -1.31(0.90) -0.88(0.81) +1 0.93 0.76 -0.74(0.77) -0.78(0.78) :2T 1.42, 0.91 -0.64(0.74) -0.52(0.70) Mean 0.95 0.75 -1.05(0.85) -0.22(0.59) ane : lze 1rms Irms) 4 44 -2 0.48 0.66 1.27(0.10) 1.62(0.05) 4 44 0 1.39 0.98 -0.70(0.76) 0.18(0.43) +1 1.31 , 1.03 -0.76(0.77) -1.12(0.87) 4 44 :2 0.66 _l.03 1.48(0.07) 1.96(0.03) Mean 0.88 0.92 0.21(0.42) l.09(0.14) ane : Size 1rms irms) -2 1.00 1.08 0.25(0.40) 0.58(0.28) -1 0.71 0.82 0.55(0.29) 1.17(0.12) 4 0 0.72 0.82 0.39(0.35) 1.50(0.07) 44 44 +1 0.55 0.91 1.88(0.03) 1.88(0.03) :2 0.80 0i67 -0.66(0.751, 0.37(0.36) Mean 0.75 0.86 0.97(0.17) 0.84(0.20) 110 APPENDIX J (cont'd.) Panel D: Size IV Firms (78 firms) -2 0.66 0.63 -0.20(0.58) 0.85(0.20) -1 0.60 0.77 0.82(0.21) 0.84(0.20) 4 0 0.70 1.11 1.40(0.08) 0.94(0.17) 4 +1 0.58 1.27 1.50(0.07) 1.14(0.13) :2 0.93 0.94 0.03(0.49) 0.44(O.33) 4 4 Mean 0.69 0.95 1.47(0.07) 1.33(0.09) Panel B: Size V Firms (78 firms) * 'k -2 0.60 0.82 1.28(0.10) 1.52(0.07) . 4 -1 0.60 0.81 0.94(0.17) 1.34(0.09) 44 444 0 0.52 0.97 1.63(0.05) 2.26(0.01) 44 4 +1 0.46 0.98 1.87(0.03) 1.57(0.06) 4 12 0.89 0.87 -0.08(0.531 1.42(0,98) 44 44 Mean 0.61 0.89 1.86(0.03) 2.15(0.02) a 390 out of 407 firms have a complete set of data for size and ratio analyses in the industrial COMPUSTAT tape. A firm is classified as "fiiss_1" if 12.77 < Market Value (MV) < 83.23, Mean MV = 44.49 in M$, and Median MV = 44.12 in M$, as "fiiss_11" if 83.33 < MV < 178.09, Mean MV = 125.97 in M$, and Median MV = 126.17 in M$, as "EiZ§_III" if 178.16 < MV < 370.90, Mean MV = 256.58 in M$, and Median MV = 234.34 in M$, as ”§i§§_ly" if 371.40 < MV < 804.30, Mean MV = 580.60 in M$, and Median MV = 580.50 in M$, and as "Sizs V" if 811.00 < MV < 17,123.00, Mean MV = 2,560.00 in M$, and Median MV = 1,685.00 in M$. The PRIOR/ASR 190 designation refers to inflation-adjusted information releases through the 10-K report for the year before (prior year)/the year after (initial disclosure or ASR 190 year) the issuance of ASR No. 190 by SEC. SVR is defined in equation <1>. b A paired comparison is made by taking the difference in SVRs between the initial disclosure year and the prior year. The symbols of *, **, and *** indicate statistical significance levels of 10%, 5%, and 1%, respectively, in one tailed-tests. BI BLI OGRAPHY BIBLIOGRAPHY Atiase, R., "Predisclosure Information, Firm Capitalization, and Security Price Behavior Around Earnings Announcements," Jenrne1.9f_Aeeennting_Eeeeeren. Spring 1985, pp. 21-36. Beaver, W., "The Information Content of Annual Earnings Announcements," rna c t es a c , Supplement 1968, pp. 67-92. Beaver, W., A. Christie and P. Griffin, ”The Information Content SEC, Accounting Series Release No. 190," Jontnal sf ncsgnnting ans Essnoniss, 2, 1980, pp. 15-39. Beaver, W., P. Kettler and M. Scholes, "The Association Between Market-Determined and Accounting-Determined Risk Measures," The Ascounting neview, October 1970, pp. 654-682. Brown , S. J. and J. B. Warner, "Using Daily Stock Returns: The Case of Event studies." Jenrnel_ef_rinaneial Economics, 14, 1985, pp. 3-31. Bublitz, B., T. Frecka and J. McKeown, "Market Association Tests and FASB Statement No.33 Disclosures: A Reexamination." 1enrnel_ef_Aeeenntin9_Eeeeeren. Supplement 1985, pp. 1-27. Chambers, A. and S. Penman, "Timeliness of Reporting and the Stock Price Reaction to Earnings Announcements," Journal of Accounting gsssntch, Spring 1984, pp. 21-47. Collins, D., S. P. Kothari and J. Rayburn, ”Firm Size and the Information Content of Prices With Respect To Earnings, " Jenrnel_ef_Aeeeuntins_end_zeeneniee 9 1987 pp. 111- 138. DeBerg, C. and K. Shriver, "The Relevance of Current Cost Accounting Data: A Review and Analysis of Recent studies," 1enrnel_ef_Aeeenntin9_Literatnre. Vol-6. 1987. pp. 55-87 0 111 112 Demski, Joel 8., "Demand for Accounting Procedures in a Managerial Setting," Working Paper(Stanford University), 1984, pp. 1-38. Dyckman, T., D. Philbrick, and J. Stephan, ”A Comparison of Event Study Methodologies Using Daily Stock Returns: A Simulation Approach." l9urnal_9f_Aeeennting_Beeeareh. Supplement 1984, pp. 1-30. FASB, Statement of Financial Accounting Standards No. 33 Financial Reporting and Changing Prices, September 1979. Foster, G.,"Quarterly Accounting Data: Time Series Properties and Predictive Ability Results," Ins_nsssnnting_3syisg, January 1977, pp. 1-21. Foster III, T. and D. Vickery, "The Incremental Information Content of the 10-K." Tne_neeennting_8exieu. October 1978, pp. 921-934. Freeman, R., "Alternative Measures of Profit Margin: An Empirical Study of the Potential Information Content of Current Cost Accounting," 0 o Ch, Spring 1983, pp. 42-64. Gheyara, K. and J. Boatsman, "Market Reaction to the 1976 Replacement Cost Disclosures,” o nal c u 'n and Economics, 2, 1980, pp. 107-125. Lev, B., "Toward a Theory of Equitable and Efficient Accounting Policy.” Ine_neeennting_3exieu. January 1988. pp. 1-22. Lintner, J., "The Valuation of Risky Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets," BexieE_ef_Eeeneniee_end_§tetietiee 47 1965 pp. 13- 37. Lustgarten, S., "The Impact of Replacement Cost Disclosure on Security Prices,” lenrnel_ef_Aeeennting_ene_Eeeneniee. 4, 1982, pp. 121-141. McNichols, M. and J. Manegold, "The Effect of Information Environment on the Relationship between Financial Disclosure and Security Price Variability," isnznsi_st Aeeenntin9_ene_fieenenieeI 5. 1983 PPo 49-74. Morse, D., "Price and Trading Volume Reaction Surrounding Earnings Announcements: A Closer Examination," lQBIflQl_Q£_A§QQQDEIBQ_B§§§QIQnI Antumn 1981: 113 Ohlson, J. A.,"On Financial Disclosure and the Behavior of Security Prices," Ienrnal_ef_Aeeennt1n9_ene_neeneniee. 1, 1979, pp. 211-232. Patell, J. M., "Corporate Forecasts of Earnings Per Share and Stock Price Behavior: Empirical Tests," antnn1_sf AQQQHntinQ_BEE£sIQh Autumn 1976- PP- 245 276 R0, Byung T.,"The Adjustment of Security Returns to the Disclosure of Replacement Cost Accounting Information, " Jenrnel_9f_Aeeeunt1n9_and_fleenem1ee 2. 1980 pp. 159- 189. Securities and Exchange Commission, Accounting Series Release No. 190, March 1976. Sharpe, W., "Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk," {surnal o: Einance, September 1964, pp. 425-442. Wild, J., "Precision of Information and the Stock Price Reaction to Earnings Releases," Unpublished Manuscript, Michigan State University, February 1989.