Disclosure Quality and Earnings Management

Transcription

1 Farsiarticles.com Iran-article.ir Iranarticles.com Disclosure Quality and Earnings Management Gerald J. Lobo Arthur Andersen Professor of Accounting Department of Accountancy & Taxation Bauer College of Business University of Houston 334 Melcher Hall Houston, Texas Jian Zhou Assistant Professor of Accounting School of Management SUNY at Binghamton PO Box 6000 Binghamton, NY (Phone) (Fax) Gerald J. Lobo, and Jian Zhou Disclosure quality and earnings management. Asia-Pacific Journal of Accounting and Economics V8 (1): We acknowledge the helpful comments provided by Anwer S. Ahmed, Steve Fortin, Mary Harris, Kiridaran Kanagaretnam, Dong Hoon Yang, participants at the 2001 Asia-Pacific Journal of Accounting and Economics Symposium in Hong Kong, and workshop participants at Hong Kong Baptist University, University of Massachusetts at Boston, McGill University, and State University of New York Binghamton. We extend a special thank you to the reviewer, Bin Srinidhi, for his many insightful comments and suggestions. We also thank Feixue Yan for help with data collection. Jian Zhou and Gerald Lobo acknowledge the financial support provided by the George E. Bennett Accounting Research Center and the School of Management Research Committee, respectively, at Syracuse University.

2 Farsiarticles.com Iran-article.ir Iranarticles.com Disclosure Quality and Earnings Management Abstract This study examines the relationship between disclosure quality and earnings management. Corporate disclosure and earnings management are both subject to managers discretion; therefore, managers are likely to consider their interaction when exercising managerial discretion. This study employs a simultaneous equations model to test the hypothesis that disclosure quality and earnings management are negatively related. It uses ratings published by the Association for Investment Management and Research to measure corporate disclosure, and discretionary accruals from the modified Jones model to measure earnings management. Consistent with theoretical predictions, the empirical analysis indicates that there is a statistically significant negative relationship between corporate disclosure and earnings management. Firms that disclose less tend to engage more in earnings management and vice versa. This result holds even after controlling for the effects of potentially confounding variables, and for all three components of corporate disclosure: annual disclosure, quarterly disclosure, and investor relations disclosure. By documenting a consistent negative relationship between corporate disclosure and earnings management, the study provides evidence on how management uses the flexibility afforded it under current minimum disclosure requirements to exercise discretion in reporting earnings. This has implications for the interpretation of and information conveyed by reported accounting earnings. The result that more informative corporate disclosure is related to less earnings management is also consistent with one of the SEC s objectives in encouraging companies to disclose more information.

3 Disclosure quality and Earnings Management 1. Introduction This paper presents empirical evidence on the relation between firms financial disclosure and earnings management. Prior research indicates that corporate disclosure is related to information asymmetry between investors and managers [e.g., Glosten and Milgrom(1985); Lang and Lundholm(1993); Welker(1995); Lang and Lundholm(1996)]. Prior research also demonstrates a relationship between information asymmetry and earnings management [e.g., Dye(1988); Trueman and Titman(1988); Richardson(1998)]. Drawing upon the results of these two streams of research, we hypothesize that the extent of earnings management is negatively related to the informativeness of corporate disclosure. We use rankings of firms overall disclosures published by the Corporate Information Committee of the Association for Investment Management and Research as our measure of the informativeness of corporate disclosure policies. This measure has been employed in prior research on the effects of disclosure [e.g., Lang and Lundholm (1993); Welker (1995); Lang and Lundholm (1996); Sengupta (1998); Healy, Hutton and Palepu (1999)]. We use discretionary accruals estimated with the modified Jones model as our measure of earnings management [Dechow, Sloan, Sweeny (1995)]. Our empirical analysis is conducted on a sample of 1,444 firmyear observations over the period. Consistent with our hypothesis, we find a significant negative relationship between corporate disclosure and discretionary accruals. Our results are consistent with the theoretical predictions and empirical findings of prior research. They provide evidence on how management may use the flexibility afforded it under current minimum disclosure requirements to exercise discretion in corporate disclosure and in

4 reporting earnings. This has implications for the interpretation of and information conveyed by reported accounting earnings. Our results also provide empirical support for the SEC s approach to curbing earnings management. The SEC has been exhorting companies to disclose more information citing reduced earnings management as one of the potential benefits of more informative disclosure.the rest of this paper is organized as follows. We discuss the related literature and present our hypotheses in the next section. In section 3, we describe the data, the variables used, and the empirical procedures. We present and interpret the results of the empirical analysis in section 4. This is followed by a summary and conclusions in section Background and Hypothesis Development Two streams of research that include both analytical and empirical work provide the underlying rationale for our hypothesis. The first identifies the relation between information asymmetry and disclosure quality, while the second links earnings management to information asymmetry. Together, these two lines of research yield predictions about the relation between disclosure quality and earnings management. 2.1 Corporate Disclosure and Information Asymmetry Glosten and Milgrom (1985) have modeled the relation between corporate disclosure and information asymmetry. Their model demonstrates that information asymmetry decreases as the level of corporate disclosure increases. Welker (1995) provides empirical evidence consistent with this result. His findings indicate that information asymmetry, measured as the bid-ask spread, is reduced and market liquidity increased as the level of disclosure is increased. Lang and Lundholm (1993) report that disclosure levels are higher for larger firms, for firms that perform 2

5 Farsiarticles.com Iran-article.ir Iranarticles.com well, and for firms with weaker earnings-return relations. 1 They use the correlation between earnings and returns as a measure of information asymmetry. A low correlation indicates that little information about firm value is captured by the mandatory earnings disclosure, so the remaining information asymmetry is high. Lang and Lundholm (1993) interpret this result as being consistent with firms having greater incentives to disclose more information to mitigate adverse selection when there is greater information asymmetry. Their results are also consistent with the theoretical predictions of Glosten and Milgrom (1985) that increased disclosure is associated with reduced information asymmetry. Lang and Lundholm (1996) provide evidence that firms with more informative disclosure policies have a larger analyst following, more accurate analyst earnings forecasts, less dispersion among individual analyst forecasts, and less volatility in forecast revisions. If dispersion among individual analyst forecasts is a valid measure for information asymmetry, then these results imply that more informative disclosure policies reduce information asymmetry. Firms have many incentives to disclose more information. Verrecchia (1983, 1990a) demonstrates that even if disclosure is costly because of product market consequences, managers may voluntarily expand disclosure to correct undervaluation by the capital market. Additionally, expanded disclosure can improve intermediation for a firm s stock in the capital market. Studies by Barry and Brown (1984, 1985), Merton (1987), Diamond and Verrecchia (1991), and Kim and Verrecchia (1994) suggest that increased voluntary disclosure reduces information asymmetries between management and outside investors, and among different types of investors. This, in turn, improves liquidity in a firm s stock, making it more attractive to institutional investors. Healy, Hutton and Palepu (1999) find that disclosure rating increases are accompanied by increases in 1 Lang and Lundholm (1993) also find that disclosures are greater for firms that issue securities. 3

6 sample firms stock returns, institutional ownership, analyst following, and stock liquidity. In sum, expanded disclosure benefits firms in many ways. Managers wishing to enhance the value of their firms may do so by communicating their superior, private information through increased disclosure. However, as discussed in the following subsection, managers wishing to retain flexibility to engage in opportunistic earnings management have incentives to limit disclosure because the effectiveness of their earnings management efforts depends on the level of information asymmetry between management and related stakeholders. 2.2 Earnings Management and Information Asymmetry Several analytical models demonstrate that the extent of opportunistic earnings management increases with the level of information asymmetry. For example, Dye (1988) and Trueman and Titman (1988) show analytically that the existence of information asymmetry between management and shareholders is a necessary condition for earnings management. Dye (1988) assumes that there exist overlapping generations of shareholders. Selling shareholders allow management to follow a certain earnings management strategy to create a favorable impression on the buying group. In his model, the manager has an information advantage over shareholders. Consequently information asymmetry is a necessary condition for earnings management in this setting. As noted in Schipper (1989, p. 95), an additional condition which must be met for earnings management to exist in an analytical model is that the asymmetry in information persists; one assumption that permits this persistence is a form of blocked communication 2 that cannot be eliminated by changing the contractual arrangements. This is so because shareholders cannot perfectly observe a firm s performance and prospects in an 2 Blocked communication means that managers cannot communicate all their private information; certainly, some communication is permitted. 4

7 environment in which they have less information than management. In such an environment, management can use its flexibility to manage reported earnings. Furthermore, management s discretionary ability to manage earnings increases as the information asymmetry between management and shareholders increases. Richardson (1998) provides empirical evidence consistent with this line of reasoning. He finds that the extent of information asymmetry, as measured by the bid-ask spread and the dispersion in analysts forecasts, is positively related to the degree of earnings management. Imhoff and Thomas (1994) provide evidence that analysts quality ratings are positively related to the conservatism of accounting estimates and methods, and to the amount of disclosure provided about the details underlying reported numbers. An implication of their findings is that firms with more conservative accounting estimates and methods (defined in our study as firms which engage in less earnings management) disclose more information. If firms with more conservative reporting engage in less earnings management, then this suggests a negative relationship between earnings management and corporate disclosure. That is, firms engaging in less earnings management disclose more information and firms disclosing more information engage in less earnings management. 2.3 Statement of Hypothesis Taken together, the two lines of research described in sections 2.1 and 2.2 suggest that managers of firms that disclose more information have less flexibility to manage earnings. An alternative way of stating this is that shareholders of firms that have more informative disclosure 5

8 policies 3 can more easily detect earnings management; therefore, management is less likely to engage in such behavior. Our hypothesis, stated in its alternative form, is as follows: H a : Corporate disclosure and earnings management are negatively related. Although the focus of our study is on opportunistic earnings management as defined in Healy and Wahlen (1999) 4, we note that earnings management may also be used to communicate managers superior, private information about their firms value relevant attributes. For example, Gul, Leung and Srinidhi (2000) report that managers of firms with greater investment opportunities use earnings management to signal their future opportunities for growth. More specifically, they find that investment opportunity set induced discretionary accruals are employed for information communication Therefore, Hypothesis H a is far from obvious. Its validation will support our view that the negative relationship between disclosure and opportunistic earnings management will prevail in equilibrium over the potential positive relationship between information earnings management and disclosure. 3. Data Requirements and Variable Measurement 3 Some might argue that the difference in disclosure is due to the adverse impact of increased disclosure on the product market. The adverse impact of disclosure on the product market (for example, disclosure of proprietary information has an adverse impact on the firm's future earnings) is controlled because we investigate the relationship between earnings management and corporate disclosure differences within industries. That is, we assume that firms in the same industry have similar concerns about the effects of corporate disclosure on the product market. This suggests that any observed differences in disclosure are more likely to result from concerns about earnings management. 4 Healy and Wahlen (1999) define earnings management as follows: Earnings management occurs that managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on the reported accounting numbers. 6

9 Farsiarticles.com Iran-article.ir Iranarticles.com 3.1 Measuring Corporate Disclosure We use ratings of corporate disclosure reported by the Corporate Information Committee (CIC) of the Association for Investment Management and Research during the period. These ratings reflect assessments of analysts specializing in specific industries about the informativeness of disclosures made by firms in their respective industries. Analysts evaluate the timeliness, detail and clarity of the corporate disclosure. The disclosure scores are based on a weighted average of analysts assessments of three dimensions of disclosure: annual published information (which includes annual reports and 10-Ks), quarterly and other published information (which include quarterly reports, press releases and proxy statements), and investor relations and related aspects (which include accessibility of management and management s responsiveness to analysts questions). The weights assigned to these three categories range from percent, percent, and percent, respectively. 5,6 The corporate disclosure ratings employed in our study have also been used by Lang and Lundholm (1993; 1996), Welker (1995), Sengupta (1998), and Healy, Hutton and Palepu (1999). Because different groups of analysts rate disclosure policies of firms in different industries, we industry-adjust the ratings by subtracting the mean rating for the industry to which the specific firm belongs. This allows us to pool firms across industries while allowing for intraindustry variation across firms. 5 The appendix to each issue of the AIMR report describes the evaluation criteria in considerable detail. 6 The issue includes an evaluation of corporate reporting practices of 374 public companies from 18 industries by 164 analysts. There is considerable variability in the number of companies evaluated in each industry and the number of analysts following each industry. For example, 4 firms are evaluated in the petroleum industry (independent refining companies) and 31 firms in food, beverage and tobacco industry. Five analysts evaluated the homebuilding industry, whereas 19 analysts evaluated the food, beverage and tobacco industry. 7

10 3.2 Measuring Earnings Management Consistent with the extant literature (see, for example, Becker et al. (1998)), we use discretionary accruals to measure the extent of firms earnings management. We employ the model identified by Dechow, Sloan and Sweeny (1995) to estimate discretionary accruals. This model, which is commonly referred to as the Modified Jones model, has been shown to be the most powerful among competing models for detecting earnings management. Subramanyam (1996) presents evidence that discretionary accruals estimated with this model are priced by the market. However, the coefficient on discretionary accruals is lower in magnitude than the coefficient on non-discretionary earnings. This evidence suggests that discretionary accruals are viewed by market participants as a less reliable component of earnings, which implies that discretionary accruals are more likely to be subject to managers manipulation and, therefore, are valid measures of earnings management. To measure discretionary accruals, we first have to measure total accruals. We employ two alternative measures of total accruals in this study: the traditional balance sheet approach that has been used in the majority of prior studies and the cash flow approach proposed by Collins and Hribar (1999). This facilitates an examination of the sensitivity of our results to the accruals measure. where: Under the traditional balance sheet approach, total accruals are measured as follows: TACC it = CA it - CL it - Cash it + STDEBT it - DEPTN it (1) CA it = change in current assets during period t (Compustat #4), CL it = change in current liabilities during period t (Compustat #5), Cash it = change in cash and cash equivalents during period t (Compustat #1), STDEBT it = change in the current maturities of long-term debt and other short-term debt included in current liabilities during period t (Compustat #34), 8

11 DEPTN it = depreciation and amortization expense during period t (Compustat #14). Collins and Hribar (1999) find that studies using the balance sheet approach to test for earnings management are potentially contaminated because the balance sheet approach to measuring total accruals introduces significant measurement error in the accruals. They suggest that the cash flow method is a better way to calculate total accruals and document evidence to support their view. Under the cash flow approach, total accruals are measured as follows: where: TACC it = EBXT it - OCF it (2) EBXT it = earnings before extraordinary items and discontinued operations for period t (Compustat #18), OCF it = operating cash flow for period t (Compustat #308). 7 We estimate discretionary accruals (DACC) as the difference between total accruals (TACC) and nondiscretionary accruals (NDACC). To estimate nondiscretionary accruals, we first estimate the Modified Jones model, which is formulated as follows: where: TACC it = α 1 (1/A i,t-1 ) + α 2 ( REV it - REC it ) +α 3 PPE it + ε it (3) TACC it = total accruals for firm i in year t divided by total assets for firm i at the end of year t-1, REV it = change in revenue for firm i in year t divided by total assets for firm i at the end of year t-1, REC it = change in net receivable for firm i in year t divided by total assets for firm i at the end of year t-1, PPE it = property, plant and equipment for firm i in year t divided by total assets for firm i at the end of year t-1. 7 Compustat defines net cash flow from operating activities as the change in cash from all items classified in the Operating Activities section on a Statement of Cash Flows (Format code = 7.000). This item includes changes in operating assets and liabilities. Increases in cash are presented as positive numbers, decreases are presented as negative numbers. Also, this item is not available for banks, utilities, life insurance, and property and casualty companies. 9

12 Equation (3) is estimated each year using ordinary least squares estimation. The estimates of α 1, α 2, and α 3 obtained from these regressions are then used to estimate nondiscretionary accruals as follows: NDACC it = â 1 (1/A i,t-1 ) + â 2 ( REV it - REC it ) + â 3 PPE it (4) Finally, discretionary accruals are estimated as: DACC it = TACC it - NDACC it. (5) 3.3 Model for Testing the Hypothesis Our hypothesis that corporate disclosure and earnings management are negatively related is based upon the relations of each of these variables to information asymmetry. Whether management s disclosure decision results from its desire to allow itself flexibility to manage earnings, or whether management s ability to manage earnings results from its choice of disclosure policy is unclear. Both of these cause-and-effect relations are feasible, suggesting that corporate disclosure and earnings management decisions are likely to be jointly endogenously determined. To account for this potential simultaneity, we estimate the relation between disclosure quality and earnings management using the following simultaneous equation system: where: DACC it = α 0 + α 1 DP it + α 2 CRP it + α 3 FRP it + α 4 LEV it + α 5 SIZE it + ε it (6) DP it = β 0 + β 1 DACC it + β 2 SIZE it + β 3 VWRET it + ε it (7) DACC = discretionary accruals, DP = disclosure policy, 10

13 Farsiarticles.com Iran-article.ir Iranarticles.com CRP = current industry relative performance, based on current period net income deflated by beginning total assets, FRP = future industry relative performance, based on next period net income deflated by beginning total assets, LEV = total liability over total assets, SIZE = market value of the firm at the beginning of the year, VWRET = market adjusted stock return. Equation (6) specifies discretionary accruals, DACC, as a function of disclosure policy, DP, and four exogenous variables that prior research indicates are related to discretionary accruals. Equation (7) specifies disclosure policy as a function of discretionary accruals and two exogenous variables that have been identified in prior research. We use two-stage least squares (2SLS) to estimate this two-equation system. In the first stage, we regress DP on DACC and all the exogenous variables in equations (6) and (7), i.e., on SIZE, VWRET, CRP, FRP and LEV. In the second stage, we estimate equation (6) using the fitted value of DP from the first stage regression. The use of ordinary least squares in the second stage yields consistent estimates of the parameters in equation (6) because the fitted value of DP from the first stage is uncorrelated with the error term in the second stage regression. We perform an analogous estimation procedure for equation (7). In addition to disclosure policy, equation (6) includes four exogenous variables that prior research indicates are related to discretionary accruals. Fudenberg and Tirole (1995) develop a model, which shows that managers have incentives to reduce current earnings when they are high and future earnings are expected to be low, and to increase current earnings when they are low and future earnings are expected to be high. 8 DeFond and Park (1997) present empirical evidence consistent with the Fudenberg and Tirole (1995) model. They document that discretionary 8 This also indicates a signaling of future earnings by managing current earnings that is consistent with efficient earnings management (Gul, Leung and Srinidhi, 2000). 11

14 accruals are negatively related to current performance relative to the industry and positively related to next period s industry relative performance. This is because managers save income (through negative discretionary accruals) for future periods when current industry relative performance is good. When future industry relative performance is expected to be high, managers shift income (through positive discretionary accruals) to the current period. Therefore, the extent of earnings management as measured by discretionary accruals is related to both current and future relative performance. DeFond and Park (1997) also report that leverage is negatively and firm size positively related to discretionary accruals. We include current relative performance, future relative performance, leverage, and firm size in equation (6) to control for their effects on discretionary accruals. Lang and Lundholm (1993) provide evidence that the informativeness of corporate disclosure policies is increasing in firm size and firm performance. Larger firms tend to disclose more because of greater demand for information [see Atiase (1980)] and/or because their average cost of disclosure is decreasing in firm size. Furthermore, since larger firms are monitored more closely by a large number of investors and analysts, they are less likely to engage in earnings management. Managers of firms that are performing well are likely to provide more information than are managers of poor-performing firms. Analytical models of voluntary disclosure in the presence of adverse selection demonstrate a positive relation between disclosure and firm performance Data Requirements and Sample Description 9 McNichols (1984), Dye (1985; 1988) and Verrecchia (1990b) are examples of such research. 12

15 To be included in the sample, firms are required to have disclosure policy scores available in the AIMR reports. These disclosure policy ratings serve as our dependent variable. Additionally, sample firms are required to have data available in Compustat for estimating discretionary accruals, and to have data available in Compustat/CRSP for measuring the control variables. The effects of these data requirements on sample size are summarized in Table 1. Our initial sample consists of 2,531 firm-years. Software service firms are excluded because the AIMR report does not present disclosure scores for these firms. We exclude financial services firms because their accrual processes differ considerably from the Modified Jones model. Similarly, natural gas is a regulated industry whose accruals also follow a different process and is excluded. Additionally 391 firms do not have the requisite data in Compustat. This left us with 1,444 firm-year observations for the empirical analysis. There were 315 observations in 1990, 299 observations in 1991, 283 observations in 1992, 201 observations in 1993, 178 observations in 1994, and 168 observations in Table 2 reports descriptive statistics for the dependent, independent, and control variables. The disclosure policy rating is a weighted average of the scores received for each of the three disclosure dimensions evaluated. The score received for a given measure is computed as the ratio of points received to total points assigned to that dimension. This allows us to aggregate scores across industries with different weights assigned to each dimension, while still preserving any inter-industry differential weighting across the three dimensions. The mean disclosure policy rating for our sample is This is similar to the mean rating of 70 for Lang and Lundholm s (1993) sample, which used data from The lower quartile of disclosure is 61, while the upper quartile of disclosure is The mean and median for the three components of 13

16 disclosure are as follows: and 72.5, respectively, for annual disclosure; and 71.6, respectively, for quarterly disclosure; and and 74.4, respectively, for investor relations disclosure. The mean and median discretionary accruals calculated using the balance sheet approach are and 0.006, respectively, while the corresponding values for the cash flow approach are and 0.00, respectively. 10 Median firm size is $2.26 billion, indicating that our average sample firm is quite large. The lower quartile and upper quartile of firm size are $ million and $6.1 billion, respectively. The mean and median leverage are 0.58 and 0.57, respectively. The average return is 15% and the mean and median of market-adjusted return are 0.02 and 0.02, respectively. 4. Empirical Results 4.1 Simple Correlations Because our primary focus is on the direction of the relationship between earnings management and disclosure policy and not on the magnitude of the effect of one variable on the other, we conduct our analysis using ranked data. Using the procedure adopted in Lang and Lundholm (1993; 1996), we first rank the dependent, independent and control variables by industry and by year. We then convert the ranks to percentiles using the transformation (rank - 1)/(number of firms - 1). This transformation yields the percentile equivalent of a firm s rank within its industry. For each variable, the highest ranked firm in each industry receives a zero and the lowest ranked firm receives a one. Table 3 presents correlations between the transformed dependent, independent and control variables. Consistent with our hypothesis, disclosure policy is significantly negatively 14

17 related to discretionary accruals. The correlation coefficient of is significantly less than zero (p = 0.01). As discussed earlier, disclosure policy is significantly positively related to firm size (p = 0.01). This result is consistent with the results of Lang and Lundholm (1996). Generally consistent with previous literature, disclosure is significantly positively correlated with stock return. As expected, the balance sheet and cash flow based measures of discretionary accruals are highly positively correlated. Consistent with the results of DeFond and Park (1995), each of these measures is significantly negatively related to current relative performance. However, neither variable is significantly positively related to future relative performance as expected. Furthermore, both measures of discretionary accruals are significantly negatively correlated with leverage but only the cash flow based measure of discretionary accruals is reliably negatively correlated with firm size. 4.2 Regression Analysis Aggregate Disclosure Ratings Panels A and B of Table 4 present the results of estimating equations (6) and (7) as a system of simultaneous equations. Panel A reports results for discretionary accruals measured using the balance sheet approach and panel B presents analogous results for the cash flow approach to measuring discretionary accruals. Both panels use aggregate disclosure ratings as the measure of disclosure policy. To examine the sensitivity of our estimation results to the choice of simultaneous equation estimation, we present ordinary least squares (OLS) estimation results of equations (6) and (7) in panel C. 10 These values are similar in magnitude to those reported in Subramanyam (1996). 15

18 The 2SLS estimation results for equation (6) that are contained in panels A and B provide strong evidence in support of our hypothesis. α 1, the coefficient relating disclosure policy and discretionary accruals, is significantly less than zero at the p = 0.01 level. This indicates that firms that disclose more information engage less in earnings management. The coefficient estimates of the exogenous variables in equation (6) are also consistent with predictions. Current relative performance (CRP) and leverage (LEV) are each significantly negatively related to discretionary accruals, while future relative performance (FRP) and firm size (SIZE) are positively related. These results are consistent with DeFond and Park (1997). Similar results are obtained for both the balance sheet and the cash flow approaches to measuring discretionary accruals. This shows that our results are robust to the choice of accruals model. Estimation results for equation (7) also provide strong support for our hypothesis. As predicted, β 1 is significantly negative, suggesting that firms that engage more in earnings management disclose less information or information of lower quality. Recall from the discussion in section 2 that the effectiveness of earnings management relies on the existence of information asymmetry between management and related stakeholders. By disclosing less information or lower quality information, managers ensure a higher level of information asymmetry, thus providing themselves with greater flexibility to engage in earnings management. It would be far more difficult for related stakeholders to undo firms earnings management because of the higher level of information asymmetry. Both firm size (SIZE) and market adjusted return (VWRET) are also significantly positively related to disclosure policy as predicted. These results are consistent Lang and Lundholm (1993), who find that disclosure is positively related to firm size and firm performance. They indicate that large firms and well-performing firms tend to 16

19 Farsiarticles.com Iran-article.ir Iranarticles.com disclose more information or information of higher quality. Once again, the estimation results for equation (7) are highly consistent across the balance sheet and cash flow approaches. Panel C presents the OLS estimation results for equations (6) and (7). Results are reported only for the cash flow approach because they differ little from the results of the balance sheet approach. Accordingly, we compare these OLS results to the 2SLS results for the cash flow approach that are contained in panel B. Comparison of these two sets of estimation results indicates little difference in the coefficient estimates, β 2 and β 3, in equation (6). The estimated OLS coefficient on discretionary accruals, β 1, is 0.07 compared to an estimate of 0.12 when the effects of potential simultaneity are considered. The 2SLS estimation procedure reduces measurement error, thereby resulting in a larger absolute value estimate of β 1. However, regardless of the bias in the OLS estimate, β 1 is still significantly less than zero as hypothesized. While the estimation bias from using OLS is not as pronounced in equation (7), it is more severe in equation (6). The OLS estimate of α 1 in equation (6) is only 0.02, whereas the corresponding 2SLS estimate is Furthermore, the OLS estimate is not sufficiently negative to support our hypothesis of a negative relation between discretionary accruals and disclosure policy, whereas the 2SLS estimate indicates rejection at the 0.01 level of significance. These results demonstrate that the simultaneity bias is sufficiently large to alter the conclusion of our hypothesis test Component Disclosure Ratings Our primary focus so far has been on the relationship between aggregate measures of disclosure policy and earnings management. Recall from our earlier discussion that these aggregate disclosure policy ratings are weighted averages of ratings in three areas of disclosure: 17

20 annual published information, quarterly and other published information, and other aspects such as investor relations. We examine the relation between each of these dimensions of disclosure policy and earnings management using the simultaneous equation approach. This will allow us to identify whether the observed negative relationship between disclosure policy and earnings management varies across these components. Table 5 presents results of the relation between each dimension of disclosure and earnings management. The models are analogous to those used for studying the relation between aggregate disclosure policy and earnings management. The estimation results indicate that the negative relation observed between the aggregate measure of disclosure policy and discretionary accruals is similar across the three components of corporate disclosure. This is as expected given the significant positive relationship among the three components of corporate disclosure and the significant positive relationship between each of the three components of corporate disclosure and the total disclosure. 5. Summary and Conclusions This study examines the relationship between disclosure quality and earnings management. It hypothesizes that disclosure quality and earnings management are negatively related. It uses ratings published by the Association for Investment Management and Research to measure disclosure quality, and discretionary accruals from the Modified Jones model to measure earnings management. The empirical analysis indicates that corporate disclosure and earnings management are significantly negatively related. Firms with lower disclosure ratings tend to engage more in earnings management and firms that engage more in earnings management tend 18

21 to have lower quality disclosures. This result holds even after controlling for the effects of potentially confounding variables. This empirical evidence presented in this study facilitates discriminating between two competing hypotheses that have opposite implications for the relation between earnings management and corporate disclosure quality. If earnings management is opportunistic, then the predicted relation is negative. Alternatively, if earnings management is for information communication, then the predicted relation is positive. The significant negative relation between earnings management and disclosure quality documented in this study is consistent with opportunistic earnings management. The finding of a negative relation between earnings management and disclosure policy suggests that the extent to which firms engage in earnings management is an important determinant of their decision to be more or less forthcoming in their disclosure policies. This has implications for policy-making bodies that set minimum disclosure requirements for firms, because these requirements may play a significant role in a firm s ability to manage its earnings. It also provides empirical support for the SEC s approach to earnings management. The SEC has been urging companies to disclose more information in the face of rising earnings management activity. Our results show that more disclosure has a constraining effect on firms earnings management. Another implication of our findings is that limiting accounting discretion will increase the informativeness of earnings because it constrains earnings management and increases the comparability of earnings across firms. This is consistent with the theoretical predictions of Fishman and Hagerty (1990) who show that, under certain circumstances, rules that limit discretion increase the informativeness of disclosures and thus improve economic decisions. Our 19

22 empirical findings also serve as a triangulation of Imhoff and Thomas (1994) results. They find that analysts quality ratings are positively related to the conservatism of accounting estimates and methods, and to the amount of disclosure provided about the details underlying reported numbers. A triangulation of their result is that firms with conservative accounting estimates and methods (in our case, firms which engage in less earnings management would be considered more conservative) disclose more information. Finally, our results are also consistent with the theoretical prediction of Verrecchia (1990b) that an increase in the quality of private information received by the manager results in more disclosure on his/her part. When managers engage less in earnings management, the information quality of the signal (earnings) is higher, and managers disclose more information. 20

23 References Atiase, R.K Predisclosure informational asymmetries, firm capitalization, financial reports, and security price behavior. Ph.D. dissertation, University of California, Berkeley. Barry, C.B. and S.J. Brown Differential information and the small firm effect. The Journal of Financial Economics 13: Barry, C.B. and S.J. Brown Differential information and security market equilibrium. Journal of Financial and Quantitative Analysis 20 (Dec): Becker, C., M. DeFond, J. Jiambalvo, and K. R. Subramanyam The effect of audit quality on earnings management. Contemporary Accounting Research 15 (Spring): Bradshaw, M., S. Richardson and R. Sloan, 2000, Do analysts and auditors use information in accruals? Working Paper. University of Michigan. Collins, D. and P. Hribar Errors in estimating accruals: Implications for empirical research. Working Paper. University of Iowa. Dechow, P., R. Sloan, and A. Sweeny Detecting earnings management. The Accounting Review 70 (April): DeFond, M. and C. Park Smoothing income in anticipation of future earnings. Journal of Accounting and Economics 23 (July): Diamond, D.W. and R.E. Verrecchia Disclosure, liquidity, and the cost of capital. The Journal of Finance 66 (September): Dye, R Disclosure of nonproprietary information. Journal of Accounting Research 23: Dye, R Earnings management in an overlapping generations model. Journal of Accounting Research 26: Fishman, M. and K. Hagerty The optimal amount of discretion to allow in disclosure. Quarterly Journal of Economics 105, 2 (May). Fudenberg, K. and J. Tirole A theory of income and dividend smoothing based on incumbency rents. Journal of Political Economy 103: Glosten, L. and P. Milgrom Bid, ask, and transaction prices in a specialist market with heterogeneouly informed traders. Journal of Financial Economics 26 (March):

24 Gul, F., S. Leung, B. Srinidhi The Effect of Investment Opportunity Set and Debt Level on Earnings-Returns Relationship and the Pricing of Discretionary Accruals. Working Paper. City University of Hong Kong. Healy, P., A. Hutton, K. Palepu Stock Performance and Intermediation Changes Surrounding Sustained Increases in Disclosure. Contemporary Accounting Research (Fall): Healy, P. and J. Wahlen A review of the earnings management literature and its implications for standard setting. Accounting Horizons (December): Imhoff, E. Jr. and J. Thomas Accounting Quality. In Asset Valuation. Stephen A. Butler, ed. The Center for Economic and Management Research, The University of Oklahoma: Kim, O. and R.E. Verrecchia Market liquidity and volume around earnings announcements. Journal of Accounting and Economics, 17 (January): Lang, M. and R. Lundholm Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research 31 (Autumn): Lang, M. and R. Lundholm Disclosure quality and Analyst Behavior. The Accounting Review 71: McNichols, M The anticipation of earnings in securities markets. Ph.D. dissertation. University of California, Los Angeles. Merton, R.C A simple model of capital market equilibrium with incomplete information. The Journal of Finance. (July): Richardson, V Information Asymmetry and Earnings Management: Some Evidence, Working paper, University of Kansas. Schipper, K Commentary on earnings management. Accounting Horizons 3: Sengupta, P Corporate disclosure quality and the cost of debt. The Accounting Review 73: Subramanyam, K.R The pricing of discretionary accruals. Journal of Accounting and Economics 22: Trueman, B. and S. Titman An explanation for accounting income smoothing, Journal of Accounting Research 26 (supplement):

25 Farsiarticles.com Iran-article.ir Iranarticles.com Verrecchia, R.E Discretionary disclosure. Journal of Accounting and Economics 5 (December): Verrecchia, R.E. 1990a. Endogenous proprietary costs through firm interdependence. Journal of Accounting and Economics 12 (January): Verrecchia, R.E. 1990b. Information quality and discretionary accruals. Journal of Accounting and Economics 12: Welker, M Disclosure policy, information asymmetry, and liquidity in equity markets. Contemporary Accounting Research 11 (Spring):

26 Table 1 Effects of Selection Criteria on Sample Size Total Initial sample Canadian banking NA NA NA Banking NA NA NA 200 Savings institutions 7 NA 7 NA NA NA 14 Financial services NA NA 56 Insurance Natural gas Diversified companies NA NA 47 Software services R R R R Computer and electronics 15* 15* NA NA NA NA 30 Data are not available from Compustat/CRSP Remaining sample NA: not applicable in that year R: ranked in that year *: not ranked in that year 24

27 Farsiarticles.com Iran-article.ir Iranarticles.com Table 2 Descriptive Statistics for Endogenous and Exogenous Variables Variable Mean Std Dev Lower Quartile Medium Upper Quartile DP ANN QRT INV DACCBS DACCCF CRP FRP SIZE(Millions) LEV RET VWRET Variable definitions: DP: disclosure policy ANN: annual disclosure QRT: quarterly disclosure INV: investor relations disclosure DACCBS: discretionary accruals calculated using balance sheet approach DACCCF: discretionary accruals calculated using cash flow approach CRP: current industry relative performance, based on net income deflated by beginning total assets FRP: future industry relative performance, based on net income deflated by beginning total assets SIZE: market value of the firm at the beginning of the year LEV: total liability over total assets RET: current year s return VWRET: market adjusted stock return 25

28 Table 3 Correlation among Endogenous and Exogenous Variables* DACCBS DP ANN QRT INV CRP FRP LEV SIZE VWRET DACCCF (0.03) (0.02) (0.66) (0.19) 0.02 (0.52) DACCBS (0.02) (0.02) (0.28) (0.03) (0.40) (0.02) (0.28) DP ANN (0.57) QRT (0.22) (0.03) INV CRP FRP LEV (0.73) (0.04) SIZE (0.18) * Figures in parentheses are p-values Variable definitions: DP: disclosure policy ANN: annual disclosure QRT: quarterly disclosure INV: investor relations disclosure DACCBS: discretionary accruals calculated using the balance sheet approach DACCCF: discretionary accruals calculated using the cash flow approach CRP: current industry relative performance, based on net income deflated by beginning total assets FRP: future industry relative performance, based on net income deflated by beginning total assets SIZE: market value of equity at the beginning of the year LEV: total liabilities over total assets RET: current year s return VWRET: market adjusted annual return 26

29 Farsiarticles.com Iran-article.ir Iranarticles.com Table 4 Relation Between Overall Disclosure Policy and Discretionary Accruals Panel A: Simultaneous Equation Estimation (Two-Stage Least Squares Estimation) (Accruals are measured using the Balance Sheet Method) Equation (6): DACCBS it = α 0 + α 1 DP it + α 2 CRP it + α 3 FRP it + α 4 LEV it + α 5 SIZE it + ε it α 0 α 1 α 2 α 3 α 4 α 5 Coefficient t-statistics Equation (7): DP it = β 0 + β 1 DACCBS it + β 2 SIZE it + β 3 VWRET it + ε it β 0 β 1 β 2 β 3 Coefficient t-statistics Variable definitions: DP: disclosure policy ANN: annual disclosure QRT: quarterly disclosure INV: investor relations disclosure DACCBS: discretionary accruals calculated using the balance sheet approach DACCCF: discretionary accruals calculated using the cash flow approach CRP: current industry relative performance, based on net income deflated by beginning total assets FRP: future industry relative performance, based on net income deflated by beginning total assets SIZE: market value of equity at the beginning of the year LEV: total liabilities over total assets RET: current year s return VWRET: market adjusted annual return 27

30 Table 4 (continued) Relation Between Overall Disclosure Policy and Discretionary Accruals Panel B: Simultaneous Equation Estimation (Two-Stage Least Squares Estimation) (Accruals are measured using the Cash Flow Method) Equation (6): DACCCF it = α 0 + α 1 DP it + α 2 CRP it + α 3 FRP it + α 4 LEV it + α 5 SIZE it + ε it α 0 α 1 α 2 α 3 α 4 α 5 Coefficient t-statistics Equation (7): DP it = β 0 + β 1 DACCCF it + β 2 SIZE it + β 3 VWRET it + ε it β 0 β 1 β 2 β 3 Coefficient t-statistics Variable definitions: DP: disclosure policy ANN: annual disclosure QRT: quarterly disclosure INV: investor relations disclosure DACCBS: discretionary accruals calculated using the balance sheet approach DACCCF: discretionary accruals calculated using the cash flow approach CRP: current industry relative performance, based on net income deflated by beginning total assets FRP: future industry relative performance, based on net income deflated by beginning total assets SIZE: market value of equity at the beginning of the year LEV: total liabilities over total assets RET: current year s return VWRET: market adjusted annual return 28