Accounting Conservatism and the Relation Between Returns and Accounting Data

Transcription

1 Review of Accounting Studies, 9, , 2004 Ó 2004 Kluwer Academic Publishers. Manufactured in The Netherlands. Accounting Conservatism and the Relation Between Returns and Accounting Data PETER EASTON* The University of Notre Dame, 305A Mendoza College of Business, Notre Dame, IN JINHAN PAE School of Business, Queen s University, 228 Goodes Hall, 143 Union Street, Kingston, ON, Canada K7L 3N6 Abstract. This study adds change in cash investments and change in lagged operating assets to the regression of returns on earnings levels and earnings changes examined in Easton and Harris (1991). We argue that a positive coefficient on change in cash investments captures conservatism associated with investments in positive net present value projects the effects of which will not flow into the accounting statements until the expected future benefits are realized. A positive coefficient on change in lagged operating assets implies accounting conservatism associated with the application of accounting rules to operating assets in place. Our empirical results are, in general, consistent with these arguments. We examine differences in conservatism across samples with different market to book ratios, we compare firms with non-negative returns with firms with negative returns, we compare firms reporting losses with firms reporting profits, and we examine firms in different industries, firms with different levels of research and development expenditure, different amounts of depreciation, different amounts of advertising expense, and firms that adopt LIFO inventory valuation compared with those that adopt an alternative to LIFO. Keywords: accounting conservatism, returns, earnings JEL Classification: M41 Numerous studies use the regression of returns on earnings (deflated by beginningof-period stock price) and deflated earnings changes introduced by Easton and Harris (1991) as the basis for tests of the value relevance of accounting. For example, Alford et al. (1993) use the R 2 from this regression to compare the value relevance of GAAP across sixteen countries, and Francis and Schipper (1999) use the change in the R 2 and change in the coefficients on earnings and earnings changes as indications of the change in the value relevance of U.S. financial statements over time. The foundation of the Easton and Harris (1991) regression of returns on earnings (deflated by beginning-of-period stock price) and deflated earnings changes is a model that expresses price as a linear function of book value and earnings. The key contribution of this model, which is formally described in Ohlson (1995), is that it provides a role for the balance sheet as well as the income statement (which had been the focus of prior studies). The shortcoming, however, is that does not incorporate the conservative accounting principles under which the GAAP balance sheets are constructed. We estimate earnings-return regression specifications that not only *Corresponding author.

2 496 EASTON AND PAE recognize the balance sheet (with which the income statement articulates), but also conservatism in the balance sheet (with which conservatism in the income statement articulates). We identify two forms of conservative accounting and we modify the Easton and Harris (1991) regression accordingly. Since conservatism is a fundamental feature of accounting, these modifications may affect conclusions from studies of value relevance. We identify firm and sample characteristics that may suggest a priori that accounting is likely to be conservative and, hence, the modifications may be important in the design of empirical analyses. The first form of conservatism arises because accounting does not record the payoffs from positive net present value projects until the associated future sales have occurred. It follows that the benefits of new cash investments in positive net present value projects will not be captured in book value and earnings. Thus, new cash investments should be added to the pricing model and, hence, change in cash investments should be added to the returns regression. The second form of conservatism arises because accounting rules, choices and procedures (such as aggressive depreciation policy) may lead to an understatement of book value and accounting earnings in prior periods, in the current period and in future periods. 1 We argue that this form of conservatism suggests that lagged operating assets should be added to the pricing model and, hence, deflated lagged change in operating assets should be added to the returns regression. The argument rests on the idea that, since lagged operating assets captures the cumulative effects of conservatism at the beginning of the fiscal period, it captures the effect of conservatism on the other variables in the valuation model earnings and book value. Our empirical analyses are based on a sample of 54,313 firm-year observations from 1988 to We modify the Easton and Harris (1991) regression adding lagged change in operating assets and change in cash investments to earnings and earnings changes. The mean of the year-by-year estimates of the coefficients on change in lagged operating assets is not significantly different from zero while the mean of the estimate of the coefficient on change in cash investments is (t-statistic of 6.63). We find evidence of both forms of conservatism for firms where the market value of operating assets is high relative to their book value (this is consistent with use of the ratio of the market value of common equity to the book value of common equity as a proxy for accounting conservatism in Beaver and Ryan (2000)). Our empirical analyses demonstrate that the magnitude of the estimates of the coefficients on change in cash investments and change in lagged operating assets increases as the market to book ratio increases. Industry-specific analysis shows evidence of varying degrees of accounting conservatism across industries. Accounting conservatism is particularly evident in the pharmaceutical industry, for example. Basu (1997) suggests that accounting will be more (less) conservative for firms with good (bad) news over the fiscal period. Although his concept of conservatism differs from ours, we seek evidence of a difference in accounting conservatism between firms with non-negative returns (good news) and firms with negative returns (bad news). While the estimate of the coefficient on change in cash investments is significant for both net good news and net bad news firms, the estimate of the coefficient is significantly less for firms with net bad news than for firms with net good news. This

3 ACCOUNTING CONSERVATISM 497 suggests a greater tendency for firms with net good news to invest in positive net present value projects. The estimate of the coefficient on change in lagged operating assets is not significantly different from zero for firms with non-negative returns, and it is significantly negative for firms with negative returns. The negative coefficient on change in lagged operating assets suggests that, for these firms which lost value in the market, accounting recorded a smaller decline in value than did the market. Hayn (1995) focuses on the news in earnings rather than the news in returns to motivate an analysis of the returns/earnings relation for firms reporting losses compared with firms reporting profits. In order to examine the effects of losses on accounting conservatism, we partition the sample into profit and loss firms. Our empirical analyses suggest that there is no difference in accounting conservatism associated with the application of accounting rules between firms reporting losses and firms reporting profits; in fact, the estimates of the coefficient on lagged operating assets are not significantly different from zero for either the profit or loss subsample. The estimates of the coefficients on change in cash investments are positive for both profit and loss firms consistent with the notion that cash investments are generally in positive net present value projects. In general, we find pervasive evidence that change in cash investments provides significant incremental explanatory power for returns over earnings and earnings changes. This evidence is consistent with the notion that firms invest in positive NPV projects and book value and earnings do not capture the value of the investment until later periods. Although the evidence of incremental explanatory power in lagged operating assets is, by no means, pervasive, the estimate of the coefficient on operating assets is positive as expected for sub-samples of firms where the ratio of the market value of operating assets to the book value of operating assets is highest. The paper proceeds as follows. In the next section, we develop the empirical regression model that captures the effects of accounting conservatism. Section 2 describes the data and the sample selection procedure. Section 3 reports the results of the empirical analysis. We summarize these results and draw conclusions in Section Model We begin with a simple model that is pervasive in the recent empirical literature on the value relevance of accounting. This model expresses price p jt as a linear function of book value b jt and earnings x jt : p jt ¼ a 0 þ a 1 b jt þ a 2 x jt þ e jt : ð1þ Easton and Harris (1991) and Easton (2001) provide intuitive arguments (supported by the theoretical model in Ohlson (1995)) that suggest that weights a 1 and a 2 depend on the persistence/transitoriness of earnings. If earnings are permanent, the weight, a 1, on book value is low and the weight, a 2, on earnings is high. If earnings are transitory, the weight on book value is high and the weight on earnings is low. Neither Easton and Harris (1991) nor Ohlson (1995) permit conservative accounting.

4 498 EASTON AND PAE We identify two forms of conservative accounting and we modify Equation (1) accordingly. The first form of conservatism arises because accounting rules, choices, and procedures (such as aggressive depreciation policy) may lead to an understatement of book value and accounting earnings in prior periods, in the current period, and in future periods. Since these conservative accounting rules tend to affect operating assets much more than financial assets (which tend to be valued at close to their market value), we focus on conservatism in the valuation of operating assets and we add the lagged book value of operating assets oa jt)1 to Equation (1). Conservatism in the valuation of operating assets at the beginning of the period suggests that earnings of the current period and book value at the end of the current period will be understated and hence we would expect a positive weight on operating assets. On the other hand, if operating assets at the beginning of the period are over-valued, it is likely that earnings of the current period and book value at the end of the period will be overstated and the weight on operating assets will be negative. The second form of conservatism arises because accounting does not record the payoffs from positive net present value projects until the associated future sales have occurred. It follows that the benefits of new cash investments in positive net present value projects will not be captured in book value and earnings. Thus, cash investments ci jt should be added to the pricing model to obtain: p jt ¼ b 0 þ b 1 b jt þ b 2 x jt þ b 3 oa jt 1 þ b 4 ci jt þ e jt : ð2þ Feltham and Ohlson (1996) present a model that supports the preceding arguments. The essential elements of this model are provided in the Appendix A. The two forms of conservatism are discussed in detail in Easton (2001) who shows that they, respectively, capture accounting value added and economic value added. Most of our analyses are based on regressions that are an empirical analogue of Equation (2). Taking first differences, invoking clean surplus (in other words, defining x jt as comprehensive income), re-arranging, and dividing by beginning-ofperiod price, we obtain: x jt Dx jt d jt 1 Dci jt Doa jt 1 ret jt ¼ b 0 þ b 1 þ b p 2 þ b jt 1 p 3 þ b jt 1 p 4 þ b jt 1 p 5 þ e jt ; ð3þ jt 1 p jt 1 where ret jt ¼ðp jt þ d jt p jt 1 Þ=p jt 1 and D represents first differences. The subscript j denotes an observation for firm j. b 4 captures the effect of conservatism due to future positive NPV projects, and b 5 captures the effect of conservatism due to accounting rules. The coefficients on earnings levels, earnings changes, and lagged dividends are all predicted to be positive Data Selection and Sample Description Initially, we collect all Compustat firm-year observations from fiscal years 1988 through 2002 for which we have complete data for the following items. Return (ret t )

5 ACCOUNTING CONSERVATISM 499 is obtained from CRSP by compounding monthly returns during the fiscal period. Comprehensive income (x t ) is net income (#172) minus preferred dividends (#19) plus the change in value of marketable securities (#238) plus the change in the cumulative foreign currency translation adjustment (#230). Dividends (d t ) are the sum of dividends to common shareholders (item #21) and net capital contributions. Net capital contributions are purchases of common and preferred stock (item #115) 3 minus sales of common and preferred stock (item #108). Operating assets (oa t ) are book value of equity (b t ) minus financial assets (fa t ). Book value of equity (b t ) is common equity (#60) plus preferred treasury stock (#227) minus preferred dividends in arrears (#242). Financial assets (fa t ) are cash and short-term investments (#1) plus investments and advances-others (#32) minus debt in current liabilities (#34) minus longterm debt (#9) minus preferred stock (#130) plus preferred treasury stock (#227) minus preferred dividends in arrears (#242) minus minority interest (#38). 4 Cash investments are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). 5 The ratio of the market value of operating assets to the book value of operating assets (V/oa) is the market value of common equity minus financial assets (fa t ) divided by the book value of operating assets ((p t ) fa t )/oa t ). All variables except the market value of equity (p t ), annual stock returns (ret t ), and ratio of the market value of operating assets to the book value of operating assets (V/oa) are deflated by the beginning market value of equity (p t 1 ). Observations with negative book value of equity or negative (estimated) book value or market value of operating assets are excluded. We exclude utilities (SIC ) and financial institutions (SIC ). 6 We further delete observations in the top and bottom one percent of the distribution for any one of the following variables: annual returns, earnings levels, earnings changes, lagged dividends, change in cash investments and change in lagged operating assets in order to mitigate the effect of extreme values. The final sample is 54,313 firm-year observations, which consist of 36,416 firmyear observations reporting profits (profit firms, hereafter) 17,897 firm-year observations reporting losses (loss firms, hereafter) 27,697 observations with non-negative returns (good news firms, hereafter) 26,616 observations with negative returns (bad news firms, hereafter). The lack of data necessary to measure cash investments (ci t ) restricts our analysis to the post-1987 period. Panel A of Table 1 reports descriptive statistics for the sample of 54,313 firm-year observations from 1988 to The median market value of equity is $ million. Over the 15 years, the mean and median annual raw stock returns are 10.9% and 0.9%, respectively. Median net comprehensive income and the median change in net comprehensive income are, respectively, 3.8% and 0.4% of the beginning market value of equity. Median lagged dividends are zero. The decomposition of book value of equity into operating assets and financial assets shows that firms have on average net financial obligations; hence operating assets are greater than book value of equity. The positive change in operating assets (median of 3.3% of price) implies that operating assets are, on average, increasing. The ratio of the market value of

6 500 EASTON AND PAE Table 1. Descriptive statistics for key variables. Variable Mean Std. Dev. Q1 Median Q2 Min. Max. Panel A: All firms (#obs=54,313) pt 1, , ,329.5 rett ) ) x t ) ) ) Dxt ) ) dt-1 ) ) ) Dci t ) ) b t fat ) )0.446 ) ) oat Doa t ) ) Doa t ) ) V/oa ,102.5 P/B ,071.4 Panel B: Profit and loss firms Profit firms (#obs = 36,416) Loss firms (#obs = 17,897) Mean Q1 Median Q3 Mean Q1 Median Q3 pt 2, rett ) )0.089 )0.500 ) x t )0.171 )0.218 )0.100 )0.040 Dx t ) )0.072 )0.154 ) dt-1 )0.008 ) )0.054 )0.073 ) Dcit ) )0.013 )0.053 ) b t fat )0.260 )0.415 ) )0.392 )0.524 ) oat Doa t ) )0.019 )0.124 ) Doa t ) ) V/oa P/B

7 ACCOUNTING CONSERVATISM 501 Table 1. continued. Good News Firms (#obs = 27,697) Bad News Firms (#obs = 26,616) Mean Q1 Median Q3 Mean Q1 Median Q3 Panel C: Good news and bad news firms pt 2, , ret t )0.322 )0.481 )0.286 )0.135 xt )0.049 ) Dxt ) )0.031 )0.070 ) d t-1 )0.012 ) )0.035 )0.030 ) Dci t ) )0.008 ) bt fat )0.283 )0.443 ) )0.325 )0.449 ) oa t Doa t ) ) Doat) ) ) V/oa P/B pt is the market value of equity (Compustat annual data item #199 #25). rett is the annual stock return obtained by compounding CRSP monthly returns over the fiscal period. x t is comprehensive income calculated as net income (#172) minus preferred dividends (#19) plus the change in the marketable securities adjustment (#238) plus the change in the cumulative translation adjustment (#230). Dxt is the change in comprehensive income. dt)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions include purchases of common and preferred stock (#115) minus sales of common and preferred stock (#108). Operating assets (oa t ) is calculated as book value of equity (b t ), minus financial assets (fa t ). b t is common equity (#60) plus preferred treasury stock (#227), minus preferred dividends in arrears (#242). fa t is cash and short-term investments (#1) plus investments and advances-others (#32), minus debt in current liabilities (#34) minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Doat is the change in operating assets. Cash investments (cit) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). Dci t is the change in cash investments. V/oa is the ratio of the market value of operating assets to the book value of operating assets. The market value of operating assets is calculated as the market value of equity minus financial assets (pt fat). P/B is price to book ratio (pt divided by bt). All variables except for pt, rett, V/oa, and P/B are deflated by beginning market value of equity (pt)1).

8 502 EASTON AND PAE operating assets to the book value of operating assets is generally greater than one; although, for about 19% of the sample, the market value of operating assets is less than their book value. Panel B of Table 1 reports descriptive statistics for the profit and loss sub-samples. Profit firms are, on average, bigger than loss firms. The median market values of equity for profit and loss firms are $ million and $46.26 million, respectively. Loss firms have, on average, higher market to book (P/B and V/oa) ratios than profit firms. This is due to both higher market value of equity for profit firms and lower book value of equity for loss firms. Panel C of Table 1 reports descriptive statistics for firm-years with non-negative returns ( good news) and for the firm-years with negative returns ( bad news). The median market values of equity for good and bad news firm-years are $ million and $68.20 million, respectively. Good news firm-years have, on average, higher market to book (P/B and V/oa) ratios than bad news firms. Table 2 reports the Pearson and Spearman correlations among key variables. The correlations between the returns and each of the independent variables are significant at, at least the 0.01 level. The correlations between change in lagged operating assets and both earnings changes and change in cash investments are high ()0.291 and )0.236, respectively) suggesting that multicollinearity may affect the stability of the estimates of the coefficients on these variables Empirical Results 3.1. Conservatism in the Entire Sample Table 3 summarizes the output from regression (3) for each of years 1988 to Conservatism associated with investment in positive NPV projects is evident in the data. However, when all observations are analyzed together, there is no evidence of conservatism due to accounting rules. Evidence of this form of conservatism is seen later in sub-samples of the data. The estimate of the coefficient on change in cash investments is positive in every annual regression except year The mean of these estimates (0.242) is significantly positive at, at least, the 0.01 level (t-statistic of 6.63). This evidence is consistent with the notion that conservative accounting does not reflect the effects of investments in positive net present value projects until future periods. The estimates of the coefficient on change in cash investments are significantly positive in 13 of the 15 annual regressions, providing some comfort that multicollinearity (which may lead to instability of the coefficient estimates) is not unduly affecting the analyses. The mean of the estimates of the coefficient on change in lagged operating assets is not significantly different from zero at conventional levels (t-statistic of )1.74). That is, when all observations are analyzed together, there is no evidence of conservatism associated with over-depreciation of assets in place.

9 ACCOUNTING CONSERVATISM 503 Table 2. Pearson and Spearman Correlations among key variables a. p t ret t x t Dx t d t)1 Dci t Doa t)1 V/oa p t ) (<0.01) (<0.01) (0.61) (<0.01) (0.62) (0.12) (0.21) ret t ) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) x t )0.003 (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (0.53) Dx t )0.001 )0.041 ) (<0.01) (<0.01) (<0.01) (0.79) (<0.01) (<0.01) (0.75) d t) )0.009 )0.024 )0.095 )0.010 (<0.01) (<0.01) (<0.01) (0.03) (<0.01) (<0.01) (0.02) Dci t )0.047 )0.236 )0.001 (<0.01) (<0.01) (<0.01) (0.01) (<0.01) (<0.01) (0.88) Doa t) ) )0.175 )0.106 )0.199 )0.005 (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (0.27) V/oa ) ) )0.142 (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) (<0.01) a Pearson correlations above the diagonal, Spearman correlations below the diagonal (two tailed p values in parentheses). p t is the market value of equity (Compustat annual data item #199 #25). ret t is the annual stock return obtained by compounding CRSP monthly returns over the fiscal period. x t is comprehensive income calculated as net income (#172) minus preferred dividends (#19), plus the change in the marketable securities adjustment (#238), plus the change in the cumulative translation adjustment (#230). Dx t is the change in comprehensive income. d t)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions include purchases of common and preferred stock (#115), minus sales of common and preferred stock (#108). Operating assets (oa t ) is book value of equity (b t ) minus financial assets (fa t ). b t is common equity (#60) plus preferred treasury stock (#227), minus preferred dividends in arrears (#242). fa t is cash and short-term investments (#1) plus investments and advances-others (#32), minus debt in current liabilities (#34), minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Doa t is the change in operating assets. Cash investments (ci t ) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). V/oa is the ratio of the market value of operating assets to the book value of operating assets. The market value of operating assets is the market value of equity minus financial assets (p t) fa t ). All variables except for p t, ret t, and V/oa are deflated by beginning market value of equity (p t)1 ). The mean adjusted R 2 is 11%. The estimates of the coefficients on earnings, earnings changes and lagged dividends are all positive as predicted A Comparison with Easton and Harris (1991) We now investigate the effect of the omission of change in cash investments and change in lagged operating assets from regression (3). The results of this investigation are reported in Table 4. The estimate of the coefficient on earnings levels in the simple regression of returns on deflated earnings levels (model M1) is significantly positive at the 0.01 level (t-statistic of 8.84). The estimate of the coefficient on

10 504 EASTON AND PAE Table 3. The regression of returns on earnings, earnings changes, lagged dividends, lagged change in operating assets, and change in cash investments: results from year-by-year regressions. retjt ¼ b 0 þ b 1 xjt pjt 1 Dxjt þ b 2 pjt 1 djt 1 þ b 3 pjt 1 Dcijt þ b 4 pjr 1 Coefficient estimates with t-statistics in parentheses Doajt 1 þ b 5 pjt 1 þ ejt Year #obs Int. xt Dxt dt)1 Dcit Doat)1 Adj. R (7.87) (9.86) (2.33) (0.09) (1.15) )0.052 ()0.81) , (14.43) (12.43) (5.43) (0.96) (4.18) )0.031 ()1.20) ,196 )0.116 ()15.91) (15.28) (4.23) )0.021 ()0.22) (4.35) )0.039 ()1.39) , (28.47) (6.13) (7.29) )0.458 ()3.30) (3.07) (0.91) , (14.46) (8.11) (10.34) (3.94) (5.58) )0.088 ()2.58) , (20.81) (9.38) (7.52) )0.044 ()0.49) (7.81) )0.015 ()0.31) ,001 )0.031 ()4.72) (13.46) (6.47) (6.55) (6.17) )0.045 ()1.28) , (24.09) (3.80) (12.14) )0.025 ()0.28) (5.29) )0.074 ()1.64) , (17.08) (9.73) (7.95) (4.80) (9.05) )0.008 ()0.21) , (19.64) (17.55) (6.06) (5.78) (9.51) (2.34) ,540 )0.062 ()7.76) (6.94) (8.50) (4.29) (3.94) )0.107 ()2.83) , (18.63) )0.685 ()5.69) (10.63) )0.166 ()1.04) (4.60) )0.061 ()0.98) ,122 )0.012 ()1.09) (8.07) (7.45) (4.91) (4.07) (1.54) , (13.28) (5.40) (10.49) (7.92) (2.60) (1.09) ,954 )0.096 ()12.25) (15.38) (1.44) (6.34) )0.023 ()0.51) )0.127 ()4.80) Mean (t-value) 3, (2.89) (5.06) (6.63) (2.49) (6.63) )0.028 ()1.74) The dependent variable is annual stock returns retjt obtained by compounding CRSP monthly returns over the fiscal period. xt is comprehensive income calculated as net income (#172) minus preferred dividends (#19), plus the change in the marketable securities adjustment (#238), plus the change in the cumulative translation adjustment (#230). Dxt is the change in comprehensive income. dt)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions are purchases of common and preferred stock (#115) minus sales of common and preferred stock (#108). Operating assets (oat) is book value of equity (bt) minus financial assets (fat). bt is common equity (#60) plus preferred treasury stock (#227) minus preferred dividends in arrears (#242). fa t is cash and short-term investments (#1), plus investments and advances-others (#32), minus debt in current liabilities (#34), minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Doat is the change in operating assets. Cash investments (cit) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). Dci t is the change in cash investments. All independent variables are deflated by beginning market value of equity (p t)1 ).

11 ACCOUNTING CONSERVATISM 505 Table 4. Addition of lagged dividends, lagged change in operating assets, and change in cash investments to the Easton and Harris (1991) regression of returns on earnings and earnings changes. retjt ¼ b 0 þ b 1 xjt pjt 1 Dxjt þ b 2 pjt 1 djt 1 þ b 3 pjt 1 Dcijt þ b 4 pjt 1 Coefficient estimates with t-statistics in parentheses. Doajt 1 þ b 5 pjt 1 þ ejt Model Int. x t Dx t d t)1 Dci t Doa t)1 Adj. R 2 M1 Coef. t-value (2.76) (8.84) M2 Coef. t-value (2.73) (11.79) M3 Coef. t-value (2.75) (5.32) (6.55) M4 Coef. t-value (2.93) (5.22) (6.83) (2.34) M3 Coef. = M4 Coef.: ()2.72) (2.09) M5 Coef. t-value (2.89) (5.06) (6.63) (2.49) (6.63) )0.028 ()1.74) M4 Coef. = M5 Coef.: ()2.15) (0.85) Coefficients are means of annual regressions over the period , and t-values in parentheses are based on the standard error of the mean (Fama and MacBeth, 1973; Bernard, 1987). The dependent variable is annual stock returns obtained by compounding CRSP monthly returns over the fiscal period. xt is comprehensive income calculated as net income (#172) minus preferred dividends (#19), plus the change in marketable securities adjustment (#238), plus the change in cumulative translation adjustment (#230). Dxt is the change in comprehensive income. dt)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions are purchases of common and preferred stock (#115) minus sales of common and preferred stock (#108). Operating assets (oat) is book value of equity (bt) minus financial assets (fat). bt is common equity (#60) plus preferred treasury stock (#227), minus preferred dividends in arrears (#242). fat is cash and short-term investments (#1) plus investments and advances-others (#32), minus debt in current liabilities (#34), minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Doa t is the change in operating assets. Cash investments (ci t ) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). Dcit is the change in cash investments. All independent variables are deflated by beginning market value of equity (pt)1).

12 506 EASTON AND PAE earnings changes in the simple regression of returns on deflated earnings changes (model M2) is also significantly positive at the 0.01 level (t-statistic of 11.79). This result is consistent with Easton and Harris (1991). 8 Next, we regress annual stock returns on both earnings levels and earnings changes (model M3). The mean estimates of the coefficients on earnings levels and earnings changes are and 0.504, respectively, with associated t-statistics of 5.32 and 6.55, respectively. Consistent with the Ohlson (1995) model, we add lagged dividends to earnings levels and earnings changes as explanatory variables for returns (model M4). The addition of lagged dividends does not materially increase the adjusted R 2, but the estimate of the coefficient on lagged dividends is significantly different from zero at the 0.05 level. 9 Addition of change in cash investments and change in lagged operating assets increases the average adjusted R 2 from 10.2% (in model M4) to 11% (in model M5). The estimate of the coefficient on earnings levels decreases from to (with a t-statistic for the difference between these coefficient estimates of )2.15) while the estimate of the coefficient on earnings changes increases from to (with a t-statistic for the difference between these coefficient estimates of 0.85) The Current Market to Book Ratio as a Proxy for Conservatism Each of the forms of accounting conservatism (conservatism due to accounting rules and failure to capture investment in positive NPV projects) results in understatement of book value. It follows that one would expect to see more evidence of conservatism when the ratio of the market value of equity to the book of equity is high. Conservatism, however, is likely to be less prevalent in the valuation of financial assets due to less conservative accounting rules and because investment in financial assets are generally viewed as a means of holding reserves for future investments in operations and are thus unlikely to be positive net present value. It follows that the ratio of the market value of net operating assets to the book value of net operating assets (as opposed to the ratio of the market value of common equity to the book value of common equity) may be a more appropriate a priori indicator of conservatism. We partition the sample each year into deciles based on the ratio of the market value of net operating assets measured as the market value of equity minus the book value of financial assets (that is, p t fa t ) to book value of operating assets (that is, oa t ), and we examine whether the estimates of the coefficients on the variables that are chosen to capture the two forms of accounting conservatism Doa t)1 and Dci t vary across these sub-samples. We expect that the higher the ratio of market value of net operating assets to book value of net operating assets, the more significant the coefficients on these variables. 10 Table 5 summarizes the output from regression (3) conducted within deciles of market value of net operating assets to book value of net operating assets. Decile 1 includes firms with the lowest ratios of market value of net operating assets to book

13 ACCOUNTING CONSERVATISM 507 Table 5. Explanatory power of lagged change in operating assets and change in cash investments for firms grouped on the ratio of market value of operating assets to the book value of operating assets (V/oa). V/oa Decile Median Int. xt Dxt dt)1 Dcit Doat)1 Adj. R )0.176 ()4.81) (7.36) (3.26) (7.84) (0.64) )0.117 ()5.72) )0.095 ()2.46) (9.72) (6.37) (9.53) (2.07) )0.043 ()2.20) )0.047 ()1.26) (6.69) (4.25) (6.27) (2.77) )0.022 ()0.64) (0.59) (7.04) (5.26) (7.75) (2.75) (1.13) (1.51) (6.48) (5.30) (7.85) (4.09) (2.32) (2.93) (5.74) (3.75) (6.08) (4.68) (2.24) (3.58) (5.53) (6.13) (3.77) (5.37) (3.92) (3.24) (3.81) (7.99) (4.39) (5.95) (6.04) (4.22) (3.12) (4.17) (2.48) (4.72) (2.36) (4.11) (0.68) (6.23) (0.92) (4.77) (2.74) Coefficients are means of annual regressions over the period , and t-values in parentheses are based on the standard error of the mean (Fama and MacBeth, 1973; Bernard, 1987). The dependent variable is annual stock returns obtained by compounding CRSP monthly returns over the fiscal period. xt is comprehensive income calculated as net income (#172) minus preferred dividends (#19), plus the change in the marketable securities adjustment (#238), plus the change in the cumulative translation adjustment (#230). Dx t is the change in comprehensive income. d t)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions are purchases of common and preferred stock (#115) minus sales of common and preferred stock (#108). Operating assets (oat) is book value of equity (bt) minus financial assets (fat). bt is common equity (#60) plus preferred treasury stock (#227) minus preferred dividends in arrears (#242). fat is cash and short-term investments (#1) plus investments and advances-others (#32), minus debt in current liabilities (#34), minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Doat is the change in operating assets. Cash investments (cit) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). Dcit is the change in cash investments. All independent variables are deflated by beginning market value of equity (pt)1). V/oa is the ratio of the market value of operating assets to the book value of operating assets. The market value of operating assets is the market value of equity minus financial assets (p t ) fa t ).

14 508 EASTON AND PAE value of net operating assets, and decile 10 includes firms with the highest ratios. 11 If the current market to book ratio is a good proxy for accounting conservatism, the coefficients on the change in cash investments and change in lagged operating assets will increase as we move from decile 1 to decile 10. Table 5 reports that the median market to book (V/oa) ratio is less than one for deciles 1 and 2, implying that accounting is more likely to be aggressive. 12 For deciles 3 and higher, the median market to book (V/oa) ratios are greater than one, implying that accounting is more likely to be conservative. The estimate of the coefficient on change in lagged operating assets is significantly negative for deciles 1 and 2 consistent with the change in value represented by the accruals being less than the change in value reflected in returns. Accounting conservatism in the higher V/oa deciles is seen in significantly positive estimates of the coefficients on change in lagged operating assets for deciles 5 and higher. The estimates of the coefficients on change in cash investments increase monotonically from for the decile with lowest V/oa to for the decile with highest V/oa. To summarize, there is more evidence of both forms of conservatism when the price-to-book ratio is high Conservatism and Industry Since accounting methods differ considerably across industries, we expect to see differences in the degree of conservatism and differences in the explanatory power of lagged change in operating assets and change in cash investments for returns. We partition the sample into 16 industries using the primary SIC code. Table 6 reports the industry composition of the sample. The classification scheme is similar to Barth et al. (1998), but we also include the agriculture industry. The other industry category includes firms that are classified as non-operating establishments by Compustat (SIC 9995). Table 7 reports medians of key variables by industry. Utilities and firms in the chemicals and transportation industries have larger market value of equity. The median annual stock returns are positive in all industries other than the mining and construction, computer and services industries. The median net income is positive in all industries other than the pharmaceutical industry. The median market to book ratio is greater than one for all industries. The pharmaceutical industry has the highest median market to book ratio followed by the computer industry. The positive median change in cash investments for all industries suggests increasing cash investments in operations over this time period. Consistent with increasing cash investments, the median change in lagged operating assets is positive for all industries. Table 8 reports the results from regression (3) conducted at the industry level. The estimates of the coefficients on change in cash investments (consistent with investment in positive net present value projects) are significantly positive in all industries other than agriculture, mining and construction, food and chemicals. The estimates

15 ACCOUNTING CONSERVATISM 509 Table 6. Identity of industry sub-samples. Industry Primary SIC codes # firm- years % of obs. Agriculture 1) Mining and , excluding , Construction Food , Textiles and Printing , Chemicals , , Pharmaceuticals , Extractive Industries , , Durable Manufacturers , excluding and , Computers , , , Transportation , Retail , Insurance and , Real Estate Services , excluding , Others 9000 and above Total 54, Mean 3, of the coefficients on change in lagged operating assets are not significantly different from zero with the exception of pharmaceuticals (t-statistic of 2.87) and durable manufactures (t-statistic of )3.19). Note that the pharmaceutical industry has the highest median V/oa (6.967). These results are consistent with those in Table Positive Returns vs. Negative Returns and Conservatism Basu (1997) observes that the explanatory power of earnings for returns differs according to whether news is, on average, good (that is, returns of the fiscal period are positive) or news is, on average, bad (negative returns). Although Basu s (1997) concept of conservatism is very different from ours, we also partition observations according to the sign of their fiscal period returns and re-run regression (3) for each of the two partitions of the data. The essence of Basu s (1997) argument is that bad news and the associated reported earnings tend to be less persistent than good news and the associated reported earnings. In the valuation model (2) this implies that the coefficient on book value (earnings) will be higher (lower) for bad news firms than for good news firms. Correspondingly, the coefficient on earnings levels (changes in earnings) in the returns regression (3) will be higher (lower) for bad news firms than for good news firms. Since the effects of conservatism associated with the accounting measure of change in value being less than the change in market value are likely to be exacerbated for good news firms, we predict that the estimate of the coefficient on change in operating assets will be significantly positive. On the other hand, for bad news firms, it is possible that the accounting measure of change in the value of operating assets may

16 510 EASTON AND PAE Table 7. Median of key variables by industry. Industry a p t ret t x t Dx t d t)1 Dci t Doa t)1 V/oa Agriculture Mining and Construction ) Food Textiles and Printing Chemicals Pharmaceuticals )0.028 )0.038 )0.001 ) Extractive Industries Durable Manufacturers Computers ) ) Transportation Retail Insurance and Real Estate Services ) ) Others ) a See Table 6 for the definition of industry. p t is the market value of equity (Compustat annual data item #199 #25). ret t is the annual stock return obtained by compounding CRSP monthly returns over the fiscal period. x t is comprehensive income calculated as net income (#172) minus preferred dividends (#19) plus the change in the marketable securities adjustment (#238) plus the change in the cumulative translation adjustment (#230). Dx t is the change in comprehensive income. d t)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions are purchases of common and preferred stock (#115) minus sales of common and preferred stock (#108). Operating assets (oa t ) is book value of equity (b t ) minus financial assets (fa t ). b t is common equity (#60) plus preferred treasury stock (#227) minus preferred dividends in arrears (#242). fa t is cash and short-term investments (#1) plus investments and advancesothers (#32), minus debt in current liabilities (#34), minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Doa t is the change in operating assets. Cash investments (ci t ) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). Dci t is the change in cash investments. V/oa is the ratio of the market value of operating assets to the book value of operating assets. The market value of operating assets is the market value of equity minus financial assets (p t ) fa t ). All variables except for p t, ret t, and V/oa are deflated by beginning market value of equity (p t)l ). be greater than the market assessment of this change in value. Panel C of Table 1 reports that the mean earnings for bad news firms ()4.9% of beginning of year market value of equity) are greater than the mean returns ()32.2%). Since accounting records a smaller decline in value than assessed by the market, we predict that the estimate of the change in operating assets will be significantly negative. We predict that cash investments for firms with good news and for firms with bad news will be positive NPV and, hence, the estimate of the coefficient on change in cash investments will be positive for both of these sub-samples. As predicted, panel A of Table 9 reports that the estimate of the coefficient on earnings in the regression of returns on earnings for non-negative returns firms is not significantly different from zero (0.079) and panel B of Table 9 reports that the estimate of this coefficient for firms with negative returns is significantly positive

17 ACCOUNTING CONSERVATISM 511 Table 8. Conservatism and industry: mean of annual OLS regressions. Industry a Int. xt Dxt dt)1 Dcit Doat)1 Adj. R 2 Agriculture (1.78) (1.83) (2.15) )0.139 ()0.30) )0.176 ()0.42) )0.133 ()0.49) Mining and Construction (1.48) (3.96) (1.01) (0.82) (0.76) )0.059 ()0.92) Food (1.40) (4.01) (3.77) (2.31) (1.05) (0.19) Textiles and Printing (1.60) (5.43) (2.66) (0.53) (3.23) )0.043 ()1.41) Chemicals (1.65) (4.71) (1.30) )0.063 ()0.15) (0.09) )0.232 ()1.56) Pharmaceuticals (2.12) (0.33) (2.78) (1.78) (5.08) (2.87) Extractive Industries (1.54) (4.48) (1.08) (0.70) (4.32) (1.13) Durable Manufacturers (2.57) (6.52) (5.61) (2.71) (5.07) )0.088 ()3.19) Computers (1.86) (6.51) (6.27) (2.96) (2.77) )0.005 ()0.13) Transportation (2.72) (0.94) (3.34) (0.45) (2.69) (1.04) Retail (2.48) (6.41) (4.31) (3.23) (3.72) )0.023 ()0.55) Insurance and Real Estate (3.18) (3.52) (1.90) (0.76) (4.68) )0.090 ()1.89) Services (2.76) (5.12) (4.08) (0.70) (4.83) )0.031 ()0.98) Others (1.07) (0.95) (2.26) (0.66) (2.06) (0.16) a See Table 6 for the definition of industry. Coefficients are means of annual regressions over the period , and t-values in parentheses are based on the standard error of the mean (Fama and MacBeth, 1973; Bernard, 1987). The dependent variable is annual stock returns obtained by compounding CRSP monthly returns over the fiscal period. xt is comprehensive income calculated as net income (#172) minus preferred dividends (#19) plus the change in the marketable securities adjustment (#238) plus the change in the cumulative translation adjustment (#230). Dx t is the change in comprehensive income. d t)1 is the sum of cash dividends to common shareholders (#21) and net capital contributions. Net capital contributions are purchases of common and preferred stock (#115) minus sales of common and preferred stock (#108). Operating assets (oat) is book value of equity (bt) minus financial assets (fat). bt is common equity (#60) plus preferred treasury stock (#227) minus preferred dividends in arrears (#242). fa t is cash and short-term investments (#1) plus investments and advances-others (#32), minus debt in current liabilities (#34), minus long-term debt (#9), minus preferred stock (#130), plus preferred treasury stock (#227), minus preferred dividends in arrears (#242), minus minority interest (#38.). Cash investments (cit) are calculated from the cash flow statement as capital expenditures (#128) minus sale of property, plant, and equipment (#107) plus acquisitions (#129) minus other investing activities (#310). Doa t is the change in operating assets. Dci t is the change in cash investments. All independent variables are deflated by beginning market value of equity (p t)1 ).