Financing Decisions and Discretionary Accruals: Managerial Manipulation or Managerial Overoptimism

Transcription

1 Financing Decisions and Discretionary Accruals: Managerial Manipulation or Managerial Overoptimism Dalia Marciukaityte Louisiana Tech University Samuel H. Szewczyk Drexel University December 3, 2007 Address for correspondence: Dalia Marciukaityte, College of Business, Louisiana Tech University, P.O. Box 10318, Ruston, LA 71272, USA. Phone:

2 1 Financing Decisions and Discretionary Accruals: Managerial Manipulation or Managerial Overoptimism Abstract We examine whether discretionary accruals of firms obtaining substantial external financing can be explained by managerial manipulation or managerial overoptimism. Consistent with Teoh, Welch, and Wong (1998b), we find that discretionary current accruals peak when firms obtain equity financing. However, we also find that discretionary accruals peak when firms obtain debt financing. Furthermore, discretionary accruals are higher for firms that rely on debt financing rather than equity financing. The results are robust to controlling for firm characteristics, excluding small and distressed firms, and using alternative measures of discretionary accruals. These findings support the hypothesis that managerial overoptimism distorts financial statements of firms. Keywords: managerial overoptimism, earnings manipulation, discretionary accruals, debt financing, equity financing

3 I. Introduction There is substantial evidence that firms issuing new equity or debt securities exhibit poor stock performance for a number of years following these issues (e.g., Loughran and Ritter, 1995; Spiess and Affleck-Graves, 1995 and 1999; and Datta, Iskandar-Datta, and Raman, 2000). Most studies that examine the reasons for abnormal post-issue performance focus on equity issues. Loughran and Ritter (1997) find that operating performance peaks at the time of public equity issues. They suggest that investors incorrectly expect good operating performance at the time of the issues to continue and pay too much for shares of issuing firms. Furthermore, Rangan (1998) and Teoh, Welch, and Wong (1998a, 1998b) propose that this improvement in operating performance may be caused by earnings manipulation intended to mislead investors. The authors find that the peak in operating performance at the time of equity issues is driven by high discretionary accruals, and that firms with higher discretionary accruals experience worse post-issue performance. The earnings manipulation hypothesis is not the only potential explanation for high discretionary accruals at the time of security issues and poor stock performance after the issues. Teoh, Welch, and Wong (1998a) mention that high discretionary accruals associated with security issues and poor post-issue performance may be due to managerial overoptimism at the time of issues. Overoptimistic managers, for example, may use overly generous credit policies and allow inadequate provisions for bad debts since such managers overestimate the expected success of their business strategies. To examine the effect of managerial earnings manipulation and managerial overoptimism on

4 3 the size of discretionary accruals, we compare discretionary accruals of firms obtaining substantial equity and debt financing. Managers can benefit the most from earnings manipulation when they rely on external equity financing. Higher earnings lead to higher share prices, and firms need to issue fewer shares to raise the desired amount of money. The costs of debt financing are less sensitive to the changes in expected future performance than the costs of equity financing. If the earnings manipulation hypothesis explains high discretionary accruals at the time of security issues, discretionary accruals should be higher for firms relying on external equity financing than for firms relying on debt financing. Since manager overconfidence cannot be estimated directly, similar to Malmendier and Tate (2003 and 2005), we rely on observable manager decisions to gauge managerial overoptimism. Heaton s (2002) model suggests and empirical studies (e.g., Malmendier, Tate, and Yan 2005; Gombola and Marciukaityte 2007) confirm that managers using external equity financing, on average, are less overoptimistic than managers relying on debt financing. We confirm that a firm s reliance on debt financing proxies for managerial overoptimism by examining insider trading. Focusing on financing choices to proxy for managerial overoptimism allows us to examine a large number of firms as necessary data are readily available. If managerial overoptimism is the prevailing factor affecting the size of discretionary accruals, discretionary accruals should be higher for firms relying on debt financing than for firms relying on equity financing. To be more precise, we are testing a joint hypothesis that the preference for debt financing rather than equity financing is a good proxy for managerial overoptimism and that managerial overoptimism has the prevailing impact on the size of discretionary

5 4 accruals of firms obtaining substantial financing. Observing higher discretionary accruals for firms relying on debt financing provides support to both hypotheses. We examine a sample of firms obtaining financing during 1988 to To create the sample of firms with substantial external financing during one fiscal year, we identify the top 25% based on their external financing normalized by beginning total assets. Firms obtaining substantial financing are especially suitable for the analysis in this paper as the effects of earnings manipulation and managerial overoptimism are likely to be stronger for high-growth firms. High-growth firms derive a substantial part of their value from highly uncertain growth opportunities, making it harder for investors to assess their value and easier for managers of these firms to mislead investors. Furthermore, high-growth firms can be expected to be managed by more overoptimistic managers than slower growing firms as overoptimistic managers believe that they have more good projects available to them than less optimistic managers do. From external-financing sample, we identify equity- and debt-financing samples. Firms with equity financing at least twice exceeding debt financing compose the equityfinancing sample, and firms with debt financing at least twice exceeding equity financing compose the debt-financing sample. The results are not sensitive to alternative specifications of the samples. We use three measures of discretionary accruals: discretionary total accruals, discretionary current accruals, and discretionary current accruals excluding depreciation. Consistent with Teoh, Welch, and Wong, we find that both discretionary current accruals measures peak in the year firms obtain substantial equity financing. We also find that discretionary total accruals and both measures of discretionary current accruals peak in

6 5 the year firms obtain substantial debt financing. Comparison of the debt- and equityfinancing samples shows that all three discretionary accruals measures are significantly higher for firms obtaining debt financing than for firms obtaining equity financing. This finding is consistent with the hypothesis that managerial overoptimism provides a better explanation for high discretionary accruals of firms obtaining substantial financing than managerial earnings manipulation. Controlling for firm characteristics, we confirm the inverse relation between discretionary accruals and reliance on equity financing rather than debt financing. Firms with higher ratio of equity financing to external financing tend to have lower discretionary total accruals and lower discretionary current accruals. The results are similar with both winsorized and unwinsorized discretionary accrual values. Furthermore, the inverse relation between discretionary accruals and the ratio of equity financing to external financing is highly significant for the subperiod and the subperiod. We confirm the inverse relation between discretionary accruals and reliance on equity financing using restricted samples excluding small firms and distressed firms. All our results support the hypothesis that managerial overoptimism is the prevailing factor affecting discretionary accruals of firms obtaining substantial financing. This study contributes to the literature in three ways. First, we show that discretionary accruals peak when firms obtain substantial debt financing and exceed discretionary accruals of firms obtaining substantial equity financing. Second, based on the results in the study, we suggest that managerial overoptimism is a significant factor

7 6 distorting financial statements of firms. Third, we confirm the notion that the choice between debt and equity financing is affected by managerial overoptimism. The rest of this paper is organized as follows. The next section discusses managerial overoptimism. Section III describes samples and methodology. Sections IV through VI present and discuss empirical results. Section VII concludes. II. Managerial Overoptimism Myers and Majluf (1984) suggest that managers avoid issuing undepriced securities to finance projects to escape transferring value from old to new security holders. As the cost of external equity financing is more sensitive to the market perception of firm value than the cost of debt financing, when managers believe that the market underestimates the value of their firm, they prefer debt financing to external equity financing. Likewise, Heaton (2002) shows that overoptimistic managers prefer debt financing to external equity financing, as overoptimistic managers tend to believe that the market underestimates the value of their firm. Believing that the market undervalues shares of their firms, overoptimistic managers avoid issuing equity when they have other financing choices. Recent studies provide empirical support for Heaton s model. Malmendier, Tate, and Yan (2005) examine financing decisions of Forbes 500 firms and show that overconfident CEOs prefer debt to external equity financing. Gombola and Marciukaityte (2007) provide support for Heaton s model by showing that high-growth

8 7 firms relying primarily on debt financing experience worse post-financing stock performance than high-growth firms relying on external equity financing. While many studies examine intentional earnings manipulation by managers (e.g., Rangan, 1998; Teoh, Welch, and Wong, 1998a and 1998b; and Peasnell, Pope, and Young, 2005), the impact of managerial overoptimism and overconfidence on reported earnings is not well examined. Teoh, Welch, and Wong (1998a) only mention a possibility that high discretionary accruals at the time of equity issues may be explained by managerial overoptimism, without further examining the managerial overoptimism hypothesis. Considering how common are overoptimism and overconfidence, the lack of research is surprising. Russo and Schoemaker (1992) test overconfidence among managers from different industries and find that more than 99% of managers are overconfident. Even being smarter does not help to avoid overoptimism; intellectual abilities and overoptimism are positively related (Klaczynski and Fauth, 1996). Recent studies show that some managerial decisions are affected by manager overoptimism and overconfidence (e.g., Heaton, 2002; Gervais, Heaton, and Odean, 2003; Malmendier and Tate, 2003 and 2005; and Malmendier, Tate, and Yan, 2005). Moderate managerial overoptimism may help to counteract the effect of managerial risk aversion and be beneficial to the company as managers are likely to be more risk averse with respect to their company than well-diversified shareholders (Gervais, Heaton, and Odean, 2003). However, strong managerial overoptimism can lead to poor business decisions and distorted financial statements. Overoptimistic managers overvalue the probability of their success and may employ overly aggressive business and accounting strategies that lead to higher discretionary accruals. Even without being dishonest,

9 8 overoptimistic managers may underestimate the portion of accounts receivable that is uncollectible and the cost of warranty plans. They may also overestimate the degree of completion of their long-term projects and the market value of their assets. 1 The accrual accounting system mandated by generally accepted accounting principles (GAAP) requires managers to adjust cash flows when calculating net income to reflect the actual condition of the firm. Overoptimistic managers are likely to use cash flow adjustments to reflect their overoptimistic view of the firm. III. Samples and Methodology Samples To obtain the external-financing sample, we use the following steps. We start with all firm-years included in both Compustat and CRSP databases during 1988 to 2001 fiscal years. We exclude the earlier years since the data necessary to estimate discretionary accruals from cash flow statements are not available before We estimate one-year equity financing as a change in common equity (item 60) minus a change in retained earnings (item 36) and debt financing as a change in total debt (total long-term debt (item 9) plus debt in current liabilities (item 34)). External financing is a sum of equity and debt financing. We normalize external financing using beginning total assets. When normalized external financing cannot be calculated for a firm-year or the length of a fiscal year is not equal to twelve months, we exclude that firm-year from 1 Teoh, Welch, and Wong (1998a) discuss in more detail how managers can affect net income by their choice and application of accounting methods and by their choice of timing for asset sales and purchases.

10 9 further consideration. To identify the firms that obtain substantial financing during a fiscal year, we find the top 25% of firm-years based on the normalized external financing. This procedure generates a sample with 19,747 firm-years. The fiscal year during which external financing is estimated is defined as the event year (Year 0). We limit the sample to firm-years with the data necessary to estimate discretionary total accruals for Year 0. Furthermore, we exclude from the sample regulated utilities (SIC codes ), depository institutions (SIC codes ), and holding or other investment offices (SIC codes ). As we examine the fiveyear post-financing period, we require that events for the same firm are at least five years apart. If a firm has more than one event in any five-year period, we include only the earliest one, limiting the sample to 8,316 events. The equity-financing sample includes firm-years with external equity financing at least twice higher than debt financing (3,740 firm-years), and the debt-financing sample includes firm-years with debt financing at least twice higher than equity financing (3,592 firm-years). This sample identification procedure allows us to examine overall financing policy of each firm during one year instead of relying on a single event. A firm that makes equity issue may also make an even larger debt issue during the same year. The procedure used in this paper allows us to exclude such firm from the equity-financing sample. Table 1 describes the distribution of external-, equity-, and debt-financing samples by the fiscal year. If the fiscal year ends in the first five months, the fiscal year is treated as belonging to the previous calendar year (consistent with Compustat). We find that all samples are well distributed across time with none of the years including

11 10 more than 12% of the events in any sample. Table 2 shows the ten most frequent industries (defined by the two-digit Standard Industrial Classification (SIC) code) in the financing samples. More than 5% of the external-financing firms belong to each of the following industries: business services, electronic and other electric equipment, chemicals and allied products, industrial machinery and equipment, and instruments and related products. The equity-financing sample has substantially higher concentration in business services as well as chemicals and allied products than the debt-financing sample. These differences should not affect our results as all methodologies used to estimate discretionary accruals control for the industry. Table 3 presents select characteristics of financing firms. All variables in this table are estimated at the beginning of or during Year 0. Firm characteristics vary substantially across financing samples. Debt-financing firms tend to be larger than equity-financing firms. The mean (median) market value of equity at the beginning of the event year is $906M ($72M) for debt-financing firms and $647M ($67M) for equityfinancing firms. Equity-financing firms in this study are larger than firms making seasoned equity offerings in Teoh, Welch, and Wong s (1998) study. Their firms have the mean (median) market value of equity of $284M ($52M). Following Loughran and Ritter s (1997) study, we estimate the market-to-book ratio as shares (item 54) multiplied by price (item 199) and divided by book value of equity (item 60). The mean market-to-book ratio is higher for the equity-financing sample (4.74) than for the debt-financing sample (2.63). Debt-financing firms are more leveraged, even at the beginning of the event year; their mean debt-to-asset ratio is 0.24, versus 0.17 for equity-financing firms. All firms raise significant amounts of external

12 11 financing during the event year. External financing normalized by beginning total assets is higher for equity-financing firms; the mean (median) relative financing is 1.38 (0.76) for equity-financing firms and 0.71 (0.53) for debt-financing firms. To address the issue of different characteristics across the samples, we perform tests controlling for firm characteristics. Table 3 also shows that equity-financing firms tend to rely very strongly on external equity financing, and debt-financing firms tend to rely very strongly on debt financing. The mean (median) ratio of equity financing to total financing for equityfinancing firms is 1.06 (1.00) during the event year. The mean (median) ratio of debt financing to total financing for debt-financing firms is 0.95 (0.98). Estimation of Discretionary Accruals Hribar and Collins (2002) show that accruals measures estimated using balance sheet items (e.g., Teoh, Welch, and Wong, 1998a and 1998b) are biased when firms undergo mergers and acquisitions or discontinue some of their operations and suggest using accruals measures estimated from cash flow statements. Our sample of firms obtaining substantial financing is likely to include a significant number of firms undergoing mergers and acquisitions. Thus, we examine discretionary accruals estimated from cash flow statements. We follow the methodology used in Hribar and Collins paper and estimate total accruals for firm j in year t (TACC j,t ): TACC j,t = EXBI j,t CFO j,t, (1)

13 12 where EXBI j,t is earnings before extraordinary items and discontinued operations (item 123) and CFO j,t is operating cash flow from continuing operations (item 308 minus item 124). Following Kothari, Leone, and Wasley (2005), we exclude observations where the absolute value of total accruals exceeds the value of total assets at the beginning of the year as such values are likely to be due to recording mistakes. We obtain discretionary total accruals using a modified version of Jones (1991) model. Following Teoh, Welch, and Wong (1998b), we normalize each variable using total assets at the beginning of the year. Each year, for each two-digit SIC code, excluding sample firms, we estimate ordinary least-squares regression: TACC ( 1/ TA ) + α ( S / TA ) + α ( PPE / TA ) ε, (2) j, t / TA j, t 1 = α 0 j, t 1 1 j, t j, t 1 2 j, t j, t 1 + j, t where TA j,t-1 is total assets (item 6) at the beginning of year t, is a change in sales S j, t (item 12) during year t, and PPE j,t is the gross property, plant, and equipment (item 7). To obtain reliable estimates, we require at least 10 firms in each two-digit SIC code group. Using the estimated coefficients from Equation 3, we obtain nondiscretionary total accruals for each sample firm i in year t (NDTA i,t ): NDTA ( 1/ TA ) + ˆ α (( S TR ) / TA ) ˆ α ( PPE / TA ), (3) ˆ i, t = α 0 i, t 1 1 i, t i, t i, t i, t i, t 1 where TR i, t is a change in trade receivables (item 151). Then, we estimate discretionary total accruals for firm i in year t (DTA i,t ):

14 13 DTA i, t = TACC i, t / TAi, t 1 NDTA i, t. (4) To estimate current accruals for firm j in year t (ACC j,t ), we follow Hribar and Collins (2002): ACC j,t = -(CHGAR j,t + CHGINV j,t + CHGAP j,t + CHGTAX j,t + CHGOTH j,t + DEP j,t ), (5) where CHGAR j,t is the decrease (increase) in accounts receivable (item 302), CHGINV j,t is the decrease (increase) in inventory (item 303), CHGAP j,t is the increase (decrease) in accounts payable (item 304), CHGTAX j,t is the increase (decrease) in taxes payable (item 305), CHGOTH j,t is the net change in other current assets (item 307), and DEP j,t is depreciation expense (item 125). Following Kothari, Leone, and Wasley (2005), we exclude observations where the absolute value of current accruals exceeds the value of total assets at the beginning of the year as such values are likely to be due to recording mistakes. We use the same procedure to estimate discretionary current accruals as we used for discretionary total accruals. Each year, for each two-digit SIC code, excluding sample firms, we estimate ordinary least-squares regression: ACC ( 1/ TA ) + α ( S / TA ) αˆ ( PPE / TA ). (6) j, t / TA j, t 1 = α 0 j, t 1 1 j, t j, t j, t j, t 1 Using the estimated coefficients from Equation 6, we obtain nondiscretionary current accruals for each sample firm i in year t (NDCA i,t ):

15 14 NDCA ( 1/ TA ) + ˆ α ( S TR ) / TA ) ˆ α ( PPE / TA ). (7) ˆ i, t = α 0 i, t 1 1 i, t i, t i, t i, t i, t 1 Then, we estimate discretionary current accruals for firm i in year t (DCA i,t ): DCA i, t = ACC i, t / TAi, t 1 NDCA i, t. (8) Teoh, Welch, and Wong s (1998b) measure of current accruals excludes depreciation expense. To obtain a comparable measure to the one used in their study, we also estimate discretionary current accruals excluding depreciation. We follow the above described procedure, except, we exclude depreciation expense from Formula 5 and the ratio of gross property, plant, and equipment to total assets from Formulas 6 and 7. IV. Discretionary Accruals for Equity- and Debt-Financing Samples To test whether managerial dishonesty or managerial overoptimism is the prevailing factor affecting the choice between debt and equity financing and discretionary accruals of firms obtaining substantial financing, we start with examining discretionary total accruals and two measures of discretionary current accruals for the three years before and five years following the financing year for equity- and debt-financing samples. The results are presented in Table 4. Following other discretionary accrual studies (e.g., Teoh, Welch, and Wong, 1998b; Kothari, Leone, and Wasley, 2005), we winsorize discretionary accruals at the top 1% and the bottom 1%. Winsorizing does not have significant effect on our results.

16 15 Teoh, Welch, and Wong (1998b) examine discretionary accruals estimated from balance sheet data for a sample of seasoned equity offerings. They find that discretionary current accruals peak in the year of the offering and mean (median) reach 5.6% (2.5%) of total assets, significant at the 1% level. Teoh, Welch, and Wong find no evidence of positive discretionary long-term accruals at the time of equity offerings; the mean (median) discretionary long-term accruals are -1.3% (-1.2%), significant at the 1% level. In Panel A of Table 4, we examine discretionary total accruals and two measures of discretionary current accruals for the three years before to five years following the equity-financing year. We find negative mean and median discretionary total accruals for the whole period examined, providing no evidence of earnings manipulation before or at the time of equity financing. However, consistent with Teoh, Welch, and Wong (1998b) findings, we find that discretionary current accruals peak in Year 0. The mean (median) discretionary current accruals for the equity-financing sample in Year 0 are 1.68% (0.18%), significant at the 1% (10%) level. The difference in total and current discretionary accruals is consistent with Teoh, Welch, and Wong s finding of negative discretionary long-term accruals for the three years before to three years following seasoned equity offerings. Other studies (e.g., Guenther, 1994) also find that current accruals are more likely to be manipulated by managers than long-term accruals. The current accruals are likely to be easier to manipulate than the long-term accruals to time their change before corporate events. Teoh, Welch, and Wong s measure of discretionary current accruals does not include depreciation. To obtain a comparable measure, we also estimate these accruals excluding depreciation. The last section in Panel A of Table 4 shows that discretionary

17 16 current accruals excluding depreciation peak in Year 0. The mean (median) discretionary accruals for the equity-financing sample in Year 0 are 4.11% (2.20%), significant at the 1% level. These findings are very similar to Teoh, Welch, and Wong s (1998b) findings. In Panel B of Table 4, we examine discretionary total accruals and two measures of discretionary current accruals for the three years before to five years following the debt-financing year. We find that all three measures of discretionary accruals peak in Year 0 when firms obtain substantial debt financing with the mean (median) discretionary total accruals of 2.45% (1.82%), the mean (median) discretionary current accruals of 4.25% (3.01%), and the mean (median) discretionary current accruals excluding depreciation of 5.12% (3.52%), significant at the 1% level. Table 5 compares discretionary accruals for debt- and equity-financing samples in Year 0. We use the analysis of variance to test the difference in means and the Kruskal- Wallis test to test the difference in medians. The results show that all three measures of mean and median discretionary accruals are significantly higher for the debt-financing sample than for the equity-financing sample. These findings support the hypothesis that managerial overoptimism affects the choice between debt and equity financing and has the prevailing effect on the size of discretionary accruals. Some debt-financing firms raise equity financing during Year 0 and some equity-financing firms raise debt financing during Year 0. To insure that mixedfinancing firms are not driving the results, we also compare the debt-financing sample restricted to firms raising only debt financing and the equity-financing sample restricted to firms raising only equity financing (Table 5). The results using restricted samples are

18 17 very similar to those using full debt- and equity-financing samples, suggesting that mixed-financing firms do not change the results. V. Discretionary Accruals and Preference for Equity Financing Instead of Debt Financing Table 3 reveals that firms relying on equity and debt financing have very different characteristics. As discretionary accruals might be affected by firm characteristics, it is important to control for them when examining the relation between discretionary accruals and reliance on external equity financing instead of debt financing. We use the logarithm of ratio of external financing to assets, the logarithm of market value of equity, the debtto-asset ratio, and the market-to-book ratio as controlling variables. Investors and analysts follow closely large firms making it hard for their managers to manipulate earnings. The debt-to-asset ratio depends on a firm s access to debt market and also may proxy for how well investors and analysts are informed about the firm. Low transparency of high-growth firms and firms with high market-to-book ratios makes it easy for mangers to manipulate earnings and conceal earnings manipulation. In Table 6, we use ordinary least-squares regression analysis to examine the relation between discretionary accruals and reliance on external equity financing in Year 0 while controlling for firm characteristics. In the first three models, we examine winsorized discretionary accruals, while in the next three models we examine nonwinsorized discretionary accruals. As in Table 4, we use three discretionary accruals measures: discretionary total accruals, discretionary current accruals, and discretionary

19 18 current accruals excluding depretiation. To proxy for reliance on equity financing instead of debt financing, we use the ratio of equity financing to external financing in Year 0. 2 If earnings manipulation intended to mislead investors is the prevailing factor distorting discretionary accruals, we expect the estimated coefficients for the ratio of equity financing to external financing to be positive. If managerial overoptimism is the prevailing factor, we expect the estimated coefficients to be negative. The first three models in Table 6 show the inverse relation between discretionary accruals and the ratio of equity financing to external financing, significant at the 1% level. The statistical significance of the findings is even stronger than it was in univariate comparisons of the debt- and equity-financing samples (Table 5). Although managers have stronger incentives to manipulate earnings before equity issues than before debt issues, we find that equity financing is not associated with higher discretionary accruals than debt financing. Instead, consistent with the managerial overoptimism hypothesis, we find higher discretionary accruals for firms relying on debt financing. The statistical significance of the controlling variables varies depending on the measure of discretionary accruals. Only the relation between the market value of equity and discretionary accruals is significant at the 1% level in all three models. Lower discretionary accruals for larger firms may be due to the closer following by analysts and investors. 2 To proxy for reliance on equity financing instead of debt financing, we also used a dummy variable equal to one when a firm belongs to the equity-financing sample and equal to zero otherwise. As the results were essentially the same, we did not present them in a table.

20 19 Examining the correlations among the independent variables used in the models, we find that these correlations are in between and 0.24 (not presented in a table). To insure that there are no estimation problems due to multicollinearity, we estimate condition indexes for all three models. In all models, we find that the highest condition index is bellow 8, indicating no problems due to multicollinearity (not presented in a table). Comparison of the first three models and the next three models in Table 6 shows that winsorizing does not have significant effect on our results. The statistical significance of the results is only slightly higher when winsorized values are used. VI. Robustness Tests To examine the robustness of the inverse relation between discretionary accruals and reliance on equity financing instead of debt financing, we perform subperiod analysis, examine restricted samples, and use alternative measures of discretionary accruals. Tables 7, 8, and 9 show the results of these tests. Except for the differences described bellow, all models in these tables are constructed in the same way as the first three models in Table 6. Furthermore, to test whether mangers preferring debt financing are more overoptimistic than managers preferring equity financing, we examine insider trading around equity issues.

21 20 Subperiods To check the robustness of the results across time, we split the external-financing sample into two subperiods. The first three models in Table 7 examine the first half of the sample ( ), and the last three models in Table 7 examine the second half of the sample ( ). Consistent with the managerial overoptimism hypothesis, analysis of the subperiods shows that reliance on equity financing instead of debt financing is associated with lower discretionary accruals in both subperiods. In all six models, the estimated coefficients for the ratio of equity financing to total financing are negative and significant at the 10% or higher level. Effect of Small and Distressed Firms In Table 8, we examine the effect of small firms and firms with high probability of bankruptcy. As some small firms have limited access to external financing, their managers cannot freely choose between debt and equity financing but instead have to rely on financing they can obtain. To insure that firms with limited access to external financing are not affecting the results, in the first three models in Table 8, we examine only firms belonging to the top three size (market value of equity) quartiles in the CRSP database. We identify the top three quartiles each year. The results with restricted sample are very similar to the results obtained with the full samples. Small firms are not driving the inverse relation between discretionary accruals and reliance on equity financing. Another group of firms that may experience limited access to external financing markets are firms in financial distress. Furthermore, distressed firms may have binding

22 21 debt covenants. To insure meeting the requirements of covenants, distressed firms may have strong motivation to overstate their earnings even when they rely on debt financing. Earlier evidence about the impact of debt covenants on accruals is mixed. For example, DeAngelo, DeAngelo, and Skinner (1994) find no evidence of inflated discretionary accruals for firms with binding debt covenants, while DeFond and Jiambalvo (1994) find positive discretionary accruals in the year prior to the violation of debt covenants. To identify distressed firms, we examine Altman s bankruptcy scores (Altman, 1968) of external-financing firms. Firms with low probability of bankruptcy are very unlikely to have binding debt covenants. To insure that there is no significant number of firms with binding covenants in the examined sample, we limit the external-financing sample to firms with Altman s bankruptcy scores higher than 3 (firms with low probability of bankruptcy). The last three models in Table 8 present the analysis of firms with the low probability of bankruptcy. We find no evidence that the findings of higher discretionary accruals for debt-financing firms than for equity-financing firms are driven by distressed firms. Alternative Measures of Discretionary Accruals Following Teoh, Welch, and Wong (1998a, 1998b), we estimated discretionary accruals using modified Jones s (1991) model. The modification assumes that an increase in trade receivables for the sample firms is not a part of nondiscretionary accruals. Such assumption overstates discretionary accruals for firms with high sales growth. To insure that our results are not biased by the assumption about trade receivables, we also examine discretionary accruals estimated using unmodified Jones

23 22 (1991) model. In unmodified Jones model, trade receivables are assumed to be a part of nondiscretionary accruals, possibly leading to understated discretionary accruals. The first three models in Table 9 show that the change in the methodology does not have a significant impact on our results. Furthermore, in the last three models in Table 9, we examine discretionary accruals estimated relative to those of industry- and performance-matched firms. We match by the two-digit SIC code and return on assets in Year 0. Kothari, Leone, and Wasley (2005) suggest that such matching enhances the reliability of the results, although, possibly at the price of reduced power of the tests. To insure that matching firms differ significantly from external-financing firms, we exclude the top 35% based on external financing normalized by beginning total assets from the matching sample. Excluding only firms in the external-financing sample (the top 25% based on external financing normalized by beginning total assets) would result in some matching firms raising almost as much external financing as external-financing firms. Table 9 shows that estimating discretionary accruals as a difference between discretionary accruals of a sample firm and a matching firm does not eliminate the inverse relation between discretionary accruals and the ratio of equity financing to external financing. All robustness tests confirm the inverse relation between discretionary accruals and preference for equity financing instead of debt financing. These findings are consistent with the hypothesis that managerial overoptimism has a more significant effect on the size of discretionary accruals than managerial earnings manipulation.

24 23 Alternative Measure of Managerial Overoptimism So far we used a firm s reliance on debt financing to proxy for managerial overoptimism. In this section, we use insider trading as a measure of managerial overoptimism to confirm that managers of debt-financing firms are more optimistic than managers of equity-financing firms. Optimistic managers are less likely to sell shares of their firm and more likely to purchase them than less optimistic managers (e.g., Boyer, Ciccone, and Zhang, 2004). We use insider trading data provided by Thomson Financial Corporation. The definition of insiders in this database includes managers, directors, affiliated persons, investment advisors, and large shareholders. As we are trying to gauge managerial optimism, we examine only trades by managers and directors. Following John and Lang (1991), we estimate insider sale and purchase index as the difference in the number of shares purchased and sold, divided by the sum of shares traded. We estimate the insider sale and purchase index for each firm during each fiscal year including all open market and private purchases and sales of common shares (e.g., Boyer, Ciccone, and Zhang, 2004). We exclude index values based on fewer than five transactions. The index values range from -1 (no purchases) to 1 (no sales) with higher values indicating more purchasing relative to selling and thus higher managerial optimism. For the whole database the average index value is negative since managers often receive firm shares as part of their compensation without having to buy them. Table 10 presents the mean insider sale and purchase index for equity- and debtfinancing firms before and after the substantial financing year (Year 0). For just control see the index reaches the lowest value in Year 0, suggesting a decrease in managerial optimism. For the debt-financing sample the index reaches the highest values in Year -1

25 24 and Year 0, suggesting an increase in managerial optimism. Furthermore, the index values are significantly higher for the debt-financing sample than for the equity-financing sample for three years before Year 0 to two years after Year 0. Thus, debt-financing managers seem to be more optimistic than equity-financing managers. VII. Conclusions We examine discretionary accruals of firms obtaining substantial external equity or debt financing. Consistent with earlier studies, we find that discretionary current accruals peak when firms obtain substantial equity financing. Both discretionary total accruals and discretionary current accruals peak when firms obtain substantial debt financing. Furthermore, discretionary accruals are significantly higher at the time of debt financing than at the time of equity financing. Controlling for firm characteristics, we again find the inverse relation between discretionary accruals and reliance on equity financing instead of debt financing. All robustness tests confirm the inverse relation. Our findings support the hypothesis that managerial overoptimism affects the choice between debt and equity financing and has the prevailing impact on discretionary accruals of firms obtaining substantial financing. This study contributes to understanding of factors affecting financing decisions and factors affecting accounting of discretionary accruals. Managerial earnings manipulation is not the only factor distorting financial statements of firms; in some cases, managerial overoptimism is the prevailing factor distorting financial statements. Overoptimistic managers undertake aggressive business and financial reporting practices

26 25 not because they want to mislead the market, but because they believe in the truthfulness of their actions. Better understanding of the factors distorting financial statements is important to help investors to use the information about discretionary accruals more effectively. Furthermore, understanding of factors distorting discretionary accruals is the first step in improving discretionary accruals accounting. Managerial overoptimism also may help to understand poor stock performance after security issues. As earlier studies show that the size of discretionary accruals is related to post-issue stock performance (e.g., Teoh, Welch, and Wong, 1998b), our findings suggest that managerial overoptimism may be a significant factor causing poor stock performance after security issues. This proposition is consistent with Gombola and Marciukaityte (2007) finding that high-growth firms obtaining substantial debt financing underperform high-growth firms obtaining substantial external equity financing for a number of following years.

27 26 References Altman, E.I., Financial ratios, discriminant analysis and the prediction of corporate bankruptcy, Journal of Finance 23, Boyer, Carol, Stephen J. Ciccone, and Wei Zhang, Insider Trading and Earnings Reporting: Evidence of Managerial Optimism or Opportunism. Advances in Investment Analysis and Portfolio Management (forthcoming). Datta, S., M. Iskandar-Datta, and K. Raman, Debt structure adjustments and longrun stock price performance, Journal of Financial Intermediation 9, DeAngelo, H., L. DeAngelo, and D.J. Skinner, Accounting choice in troubled companies, Journal of Accounting and Economics 17, DeFond, M.L. and J. Jiambalvo, Debt covenant violation and manipulation of accruals, Journal of Accounting and Economics 17, Gervais, S., J.B. Heaton, and T. Odean, Overconfidence, investment policy, and executive stock options, Working Paper, Duke University, University of California at Berkley. Gombola, M. and D. Marciukaityte, Managerial optimism and financing choice for rapidly growing firms, Journal of Behavioral Finance, forthcoming. Guenther, D., Measuring earnings management in response to corporate tax rate changes: evidence from the 1986 tax reform act. Accounting Review 69, Heaton, J.B., Managerial optimism and corporate finance, Financial Management 31, Hribar, P. and D.W. Collins, Errors in estimating accruals: Implications for empirical research, Journal of Accounting Research 40, John, Kose and Larry H.P. Lang, Insider trading around dividend announcements. Journal of Finance 46, Jones, J Earnings management during import relief investigations, Journal of Accounting Research 29, Klaczynski, P.A. and J.M. Fauth, Intellectual ability, rationality, and intuitiveness as predictors of warranted and unwarranted optimism for future life events, Journal of Youth and Adolescence 25,

28 27 Kothari, S.P., A.J. Leone, and C.E. Wasley, Performance matched discretionary accrual measures, Journal of Accounting and Economics 39, Loughran, T. and J.R. Ritter, The new issues puzzle, Journal of Finance 41, Loughran, T. and J.R. Ritter, The operating performance of firms conducting seasoned equity offerings, Journal of Finance 52, Malmendier U. and G.A. Tate, Who makes acquisitions? CEO overconfidence and the market s reaction, Working Paper, Stanford University, Harvard University. Malmendier U. and G.A. Tate, CEO overconfidence and corporate investment, Journal of Finance, Forthcoming. Malmendier U., G.A. Tate, and J. Yan, Corporate financial policies with overconfident managers, AFA 2006 Boston Meetings Paper. Myers, S.C. and N.S. Majluf, Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics 13, Peasnell, K.V., P.F. Pope, and S. Young, Board monitoring and earnings management: Do outside directors influence abnormal accruals? Journal of Business Finance and Accounting 32, Rangan, S., Earnings management and the performance of seasoned equity offerings, Journal of Financial Economics 50, Russo, J.E. and P.J.H. Schoemaker, Managing overconfidence, Sloan Management Review 33, Spiess, D.K. and J. Affleck-Graves, Underperformance in long-run stock returns following seasoned equity offerings, Journal of Financial Economics 38, Spiess, D.K. and J. Affleck-Graves, The long-run performance of stock returns following debt offerings, Journal of Financial Economics 54, Teoh, S.H., I. Welch, and T.J. Wong, 1998a. Earnings management and the underperformance of initial public offerings, Journal of Finance 53, Teoh, S.H., I. Welch, and T.J. Wong, 1998b. Earnings management and the underperformance of seasoned equity offerings, Journal of Financial Economics 50,

29 28 TABLE 1. Chronological Distribution of Financing Samples. \ External-financing Equity-financing Debt-financing sample sample sample Number Percent Number Percent Number Percent Year of events of events of events of events of events of events Total 8, , , Note: We create the external financing sample using the following steps: (1) identify firm-years included in both CRSP and Research Insight databases during 1988 to 2001; (2) exclude firm-years with the fiscal year not equal to 12 months and firm-years with a normalized external financing variable (external financing / beginning total assets) not available; (3) limit the set to the top 25% based on the external financing; (4) if the variables necessary to identify financing samples or the variables necessary to calculate discretionary total accruals are not available, exclude the firm; (5) exclude regulated utilities (SIC codes ), depository institutions (SIC codes ), and holding or other investment offices (SIC codes ); and (6) if the same firm is included more than once in any five-year period, include only the earliest firm-year. The fiscal year during which the firm s external financing is estimated is defined as the event year (Year 0). We estimate equity financing as a change in common equity minus a change in retained earnings during Year 0, and we estimate debt financing as a change in total debt during Year 0. External financing is a sum of equity and debt financing. The equity-financing sample includes firms with external common-equity financing exceeding debt financing at least twice during Year 0, and the debt-financing sample includes firms with debt financing exceeding external common-equity financing at least twice during Year 0. This table reports the distribution of events for each sample by the fiscal year.

30 29 TABLE 2. Industry Distribution of Financing Samples. External-financing Equity-financing Debt-financing sample sample sample Number Percent Number Percent Number Percent SIC of of of of of of Industry code events events events events events events Business services 73 1, Electronic and other electric equipment Chemicals and allied products Industrial machinery and equipment Instruments and related products Oil and gas extraction Communications Wholesale trade--durable goods Health services Engineering and management services Other 2, , Total 8, , , Note: This table reports the distribution of events across two-digit Standard Industrial Classification (SIC) codes. The external-financing sample includes firms that obtained substantial one-year external financing during 1988 to The equity-financing sample includes firms with external common-equity financing exceeding debt financing at least twice during Year 0, and the debt-financing sample includes firms with debt financing exceeding external common-equity financing at least twice during Year 0.

31 30 TABLE 3. Characteristics of Financing Samples. External-financing Equity-financing Debt-financing sample sample sample Mean Median Mean Median Mean Median Market value of equity, $M Market-to-book of equity Debt-to-asset ratio External financing / total assets Equity financing / ext. financing Debt financing / ext. financing Note: The external-financing sample includes firms that obtained substantial one-year external financing during 1988 to The equity-financing sample includes firms with external common-equity financing exceeding debt financing at least twice during Year 0, and the debt-financing sample includes firms with debt financing exceeding external common-equity financing at least twice during Year 0. All variables in this table are estimated at the beginning of or during Year 0. External financing is a sum of equity and debt financing. Equity financing is a change in common equity minus a change in retained earnings, and debt financing is a change in total debt.