Routine Insider Sales and Managerial Opportunism

Transcription

1 Routine Insider Sales and Managerial Opportunism Ashiq Ali Jindal School of Management University of Texas at Dallas (972) Kelsey D. Wei Jindal School of Management University of Texas at Dallas (972) Yinghua Li Zicklin School of Business Baruch College (646) Yibin Zhou Jindal School of Management University of Texas at Dallas (973) November 2013 Abstract Prior research shows that routine insider trades are not informative about future firm performance (Cohen et al. 2012). We examine whether managers enhance their personal gains by manipulating reported earnings upwards prior to their routine sales. We show that the amount of routine sales by top-five managers is positively associated with discretionary accruals as well as overproduction, which reduces per unit cost of sales, and is negatively associated with discretionary expenditures such as R&D. These associations are more pronounced in periods when routine sales are conducted by more than one top-five managers and are not significant for routine sales by non-top five managers. Overall, our results suggest that even routine trades could be associated with managerial opportunism.

2 1. Introduction While earlier studies find that corporate insiders are more informed about their firms prospects (see, e.g, Seyhun, 1986; Ke, Huddart, and Petroni, 2003; Piotroski and Roulstone, 2005), recent studies show fairly weak evidence on the profitability of insider trades after risk adjustments of trading profits (see, e.g, Lakonishok and Lee, 2001; Jeng et al., 2003). Cohen, Malloy and Pomorski (2012) contend that detecting informed insider trades has been difficult because many of these trades are driven by insiders liquidity needs and diversification considerations, as opposed to substantial value-relevant information about their companies. They develop a simple scheme to classify insider trades as either opportunistic or routine based on calendar time patterns in insider trading dates. They find that routine insider trades are not associated with future abnormal returns while opportunistic trades are highly profitable. In this paper, we hypothesize that routine sales of insiders with significant control rights could also be associated with managerial opportunism. Specifically, these insiders may enhance their trading gains by overstating reported earnings ahead of their upcoming routine trades. Following Cohen et al. (2012), we define routine insider sales as those that occur around the same calendar time as in the past during our sample period of 1992 to We then examine whether these routine insider sales are preceded by upward earnings manipulation. Consistent with Cohen et al. (2012), we do not observe negative abnormal returns for the 30-day or 60-day period following routine insider sales, suggesting that routine sales are not informative about future firm performance. 1 On the other hand, we find significantly positive abnormal returns for the period immediately before routine inside sales. For example, the average market adjusted cumulative abnormal returns for the window of 10 trading days before the quarterly 1 We find significantly negative abnormal returns following opportunistic insider sales, confirming Cohen et al. (2012) s finding that opportunistic insider trades are informative about subsequent firm performance. 1

3 earnings announcement to 1 trading day before routine inside sales following the earnings announcement is about 4.8%. These findings suggest that while routine sales are not informative about future firm performance, they could still be exploited by top management to enhance their personal trading gains through the manipulation of reported earnings ahead of such trades. Following prior work, we consider three methods by which a firm may overstate reported earnings: overstating discretionary accounting accruals; reducing discretionary spending, such as R&D and selling, general, and administrative (SG&A) expenses; and overproducing, with the intent to reduce per unit cost of sales. For the sample period of 1992 to 2009, we find that quarterly discretionary accruals are positively associated with the volume of routine trades by top-five executives made during the 60 days following quarterly earnings announcements. A one standard deviation increase in routine inside sales by top-five executives is associated with an increase of 0.32 percent in quarterly discretionary accruals scaled by total assets. This finding is obtained after we include in the discretionary accruals model controls for previously identified determinants related to firm characteristics and confounding corporate activities such as mergers and acquisitions, restructuring, external financing, and write-offs. It suggests that opportunistic managers may distort financial reporting to facilitate their upcoming insider trades, when the occurrence of such trades is predictable (i.e., being routine ). Graham, Harvey and Rajgopal (2005) provide survey evidence that managers engage in real economic activities such as reducing R&D or advertising expenditures to meet short-term earnings benchmarks. Empirically, Bens, Nagar and Wong (2002) and Roychowdhury (2006) show that managers focusing on short-term earnings or stock price targets may try to overproduce in the short-run to lower per unit cost of sales, and reduce discretionary 2

4 expenditures such as R&D. Therefore, we examine whether upcoming routine sales induce top company executives to engage in real activities management to boost reported earnings. We find that top executives impending stock sales are negatively related to the discretionary expenditures during the prior quarter. A one standard deviation increase in top-five managers routine sales is associated with abnormal reduction in R&D and SG&A expenditures of about 1.34 million (i.e., 0.36% of total assets) for the previous quarter. We also find a positive relation between insider routine sales and overproduction in the prior fiscal quarter. Furthermore, we show that all three forms of earnings manipulation activities are more pronounced when more than one top-five insiders make routine sales following the earnings announcements, consistent with the interpretation that the observed relation between routine insider sales and earnings management stems from managerial opportunism, which typically requires the complicity of senior executives (Armstrong et al., 2009). An alternative explanation for our finding is that managers sell their shares following good earnings realizations. Thus, the observed higher discretionary accruals, lower discretionary expenditures, and greater production in the period before routine sales may be coincidental and not deliberate. To the extent that the timing of routine trades is largely predetermined, our findings should be less susceptible to this reverse causality problem. Nevertheless, we conduct several analyses to address the above concern. First, given that routine insider trades follow similar calendar time patterns and are likely to be positively correlated over time, we use adjacent years routine trades in the same calendar month as the instrument for the actual insider sale and conduct two-stage least squares (2SLS) regressions. With this instrumental variable (IV) regression, where the instrumented routine insider sale is unlikely to be influenced by the concurrent earnings outcome, we continue to find that routine insider sales are positively 3

5 associated with the overstatement of reported earnings through high discretionary accruals, overproduction, and the cutting of R&D and SG&A expenses. Second, we focus our analyses on a subset of routine sales that are not only routine in terms of their timing, but also involve relatively constant trading volumes. To the extent that insiders are likely to adjust the magnitude of their sales, if they sell in response to recent earnings realizations, focusing on this subset of routine sales with relatively constant volume should help alleviate the reverse causality concern. We show that our findings are robust to restricting to this subset of more stringently defined routine insider sales. Finally, if managers tend to sell stocks after favorable earnings announcements, and the observed earnings management is coincidental rather than deliberate, we would expect to observe a significant relation between earnings manipulation measures and inside trading by non-top five managers as well. However, we find no evidence of earnings manipulation prior to routine sales by these lower-level insiders. Overall, our results suggest that the top-five corporate executives manipulate earnings prior to routine sales, presumably to boost their trading profits. This study makes several contributions to the existing literature on insider trading and managerial opportunism. First, while insider trading is often suspected of containing non-public information about firm value, empirically it has been difficult to decipher whether and when insider trades are informative. 2 Cohen et al. (2012) suggest that opportunistic, rather than routine, trades should be the focus of scrutiny by investors and regulators. However, by examining potential opportunistic behaviors before routine insider trades, we find that even routine insider trades could be exploited by opportunistic managers to enhance their personal benefits. 2 It is an even more daunting task to detect illegal insider trading. For example, during the past decade, the United States Securities and Exchange Commission (SEC) has on average taken 700 enforcement actions annually. Only around 40 to 50 of these actions are against insider trading, the second lowest number of actions across all eight categories. See 4

6 Therefore, while Cohen et al. (2012) help regulators and investors focus on the information content of opportunistic trading, our study directs their attention to potential managerial opportunism associated with routine trades. Our findings along with those in Cohen et al. (2012) suggest that there exist two alternative channels of opportunistic behaviors aiming at increasing managers personal wealth through insider trades: earnings management prior to routine insider trades, and opportunistic timing of insider trades prior to important information events. Therefore, although regulators have erected regulations to deter insiders from trading on material nonpublic information while affording them the opportunity to preplan their trades for liquidity and diversification purposes (for example, via Rule 10b5-1), our study cautions that such regulations might still be subject to manipulation. By exploring routine insider trades, we also contribute to the literature that attempts to establish the effect of managerial equity incentives on earnings manipulations. Despite prior evidence on the association between earnings management and subsequent personal or corporate transactions (see, e.g., Bergstresser and Philippon, 2006 and Brockman, Khurana, and Martin 2008), establishing the direction of causality has been a challenging task. Since we focus on routine insider trades, our tests better align ex post actions with ex ante incentives and mitigates the concern that our findings might be attributable to the opportunistic timing of insider trading around favorable performance realizations or information events (see, e.g., John and Lang, 1991; Lee, Mikkelson and Partch, 1992; and Ke, Huddart and Petroni, 2003). Moreover, compared with prior studies that infer managers incentives from their compensation structures (e.g., Bergstresser and Philippon, 2006 and Jiang et al., 2010), our focus on routine insider trades allows us to conduct more powerful tests of managerial opportunism. First, examining insider trading allows us to test directional predictions since we can determine 5

7 the exact timing and direction of managers trading needs. Therefore, our setting helps attenuate potential biases that may arise from the correlated omitted variables problem that is a common concern with tests using unsigned measures of manipulative activities (Hribar and Nichols, 2007). 3 Further, managerial incentives derived from equity-based compensation cannot capture cross-sectional variations in realized private benefits from manipulative behaviors. Our approach of focusing on routine insider trades allows us to better quantify the actual gain to managers from manipulations of earnings (Amstrong et al., 2010). 2. Data and Variable Definitions In this section, we first describe the procedure for constructing our sample of routine trades by the top five executives. We then discuss the measurement of opportunistic behaviors by managers: earnings management, including both accrual management and real activities manipulation. 2.1 Identifying Routine Trades by the Top-Five Executives We obtain open market insider trading data from the Thomson Reuters Insider Filing Data Files for the period from 1992 to We focus on the top five executives since they are in the most favorable positions to adjust accruals and real corporate activities for personal trading gains. We identify trades made by the top five executives (CEO, CFO, COO, President, and Chairman of board) via role codes of "CEO", "CFO", "CO", "P", and "CB" provided in the database. Following the prior literature on insider trading (see, e.g., Lakonishok and Lee, 2001 and Sias and Whidbee, 2010), we clean up filing errors by excluding duplicate filings, 3 For example, since earnings management involves the shifting of earnings from one period to another, prior studies on the relation between earnings manipulation and equity-based compensation have relied mostly on unsigned earnings management measures (see, e.g., Bergstresser and Philippon 2006 and Jiang, Petroni, and Wang 2010). This is because one cannot predict the exact timing and direction of earnings management over a short time span based on managers stock and option holdings. 6

8 transactions of either fewer than 100 shares or more than 20% of the firm s total shares outstanding, and transactions whose trade prices deviate from CRSP close prices on the same day by more than 50%. For each executive, we compute her daily sale (purchase) as the number of shares sold (purchased) on a single day as a fraction of the prior-quarter shares outstanding.. Since our tests focus on managers intentional manipulation of corporate activities in order to facilitate personal trading, we need to identify routine trades. One option is examining insiders prescheduled trades that are not based on material nonpublic information under SEC Rule10b5-1. However, Jagolinzer (2009) finds that insiders strategically adjust the timing of these plans to trade on private information: initiating sales plans before bad news and terminating sales plans before good performance. In addition, 10b5-1 plans do not exist before 2000 and are only voluntarily disclosed by a small sample of firms. Following Aboody and Kasznik (2000) and Cohen, Malloy, and Pomorski (2012), we then identify top executives' routine trades by examining calendar time patterns in their historical trading dates. 4 Specifically, a sale is classified as routine if an insider s sale date falls in the same calendar month for at least three consecutive years. 5 This approach of identifying routine trades allows us to better capture the liquidity- and diversification-related trading needs of insiders since they tend to follow consistent time patterns. For example, a CEO may routinely sell her shares of the firm after she exercises option grants. Since stock options are often granted under a fixed schedule and thus periodically become vested, the CEO tends to conduct routine insider sales. Alternatively, she may routinely use 4 According to Cohen, Malloy and Pomorski (2012), the differences in abnormal returns associated with routine versus opportunistic insider trading are not driven by the establishment of these plans. 5 We allow gaps between routine trades occurring in the same calendar month over multiple years. For example, a CEO may sell in January of 1997, 1998, 1999, and However, routine trades in non-consecutive years (that is, 2001 in this example) only account for 10% of all routine trades. In unreported analyses, we find that all of our main findings remain unchanged after we exclude non-consecutive routine trades from our analyses. 7

9 annual bonuses (that are often awarded around the same time of the year) to purchase company stocks. Given the large equity holdings of top executives and the associated diversification and liquidity needs, the frequency of routine insider purchases is significantly lower than that of routine insider sales, especially in earlier years. Our analyses thus will focus on potentially manipulative behaviors before routine insider sales. When investigating the impact of insider trading incentives on accruals management and real activities manipulation, we employ a fiscal quarter approach and construct time series of top executives' routine trades subsequent to the end of each fiscal quarter. Specifically, we aggregate routine sales by the top five executives over the 60 days following each quarter's earnings announcement, as illustrated in Figure 1. Our final sample consists of 7,067 quarterly observations of the top five executives' routine sales across 1,820 firms during the period from 1992 to This accounts for about 27% of all insider trades by top five executives during the same period. 6 Table 1 presents descriptive statistics of our sample of routine trades. The first column of the table reports the yearly frequency of firms with routine trades. During our sample period the number of firms with routine sales by top-five executives has increased steadily since early 1990s before beginning to decline slightly after The recent decline in routine sales is likely driven by option-based compensation cutbacks following the stock option expensing regulation of 2004 (SFAS 123R). Columns 2 and 3 report the average number and amount of routine sales as a percentage of the prior-quarter market capitalization for each firm in each year, respectively. In untabulated results, we find that routine insider sales appear to be highest during the first 6 Note that the portion of routine trades we identify is lower than that reported in Cohen et al. (2012). This is because of the focus of Cohen et al. (2012) on routine traders. That is, once an insider is identified as a routine trader based upon her trading pattern during a specific period of time, all of her future trades are considered as routine trades. In contrast, we use a rolling process to identify routine trades as opposed to routine traders. 8

10 quarter and lowest during the last quarter of a fiscal year in terms of the dollar amount, presumably because option grant dates often coincide with fiscal-year end dates (Norton and Porter, 2010). 7 In addition, about 75% of our sample firms only have routine sales in one of the four fiscal quarters. We define top executives' routine trades by examining time patterns in their trading dates. To shed light on the variation in the quantities of those trades, for each top executive's routine trading string, we calculate the following variability ratio: / 2 where Max is the largest transaction in a top manager's routine sales string, Min is the smallest transaction, and Mean is the average quantity of her routine trades. Figure 3 depicts the distribution of this ratio. It indicates that in terms of the trading quantity, top executives' routine sales are relatively stable over time. 2.2 Measuring Earnings Management To investigate whether top executives manage earnings to temporarily boost the stock price ahead of their routine sales, we examine both accrue based earnings management and real activities manipulation. Following DeFond and Jiambalvo (1994) and Subramanyam (1996), we measure accrual management as discretionary accruals estimated from the cross-sectional Jones (1991) model. First, we estimate nondiscretionary accruals as a function of the change in revenue and the level of property, plant, and equipment: 1, (1) 7 Consistent with this observation, Cohen et al. (2012) find more routine trades than opportunistic trades in the quarter following a fiscal year-end. Specifically, about 75% of the insider sales after fiscal year-end are routine sales. 9

11 where ACCR it is total accruals for firm i in quarter t measured as the difference between earnings before extraordinary items and discontinued operations, and cash flows from operations reported in the statement of cash flows. TA it-1 is total assets in quarter t-1, REV it is the change in sales revenues from the previous year, and PPE it is gross value of property, plant and equipment. We estimate the above model cross-sectionally within each two-digit SIC code and year combination. A combination with less than ten observations is dropped from the sample. Our measure of discretionary accruals, DA, is defined as the residuals from Equation (1) 8 : 1. (2) All financial data required to construct the discretionary accruals are extracted from COMPUSTAT. Prior studies have documented that managers also manipulate real activities to temporarily boost current earnings in three ways: 1) cutting discretionary expenditure (R&D, advertising, and maintenance), 2) accelerating sales through excessive price discount or more lenient credit terms, and 3) lower cost of goods sold per unit by overproduction. Following Roychowdhury (2006), Cohen et al. (2008), and Cohen and Zarowin (2010), we consider two proxies for real earnings management: abnormal levels of discretionary expenses and production costs. 9 First, the cutting of discretionary expenditures should lead to abnormally low 8 We also utilize discretionary accruals estimated from the modified Jones (1991) model as an alternative measure of earnings management, and find similar inferences. In addition, our results are robust to the performance-adjusted discretionary accruals measures proposed by Kothari et al. (2005). 9 Prior studies also use abnormal cash flows from operations as a proxy for real earnings management. However, as noted in Roychowdhury (2006), while sales manipulation and overproduction have a negative effect on abnormal CFO, reduction of discretionary expenditures has a positive effect. We don't include abnormal CFO as a real earnings management measure because of the resulting ambiguous net effect. 10

12 discretionary expenses relative to sales. To estimate normal level of discretionary expense, we run the following regressions for every industry and year: 10 1, (3) where DISEXP it is discretionary expenses for firm i in quarter t measured as the sum of advertising expenses, R&D expenses and SG&A. SALE it-1 is sales revenues in quarter t-1. The abnormal discretionary expenses, ABDISEXP, is defined as the residuals from Equation (3): 1. (4) Sales manipulation via excessive price discounts and overproduction could lead to abnormally high production costs relative to sales. Production costs are defined as the sum of cost of goods sold (COGS) and change in inventory. Normal level of COGS can be expressed a linear function of contemporaneous sales and normal inventory growth can be expressed as a linear function of the contemporaneous and lagged changes in sales: 1, (5) 1, (6) where COGS it is cost of goods sold for firm i in quarter t and INV it is the change in inventory. Therefore, we model the normal level of production costs using the following equation: 1, (7) where PROD it is production costs in quarter t. The abnormal production costs ABPROD, is computed as the residuals from Equation (7): 10 According to Dechow et al. (1998), discretionary expenses can be expressed as a linear function of contemporaneous sales under the simplifying assumptions. But the model creates a mechanical problem when firms manage sales upward to increase earnings in any year, resulting in abnormally low residuals. To address this issue, we model discretionary expenses as a function of lagged sales instead. 11

13 1. (8) If top executives manage earnings upwards by manipulating real firm activities, we expect that holdings sales constant, firms should exhibit abnormally low discretionary expenses and/or abnormally high production costs in the quarters immediately prior to their routine sales. 2.3 Control Variables Throughout the paper, we have included a battery of control variables that may also influence firms financial reporting and real investment decisions. Detailed descriptions on the construction of these variables are provided in the Appendix. 3. Abnormal Returns Surrounding Routine Insider Sales Cohen et al. (2012) show that routine insider trades are less predictive of future stock returns relative to opportunistic trades. The objective of our study is to investigate whether top executives opportunistically manage earnings outcomes, i.e., the timing of information flows, before their routine trades to increase their personal trading gains. Specifically, we conjecture that top executives could maximize the profits from their routine sales by managing earnings upward (through both accrual and real activities manipulation) and rushing (delaying) good (bad) news. As a result, our hypotheses predict positive abnormal returns immediately prior to routine insider sales and no significant negative returns following routine sales. We note that routine sales may not be followed by immediate low subsequent returns because it may take a long time for earlier manipulative actions to be eventually reversed, especially for those involving real firm 12

14 activities whose long-term impacts on the firm tend to be subtle (see, e.g., Zang, 2012 and Alan, Larson and Sloan, 2013). 11 Specifically, for each insider routine sale event including top-five officers routine sales aggregated over the adjacent 60-day period following each earnings announcement day, we compute the cumulative market adjusted returns during the period from 5-day preceding the quarterly earnings announcement date to 1-day preceding the first routine insider sale of the event. We similarly compute stock returns during various windows starting from the last routine insider sale of each insider trading event. For comparison, we also compute abnormal returns surrounding non-routine insider sales (i.e., opportunistic sales) that are conducted during the same 60-day period following each earnings announcement, as Cohen et al. (2012) show that the timing of opportunistic insider trades is often strategically chosen such that opportunistic trades strongly predict subsequent abnormal returns. Figure 2 indicates that the average market-adjusted cumulative returns from 5-day preceding the quarterly earnings announcement date to the first routine sale in each routine sale event is 2.8% and statistically significant at the 1% level, consistent with our hypothesis that insiders manipulate firm activities to achieve favorable earnings news before their upcoming trades. As expected, we also find a significant 5.1% average abnormal return during the same period before opportunistic insider sale events. As to stock returns following insider sales, we do not observe any significant abnormal returns following routine insider sale events. The average cumulative abnormal returns are insignificant across all windows following each routine sale event. In stark contrast, opportunistic trades happening during the same period following earnings announcements are followed by significantly negative abnormal returns. For example, 11 In unreported analysis, we find the fiscal quarter immediately prior to a routine insider trading event is typically associated with the largest earnings announcement returns, relative to the four quarters following the insider trading event. 13

15 during the 60-day period after the last opportunistic inside sale, the underlying stock experiences a significantly negative abnormal return of 1.55%, as compared to an insignificantly return of % following the last routine insider sale during the same period. Our findings on stock returns before and after routine insider sales are thus consistent with those of Cohen et al. (2012) that routine insider trades do not help predict future stock returns while opportunistic trades do. However, in contrast to Cohen et al. (2012) which focus on whether opportunistic versus routine insider sales help predict future stock returns, our focus is on possible managerial opportunism prior to upcoming routine sales. That is, routine sales, due to their predictable timing, could provide managers the incentive to manipulate firm activities to produce the desired firm performance. The significantly positive abnormal returns prior to routine trades, as illustrated in Figure 2, provide preliminary support to this hypothesis. We then proceed to examine whether these abnormal return patterns are associated with abnormality in firms financial reporting and real activities. 4. The Effect of Top Executives' Routine Trades on Earnings Management 4.1 Baseline analyses While a large body of literature has linked top managers equity incentives arising from stock-based compensation to irregularities in financial and real firm activities, the empirical evidence is mixed. Some studies find that earnings management is more pronounced in firms where executive pay is more closely tied to stock prices (see, for example, Bergstresser and Philippon, 2006; Burns and Kedia, 2006; Jiang, Petroni and Wang, 2010; and Feng et al., 2011). Others do not find evidence of a positive association between managerial equity incentives and myopic behaviors. For example, using propensity-score matching research design, Armstrong, 14

16 Jagolinzer, and Larcker (2010) indicates that the level of CEO equity incentives has a modest negative relationship with the incidence of accounting irregularities. The mixed evidence above reflects difficulties in establishing the causal effect of equity incentives inferred from top managers' annual compensation on the occurrence of accounting irregularities during the year. In this study, we take a more direct approach by investigating whether top executives explicit trading incentives may induce earnings management during the immediate prior period. This research design should have greater power to identify top executives opportunistic behaviors to serve their personal interests because we can identify the exact time and direction of top executives short-run equity incentives. Therefore, our research design can better capture executives short-term opportunistic behaviors that may be more difficult to detect over longer horizons. We test whether trading incentives motivate executives to manipulate accounting earnings for personal gain by estimating the following equation:, (9) where EM is the various financial and real earnings management measures during fiscal quarter t, including the signed discretionary accruals, abnormal discretionary expenditure and overproduction. The measurement procedures are described in Section 2.2. Routine Sale represents top executives' routine sales that take place within 60 days following the earnings announcement for fiscal quarter t. We focus on the 60-day period post earnings announcement because previous studies have documented that the majority of the firms have self-imposed insider trading policies 15

17 prohibiting insider trading at all times except for the period following the quarterly earnings announcement (Bettis, Coles and Lemmon, 2000). Following prior studies (Ashbaugh-Skaife et al., 2008; Geiger et al., 2006; Katz, 2009), we control for a comprehensive set of firm characteristics that may be related to discretionary accruals and expenditures: total assets, BM, sales growth, leverage, cash flows, operating cycle, capital intensity, firm age as of the pretrading fiscal quarter, and a set of dichotomous variables indicating significant corporate events that may affect earnings management, such as merger and acquisition, restructuring, external financing, as well as assets write-offs. When the dependent variable is discretionary accruals, we also include lagged discretionary accruals in the regression to account for the autocorrelation and possible reversal of accruals. Detailed definitions of these control variables are presented in the Appendix. We include year fixed effects in all regression analyses and calculate t-statistics using heteroskedasticity-robust standard errors clustered by firm. Note that in contrast to prior studies on the relation between earnings management and managerial equity incentives derived from managers equity based compensation, (e.g., Bergstresser and Philippon, 2006; Jiang, Cohen et al., 2008; Petroni, and Wang, 2010), our model tests the directional relation between managerial trading incentives and abnormal accruals. This approach allows us to avoid any potential endogeneity problem arising from omitted variables that may be correlated with unsigned accruals. For example, since earnings management involves the shifting of earnings from one period to another, prior studies on the relation between earnings manipulation and equity incentives have relied mostly on unsigned earnings management measures because one cannot determine the exact timing of earnings management based on managers stock and options holdings. This could potentially bias tests in favor of rejecting the null hypothesis of no earnings management due to the correlation between 16

18 certain firm characteristics and unsigned earnings management variables (Hribar and Nichols, 2007). The advantage of our insider trading setting is that we can identify the exact time and direction of managers trading needs, and therefore do not have to rely on unsigned accruals management measures. Table 2 reports descriptive statistics for routine trades along with descriptive statistics for the variables used in the estimation of Equation (9) during our sample period. The results indicate that the average discretionary accruals is 0.00 with a standard deviation of The next two rows report our proxies for real earnings management. The average abnormal discretionary expenditure (over-production) is 0.02 (-0.03) with a standard deviation of 0.07 (0.06). Our sample contains 6,458 firm-quarters of top executives' routine sales. The average (median) aggregated routine sales of the top 5 executives during each quarter is 0.23% (0.08%). In our regression analyses, the sample size varies depending on the availability of earnings management measures and control variables. Table 3 reports the coefficient estimates for Equation (9). Column 1 shows the effect of top executives' routine sales on the accruals management decision during the prior quarter. The result indicates that the coefficient on Routine Sale is 0.45 (t-statistic = 4.05). This evidence suggests that top executives stock sale incentives induce income-increasing accruals management during the prior quarter. The coefficient on the Routine Sale variable suggests that a one percentage point increase in routine sales by top-five managers is associated with a 45 basis point increase in the value of firm financial accruals, which is very much comparable to the effect of CEO equity-based incentives on the absolute value of accruals as documented in Bergstresser and Philippon (2006). This effect is also economically important. Specifically, one standard deviation increase in their routine sales (approximately $47,459) is associated with an 17

19 increase of 0.32% (as a percentage of total assets) in quarterly discretionary accruals. Based upon the median total assets of $ million, this would translate into an increase of discretionary accruals of 1.19 million dollars prior to routine insider sales. In addition to accruals management, existing theoretical and empirical studies suggest that managers who care about stock prices may also manage real operational activities to boost earnings and thus stock performance (e.g. Stein 1989; Bebchuk and Stole 1993; Graham, Harvey and Rajgopal 2005; Roychowdhury 2006; Cohen and Zarowin 2010; Zang 2012). For example, in their survey, Graham et al. (2005) report that 80% (55%) of survey participants state that they would decrease discretionary R&D, advertising, and maintenance (delay starting a new project) to meet an earnings target. Roychowdhury (2006) suggests that in order to avoid reporting losses, managers are reducing of discretionary expenditures (advertising, R&D, and SG&A) to improve reported margins and overproducing to report lower cost of goods sold. Cohen and Zarowin (2010) document that firms use real, as well as accrual-based, earnings management activities around seasoned equity offerings. In fact, self-dealing managers may even prefer real activities management over accrual manipulation. This is the case because real earnings management is arguably less likely to be scrutinized by regulators and auditors, and thus less likely to get caught. Therefore, we next investigate top executives strategically alter real operational activities in response to their trading incentives. Specifically, we examine whether insiders' routine sales affect firms decisions on discretionary expenditures and production scales in the immediate prior quarter. The results for real activities management are presented in Columns 2 and 3 of Table 3. Column 2 reports the coefficient estimates for Equation (9) when the dependent variables is abnormal discretionary expenditures, as calculated using Equation (4). We find strong evidence 18

20 that top executives' impending stock sales are negatively related to discretionary expenditures during the previous quarter. Specifically, a one standard deviation increase in top managers' routine sales would induce them to cut advertising, R&D and SG&A expenditures by approximately 1.34 million (i.e., 0.36% of total assets) during the previous quarter. Column 3 reports the coefficient estimates for Equation (9) when the dependent variables is the overproduction, as calculated using Equation (8). The coefficient on Routine Sale is positive (0.31) and statistically significant (t = 2.66). This implies that a one standard deviation increase in top managers' routine sales is associated with an overproduction of approximately 0.82 million (i.e., 0.22% of total assets) in the prior quarter. In summary, we find strong evidence that top executives' routine stock sales (purchases) are preceded by upward accruals management as well as real operational activities manipulation such as the cutting of discretionary expenditures and short-term overproduction. While existing studies usually focus on the relationship between ex ante measures of executives equity incentives (e.g., equity portfolio delta) and earnings management, the realized benefits from opportunistic behavior not only depend on managerial incentives, but also managers ability to liquidate their equity holdings to benefit from manipulating earnings (Armstrong et al and 2010). Therefore, our finding of earnings manipulation before routine insider sales represents more direct evidence on managerial opportunism The complicity of incentives among top executives Our evidence so far suggests that at least some top managers engage in manipulative activities to enhance personal gain from their routine stock trading. Armstrong et al. (2009) argue that managerial opportunism typically requires the complicity of senior executives. Therefore, we expect that the relation between these opportunistic behaviors and routine trades to be 19

21 stronger when there are more than one senior executives trading together. We test this conjecture by augmenting Equations (9) and (10) with an interaction term between Routine Sale and an indicator variable that equals 1 if the number of trading insiders is greater than 1, and zero otherwise. Table 4 presents the results for the three earnings management measures. First, consistent with results reported in Table 3, we observe a significantly positive (negative) association between insiders routine sale and the prior quarter's discretionary accruals and overproduction (abnormal discretionary expenditure level). More interestingly, we observe a significantly positive (negative) coefficient on Dummy for multiple Traders*Routine Sale when the dependent variable is discretionary accruals (abnormal discretionary expenditure). The coefficient on the interaction term is positive but insignificant for overproduction. These results suggest that when multiple executives are trading together, firms involve in even more pronounced accruals manipulation and discretionary expenditure reduction in the prior period. Overall, the evidence in this section further supports our hypothesis that it is managers equity incentives, which become stronger when multiple executives need to conduct immediate stock sales, which drive the irregularities in financial reporting and real activities. 5. Assessing the Causal Relation between Routine Insider Sales and Earnings Management Since previous studies have documented that insiders are contrarian traders (Rozeff and Zaman, 1998 and Piotroski and Roulstone, 2005), one potential concern for the observed relation between insider trading and prior manipulative activities on earnings is that it may be driven by managers strategic timing of their trades around favorable earnings outcomes. However, our unique research setting of employing routine insider trades makes this less likely to be the case. 20

22 Since we determine routine insider trades based on calendar time patterns in past trading dates, the direction of the trade as well as the decision to trade is largely predetermined. This setting ensures that our results are unlikely to be entirely explained by the effect of earnings management on insider sales. Nonetheless, in this section we explicitly address the potential reverse causality problem from several angles, through both instrumental variable regressions and cross-sectional analyses. 5.1 An Instrumental Variable Approach Since routine inside sales are likely to capture insider trading due to liquidity and diversification concerns, rather than active responses to realized firm performance, we should expect to observe high correlations between routine sales over time. This feature of routine insider trades allows us to conveniently use routine sales conducted in the same calendar month but in adjacent years as the instrument for current period s actual routine sales. Since past or future years insider trades during the same calendar month are correlated with actual insider sales in the current month, and are not endogenously affected by earnings outcomes in the current period, they serve as ideal instrumental variables (IVs) in the two-stage least squares (2SLS) regression we employ to address the endogeneity problem. Specifically, for each routine insider sale conducted during the 60-day period after earnings announcement in quarter t, we identify the routine insider sale in the previous year of the corresponding month as the IV for the actual sale. If routine insider trade in the immediate prior year is missing, we then use routine insider sale in the same month during the following year relative to quarter t as the substitute. If both routine insider sales in the year immediately before and those in the year immediately after quarter t are available, we use the average of the two as the instrument variable. After we find the IV for each top-five insider s routine sale in the 21

23 60-day period after earnings announcement of quarter t, we then compute the total, Routine Sale_IV, as the instrument variable in the first stage regression of actual routine inside sales on past public information (i.e., all the exogenous variables in the 2 nd stage regression). The correlation between the actual routine sale of top-five managers and the instrumented one is around Table 5 presents the second-stage regression results for all three earnings management measures. To ensure that the inference from the comparison between OLS and 2SLS IV regression estimates is not confounded by the scaling differences between raw routine sale and the instrumented one, we standardize routine sales in both regressions to have a mean of zero and standard deviation of one. Consistent with our hypothesis, the instrumented routine insider sales are significantly positively related to accruals management, discretionary expenditure, and over production. In terms of the economic impact, the effect of routine sales on quarterly discretionary accruals under the IV approach is about the same as that under the OLS approach. Similarly, increases in routine insider sales lead to significantly lower discretionary expenditures and greater over productions. Since the instrumented routine sales, routine sales in adjacent years during the same calendar month, should bear little relation to the current year s earnings outcome, the evidence from the instrumental variable regression strongly suggests that our findings on the relation between routine insider sales and accruals and real earnings management are more likely to be attributed to insiders manipulating earnings to enhance personal trading gains, as opposed to good earnings outcomes leading to insider sales. 5.2 Stable routine trades In our prior analyses, we identify top executives' routine trades by examining the pattern of their trading dates. In this sub-section, we check the robustness of our findings by adopting a 22

24 more stringent definition of routine trades in terms of both the time pattern and the trading quantity. For a trade to be considered as routine trades, we require not only that top managers trade in the same calendar month in at least three consecutive years, but also that their trading needs be relatively constant over the same period. Specifically, we restrict our analyses to a subsample of routine trades with Ratio, as defined in section 2.1, smaller than or equal to 1. We then re-run Equations (9) and (10) on this more restricted sample of routine trades. Table 6 indicates that top managers' prescheduled trades are significantly associated with accruals manipulation and real earnings management activities in the preceding period. The effects of these more stable managerial routine trades on opportunistic firm behaviors are economically and statistically similar to what we find earlier using the full sample of routine trades. 5.3 The effect of non-top five executives' routine trading on managerial opportunism To further mitigate the potential concern on the direction of causality, we investigate whether routine sales executed by corporate directors and officer other than the top-five executives are also related to managerial manipulation activities in the prior period. The idea is that if good or bad stock performance motivates insiders to sell or buy, then it should similarly affect both the top five and non-top five executives. However, if the causality is the other way around, then the top five executives' routine trades should have a much stronger with any observed abnormality in accounting and real firm activities because they are the ones who have the ultimate control over various firm decisions. Following prior literature (e.g., Rozeff and Zaman 1998; Piotroski and Roulstone 2005; and Sias and Whidbee 2010), we define non-top five executives as the directors and officers of a company other than its CEO, CFO, COO, President, and Chairman of board. We measure non-top five executives' routine trades following the same procedure as described in Section

25 Table 7 reports the coefficient estimates of Equation (9) for the sample of non-top five managers' routine trades. The coefficients on Routine Sale are unanimously insignificant for all three measures of earnings management. That is, unlike top-five executives, non-top five managers routine trades are not related to any earnings management activities in the prior quarter. Therefore, it is unlikely that our earlier findings can be simply interpreted as insiders selling their holdings when they observe favorable earnings realizations. Because if that were the case, we would have observed a similarly significant relation between measures of earnings management and routine insider sales conducted by non-top five executives. Therefore, our finding that only top-five managers' trading incentives, but not those of other executives, play a role on earnings management further corroborates a causal relation between managerial incentives and firm decisions. Lastly, to ensure that the observed differences in firm activities before routine insider sales by top-five versus non-top five insiders are not driven by differences in firm characteristics across firms or periods when the two groups of insiders conduct their trades, respectively, we conduct a matched sample analysis. That is, we restrict the analysis in Table 7 to a subset of routine sales by non-top five insiders that happen within the same calendar year (as those by topfive managers. The results in the last three columns of Table 7 are highly consistent with those with the full sample of non-top five insider trades, both qualitatively and quantitatively. 6. Conclusions In this study, we show that managers trading incentives significantly influence both the accruals and real activities management of prior fiscal quarter s reported earnings of their firm. Specifically, we find that routine insider sales by top-5 managers are positively associated with prior quarter s discretionary accruals, negatively associated with prior quarter s discretionary 24

26 expenditures (including R&D and SG&A expenditures) and positively associated with prior quarter s overproduction (which can reduce per unit cost of sales). These findings indicate that managers overstate reported earnings to increase personal trading gains. We also show that the associations between insider trading and earnings management activities are more pronounced when more than one top-five insiders make routine sales following the earnings announcements. This result is consistent with the interpretation that the observed relations between routine insider sales and earnings management activities stem from managerial opportunism, which typically requires the complicity of senior executives. Furthermore, our results remain robust when we correct for insider trading that occurs in response to earnings outcomes. We estimate instrumental variable regressions, using routine insider sales of individual top-five managers in adjacent years as the instrument. If routine sales are carried out for liquidity or diversification reasons, then routine sales conducted in the same calendar month of adjacent years are likely to be correlated with the actual sale in the current period but are unlikely to be influenced by current period s realized earnings. Our instrumental variable regression analysis helps establish the causal influence of routine insider trading on overstatement of reported earnings. In a related test, we further demonstrate that overstatement of reported earnings is positively associated with a subset of routine insider sales that are not only routine in timing, but are also routine in magnitude. This result suggests that our findings are not driven by managers tendency to conduct insider trades in response to favorable earnings news. Also consistent with the agency problem hypothesis, we show that earnings overstatement activities are positively associated with routine trading of top-5 managers, but not with routine trading of non top-five managers. 25

27 Our study uncovers the multifaceted nature of managerial opportunism associated with insider trading by examining the link between manipulative firm behaviors and routine insider trades. The existing literature suggests that opportunistic trades are informative about future firm performance and routine trades are not. We show that even when managers do not time their insider trades to take advantage of their private information, as is the case with routine trades, they may still enhance their personal gain by engaging in opportunistic firm behaviors before such trades. 26

28 References Aboody, D., Kasznik, R., CEO stock option awards and the timing of corporate voluntary disclosures. Journal of Accounting and Economics 29, Allen, E., Larson, C., Sloan, R., Accrual reversals, earnings and stock returns. Journal of Accounting and Economics, forthcoming. Anilowski, C., Feng, M., Skinner, D., Does earnings guidance affect market returns? The nature and information content of aggregate earnings guidance. Journal of Accounting and Economics 44, Armstrong, C., Larcker, D., Discussion of The impact of the options backdating scandal on shareholders" and "Taxes and the backdating of stock option exercise dates. Journal of Accounting and Economics 47, Armstrong, C., Jagolinzer, A., Larcker, D., Chief executive officer equity incentives and accounting irregularities. Journal of Accounting Research 48, Baginski, S., Hassell, J., Kimbrough, M., The effect of legal environment on voluntary disclosure: Evidence from management earnings forecasts issued in U.S. and Canadian markets. The Accounting Review 77, Bebchuk, L., Stole, L., Do short term managerial objectives lead to under- or overinvestment in long-term projects? Journal of Finance 48, Bens, D., Nagar, V., Wong, M., Real investment implications of employee stock option exercises. Journal of Accounting Research 40, Bergstresser, D., Desai, M., Ruah, J., Earnings manipulations, pension assumptions, and managerial investment decisions. Quarterly Journal of Economics 121, Bergstresser, D., Philippon, T., CEO incentives and earnings management. Journal of Financial Economics 80, Bettis, C., Coles, J., Lemmon, M., Corporate policies restricting trading by insiders. Journal of Financial Economics 57, Burns, N., Kedia, S., The impact of performance-based compensation on misreporting. Journal of Financial Economics 79, Chapman, C., Steenburgh, T., An investigation of earnings management through marketing actions. Management Science 57, Cheng, C., Huang, H., Li, Y., Lobo, G., Institutional monitoring through shareholder litigation. Journal of Financial Economics 95, Cheng, Q., Warfield, T., Equity incentives and earnings management. The Accounting Review 80,

29 Cohen, D., Mashruwala, R., Zach, T., The use of advertising activities to meet earnings benchmarks: Evidence from monthly data. Review of Accounting Studies 15, Cohen, L., Malloy, C., Pomorski, L., Decoding inside information. Journal of Finance 67, Dechow, P., Sloan, R., Sweeney, A., Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research 13, Erickson, M., Hanlon, M., Maydew, E., Is there a link between executive equity incentives and accounting fraud? Journal of Accounting Research 44, Fama, E., French, K., Industry costs of equity. Journal of Financial Economics 43, Fazzari, S., Hubbard, R., Petersen, B., Financing constraints and corporate investment. Brookings Papers on Economic Activity 1, Feng, M., Ge, W., Luo, S., Shevlin, T., Why do CFOs become involved in material accounting manipulations? Journal of Accounting and Economics 51, Graham, J., Harvey, C., Rajgopal, S., The economic implications of corporate financial reporting. Journal of Accounting and Economics 40, Hart, O., The market mechanism as an incentive scheme. Bell Journal of Economics 14, Hribar, P., Nichols, C., The use of unsigned earnings quality measures in tests of earnings management. Journal of Accounting Research 45, Jagolinzer, A., SEC rule 10b5-1 and insiders strategic trade. Management Science 55, Jensen, M., Meckling, W., Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, Jiang, J., Petroni, K., Wang, I., CFOs and CEOs: who have the most influence on earnings management? Journal of Financial Economics 96, John, K., Lang, L., Insider trading around dividend initiation: Theory and evidence. Journal of Finance 46, Jones, J., Earnings management during import relief investigations. Journal of Accounting Research 29, Katz, S., 2009, Earnings quality and ownership structure: The role of private equity sponsors. The Accounting Review 84,

30 Karpoff, J., Lee, D., Martin, G., 2008a. The consequences to managers for financial misrepresentation. Journal of Financial Economics 88, Karpoff, J., Lee, D., Martin, G., 2008b. The cost to firms of cooking the books. Journal of Financial and Quantitative Analysis 43, Ke, B., Huddart, S., and Petroni, K., What insiders know about future earnings and how they use it: Evidence from insider trades. Journal of Accounting and Economics 35, Kim, J.B., Li, Y., and Zhang, L., CFOs versus CEOs: Equity incentives and crashes. Journal of Financial Economics 101, Kothari, S., Leone, A., Wasley, C., Performance matched discretionary accrual measures. Journal of Accounting and Economics 39, Lee, D., Mikkelson, W., Partch, M., Manager s trading around stock repurchases, Journal of Finance 47, Piotroski, Joseph D., and Darren Roulstone, 2005, Do insider trades react to both contrarian beliefs and superior knowledge about future cash flow realizations? Journal of Accounting and Economics 39, Richardson, S., Teoh, S., Wysocki, P., The walk-down to beatable analyst forecasts: The role of equity issuance and insider trading. Contemporary Accounting Research 21, Roychowdhury, S., Earnings management through real activities manipulation. Journal of Accounting and Economics 42, Rozeff, M., Zaman, M., Overreaction and insider trading: Evidence from growth and value portfolios. Journal of Finance 53, Sias, R., Whidbee, D Insider trades and demand by institutional and individual investors. Review of Financial Studies 23, Skinner, D., Why firms voluntarily disclose bad news. Journal of Accounting Research 32, Skinner, D., Earnings disclosures and stockholder lawsuits. Journal of Accounting and Economics 23, Stein, J., Efficient capital markets, inefficient firms: a model of myopic corporate behavior. Quarterly Journal of Economics 104, Zang, A., Evidence on the tradeoff between real activities manipulation and accrual-based earnings management. The Accounting Review 87,

31 Appendix: Variable Definitions Insider Trading Variables: The number of shares purchased by the CEO/CFO in her CEO/CFO Prescheduled Buy prescheduled trade during the 30 days following the earnings announcement date, scaled by shares outstanding. CEO/CFO Prescheduled Sale CEO/CFO Prescheduled Net Buy The number of shares sold by the CEO/CFO in her prescheduled trade during the 30 days following the earnings announcement date, scaled by shares outstanding. The net number of shares purchased by the CEO/CFO in her prescheduled trade during the 30 days following the earnings announcement date, scaled by shares outstanding. Earnings Management Variables: Discretionary Accruals Abnormal Discretionary Expenditure Over-production The abnormal level of discretionary accruals, measured as the estimated residual from the regression: 1, where Accruals t is the earnings before extraordinary items and discontinued operations minus the operating cash flows reported in the statement of cash flows in quarter t, A t-1 is total asset at the beginning of quarter t, S t is change in sales revenue in quarter t, and PPE t is the gross property, plant, and equipment. The above regression is estimated cross-sectionally for industry-quarters with at least 10 observations. The abnormal level of discretionary expenditure, measured as the estimated residual from the regression: 1, where DISX t is the sum of R&D and SG&A expenditures in quarter t, A t-1 is total asset at the beginning of quarter t, and S t-1 is sales revenue in quarter t-1. The above regression is estimated cross-sectionally for industry-quarters with at least 10 observations. The abnormal level of production cost, measured as the estimated residual from the regression: 1, where PROD t is the sum of cost of goods sold in quarter t and the change in inventory from t-1 to t, A t-1 is total asset at the beginning of quarter t, and S t is sales revenue in quarter t. The above regression is estimated cross-sectionally for industry-quarters with at least 10 observations. 30

32 Control Variables: Assets BM Sales Growth Leverage Cash Flows Operating Cycle Capital Intensity Firm Age Merger Restructuring External Finance Write-Off Internal Funds MV ROA Returns Future Returns Prior Forecasts Total assets of the previous quarter. Book value of equity/market value of equity, measured at the end of the previous quarter. Annual growth rate in sales revenue. Book value of debt/(book value of debt + book value of equity), where debt is the sum of current and long-term debts. Cash flow from operations divided by total assets. Operating cycle days (receivable collection period plus inventory turnover in days). Net property, plant, and equipment divided by total assets. The number of years since incorporation (first appearance on CRSP). An indicator variable that equals to 1 if a firm is involved in a merger or acquisition during the current quarter, and 0 otherwise (Compustat SALEQ_FN = ("AA", "AB", "AR", "AS")). An indicator variable that equals to 1 if a firm is involved in a restructuring in the current quarter, and 0 otherwise. The variable is coded 1 if any of the following Compustat data items are non-zero: RCA, RCD, RCEPS, RCP, RRA, RRD, RREPS, and RRP. An indicator variable that equals to 1 if the number of shares outstanding increased by at least 10% or long-term debt increased by at least 20% during the current quarter, and 0 otherwise. An indicator variable that equals to 1 if a firm reports a write-down during the current quarter, and 0 otherwise. Earnings before extraordinary items plus depreciation plus R&D expenditure, scaled by total assets. Market value of equity at the end of the previous quarter. EBITDA/lagged assets. Cumulative returns over the most recent fiscal quarter prior to the prescheduled trade. Cumulative returns over the fiscal quarter of the prescheduled trade. An indicator variable that equals to 1 if the firm has issued at least one management forecast during the [-120, -61] window relative to the prescheduled trade. 31

33 Figure 1: Timeline of Analyses Earnings manipulation period Earnings Announcement t Qtr t-1 Qtr t 60-day window to aggregate routine insider sales 32

34 Figure 2: Abnormal Returns Surrounding Routine vs. Opportunistic Insider Sales Routine Sales Opportunistic Sales (ann 5, ft 1) (lt+1, lt+5) (lt+1, lt+30) (lt+1, lt+60) (lt+1, lt+90) Note: This figure illustrates the market adjusted abnormal returns before and after routine vs. opportunistic insider sales. We present the average cumulative abnormal returns before and after insider sales in the following windows: (ann-5, ft-1), (lt+1, lt+5), (lt+1, lt+30), (lt+1, lt+60), and (lt+1, lt+90), where ann denotes the earnings announcement date, ft denotes the first day of each 60-day insider trading event, and lt denotes the last day of each 60-day insider trading event. 33

35 Figure 3: Histogram of Variability Ratio of Routine Insider Sales 34