Internet Appendix for. Industry Expertise of Independent Directors and Board Monitoring

Transcription

1 Internet Appendix for Industry Expertise of Independent Directors and Board Monitoring Cong Wang Fei Xie Min Zhu Appendix A. Definitions of Earnings Management Measures I. Abnormal Accruals We follow Dechow, Sloan, and Sweeney (1995) and use the modified Jones (1991) model to estimate abnormal accruals (AA). We first estimate the following cross-sectional regression by year for each industry. Assets i,t-1 is total assets for firm i (#6) at the beginning of fiscal year t. Total accruals for firm i in year t, TA i,t, is calculated as income before extraordinary items (#123) minus cash flows from operating activities adjusted for extraordinary items and discontinued operations (#308-#124). ΔSale i,t is change in sales (#12) and PPE i,t is gross value of property, plant, and equipment (#7). The estimated coefficients from the above regression are then used to estimate abnormal accruals. We include change in accounts receivable (#2), ΔAR i,t, to capture the extent to which a change in sales is due to aggressive recognition of questionable credit sales. Abnormal accruals, AA i,t, is calculated as: ( ) We follow previous studies (Kothari, Leone, and Wasley (2005) and Gong, Louis, and Sun (2008)) to adjust abnormal accruals for firm performance. More specially, for each 2-digit

2 SIC industry in each year, we sort all firms into quintile portfolios by their returns on assets in the previous year. The performance-matched abnormal accruals for a sample firm is calculated as its abnormal accruals minus the median abnormal accruals of its industry performance matched portfolio. II. F-score We follow Dechow, Ge, Larson and Sloan (2011) and first calculate the predicted value from the following regression: p i,t = (-7.893) RSST_ACC i,t ΔREC i,t Δ INV i,t SOFT i,t ΔCSale i,t + (-0.932) Δ ROA i,t ISSUE i,t RSST_ACC i,t is Richardson, Sloan, Soliman, and Tuna (2005) measure of accruals for firm i in fiscal year t, which is the sum of change in non-cash working capital WC((#4-#1)-(#5- #34)), change in net non-current operating assets NCO((#6-#4-#32)-(#181-#5-#9)), and change in net financial assets FIN((#193+#32)-(#9+#34+#130)), scaled by average total assets. ΔREC i,t is the change in receivables (#2) for firm i in fiscal year t scaled by average total assets, ΔINV i,t is the change in inventory (#3) for firm i in fiscal year t scaled by average total assets, SOFT i,t is the percentage of soft assets, which is equal to total assets minus property, plant, and equipment and cash (#6-#8-#1) for firm i in fiscal year t scaled by total assets (#6), ΔCSALE i,t is the percentage change in cash sales (#12-Δ#2) for firm i in fiscal year t, ΔROA i,t is the change in return on assets (#18/averaged#6) for firm i in fiscal year t, and ISSUE i,t is an indicator variable that equals 1 if the firm has issued new debt or equity (#108>0 or #111>0) during fiscal year t and 0 otherwise. The probability of misstatement is then calculated as and F-score is equal to. 1

3 III. Earnings Restatements Our sample of earnings restatements is constructed as follows. We begin with the two reports published by the U.S. General Accounting Office (GAO) in 2003 and 2007, which contain a total of 2,705 earnings restatements announced between January 1, 1997 and June 30, We then expand the GAO sample to cover the period from July 1, 2006 to December 31, 2010 as well. 1 In doing so, we follow GAO s methodology and manually retrieve all news articles from Lexis-Nexis that contain variants of restate, adjust, amend, or revise within 50 words of financial statement or earning. We read all these articles and identify additional 721 earnings restatements during this later time period. Therefore, over the period from 1997 to 2010, we have a combined sample of 3,426 earnings restatements, which include both restatements due to accounting irregularities (intentional misreporting) and those due to accounting errors (unintentional misstatements). Hennes, Leone, and Miller (2008) develop a methodology that classifies a restatement as an irregularity if it satisfies at least one of the three criteria: (i) variants of the words irregularity or fraud were explicitly used in restatement announcements or relevant filings in the four years around the restatement; (ii) the misstatements came under SEC or DOJ investigations; and (iii) independent investigations were launched by boards of directors of restatement firms. In a sample of restatements between 2002 and 2005, they demonstrate the importance and effectiveness of their classification scheme by showing that compared to error restatements, irregularity restatements are met with significantly more negative announcement returns (on average: -14% vs. -2%), are followed by shareholder class action lawsuits at a significantly 1 Even though our sample period ends in 2007, we collect earnings restatement announcements for 3 more years after 2007 since earnings manipulations that took place during our sample period may take a few years to be uncovered and disclosed. 2

4 higher rate, and lead to significantly more CEO/CFO turnovers. We obtain from Andrew Leone s website the irregularity vs. error classification for the GAO sample of restatements. For the additional 721 restatements we collected ourselves, we classify them into irregularity or error driven following Hennes, Leone, and Miller s methodology. We then read all irregularity restatement news releases to identify the years in which earnings manipulation happened. After merging the sample of irregularity-restatement firms and our original sample of firms with independent director industry expertise information from BoardEx, we find 231 firm years in which firms engaged in intentional earnings manipulation that resulted in irregularity restatements later. Appendix B. Correlation Matrix Table B1 presents the correlations between the variables used in our analysis. Among other things, we find that IEDs are significantly more likely to appear on the audit committee at larger and younger firms that have lower leverage, better operating performance measured by the return on assets (ROA), and more growth opportunities as proxied by Tobin s Q. We also observe a significantly higher presence of IEDs on the audit committee when boards are smaller, more independent, and separate the position of the CEO and the chairman of the board. Additionally, the presence of IEDs on the audit committee is higher when audit committees are smaller, more independent, have more outside directorships, and have more financial experts. With respect to the earnings management measures, we find that the magnitude of abnormal accruals is higher in younger firms, firms with higher Tobin s Q, and firms with higher volatility in sales. Smaller boards, smaller and more independent audit committees, and audit committee with more financial expertise appear to be associated with lower abnormal accruals. IED 3

5 presence on the audit committee is negatively correlated with abnormal accruals, but the relation is not statistically significant at the conventional level. However, audit committee industry expertise is significantly and negatively correlated with F-score. In the ensuing multivariate regressions, we control for firm characteristics that may relate to both earnings management and the presence of independent directors with industry expertise. [Insert Table B1 here] Appendix C. Earnings Management Prior to Seasoned Equity Offerings and Stock Repurchases Several prior studies show that firms manage their earnings in certain predictable directions prior to important corporate events. For example, firms tend to manage their earnings upward through positive accruals prior to seasoned equity offerings (SEOs) in order to boost stock price so that they can raise equity financing at more favorable terms (Teoh, Welch, and Wong (1998)), while managing their earnings downward through negative accruals prior to open market repurchases to depress stock price so that they can buy back their shares at lower costs (Gong, Louis, and Sun (2008)). Therefore, we extend our analysis to examine whether the presence of IEDs on the audit committee can reduce the magnitude of positive abnormal accruals prior to SEOs and the magnitude of negative abnormal accruals prior to open market repurchases. We follow Gong, Louis, and Sun (2008) and use Compustat Quarterly to compute performance adjusted abnormal accruals in each quarter and use it as our dependent variable. We then extract SEOs and stock repurchases from the Securities Data Corporation (SDC) Database. We create an indicator variable, SEO, which is equal to 1 for firm-quarters immediately prior to a SEO and zero otherwise. A second indicator variable, Repurchase, is set equal to 1 for firm-quarters 4

6 immediately prior to an open market stock repurchase and zero otherwise. Our sample firms conduct 77 seasoned equity offerings and 740 open market stock repurchases over 11,222 firmquarters during the sample period. To test our hypothesis, we partition our sample into two subsamples based on whether a firm s AC IED Percentage is above or below the sample median. We estimate separate subsample regressions of quarterly abnormal accruals and report the results in Table C1. The key independent variable is the SEO indicator in columns (1) and (2), with column (1) for the subsample with above-sample-median AC IED Percentage and column (2) for the subsample with below-sample-median AC IED Percentage. We find that the coefficient estimate of the SEO indicator is significantly positive only in the subsample of firms with a lower percentage of IEDs on the audit committee, and while there is no evidence of significantly positive abnormal accruals prior to SEOs at firms with a greater presence of IEDs on their audit committees. These findings suggest that independent director industry expertise on the audit committee reduces firms earnings management behavior prior to SEOs. [Insert Table C1 here] 5

7 Table B1. Correlation Matrix The sample consists of 2,744 firm-year observations from 2000 to In parentheses below the correlations are their corresponding p-values. ABSDA (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (1) F-score (0.156) (2) Irregularity (0.002) (0.042) (3) AC IED Percentage (0.210) (0.006) (0.240) (4) Assets (0.263) (0.000) (0.000) (0.051) (5) Age (0.000) (0.000) (0.000) (0.000) (0.000) (6) Leverage (0.135) (0.017) (0.002) (0.000) (0.000) (0.000) (7) Tobin Q (0.000) (0.002) (0.017) (0.000) (0.000) (0.000) (0.000) (8) ROA (0.537) (0.348) (0.002) (0.003) (0.774) (0.000) (0.000) (0.000) (9) Std Sales (0.000) (0.128) (0.188) (0.273) (0.000) (0.000) (0.000) (0.803) (0.000) (10) Board Size (0.000) (0.056) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.772) (0.000) (11) Board Independence (0.376) (0.264) (0.000) (0.000) (0.626) (0.000) (0.815) (0.032) (0.000) (0.004) (0.000) (12) CEO/Chairman (0.173) (0.160) (0.051) (0.000) (0.003) (0.000) (0.000) (0.000) (0.008) (0.132) (0.004) (0.000) (13) Board Busyness (0.488) (0.544) (0.015) (0.634) (0.000) (0.000) (0.000) (0.014) (0.618) (0.424) (0.000) (0.942) (0.000) (14) AC Size (0.005) (0.283) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.779) (0.234) (0.000) (0.000) (0.000) (0.000) (15) AC Independence (0.061) (0.185) (0.000) (0.000) (0.822) (0.003) (0.001) (0.385) (0.000) (0.000) (0.000) (0.000) (0.275) (0.000) (0.119) (16) AC Busyness (0.797) (0.699) (0.002) (0.000) (0.000) (0.000) (0.001) (0.204) (0.323) (0.150) (0.000) (0.000) (0.004) (0.000) (0.000) (0.000) (17) AC FIN Expert Percentage (0.140) (0.314) (0.163) (0.000) (0.986) (0.017) (0.147) (0.502) (0.866) (0.334) (0.000) (0.000) (0.007) (0.000) (0.000) (0.000) (0.738) (18) Auditor Market Share (0.087) (0.965) (0.302) (0.007) (0.000) (0.002) (0.000) (0.000) (0.002) (0.771) (0.000) (0.012) (0.269) (0.032) (0.000) (0.227) (0.003) (0.715) (19) CEO Incentive (0.087) (0.518) (0.000) (0.000) (0.808) (0.000) (0.000) (0.000) (0.000) (0.171) (0.000) (0.000) (0.104) (0.000) (0.000) (0.300) (0.817) (0.125) (0.186) (20) Repurchase Incentive (0.945) (0.011) (0.147) (0.069) (0.952) (0.096) (0.000) (0.000) (0.001) (0.775) (0.296) (0.007) (0.478) (0.476) (0.227) (0.175) (0.578) (0.255) (0.263) (0.791) (21) SEO Incentive (0.648) (0.607) (0.553) (0.344) (0.003) (0.158) (0.000) (0.000) (0.061) (0.762) (0.148) (0.001) (0.512) (0.627) (0.472) (0.153) (0.857) (0.521) (0.855) (0.005) (0.000) 6

8 Table C1. Accrual Earnings Management Prior to Seasoned Equity Offerings and Open Market Repurchases The sample consists of 11,222 firm-quarter observations from 2000 to The dependent variable is the level of quarterly abnormal accruals. We partition the sample into subsamples based on a firm s AC IED Percentage. In parentheses are t-statistics based on standard errors adjusted for heteroskedasticity (White (1980)) and firm clustering (Petersen (2009)). Superscripts ***, **, and * stand for statistical significance based on two-sided tests at the 1%, 5%, and 10% level, respectively. Variables (1) (2) (3) (4) Low AC IED High AC IED Percentage Percentage High AC IED Percentage Low AC IED Percentage SEO * (0.581) (1.809) Repurchase * ** (-1.913) (-2.265) Size (-1.489) (-1.431) (-1.462) (-1.364) Age *** *** *** *** (-6.697) (-7.399) (-6.699) (-7.398) Leverage *** *** (1.040) (3.958) (1.014) (3.900) Tobin Q * * (-1.779) (-0.097) (-1.775) (-0.066) ROA 0.289*** 0.200*** 0.290*** 0.199*** (6.493) (2.684) (6.548) (2.665) Std Sales (0.053) (-0.037) (0.044) (-0.053) Board Size (-1.591) (-1.088) (-1.547) (-1.090) Board Independence (0.090) (-0.319) (0.093) (-0.305) CEO/Chairman * * (1.137) (-1.802) (1.133) (-1.783) Board Busyness (-1.062) (-1.361) (-1.063) (-1.318) AC Size * ** (-1.959) (0.994) (-1.998) (0.958) AC Independence (-0.546) (-0.418) (-0.538) (-0.429) AC Busyness (-0.259) (0.250) (-0.241) (0.222) AC FIN Expert Percentage ** ** (-2.324) (0.840) (-2.289) (0.800) AC IED Percentage (0.654) (0.089) (0.648) (0.078) 7

9 Auditor Market Share (-0.471) (-0.289) (-0.472) (-0.312) Constant 1.431*** 1.518*** 1.424*** 1.514*** (7.168) (7.734) (7.162) (7.717) Firm Fixed Effects Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Observations 5,969 5,253 5,969 5,253 R-squared