Sample Selection and Data:

where the definitions of all variables are summarized in the Appendix.

3.3 Sample Selection and Data:

We obtain data from several sources. The original sample consists of S&P 1500 firm observations during fiscal years 2006~2010.⁵ Since we collect each firm’s financial data from COMPUSTAT, we exclude 613 observations with missing data on this database. We also exclude 620 observations due to data missing on GMI’s Clawback Dataset. About 1,024 financial institutions are excluded because these financial institutions were subjected to mandatory clawbacks enforced by the Department of Treasury under the Emergency Economic Stabilization Act of 2008. Audit committees’ compensation data are hand-collected from each firm’s Form DEF 14A using the EDGAR database.  We exclude 391 observations for which compensation information on board and audit committee members are not available. Data about CFOs and CEOs’ tenures, chairmanship, cash bonuses, and share ownerships are collected from ExecuComp. About 513 observations are deleted due to missing data on CEOs’ and CFOs’ profiles and compensation. We obtain sample firms’ managerial ability scores from Professor Peter Demerjian’s website (maintained by the Business School of Emory University). This        

5We choose 2006 as the starting year because only a few listed companies began to establish their clawback provisions since early 2005. We choose 2010 as the ending year because on July 29, 2011, the SEC has decided to postpone the mandatory implementation of Section 954 of the Dodd-Frank Act to 2012. Therefore, all extant clawback provisions during 2006~2010 should be voluntary in nature.

leads to a further reduction in our sample by 512 observations. Finally, we exclude 83 observations that fall in the top and bottom 1 percent of the empirical distribution for both the dependent and independent variables. Our final sample consists of 4,155 firm-year observations.

Panel A of Table 1 provides a breakdown of our sample by years and industries. As depicted in this panel, our sample does not vary substantially from 2006 to 2010. In addition, 34.87% of the companies belong to the Manufacturingdurables industry, followed by Manufacturing nondurables (19.52%), Wholesalers and Retailers (15.50%), and Business Services (12.73%).

[Insert Table 1 here]

Panel B of Table 1 shows the average amounts of the three components of audit committees’ compensation (i.e., cash, stock, and stock options). The average payments of both cash and stock increase gradually while the payment of options decreases. This may result from the passage of Statement of Financial Accounting Standards No. 123(R), which requires the expensing of option grants beginning from 2006. Looking at the compensation ratios, we find that the stock-to-total ratio increases 4.35% while the option-to-total ratio decreases 6.06% from 2008 to 2009, respectively. This result appears to imply that firms’ compensation structure has changed since the financial crisis of 2008. Most companies choose to pay higher portions of cash and stock to their audit committees.

Panel C of Table 1 presents the descriptive statistics of the test and control variables. The mean of FRAUD_RISK is 0.0229, suggesting that only 2.3% of our sample firms have substantial (cutoff F-score is 1.85) or high (cutoff F-score is 2.45) risk of occurring material accounting misstatements. This frequency parallels DeChow et al.’s (2011) finding that only 20% of the firms have an F-score greater than 1.4. In addition, about 22.62% of our sample firms have voluntarily adopted the clawbacks, consistent with Chan et al. (2012a). Based on our more stringent definition of managerial ability, the mean of MGR_Ability indicates that about 58.75% of the firms’ managers perform above average, compared with their two-digit SIC peers. On average, the audit committee has four directors and about 20.70% of the audit committee members have accounting expertise. The portion of equity-based compensation to total compensation awarded to audit committee members is 53.14%.

Notably, about 73.62% and 26.33% of the firms have issued debts or stocks during our sample period. About 54.78% of the CEOs also serve as the chairman.

Panel D of Table 1 compares firms with and without adopting clawbacks. In most cases, parametric and non-parametric tests results are consistent. Several findings are worth noting. First, firms’ voluntarily adopting clawbacks have lower fraud risk, suggesting that clawbacks appear to mitigate firms’ fraudulent financial reporting. Meanwhile, these firms have more capable managers.⁶ Although clawbacks may constrain managers’

accruals management, this result appears to imply that effective clawback provisions can protect the best interests of more capable managers. Since the voluntary adoption of clawback provisions may signal firms’

financial reporting quality, market participants perceive firms’ financial statements as being of higher quality.

Consequently, capable managers’ will benefit from firm-initiated clawback provisions by receiving more base pay. Second, firm with clawbacks are larger, relying less on leverage financing, having smaller sales and cash flow volatilities, more profitable, more likely to incur restructuring and foreign exchange adjustments, and having higher operation complexity. Third, the adoption of clawbacks is associated with lager audit committees whose members have less accounting expertise. Fourth, firms with and without clawbacks are equally likely to be audited by industry specialists and paying equal proportion of equity-based compensation. Finally, CEOs in firms with clawbacks are more likely to serve as the chairman, having smaller share of firms’ equity, and receiving less cash bonuses.

The Pearson and Spearman correlation coefficients between the dependent and independent variables are reported in Table 2. Since we use the Heckman model to control for endogenity due to firms’ voluntary adoption of clawbacks, we first report the correlation coefficients of variables used in the first-stage model in Panel A. In Panel B, we show the correlation coefficients of the dependent and independent variables used to examine our two research questions. We find that CLAWBACK and MGR_Ability are negatively related to FRAUD_RISK (significant at the 0.001 level), indicating that firms with clawbacks or hiring more capable managers have smaller risk of material misstatements due to fraud. Moreover, the correlations between FRAUD_RISK and EQUITY% are positive and significant. Overall, these results suggest that, while clawbacks and managerial ability have a positive effect in reducing fraud risk, equity-based compensation paid to the audit committees has an adverse impact on credible financial reporting.

6Using raw MA scores, we find that the means of firms with and without adopting clawbacks are 0.6390 and 0.5726, respectively.

[Insert Table 2 here]