EMPIRICAL RESULTS

4.1 Association between Audit Committees’ Equity-based Compensation and Fraud Risk

Table 3 reports that the coefficient of EQUITY% (0.881, ² = 14.85, p < 0.01) is significant with a positive sign. This result suggests that larger portions of stocks and options paid to the audit committees are detrimental to audit committees’ oversight effectiveness, leading to higher risk of incurring material accounting misstatements due to fraud. Unreported marginal effect shows that an increase in one standard deviation in EQUITY% would increase the fraud risk by 1.31%. This evidence complements and extends Archambeault et al.’s (2008) finding that the proportions of options are associated with restatement likelihood. The pseudo R² is 0.073 and the likelihood ratio ² is significant at the 0.01 level, indicating that model (1) has explanatory power in explaining variations in FRAUD_RISK.

[Insert Table 3 here]

Two things are worth noting. First, while the percentage of outsiders or non-affiliated directors sitting on the audit committees appears to be a more straightforward proxy for audit committee independence and has been used by early studies (e.g., Bedard et al. 2004; Carcello and Neal 2000, 2003; Klein 2002a; Xie et al. 2003), this pre-SOX measure is noisy because outsiders or non-affiliated directors may still own firms’ equity that tie their wealth with firms’ financial performance. Due to this measurement issue, earlier studies’ findings lack empirical consistency (see Section 2.1) and cannot provide evidence as to whether audit committees’ compensation practice should be regulated. In contract, since Section 301 ensures that audit committee members cannot hold any shares when they are newly appointed, they own firms’ stocks and options entirely from their equity-based compensation. We take advantage of this setting to examine whether audit committees remain independent in overseeing managers’ earnings management when they are compensated by stocks and options. Because the F-score can identify within-GAAP earnings management as well as the more aggressive techniques identified by the SEC (DeChow et al. 2011), we thus contribute to the audit committee and earnings management literature by presenting the negative association between equity-based compensation and earnings quality.

Second, recent research indicates that companies have switched their earnings management methods from

pre-SOX accruals to post-SOX real earnings management (e.g., Cohen et al. 2008). Since SOX has stipulated numerous provisions to help audit committees fulfill their oversight duties, effective audit committees should be able to detect this switch in earnings management methods and revise their oversight policies and procedures accordingly. Given this expectation, the significantly positive coefficient of EQUITY% implies that audit committees appear to remain incapable of detecting or mitigating real earnings management activities, possibly due to the high portions of equity-based compensation they receive. This may partially explain Cohen et al.’s (2008) finding that the level of post-SOX earnings management has returned to the pre-SOX level.

4.2 Effects of Clawbacks and managerial ability on the Association between Audit Committees’

Equity-based Compensation and Fraud Risk

We now test whether the adoption of clawbacks can mitigate the adverse effect of equity-based compensation on audit committee’s oversight effectiveness. Panel A of Table 4 reports the first-stage regression results. Because the pseudo R² is 0.167 and the likelihood ratio ² is significant at the 0.01 level, model (2-1) provides a fair measurement of the inverse Mill’s ratios to be used in the second-stage model.

[Insert Table 4 here]

Panel B indicates that the coefficient of CLAWBACK is negative but not significant, suggesting that the adoption of clawbacks economically reduces firms’ risk of incurring fraudulent financial reporting. Importantly, while the coefficient of EQUITY% remains positive and significant (0.510, ² = 12.31, p < 0.01), the coefficient

of EQUITY%CLAWBACK reverses to significantly negative (-0.528, ² = 4.14, p < 0.05). Unreported marginal effects indicate that, when CLAWBACK increases from zero to one, the fraud risk would decrease by 1.96%. In contrast, given a firm has adopted clawbacks, an increase in one standard deviation in EQUITY% will increase the fraud risk by 0.61%. In other words, the existence of clawbacks reduces a firm’s total fraud risk by 1.35%.

These results provide evidence that the adoption of clawback effectively mitigates the adverse effect of equity-based compensation on audit committees’ oversight effectiveness, leading to lower fraud risk.

We next examine whether managerial ability can also enhance audit committees’ oversight effectiveness.

Different from the results for clawbacks, Table 5 shows that the coefficient of EQUITY% remains significantly positive (0.874, ² = 15.86, p < 0.01). While the coefficient of MGR_Ability is insignificantly negative, the

coefficient of EQUITY%MGR_Ability is negative and significant (-0.578, ² = 4.85, p < 0.05). This result indicates that greater managerial ability can effectively mitigate the adverse effect of equity-based compensation on audit committees’ oversight effectiveness. Unreported marginal effects show that, when MGR_Ability increases from zero to one, the fraud risk decreases by 1.85%. Given a firm has employed more capable managers, an increase in one standard deviation in EQUITY% will increase the fraud risk by 0.62%. Therefore, hiring more capable managers reduces a firm’s total fraud risk by 1.23%. These results imply that higher managerial ability by itself can reduce firms’ fraud risk, consistent with Demerjian et al. (2013).

[Insert Table 5 here]

To compare the relative efficacy of clawbacks and managerial ability in motivating audit committee members to exert their effort in monitoring firms’ financial reporting, we include both CLAWBACK and MGR_Ability and their interactions with EQUITY% into the regression model. Table 6 documents two important results. First, while the coefficients of CLAWBACK and MGR_Ability are not significant, the apparently larger coefficient of CLAWBACK (-0.184) is about nine times of the coefficient of MGR_Ability (-0.021). Unreported marginal effects show that, when CLAWBACK and MGR_Ability each increases from zero to one, the fraud risk will decrease by 2.08% and 1.93%, respectively. In other word, the marginal effect of managerial ability is smaller than the marginal effect of clawbacks in reducing firms’ fraud risk. This finding implies that, without considering the adverse effect of equity-based compensation on audit committees, employing more capable managers appears to be less effective in preventing fraudulent financial reporting than adopting clawbacks.

[Insert Table 6 here]

Second, the coefficient of CLAWBACKEQUITY% is negative and significant (-0.644, ² = 4.26, p < 0.05).

This result indicates that the negative effect of equity-based compensation on audit committees’ oversight effectiveness is mitigated by firms’ adoption of clawbacks. Unreported marginal effects show that, given a firm has adopted clawbacks, an increase in one standard deviation in EQUITY% will increase the fraud risk by 1.34%.

Therefore, the adoption of clawbacks reduces a firm’s total fraud risk by 0.74% (2.08% - 1.34%). In contrast, the coefficient of MGR_AbilityEQUITY% is also negative and significant (-0.602, ² = 4.89, p < 0.05). Unreported marginal effects indicate that, given a firm has employed more capable managers, an increase in one standard

deviation in EQUITY% will increase the fraud risk by 1.16%. Therefore, hiring more capable managers reduces a firm’s total fraud risk by only 0.77% (1.93% - 1.16%). These results do not imply that the negative effect of equity-based compensation on audit committees’ oversight effectiveness outweighs the positive effect of hiring more capable managers on improved earnings quality. Rather, it should be interpreted as firms’ use of stocks and options to compensate their audit committees does not increase firms’ fraud risk when these firms employ capable managers. The reason is simple: Since more capable managers have less incentive to manage earnings, firms’ risk of fraudulent financial reporting decreases. In this situation, audit committees’ reduced oversight effectiveness resulting from the equity-based compensation may play a trivial role in raising the fraud risk. Note that our interpretation does not suggest that enhancing audit committees’ oversight effective is not important.

From corporate governance’s perspective, audit committees’ major responsibilities not only include the monitoring of firms’ financial reporting, but also include the choice of auditors, the determination of appropriate audit fee levels, and the communications with the auditors on major auditing matters.