Not Clawing the Hand that Feeds You: The Case of Co-opted Boards and Clawbacks

Abstract

We examine how board co-option, defined as the fraction of the board comprising directors appointed after the CEO assumed office, is related to clawback adoption. We find that co-opted boards have a lower probability of adopting clawback provisions. Further, the negative association between board co-option and clawback adoption is more pronounced when at least one co-opted member is on the compensation committee and when there is a higher likelihood that a clawback provision will be triggered. Finally, we find that board co-option is an important mechanism through which longer tenured CEOs reduce the likelihood of clawback adoption.

Acknowledgements

We thank K. R. Subramanyam and Shiva Rajgopal, participants at the European Accounting Association Annual Congress and the American Accounting Association Annual Congress and workshop participants at Singapore Management University for their comments.

ORCID

Jeffrey Ng http://orcid.org/0000-0003-1430-3646

Notes

1 While Coles et al. (2014) provide no direct evidence of whether an increase in investment due to board co-option enhances or destroys shareholder value, based on related work by Pan, Wang, and Weisbach (2013) they suggest that such increased investment reflects overinvestment that harms shareholders.

2 Even if there is a very low likelihood of enforcing clawbacks in the future, the adoption of a mechanism that has a non-zero probability of being detrimental to the future welfare of one’s benefactor/friend could be perceived as ‘unfriendly.’

3 It is important to control for these characteristics to demonstrate the incremental effect of board co-option because studies have documented that such characteristics are associated with clawbacks (e.g. Addy et al., 2014; Babenko et al., 2015; Brown, Davis-Friday, & Guler, 2011; Chan et al., 2015; Dehaan et al., 2013). For example, Addy et al. (2014) show that greater management entrenchment is associated with a lower likelihood of clawbacks, while board interlocks with other companies with clawbacks are associated with a higher likelihood.

4 During our sample period, no firms drop their clawback provisions.

5 In addition to providing a richer analysis of the relation between board co-option and clawbacks, these cross-sectional analyses help to mitigate endogeneity concerns.

6 For example, Fracassi and Tate (2012) show that CEO–director connections weaken board monitoring and destroy corporate value. Hwang and Kim (2009) find that firms with board members who are personally connected to the CEO have higher CEO compensation, lower pay-performance sensitivity, and lower turnover-performance sensitivity.

7 Research in political science and finance also suggests that one reason policies are not implemented is because the policy makers do not want to ‘bite the hand that feeds them.’ For example, voting on antismoking legislation is associated with tobacco industry lobbying and campaign contributions (Givel & Glantz, 2001; Glantz & Begay, 1994; Monardi & Glantz, 1998). Ovtchinnikov and Pantaleoni (2012) report that Microsoft’s political contributions increased significantly during the firm’s antitrust litigation with the Department of Justice, with the latter finally announcing that it would not order the breakup of the company.

8 The audit literature has examined how auditors who are more reliant on their clients, for instance regarding non-audit fees, are less stringent with regard to allowing financial manipulation and misreporting in client audits (e.g. Klein, 2002; Abbott, Parker, & Peters, 2004). One might argue that this approach is similar in spirit to having an implicit policy of not ‘biting the hand that feeds you.’

9 However, the evidence is far from conclusive. Agarwal, Jaffe, and Karpoff (1999) find that director turnover is unchanged after fraud, and Fich and Shivdasani (2007) find that directors do not leave a sued firm beyond normal levels. Other studies also conclude that the related financial loss borne by outside directors is generally small, if any (Black, Cheffins, & Klausner, 2006; Srinivasan, 2005).

10 To the extent that the results of these analyses are consistent with expectations regarding the factors that could moderate the effect of board co-option on clawback adoption, the likelihood is reduced that an (uncontrolled) omitted factor correlated with board co-option drives the association between board co-option and clawback adoption.

11 The compensation committee would propose a clawback. For example, Compensation Advisory Partners (2015, pp. 153–154) indicates the key questions that compensation committee members should discuss when considering a clawback provision. This provides some evidence that compensation committees are indeed heavily involved in the decision to adopt clawbacks.

12 Formed in 2010 through the merger of the Corporate Library, Governance Metrics International and Audit Integrity, GMI Ratings provides global research coverage of the environmental, social, governance, and accounting-related risks that affect the performance of public companies. Prior studies using these data (e.g. Dehaan et al., 2013; Chan et al., 2015) cite the Corporate Library as the data source.

13 The marginal effect of a one-standard-deviation (SD) increase in the board co-option measure is computed as p × (1 − p) × b × SD, where p is the base rate (44.2% for Co-option and 21.0% for TW Co-option), and b is the estimated coefficient from the logistic regression (Liao, 1994).

14 As a robustness check, we also use the percentage of co-opted directors on the compensation committee to measure co-option to the committee. We find similar results using this alternative proxy.

15 Ai and Norton (2003) question the properties of the estimators of the coefficient on the interaction term in a logistic model such as Equation (2), as well as their related test statistics. However, Greene (2010) concludes that an overall statistical inference cannot be obtained from the Ai and Norton (2003) measure. Furthermore, Kolasinski and Siegel (2010) argue that it is appropriate to draw inferences from the interaction term in nonlinear models. Therefore, we use the interaction coefficient to assess the directional effect of our results. As a further robustness check, we calculate the modified statistical output, as Ai and Norton (2003) suggest and Evans, Nagarajan, and Schloetzer (2010) use, by using the ‘inteff’ procedure in STATA. We find that the inferences based on this test statistic are similar to those reported in Tables 4 and 5.

16 Another possible channel is through reduced need for monitoring. Dikolli, Mayew, and Nanda (2014) provide evidence that longer tenured CEOs have less uncertainty about their abilities; thus, the board needs to monitor them less. Accordingly, there might be less of a need to adopt clawbacks to monitor CEOs who have served in their position longer.

17 Specifically, a path analysis is used to answer how one variable (CEO tenure in our case) affects another (i.e. clawbacks). We argue that the impact occurs through a mediating variable, board co-option. In contrast, an interaction analysis is used to answer when a moderating variable affects the association between two variables. Baron and Kenny (1986) provide a discussion on mediation versus moderation.

18 To carry out the path analysis, we estimate Equation (4b) using a linear probability model. Standardized coefficients are reported in Panels B and C of Table 8.