Founding Family Firms, CEO Incentive Pay, and Dual Agency Problems

Abstract

This paper contributes to the literature on agency theory by examining relations between family involvement and CEO compensation. Using a panel of 362 small U.S. listed firms, we analyze how founding families influence firm performance through option portfolio price sensitivity. Consistent with the dual agency framework, we find that family firms have lower CEO incentive pay, which is further reduced by higher executive ownership. Interestingly, such incentive pay offsets the positive impact that families have on firm valuation. Collectively, our results show that, compared with nonfamily firms, lower incentive pay adopted by family firms due to lower agency costs mitigates the direct effect of family involvement on firm performance. Once accounting for CEO incentive pay, we do not observe performance differences between family and nonfamily firms.

Notes

1 According to OECD (2009), small and medium‐sized enterprises (SMEs) account for more than 99 percent of all enterprises in the European Union and more than half of the labor force in the private sector in the OECD area.

2 We utilize several online sources, such as http://www.fundinguniverse.com/.

3 In some cases, we cannot obtain founder information. It is also likely that we lose track of founding family members. Thus, we underestimate the true proportion of family firms in the sample. This would potentially work against our testing hypotheses.

4 For instance, see Brockman, Martin, and Unlu (2010) for a detailed description of the computation of the option delta.

5 As a robustness check, we use equity‐based incentive pay as one alternative measure of incentive pay.

6 In both the high and low ownership groups, the level of incentive pay is lower among passive family firms than among both active family and nonfamily firms. These results are not tabulated but are available on request.

7 Our two‐stage‐least‐squares regression estimates for these control variables are similar to the SEM estimates, which are available on request.

8 Two separate logit regressions and one multinomial logit regression all yield the same fitted values.

9 We use xtabond2 in Stata to generate the system GMM estimators and the test results.

Additional information

Notes on contributors

Mieszko Mazur

Mieszko Mazur is an assistant professor of Finance at the IESEG School of Management.

Betty H.t. Wu

Betty H.T. Wu is a lecturer in Accounting and Finance at the University of Glasgow Adam Smith Business School.