Revisiting CEO power and firm value

Abstract

Prior literature associates CEO power with agency problems and documents the negative relationship between CEO power and firm value (e.g., Bebchuk et al., 2011). However, the ‘optimal’ level of CEO power may differ for every firm and for individual CEO depending on firm and CEO characteristics. In this study, we estimate the normal (‘optimal’) level of CEO power and show that the association between CEO power and firm value is nonmonotonic. Our results reveal that the normal level of CEO power is positively associated with firm value while excess and deficient CEO power is negatively associated with firm value. Using the residuals from our estimation model of CEO power, we illustrate that our measure of residual CEO power has an inverse U-shaped relationship with firm value.

Keywords:

• CEO power
• CEO Pay Slice (CPS)
• firm value
• Tobin’s q

JEL Classification:

• G30
• G34

Acknowledgements

We wish to thank Mark P. Taylor (Editor), David A. Peel (Co-Editor) and the anonymous referee for their helpful comments. We would also like to thank Cherree Shin for her research assistance.

Notes

¹ In Fig. 1, we illustrate that CEO power (lagged CPS) and firm value (industry-adjusted Tobin’s q) may not be simply linear. The observed distribution provides a graphical evidence that the relationship between CEO power and firm value is not clearly negative but rather a somewhat of an inverse U-shape.

² In a similar vein, Finkelstein and D’Aveni (1994) argue that powerful CEOs, by establishing unity of command among top executives, are able to clarify decision-making authority.

³ Weisbach (1988) finds that powerful CEOs are less likely to be replaced following poor performance. Feng et al. (2011) show that the CEOs with excessive power suppress CFOs, thus engage in more accounting manipulations. Two exceptions are Jiraporn and Chintrakarn (2013) who report nonmonotonic relationship between CEO power and CSR investments and Chintrakarn et al. (2014) who document a similar relationship between CEO power and capital structure decisions.

⁴ We utilize Compustat’s Execucomp database item (TDC1) to measure total compensation given to each executive including the CEO. We exclude firm-year observations with compensation data fewer than five executives. However, our overall results do not vary when we include such observations following the method used by Feng et al. (2011).

⁵ CEO compensation and CPS are both settled by corporate boards, specifically compensation committees, and are determined contemporaneously. Accordingly, we show that the coefficient estimates for the independent variables are similar for both dependent variables.

⁶ Bureau of Labor Statistics provides the CPI (Consumer Price Index) Inflation Calculator (http://www.bls.gov/data/inflation_calculator.htm). By setting the year 1993 (beginning year of the sample period) as a base year, we calculate the adjustment factors for the entire sample period. The adjustment factors are then multiplied to explanatory variables in each year. We thank the anonymous reviewer for the helpful comments related to this issue.

⁷ Due to the fact that our model includes industry dummy variables, we do not find any significant association between industry-median CPS and CPS. The overall results do not vary if we exclude the industry-median CPS from the model.

⁸ When we replace residual CPS in Column (2) of Table 3 with absolute value of residual CPS, we find significantly negative coefficient (−0.3687 with t-statistics of −1.79) which is consistent with our main findings.

Additional information

Funding

The authors are grateful for financial support from Georgia Regents University and KAIST.