Print
E-MailAna Albuquerque is Associate Professor of Accounting at Boston University Questrom School of Business. This post is based on a recent paper authored by Professor Albuquerque; Rui Albuquerque, Associate Professor of Finance at the Boston College Carroll School of Management; Mary Ellen Carter, Associate Professor at the Boston College Carroll School of Management; and Flora Dong, Assistant Professor at Pennsylvania State University. Related research from the Program on Corporate Governance includes The CEO Pay Slice by Lucian Bebchuk, Martijn Cremers and Urs Peyer (discussed on the Forum here) and Paying for Long-Term Performance by Lucian Bebchuk and Jesse Fried (discussed on the Forum here).
In a recent paper, my co-authors and I provide empirical evidence that CEO compensation does not fully reflect riskiness in pay packages. Our evidence derives from an examination of the fundamental prediction in the static moral hazard model of Grossman and Hart (1983) that the mean pay and the volatility of pay are positively associated through the participation constraint. Firms providing incentive pay as a way to reduce principal-agent conflicts need to compensate the CEO for the additional risk borne by the CEO through incentive pay. This paper is the first to provide a test of this fundamental hypothesis using U.S. data, providing us an advantage over other cross-country studies as we can hold constant the institutional characteristics of the contracting environment.
We use three different empirical approaches to provide evidence on the relation between pay volatility and pay levels. The first empirical approach uses Incentive Lab’s detailed contract information on the relation between performance metrics and performance-based compensation in actual CEO contracts. The main advantage of these rich data is that we can evaluate simultaneously, through a simulation exercise, the beginning-of-the-year expected value and variance of the CEO’s end-of-year pay. We are therefore able to explicitly consider the underlying sources of the riskiness of CEO pay packages. The second empirical approach is to model the conditional variance of pay with realized variance of past CEO pay in the spirit of Schwert (1989) and Andersen and Bollerslev (1998). The third empirical approach uses a variant of Engle’s (1982) autoregressive conditional heteroskedasticity (ARCH) model to jointly estimate the mean of pay and the volatility of pay. The ARCH model estimates an equation for the level of pay and another equation for the volatility of pay. This simultaneous determination of mean and variance of pay is consistent with the theory and constitutes an advantage over the realized volatility approach.
While our results support the hypothesis that CEOs require higher pay for higher risk in their pay, the magnitude is low. We find remarkably consistent point estimates around 0.05 for the elasticity of average pay to variance of pay across all three approaches. To understand the economic significance, we conduct a back-of-the envelope calculation. At 0.05 elasticity, the implied portion of compensation that should be incentive pay is 10%. However, in our sample of Incentive Lab firms, actual incentive pay is much larger at 45% of total pay. The discrepancy suggests that CEOs are not fully compensated with fixed pay for the growth in incentive pay and its associated volatility in pay, hence the low estimated elasticity. Overall, CEOs appear to be saturated with incentives, without commensurate compensation for risk. We also find that CEOs are willing to earn less on average when faced with pay packages that have positive skewness, e.g., packages with stock and option grants that offer the opportunity to avoid downside risk. In addition, we find that founder (vs non-founder) CEOs are compensated less for risk and have lower preference for positive skewness in pay.
Our evidence supports the conjecture in Murphy and Jensen (2018) that the growth in incentive pay in the last two decades seems not to be driven by the provision of economic incentives and the risk and reward tradeoff. One possible explanation for the trend in pay may be tax rules, specifically IRS code Section 162(m), which formerly favored incentive pay, thereby restricting the growth in pay levels to occur primarily through incentive pay. The expansion of the deductibility of compensation under the Tax Cut and Jobs Act of 2017 may change this trend. Another possible explanation is proxy advisors’ push for greater incentive pay.
The complete paper is available here.
