Research at Drexel University’s LeBow College of Business dispels the widely-accepted view that executive compensation rewards chief executive officers for good luck, but does not penalize them for bad luck to the same extent.
- The increases in executive compensation attributable to good luck are similar in size to the decreases in executive compensation attributable to bad luck. “Luck” is the component of firm performance that is outside the control of the chief executive officer.
- This finding challenges the view that executive pay reacts more positively to good luck than negatively to
bad luck. Evidence of an asymmetric reaction has previously been used to conclude that executive compensation is inefficient.
- In order to improve our understanding of how CEOs should be paid, it is important to test even the most
intuitive concepts against a wide range of assumptions. This is a critical consideration for stakeholders and lawmakers before they propose or enact changes that affect executive compensation.
Summary of Complete Findings
Executive compensation is frequently under scrutiny from lawmakers, regulators, shareholders, and other company stakeholders. Previous research has argued that CEO compensation is inefficient because it increases in response to positive performance attributable to good luck, however it does not decrease as much in response to poor performance due to bad luck. This finding is commonly referred to as “asymmetry in pay for luck”. Some studies find that the asymmetry is stronger in firms with poor governance. “Luck” is defined as an external factor outside the CEO’s control that is typically measured by market and industry returns, while firm performance is measured by the firm’s stock returns.
This research shows that there is no asymmetry in pay for luck. The results of previous statistical analyses are not robust to changes in the specifications of the empirical models tested in those studies. This finding does not necessarily imply that CEO pay structures are optimal; rather it suggests that if there are inefficiencies in executive compensation, they cannot be proven by the existence of asymmetry in pay for luck. More generally, this research warns against making definite conclusions without a full range of testing, even on concepts that are intuitively appealing.
The authors used the Standard & Poor’s ExecuComp database to test the methodology commonly used to investigate asymmetry in pay for luck. They noted at least 17 decisions researchers need to make when designing the specifications of their empirical models. For each decision, there are several plausible alternative choices. To stress-test the methodology, the authors re-estimated the model using different ways of splitting performance into luck and skill, alternative luck measures and performance measures, different timeframes, and alternative samples in terms of governance, time-periods and industry.
Fewer than 2% of the 205 alternative specifications estimated in the study show some asymmetry in pay for luck; and in an overwhelming majority of circumstances, the increases in executive compensation attributable to good luck are similar in size to the decreases in executive compensation attributable to bad luck.
This study does not suggest that the statistical results of previous analyses were incorrect because they could be replicated. However, the results in earlier studies did not hold true when tested on a wider sample period and outside the group of the largest firms. Overall, the results of prior studies could not be generalized.
In conclusion, this research is an empirical tour de force which shows the critical importance of academic analysis to inform the debate and potential regulatory decisions on contentious topics such as executive compensation.
“Symmetry in Pay for Luck” by Naveen D. Daniel (Drexel University), Lily Yuanzhi Li (Villanova University), Lalitha Naveen (Temple University), was published in The Review of Financial Studies, July 2020.
> Access paper for subscription holders
> Access working paper