Print
E-MailJung Ho Choi is Assistant Professor of Accounting, Brandon Gipper is Assistant Professor of Accounting, and Shawn Shi is a Ph.D. Student at the Stanford Graduate School of Business. This post is based on their recent paper. Related research from the Program on Corporate Governance includes Lucky CEOs and Lucky Directors by Lucian Bebchuk, Yaniv Grinstein and Urs Peyer (discussed on the Forum here); Paying for Long-Term Performance by Lucian Bebchuk and Jesse Fried (discussed on the Forum here); and Golden Parachutes and the Wealth of Shareholders by Lucian Bebchuk, Alma Cohen, and Charles C. Y. Wang (discussed on the Forum here).
Per the Wall Street Journal (May 17, 2019), when it comes to CEO compensation at big companies, “the best performers got big pay and big raises […], but the laggards didn’t do much worse.” The data underlying the central argument of that article pointed to a year-on-year rise in median compensation for S&P 500 CEOs of 6.6 per cent, taking pay to its highest level since the 2008 recession, in spite of median shareholder return generated by those same companies slumping 5.8 per cent over the same period, the worst such dip since the financial crisis.
This state of affairs is presented as a classic manifestation of the concept of “pay for luck”, a common form of compensation for non-performance—a long-established theory of executive rent extraction. There is considerable public interest in executive pay, and in particular its relation to both individual and company performance. However, the two foundational papers providing empirical evidence of the “pay for luck” phenomenon—Bertrand and Mullainathan, and Garvey and Milbourn—date from 2001 and 2006, and are based on data from 1984-1991 and 1992-2001, respectively. In the years since then, there have been major changes to corporate governance—including the expensing of stock options, new pay disclosure rules, and reforms (such as TARP and Dodd-Frank) prompted by the financial crisis—that we predict should be associated with a decrease in pay for luck in the period since that foundational research was published.
Based on our underlying hypothesis that pay governance changes—such as the transparency-based regulations introduced over the past 20 years—can limit executive rent extraction behaviour through improved shareholder monitoring, our research sets out to address two issues: first, whether pay for luck still exists in recent years; second, the link between pay for luck and pay governance.
While Bertrand and Mullainathan (2001) articulated the pay for luck phenomenon by documenting that executives are rewarded for performance that is influenced by industry and market forces rather than their own efforts, in addressing these issues our research is closely related to two other studies. First, Davis and Hausman (2018), who find evidence of pay for luck from 1992 to 2016 and conclude that executives are still extracting rents from shareholders; our research highlights a decrease in pay for luck in the latter years of that period. Second, Daniel, Li, and Naveen (2019), who reject the existence of asymmetric pay for luck; we offer an alternative explanation of their findings, proposing that regulations may change institutional features allowing pay for luck.
We find evidence of pay for luck: for the period 1997-2006, for every $1,000 of “luck profits” (attributable to factors other than executive performance), the CEO receives $3.46 in total pay; for every 1 per cent of lucky return on assets, the CEO receives 1.7 per cent more in total pay; a 1 per cent luck-related change in their firm’s market value yields a 37-basis-point change in the CEO’s pay. But we also find evidence of pay for luck declining over time. While the entire first decade covered by our research exhibits positive, significant pay for luck, each of our three measurements (luck profits, lucky return on assets, luck-related change in firm market value) sees a sharp decline for the period 2007-2016: 98 per cent, 53 per cent, and 48 per cent respectively. This indicates that the structural shift away from pay for luck occurred approximately in the middle of our sample, around or immediately after 2007. This held true when measuring luck in a number of different ways. We find this structural break to be robust to different measures of luck, industry changes, and the financial crisis.
We identify two meaningful regulatory events in our two-decade sample, with which we associate the changes in pay for luck from one decade to the next: the passage of Sarbanes-Oxley and voluntary option expensing halfway through the first decade, and the implementation of expanded pay disclosures through CD&A between the first and second decades as a response to the financial crisis. We find that regulations introduced with the intent of rebalancing performance and pay by encouraging greater oversight of executive compensation by way of greater transparency have indeed succeeded: there are large and significant decreases in pay for luck among companies that have adopted the disclosures of CD&A more intensively than the median.
Conclusion
We identify a clear structural break in “pay for luck” between the two periods studied, that is robust to different measures of luck, systemic changes in industry shocks, and the 2008-09 financial crisis. Furthermore, while it should be noted that—because new laws and regulations tend to cluster and occur along with other structural shifts in the economy—our evidence is not necessarily causal and our inferences do not necessarily generalise to other transparency-based regulation, we do find a compelling case for the decline in “pay for luck” being associated with two events. First, option “pay for luck” disappears during the first period around the passing of Sarbanes-Oxley and the adoption of option expensing. Second, total “pay for luck” declined around the introduction of CD&A, which mandated increased transparency concerning performance pay. We anticipate further research that more closely measures the causal link between regulation and “pay for luck”.
The complete paper is available for download here.
