L’étude de Mathias Kronlund, professeur au département de finance de l’Université de l’Illinois à Urbana-Champaign et Shastri Sandy, professeur au département de finance de l’Université du Missouri à Columbia, aborde un sujet dont nous avons beaucoup parlé au cours des cinq dernières années : le Say-on-Pay.
Il est temps de revisiter les résultats de ce mode de consultation des actionnaires à propos des rémunérations globales des hauts dirigeants. Les auteurs font une analyse très fine des conséquences liées au Say-on-Pay.
Dans l’ensemble, les résultats montrent que cette mesure a eu des effets positifs sur les décisions des comités de rémunération qui proposent des schèmes de rémunération plus en ligne avec la performance organisationnelle.
« The net effect on total CEO pay from these changes in various pay components is positive. In other words, firms increase total CEO compensation when they face increased scrutiny, mainly as a result of the higher stock awards. Thus, to the extent that the goal of the say-on-pay mandate was to reduce total executive pay, this regulation has had the opposite effect. We generally find much weaker results among non-CEO executives compared with CEOs, which is consistent with CEO pay receiving the most scrutiny around say-on-pay votes ».
Bonne lecture !
As a result of the Dodd-Frank Act of 2010, public firms must periodically hold advisory shareholder votes on executive compensation (“say on pay”). One of the main goals of the say-on-pay mandate is to increase shareholder scrutiny of executive pay, and thus alleviate perceived governance problems when boards decide on executive compensation. In our paper, Does Shareholder Scrutiny Affect Executive Compensation? Evidence from Say-on-Pay Voting, which was recently made publicly available on SSRN, we examine how firms change the structure and level of executive compensation depending on whether the firm will face a say-on-pay vote or not.
The theoretical impact of having a say-on-pay vote on executive compensation is ambiguous. On the one hand, it is possible that having a vote results in more efficient compensation practices, for example, in the form of stronger alignment between pay and performance, or in the form of lower pay if past pay was excessive. Say-on-pay may also improve compensation practices simply because directors pay more attention to executive compensation when they know that the pay packages they award face increased scrutiny. On the other hand, it is also possible that say-on-pay results in less efficient compensation practices. For example, having a say-on-pay vote may lead firms to excessively conform to the guidelines of proxy advisors, who tend to prefer specific pay practices that may not sufficiently account for each firm’s unique circumstances. Finally, it is possible that say-on-pay has no effect at all, either because governance problems are so severe that say-on-pay is an insufficient mechanism to improve firms’ pay practices, or conversely, because firms already have optimal pay practices and therefore have no reason to change them in response to increased scrutiny.
To examine the effect of say-on-pay on executive compensation, our identification strategy exploits within-firm variation regarding when (i.e., in which years) firms hold say-on-pay votes based on a pre-determined cyclical schedule. Specifically, many firms have elected to hold votes in cycles of every two or three years rather than every year, resulting in predictable year-to-year variation in whether a vote is held or not. Our empirical strategy then compares executive compensation across years when, according to its voting cycle, a firm is expected to hold a vote, versus the same firm in years when it is expected to not hold a vote.
Our results show that in years when firms are expected to hold a say-on-pay vote, they decrease CEO salaries, and increase stock awards. We also find that firms are significantly less likely to have change-in-control payments (“golden parachutes”) for their CEOs in years with a vote. These results are consistent with altering pay practices to better comply with proxy advisors’ guidelines. Further, deferred compensation and pension balances are higher in years with a vote, which is consistent with say-on-pay resulting in increased use of less-scrutinized components of pay.
The net effect on total CEO pay from these changes in various pay components is positive. In other words, firms increase total CEO compensation when they face increased scrutiny, mainly as a result of the higher stock awards. Thus, to the extent that the goal of the say-on-pay mandate was to reduce total executive pay, this regulation has had the opposite effect. We generally find much weaker results among non-CEO executives compared with CEOs, which is consistent with CEO pay receiving the most scrutiny around say-on-pay votes.
We also find economically large, but statistically weaker, evidence that executives choose to exercise fewer options in years when they face say-on-pay votes. Executives thus appear to shift realized pay from voting years to non-voting years—which suggests that executives believe that observers of pay (e.g., shareholders, news media) do not distinguish between awarded pay and ex-post realized pay.
One goal of the say-on-pay regulation was to foster more transparent CEO compensation and better alignment of CEO incentives with the interests of shareholders. Overall, our results show that holding a say-on-pay vote does cause firms to change how they pay executives. But despite the law’s intention of improving executive pay practices, the say-on-pay mandate has not unambiguously resulted in more efficient CEO compensation. And contrary to the goals of the say-on-pay regulation, the net result of these changes may be higher, not lower, total compensation. The fact that salaries are lower but stock awards higher is consistent with firms being particularly concerned about the optics of pay (Bebchuk and Fried (2004)) in years when compensation will be put to a vote, but is also consistent with models of optimal pay as in Dittmann and Maug (2007). Because CEOs receive more stock awards in voting years, which in turn will make their wealth more closely aligned with that of shareholders going forward, it is possible that pay in these years is more efficient, despite being higher. The fact that firms change pay practices between years with and without votes further is evidence that pay practices are not always perfectly optimal. If they were, whether a vote is held or not should be irrelevant for pay.
The full paper is available for download here.