Say-on-pay, an annual, non-binding shareholder vote on CEO compensation, is one of the primary ways of giving shareholders a voice in corporate governance. It is in essence a vote of confidence on the board of directors’ compensation decisions and the performance of the CEO.[1]

Despite its potential importance, the impact of say-on-pay on compensation is unclear. First, the outcome of the vote is non-binding, meaning there is no explicit consequence if the vote goes badly for the board – the consequences, if they exist, are entirely implicit. Second, and perhaps more important, the proportion of shareholders voting in favor of compensation has historically been very high – on average around 93 percent, with only 7 percent of votes considered a “failure.”

However, an important consideration is that say-on-pay votes occur after compensation policy has been set and company performance realized. The impact of say-on-pay depends on how much the board of directors considers the threat of say-on-pay failure in their compensation decisions, which are made well in advance of the vote.

In a new article, I venture to understand, quantitatively, how much impact say-on-pay has on compensation policy and the economic channels through which this impact occurs. I pose a model of CEO compensation with non-binding shareholder approval votes. I then fit this model to the data to estimate the underlying model parameters that describe the preferences of the board (with respect to compensation, CEO entrenchment, and efforts to maximize shareholder value) and shareholders (with respect to compensation, firm value, and their preferred vote outcome).

I find that, despite its low failure rate and non-binding nature, say-on-pay is an effective governance mechanism: Providing this regularly occurring vote to shareholders lowers total CEO pay levels by about 6.6 percent. This impact on compensation reins in the CEO’s actions by reducing an agency cost, bringing compensation policy closer to that which maximizes shareholder value. I further find that say-on-pay increases firm value by 2.4 percent, on average. These numbers align closely with research studying how the adoption of say-on-pay across the world affected compensation and value (Correia and Lel, 2016).

How does one square this impact on compensation and firm value with the low failure rate and non-binding nature of the vote? My estimated model provides an answer. First, I find that the board takes into account the cost of a potential vote failure, which affects compensation decisions – even though say-on-pay is non-binding.

Interestingly, I also find that shareholders perceive a cost to a say-on-pay vote failure. For example, shareholders see a cost to dissenting from the board on a prominent corporate policy, which causes them to approve some CEO pay packages that they would not approve if the cost did not exist. That shareholders perceive a cost to say-on-pay failure is supported by recent surveys (Edmans, Gosling & Jenter, 2023), in which institutional investors explicitly say that failing a say-on-pay vote is costly.

These costs clarify the economic mechanism of say-on-pay: It resembles a punishment mechanism – shareholders can punish the board by voting against CEO compensation and dissenting, but dissent itself is not free. Though failed votes rarely occur and are costly to shareholders, providing them with the punishment mechanism enhances firm value.

This impact on compensation and value is masked by an exclusive focus on the high approval rate of say-on-pay votes. It is the threat of vote failure that affects the board’s compensation decisions. Just because failures do not happen very often does not make say-on-pay ineffective.

These findings have important implications for interpreting shareholder voting outcomes. The vote result may not be the best measure of a vote’s impact – one must consider how the consequences of a potential failed vote influence corporate decisions. Many articles that cite their low failure rate to conclude that say-on-pay votes lack impact may be missing the point.

However, legal practitioners and academics have long wondered whether say-on-pay would be more effective as a binding vote (Allaire and Dauphin, 2016). There is scant empirical evidence comparing binding with non-binding voting, as such analysis would require comparison of outcomes (with respect to compensation and firm value, in the case of say-on-pay) for the same company under a binding and non-binding voting regime, and such a setting is unlikely to exist in the data.

My paper is thus uniquely positioned to compare non-binding and binding voting. I undertake a policy counterfactual by making say-on-pay a binding vote: Vote failure would lead to the CEO receiving the same remuneration as in the previous period – which is the outcome in say-on-pay in countries such as the UK and Switzerland.

In my model I find that a binding vote would in fact reduce the effectiveness of say-on-pay and lead to higher total compensation and lower firm value.

Why does this occur? In my model, when a binding say-on-pay vote fails, the CEO’s compensation remains unchanged. This creates a challenge: Boards hold annual meetings to review and adjust compensation based on new information about the performance of the CEO and the company. When learning about a CEO’s ability is most valuable (early in her tenure, for example), this information is especially useful for updating and refining compensation policy. As a result, shareholders may hesitate to use failed say-on-pay votes as a disciplining device because doing so could bind the company to an outdated pay structure. This (perhaps unintended) consequence of a binding say-on-pay lowers its disciplinary power (CEO pay increases), even though a binding vote explicitly grants shareholders some control over compensation policy.

These findings have important implications for shareholder democracy and whether binding or non-binding shareholder votes are better for corporate governance. While a binding vote may give shareholders more explicit control over corporate policies, a failed, binding vote could force the company to stick with policies that no longer align with its current situation.

RFEFERENCES

Allaire, Y. and Dauphin, F. (2016). Should say on pay votes be binding? https://clsbluesky.law.columbia.edu/2016/09/13/should-say-on-pay-votes-be-binding/ [Accessed: 2024-12-12].

Correa, R. and Lel, U. (2016). Say on pay laws, executive compensation, pay slice, and firm valuation around the world. Journal of Financial Economics, 122(3):500–520.

Dey, A., Starkweather, A., and White, J. T. (2024). Proxy advisory firms and corporate shareholder engagement. The Review of Financial Studies, 37(12):3877–3931

Edmans, A., Gosling, T., and Jenter, D. (2023). CEO compensation: Evidence from the field. Journal of Financial Economics, 150(3):103718.

ENDNOTE

[1] Say-on-pay (SOP) was adopted in the U.S, as part of the Dodd-Frank Act in 2010. SOP proposals are put forth by management at the annual shareholder meeting, and shareholders vote on the CEO’s compensation from the previous fiscal year. I use “failure” to refer to SOP proposals that do not garner the required support from shareholders. In the U.S., SOP votes are non-binding, so there is no threshold which forces the board to change pay policy. However, the understood threshold for SOP failure is 70 percent support (or 30 percent voting against, see Dey 2024), which is the threshold for action from Institutional Shareholder Services (ISS).  Further, SOP votes are approval votes on the previous year’s CEO compensation, not advisory votes on proposed compensation.

This post comes to us from Professor John W. Barry at Rice University. It is based on his recent article, “Shareholder Voice and Executive Compensation,” available here.