The Advisors' Blog

Here’s a pretty interesting study from three B-school profs about what actual CEO take-home pay can tell investors, compensation committees, and others that use executive pay info for decision-making. Let’s start with the takeaways – excerpted from this CLS Blue Sky blog:

We find that a firm’s accounting performance during the evaluation period strongly predicts the probability that the CEO will receive a payout, the probability that the payout is at the target level or higher, and the actual payout amount after controlling for firm characteristics, other aspects of CEO pay, CEO power, and corporate governance. We find no evidence that the strong payout-for-performance relation is driven by earnings management. There is also no strong evidence that firms design contacts to award CEOs regardless of performance.

In contrast to the significant relation between the actual plan payout and performance, the disclosed target plan payout, which is often used as a proxy for the estimated fair value of performance plans, is not correlated with firm performance in the evaluation period. The finding suggests that the current practice of extrapolating executives’ payout-for-performance relation from the reported ex-ante expected values could be misleading.

We also find that the actual plan payout contains information about CEO quality. Achieving the target payment level or higher may signal that the CEO is of good quality as she can meet or beat the internal performance expectations set by the board. Confirming that investors infer CEO quality from actual plan payouts, we find that, when the LTAP payout is at or above the target payment level, the stock market reacts positively, shareholders are less likely to vote against executive compensation in subsequent say-on-pay (SOP) voting, and the CEO is less likely to leave the firm over the next two years.

A small subset of our sample LTAPs (8.6 percent of those with actual payouts) has abnormally high payouts that generally exceed the plan’s maximum payment level. These firms do not have better accounting or stock performance over the performance evaluation period than firms with normal payouts and experience significantly lower performance over the next three years. The CEOs who receive abnormal payouts tend to be powerful and work for firms with weak governance, and they are more likely to sell their shares after receiving abnormal payments. After reading the proxy statements, we find that firms granting abnormal LTAP payouts on average provide less detailed disclosure, adopt complex plans with soft payout targets, and allow adjustments to the amount ultimately paid.

In a perfect world, these findings could be compelling support for the SEC’s new pay versus performance disclosure rule, with its notion of “compensation actually paid.” But we’ll have to file this with our hopes & dreams of “what could have been” – because it doesn’t match up with the final rule that the SEC actually adopted.

The difference is that the the researchers calculated take-home pay by simply using ex-post payouts, rather than using ex-ante estimates of equity awards and other benefits. That’s more in line with how you would think “compensation actually paid” would be defined. But their approach also has some limitations. For example, it may be complicated for plans that have multiple performance metrics.

Speaking of plan complexity and multiple performance metrics, here’s one other interesting finding about plan design:

LTAPs can differ in various contract features, such as performance horizon, relative performance evaluation, or the number of performance contingencies. We find that plans contingent on multiple accounting performance metrics demonstrate a weaker payout-for-performance relation.

We do not find that the payout-for-performance relation depends on the length of the evaluation period, whether the performance hurdles are disclosed ex-ante, or whether the performance is benchmarked to peer firms (i.e., relative performance evaluation). Taken together, the influence of contact design on the payout-for-performance relation appears to be weak.

Programming Note: This blog will be off tomorrow and Friday. Happy Thanksgiving!

– Liz Dunshee