Considerable debate remains among academics and practitioners regarding the economic forces that drive CEO compensation practices in the United States. Some view the market for CEO talent as the main economic force that drives the level and form of CEO compensation (e.g., Rosen, 1992; Gabaix and Landier, 2008). Others argue that these forces have little effect on CEO compensation because of frictions such as managerial entrenchment, asymmetric information, and transaction costs of replacing managers, believing instead that compensation practices are by and large driven by the bargaining power that the CEO has vis-à-vis the board (e.g., Bebchuk and Fried, 2003).
The debate has intensified in recent years due to several controversial compensation practices, a first example of which is the tendency of firms to benchmark CEO compensation to that of other CEOs. While some find benchmarking consistent with competitive compensation (Holmstrom and Kaplan, 2003; Bizjak et al., 2008), others argue it is a way for CEOs to increase their compensation by benchmarking themselves to highly paid CEOs (e.g., Faulkender and Yang, 2010).
A second controversial practice that has received much attention in the literature is the tendency of firms to compensate their CEOs for firm performance that is outside their control. For example, Bertrand and Mullainathan (2001) show that when oil prices go up, oil companies tend to increase the compensation to their CEOs even though the increase in oil prices (and hence in an oil company’s value) is outside the CEO’s control. Bertrand and Mullainathan term this “pay for luck” and argue that this practice is driven by CEO self–interest. Others, such as Himmelberg and Hubbard (2000), argue that this practice reflects competitive compensation practices, as it embeds both CEO performance and the change in the value of CEO talent in the in the market (since higher market price reflects higher marginal contribution to CEO talent and therefore the higher price for CEO talent).
A third controversial practice has been the tendency of firms to provide large compensation packages to their CEOs in recent years. Some have attributed the large increase in pay to the power that CEOs have over their boards of directors (e.g., Bebchuk and Fried, 2003). Others argue that pay has risen because the return to CEO talent has increased. In particular, Gabaix and Landier (2008) point to the fact that the considerable rise in CEO compensation is commensurate with the large increases in firm size in recent years. They then argue that since CEO talent becomes more valuable to the firm as the firm becomes larger, CEO compensation should increase with firm size, consistent with the CEO talent argument.
In our paper, Does the Market for CEO Talent Explain Controversial CEO Pay Practices?, forthcoming in the Review of Finance, we shed new light on the role that the market for CEO talent plays in explaining these three controversial compensation practices. Our approach is to examine the extent to which these controversial practices can be explained by cross-sectional variation in the importance of firm-specific talent as compared to more generic talent that boards are looking for when choosing their CEO. Our proxy for the importance of CEO firm-specific talent is the percentage of insider CEOs in the industry in which the firm operates. Parrino (1997) uses this measure to capture the costs of hiring CEOs from outside the firm, showing that industries in which CEOs usually come from outside the firm tend to be more homogeneous in the sense that CEO talent from other firms can readily replace CEO talent from inside the firm. In contrast, industries in which CEOs tend to come from inside their own firms are more heterogeneous in nature, implying that CEO talent from inside the firm is harder to replicate.
By measuring the importance of firm-specific talent at the industry level, our proxy measures constraints faced by the board when considering a new CEO. In industries with few outside CEOs, the potential outside options of both the CEO and the firm are more limited. Furthermore, as these are industry-wide constraints they seem largely outside the control of the board. In our appendix, we make our main industry-level CEO talent proxy variable (i.e., the percentage of insider CEOs in each of the 48 Fama-French industry groups) available for other researchers.
Our main contribution to the literature is to consider how this industry-level proxy for the importance of the CEO having worked inside the firm can shed light on the debate over the controversial pay practices outlined above. In general, we expect CEO compensation practices to be different across industries with few versus many insider CEOs. Under the view that CEO talent drives CEO compensation, what happens at the industry level should matter less for CEO pay at firms in industries where boards typically choose insider CEOs. Under the view that the CEO’s bargaining power vis-à-vis is important, CEOs in industries with mostly insider CEOs are likely to have greater bargaining power. As a result, if the controversial pay practices are driven by entrenched CEOs taking advantage of strong bargaining positions, we would expect these practices to be more prevalent in industries with few outsider CEOs. Overall, we conclude that CEO talent pools are related to marginal decisions on CEO pay (such as benchmarking and pay-for-luck), but we cannot find evidence that it can explain the overall rise in CEO pay.
First, we examine the practice of benchmarking. If compensation practices are driven by competition in the market for CEOs, we hypothesize that then CEO compensation should be benchmarked to that of other CEOs in the industry only when the CEO has viable employment options in the industry. We find strong evidence that benchmarking is prevalent primarily in industries in which new CEOs tend to come from outside the firm. In contrast, the compensation to the CEO is not affected by changes in the compensation of CEOs in peer firms in industries in which CEOs tend to come from inside the firm. This finding supports the important role of the market for CEO talent in affecting benchmarking and is consistent with the interpretation in Bizjak et al. (2008).
Second, we examine the practice of paying the CEO for luck, i.e., for performance that is outside the CEO’s control. Our approach is to decompose stock performance into industry performance and firm-specific performance and to examine the extent to which industry performance (the component of performance outside the CEO’s control) explains changes in CEO compensation. According to Holmstrom (1979), CEOs should only be paid for the part of performance that they can influence, not for the performance that can be attributed to other factors such as industry-wide shocks. However, when a CEO’s outside options are associated with industry conditions, a correlation between CEO compensation and industry performance could rise naturally (Himmelberg and Hubbard, 2000; Hubbard, 2005). For example, when the industry is booming, the CEO has more options to use his or her talents in other firms; therefore the CEO should receive higher compensation. This argument thus implies that the relation between CEO compensation and industry compensation (i.e., pay for luck) depends on the extent to which the CEO has outside options in the industry. We find evidence that “pay for luck” is strongest in industries that have the largest percentage of outsider CEOs. In contrast, the relation between firm performance and industry performance is weaker in industries where CEOs tend to come from inside the firm. This result is again consistent with CEO labor market competition explaining the relation between CEO compensation and industry-wide performance.
Our main robustness check is to employ an alternative measure of CEO talent pools based on the number of other publicly traded firms in the same industry that have a headquarter within 100 miles of the firm’s headquarter (see Coval and Moskowitz (1999, 2001)). We argue that the outside talent pool for firms that are closer to other peer firms is likely to be larger. Likewise, top executives at public firms in a geographical area that is reasonably close to other public firms in the same industry have better potential outside options. The correlation between the percentage of insiders in the industry and the number of other public firms close-by equals -16%, indicating that this alternative proxy is substantially different (though with the expected sign). Using the alternative proxy, we find strong corroborating evidence for that benchmarking is only prevalent at firms with a large local CEO talent pool or where there are better outside options in the industry close-by. The results for pay for luck using the alternative proxy are similarly consistent, though with weaker statistical power.
Finally, we study the extent to which variations in the talent pool structure across industries explains the increase in CEO compensation in recent years. Building on the insight of Rosen (1992), Gabaix and Landier (2008) present a model in which more talented CEOs are attracted to larger firms, predicting that changes in CEO compensation should depend both on changes in the size of the firm in which the CEO operates and changes in the size distribution of firms in the economy (capturing the productivity of talent across firms, and hence the outside opportunities of CEOs with different talents). Their specification assumes that CEO skills are substitutable across firms and that profitability is a function of skills and firm size. Therefore, in equilibrium, more talented CEOs will be attracted to larger firms. If CEO skills are less transferable across firms, then they will be less able to move to larger firms if they are more talented, and the relation between talent and firm size should be weaker. Given our findings of the importance of firm-specific skills across industries in explaining CEO compensation practices, we expect industries with more firm-specific skills to have a weaker relation between compensation and firm size.
We find very little evidence of weaker relation between compensation and firm size in industries with more firm-specific skills. (We find an effect that is statistically significant but economically meaningless.) We further examine whether the firm-size distribution at the industry level (capturing talent variation within the industry) explains variation in compensation beyond what is captured by the firm-size distribution in the whole economy. We fail to find any economic relation between firm-size distribution at the industry level and compensation levels. The relation does not exist, regardless of whether CEOs in the industry tend to come from inside or outside of the firm. These results go against the argument that the market for CEO talent is a central force in the increase in CEO compensation in recent years. Instead, our results indicate that the size of the market-wide reference firm may be a proxy for something else that is not directly related to the equilibrium model in Gabaix and Landier.
The full paper is available for download here.