Print
E-MailRui Albuquerque is Associate Professor at Boston College Carroll School of Management; Luis M. B. Cabral is the Paganelli-Bull Professor of Economics and International Business at the New York University Stern School of Business; and José Corrêa Guedes is Professor at the Católica Lisbon School of Business & Economics at the Catholic University of Portugal. This post is based on their recent article, forthcoming in the Review of Financial Studies. Related research from the Program on Corporate Governance includes Paying for Long-Term Performance by Lucian Bebchuk and Jesse Fried (discussed on the Forum here).
The SEC, the NYSE, and the U.S. government, accompanied by the actions of consultants, such as the Institutional Shareholder Services, recently have pushed to create, by means of relative performance evaluation (RPE), a tighter link between CEO pay and the factors under CEO control. This paper addresses the consequences of RPE for firm investment decisions and systemic risk in an industry model.
We propose a novel channel through which CEO incentive pay may have an effect on systemic risk. We consider the implications of relative performance evaluation, a practice that emerges in the equilibrium of our industry model. We show that RPE allows for a better alignment of interests between shareholders and managers, thereby reducing agency costs and rendering firms more productive; but it also leads managers disproportionately to choose investments that are correlated across firms, thus increasing systemic risk.
Our model features competition between two firms. Each firm is owned by a risk-neutral principal (the firm’s shareholders) and managed by a risk-averse agent (the firm’s CEO). The agent is required to spend costly unobservable effort that increases the firm’s returns. The manager must also choose how to allocate the firm’s assets. Our central assumption is that each firm has access to two investment opportunities, one with only idiosyncratic risk and another one with risk that is correlated across firms. To focus on risk, we assume both projects have the same expected return.
As in the classical principal-agent setting with hidden action, in our model the agent is induced to deploy unobservable effort by linking her pay to the firm’s performance. However, because the agent is risk averse, her contract can be improved upon by incorporating RPE: making compensation depend on relative rather than absolute performance leads in equilibrium to lower pay volatility, especially when the common investment opportunities are highly correlated across firms.
The model’s novelty stems from the strategic interaction between firms and the endogeneity of the industry return. Relative performance compensation leads managers to put more weight on investments that are common to the rival firm, as opposed to firm-specific investments subject to idiosyncratic risks. Moreover, the weights placed by each firm in the common project are strategic complements: the more one firm invests in common investments, the more the other firm wants to do the same. A greater weight placed by all firms on the common project implies greater correlation of the firms’ overall returns. This greater correlation reduces the level of firm CEO’s risk for a given level of pay. This in turn is good news for risk-neutral firm shareholders: the same level of agent utility can be offered with lower expected pay. We show that the induced strategic choice by shareholders of CEO contracts features strategic complementarity in the degree of RPE: if one firm designs a compensation package with more RPE, the optimal response of the rival firm’s shareholders is to increase the level of RPE in the compensation of their own manager.
Another novelty of our model is that we consider firm leverage in the principal-agent model. We show that with leverage, the manager is incentivized to invest more in both risky projects—to the extent that these earn a return higher than the borrowing rate—particularly in the correlated project. Because some of the risk associated with the correlated project can be hedged via RPE, the manager is offered more RPE, and engages in relatively more investment in the correlated project, than in the model without leverage.
In the banking industry, the regulatory push for greater use of RPE described above is being met with a contemporaneous and unprecedented effort by central banks around the world to regulate CEO pay as part of a reform effort that followed the subprime crises in the United States and in other countries. We analyze the impact on systemic risk of many of the new regulatory actions by central banks that constrain CEO pay. We show that these policies are costly to shareholders who then optimally adjust RPE as a way to minimize this cost, effectively undoing the intended consequences of the systemic-risk-reducing policies. We view this ineffectiveness result as a reflection of the argument put forth by Richard Posner that these regulatory actions are bound to fail due to loopholes. More than a loophole, we argue that existing dimensions of executive pay will adjust to an artificial regulation; and that, as a result, the intended goal of reducing systemic risk may fail to materialize; rather, a negative effect may take place in corporate behavior.
We develop several empirical predictions of the model. Here we highlight three of them, which we believe are new in the literature. First, in the cross section, firms that have more leverage engage in both more investment in correlated projects and use more RPE to reduce the CEO’s exposure to this risk. The model thus identifies a novel channel through which leverage induces systemic risk. Second, the use of relative performance evaluation should vary over time with the availability of correlated projects. For example, stronger correlations in stock returns in market downturns should be associated with greater incentives to correlate strategies for closet indexers in the mutual fund industry. While the econometrician may not know at every point in time the information available to the board with respect to such investment opportunities for a general industry, ex post the information may be revealed in the balance sheet. Another way to test this prediction is to use changes in the tax code or in regulations that create opportunities for new products or markets, for example, barriers to interstate commerce, or other significant industry transformations including product innovation, and test how RPE changes subsequently. Third, as an effect of RPE, executive pay volatility decreases as industry volatility increases, all else constant. This prediction is new in that it relates directly to incentive pay as a source of herding behavior and helps identify our mechanism from other sources of correlated actions.
The complete paper is available for download here.
