Marital Prenups? A Look at CEO Severance Agreements
U.S. Securities and Exchange Commission - Division of Economic and Risk Analysis
March 15, 2011
Some theoretical literature on firm-specific human capital investment suggests that severance contracts generate strong incentives for CEOs to ensure firm profitability, while the agency problem theory argues severance agreements are a less effective executive compensation measure. Using a unique sample of S&P500 CEO severance agreements between 1993 and 2007, I examine whether and how the existence and structure of severance agreements influence subsequent firm performance, CEO investment behavior, as well as CEO turnover likelihood. I show that while firms with severance agreements tend to perform worse than firms without severance agreements, the structure of severance agreements also makes a significant difference in performance. In particular, I find that firms with cash-only severance contracts underperform other firms that also include equity elements in the agreement. In addition, I find that severance agreements affect CEOs’ risk-taking behavior in that they encourage CEOs to overinvest in R&D. The findings are robust to controls for possible endogeneity concerns. Consistent with these findings I show in event study analysis that shareholders react negatively when firms award CEOs severance contracts and positively when firms cease to do so. Moreover, executives with severance contracts are more likely to be forced out of their firms. Results indicate that severance agreements, rather than promoting incentive alignment between CEOs and shareholders, actually exacerbate agency problems which lead to overinvestment and poor subsequent firm performance.
Number of Pages in PDF File: 49
Keywords: Executive Compensation, Severance, Managerial Incentives, Firm Performance
JEL Classification: G34, J33, J41
Date posted: March 15, 2011 ; Last revised: March 15, 2012
© 2016 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollobot1 in 1.188 seconds