Documents

Incentives_April2002.pdf

Categories
Published
of 40
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Description
Executive Compensation, Incentives, and Risk Dirk Jenter MIT Sloan School of Management This draft: April 2002 First draft: November 2000 Comments Welcome 1 Executive Compensation, Incentives, and Risk 1 This paper analyzes the link between equity-based compensa- tion and created incentives by (1) deriving a measure of incentives suitable for both linear and non-linear compensation contracts, (2) analyzing the effect of risk on incentives, and (3) clarifying the role of the agent’s private tradi
Transcript
  Executive Compensation, Incentives,and Risk Dirk JenterMIT Sloan School of ManagementThis draft: April 2002First draft: November 2000Comments Welcome 1  Executive Compensation, Incentives, and Risk 1 This paper analyzes the link between equity-based compensa-tion and created incentives by (1) deriving a measure of incentives suitable for both linear and non-linear compensation contracts,(2) analyzing the e   ff  ect of risk on incentives, and (3) clarifying the role of the agent’s private trading decisions in incentive creation.With option-based compensation contracts, the average pay-for-performance sensitivity is not an adequate measure of ex-ante in-centives. Pay-for-performance covaries negatively with marginal utility and hence overstates the created incentives. Second, more noise in the performance measure implies that the manager is less certain about the e   ff  ect of e   ff  ort on performance, which in turn makes her less willing to exert e   ff  ort. Finally, the private trading decisions by the manager have   fi  rst-order e   ff  ects on incentives. By reducing her holdings of the market asset, the manager achieves an e   ff  ect similar to ”indexing” the stock or option grant, making explicit indexation of the contract redundant. 1 A previous version of this paper was titled ”Understanding High-Powered Incentives”.I am grateful to Jeremy Stein, Peter Tufano, John Campbell, Brian Hall, Tom Knox, AxelAdam-Mueller and Stuart Gillan for their suggestions, and to seminar participants at theEuropean Financial Management Association Meeting (Lugano), European Meeting of theEconometric Society (Lausanne), German Finance Association Meeting (Vienna), Finan-cial Management Association Meeting (Toronto), Southern Finance Association Meeting(Destin), and the Harvard Finance Lunch Seminar for their comments. I would like tothank Nittai Bergman, Brian Hall, Jeremy Stein, George Baker, Lisa Meulbroek, SendhilMullainathan, Robert Gibbons, Ernst Maug and Paul Oyer for extremly helpful discus-sions about executive compensation. All remaining errors are my own. 2  Introduction Principal-agent theory frames the design of incentive plans for corporatemanagers as a trade-o ff   between incentives and insurance. To induce man-agers to act in the interest of shareholders managerial wealth is linked tomeasures of corporate performance. At the same time, the noise in the avail-able measures of performance burdens managers with ine ffi cient levels of risk.The optimal incentive scheme balances the bene fi t of increased manageriale ff  ort with the deadweight loss from ine ffi cient risk sharing. 2 This characterization of the incentive compensation problem is silentabout the translation of contracts into incentives and the direct e ff  ect of risk on incentive creation. Two di ff  erent contracts can provide the sameprobability-weighted average link between managerial wealth and perfor-mance, but at the same time result in very di ff  erent levels of ex-ante incen-tives. Similarly, the same incentive scheme creates di ff  erent levels of incen-tives depending on the noisiness of the link between e ff  ort and performance.This paper contributes to our understanding of the link between incen-tive compensation and created incentives by (1) deriving and analyzing anappropriate measure of incentives for both linear and non-linear compensa-tion contracts, (2) analyzing the e ff  ect of risk on the incentives created bya given contract and (3) clarifying the role of the agent’s private tradingdecisions on incentive creation. The mechanism by which executive compen-sation schemes translate into managerial incentives has not been analyzed indetail in the extant literature, resulting in much unnecessary confusion. Thepaper at hand attempts to  fi ll this gap.For the class of incentive problems considered in this paper, e ff  ort incen-tives are appropriately measured as the expected marginal e ff  ect of e ff  ort onutility. The e ff  ort incentives can be decomposed into (i) managers’ expectedmarginal need for wealth, (ii) the expected marginal e ff  ect of e ff  ort on wealthand (iii) the covariation between need for wealth and e ff  ect of e ff  ort on wealth 2 There is a vast empirical literature analyzing CEO compensation in the context of theprincipal-agent model. Murphy (1999) provides an excellent summary. More recent em-pirical work has focused on estimating the sensitivity of CEO compensation to corporateperformance. See Jensen and Murphy (1990) and Hall and Liebman (1998) for two excel-lent examples. Other predictions of the agency model have been tested by Gibbons andMurphy (1992), Tufano (1996), Aggarwal and Samwick (1999), Cohen, Hall and Viceira(2000) and Bertrand and Mullainathan (1998, 2000) amongst others. 3  across states. The third incentive lever is often overlooked but turns out to becrucial for understanding incentive schemes, the e ff  ect of risk on incentives,and the relevant measure of risk for compensation schemes.The results can be summarized as follows: With non-linear compensationcontracts the link between measured performance and managerial wealthis state-dependent. The expected (probability-weighted average) pay-for-performance is not an adequate measure of ex-ante incentives. In particu-lar, with convex, option-based compensation schemes pay-for-performancecovaries negatively with marginal utility and overstates the actually cre-ated incentives by a substantial amount. A negative shock to the perfor-mance measure is associated with low pay-for-performance sensitivity andhigh marginal utility, while a positive shock is associated with high pay-for-performance sensitivity and low marginal utility. Hence options deliver atight link between wealth and performance in states of nature in which themanager’s need for additional funds is low. An e ffi cient incentive schemewould display the opposite, concave pattern, rewarding managerial e ff  ort inhigh marginal utility states and o ff  ering a smaller pay-for-performance slopein low marginal utility states.The second set of results is concerned with the direct e ff  ect of risk onincentives. Firstly, the e ff  ect of the covariation of the pay-performance linkwith marginal utility described in the previous paragraph is increasing in thevolatility of the performance measure. Thus the wedge between average pay-for-performance and the actual incentives created by option-based contractswidens as risk increases. This  fi nding is important for the measurement of incentives in the empirical literature, and for understanding the empiricalrelationship between incentives and risk (Prendergast, 2000b). Secondly, Idemonstrate that the e ff  ect of risk on incentives depends on wether the linkbetween e ff  ort and performance is itself stochastic. The standard principal-agent model utilizes an additive noise structure, implicitly assuming that themarginal e ff  ect of e ff  ort on measured performance is deterministic. Put dif-ferently, the additive model assumes that the manager knows with certaintyhow her e ff  ort a ff  ects  fi nal outcomes at the time she makes her decision. Adirect and counterintuitive consequence of this assumption is that an increasein risk leads to an increase in managerial e ff  ort due to a “precautionary e ff  ort”e ff  ect.I propose instead that a non-standardmultiplicative noise structure better4
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks