1、Competition and Incentives with Motivated AgentsBy TIMOTHY BESLEY AND MAITREESH GHATAK*A unifying theme in the literature on organizations such as public bureaucraciesand private nonprofits is the importance of mission, as opposed to profit, as anorganizational goal. Such mission-oriented organizati
2、ons are frequently staffed bymotivated agents who subscribe to the mission. This paper studies incentives in suchcontexts and emphasizes the role of matching the mission preferences of principalsand agents in increasing organizational efficiency. Matching economizes on theneed for high-powered incen
3、tives. It can also, however, entrench bureaucraticconservatism and resistance to innovations. The framework developed in this paperis applied to school competition, incentives in the public sector and in privatenonprofits, and the interdependence of incentives and productivity between theprivate for
4、-profit sector and the mission-oriented sector through occupationalchoice. (JEL D23, D73, H41, L31)The late twentieth century witnessed a his-toric high in the march of market capitalism,with unbridled optimism in the role of the profitmotive in promoting welfare in the productionof private goods. M
5、oreover, this generated abroad consensus on the optimal organization ofprivate good production through privatelyowned competitive firms. When it comes to theprovision of collective goods, no such consen-sus has emerged.1Debates about the relativemerits of public and private provision stilldominate.T
6、his paper suggests a contracting approach tothe provision of collective goods. It focuses ontwo key issues: how to structure incentives, andthe role of competition between providers. Atits heart is the idea that organizations that pro-vide collective goods cohere around a mission.2Thus production of
7、 collective goods can beviewed as mission-oriented.Our approach cuts across the traditionalpublic-private divide. Not all activities withinthe public sector are mission-oriented. For ex-ample, in some countries, governments own carplants. While this is part of the public sector, theoptimal organizat
8、ion design issues here are nodifferent from those faced by General Motors orFord. Not all private sector activity is profit-oriented. Universities, whether public or pri-vate, have many goals at variance with profitmaximization. Our examples will draw fromboth the public and the private sectors.The
9、missions pursued in the provision ofcollective goods come from the underlying mo-tivations of the individuals (principals andagents) who work in the mission-oriented sec-tor. Workers are typically motivated agents, i.e.,agents who pursue goals because they perceiveintrinsic benefits from doing so. T
10、here are manyexamplesdoctors who are committed to sav-ing lives, researchers to advancing knowledge,judges to promoting justice, and soldiers to de-fending their country in battle. Viewing workers* Besley: London School of Economics, HoughtonStreet, London WC2A 2AE (e-mail: t.besleylsc.ac.uk);Ghatak
11、: London School of Economics, Houghton Street,London WC2A 2AE (e-mail: m.ghataklse.ac.uk). Theauthors would like to thank numerous seminar participantsas well as Daron Acemoglu, Alberto Alesina, MadhavAney, Ben Bernanke, Michela Cella, Leonardo Felli, JulianLe Grand, Caroline Hoxby, Rocco Machiavell
12、o, MegMeyer, Hannes Mueller, Andrei Shleifer, Daniel Sturm,Steve Tadelis, Jean Tirole, and two anonymous referees forhelpful comments. Financial support from the Economicand Social Research Council through research grant RES-000-23-0717 is gratefully acknowledged.1We use the term “collective good” a
13、s opposed to thestricter notion of a “public good.” Collective goods in thissense also include merit goods. This label also includes agood like education to which there is a commitment tocollective provision, even though the returns are mainlyprivate.2See, for example, James Q. Wilson (1989) on publ
14、icbureaucracies and Robert M. Sheehan (1996) on nonprofits.Jean Tirole (1994) is the first paper to explore the implica-tions of these ideas for incentive theory.616as mission-oriented makes sense when the out-put of the mission-oriented sector is thought ofas producing collective goods. The benefit
15、s andcosts generated by mission-oriented productionorganizations are not fully reflected in the mar-ket price. In addition, donating our incomeearned in the market to an organization thatpursues a mission that we care about is likely tobe an imperfect substitute for joining and work-ing in it. This
16、could be due to the presence ofagency costs or because individuals care notonly about the levels of these collective goods,but also about their personal involvement intheir production (i.e., a “warm glow”).It is well known from the labor literature oncompensating differentials that employmentchoices
17、 and wages depend on taste differences(Sherwin Rosen, 1986). This paper exploreshow these economize on the need for explicitmonetary incentives while accentuating the im-portance of nonpecuniary aspects of organiza-tion design in increasing effort. Thus, missionchoice can affect the productivity of
18、the orga-nization. For example, a school curriculum ormethod of discipline that is agreed to by theentire teaching faculty can raise schoolproductivity.Mission preferences typically differ, how-ever, between motivated agents. Doctors mayhave different views about the right way to treatill patients,
19、and teachers may prefer to teachdifferent curriculums. This suggests a role fororganizational diversity in promoting alterna-tive missions and competition between organi-zations in attracting motivated agents whosemission preferences best fit with one another.We show that there is a direct link betw
20、een suchsorting and an organizations productivity.The insights from the approach have applica-tions to a wide variety of organizations includ-ing schools, hospitals, universities, and armies.In this paper we abstract from the question ofpublic versus private ownership.3The primi-tives are the produc
21、tion technology, the moti-vations of the actors, and the competitiveenvironment. We also abstract (for the mostpart) from issues of financing.We benchmark the behavior of the mission-oriented part of the economy against a profit-oriented sector where standard economicassumptions are madeprofit seeki
22、ng and nononpecuniary agent motivation. This is impor-tant for two reasons. First, we get a precisecontrast between the incentive structures ofprofit-oriented and mission-oriented production.Second, the analysis casts light on how changesin private sector productivity affect optimalincentive schemes
23、 operating in the mission-oriented sector. This has implications for de-bates about how pay-setting in public sectorbureaucracies responds to the private sector.Our approach yields useful insights into on-going debates about the organization of themission-oriented sector of the economy. Forexample,
24、it offers new observations on the roleof competition in enhancing productivity inschools. More generally, it suggests that one ofthe main virtues of private nonprofit activity isthat it can generate a variety of different mis-sions which improve productivity by matchingmanagers and workers who have
25、similar mis-sion preferences. An analogous argument canbe made in support of decentralization of publicservices. On the flip side, however, public bu-reaucracies, whose policies can be imposed bypoliticians, may easily become demotivated inthe event of a regime change. Also, whilematching on mission
26、 preferences is potentiallyproductivity enhancing, it also leads to conser-vatism and can raise the cost of organizationalchange.This paper contributes to an emerging litera-ture which studies incentive issues outside ofthe standard private goods model.4One strandputs weight on the multitasking aspe
27、cts of non-profit and government production along thelines of Bengt Holmstrom and Paul Milgrom(1991). Another emphasizes the career concernaspects of bureaucracies (Mathias Dewatripontet al., 1999; Alberto Alesina and Guido Tabel-lini, 2004). These two areas are brought to-gether in Daron Acemoglu e
28、t al. (2003).However, these all work with standard motiva-tional assumptions. This paper shares in com-mon with George Akerlof and Rachel Kranton(2005), Roland Benabou and Tirole (2003),Josse Delfgaauw and Robert Dur (2004), Avi-nash Dixit (2001), Patrick Francois (2000),Kevin Murdock (2002), Canice
29、 Prendergast3See Besley and Ghatak (2001) on the question ofoptimal ownership in the context of public goods.4See Dixit (2002) for a survey of this literature.617VOL. 95 NO. 3 BESLEY AND GHATAK: COMPETITION AND INCENTIVES WITH MOTIVATED AGENTS(2001), and Paul Seabright (2003) the notionthat nonpecun
30、iary aspects of motivation mat-ter.5In common with Vincent Crawford andJoel Sobel (1982) and Philippe Aghion and Ti-role (1997), our approach emphasizes how non-congruence in organizational objectives canplay a role in incentive design. We explore,however, the role of matching principals andagentsse
31、lection rather than incentivesas away to overcome this.6The remainder of the paper is organized asfollows. In Section I, we lay out the basic modeland study optimal contracts and the matching ofprincipals and agents. Section II explores appli-cations of the model, and Section III concludes.I. The Mo
32、delA. The EnvironmentA “firm” consists of a risk-neutral principaland a risk-neutral agent. The principal needs theagent to carry out a project. The projects out-come (which can be interpreted as quality) canbe high or low: YHH11005 1 (“high” or “success”)and YLH11005 0 (“low” or “failure”). The pro
33、babilityof the high outcome is the effort supplied by theagent, e, at a cost c(e) H11005 e2/2. Effort is unob-servable and hence noncontractible. We assumethat the agent has no wealth and so cannot put ina performance bond. Thus, a limited-liabilityconstraint operates, which implies that the agentha
34、s to be given a minimum consumption levelof wH6018 H11350 0 every period, irrespective of perfor-mance. Because of the limited-liability con-straint, the moral hazard problem has bite. Thisis the only departure from the first-best in ourmodel.We assume that each principal has sufficientwealth so as
35、not to face any binding wealthconstraints, and that the principal and agenteach can obtain an autarchy payoff of zero.The mapping from effort to outcome is thesame for all projects. We also assume thatagents are identical in their ability to work onany type of project. Projects differ exclusivelyin
36、terms of their missions. A “mission” consistsof attributes of a project that make some prin-cipals and agents value its success over andabove any monetary income they receive in theprocess. This could be based on what the orga-nization does (charitable versus commercial),how they do it (environment-
37、friendly or not),who the principal is (kind and caring versusstrict profit-maximizer) and so on. Allowingagents to have preferences over their work en-vironment follows a long tradition in labor eco-nomics (see, for example, Rosen, 1986).In our basic model, missions are exogenouslygiven attributes o
38、f a project associated with aparticular principal. In Section I C, we examineconsequences of endogenizing mission choice.There are three types of principals andagents, labelled i H20678 0, 1, 2 and j H20678 0, 1, 2.The types of principals and agents are perfectlyobservable. If the project is success
39、ful, a prin-cipal of type i receives a payoff of H9266iH11022 0. Allprincipals receive 0 if a project is unsuccessful.Principals of type 0 have the same preferencesas in the standard principal-agent model, i.e.,H92660is entirely monetary. However H92661and H92662mayhave a nonpecuniary component. To
40、focus ex-clusively on horizontal aspects of matching be-tween principals and agents, we assume thatH92661H11005 H92662H11013 H9266.7Some agents care about the mission of theorganization for which they work. Formally thisimplies that the payoff of such agents dependson their own type, and the type of
41、 the principalfor whom they work. Like principals, all agentsare assumed to receive 0 if the project fails,irrespective of with whom they are matched.Agents of type 0 have standard pecuniaryincentivestheir utility depends positively onmoney and negatively on effort. Since they aremotivated solely by
42、 money, they do not careintrinsically about which organization theywork for. In contrast, an agent of type 1 (type 2)receives a nonpecuniary benefit of H9258H6126from5Some of these ideas consider the possibility that intrin-sic motivation can be affected by the use of explicit incen-tives (see also
43、Richard Titmuss, 1970, and Bruno S. Frey,1997). We treat the level of intrinsic motivation as given.6See Daniel Ackerberg and Maristella Botticini (2002),Kaniska Dam and David Perez-Castrillo (2001), andEdward P. Lazear (2000) for approaches to principal agentproblems where sorting is important.7We
44、use the term horizontal matching to describe asituation where there is no difference in the productivity oforganizations when principals and agents are efficientlymatched. The standard vertical matching model looks atsituations where some principals and agents are more pro-ductive regardless of whom
45、 they match with. We will brieflyreturn to the implications of vertical sorting in Section II B.618 THE AMERICAN ECONOMIC REVIEW JUNE 2005project success if he works for a principal oftype 1 (type 2), and H9258 if matched with a princi-pal of type 2 (type 1), where H9258H6126H11022 H9258 H11350 0.8T
46、he payoff of an agent of type j who ismatched with a principal of type i when theproject succeeds can therefore be summarizedas:H9258ijH11005H208770 i H11005 0 and/or jH110050H9258 iH20678H208531, 2H20854, jH20678H208531, 2H20854, iHS11005jH9258H6126iH20678H208531, 2H20854, jH20678H208531, 2H20854,
47、iH11005j.We will refer to the parameter H9258ijas agentmotivation and agents of type 1 and 2 as moti-vated agents. We will refer to the economy asbeing divided into a mission-oriented sector(i.e., i H11005 1, 2) and a profit-oriented sector (i.e.,i H11005 0).We makeASSUMPTION 1:maxH20853H92660, H926
48、6 H11001H9258H6126H20854H110211.This ensures that there is an interior solution foreffort in all possible principal-agent matches.The analysis of the model is in three steps.We first solve for the optimal contract for anexogenously given match of a principal of typei and an agent of type j. Contract
49、s betweenprincipals and agents have two components: afixed wage wij, which is paid regardless of theproject outcome, and a bonus bij, which theagent receives if the outcome is YH. Initially, wetake the agents reservation payoff uH6126jH11350 0tobeexogenously given. Second, we consider theextension to endogenous missions which makesH9258ijendogenous. Third, we study matching ofprincipals and agents where the reservation pay-offs are endogenously determined.B. Optimal ContractsAs a benchmark, consider the first-best casewhere effort is contractible. This will result ineffort bein
