interdependent preferences and jsobel/papers/iprdraft.pdf · pdf filesobel:...

Click here to load reader

Post on 30-Aug-2018




0 download

Embed Size (px)


  • Journal of Economic LiteratureVol. XLIII (June 2005), pp. 392436

    Interdependent Preferences and Reciprocity



    Sobel: University of California, San Diego. I present-ed a version of this paper at the First World Congress ofthe Game Theory Society and to my colleagues at theCenter for Advanced Study in the Behavioral Sciences. Ithank Eli Berman, Antonio Cabrales, Miguel Costa-Gomes, Vincent Crawford, David M. Kreps, HerbertGintis, Mark Machina, Efe Ok, Lus Pinto, MatthewRabin, Paul Romer, Klaus Schmidt, Uzi Segal, JoelWatson, and Kang-Oh Yi for discussions, references, andcomments. I am especially grateful to two referees whosupplied detailed, intelligent, and constructive commentson an earlier version of the manuscript and to JohnMcMillan for his advice and encouragement. I worked onthis project while a Fellow at the Center for AdvancedStudy in the Behavioral Sciences. I thank my classmates atthe Center for conversations and encouragement and theCenter for financial and clerical support. NSF funding isalso gratefully acknowledged.

    1. Introduction

    Much of economic analysis stems from thejoint assumptions of rationality and indi-vidual greed. Common sense and experimen-tal and field evidence point to the limits of thisapproach. Not everything of interest to econ-omists can be well understood using thesetools. This paper reviews evidence that narrowconceptions of greed and rationality performbadly. The evidence is consistent with the viewthat economic incentives influence decisionmaking. Hence there is a role for optimizingmodels that relax the assumption of individualgreed. I discuss different ways in which onecan expand the notion of preferences.

    I pay particular attention to how reciproc-ity influences decision making. Reciprocityrefers to a tendency to respond to perceivedkindness with kindness and perceived mean-ness with meanness and to expect thisbehavior from others. I introduce models ofintrinsic reciprocity in section 3.4.Intrinsic reciprocity is a property of prefer-ences. The theory permits individual prefer-ences to depend on the consumption ofothers. Moreover, the rate at which a personvalues the consumption of others dependson the past and anticipated actions of others.An individual whose preferences reflectintrinsic reciprocity will be willing to sacri-fice his own material consumption toincrease the material consumption of othersin response to kind behavior while, at thesame time, be willing to sacrifice materialconsumption to decrease someone elsesmaterial consumption in response to unkindbehavior.

    It is more traditional to view reciprocity asthe result of optimizing actions of selfishagents. Responding to kindness with kind-ness in order to sustain a profitable long-term relationship or to obtain a (profitable)reputation for being a reliable associate areexamples of instrumental reciprocity.Economics typically describes instrumentalreciprocity using models of reputation and

    ju05_Article 2 6/10/05 1:40 PM Page 392

  • Sobel: Interdependent Preferences and Reciprocity 393

    repeated interaction. This approach is quitepowerful as essentially all exchanges in natu-ral settings can be viewed as part of somelong-term interaction. Consequently onecould argue that the models of section 3.4are unnecessary. This essay presents thecounterargument that models of intrinsicreciprocity can provide clearer and moreintuitive explanations of interesting econom-ic phenomena. An openness to the possibili-ty of intrinsic reciprocity leads to a new anduseful perspective on important problems.

    The next section contains a stylized exam-ple that illustrates the limitations of standardmodels. I use the example to provide aninformal introduction to alternative theoret-ical approaches and motivate the paper.Section 3 introduces these models formally.Section 4 describes some economic settingsin which the modeling approaches of section3 may be particularly useful. Section 5reviews literature on the evolution of prefer-ences. Section 6 responds to stylized argu-ments against the approach and section 7 isa conclusion.

    2. An Informal Guide to the Concepts

    We regularly read accounts of dissatisfiedor recently fired employees destroying prop-erty, sabotaging computer files, or evengoing postal and killing people at theirworkplace. The sense of outrage at an appar-ent injustice is real. Many people are willingto take destructive actions as part of the out-rage. This kind of destructive behavior isunlikely to be in the material interest of afired employee: it takes time, it is not com-pensated, and it carries the risk of criminalpenalties. How should we think about it?For concreteness, imagine that Paul was ahigh-ranking executive who had worked fora company for more than ten years. He losthis job when business turned bad. On his lastday on the job, Paul destroyed vital companydocuments and continued to sabotage com-puter files until he was caught six monthslater. In this section, I will use Pauls story to

    introduce and motivate the ideas I review inthe manuscript.

    2.1 Descriptions

    There are several ways to react to Paulsdestructive activity. One response is to treat itas an emotional response not subject to eco-nomic analysis. We should not give up so eas-ily. Paul may be crazy, but his former boss,Marsha, probably is not. Unless Pauls actionsare completely unrelated to the environment,Marsha will want to understand how toreduce the chance of adverse behavior.Marsha may need to hire lawyers or psychi-atric consultants (instead of economists) to tellher how to deal with Paul or reduce the risk ofcostly outbursts by employees, but she shouldevaluate her options using economic models.

    I will concentrate on descriptions that areconsistent with the hypotheses of optimiza-tion and equilibrium. Once we allow thatPaul maximizes something more that his ownmonetary reward, there are many stories likethis available. This section introduces somepotential descriptions informally. Section 3provides a more systematic treatment.

    2.1.1 Static Income Maximization

    Hypothesis. The narrowest version ofeconomic theory assumes that Paul seeks tomaximize his utility and that his utilitydepends on the quantities of the privatematerial goods he consumes. In staticincome maximization, Paul balances theimmediate cost and benefits of actionsrather than the long-term implications ofthese decisions. In simple settings, thishypothesis reduces to the assumption thatPaul maximizes his monetary income.

    Analysis. Paul would carry out hisdestructive action only if he imagined that itwould lead to direct material gain. It is hardto rationalize Pauls behavior under theseassumptions. His actions may advance hisimmediate material interests if Marsha giveshim back his job or if he receives a paymentto stop sabotaging the company, but a moreelaborate description seems necessary.

    ju05_Article 2 6/10/05 1:40 PM Page 393

  • 394 Journal of Economic Literature, Vol. XLIII (June 2005)

    2.1.2 Interdependent Preferences

    Hypothesis. Paul maximizes a utilityfunction that depends on Marshas consump-tion of material goods in addition to his own.

    Analysis. If Pauls utility is decreasing inthe material wealth of Marsha, then Paul willbe willing to sacrifice his own material wellbeing to punish Marsha. These preferencesexplain why Paul would wish to harmMarsha, but do not explain why he waitsuntil after he is fired to do so. There are twopossibilities. In the midst of an ongoingemployment relationship, Paul does notharm Marsha because he fears that Marshawill fire him, which would be a sufficientlygreat punishment to deter him from hurtingMarsha.

    Alternatively, the marginal rate of substi-tution between Paul and Marshas materialincome in Pauls preferences may change asa result of Pauls termination. The impover-ished Paul is willing to sacrifice to makeMarsha worse off. This explanation onlyworks if Pauls income after being fired islower than after a voluntary separation (oth-erwise any separation would trigger Paulsdisruptive behavior).

    For the example, it makes sense to assumethat Pauls utility is decreasing in Marthasincome. The interdependent preferenceapproach permits Paul to be willing to sacri-fice material welfare to decrease the incomeof others.

    2.1.3 Preferences over GeneralConsumption Goods

    Hypothesis. Paul maximizes a utilityfunction that is a function of consumptiongoods that are derived from marketed goodsthrough a personal production process.

    Analysis. This approach generates severalpossible stories. One possibility is that Paulsbehavior demonstrates that he has a mar-ketable characteristicthat is, he is not thetype of person who can be pushed around, heis not afraid to stand up for injustice, or he iscapable of identifying weaknesses in a firms

    security. By hurting Marsha, Paul gainsbecause he positions himself to get anotherjob (which he may be less likely to lose) or sella book about his experiences. Under thesecircumstances, Paul may have preferencesdefined over both his monetary wealth andhis sense of honor. If the preferences areincreasing in both arguments, then he will bewilling to make material sacrifices in order toincrease in honor. If Paul only cares abouthonor because it enables him to increase hismonetary payoff, then this explanationreduces to income maximization.

    Another possibility is that Paul takes pleas-ure directly from the act of sabotage. That is,his preferences contain an additional argu-ment (sabotage). Paul will not maximizehis material payoff, but he is selfish and goaloriented. This explanation does not explainwhy Paul turns to sabotage only after he wasfired. Perhaps the advantage of maintainingthe employment relationship deterred hisurge to destroy files until he was fired, butthis explanation suggests that the

View more