Deception: the Role of Consequences
Total Page：16
File Type：pdf, Size：1020Kb
Deception: The Role of Consequences By URI GNEEZY* Deception is part of many economic interac- sociated with lying per se. This assumption is tions. Business people, politicians, diplomats, very useful in many economic models. Consider lawyers, and students in the experimental labo- contract theory, where it is assumed that with- ratory who make use of private information do out an explicit contract, neither side will fulfill not always do so honestly. This observation its respective obligations. For example, George indicates that behavior often rejects the moral Akerlof’s (1970) paper on asymmetric informa- approach to deception. As St. Augustine wrote, tion and the market for lemons assumes that “To me, however, it seems certain that every lie sellers of used cars will always lie if it is in their is a sin. ” (St. Augustine, 421). Later, philos- benefit to do so. ophers like Immanuel Kant (1787) again In the mechanism design literature (e.g., adopted this uncompromising moral stance Bengt Holmstrom, 1979), the standard assump- when arguing against lying. tion is that people will tell the truth only if this At the other extreme, economic theory is is incentive-compatible given material out- built on the assumption of “homo economicus,” comes. In the literature on tax evasion, the a figure who acts selfishly and is unconcerned choice of whether to avoid paying taxes is con- about the well-being of others.1 An implication sidered a decision under uncertainty; cost is of this assumption is that lies will be told when- treated as a product of the probability of being ever it is beneficial for the liar, regardless of caught and the cost of punishment, whereas their effect on the other party.2 Another impli- benefit is simply the money saved by avoiding cation is that there is no negative outcome as- payment. However, there is no cost associated with the very act of lying (Michael Alingham and Agnar Sandmo, 1972). Another example is * Graduate School of Business, University of Chicago, the game theoretic treatment of “cheap talk” 5807 South Woodlawn Avenue, Chicago, IL 60637 (e-mail: (Crawford and Joel Sobel, 1982). uri.gneezy.gsb.uchicago.edu). I thank Douglas Bernheim An intermediate approach is taken by utili- and two anonymous reviewers for insightful comments that tarian philosophers (e.g., Jeremy Bentham, considerably improved the paper. I also thank Andreas Blume, Gary Charness, Rachel Croson, Martin Dufwen- 1789). Utilitarianism prescribes that, when berg, Georg Kirchsteiger, David Levine, Muriel Niederle, choosing whether to lie, one should weigh ben- Yuval Rottenstreich, Maurice Schweitzer, Richard Thaler, efits against harm, and happiness against unhap- George Wu, and seminar participants at numerous univer- piness. As Martin Luther stated, “What harm sities for their comments and suggestions. Ira Leybman provided valuable help in running the experiment. I became would it do, if a man told a good strong lie for interested in deception when my father was terminally ill the sake of the good and for the Christian and his physicians created in him a belief that they consid- church. a lie out of necessity, a useful lie, a ered to be untrue. I dedicate this paper to his memory. helpful lie, such lies would not be against God, 1 Important deviations from this assumption in economic he would accept them.”3 Similarly to the eco- modeling are found in Kenneth Arrow’s (1972) discussion of trust, Gary Becker’s (1976) modeling of altruistic pref- nomic theory approach, this type of calculation erences, and Akerlof’s (1982) study of the fair-wage hy- implies that lies, apart from their resultant harm pothesis. For a general discussion, see Becker (1993): “The and benefit, are in themselves neutral. A lie and economic approach I refer to does not assume that individ- a truthful statement that achieve the same mon- uals are motivated solely by selfishness or material gain. It is a method of analysis, not an assumption about particular etary payoffs (for both sides) are considered motivations. Along with others, I have tried to pry econo- mists away from narrow assumptions about self-interest. Behavior is driven by a much richer set of values and (1982) discussion of reputation and imperfect information; preferences” (p. 385). see also Vincent P. Crawford (2003). 2 Note that this does not mean that a completely selfish 3 Cited by his secretary, in a letter in Max Lenz, ed., person will always lie. There may be strategic reasons not to Briefwechsel Landgraf Phillips des Grossmuthigen von lie. For example, see the David Kreps and Robert Wilson Hessen mit Bucer, Vol. 1. 384 VOL. 95 NO. 1 GNEEZY: DECEPTION: THE ROLE OF CONSEQUENCES 385 equivalent (Sissela Bok, 1978, Ch. 4). This is a which lying increases the payoffs to the liar at consequentialist approach. An alternative ap- the expense of her counterpart, and I asked the proach—one that distinguishes between two following question: How do changes in relative decisions with the same payoff set according to payoffs influence this decision? The main em- the process leading to the outcomes—would be pirical finding is that people not only care about called nonconsequentialist. their own gain from lying; they also are sensi- St. Augustine’s approach is normative in the tive to the harm that lying may cause the other sense that it prescribes “what a person should side. The average person prefers not to lie, when do.”4 His injunction is (unfortunately?) not sup- doing so only increases her payoff a little but ported by a casual observation of real life: peo- reduces the other’s payoff a great deal. ple do lie. Economic theory is normative in the sense that it prescribes “what a rational eco- I. Classification of Lies, and a Definition nomic agent should do.” This approach is also not supported by casual observation: even econ- It is interesting to note that the literature omists tell the truth from time to time, in the offers many ways to classify lies.6 I base my absence of any strategic justification for doing classification on the consequences that the lie so. The utilitarian approach predicts that if peo- produces. Using this criterion, one can devise ple do care about the well-being of others, the four major categories. First, there are lies that decision to lie (or not) may depend on its cost to help both sides, or at least do not harm anyone, the other side. As a result, people will not go to for instance, a white lie that costs the liar noth- either extreme of always lying or always telling ing and makes the counterpart feel good (“You the truth. As I show below, however, people look great today!”). In the second category I do distinguish between lying and “innocent” place lies that help the other person even if it choices, even when the decisions do not differ harms the liar. The motivation for this kind of in monetary outcomes. In particular, people are lie may be pure altruism (Becker, 1976), an less likely to choose the outcome that maxi- impure motive according to which people enjoy mizes their own monetary payoff if it involves a the act of giving (James Andreoni, 1990), or an lie than if it involves an innocent choice. Hence, “efficiency motive,” according to which people the consequentialist assumption of utilitarian- prefer outcomes that enlarge total surplus (Gary ism is rejected. Charness and Matthew Rabin, 2002). In the I empirically studied the role of conse- third category are lies that do not help the liar quences in the decision concerning whether to but can harm both sides or, at the very least, the lie.5 I considered a two-person interaction in other person. The motive for this might be a spiteful reaction to unfair behavior. The fourth category I consider in this paper includes lies that increase the payoff to the liar 4 Although Augustine was categorically against lies, he distinguished between different types of lies. The continu- and decrease the payoff to the other party. I ation of the citation reads, “. though it makes a great argue that this is the relevant category for many difference with what intention and on what subject one economic events, such as those covered by lies.” Similarly, although Jewish texts prohibit lying, certain mechanism design and contract theory. If per- lies, especially those told to preserve household unity, are regarded as exceptions (Lewis Jacobs, 1960). son A signs a contract with person B, it is 5 Many other aspects of deception are studied in the simply to prevent B from acting in ways that literature. Psychologists study personality characteristics of honesty (Hugh Hartshorne and Mark May, 1928), how to detect lies (Paul Ekman, 1992; Albert Vrij, 2001), etc. See Charles Ford (1995) for an introduction to the psychology Hertwig (2002) compare the costs and benefits of deceiving of deceit, Roy Lewicki (1983) for a behavioral model, and participants in laboratory experiments, concluding that ex- Bella DePaulo et al. (1996) for taxonomy of lies and their perimental economists’ prohibition of deception is sensible. classifications according to content, motivation, and mag- In business, research focuses on deception in negotiations nitude. In accounting, John Evans et al. (2001) examine how (e.g., Maurice Schweitzer and Rachel Croson, 1999). See preferences for wealth and honesty affect managerial report- Bok (1978, particularly chapter 4) for what I find the most ing (see also the discussions by Stanley Baiman and Barry thoughtful modern treatment on the morality of deception. Lewis, 1989; Kenneth Koford and Mark Penno, 1992). In 6 See Lewicki (1983), DePaulo et al. (1996), and experimental economics, Andreas Ortmann and Ralph Schweitzer (2001).