Lying aversion and the size of the lie

摘要

Although lying is common, compelling real-world and laboratory evidence shows that people sometimes avoid telling lies that would increase their material payoffs. The authors argue that there are intrinsic costs associated with lying and provide theory and evidence about the form of these costs. The intrinsic cost of lying depends on the size of the lie. Three different ways to measure this size are identified 1. The payoff dimension: monetary gains of lying, 2. The outcome dimension: the distance between what the agent observes and what she says, and 3. The likelihood dimension: the ex ante probability that the agent's report is true. A basic model is presented that generates predictions regarding how varying the size of the lie on each of these three dimensions affects behavior. The model shows the equilibrium involves a cutoff value. If the agent draws an outcome above the cutoff, she never lies. If she draws an outcome below the cutoff, she may lie and, if she lies, she makes a claim above the cutoff. However, in equilibrium, dishonest claims of the maximal value arise with positive probability. Further experiments that manipulate three characteristics of the game were conducted. The results showed that people lie and a large fraction of those who lie report the maximum lie. In addition, people report more partial lies when no one observes their outcomes than when outcomes can be observed by the experimenter ex post. It is also found that participants tell more partial lies when the ex ante probability of the highest state decreases. Finally, with respect to the outcome dimension, it was found that the fraction of dishonest reports does not depend on how outcomes are labeled, but the labels influence the frequency of partial lies (32 refs.).