Croson and Buchan (1999) use a trust game to investigate how gender interacts with cultures to influence individual behavior. The results are similar across cultures: no significant effect of gender is found on the amount send by the proposers, but women reciprocate significantly more than men. Beside cross-cultural studies, many other experimental studies emerging in recent years have shed light on how social norms function. Bicchieri and Xiao (2008) designed an experiment to test Bicchieri (2005)’s theory and test the relative importance of normative expectations and empirical expectations. In their dictator game, they tried to manipulate the subjects’ empirical expectations and normative expectations by showing them dictators’ decision or/and expectations in previous sessions. The expectation elicitation was made incentive compatible by rewarding the dictators based on the accuracy of the expectations they reported. The results show that subjects’ choice are influenced by both normative and empirical expectations, and when there is a conflict between the two expectations, empirical expectations are more important in predicting individual behavior.
Dana et al. (2006) show that people take into account other’s expectations while making decisions. They added one twist to the standard dictator game: the dictators can pay \$1 to exit the game with the advantage that the receiver will never know that the game has been played. About one third of dictators decided to quietly exit the game. In another private game setting where the receiver never knows about the game or where the money received is from, almost not exit was chosen. This implies that some dictators choose to give not because they care for fairness or others’ well-being, but simply because they don’t want to violate others’ expectations.
In another experiment, Dana et al. (2007) used binary dictator game to lend evidence to a similar argument: people behave generously mainly because they dislike appearing unfair to others. In their baseline treatment, the dictator can either choose A, which results in a self-interested allocation (\$6, \$1), or choose B, which leads to a fair allocation B (\$5, \$5). For the other three treatments, the level of transparency is reduced using different designs. In the hidden information treatment, the dictator can again choose A to receive $6 or B to receive $5, but the payoffs for the receiver are uncertain ex ante: with probability 0.5 the payoffs are the same as the baseline treatment (\$1 and \$5 respectively), and with probability 0.5 the payoffs for the receiver flip (\$5 and \$1 respectively). The dictator can choose to reveal the real payoffs or not. Many subjects chose not to reveal the really payoffs and took the selfish action. In their multiple dictator treatment, two subjects play the role of dictator and only when both of them choose A the allocation is inequitable (\$6, \$6, \$1), otherwise it will be a fair allocation (\$5,\$5,\$5). In their final treatment, the plausible deniability treatment, a "cutoff” feature is added to the baseline game: the computer will randomly choose an allocation if the dictator does not choose an action by the end of the cutoff period. Hence the receiver in this game would never know whether the allocation is chosen by the dictator or by the computer. The proportion implementing fair allocation in the three treatments that relax transparency, as the experimental results show, drops significantly compared to the baseline treatment. It seems that when the subjects have the moral wiggle room, they are more likely to behave according to self interest. People are willing shun away from a social norm, even at a cost, in order to make “justified” self-interested decision.
The above experiments have clearly shown that social norms play an important role in individual decision making and the expectations of other people matter. Current theories on prosocial preferences, such as inequality aversion, are inadequate in explaining these results (Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000). In the game studied by (Dana et al., 2006), it is clear that the allocation (\$9,\$1) is preferred to (\$9,\$0) according to the inequality aversion model, but many subjects chose (\$9,\$0) given that the receiver would not know the proposed ever made any decision on the allocation. Similarly, the binary dictator game with hidden information in (Dana et al., 2007), revealing the payoff information for the receiver will aid the proposed in making decisions and increase her expected utility according to the inequality aversion model, but subjects chose to be ignorant on the other party’s potential payoff. Inequality aversion is outcome-based preferences. For the same reason, social welfare concerns, which drive people to maximize the total welfare of all the players, are not enough to explain the above experimental regularities.
No comments:
Post a Comment