اکتشاف مجازات شخص ثالث و دوم در بازی های ساده ده نفره

This paper explores the motivations behind punishment from unaffected third parties and affected second parties using a within-subjects design in ten simple games. We apply a classification analysis and find that a parsimonious model assuming that subjects are either inequity averse or selfish best explains the pattern of punishment from both third and second parties. Despite their unaffected position, we find that many third parties do not punish in an impartial or normative manner.

Third parties play a crucial role in many institutions: They serve in courts, as referees or arbitrators.1 The US legal system, for instance, relies on the judgment by juries when it comes to the application of sanctions. Third parties are also important with regard to informal sanctions (Homans, 1961) and, in fact, their interventions seem to be essential in the explanation of norm enforcement, as they are often more numerous than affected second parties (Bendor and Swistak, 2001) or the only parties present and hence their sanctions are potentially more damaging than those from second parties. Despite their importance, there exist several uncertainties regarding how third parties sanction others. In particular, it is not clear whether third parties sanction in a different manner than second parties. On the one hand, third parties might sanction in a more impartial, “normative”, and controlled manner, and less egocentrically (Fehr and Fischbacher, 2004). Adam Smith apparently had this idea in mind when he introduced the concept of the “impartial spectator” in his Theory of Moral Sentiments, a party who is not personally affected, making decisions from beyond the limitations of egocentric biases. In this line, the prevalence of institutions (like juries) that rely on third parties seems in accordance with the idea that third parties make more appropriate decisions. On the other hand, it also seems plausible that even third parties cannot completely eliminate egocentric biases ( Ross et al., 1977 and Babcock et al., 1995). In fact, the concerns about the selection of jury members in many law cases suggest that third parties can make very inappropriate decisions in the context of sanctioning (e.g. Kennedy, 1997). In this paper we report the data from two experimental treatments, one on second party punishment, and the other on third party punishment. The two treatments include ten games which allow us to study and compare the motivations behind third and second party punishment. For this, we apply the classification method by El-Gamal and Grether (1995) and categorize subjects according to the theory that best explains their behavior in all ten games. By comparing the classification results for each treatment, we get some insights into the motivations of second and third parties. Our results also provide a picture of heterogeneity in agents’ other-regarding preferences, another factor in which second and third parties might differ (e.g. the fraction of subjects who sanction in a normative, impartial manner could be different between third and second parties). We consider many possible theories/motivations for third and/or second party punishment, including those advanced by recent theories of other-regarding preferences. Thus, theories of inequity-aversion like Fehr and Schmidt (1999) predict punishment of richer co-players if that reduces the payoff distance, while pure reciprocity theories (Rabin, 1993, Dufwenberg and Kirchsteiger, 2004 and Cox et al., 2007) are based on the idea that people harm those who harmed them. Further, Bolton and Ockenfels (2000) predict punishment of any co-player if that brings the aggressor's relative payoff closer to the average relative payoff, Levine (1998) posits the existence of spiteful types who punish indiscriminately and type-reciprocal agents who punish selfish or spiteful co-players, Falk and Fischbacher (2006) combine ideas from inequity-aversion and reciprocity, and López-Pérez (2008) predicts punishment of norm deviators. Our study provides several insights about the differences and similarities in motivations behind third and second party punishment. To start, the classification analysis reveals two key non-selfish motivations behind the occurrence of both third and second party punishment. These motivations are two types of inequity-aversion for the third parties, and inequity-aversion and spitefulness for the second parties. If parsimony is the main goal, however, the Akaike Information Criterion suggests that a model including both inequity-averse and selfish agents (who never punish) is sufficient to explain both second and third party punishment. That is, in our games both types of punishment are predominantly targeted towards richer co-players, and as a result we do not find sharp differences in the motivations of both parties. We also observe that the strength of punishment depends heavily on the size of the payoff disadvantage for second parties and even more so for third parties. Interestingly, our data shows that third parties often punish as intensely as second parties, which contrasts with previous findings (Fehr and Fischbacher, 2004), and that most third parties do not punish in a more normative and impartial manner than second parties. Therefore, our results suggest that second and third party punishment have more similarities than differences and show that models that include inequity-aversion as a motivation (specifically Fehr and Schmidt, 1999 and Falk and Fischbacher, 2006) explain best the occurrence and strength of both types of punishment, whereas motivations like pure reciprocity or spite play a relatively minor role. 2 The rest of the paper proceeds as follows. The next section compares our study with the related literature. Section 3 presents the experimental design and procedure. Section 4 reports the results from the classification analysis and analyzes which factors affect the occurrence and strength of second and third party punishment. The fifth section concludes.

We investigate third and second party punishment in a set of ten different games to compare the individual motivations behind both types of punishment and to provide insights into the different existing theoretical approaches. The results suggest that inequity aversion is a crucial (although by no means the only) cause of third and second party punishment. This evidence has implications for the different theoretical models. To start, models that incorporate inequity aversion like Fehr and Schmidt (1999) and Falk and Fischbacher (2006) fare relatively better in explaining the occurrence of punishment in 3P and give also rather good predictions in 2P. These models (especially Falk and Fischbacher, 2006) are also more accurate in predicting the strength of punishment. Pure reciprocity models like Dufwenberg and Kirchsteiger (2004), and Cox et al. (2007) are less accurate, especially with regard to third party punishment. Levine (1998) is inconsistent with the heavy punishment of socially and Pareto efficient actions, and with the role that strict equality plays in reducing punishment (pure reciprocity also faces this problem). In turn, norm approaches like López-Pérez (2008) face an unanticipated problem in 2P and 3P: There seems to be no way to explain punishers’ choices as a reaction to a prior deviation from any sensible norm of distributive justice (taking standard concepts like social efficiency, equity, or maximin into account). A clear illustration of this is that both allocations are punished in some games or that by-standers are damaged by third parties. Under one possible interpretation of the term, in fact, our data suggests that third party punishment is not particularly “normative”. Yet this is not to say that norms are unimportant in explaining punishment, as many third and second parties (even if inequity-averse) might rationalize their punishment in normative terms, as the classical philosopher Seneca noted: “Reason wishes the decision that it gives to be just; anger wishes to have the decision which it has given seem the just decision”. People might not punish normatively, but they are likely to believe that they do so. Experimental procedures and discussion: The experiment was conducted at the Universidad Autónoma de Madrid, and the participants were students from different disciplines (60 subjects participated in the 2P and 75 subjects in the 3P treatment).26 We chose game 1 (150, 150 vs. 590, 60) because, given our results from the 2P and the 3P treatments, we expected a large amount of punishment, and also because third and second parties punished the allocation (590, 60) equally strongly when we used the strategy method. In this respect, our data shows that second and third parties do not punish the choice of the allocation (590, 60) significantly stronger in this experiment with specific response method than in our treatments 2P and 3P with the strategy method. Indeed, in this “hotter” environment (because of the applied specific response method and maybe the location), third parties punish the allocation (590, 60) significantly stronger and not weaker than second parties (z = 3.596, p < 0.01).