Institutional approaches to social dilemmas have so far focused on how to create incentive structures that channel individuals' behavior into socially desirable outcomes assuming that everyone is selfish. However, the presumption of universal selfishness is not only empirically invalid but may also result in inefficient policy prescriptions in the short run and a culture of distrust in the long run. The game theoretic models that depart from the universal self-interest assumption, on the other hand, have not been successful in incorporating both individual rationality and inter-individual heterogeneity.; This dissertation develops game theoretic models of social dilemmas with individuals who are rational in the sense that they have preferences and try to maximize expected utility but heterogeneous in the sense that they have different preferences over possible social outcomes. This dissertation tests the implications of the models using a series of experimental data sets. The empirical results are: (1) there is a significant proportion of individuals who are rational but not selfish, (2) non-selfish individuals' motivations are best described on a dimension of equity/fairness rather than being thought of as unconditional altruism, and (3) the possibility of sustained mutual cooperation in finitely repeated social dilemmas is affected by the material conditions of the action situation, the composition of types within the population, and individuals' willingness to take risks to initiate coordination.
展开▼
