Kantians, selfish and "nice" agents: some implications for normative public policy.

Public choice theory postulate that agents in a non-market setting are instrumentally rational and selfish. However, this approach creates some problems related to the raison d'etre of the normative public choice research programme. This paper considers interaction in two different environments, of instrumental rational agents and moral agents behaving according to kantian practical reason. The first environment, is a step-level contribution to a pure public good in repeated single-shot prisoners' dilemma (PD) games where agents learn the total contribution after each round. The second is infinitely repeated PID games, where agents are randomly selected from the same group of individuals. An evolutionary analysis shows that kantians, although not playing the games, may have an important influence on the evolutionary stability of the tit-for-tat Nash equilibrium.