Repeated games Some applications to under-cutting and harassment in the labour market

Abstract

This paper consists of some simple applications of the complete information repeated game (RG) framework to the labour market.’ In many economic (and social) problems, agents participate in an environment for many periods. In such contexts one often observes some degree of collusion (obeying behavioural norms which are not in the agents’ immediate interests) among agents. One explanation of such collusive activities is: agents do not deviate from the norm because of the fear of punishment by other players that the deviation might invoke. The literature on RGs has tried to model the above explanation of collusive activities within a game-theoretic framework. The result of the research has been that with the introduction of time, many types of activities (which may at first seem irrational in terms of ‘selfish’ utility maximisation) such as co-operation, altruism, threats, revenge, etc. can be explained as solutions to some RGs. It can be argued that the RG framework has been too successful; it can explain too many behavioural rules. In fact the Folk Theorem of repeated games [see Sabourian (1988b) for a survey] states that any individually rational payoff of the one-shot game can be supported as a Nash equilibrium (NE), in fact as a subgame perfect equilibrium (SPE), of the RG if players do not discount the future. The reason for the existence of so many equilibria in RGs is that in such games a strategy is a plan of what to do at every contingency (history). Many norms of behaviour which are not in players’ immediate interest can be supported as equilibria if players can choose history-dependent strategies which punish any player who does not adhere to the norm. But why should players, in equilibrium, choose a history-dependent strategy? This is even more surpris-