Welfare Economics of Decision Making with Changing Preferences

Abstract

paper considers the properties a decision tree needs to guarantee that there are no preferences such that the sophisticated equilibrium is Pareto dominated when all agents act completely myopically. It shows that the only decision trees with this property are the most trivial problems. analysis is generalized to agents who are not completely myopic and gives an example of union and shareholder conflict to illustrate the results. 1. This paper is concerned with the welfare aspects of decision making in a non-cooperative game where the players have different preferences over future actions and must move sequentially. Typically two outcomes are analysed in this context. One is the naive or myopic outcome which is the result of all individuals making irrational myopic choices and the other is the sophisticated or perfect equilibrium.1 results of Hammond (1976) show that if there is no essential inconsistency2 between agents preferences then these two outcomes coincide. welfare comparison when there is essential inconsistency remains an open question and this paper will attempt to provide some answers. Although this issue has not been explicitly considered the terminology perfect and sophisticated compared to myopic or naive is itself persuasive. literature has tended to enhance this view, e.g. Blackorby et al (1973): The sophisticated solution would appear to be an improvement over naive By explicitly considering the future preferences each generation realises a level of utility greater than that which is possible under inconsistent naive planning. It will be shown that this view is incorrect (for all degrees of myopia). Of course there are many cases where the perfect equilibrium in non-cooperative games is inefficient, thus it is not surprising that actions by some agents which deviate from rational behaviour can result in a better outcome for all agents. It is less obvious that if everyone behaved myopically this is possible yet it will be shown that in all but the most trivial decision trees there always exist preferences such that this is true. following section outlines the model that will be used and Section 3 contains the results. Section 4 gives a topical economic example that reduces to the decision tree structure used extensively in Section 3. This example clarifies the structure of the results and also highlights some of the difficulties with introducing precommitment.