Uncertainty, information acquisition and economic equilibria

Chapter one introduces the thesis, and the relationships between the different chapters. The second chapter presents a model of macroeconomic co-ordination failures based on local interaction where firms have some market power. The economy exhibits a multiplicity of Pareto-ranked equilibria. We show that the introduction of uncertainty about the competitive advantage of firms generates an endogenous equilibrium selection process. This suggests a possible solution for the multiplicity problem in macroeconomic models ofco-ordination. The third chapter introduces the notion of limited attention to economic modelling. Games are drawn from a large set, and players choose how much information about the game they wish to gather. In addition we assume that more information is expensive. Information can either be acquired secretly or publicly. We solve the model for both cases and identify the conditions under which the outcomes of the two models differ. Chapter four is an application of the ideas studied in the third chapter. Here, we consider the problem of firms having to decide whether to enter a new market or not. Before they take that decision, they have the opportunity to buy information about several variables that might affect the profitability of this market. Our model differs from the existing literature on endogenous information acquisition in two respects: (1) there is uncertainty about more than one variable, and (2) information is acquired secretly. When the cost of acquiring information is small, entry decisions will be made as if there was perfect information. Equilibria where each firm acquires only a small amount of information are more robust than the socially undesirable equilibria where all firms gather all information. Examples illustrate the importance of assumptions (1) and (2). The last chapter is somewhat different from the previous ones. In the 1950s and 1960s probability learning experiments showed that people adopt probability matching strategies in repeated choice situations, rather than strategies that maximise expected utility. These findings were compatible with the predictions of stochastic learning theories like Estes' and Bush and Mosteller's. As economists' interest in experiments and learning theories grows, these findings are of relevance today. We survey the known results and assess their relationship with different theories. We look at the many results gathered for the special case of a repeated, binary choice decision experiments. We show it is unlikely that probability matching can serve as the sole prediction of asymptotic behaviour, but that any theory of behaviour in repeated decision situations should include it in its predictions