Search

In the recent years, there has been an impressive advance in the study of models in which (boundedly rational) economic agents adjust their behavior over time, in reaction to the additional information they acquire as time progresses. Some of these models involve the use of autonomous continuous-time dynamical systems, known in the literature as selection dynamics, in which the adjustment process is directly linked to the relative performance of each strategy at each given point in time. The aim of the thesis is to contribute to this line of research exploring the formal properties of these dynamics, as well as modeling suitable economic environments in which these dynamics can be applied. In particular, the thesis explores the behavior of selection dynamics in three different game-theoretic frameworks: (game-form) mechanisms, extensive form games and games with pre-play communication. A (game-form) mechanism is a game whose equilibria satisfies certain desirable properties but which does not necessitate vast amount of knowledge by the authorities ("the planner") to put it in place. Instead, this social arrangement should basically self police itself, and the planner should only make sure that the rules of the game are correctly followed by the agents. An extensive form game is a game in which players move sequentially, and make use of the information they acquire as the game proceeds to improve their performance. A game with pre-play communication is a game which is preceded by a stage in which the agents are allowed to send costless signals to their opponent, in order to influence their behavior. In all these situations, the analysis of the learning dynamics we described leads to conclusions which contradict traditional game theoretic analysis, but seem to suit more closely the empirical and experimental evidence in the field.