Preprint C147/2012
On Asymptotic Behavior of Economies with Complete Markets: the role of ambiguity aversion
Carlo Pietro Souza da Silva
Keywords: market selection | Knightian uncertainty
The aim of this work is to analyze some equilibrium consequences of the behavior of heterogeneous ambiguity averse agents in inter-temporal general equilibrium models. As the focus is on the influence of ambiguity aversion, the choice of the models treated here was based on their previous use to attain similar results, allowing their use as a parameter. Because of that, it is considered here a general equilibrium model which fits in frameworks like in Araujo and Sandroni (1999), Sandroni (2000), Blume and Easley (2006) and Condie (2008). Two different approaches have been used: in the first (Chapter 2) we strive to find out what condition of beliefs is necessary to achieve equilibrium, and in the second (Chapter 3) we want to know what conditions are related to survival. There is also a chapter of preliminaries where the framework is discussed and some auxiliary results are presented. The first results are within the framework of Araujo and Sandroni (1999) where there is a complete market of contingent claims and bankruptcy is permitted, though incurring a penalty. Agents have the smooth ambiguity preferences presented by Klibanoff et al. (2005, 2009). The main result follows those others presented in the literature, but provides quite a different interpretation. It proves that a necessary condition for equilibrium existence is the convergence of ambiguity perception reduction. Other results are placed within the context of Blume and Easley (2006), where be- havior is analyzed from the point of view of Pareto Optimal allocations. In this case, agents´┐Ż behavior is determined by variational preferences (axiomatized by Macheroni et al. (2006a,b)). These preferences are more general than the expected utility (used by Sandroni (2000) and Blume and Easley (2006)) and maxmin utility (used by Condie (2008)). The main result contrasts with the result derived by Condie (2008). Under conditions in which a maxmin agent does not survive, another ambiguity averse agent can survive.