An Ambiguity Analysis under Heterogeneity with a Bayesian Spin

We often have to deal with uncertainty regarding multiple aspects of the decision problems we face. This uncertainty may concern, for instance, our earnings, the likelihood to receive them in a given moment and in a given amount. The aim of this thesis is to contribute to the growing body of literature around “multi-dimensional uncertainty”, which enlarges the scope of ambiguity outside the frame of uncertainty about probabilities. It does so by analysing, both theoretically and empirically the evidence stemming from a multi-stage experiment in which subjects have to choose between lotteries whereby amounts of monetary prizes are not always known, whereas probabilities are always public knowledge. In the experiment, three different levels of information over some monetary prizes are randomized between subjects. The experimental evidence undergoes structural estimation exercises: these elicit the individuals' degree of risk aversion within the frame of a standard constant relative risk aversion (CRRA) utility function. Furthermore, we investigate  whether a change of information, such as the one we reproduce through the different treatments conditions, translates into a change in behavior and, in turn, whether and how much this change translates into a significant change in their measured (CRRA) attitude toward risk. As to the behavioral content of the structural model for the uncertain payoffs, we propose two alternative specifications, labelled “naïve” and “sophisticated”. The empirical evidence shows a moderate but significant degree of love for ambiguity, since less information given to subjects results in a lower estimate of their risk aversion, and, as a consequence, in a stronger attraction toward risk and uncertainty. A mixture model is implemented to identify the probability of individuals mirroring one behavioral model or the other, or, saying it differently, the percentage of observations compatible with either model.  We conclude that our subjects have a strong tendency to behave as naïve. 
Keywords: heterogeneity; risk aversion; ambiguity