Essays on Decision Making: Intertemporal Choice and Uncertainty

Being labeled as a social science, much of economics is about understanding human behavior; be it in the face of uncertainty or delayed payoffs through time or strategic situations such as auctions, bargaining, and so on. This thesis will be concerned with the first two, namely uncertainty and time preferences.

The main focus of this thesis is what we can summarize with two broad titles: "irrationalities" in human behavior and an alternative perspective on 'rational behavior". My claim requires a clarification of what is meant by rational or irrational behavior. In one of the early discussions of this topic, Richter (1966) defined a rational consumer as someone for whom there exists a total, reflexive, and transitive binary relation on the set of commodities so that his choice data consists of maximal elements of this binary relation. In this respect, Richter (1966) only imposed minimal consistency conditions on behavior for it to be labeled as rational. Although his setting does not involve any uncertainty or time dimension, analogues of these conditions exist for the models we consider here as well. So one can extend the rationality notion of Richter (1966) to our models too. Yet the essence of his approach to rationality is different than the one we take up in this thesis. This minimalistic approach of Richter would leave little space for discussions on rational behavior because much behavior would be rational except for a few cleverly constructed counterexamples. Instead we will consider more widely accepted norms of rationality and analyze them in the framework of uncertainty and time preferences.

The widely accepted norms of rationality mentioned above are understood to be axioms that lead to decision rules describing people's behavior. In the case of decision making under risk and uncertainty the most commonly used decision model is expected utility, and in the case of dynamic decision making, it is the constant discounted utility model. Although there are models that combine both to explain decision making in a dynamic stochastic settings, in this thesis we study them in isolation to assess the nature of the models in more detail.