This paper introduces a general framework for dealing with dynamic inconsistency in the context of Markov decision problems. It carefully decouples and examines concepts that are often entwined in the literature: it distinguishes between the decision maker and its various temporal agents, and between the beliefs and intentions of the agents. Classical examples of naiveté and sophistication are modeled and contrasted based on this new language. We show that naive and sophisticated decision makers can form optimal strategies at each possible history, and provide welfare comparisons for a class of decision problems including procrastination, impulsiveness, underinvestment, binges and indulgence. The creation of a unified formalism to deal with dynamic inconsistency allows for the introduction of a hybrid decision maker, who is naive sometimes, sophisticated at others. Such a hybrid decision maker can be used to model situations where type determination is endogenous. Interestingly, the analysis of hybrid types indicates that self-deception can be optimal.
|Series||GSBE Research Memoranda|