Multiple solutions under quasi-exponential discounting (Q1014330)

Dynamic decision problems with infinite horizon and discrete time are considered. The Markov game with one-to-one correspondence between decision makers and time periods is used for modeling such problems. The multiplicity of solutions with distinct payoffs is studied. It has been shown that such multiplicity is generic in the sense that it occurs in an open set of such decision problems, even with small state- and action-spaces. A sufficient condition for uniqueness of subgame-perfect equilibria in the special case of a single state is provided. Some examples are presented to illustrate how non-exponential discounting allows for an ``addictive'' equilibrium alongside a ``virtuous'' equilibrium.