Springer Online Journal Archives 1860-2000
Summary Welfare maximizing models with infinite horizon allow us to obtain time consistent, perfect foresight allocations. However, their numerical implementation requires finite horizon models. In this paper we study the steady-state properties of such infinite horizon allocations as well as the appropriateness of specific finite horizon approximations. In particular, we compare the single agent, single capital good model by Cass (1965) which is cast in continuous time, with the multiagent discrete time formulation by Lucas and Stokey (1984) and with an overlapping generation version which follows Kehoe and Levine (1985). For a modification of the Lucas and Stokey model which has satiation in utility, a technique to derive error bounds is obtained for finite horizon approximation of a model with at most one capital good. This technique also applies to appropriate respecification of the Cass model and, if one is willing to accept some heuristics, to the cases with many capital goods and with overlapping generations.
Type of Medium: