Dynamic Programming Deconstructed

Qingyin Ma, John Stachurski

Published 2018-11-05Version 1

A common strategy in economic modeling is to try to manipulate Bellman equations into advantageous forms without breaking their link to optimality. In this paper we provide a theoretical framework and a set of results that transform this art into a science. In particular, we (a) clarify the link between such manipulations and optimality of the resulting policies, including when this link breaks down, (b) establish a connection between contractivity of the modified Bellman operators and Bellman's principle of optimality, (c) use manipulations of the Bellman equation to extend the set of algorithms for obtaining optimal policies, (d) extend the set of models that can be transformed to include recursive preferences and other forms of nonseparability, (e) find new applications of these manipulations and use them to obtain new results, and (f) use these methods to simplify the Bellman equation in a range of recent applications.