A generalized endogenous grid method for non-smooth and non-concave problems

This paper extends Carroll's (2006) endogenous grid method and its combination with value function iteration by Barillas and Fernandez-Villaverde (2007) to a class of dynamic programming problems, such as problems with both discrete and continuous choices, in which the value function is non-smooth and non-concave. <P> The method is illustrated using a consumer problem in which the consumer chooses both durable and non-durable consumption subject to a borrowing constraint. The durable choice is discrete and subject to non-convex adjustment costs. <P> The algorithm yields substantial gains in accuracy and computational time relative to value function iteration, the standard solution choice for problems in which the value function is non-smooth or non-concave. (Copyright: Elsevier)