Dynamic Optimization with a Non-Convex Technology: The Case of a Linear Objective Function

The paper studies the problem of optimal intertemporal allocation in an aggregative model with a non-convex technology set and a discounted sum of consumptions as the objective function. The study demonstrates the existence of a threshold initial stock such that the long-run behaviour of optimal programmes depends critically on whether the initial stock is, above or below the threshold. This is in contrast with the standard turnpike theory of convex models in which the long-run behaviour of optimal programmes is independent of the initial stock.