I develop an algorithm for solving dynamic models in which individual decision rules and the cross-sectional distribution of agents' characteristics influence each other. To illustrate the algorithm, I solve an endowment economy with incomplete markets, a continuum of heterogeneous agents, and aggregate shocks. The key innovation of the algorithm is to parameterize the (cross-sectional) density with a flexible functional form, which makes it possible to avoid simulation techniques. The paper shows how to check for accuracy and establishes links between the properties of the incomplete-markets economy and the aspects involved in obtaining a numerical solution.