Computing business-as-usual with a representative agent and a pollution externality

Computing the no-policy equilibrium (business-as-usual) in a representative-agent (RA) model is complicated by the presence of a pollution externality, since simple optimization internalizes the pollution cost. Many researchers use ad-hoc methods, but there is no way to know how reliable these are. A solution is presented in which the RA model is divided into N identical components, each identified with its own sub-agent. Sub-agents play a dynamic game, leading to a Nash equilibrium. For sufficiently large N, this approach keeps most of the pollution cost external, and in the limit it is equivalent to a myopic-firms model, in which the entire cost is external. This approach has the advantage of theoretical consistency, and empirical applications indicate that it is easily implemented.