Robert Shimer (Chicago)
and Lones Smith (Michigan)
This paper explores optimal matching policies
in constant returns to scale search economies with heterogeneous agents. We
look for a policy that maximizes the present value of output in the economy,
taking the search frictions as given. Our main result is that if agents'
characteristics are complements in production, an optimal policy may be nonstationary, with a nontrivial asymptotic limit cycle or
possibly chaotic behavior, even though the model has no extrinsic uncertainty.
This finding holds for a generic set of parameter values. It is due to a
trading externality (Diamond 1982) that is inherent in heterogeneous agent
economies. The same forces also generate a continuum of perfect foresight
equilibria.