In this paper we consider Bayesian inference techniques for Agent-Based (AB) models, as an alternative to simulated minimum distance. MCMC techniques organized around kernel-based estimation solve the problem of selecting the right moment conditions to implement statistical inference. We discuss the specificities of AB models with respect to models with exact aggregation results (as DSGE models), and how this impact estimation. We suggest the use of unconditional density estimation and show the feasibility and good performance of these techniques in a price discovery model and a model of innovation diffusion.
Authors:
Mike Tsionas (Lancaster University Management School, Lancaseter, UK)
Jakob Grazzini (Catholic University of Milan, Department of Economics and Finance, Milano, Italy)
Matteo Richiardi (Institute of New Economic Thinking and Nuffield College, Oxford, UK)