Copyrights to papers in the Duke Economics Working Paper Archive remain with the authors or their assignees. Archive users may download papers and produce them for their own personal use, but downloading of papers for any other activity, including reposting to other electronic bulletin boards or archives, may not be done without the written consent of the authors.

Duke Economics Working Paper #00-16

Finite Mixture Distributions, Sequential Likelihood and the EM Algorithm


Peter Arcidiacono
and
John B. Jones

Abstract

The use of finite mixture distributions to control for unobserved heterogeneity has become increasingly popular among those estimating dynamic discrete choice models. One of the barriers to using mixture models is that parameters that could previously be estimated in stages must now be estimated jointly: using mixture distributions destroys any additive separability of the log likelihood function. The EM algorithm reintroduces additive separability, however, thus allowing the option of estimating parameters sequentially during each maximization step. We show that, relative to full information maximum likelihood, the EM algorithm with sequential maximization (ESM) can generate large computational savings with little loss of efficiency.

Keywords: Unobserved heterogeneity, mixture distributions, EM algorithm, dynamic discrete choice

JEL: C13, C61, D90

Published in Econometrica, Vol. 71, No. 3, May 2003, pp. 933-946.