Which Model to Match?
44 Pages Posted: 17 Jan 2012 Last revised: 18 Oct 2015
Date Written: October 11, 2015
The asymptotic efficiency of indirect estimation methods, such as Indirect Inference and the Efficient Method of Moments, depends on the choice of the auxiliary model, which is some- how ad hoc and based on an educated guess. We introduce a consistent simulation based Akaike-type class of information criteria that helps the user in this choice among nested and non-nested auxiliary models by selecting the model that provides the best trade-off between efficiency and estimation error. We prove consistency of the proposed criterion as the sample sizes of the observed and simulated data increase. In a Monte Carlo exercise and an empirical illustration we show the usefulness of the method.
Keywords: Auxiliary model, efficient method of moments, indirect inference, information criteria
JEL Classification: C13, C52
Suggested Citation: Suggested Citation