Model Selection Uncertainty and Dynamic Models

Pub. Date
08 May, 2000
Pub. Type

Inference is usually carried out conditional on a selected model disregarding any effects on the uncertainty of estimation arising out of the procedure used to select between alternative models. Nevertheless, the uncertainty arising out of model selection is bound to be of comparable magnitude to that arising out of model selection is bound to be of comparable magnitude to that arising out of the estimation procedure conditional on a particular model. In this paper we discuss methods for accounting for model selection uncertainty and evaluate these methods in the context of dynamic models through Monte Carlo simulations.