Coherent prior distributions in univariate finite mixture and Markov-switching models

TL;DR

This paper develops a framework for setting coherent prior distributions in Bayesian univariate finite mixture and Markov-switching models, ensuring consistency between nested and general models, with specific conditions for common priors and practical applications.

Contribution

It introduces a systematic approach to establish coherence of priors in mixture and Markov-switching models, including conditions for normal, inverse gamma, and gamma priors, and discusses implications of additional constraints.

Findings

Derived coherence conditions for normal, inverse gamma, and gamma priors.

Analyzed effects of constraints like identifiability and stationarity on coherence.

Illustrated methodology with Markov-switching AR(2) models.

Abstract

Finite mixture and Markov-switching models generalize and, therefore, nest specifications featuring only one component. While specifying priors in the two: the general (mixture) model and its special (single-component) case, it may be desirable to ensure that the prior assumptions introduced into both structures are coherent in the sense that the prior distribution in the nested model amounts to the conditional prior in the mixture model under relevant parametric restriction. The study provides the rudiments of setting coherent priors in Bayesian univariate finite mixture and Markov-switching models. Once some primary results are delivered, we derive specific conditions for coherence in the case of three types of continuous priors commonly engaged in Bayesian modeling: the normal, inverse gamma, and gamma distributions. Further, we study the consequences of introducing additional constraints into the mixture model's prior (such as the ones enforcing identifiability or some sort of regularity, e.g. second-order stationarity) on the coherence conditions. Finally, the methodology is illustrated through a discussion of setting coherent priors for a class of Markov-switching AR(2) models.