Skew-normal Bayesian spatial heterogeneity panel data models

J Appl Stat. 2019 Aug 24;47(5):804-826. doi: 10.1080/02664763.2019.1657812. eCollection 2020.

Abstract

This paper proposes a new regression model for the analysis of spatial panel data in the case of spatial heterogeneity and non-normality. In empirical economic research, the normality of error components is a routine assumption for the models with continuous responses. However, such an assumption may not be appropriate in many applications. This work relaxes the normality assumption by using a multivariate skew-normal distribution, which includes the normal distribution as a special case. The methodology is illustrated through a simulation study and application to insurance and gasoline demand data sets. In these analyses, a simple Bayesian framework that implements a Markov chain Monte Carlo algorithm is derived for parameter estimation and inference.

Keywords: Bayesian inference; Elasticity; Gibbs sampler; MCMC; Spatial panel data model; multivariate skew-normal distribution.