Analysis of Markov switching seemingly unrelated regression model with skewed distributions, and its application to Thai cassava market

Abstract

© Published under licence by IOP Publishing Ltd. This study conducts a Markov switching seemingly unrelated regression with skewed distribution. We analyse MS-SUR model under skewed distributions. In other words, the multivariate skew-normal and skew-t with unknown skewness parameter are adopted to be the likelihood for the MS-SUR model. The simulation results show that parameters of MS-SUR model appear to be quite reasonable. In addition, we apply our model to study Thai cassava market and the results provide an evidence that given skewed student-t distribution for both demand and supply equations in the MS-SUR models present the lowest AIC and BIC. Moreover, the empirical results show that exchange rate and export price are two main factors that affect, respectively, demand for and supply of Thai cassava in market downturn regime.