Essays on hierarchical Bayesian estimation of spatio-temporal economic models

Abstract

This dissertation contributes to the economic literature dedicated to the analysis of the complex spatio-temporal models in the Bayesian hierarchical setting. The present research consists of two empirical applications in the areas of international trade and agricultural yield prediction. The first empirical study focuses on the problem of estimating the effect of social barriers to trade. I develop a Bayesian social relation gravity model that properly accounts for a two-stage decision process and the interactive nature of bilateral trade. The model uses the distance based threshold specification and assumes invariance to the labeling of equations. The estimated effects of the fundamental variables are found to be mostly consistent with the results reported traditionally in the gravity model research literature. Formal tests based on the Savage-Dickey density ratio show the significance of effects of governance levels on international trade and suggest asymmetry of the social costs of trade. The second empirical study presents a spatio-temporal statistical model of agricultural yield prediction based on spatial mixtures of distributions. The proposed method combines several hierarchical and sequential Bayesian estimation procedures that allow the general problem to be addressed with a series of simpler tasks, providing the required flexibility. The spatial correlation hypothesis is studied by comparing the alternative models using the posterior predictive criterion under a squared loss function. Findings indicate that the proposed approach has a better ability to predict agricultural yields and is able to correctly suggest the possibility of a significant decrease in yields level. The premium rates for crop insurance suggested by the normal county level model are found to be significantly lower than the premium rates calculated based on the spatial mixtures model, especially for the years and locations where unfavorable events actually occurred.