Generalized Information Theoretical Approach to Panel Regression Kink Model

Phachongchit Tibprasorn, Paravee Maneejuk, Songsak Sriboochitta

Abstract

This study aims at exploring the use of generalized maximum entropy for estimating the unknown parameters in macroeconomic panel data models with two special concerns. First, the macroeconomic empirical panel studies usually face to a problem of data limitations, where the exiting estimation methods are often hard to get a significant results. Hence, we consider the GME estimator as one of the effective solutions for this problem. Second, there is often a discontinuity found in a relationship between explanatory and response variables in the macroeconomic studies. Therefore, in this study, the panel regression kink design based on the GME estimator is proposed for examining a discontinuous slope of the relationship between variables especially when the data is limited. The performance of the proposed method is evaluated through simulation study and later in the empirical analysis of the foreign direct investment. Both experiments showed a satisfiable performance of the model as well as the considered estimation technique.