Bayesian Empirical Likelihood Estimation of Smooth Kink Regression

Woraphon Yamaka, Paravee Maneejuk

Abstract

The smooth kink regression model is introduced in this study. The model provides more flexibility in investigating the nonlinear effect of independent variable on dependent variable. The logistic function is considered as a regime weighting function for separating our two-regime model. In the estimation point of view, we employ the Bayesian empirical likelihood (BEL) as it gives a flexible way of combining data with prior information from our knowledge and the empirical likelihood in order to avoid the misspecification of the likelihood function. The performance and accuracy of the estimation from our proposed model is examined by the simulation study and real data.