Expectile kink regression: An application to service sector output

Abstract

© 2018, Springer International Publishing AG. In this study, we propose a non-linear model for explaining the relationship between the dependent and the independent variables beyond the conditional mean. We extend the kink approach to expectile regression thus the model provides a more flexible means to explain the non-linear relationship in the model across different expectile indices. We also introduce the sup-F statistic test for the existence of kink effect in each expectile. The simulation and application studies are also proposed to examine the performance of our model. We apply our methodology to study the input factor affecting service sector growth in Asian economy. The use of this model allows us to identify and explore the non-linear labour effect on the service output. We can find both labour effect and kink effect present over a range of expectiles in the service output in this application.