Bayesian nonlinear panel cointegration: an empirical application to the EKC hypothesis

Abstract

In this paper we employ a new Bayesian approach within a cointegrating regression framework along the lines of Breitung (J Bus Econ Stat 19:331–340, 2001). We extend this fully to the case of panel data framework with general cross-sectional dependence. We argue that our Bayesian technique does not depend on asymptotic critical values and in our empirical application we do find strong evidence of nonlinear cointegrated relationships between local (SO2 and NOX) and global (CO2) emissions with the level of economic growth.