A1713
Title: Nonparametric cointegration analysis of the environmental Kuznets curve
Authors: Martin Wagner - University of Klagenfurt, Bank of Slovenia and Institute for Advanced Studies, Vienna (Austria)
Fabian Knorre - TU Dortmund University and Statkraft (Germany) [presenting]
Abstract: A large and growing literature uses unit root and cointegration techniques to investigate the environmental Kuznets curve (EKC) hypothesis that postulates an inverse U-shaped relationship between the level of economic development and pollution or emissions. Given that economic theory does not, typically, lead to specific functional forms of the EKC, it appears natural to resort to nonparametric estimation. Given the nascent state of the nonparametric cointegration literature, it is provided first a large scale simulation assessment of currently available nonparametric cointegration estimators as well as of tests for the null hypothesis of a specific parametric functional form, e.g., a polynomial relationship, which is the dominant specification in empirical EKC analysis. The considered estimators and tests differ, i.a., with respect to the setting for which asymptotic results are derived, e.g., whether the regressor is allowed to be endogenous or restricted to be strictly exogenous, or whether the errors are allowed to be serially correlated or required to be uncorrelated. The simulation setup is designed to also assess the (finite sample) sensitivity of the estimators and tests with respect to violations of some assumptions. Based upon the simulation findings, we perform nonparametric cointegrating EKC analysis using annual data for CO2 emissions, SO2 emissions and GDP for 18 early industrialized countries over the period 1870-2016.