Environmental Kuznets Curve time series application for Turkey: Why controversial results exist for similar models?

This paper investigates the Environmental Kuznets Curve (EKC) hypothesis using 40 year spanned time series for the Turkey case. CO2 emission series representing environmental pressure and GDP per capita values representing economic development are used for the period 1968-2007. A cointegration has been determined among nonstationary series. First phases of an inverted U form EKC relationship have been determined for Turkey from the econometric estimations. This result is conflicting with that of the similar models for Turkey case. On the other hand, this conflict refers the important arguments in the literature and constitutes the main points of the paper. Sensitivity critiques (for example, Ahking et al.) for cointegration tests (Johansen and Engle-Granger tests) have been supported in our study. Moreover, we detected important diversion in results according to drift and trend assumptions both in CI vector and EKC model specifications. We conclude that building EKC model according to cointegration (CI) equation restrictions can be important source for diversion when sensitivity exists in estimations and cointegration tests; therefore, EKC estimations should be held in non-restrictive way. The additional structural reasons have been also discussed for developing country EKC cases. The most important one is that the narrow income sample of developing countries makes it possible to be defined by similar but different paths; therefore, policy implications to be drawn from those analyses should not ignore this feature of developing country analyses. (C) 2015 Elsevier Ltd. All rights reserved.