11. Uluslararası İstatistik Kongresi, Muğla, Türkiye, 4 - 08 Ekim 2019, ss.67-68
Global warming and climate change have been one of the most important environmental problems in the last two
decades. The Environmental Kuznets Curve (EKC) hypothesis postulates the existence of an inverse U-shape
relationship between per capita gross domestic product (GDP) and Carbon dioxide emissions (CO2). This article
investigates the validity of the Environmental Kuznets Curve (EKC) hypothesis in a cross-section of 131 developed
and developing countries with data from 2014. Nonlinearity means the evidence of EKC hypothesis in regression
analysis. So testing linearity goes to the conclusion or not of EKC hypothesis. An F test or a comparison between
linear determination coefficient and correlation ratio may be helpful. For detecting nonlinearity with independent
of strict assumptions we proposed a method based on mutual information between partial residuals as a measure
of nonlinearity. This method is based on comparing the mutual information between residuals obtained by
removing linear dependence from the original ones and residuals obtained by regressing curvilinear effect variable
X2 on linear effect variable X. The last method is based on mutual information measure suggested by Tanaka,N.,at
all., In addition to their study we also showed that nonlinearity can be detected better with Rényi and Tsallis mutual
informations because of their flexibilities due to α parameter. In this paper we studied these tests and then compared
the results obtained by each test. By determining the nonlinearity between the variables we have shown that the
EKC hypothesis is supported. The result is important because it may have important policy implications.