Testing Linear and Nonlinear Relationships Between Foreign Direct Investment and Fossil Energy Consumption in Fragile Five CountriesAli Çelik
The global warming and climate change problem is causing severe threats for the present and the future. The excessive utilization of fossil energy resources, especially on the production side, has contributed greatly to global warming by producing the greenhouse effect. In this respect, it is essential to assess the impact of foreign direct investment (FDI), which is expected to increase the production and employment of countries, on the utilization of fossil fuels. In this study, we investigate the relationship between FDI and fossil energy consumption (EC) utilizing such tests as: Johansen linear cointegration, Kapetanios, Shin, and Snell (2006) nonlinear cointegration, the linear Error Correction Model (ECM), Exponential Smooth Autoregressive (ESTAR) ECM, Granger (1969) linear causality, and Diks and Panchenko (2006) nonlinear causality. In this context, annual data covering the period between 1980 and 2020 are employed for the “Fragile Five” countries, which are: Brazil, Indonesia, India, Turkey, and South Africa. By applying the Augmented DickeyFuller (ADF) linear unit root test, we found that the series became stationary after taking the first difference. Following the unit root test results, Johansen's (1988) linear cointegration test results indicated that there existed a cointegration relationship from FDI to EC for Turkey and South Africa, while Kapetanios, Shin, and Snell's (2006) nonlinear cointegration test results revealed that there existed a cointegration relationship from FDI to EC in South Africa. In addition, the linear error correction model was proven to be valid for Turkey and South Africa, while the ESTAR nonlinear error correction model is valid only for Turkey. Finally, Granger's (1969) causality test results proved that there was a causal relationship from FDI to EC in Turkey. Dicks and Panchenko (2006) stated that there was no causal relationship between the variables.