摘要：This paper examines the carbon dioxide (CO2) Environmental Kuznets curve (EKC) hypothesis of a balanced panel of 50 middle-income countries over the period 1996–2013 using a dynamic spatial panel data model with country and time-period fixed effects. Using a Bayesian comparison approach, we systematically searched for the most suitable spatial weights matrix describing the spatial arrangement of the countries in the sample. We found substantial spatial dependence effect in CO2 emissions across the sample of middle-income countries, highlighting the influence these countries exert on their neighbors. Besides, the empirical results showed that the relationship between economic growth and CO2 emissions shaped as an inverted-U trajectory. Furthermore, it has been found that trade openness and energy intensity are the main factors on slightly increasing CO2 emissions, while the urbanization contributes to relative decrease in CO2 emissions.
