This paper proposes a new analytical framework with which to analyze the determinants of global CO2 emissions. It contributes to the existing literature by examining the determinants of CO2 emissions using a flexible functional form (transcendental logarithmic model), taking into account the presence of dynamic effects and allowing for heterogeneity in the sample of countries. The sample covers 121 countries and the period analyzed extends from 1975 through 2003. Two main results emerge. First, a static specification is rejected against a dynamic model. Second, the data also reject a general specification for all countries; hence slopeheterogeneity in the estimated coefficients has to be modeled. Conversely, the STIRPAT model is generally accepted for high-income countries, whereas for developing countries several interaction terms also play a role in explaining CO2 emissions.
CO2 emissions developing countries panel data population growth urbanization