A static and dynamic copula-based ARIMA-fGARCH approach to determinants of carbon dioxide emissions in Argentina.

This paper attempts to model both static and dynamic dependence structures and measure impacts of energy consumptions (both renewable (EC) and non-renewable (REN) energies), economic globalization (GLO), and economic growth (GDP) on carbon dioxide (CO) emissions in Argentina over the period 1970-2020. For analyses purpose, the current research deploys the novel static and dynamic copula-based ARIMA-fGARCH with different submodels. The static bivariate copula results show that the growth rates of the pairs EC-CO and GDP-CO are asymmetrically positive co-movements and have high left tail (extreme) dependencies, implying that the increase in non-renewable energy and economic growth can critically contribute to the environmental degradation, and the decrease in the consumption of non-renewable energy at a high level will consequently reduce the CO emissions at the same level. Based on several copula-based dependence measures, we document that between the two factors, the non-renewable energy has a stronger impact than the economic growth regarding the CO emissions. On the other hand, the growth rates of both economic globalization and renewable energy symmetrically negatively co-move with the growth rates of the CO emissions, but they have no extreme dependencies, indicating that these factors contribute to Argentina's environmental quality, in which the factor of renewable energy has a greater impact. Furthermore, the dynamic copula outcomes show that the (tail) dependencies of CO emissions on the non-renewable energy and economic growth are time-varying, while the pairs REN-CO and GLO-CO possess only dynamic dependencies, but no dynamic tail dependencies. Moreover, through the dynamic copula-based dependence, the environmental Kuznets curve (EKC) hypothesis can be estimated and illustrated explicitly. In addition, we leverage multivariate vine copulas for modelling dependence structures of the five variables simultaneously, which can reveal rich information regarding conditional associations among the relevant variables. Some policy implications are also provided to mitigate CO emissions.