Economic Growth and Environmental Degradation: Investigating the existence of the environmental Kuznets curve for local and global pollutants in South Africa

Economic growth has been seen to be accompanied by surges in natural resource extraction rates or levels of pollution and waste. As such, many suggest that the pursuit thereof may lead to environmental degradation through increased waste generation and pollution, given a country’s technological constraints and environmental assimilative capacity. In the field of economics, the ‘Environmental Kuznets Curve’ (EKC) has served as arguably the most dominant approach to assess this relationship between economic growth and environmental degradation since its popularisation in the early 1990s (Stern, 2017:8). The EKC implies that economic activity is environmentally beneficial in the long-run, despite adversely affecting it in the short-run. International findings remain mixed at best, and only a limited amount of other studies which attempt to assess the existence of an EKC in South Africa’s context exist, all of which use the same global air pollutant for environmental quality. The aim of this paper is to contribute to the existing literature by investigating the presence of the EKC for a set of relatively diverse – three local and three global – air pollutants in South Africa for the period 1970 to 2010. This study serves as the first to estimate the relationship for any local pollutant, as well as two global pollutants, in South Africa through the EKC framework. Using OLS and ARDL regression techniques, the results of the 24 estimated models do not provide evidence of an EKC for any of the select pollutants. However, when using levels instead of logarithms, an EKC is found in one specification for one local pollutant (NH3). Otherwise, no distinction between local and global air pollutants is found. In contrast to the EKC’s inverted-U shape, the ARDL models for two global (CO2 and N2O) and two local (SO2 and PM10) pollutants indicate statistically significant U-shaped relationships at conventional significance levels. Unfortunately, the reduced-form approach utilised in this paper does not indicate any underlying causal relationship and as such, conclusive policy suggestions cannot be made.