Environmental Kuznets Curves for air pollution in African and developed countries: exploring turning point incomes and the role of democracy

ABSTRACT

The Environmental Kuznets Curve (EKC) hypothesis advances an inverted U-shaped relationship between economic growth and environmental pollution. Scholars have estimated turning point incomes for various pollutants within and across countries. However, the majority of these studies were conducted for developed countries. Very few studies have focused on developing countries. In particular, the relationship between economic growth and air pollution in Africa remains largely unexplored. In this paper, we test whether the EKC hypothesis holds for carbon dioxide (CO₂) and particulate matter (PM₁₀) emissions in African and high-income OECD countries. We find that the EKC hypothesis holds for both CO₂ and PM₁₀ for African and OECD countries. Further, our examination of the effect of institutional quality on air pollution reveals an insignificant effect for CO₂ for both samples. However, democracy is positively and significantly correlated with PM₁₀ emissions for African countries.

KEYWORDS:

• Air pollution
• economic growth
• environmental Kuznets Curve
• environmental pollution
• turning point incomes
• institutional quality

JEL Codes:

• C12
• N57
• O13
• Q53
• Q56

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The US Environmental Protection Agency notes that although CO₂ does not pose direct health effects to humans, it is considered a pollutant because of its effect on climate change. Our reference to CO₂ as an air pollutant is based on this. We wish to thank an anonymous reviewer for drawing our attention to this.

2. We focus on CO₂ and PM₁₀ because of data availability issues and economy of space. We owe this to an anonymous reviewer who asked why we focused on only these two pollutants.

3. Kuznets (1955) addressed inequality in the distribution of income in the course of a country's economic growth and the factors that determine the secular levels and trends in income inequality. He argued that income inequality first rises as economies grow, but eventually declines (i.e. an inverted U-shaped relationship).

4. Note, however, that when Iglehart tested this hypothesis using survey data, it was only partly supported as some developing countries exhibited high environmental concern. This led him to propose the ‘objective problems and subjective values’ thesis. He argued that in developing countries, environmental concern is due to people's direct experiences of environmental problems.

5. See Li and Reuveny (2006) for a detailed review.

6. One possibility not shown in Figure 1 is a monotonically declining relationship indicating that pollution declines as income increases.

7. A complete list of these countries is presented in Table A1.

8. The excluded country is South Sudan which became independent in 2011 and therefore has no data for the period under consideration. Note also that sub-Saharan African countries excludes the Arab countries in Africa (i.e. Algeria, Egypt, Libya, Morocco and Tunisia). According to the World Bank, as of 2012, a developing country is a country with GNI per capita of less than $11,905. Although South Africa is a member of the G20 and is among the list of Newly Industrialised Countries (NICs), we include it in the African sample as a developing country because its GNI per capita for the period under consideration (1990–2010) was less than $11,905. Additionally, excluding South Africa from the African sample does not substantially alter the results.

9. Although we use log of environmental pollution and GDP per capita (consistent with many studies), earlier studies used levels for these variables (e.g. Grossman and Krueger 1995).

10. Alternative and, perhaps, more appropriate models for our analyses here could be dynamic spatial panel models. However, we follow the precedent of other studies (e.g. Panayotou 1997; Harbaugh et al. 2002; Stern 2004; Stern and Common 2001; Orubu and Omotor 2011) and focus on FE and RE models. Nonetheless, we wish to thank an anonymous reviewer who suggested this to us.

11. Data on democracy score were unavailable for Iceland and Luxembourg.

12. A general rule of thumb is that correlations near unity (e.g. 0.8) suggest the presence of multi-collinearity (see e.g. Farrar and Glauber 1967). In addition, we tested for multi-collinearity by examining the variance inflation factors (VIFs) in OLS models (not reported for economy of space). All VIFs were below 5 (except for the lower and higher order values of GDP per capita) suggesting the absence of multi-collinearity in the data. Consistent with polynomial regressions, some collinearity is expected among the GDP per capita variables (see e.g. Panayotou 1997).

13. The turning point income for a quadratic specification is given by exp(−β₁/2β₂) as shown in Equation (2)(2) , where β₁ is the coefficient of log of GDP per capita and β₂ is the coefficient of the square of the log of GDP per capita. Take the FE model for Africa, for example. −β₁/2β₂ = −1.802/[2*(−0.103)] = 8.747573. Taking the exponent of 8.747573 generates the turning point income of $6295.39.