The Environmental Kuznets Curve: fact or fiction?

Many economics students will be familiar with the Kuznets curve. It describes the relationship between income per capita and inequality first hypothesised by Nobel laureate economist Simon Kuznets. The theory postulates that as an economy develops and incomes begin to rise so to will inequality until a threshold level of income is reached. After this threshold level of income is reached, Kuznets hypothesised that inequality would decline.

A similar relationship has been theorised between income per capita and environmental outcomes and this is known as the environmental Kuznets curve (EKC). In this article I shall discuss the EKC and examine whether it exists in reality.

The theory

The EKC posits that after a turning point, economic growth will be negatively correlated with indicators of environmental degradation. The EKC’s underlying economic rationale is that as a society begins the process of economic development, pollution and environmental degradation increase with income per capita. This is because the first priority for the society is to escape poverty and raise material living standards via economic growth. However, once a society reaches a threshold level of income ‘the demand for improvements in environmental quality will increase, as will the resources available for investment’ (IBRD, 1992). This threshold level of income forms the turning point of the curve, which is displayed visually below.

(Source: An EKC depicting the relationship between income and sulphur emissions sourced from Stern, 2004 based on work by Panayotou, 1993 and Barbier, Common and Stern, 1996.)

The implications of the EKC hypothesis, if it were true, would be radical. It would mean that we could grow our way out of environmental problems. Cap and Trade Schemes or taxation measures to ameliorate pollution would be unnecessary – the answer is economic growth. This sounds too good to be true. Yet according to the EKC it is correct.

The reality

Given that the EKC hypothesis makes intuitive sense and the major consequences it would have for public policy if it were true, it seems prudent to test the hypothesis empirically. Thankfully, there have been a multitude of studies which do just this.

The major flaw in the EKC which the literature reveals is that it assumes all pollutants will behave in generally the same way in relation to income. However, empirical studies have revealed that this is not the case. Whilst certain pollutants such as sulphur or nitrogen oxides have decreased as income has increased, others such as carbon dioxide emissions and solid waste have increased (Stern, 2004).

The above point was reiterated in a study by Brajer, Mead and Xiao testing for an EKC in relation to China’s air pollution. Brajer, Mead and Xiao stated that because overall air pollution is comprised of various discrete pollutants one cannot infer an aggregate EKC; because each pollutant has a discrete relationship with income. Expressed in econometric terms, an aggregate EKC suffers from problems of heteroscedasticity. Consequently, the EKC literature is varied depending on the pollutants selected.

Whilst acknowledging the problem of heteroscedasticity, Brajer, Mead and Xiao nevertheless tested for an EKC with various aggregate pollution index measures and found the weight of the empirical evidence pointed away from ‘the inverted U type hypothesis’ and favoured ‘an environmentally bleaker cubic relationship’. Thus, even where attempts are made to aggregate discrete pollutants, the EKC hypothesis does not hold.

Another problem with the EKC is that it is not empirically robust. Sensitivity analyses on much of the early EKC literature demonstrate this, as adding explanatory variables, removing outliers and extending the dataset all cause large changes to the outcome.


It appears that the EKC exists in reality only in relation to certain pollutants. Accordingly, it is not a theory which should be generalised and applied in public policy formation. In order to craft effective pollution abatement schemes, policy-makers should examine the discrete relationship between the pollutant they are seeking to abate and income. Assuming that the pollutant will eventually decrease as incomes rise would be unempirical, lazy and dangerous.