How do countries' incomes affect environmental quality?

Abstract

This study aims to investigate the relationship between economic growth and the environment using wider, more aggregated, indicators of ecosystem health than have been used in previous studies. This is done in an attempt to get a better picture of the global and diffuse impacts of economic activity on the biosphere. A general-to-specific (Gets) method is applied to estimate the relationship between three different proxies for environmental quality and various candidate regressors. While measures of SO2 and other point-source emissions may be decreasing for many developed nations, the validity of extrapolating to a general hypothesis about environmental quality in general may be exaggerated by the particular substitution possibilities between the chosen regressand and other potential alternatives. Globally diffuse impacts like CO2 emissions and habitat destruction are often argued to pose a more serious threat to ecosystem health, and the ability to substitute the use of some environmental services may be impossible or socially suboptimal. This represents a major objection to the environmental Kuznets curve (EKC) hypothesis, which suggests that the environment is initially degraded by economic growth, but beyond a point, improves with further development. Because of this fundamental criticism of the EKC literature (succinctly expressed by Arrow et al, 1995), indicators of degradation are assessed in terms of sustainability, particularly the tension between weak and strong views.

To avoid the potential problems of multicollinearity, omitted variable bias (Stern et al, 1996, Stern and Common, 2001), and pollution displacement (Arrow et al, 1995, Stern et al, 1996), aggregate measures of environmental performance are investigated.

Aggregation of environmental impact aims to capture all of the direct and indirect effects in the GDP/environment relationship at the macro level. Pollutant-specific data are often related to other indicators linked to industrial intensity. Movements in these variables may be caused by movements in other potential variables that are omitted from the regression equations. Aggregation based on many factors going in different directions will have a central tendency, and mitigates the possibility that substitution between impacts at the micro level distorts the growth-environment relationship from a sustainability perspective (Jha and Murthy, 2003b).

Particular cumulative indexes examined include the Ecological Footprint (Global Footprint Network, 2005, Wackernagel et al, 1997), the Environmental Sustainability Index (Esty et al, 2005), and the Environmental Degradation Index (Jha and Murthy, 2003b), each investigated as an alternative proxy for environmental quality. These broader attempts to measure environmental performance and their congruence with well-known and established indicators of economic, institutional, and social/cultural phenomena are compared with some of the more common local pollutant data used in the existing empirical literature.

To investigate an alternative methodology, the automated Gets approach, developed by Hendry and Krolzig (2001), is utilised. Its performance in the EKC research paradigm is carefully considered with a battery of diagnostic testing. The Gets method has significant appeal because the underlying theoretical literature behind EKCs is diverse in focus, allowing a wide range of candidate regressors as proxies for the possible forces at work. The theoretical basis of EKCs has been manifest as various explanations of observed data correlations, the starting point being an extrapolation of cross-country observations to single-economy, dynamic models. For that reason, model selection to square the theory with the data is crucial and the Gets emphasis on encompassing rival alternatives and diagnostic testing seems an improvement on the previously restrictive approaches to model selection.

By applying a general-to-specific (Gets) approach to the econometric methods of estimating the income-environment relationship, emphasis is placed on diagnostic testing and the general need for congruence. An attempt to encompass previously conjectured relationships is attempted from the outset.