Lessons From Quantile Panel Estimation of the Environmental Kuznets Curve

Abstract

The environmental Kuznets curve (EKC) hypothesizes that the income-pollution relationship has an inverted U shape, implying that pollution is increasing in income up to a turning point beyond which pollution decreases. The empirical literature has concentrated on estimation of this relationship at the mean, employing longitudinal data on countries or U.S. states and measures of per-capita income and emissions of pollutants like N O x (nitrogen oxide) and SO 2 (sulfur dioxide). The typical …nding supports an inverted U shaped relationship. Estimation at the mean, however, likely masks heterogeneities that can be present at higher and/or lower quantiles of the emissions'distribution. In addition, mean estimation methods are sensitive to the presence of outliers. This study applies methods for conditional-quantile panel …xed e¤ects models to the estimation of the income-emissions relationship on U.S. state-level data on N O x and SO 2 pollutants over the period 1929-1994. Our results indicate that methods that focus on the conditional mean provide too optimistic estimates about emissions reduction of N O x , as conditional-quantile methods suggest that the turning point of the relationship occurs at higher values of income; while the opposite is found for SO 2 . An important lesson we draw from the application of conditional-quantile techniques is that the income-environmental degradation relationship is sensitive to the presence of outliers in the data.

Are there Environmental Kuznets Curves for US state-level CO2 emissions?

“…In fact, in the time-series set-up, the EKC is found to hold for only a maximum of eight states. More recently, Flores et al, (2014) rightly points out that the EKC empirical literature has concentrated almost exclusively on estimation of the incomepollution relationship and its turning point at the conditional mean. The authors propose a quantile-regression based approach since estimates of the conditional median are more robust relative to estimates of the conditional mean, especially when the underlying distribution has thick tails --a typical feature in emissions data.…”

The effects of FDI, economic growth and energy consumption on carbon emissions in ASEAN-5: Evidence from panel quantile regression

“…Third, the panel quantile regression estimation results are robust to outlying observations of the explained variable and are more effective than OLS regression, especially when the error term is non-normal, which will help policymakers formulate more accurate environmental protection policies. However, only a few papers have applied a panel quantile regression fixed effect model to investigate the relationship among variables (Damette and Delacote, 2012;Flores et al, 2014;Yaduma et al, 2015).…”

Enfoques alternativos para el análisis de los cambios distributivos en América Latina

“…Sin embargo, dentro de la literatura de la economía ambiental, mencionada en la sección previa, se ha empezado a ampliar la visión sobre cómo mirar a este tipo de relación mirando otros aspectos de la misma. En particular, los estudios empíricos sobre la Curva de Kuznets Ambiental se han volcado gradualmente al estudio de los cuantiles condicionales argumentando que la interacción de la polución (variable dependiente) con factores inobservables hace que el nivel de contaminación durante la primera fase del desarrollo puede determinar la evolución ambiental en las etapas siguientes (Flores et al, 2009). En forma análoga, dentro de la literatura de la distribución del ingreso se podría esperar que los niveles iniciales de desigualdad afecten el sendero de desarrollo.…”

“…resuelve el siguiente problema de optimización: 9 Algunos ejemplos de estimaciones de curvas de Kuznets ambientales son los trabajos de Flores Flores-Lagunes y Kapetanakis (2009) y Kapetanakis (2009). 10 El hecho de que una misma región no necesariamente se mantendrá siempre sobre el mismo cuantil condicional no es un argumento en contra de ésta interpretación puesto que el mismo también se aplica sobre la interpretación de la media condicional.…”