Measurement of eco-efficiency and convergence: Evidence from a non-parametric frontier analysis

This study applies a nonparametric model to estimate the eco-efficiency across the US states over the period 1990–2017. To capture the environmental damage caused by anthropogenic activities, we utilize one global (CO2) and two local (SO2 and NOX) pollutants emitted by power plants to serve as inputs to the eco-efficiency analysis and states’ GDP levels as an output. The paper's primary contribution is to employ for the first time in the empirical literature a probabilistic frontier analysis (order-m estimators) to exemplify the US regional convergence/divergence patterns on eco-efficiency. The results based on the Phillips and Sul methodology (2007; 2009) indicate divergence for the whole sample. However, at least five regional convergence clubs are formulated dividing the US states into “champions” and “laggards” according to their eco-efficiency estimates. Moreover, we examine the convergence-divergence hypothesis by employing an alternative nonparametric distributional dynamics approach based on a Markov chain. Although the stochastic kernels uncover the presence of regional clustering among the US territory, they signify the existence of at least two convergence clubs. Our results survive robustness checks under the inclusion of two alternative eco-efficiency indicators, providing significant implications to government officials and policymakers.