Measuring Productivity When Technologies are Heterogeneous: A Semi-Parametric Approach for Electricity Generation

Abstract

While productivity growth in electricity generation is associated with multiple positive effects from an economic and environmental perspective, measuring it is challenging. This paper proposes a framework to estimate and decompose productivity growth for a sector characterized by multiple technologies. Using a metafrontier Malmquist decomposition and frontier estimation based on stochastic non-smooth envelopment of data (StoNED) allows for productivity estimation with few microeconomic assumptions. Additionally, evaluation of productivity at representative hypothetical units permits distribution-free analysis for the whole distribution of power plant sizes. The proposed framework is used to analyze a unique and rich dataset of coal, lignite, gas, and biomass-fired generators operating in Germany from 2003 to 2010. The results indicate stagnating productivity for the sector as a whole, technical progress for biomass plants, and very high productivity for gas-fired plants.