Price elasticity of CO2 emissions in China: A machine learning approach

Abstract

Quantifying the impacts of carbon prices on urban CO2 emissions is imperative to developing individualized carbon pricing schemes in China's emission trading systems (ETSs). Based on a prefecture-level panel dataset from 2005 to 2017, we use a machine learning approach to predict the individual price elasticities of CO2 emissions for 284 cities in China. The results exhibit significant heterogeneity in price elasticities: The full distribution of price elasticities indicates a right-skewed and leptokurtic characteristic, with an average price elasticity of −4.0 % and −3.6 % in CO2 emissions and CO2 intensity, respectively. We further find that the key city characteristics in predicting price elasticities are population density, industrial structure, and the number of industrial firms. Using the predictions of price elasticities, we conclude with a simulation to illustrate how different carbon pricing policies can help reduce carbon emissions at the national level. The results have important policy implications for China aiming to address low-carbon issues through carbon pricing.