Classifying political positions on carbon pricing in Germanyusing machine learning

For analyzing the adoption of carbon pricing policy it is helpful to understand the salience of carbon pricing in political debates and how political preferences in favor or in opposition to carbon pricing changed over time. Here, we use a novel machine learning based algorithm to classify salience and political preferences on carbon pricing among German politicians from 1980 to 2020. Our algorithm is trained on a dataset of manually coded parliamentary speeches and uses an ensemble of machine learning classifiers to detect and classify political positions on carbon pricing and other climate change mitigation policies. Here, we present the first results of our classification analysis and compare salience and political preferences for carbon pricing across politicians, political parties, and time. We also present the preliminary results of an inferential analysis in which we predict salience in and political stance towards carbon pricing among German politicians based on party affiliation, voting district, and current political context.