Using Machine Learning to Identify Heterogeneous Impacts of Agri-Environment Schemes in the EU: A Case Study

Abstract

Abstract Legislators in the European Union have long been concerned with the environmental impact of farming activities and introduced so-called agri-environment schemes (AES) to mitigate adverse environmental effects and foster desirable ecosystem services in agriculture. This study combines economic theory with a novel machine learning method to identify the environmental effectiveness of AES at the farm level. We develop a set of more than 130 contextual predictors to assess the individual impact of participating in AES. Results from our empirical application for Southeast Germany suggest the existence of heterogeneous, but limited effects of agri-environment measures in several environmental dimensions such as climate change mitigation, clean water and soil health. By making use of Shapley values, we demonstrate the importance of considering the individual farming context in agricultural policy evaluation and provide important insights into the improved targeting of AES along several domains.