A Random Forest–Based Panel Data Approach for Program Evaluation

It is challenging to conduct controlled experiments to assess the impacts of social policy. To address this, past studies propose a panel data approach using factor models to estimate average treatment effects. The selection of control units is a critical step to balance the goodness of fit within-sample with the posttreatment forecasting error when the number of observed potential control units is large. In this study, we propose using random forests, an ensemble learning method, which offers robustness and requires fewer candidate models compared to existing methods. We demonstrate that our approach effectively selects almost all relevant control units, and we provide asymptotic normality results under the null of no average treatment effect and significance tests for policy interventions. Extensive simulations confirm the method's superior performance. In the empirical studies, we showcase the usefulness of the method by evaluating the impact of Brexit on the United Kingdom's GDP growth and China's anti-corruption campaign on the importation of luxury watches.