Causal Forests and Machine Learning for Heterogeneous Treatment Effects
Estimate heterogeneous treatment effects—how effects vary across subgroups—using machine learning methods.
Classical econometric methods estimate average treatment effects, assuming effects are constant across individuals. In reality, policies often help some people more than others. Causal forests, built on random forest algorithms, estimate treatment effects separately for each unit by partitioning data into groups with similar treatment responses. The idea: grow random forests on residuals (outcome residuals and treatment residuals) to avoid overfitting to the treatment itself. Causal forests yield an estimated effect for each individual, revealing heterogeneity. For instance, a job training program might benefit young workers with strong math skills but not older workers. These methods combine machine learning's flexibility…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in