ddml: Double/debiased machine learning in Stata

TL;DR

The paper introduces the ddml package for Double/Debiased Machine Learning in Stata, enabling flexible causal inference with various models and machine learning methods, supported by Monte Carlo validation.

Contribution

It presents a new Stata package that implements DDML for multiple econometric models, integrating with existing machine learning tools and recommending stacking for improved estimation.

Findings

Monte Carlo simulations support stacking estimation.

ddml supports five econometric models.

Compatible with many existing machine learning programs.

Abstract

We introduce the package ddml for Double/Debiased Machine Learning (DDML) in Stata. Estimators of causal parameters for five different econometric models are supported, allowing for flexible estimation of causal effects of endogenous variables in settings with unknown functional forms and/or many exogenous variables. ddml is compatible with many existing supervised machine learning programs in Stata. We recommend using DDML in combination with stacking estimation which combines multiple machine learners into a final predictor. We provide Monte Carlo evidence to support our recommendation.