ddml: Double/debiased machine learning in Stata

Achim Ahrens; Christian Hansen; Mark Edwin Schaffer; Thomas Wiemann

doi:10.48550/arXiv.2301.09397

ddml: Double/debiased machine learning in Stata

Achim Ahrens, Christian Hansen, Mark Edwin Schaffer, Thomas Wiemann

Research output: Working paper

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.

Original language	English
DOIs	https://doi.org/10.48550/arXiv.2301.09397
Publication status	Published - 23 Jan 2023

Keywords

causal inference
machine learning
doubly-robust estimation

Access to Document

10.48550/arXiv.2301.09397Licence: CC BY-SA

Cite this

@techreport{7939184aa2b745d795d32e7891a492f9,

title = "ddml: Double/debiased machine learning in Stata",

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.",

keywords = "causal inference, machine learning, doubly-robust estimation",

author = "Achim Ahrens and Christian Hansen and Schaffer, {Mark Edwin} and Thomas Wiemann",

year = "2023",

month = jan,

day = "23",

doi = "10.48550/arXiv.2301.09397",

language = "English",

type = "WorkingPaper",

}

TY - UNPB

T1 - ddml: Double/debiased machine learning in Stata

AU - Ahrens, Achim

AU - Hansen, Christian

AU - Schaffer, Mark Edwin

AU - Wiemann, Thomas

PY - 2023/1/23

Y1 - 2023/1/23

N2 - 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.

AB - 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.

KW - causal inference

KW - machine learning

KW - doubly-robust estimation

U2 - 10.48550/arXiv.2301.09397

DO - 10.48550/arXiv.2301.09397

M3 - Working paper

BT - ddml: Double/debiased machine learning in Stata

ER -

ddml: Double/debiased machine learning in Stata

Abstract

Keywords

Access to Document

Fingerprint

Cite this