Machine learning the first stage in 2SLS: Practical guidance from bias decomposition and simulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lennon, Connor, Rubin, Edward, Waddell, Glen
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916744569815040
author Lennon, Connor
Rubin, Edward
Waddell, Glen
author_facet Lennon, Connor
Rubin, Edward
Waddell, Glen
contents Machine learning (ML) primarily evolved to solve "prediction problems." The first stage of two-stage least squares (2SLS) is a prediction problem, suggesting potential gains from ML first-stage assistance. However, little guidance exists on when ML helps 2SLS$\unicode{x2014}$or when it hurts. We investigate the implications of inserting ML into 2SLS, decomposing the bias into three informative components. Mechanically, ML-in-2SLS procedures face issues common to prediction and causal-inference settings$\unicode{x2014}$and their interaction. Through simulation, we show linear ML methods (e.g., post-Lasso) work well, while nonlinear methods (e.g., random forests, neural nets) generate substantial bias in second-stage estimates$\unicode{x2014}$potentially exceeding the bias of endogenous OLS.
format Preprint
id arxiv_https___arxiv_org_abs_2505_13422
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Machine learning the first stage in 2SLS: Practical guidance from bias decomposition and simulation
Lennon, Connor
Rubin, Edward
Waddell, Glen
Econometrics
Machine Learning
Applications
Machine learning (ML) primarily evolved to solve "prediction problems." The first stage of two-stage least squares (2SLS) is a prediction problem, suggesting potential gains from ML first-stage assistance. However, little guidance exists on when ML helps 2SLS$\unicode{x2014}$or when it hurts. We investigate the implications of inserting ML into 2SLS, decomposing the bias into three informative components. Mechanically, ML-in-2SLS procedures face issues common to prediction and causal-inference settings$\unicode{x2014}$and their interaction. Through simulation, we show linear ML methods (e.g., post-Lasso) work well, while nonlinear methods (e.g., random forests, neural nets) generate substantial bias in second-stage estimates$\unicode{x2014}$potentially exceeding the bias of endogenous OLS.
title Machine learning the first stage in 2SLS: Practical guidance from bias decomposition and simulation
topic Econometrics
Machine Learning
Applications
url https://arxiv.org/abs/2505.13422