Bootstrap Inference of Matching Estimators for Average Treatment Effects

Taisuke Otsu, Yoshiyasu Rai

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

It is known that the naive bootstrap is not asymptotically valid for a matching estimator of the average treatment effect with a fixed number of matches. In this article, we propose asymptotically valid inference methods for matching estimators based on the weighted bootstrap. The key is to construct bootstrap counterparts by resampling based on certain linear forms of the estimators. Our weighted bootstrap is applicable for the matching estimators of both the average treatment effect and its counterpart for the treated population. Also, by incorporating a bias correction method in Abadie and Imbens (2011), our method can be asymptotically valid even for matching based on a vector of covariates. A simulation study indicates that the weighted bootstrap method is favorably comparable with the asymptotic normal approximation. As an empirical illustration, we apply the proposed method to the National Supported Work data. Supplementary materials for this article are available online.

Original languageEnglish
Pages (from-to)1720-1732
Number of pages13
JournalJournal of the American Statistical Association
Volume112
Issue number520
DOIs
Publication statusPublished - 2017 Oct 2
Externally publishedYes

Fingerprint

Average Treatment Effect
Weighted Bootstrap
Bootstrap
Estimator
Valid
Bias Correction
Normal Approximation
Bootstrap Method
Linear Forms
Asymptotic Approximation
Resampling
Covariates
Simulation Study
Average treatment effect
Matching estimators
Bootstrap inference

Keywords

  • Bootstrap
  • Matching
  • Treatment effect

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Bootstrap Inference of Matching Estimators for Average Treatment Effects. / Otsu, Taisuke; Rai, Yoshiyasu.

In: Journal of the American Statistical Association, Vol. 112, No. 520, 02.10.2017, p. 1720-1732.

Research output: Contribution to journalArticle

@article{cd9ce04712264227891845e72e62ff66,
title = "Bootstrap Inference of Matching Estimators for Average Treatment Effects",
abstract = "It is known that the naive bootstrap is not asymptotically valid for a matching estimator of the average treatment effect with a fixed number of matches. In this article, we propose asymptotically valid inference methods for matching estimators based on the weighted bootstrap. The key is to construct bootstrap counterparts by resampling based on certain linear forms of the estimators. Our weighted bootstrap is applicable for the matching estimators of both the average treatment effect and its counterpart for the treated population. Also, by incorporating a bias correction method in Abadie and Imbens (2011), our method can be asymptotically valid even for matching based on a vector of covariates. A simulation study indicates that the weighted bootstrap method is favorably comparable with the asymptotic normal approximation. As an empirical illustration, we apply the proposed method to the National Supported Work data. Supplementary materials for this article are available online.",
keywords = "Bootstrap, Matching, Treatment effect",
author = "Taisuke Otsu and Yoshiyasu Rai",
year = "2017",
month = "10",
day = "2",
doi = "10.1080/01621459.2016.1231613",
language = "English",
volume = "112",
pages = "1720--1732",
journal = "Journal of the American Statistical Association",
issn = "0162-1459",
publisher = "Taylor and Francis Ltd.",
number = "520",

}

TY - JOUR

T1 - Bootstrap Inference of Matching Estimators for Average Treatment Effects

AU - Otsu, Taisuke

AU - Rai, Yoshiyasu

PY - 2017/10/2

Y1 - 2017/10/2

N2 - It is known that the naive bootstrap is not asymptotically valid for a matching estimator of the average treatment effect with a fixed number of matches. In this article, we propose asymptotically valid inference methods for matching estimators based on the weighted bootstrap. The key is to construct bootstrap counterparts by resampling based on certain linear forms of the estimators. Our weighted bootstrap is applicable for the matching estimators of both the average treatment effect and its counterpart for the treated population. Also, by incorporating a bias correction method in Abadie and Imbens (2011), our method can be asymptotically valid even for matching based on a vector of covariates. A simulation study indicates that the weighted bootstrap method is favorably comparable with the asymptotic normal approximation. As an empirical illustration, we apply the proposed method to the National Supported Work data. Supplementary materials for this article are available online.

AB - It is known that the naive bootstrap is not asymptotically valid for a matching estimator of the average treatment effect with a fixed number of matches. In this article, we propose asymptotically valid inference methods for matching estimators based on the weighted bootstrap. The key is to construct bootstrap counterparts by resampling based on certain linear forms of the estimators. Our weighted bootstrap is applicable for the matching estimators of both the average treatment effect and its counterpart for the treated population. Also, by incorporating a bias correction method in Abadie and Imbens (2011), our method can be asymptotically valid even for matching based on a vector of covariates. A simulation study indicates that the weighted bootstrap method is favorably comparable with the asymptotic normal approximation. As an empirical illustration, we apply the proposed method to the National Supported Work data. Supplementary materials for this article are available online.

KW - Bootstrap

KW - Matching

KW - Treatment effect

UR - http://www.scopus.com/inward/record.url?scp=85027047408&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027047408&partnerID=8YFLogxK

U2 - 10.1080/01621459.2016.1231613

DO - 10.1080/01621459.2016.1231613

M3 - Article

AN - SCOPUS:85027047408

VL - 112

SP - 1720

EP - 1732

JO - Journal of the American Statistical Association

JF - Journal of the American Statistical Association

SN - 0162-1459

IS - 520

ER -