Assessing Racial Disparities in Healthcare Expenditures Using Causal Path-Specific Effects Restricted; Files Only

Abstract

Racial disparities in healthcare expenditures are well-documented, yet the underlying drivers remain complex and require further investigation. This study employs causal and counterfactual path-specific effects to quantify how various factors, including socioeconomic status, insurance access, health behaviors, and health status, mediate these disparities. Using data from the Medical Expenditures Panel Survey, we estimate how expenditures would differ under counterfactual scenarios in which the values of specific mediators were aligned across racial groups along selected causal pathways. A key challenge in this analysis is ensuring robustness against model misspecification while addressing the zero-inflation and right-skewness of healthcare expenditures. For reliable inference, we derive asymptotically linear estimators by integrating influence function-based techniques with flexible machine learning methods, including super learners and a two-part model tailored to the zero-inflated, right-skewed nature of healthcare expenditures.

Table of Contents

1 Introduction 1

2 MEPS data and sample description 7

2.1 Data Source 7

2.2 Variables 7

2.3 Sample description 9

3 Disparity definition, identification, and estimation 13

3.1 Path-specific effects as measures of disparity 13

3.2 Identification assumptions for path-specific effects 19

3.3 Estimation techniques and multiply robust estimators 21

4 Empirical analysis of the MEPS data 26

4.1 Implementation details 26

4.2 Empirical results 28

5 Simulation studies 35

5.1 Simulation 1: Asymptotic properties and robustness 35

5.2 Simulation 2: Finite sample performance 40

6 Discussion 44

Appendix A Proofs 46

A.1 Identification claims 46

A.2 Estimation claims  47

A.3 Inference claims 53

Appendix B Effect decomposition 57

B.1 Cumulative PSEs in MEPS data 60

Appendix C The responses in MEPS data 63

C.1 Geometric mean interpretation 63

C.2 Two-stage super learner 66

Appendix D Additional simulation 70

Bibliography 73

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Biostatistics
Subfield / Discipline	Biostatistics - MPH & MSPH
Degree	M.S.P.H.
Submission	Master's Thesis
Language	English
Research Field	Health Sciences, Public Health Biology, Biostatistics Statistics
Parola chiave	Healthcare expenditure disparities MEPS data Machine learning for causal inference Causal mediation analysis Semiparametric efficiency
Committee Chair / Thesis Advisor	Razieh Nabi, Emory University
Committee Members	David Benkeser, Emory University

Ultima modifica

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	File download under embargo until 15 November 2025	2025-04-29 11:14:18 -0400	File download under embargo until 15 November 2025

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions