Primary Submission Category: Applicants in Social Sciences
Examining Racial Disparities in Healthcare Expenditures via Causal Path-Specific Effects
Authors: Xiaxian Ou, Xinwei He, Razieh Nabi,
Presenting Author: Xiaxian Ou*
Racial disparities in healthcare expenditures have been widely documented, yet the underlying drivers remain complex and require further exploration. This study employs causal path-specific effects to assess how different factors contribute to the observed differences. Leveraging data from the Medical Expenditures Panel Survey, we examine the roles of socioeconomic status, insurance access, health behaviors, and health status. Our pathway-specific framework provides detailed assessments of how expenditures would change if specific mediating factors were counterfactually aligned across racial groups, offering a structured approach to quantifying sources of disparities. We also address several challenges in measuring the pathway-specific disparities. The relationships between race, healthcare spending, and mediating factors are complex, necessitating robustness against model misspecification. Furthermore, healthcare expenditures are characterized by zero-inflation and right skewness, requiring specialized modeling approaches. In addition, for reliable inference, it is essential to quantify uncertainty, ensuring that our estimators exhibit desirable statistical properties such as asymptotic normality and $sqrt{n}$-consistency. To address these complexities, we analyze the MEPS data using robust influence function-based estimators and integrate flexible statistical and machine learning methods, including super learners and a two-part model for zero-inflated skewed expenditures.