Abstract/Description/Artist Statement
In recent years, policymakers have increasingly looked to Medicaid coverage expansions as a tool to improve population health. For instance, Medicaid expansions implemented as part of the Affordable Care Act (ACA) have been linked to decreases in mortality, improvements in self-reported health, and higher rates of health insurance coverage. However, less is known about how these expansions affect cancer incidence and mortality rates. In this paper, I estimate the effects of ACA Medicaid expansion on breast cancer incidence and mortality among nonelderly women ages 40–65.
Using state-years as the unit of observation, I construct a balanced panel of the 50 states and Washington, D.C. from 2010 to 2021 (N=612). I estimate modernized difference-indifferences (DiD) models that compare changes in cancer incidence and mortality rates across expansion and non-expansion states before and after the start of the ACA Medicaid expansion. Models include both traditional two-way fixed effects (TWFE) specifications as well as dynamic event-study estimates clustered by state. In addition, I employ methods from recent literature on staggered treatment adoption, including the Goodman–Bacon decomposition and the DID2S estimator. I perform placebo regressions of pre-treatment outcomes and event-study specifications to validate my identification strategy.
My results suggest that ACA Medicaid expansion is associated with reductions in breast cancer incidence, although I find no short-run effects on mortality. Placebo tests for preexisting differences in cancer incidence show no statistically significant differential pre-trends. Additionally, the coefficients on pre-expansion event times are consistently close to zero and statistically insignificant. Coefficients on Medicaid expansion from TWFE models are negative but imprecisely estimated. Results from event-study models are consistent with a gradual decrease in incidence following Medicaid expansion. This pattern could be driven by increases in screening and access to primary care induced by the acquisition of health insurance. Analyses using the Goodman–Bacon decomposition show that heterogeneous treatment effects bias against finding significant results in TWFE models, which may account for some of the limited precision in these estimates. DID2S estimates show qualitatively similar patterns to the event-study results. In contrast to the incidence results, I do not find any short-run effects of Medicaid expansion on mortality. The estimated coefficients are consistently small and statistically insignificant, and the event-study plot shows no apparent pattern following the policy’s implementation.
Keywords: Medicaid Expansion; Affordable Care Act (ACA); Breast Cancer; Differencein-Differences; Staggered Treatment Adoption
Faculty Advisor/Mentor
Prof. Jay Walker
Faculty Advisor/Mentor Email
j1walker@odu.edu
Faculty Advisor/Mentor Department
Department of Economics
College/School Affiliation
Strome College of Business
Student Level Group
Graduate/Professional
Presentation Type
Oral Presentation
Medicaid Expansion and Breast Cancer Outcomes: Evidence from Staggered Adoption under the Affordable Care Act
In recent years, policymakers have increasingly looked to Medicaid coverage expansions as a tool to improve population health. For instance, Medicaid expansions implemented as part of the Affordable Care Act (ACA) have been linked to decreases in mortality, improvements in self-reported health, and higher rates of health insurance coverage. However, less is known about how these expansions affect cancer incidence and mortality rates. In this paper, I estimate the effects of ACA Medicaid expansion on breast cancer incidence and mortality among nonelderly women ages 40–65.
Using state-years as the unit of observation, I construct a balanced panel of the 50 states and Washington, D.C. from 2010 to 2021 (N=612). I estimate modernized difference-indifferences (DiD) models that compare changes in cancer incidence and mortality rates across expansion and non-expansion states before and after the start of the ACA Medicaid expansion. Models include both traditional two-way fixed effects (TWFE) specifications as well as dynamic event-study estimates clustered by state. In addition, I employ methods from recent literature on staggered treatment adoption, including the Goodman–Bacon decomposition and the DID2S estimator. I perform placebo regressions of pre-treatment outcomes and event-study specifications to validate my identification strategy.
My results suggest that ACA Medicaid expansion is associated with reductions in breast cancer incidence, although I find no short-run effects on mortality. Placebo tests for preexisting differences in cancer incidence show no statistically significant differential pre-trends. Additionally, the coefficients on pre-expansion event times are consistently close to zero and statistically insignificant. Coefficients on Medicaid expansion from TWFE models are negative but imprecisely estimated. Results from event-study models are consistent with a gradual decrease in incidence following Medicaid expansion. This pattern could be driven by increases in screening and access to primary care induced by the acquisition of health insurance. Analyses using the Goodman–Bacon decomposition show that heterogeneous treatment effects bias against finding significant results in TWFE models, which may account for some of the limited precision in these estimates. DID2S estimates show qualitatively similar patterns to the event-study results. In contrast to the incidence results, I do not find any short-run effects of Medicaid expansion on mortality. The estimated coefficients are consistently small and statistically insignificant, and the event-study plot shows no apparent pattern following the policy’s implementation.
Keywords: Medicaid Expansion; Affordable Care Act (ACA); Breast Cancer; Differencein-Differences; Staggered Treatment Adoption