Medicaid Unwinding & Healthcare Access
A state-level causal program evaluation testing whether greater Medicaid disenrollment exposure during the post-pandemic unwinding was associated with larger increases in the uninsured rate among low-income adults.
Robustness results
| Specification | Effect per +10 pp exposure | p-value |
|---|---|---|
| Primary DiD | -0.095 pp | 0.858 |
| Population weighted | -0.256 pp | 0.668 |
| Procedural exposure | -0.012 pp | 0.984 |
| Placebo timing | -0.310 pp | 0.593 |
| 5th–95th percentile trim | -0.491 pp | 0.440 |
| Exclude 2021 pre-period | -0.015 pp | 0.978 |
| Exclude NC & SD expansion | -0.052 pp | 0.914 |
Pre-treatment diagnostics
The pre-trend evidence is not perfectly clean; the project therefore treats the causal interpretation as qualified rather than conclusive.
Event study
Year-specific exposure interactions relative to 2022. Pre-period estimates are diagnostics; 2023 is the transition year.
| Year | Effect per +10 pp exposure | p-value |
|---|---|---|
| 2018 | 0.552 pp | 0.417 |
| 2019 | 0.858 pp | 0.156 |
| 2021 | 0.791 pp | 0.028 |
| 2022 | Reference | — |
| 2023 | 0.062 pp | 0.742 |
| 2024 | 0.534 pp | 0.135 |
| 2025 | 0.376 pp | 0.366 |
Median state uninsured rate
| Year | Median rate |
|---|---|
| 2018 | 18.8% |
| 2019 | 20.1% |
| 2021 | 18.1% |
| 2022 | 16.8% |
| 2023 | 16.6% |
| 2024 | 18.2% |
| 2025 | 20.0% |
Methodology
Primary treatment: disenrollments divided by completed renewal outcomes during each state's initial 14-month unwinding window. Primary outcome: ACS uninsured rate among adults ages 19–64 below 138% of the federal poverty threshold. The primary model uses state and year fixed effects with state-clustered standard errors; 2023 is treated as a transition year.