ACA Medicaid Expansion: Causal Effects & HEOR Translation

Staggered difference-in-differences (Callaway–Sant'Anna), with naive TWFE and Sun–Abraham for contrast, translated into ICER / QALY / budget-impact terms with a probabilistic sensitivity analysis.

Event study

Estimator comparison

CEAC explorer

Budget-impact calculator

Outcome

Pre-treatment leads (event time < −1) should be flat and near zero; this is the make-or-break parallel-trends diagnostic. The shaded band is the wild-cluster bootstrap 95% CI. Reference period (−1) is normalised to zero.

Callaway–Sant'Anna (headline) vs naive two-way fixed effects. Under staggered timing TWFE averages in "forbidden" already-treated-as-control comparisons; the Goodman-Bacon decomposition below quantifies their weight.

Willingness-to-pay (per QALY):

Eligible population: Uptake rate:

Total annual incremental medical cost = newly-covered × per-adult MEPS cost. The QALY gain rests on the self-rated-health effect, which fails the pre-trend test, so the cost side is firm while the QALY side is not credibly identified, exactly as the analysis concludes.