Medicaid and CHIP enrollment, Apr 2026

Medicaid enrollment is a high-value monthly program series: it resolves from administrative data, affects coverage and poverty forecasts, and gives repeated feedback on eligibility/take-up modeling. This target resolves on 2026-09-30 under a fixed-vintage rule, with an expected ~2-3 months lag. The same series can also spawn next release, +3 months, +12 months, threshold questions.

The central estimate holds enrollment close to its post-unwinding floor. The interval is wider than SNAP because state reporting and renewal policy create larger cross-state movement.

The resolver is the April 2026 updated-data national Total Medicaid and CHIP Enrollment value in CMS monthly Medicaid and CHIP enrollment data, converted from persons to millions. This is a national total, not a weighted average or a state row.

Base-rate/reference-class anchor: the latest official level is 74.294 million in March 2026, and CMS's release list shows comparable monthly files. I use March persistence as the model prior because the fast run did not fetch a clean sequence of earlier same-vintage national totals; using older catalog levels near 79 million would overweight stale pre-current-level information.

Level, momentum, and vintage split: March gives the level anchor; post-unwinding Medicaid/CHIP enrollment still appears to be drifting down, but by 2026 the extreme unwinding losses should be mostly over. The updated April vintage should be a little higher than a preliminary April print would be because updated data include retroactive and late-processed enrollment.

Policy-mechanism split: continuous eligibility for children, normal population growth, and retroactive processing support the level. Renewal redeterminations, income growth moving some adults out of eligibility, and state reporting cleanups pull down. I do not apply a large one-off shock because April 2026 is before any later community-engagement compliance pulse would affect enrollment mechanically.

Point: start from the March official persistence prior of 74.294361 million, subtract a judgmental 0.20 million for continued post-unwinding attrition, and add a judgmental 0.06 million for updated-vintage retroactive/late processing, giving 74.154361 million, rounded to 74.15. Interval: in the absence of a fetched same-vintage month-over-month sample in this fast run, use a quantitative uncertainty model with 1.0 million process standard deviation for monthly enrollment movement plus 0.5 million reporting/revision standard deviation; 1.28 * sqrt(1.0^2 + 0.5^2) = 1.43 million, rounded to an 80% band of 72.70 to 75.60 million.

Counter-consideration and scenarios: downside outside the interval would require April updated enrollment more than about 1.59 million below March, likely from broad state renewal drops or reporting cleanups. Upside outside the interval would require April updated enrollment more than about 1.31 million above March, likely from unusually large retroactive enrollment, state resubmissions, or a reporting-break rebound. The central case is near-flat to mildly down from March.

Review disposition: accepted the resolver critique by naming the updated-data national Total Medicaid and CHIP Enrollment field and adding a no-file fallback; accepted the model-prior and interval critiques by stating March persistence as the prior and using an explicit quantitative uncertainty model. The attrition and updated-vintage adjustments remain judgmental because the fast-run evidence did not fetch a clean recent same-vintage sequence.