Federal poverty guideline projections lead the 42 U.S.C. 9902(2) CPI-U interval by one year #10451

Found in @PavelMakarchuk's review of #10227 (G2 and "Out of scope"). This is pre-existing, from #5911.

The lag

42 U.S.C. 9902(2) updates the poverty guidelines by the CPI-U change "during the annual or other interval immediately preceding the time at which the revision is made". The HHS notices state that interval as calendar years Y-2 to Y-1. The 2026 notice applies a 2.63% increase between calendar years 2024 and 2025.

gov/hhs/fpg.yaml uprates with gov.hhs.uprating, and that index holds the calendar-year average CPI-U for the labeled year. So projected guideline year Y grows by CY(Y) / CY(Y-1), one year ahead of the statute.

Pavel's back-cast from the 2022 guidelines:

• With the statutory lag, it reproduces the published 2023–2026 four-person guidelines within 0.10%.
• With the current mapping, it misses them by -3.7% to -5.6%.

Fix designs (from the review)

• A. Redefine gov.hhs.uprating so that instant Y holds CY(Y-1).
• B. Add a new lagged index used only by gov/hhs/fpg.yaml.
• B'. Add cadence metadata on fpg.yaml, with 2026–2036 at CBO levels.

Design B leaves the Head Start and Early Head Start partner pins unchanged. Designs A and B' move them: under A, 2025 moves +1.46% and 2026 +2.07%. Under every design, the 2027 Oregon ERDC and California AB 2591 expectations move.

The 2026 hold

gov.hhs.uprating keeps 2026 = 328.4, CBO's January 2025 projection, because Head Start and Early Head Start spending uprate from it. The partner contract tests under partners/analytics_coverage/**/head_start and **/early_head_start pin 2026 per-slot amounts derived from it. Ending that hold is a partner-facing change, so it needs the partner gate (CLAUDE.md) before any partner file is edited.

Plan

• The guideline lag, with design B: a follow-up PR adds an FPG-only lagged index and leaves the Head Start index and its partner pins unchanged.
• The 2026 Head Start hold: decide separately, through the partner gate.