Calibrate Microcosm UK to CenTax's parent income tabulations

changelog.d/uk-centax-childcare-parent-targets.added.md

@@ -0,0 +1,7 @@
+The UK national line now calibrates parents of 3- and 4-year-olds to CenTax's HMRC tabulations of their incomes (microcosm#1399, for the £100k childcare costing in scorecard#207).
+
+**A derived national target family, `centax_childcare_parents`.** `tools/build_uk_centax_childcare_parent_facts.py` reads Figures 3.1, 3.3, 3.6 and 3.7 of the workbook CenTax released with Advani, Flew, Pepin-Hall and Summers (2026), *Removing the childcare cliff-edge* (its SHA-256 is pinned), and writes nine counts at England grain to `uk/centax_childcare_parent_facts.json`: four adjusted net income bands that partition (0, £200k] in 2021-22 pounds, and five splits of (£80k, £150k] (women, a main income source other than employment, and London, the South excluding London and the North and Midlands). Each count is CenTax's published share times the baseline build's own parents, so the targets set the income distribution's shape and leave its level where the DfE caseloads put it. The edges are aged to the calibration year by the OBR average earnings index policyengine-uk carries. Nine contract targets (`microcosm.centax_childcare_parents.*`) bind them on the national line only. CenTax's model outputs (bunching estimates, frictions, elasticities, marginal bunchers and costs) are never targets; Figure 2.1, Figure 3.12 and Table A.1's levels stay diagnostics in the resource.
+
+**Two person measures.** `centax_parent_of_eligible_3_4_year_old` marks the adults of a benefit unit with a dependent child in England of DfE age 3 to 5 who is before school in at least one of the year's terms, on the DfE age and term share `uk_runtime/dfe_spring_term.py` reads. `centax_employment_main_income_source` marks people whose year's pay is positive and at least every other income source. The resolver also computes `adjusted_net_income_annual` on the annual-income simulation.
+
+**One loader for the derived facts.** `uk_runtime/derived_facts.py` merges the high-value home stock and the CenTax cells beside the Chronicle feed at every national compile site (the full and national target inputs, the release-cut certifier, the compile-parity receipts, the UC matrix diagnostic and the incumbent surface evaluation), and the target-input receipts and the reference membership report name both resources. `tools/generate_uk_target_references.py` streams the feed line by line instead of holding its text. `tools/diagnose_uk_centax_parents.py` reports each cell's fit and support (records, Kish effective households, top-ten household share) on a national H5.