Point dollar inputs at closer uprating series

[MaxGhenis]

Points dollar inputs at the series closest to what each one measures. Follows #9525 (docs) and #9526 (per-capita uprating), both merged; this branch is one commit on main.

The rule applied

An input follows, in order of preference: its own national series when the parameter tree has one; the income source it is part of; prices for consumption expenses and debts; CMS per-capita spending for health costs. Everything else keeps the default, CBO total AGI per capita. National totals are converted to per capita at runtime by #9526.

Two defects

• Unemployment compensation. calibration.gov.irs.soi.unemployment_compensation was extended by calibration.gov.cbo.income_by_source.taxable_social_security. calibration.gov.cbo.unemployment_compensation sat in the tree with no variable using it. The SOI series now grows with CBO's unemployment compensation outlays from 2022, the first year that series has, and that series is carried past 2036 like the other CBO anchors.
• Alimony income followed the SOI self-employment series. Alimony income and expense now follow calibration.gov.irs.soi.alimony_income, whose label said "SOI short-term capital gains" and is corrected.

Reassignments from the default

Inputs	Now follow
hourly_wage, tip_income, fsla_overtime_premium, W-2 wages of qualified businesses, desired 401(k), 403(b) and IRA contributions, military and government pay	SOI employment income
workers_compensation, disability_benefits	SOI employment income
self_employed_pension_contributions_desired	SOI self-employment income
Collectibles gains, unrecaptured section 1250 gain, non-Schedule D gains, capital gain distributions, small business stock gains, property sales gains, other net gain	SOI long-term capital gains
health_insurance_premiums	CMS per-capita out-of-pocket spending, as the other premium inputs already do
veterans_benefits	gov.ssa.uprating, the Social Security cost-of-living index
Rent, child care and care expenses, tuition, utilities, phone, broadband, homeowners insurance and association fees, transportation	CPI-U
Person-level home mortgage interest, student loan interest, investment interest expense	CPI-U, as the asset, debt and auto loan inputs already do

The dataset-extension override table already mapped the computed rent column to CPI-U; the pre_subsidy_rent input the default dataset carries now matches it.

Growth each series applies, from 2024

Series	2025	2030	2035
Default: CBO total AGI per capita	1.0592	1.2476	1.4728
SOI employment income per capita	1.0397	1.2406	1.4682
SOI long-term capital gains per capita	1.2584	1.1546	1.3174
SOI unemployment compensation per capita, new	1.1072	1.3581	1.5356
CBO taxable Social Security per capita, the old unemployment driver	1.1292	1.5198	1.9535
CMS per-capita out-of-pocket	1.0309	1.2261	1.4632
Social Security cost-of-living index	1.0249	1.1516	1.2867
CPI-U	1.0300	1.1501	1.2860

The largest changes are to rent and child care, which feed SPM resources: 3.0 percent in 2025 in place of 5.9 percent, and 28.6 percent by 2035 in place of 47.3 percent. Unemployment compensation grows 53.6 percent by 2035 in place of 95.4 percent.

Judgment calls

These are choices among the series the tree already holds, not corrections of errors. I flag them for review:

• CPI-U is all-items inflation. Rent and child care prices have their own CPI components, which the tree does not carry and CBO does not project.
• Wage-replacement benefits follow aggregate employment income per capita, which includes employment growth as well as wage growth.
• The SOI unemployment compensation level is the 2021 total, a pandemic year. Only its growth is used for uprating.
• The structured first and second mortgage balance and interest inputs stay on the default. Two are deprecated (Atomic inputs: one fact, one input — eliminate arbitration between redundant input variables #9275), and their file holds three deduction formulas outside the selective coverage job, so touching it left 30 formula lines uncovered in the project report.

Tests

• New tests/core/test_uprating_series.py: one case per reassigned variable, each series defined across the projection window, unemployment compensation growth equals CBO's for 2025 to 2036 and does not flatten afterwards, the alimony label, and rent matching its override.
• tests/core/test_per_capita_uprating.py now uses charitable donations as its default-series example, since rent moved to CPI-U.
• Partner contract tests: 620 passed locally with no expected-output changes.
• Run locally: tests/code_health, tests/microsimulation/data, tests/policy/contrib/ssa, tests/policy/baseline/parameters/test_uprating_extensions.py, tests/test_system_import.py. The full suite is left to CI.

I have not measured the effect on budgetary or poverty estimates.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (61fac14) to head (024e09a).
⚠️ Report is 185 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##              main     #9527    +/-   ##
==========================================
  Coverage   100.00%   100.00%            
==========================================
  Files            3        59    +56     
  Lines           78       510   +432     
==========================================
+ Hits            78       510   +432     

Flag	Coverage Δ
unittests	100.00% <100.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.