Fix OBBBA itemized-limit reform validation (counterfactual must keep Pease)

[PavelMakarchuk]

Problem

On the reform-validation page, obbba_itemized_tax_benefit_limit (JCX-35-25 Ch.1 line 11) showed the wrong sign: populace +$7.1B vs JCT −$16.0B.

Root cause: the reform reverted both switches —

gov.irs.deductions.itemized.limitation.applies      → False
gov.irs.deductions.itemized.limitation.obbb.applies → False

Setting limitation.applies=False disables the itemized-deduction limitation entirely, so the counterfactual had no limitation. The provision was measuring "OBBBA cap vs nothing" (a revenue raise) instead of "OBBBA cap vs present-law Pease" (the JCT-scored cost).

What the provision actually is

OBBBA replaces the (TCJA-suspended, present-law-reinstated) Pease limitation with a weaker ~2/37 cap for top-bracket filers. PE-US's itemized_taxable_income_deductions_reduction variable uses obbb.applies to switch between the cap and Pease (it doesn't stack), so the correct counterfactual is limitation.applies=True, obbb.applies=False = present-law Pease. Pease is tougher than the cap, so OBBBA is a net cost.

Fix

Revert only obbb.applies, keeping limitation.applies=True (present-law Pease).

Verification (released populace_us_2024.h5, 2026)

	income tax
OBBBA baseline (cap)	2,320.3B
present-law Pease (new counterfactual)	2,333.3B
no limitation (old counterfactual)	2,317.2B

• old effect (cap vs nothing): +$3.0B standalone (+$7.1B stacked) — wrong sign
• new effect (cap vs Pease): −$13.0B — correct sign, matches JCT −$16.0B (residual is normal PE-vs-JCT)

Adds test_itemized_benefit_limit_counterfactual_keeps_pease so the applies=False regression can't return.

🤖 Generated with Claude Code