Federal gross income omits gambling winnings but still deducts wagering losses; OBBBA 90% wagering loss limit (2026+) not modeled #9637

Summary

gambling_winnings is a person input, but it is not in gov.irs.gross_income.sources, so it never reaches irs_gross_income or adjusted_gross_income. wagering_losses_deduction (the smaller of gambling_losses and gambling_winnings) is still in gov.irs.deductions.deductions_if_itemizing, so it comes off taxable income whenever the unit itemizes. Gambling winnings are therefore never taxed federally, and a filer with gambling losses can itemize and deduct them against wages.

Separately, wagering_losses_deduction has no 90% factor for tax years beginning after 2025, which P.L. 119-21 § 70114 requires.

Found while adding Arkansas's gambling rules (#9636). That PR changes only Arkansas; this issue is the federal side.

Reproduction (upstream/main ca1805c, 2026-09-25)

Single filer, 2024, TX, age 40, employment_income 60,000; gambling inputs on the person.

The same cases with the winnings also entered as miscellaneous_income, to stand in for counting them in gross income:

PE understates federal income tax by $2,025 and $5,189 in these cases. In the second case a $5,000 net gambling gain lowers tax by $1,423 relative to no gambling. A 2026 run of the 30,000 / 25,000 case gives wagering_losses_deduction = 25,000, not 22,500.

Legal basis

• 26 U.S.C. § 61(a): gross income means all income from whatever source derived. https://uscode.house.gov/view.xhtml?req=granuleid:USC-prelim-title26-section61&num=0&edition=prelim
• Schedule 1 (Form 1040) line 8b, "Gambling", is other income: https://www.irs.gov/pub/irs-prior/f1040s1--2024.pdf#page=1 and https://www.irs.gov/pub/irs-prior/f1040s1--2025.pdf#page=1. The 2024 Form 1040 instructions for line 8b say gambling winnings include lotteries and raffles: https://www.irs.gov/pub/irs-prior/i1040gi--2024.pdf#page=85
• 26 U.S.C. § 165(d) allows wagering losses only to the extent of wagering gains. The 2024 Schedule A instructions (line 16) allow gambling losses only up to the winnings reported on Schedule 1 line 8b: https://www.irs.gov/pub/irs-prior/i1040sca--2024.pdf#page=11
• P.L. 119-21 § 70114(a) rewrites § 165(d)(1): the deduction "(A) shall be equal to 90 percent of the amount of such losses during such taxable year, and (B) shall be allowed only to the extent of the gains from such transactions during such taxable year." New (d)(2) permanently counts other expenses of carrying on wagering as wagering losses. § 70114(b): taxable years beginning after December 31, 2025. https://www.govinfo.gov/content/pkg/PLAW-119publ21/html/PLAW-119publ21.htm ; current § 165(d): https://uscode.house.gov/view.xhtml?req=granuleid:USC-prelim-title26-section165&num=0&edition=prelim . P.L. 119-108 (2026-09-11) amended § 165 too, but only subsection (h).

What the code does now

• parameters/gov/irs/gross_income/sources.yaml has no gambling_winnings, and none of its 21 commits since the 2022 rename ever added it. The variables gambling_winnings, gambling_losses and wagering_losses_deduction were added together in Kentucky Itemized Deductions #3037 (merged 2024-01-06), and none has changed since.
• wagering_losses_deduction = min_(gambling_losses, gambling_winnings) summed over the tax unit, with no parameter.
• deductions_if_itemizing includes wagering_losses_deduction in its 2018, 2025 and 2029 lists. reforms/crfb/senior_deduction_extension.py hardcodes a copy of the 2029 list that includes it too.
• It is not in gov.irs.deductions.itemized_deductions (the Schedule A total), so it is outside total_itemized_taxable_income_deductions, itemized_taxable_income_deductions and itemized_deductions_less_salt.
• tax_unit_itemizes: with branch_to_determine_itemization on (the default), the choice compares income_tax on an itemizing branch, which includes wagering losses; that is why the 30,000 / 25,000 case itemizes although itemized_taxable_income_deductions (1,459) is below the standard deduction (14,600). On the non-branching path, and on a tie within $0.01, the comparison is itemized_taxable_income_deductions > standard_deduction, which leaves wagering losses out (Add QBID and wagering losses to the non-branching itemization choice #5977, closed as not planned).

Proposed fix

1. Add gambling_winnings to gov.irs.gross_income.sources, citing § 61 and Schedule 1 line 8b.
2. Add a wagering loss share parameter, 1.0 through 2025 and 0.9 from 2026-01-01 (P.L. 119-21 § 70114), so the deduction is min_(share * losses, winnings).
3. Decide separately whether wagering_losses_deduction belongs in the Schedule A total. That changes the base of itemized_taxable_income_deductions_reduction (check § 68 first) and about 20 states that build on the federal itemized lists.
4. Tests: YAML cases for 2024 and 2026 matching the tables above, plus property tests:
   • 0 ≤ deduction ≤ winnings, and deduction ≤ losses (≤ 0.9 × losses from 2026);
   • with losses fixed, income_tax never falls as gambling_winnings rises;
   • adding equal winnings and losses never lowers income_tax. The current code breaks this one.

Why this is its own change

• Blast radius. Federal AGI and gross income feed most state income taxes and MAGI. git grep finds adjusted_gross_income, adjusted_gross_income_person, irs_gross_income or positive_agi in the tax/income code of 43 state jurisdictions (a reference does not always mean the state starts from federal AGI, so each needs a look). medicaid_magi is built from adjusted_gross_income, and aca_magi from medicaid_magi. States that exempt some winnings (for example their own lottery prizes) would need subtractions.
• States with their own gambling treatment.
  • PA: pa_total_taxable_income is irs_gross_income minus PA-nontaxable sources, so it would start picking up gross winnings. PA taxes net gambling and lottery winnings (2024 PA-40 line 8 via Schedule T, where line 5 is winnings minus the cost of tickets and bets, floored at zero, with noncash PA Lottery prizes excluded: https://www.pa.gov/content/dam/copapwp-pagov/en/revenue/documents/formsandpublications/formsforindividuals/pit/documents/2024/2024_pa-40t.pdf#page=1). PA needs its own cost offset in the same change.
  • NJ already lists gambling_winnings in its gross income sources and does not use gambling_losses.
  • PA's property tax/rent rebate income and MI household resources already count gambling_winnings.
  • Arkansas has its own rule (Count gambling winnings in Arkansas gross income and add the Arkansas gambling loss deduction #9636): winnings are Arkansas gross income and losses are an Arkansas itemized deduction up to winnings, independent of § 165(d), so the 90% federal change must not reach it.
• Microdata. microcosm's engine ABI lock lists both gambling_winnings and gambling_losses as declared_absent_engine_input (engine default 0), and the archived policyengine-us-data never mapped them at its final commit. So the change should not move aggregates today; this rests on the microcosm manifest, not on an inspection of the released default H5. A future imputation must avoid double counting with miscellaneous_income: the PUF path maps E01200 (Form 4797 other gains per Tax-Calculator's codebook) and the CPS path maps ASEC OI_VAL, which includes code 19 "Anything else".

Related

• Kentucky Itemized Deductions #3037 / Kentucky Itemized Deductions #3035: added the gambling variables and the federal deduction.
• Add QBID and wagering losses to the non-branching itemization choice #5977: the non-branching itemization comparison omits QBID and wagering losses (closed, not planned).
• Add miscellaneous income #1964: added miscellaneous_income; its body guessed gambling winnings would be taxed through it.
• Count gambling winnings in Arkansas gross income and add the Arkansas gambling loss deduction #9636: Arkansas gambling winnings and losses.