Fix WI retirement income exclusion applied after standard-deduction phaseout (#8817)

[DTrim99]

Fixes #8817.

Problem

Wisconsin's Act 15 retirement income exclusion (Schedule SB line 16, $24,000/person for age 67+) was applied after the standard-deduction phaseout. On Form 1 the line-16 subtraction reduces WI income (line 7), and the standard deduction (line 8) is looked up on that reduced income — which is larger. Applying the subtraction after the phaseout kept the standard deduction too small and over-taxed elderly filers with retirement income.

Root cause

wi_retirement_income_exclusion_tax computed wi_taxable_income − line16, but wi_taxable_income already baked in the standard deduction computed on the higher pre-subtraction WI income.

Fix

Add back the extra standard deduction the line-16 subtraction unlocks — algebraically equivalent to recomputing taxable income on the post-subtraction WI income, while keeping wi_taxable_income as the base (so existing behaviour below the phaseout threshold is unchanged). Factored the standard-deduction phaseout into a reusable wi_standard_deduction_for_income helper so both the normal path (on wi_agi) and the exclusion path (on wi_agi − line16) share one implementation.

Verification / tests

• New regression test (67+ MFJ, $120k pension, exclusion $48k): exclusion taxable income $53,651, exclusion tax $2,184 (was ~$2,602 pre-fix — the phaseout stayed keyed to $120k).
• Updated the affected E2E case (single 45k, exclusion 20k): tax 474 → 390, reflecting the recovered $2,400 of standard deduction.
• The five low-income unit tests are unaffected (income below the phaseout threshold ⇒ no extra deduction), and the age-85 high-income integration case is unchanged (income stays fully phased out even after the subtraction).

References

• 2025 Form 1 (lines 6-11) and Schedule SB line 16
• 2025 Schedule SB instructions

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (f9e58e7) to head (8433324).
⚠️ Report is 88 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main     #8818   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            3         2    -1     
  Lines           55        41   -14     
=========================================
- Hits            55        41   -14     

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.

[PavelMakarchuk]

Review: Fix WI retirement exclusion applied after standard-deduction phaseout

Nice fix — the ordering correction is algebraically exact and matches WI Form 1. Requesting changes on a couple of minor items (one real reference-format bug + some boundary test coverage); the core logic is sound.

Verified correct

• The fix is algebraically exact. taxinc − line16 − (SD(wi_agi−line16) − SD(wi_agi)) cancels the SD_full terms, leaving exactly Form 1's (wi_agi − line16) − SD(wi_agi − line16). Confirmed against the 2025 Form 1: line 6 (Schedule SB subtractions) → line 7 (Wisconsin income) → line 8 standard deduction "use the amount on line 7". Because the deduction phases down, subtracting the exclusion first correctly unlocks more deduction. (Sanity check: single $45k/$20k excl → extra deduction = SD(25k) − SD(45k) = $2,400, matching the 474→390 result.)
• Correct provision: Act 15 / Schedule SB line 16 ($24k/person age 67+, no AGI cap), correctly distinguished from the $5k line-17 subtraction (statute 71.05(6)(b)54m corroborated).
• Idiomatic pattern: the module-level wi_standard_deduction_for_income helper is the right choice (a Variable can't be evaluated at an arbitrary wi_agi − line16 input), reuses gov.states.wi.tax.income.deductions.standard params (no hardcoding), no circular-import risk, no behavior change to plain wi_standard_deduction, regression-free below the threshold. Changelog present (fixed).

Requested changes

1. wi_standard_deduction.py reference is five adjacent string literals with no commas, so Python concatenates them into one mashed, unclickable URL. Make it a comma-separated tuple. (Pre-existing, but the fix now leans on this variable.) While there, add a source that actually corroborates the phaseout the fix depends on — the 2025 Standard Deduction Table (2025-Form1-Inst.pdf#page=35, keyed to line 7) or Wis. Stat. 71.05(22).
2. Add a few boundary tests around the fix:
   • below-threshold no-regression (currently only implicit — the low-income unit tests stub wi_taxable_income and don't mention the extra-SD term);
   • the fully-phased-out top end (the "age-85 unchanged" claim isn't asserted anywhere);
   • the partial-unlock boundary — wi_agi past the phaseout end while wi_agi − line16 lands inside it (all current cases keep both endpoints in the linear region);
   • one-spouse-67+ joint ($24k per-person branch; all joint tests use two eligible spouses / $48k).

Suggestions

• Age 66/67 eligibility boundary case.
• The phaseout thresholds param cites efile.com (third-party) — prefer an official WI source (pre-existing).
• If the helper is expected to be reused beyond WI, a _helpers-style module would aid discoverability; co-locating in wi_standard_deduction.py is fine as-is.

🤖 Reviewed with Claude Code

[DTrim99]

Fixes applied (addressing @PavelMakarchuk's review)

Pushed as 3175ffa783.

Requested changes

• wi_standard_deduction.py reference — the five adjacent string literals had no commas and Python concatenated them into one unclickable URL. Made it a proper comma-separated tuple, and added two sources that corroborate the phaseout the fix leans on: the 2025 Form 1 instructions Standard Deduction Table (2025-Form1-inst.pdf#page=35, keyed to line 7) and Wis. Stat. § 71.05(22).
• Boundary tests — added four cases around the fix (all 2025):
  • below-threshold no-regression — pension $15k, both SDs at max → extra deduction $0 → tax $175 (pins that low income is unaffected);
  • fully phased-out top end — age 85, $200k → SD $0 for both full and reduced → extra $0 → $350 (the "age-85 unchanged" case);
  • partial unlock — full income $140k past the phaseout end (SD_full $0) while reduced income $116k lands inside it (SD_reduced $1,986) → $490.49 (a no-extra-deduction bug would give ~$571.88);
  • joint, one spouse 67+ — uses the $24k per-person cap, not the $48k both-eligible pool → amount $24,000, tax $210.

Suggestions

• Left the phaseout-threshold param's efile.com citation and the helper's location as-is (both noted as pre-existing / non-blocking).

CI running now.

[MaxGhenis]

@PavelMakarchuk both requested changes were addressed in 3175ffa (2026-07-02); I've now rebased the branch onto current main and re-verified. New head is 8433324.

Requested changes — resolved

1. Reference tuple. wi_standard_deduction.py's reference is now a proper comma-separated tuple (was five adjacent string literals concatenating into one unclickable URL). Added the two sources that corroborate the phaseout the fix leans on: the 2025 Form 1 instructions Standard Deduction Table (2025-Form1-inst.pdf#page=35, keyed to line 7) and Wis. Stat. § 71.05(22).

2. Boundary tests. Added the four cases you asked for in wi_retirement_income_exclusion_tax.yaml, each with the hand-computed derivation inline:

• Below phaseout start — pension $15k, SD at max for both full and reduced income → extra deduction $0 → tax $175 (pins that low income is regression-free; previously only implicit).
• Fully phased-out top end — age 85, $200k; both full ($200k) and reduced ($176k) past the single full-phaseout point → SD $0 for both → extra $0 → tax $350 (the "age-85 unchanged" claim, now asserted).
• Partial unlock — full income $140k past the phaseout end (SD_full $0) while reduced income $116k lands inside it (SD_reduced $1,986) → extra $1,986 → tax $490.49. A no-extra-deduction bug would give ~$571.88.
• Joint, one spouse 67+ — only the age-68 spouse eligible → line 16 = the $24k per-person cap, not the $48k both-eligible pool → amount $24,000, tax $210.

Rebase

main moved 92 commits since the merge-base. No overlap with this PR: zero WI files were touched on main in that range, and the rebase applied cleanly with no conflicts. (The WI-adjacent EITC merges #8875 and #8906 are Indiana and CA/CO/DC/IL/MN/WA respectively — neither touches Wisconsin.) PR now reports MERGEABLE.

Verification

Full WI test directory passes locally against current-main parameters: 165 passed across 42 files, including all 10 cases in wi_retirement_income_exclusion_tax.yaml. ruff format --check and ruff check clean on both touched Python files; CI Lint and changelog-fragment checks already green on 8433324, full suite running.

Suggestions

Left the phaseout-threshold param's efile.com citation and the helper's location in wi_standard_deduction.py as-is, per your note that both are pre-existing / non-blocking.