Pin Michigan 2026 income tax rate

[PavelMakarchuk]

Summary

• add the official TY2026 Michigan individual income-tax rate determination as an explicit annual parameter anchor
• cite the Michigan Department of Treasury taxpayer notice issued April 15, 2026
• protect the dated anchor and ordinary pre-credit calculation with focused regression tests

This corrects the underlying PolicyEngine parameter tree; it does not add an oracle-specific override, formula branch, or case-based workaround.

Official source: https://www.michigan.gov/treasury/reference/taxpayer-notices/2026/04/15/425-income-tax-rate-for-individuals-and-fiduciaries-in-2026-tax-year

Checks

• UV_PROJECT_ENVIRONMENT=/Users/pavelmakarchuk/policyengine-us/.venv UV_NO_SYNC=1 make format (ruff format and ruff check passed; two unrelated base formatter changes restored exactly)
• python -m pytest -c pyproject.toml -q policyengine_us/tests/test_parameter_files.py (7 passed)
• focused policyengine-core YAML file (3 passed)
• full Michigan income-tax YAML suite (280 passed)
• git diff --check
• independent review: clean, no actionable findings

No partner contract test or formula was changed.

[PavelMakarchuk]

@MaxGhenis This is ready for maintainer review: the exact head c336f50cb0 is 31/31 green, its independent review was clean, and the change is limited to the official Michigan TY2026 rate anchor plus focused regressions.

[DTrim99]

Program Review — PR #9164 (Pin Michigan 2026 income tax rate)

Author: PavelMakarchuk

Source Documents

• MI Dept. of Treasury taxpayer notice (April 15 2026) — "4.25% Income Tax Rate for Individuals and Fiduciaries in 2026 Tax Year" · corroborated by the signed Treasury rate letter ("the rate in effect under Section 51 for the 2026 tax year is 4.25%")

Branch Status

⚠ 6 commits behind main (1 ahead). Minor; review scoped to the merge-base diff.

Summary

Clean single-value anchor. 0 critical, 0 should-address. The MI 2026 individual income-tax rate is pinned at 0.0425 (4.25%), correct per the Treasury notice, correctly dated with all prior years intact, and properly referenced. CI green (31/31).

Value audit

• 2026 rate = 0.0425 — confirmed against the Treasury notice title and the signed rate letter. Critically, the MCL 206.51(1)(c) formulary-reduction trigger was evaluated and did not fire for 2026 (FY2025 GF/GP revenue −1.56% vs 2.70% inflation), so 4.25% is the actually-in-effect rate — neither a stale 4.25% nor an erroneously-reduced value.
• Prior years intact: 2007 (4.35%), 2013/2024/2025 (4.25%), and the 2023 temporary 4.05% are all untouched; the diff is two additive hunks (one value line + one reference).
• Reference resolves (HTTP 200), title matches, proper {title, href} dict form; HTML source so no #page= needed.
• Scalar parameter, no formula/structure change — mi_taxable_income × rate flows the 0.0425 through directly.

Critical / Should Address

None.

Suggestions (optional)

• The 2026 test case (mi_taxable_income: 20_000 → 850) is a byte-for-byte echo of the existing 2022 case. It's a valid regression guard (a wrong 2026 rate changes the output), but a distinct income level (e.g. 50,000 → 2,125) would add representative variety.
• .fixed changelog type is defensible (the 2026 rate equals 2025's, so this pins/confirms); .added would also be reasonable for new-year coverage.
• The test_parameter_files.py addition (test_mi_2026_income_tax_rate_uses_official_annual_determination) is a purely-additive strengthening guard — it ties the 2026 value to its citation. Good.

Validation Summary

Check	Result
Regulatory / Value Audit	0.0425 confirmed (trigger did not fire); prior years intact
Reference Quality	Treasury notice cited, resolves, title matches; correct dict form
Code Patterns	0 critical / 0 should — clean 2026-01-01 anchor, correct format, test_parameter_files.py strengthened
Test Coverage	2026 pre-credit computation pinned (20,000 → 850); flat-rate coverage sufficient
CI Status	Passing (31/31)

Review Severity: APPROVE

A correct, minimal, well-sourced rate pin with an added parameter-guard test. No blocking or should-address items — only optional variety/wording notes. Rebase off the 6-commit lag before merge.

Review generated with Claude Code via /review-program