Fix MT income tax applying preferential LTCG rates when net CG is negative

[PavelMakarchuk]

Summary

Fixes two interacting bugs in Montana's income tax calculation that caused $0 tax when it should be ~$20,837.

Closes #7859

Bug 1: Missing net capital gain check

mt_regular_income_tax_joint.py subtracted all positive LTCG from taxable income even when STCG losses exceeded LTCG. Per MT Form 2 and TaxAct, preferential LTCG rates should not apply when the overall net capital position is a loss.

Fix: Only subtract LTCG from taxable income when net_capital_gains (STCG + LTCG) > 0. Applied to both _joint and _indiv variants.

Bug 2: Person-level vs tax-unit mismatch in CG tax

mt_capital_gains_tax_joint was a Person-level variable, but for joint filers, LTCG and taxable income can be on different persons (taxable income is assigned to head via is_head *). Neither person had both LTCG and taxable income, so both got $0 CG tax.

Fix: Convert mt_capital_gains_tax_joint and mt_capital_gains_tax_applicable_threshold_joint from Person to TaxUnit entity, aggregating LTCG and taxable income across persons.

Test case

From TaxAct (policyengine-taxsim#755):

Variable	Before fix	After fix	TaxAct
MT Taxable Income	$361,508	$361,508	$361,508
MT Income Tax	$0	$20,837	$20,837

Test plan

• policyengine-core test policyengine_us/tests/policy/baseline/gov/states/mt/tax/income/integration.yaml policyengine_us/tests/policy/baseline/gov/states/mt/tax/income/capital_gains/ -c policyengine_us

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (d8084e1) to head (ed2e02d).
⚠️ Report is 49 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff              @@
##             main     #7862       +/-   ##
============================================
+ Coverage   71.81%   100.00%   +28.18%     
============================================
  Files        4084         6     -4078     
  Lines       58886       138    -58748     
  Branches      288         4      -284     
============================================
- Hits        42288       138    -42150     
+ Misses      16583         0    -16583     
+ Partials       15         0       -15     

Flag	Coverage Δ
unittests	100.00% <100.00%> (+28.18%)	⬆️

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

☔ View full report in Codecov by Sentry.
📢 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.

[hua7450]

Review: Verified against MT Form 2 (2024) and MCA 15-30-2103

The core fix (no preferential LTCG rates when net CG <= 0) is correct per Form 2 Line 1: "If you do not have a net long-term capital gain, skip lines 2 through 10 and enter 0 (zero) on line 11."

Two issues found:

1. "Lesser of line 15 or line 16" not fully implemented

The Form 2 instructions (p. 11) say Line 2 should be:

"the lesser of federal Schedule D, line 15 or Schedule D, line 16"

• Schedule D Line 15 = net LTCG only
• Schedule D Line 16 = net STCG + net LTCG (combined)

The PR currently uses max_(ltcg, 0) (i.e., full net LTCG) when net_cg > 0. This doesn't match the "lesser of" rule when there are partial short-term losses:

Scenario	Line 15 (net LTCG)	Line 16 (combined)	Form 2 Line 2 should be	PR uses
LTCG=50K, STCG=-30K	50,000	20,000	20,000 (lesser)	50,000

Suggested fix: min_(max_(ltcg, 0), max_(net_cg, 0)) instead of max_(ltcg, 0) in all three files (mt_regular_income_tax_joint, mt_regular_income_tax_indiv, mt_capital_gains_tax_joint).

2. Same bug in mt_capital_gains_tax_indiv.py

The _indiv variant still uses LTCG directly without the net_cg > 0 check. Impact is limited to married-filing-separately-on-same-return, but it's the same logical bug. Worth fixing here or filing a follow-up.

Sources:

• MCA 15-30-2103
• 2024 MT Form 2
• 2024 MT Form 2 Instructions (p. 11)
• 26 USC 1222

[DTrim99]

Program Review: PR #7862 - Fix MT income tax applying preferential LTCG rates when net CG is negative

Source Documents

• MT Form 2 Instructions (2023): Page 6
• MT Form 2 (2024): Page 2
• Validation: TaxAct comparison confirms expected MT Income Tax = $20,837

Bugs Fixed

Bug 1: Missing net capital gain check

• mt_regular_income_tax_joint.py subtracted all positive LTCG from taxable income even when STCG losses exceeded LTCG
• Per MT Form 2, preferential LTCG rates should not apply when the overall net capital position is a loss
• Fix: Only subtract LTCG when net_capital_gains (STCG + LTCG) > 0

Bug 2: Person-level vs TaxUnit mismatch in CG tax

• mt_capital_gains_tax_joint was a Person-level variable
• For joint filers, LTCG and taxable income can be on different persons
• Neither person had both values, so both got $0 CG tax
• Fix: Convert to TaxUnit entity, aggregate LTCG and taxable income across persons

Critical (Must Fix)

None identified.

Should Address

None identified.

Suggestions

None - this is a well-implemented bug fix.

Files Changed

Type	Files
Variables	mt_capital_gains_tax_applicable_threshold_joint.py, mt_capital_gains_tax_joint.py, mt_regular_income_tax_indiv.py, mt_regular_income_tax_joint.py
Tests	mt_capital_gains_tax_applicable_threshold_joint.yaml, mt_capital_gains_tax_joint.yaml, integration.yaml
Changelog	fix-mt-ltcg-negative-net-gains.fixed.md

Test Coverage

New test scenarios added:

1. Net CG negative: LTCG=50K, STCG=-80K → no preferential treatment, $0 CG tax
2. Positive net CG with STCG losses: LTCG=50K, STCG=-20K → preferential rates apply
3. Cross-person aggregation: LTCG on head, wages on spouse → proper TaxUnit aggregation
4. Both spouses with LTCG: Combined LTCG aggregation at TaxUnit level
5. Issue MT income tax incorrectly applies preferential LTCG rates when net capital gains are negative #7859 reproduction: Exact scenario from bug report validates fix

Validation Summary

Check	Result
Regulatory Accuracy	✅ Logic matches MT Form 2 instructions
Code Patterns	✅ Proper use of add(), where(), entity levels
Test Coverage	✅ Comprehensive (4 new integration tests, 2 new unit tests)
CI Status	✅ All checks passing
Changelog	✅ Present

Review Severity: APPROVE

This is a well-implemented bug fix with:

• Proper validation against TaxAct ($20,837 expected vs actual)
• Comprehensive test coverage for edge cases
• Correct entity-level changes for joint filing scenarios
• Clear comments explaining the logic

🤖 Generated with Claude Code

[DTrim99]

Minor Code Quality Suggestion

The code validator found that an existing variable net_capital_gains (in policyengine_us/variables/household/income/tax_unit/net_capital_gains.py) already calculates long_term_capital_gains + short_term_capital_gains.

Optional improvement: Replace the inline calculation:

ltcg = add(tax_unit, period, ["long_term_capital_gains"])
stcg = add(tax_unit, period, ["short_term_capital_gains"])
net_cg = ltcg + stcg

With:

ltcg = add(tax_unit, period, ["long_term_capital_gains"])
net_cg = tax_unit("net_capital_gains", period)

This applies to mt_capital_gains_tax_joint.py and mt_regular_income_tax_joint.py. The Person-level files (_indiv) would need to stay as-is since net_capital_gains is TaxUnit-level.

This is not blocking - just a maintainability improvement. The PR is approved as-is.