VT income tax difference with TAXSIM35

[martinholmer]

The following integration test fails. It is not yet known whether the
tax logic problem is in TAXSIM35 or in the current PEUS code, or
whether there are problems in both models. This topic is meant to
facilitate discussion of why this test fails, primarily by reporting
the results of hand calculations required to fill out the relevant tax
form.

- name: Tax unit with taxsimid 99952 in h21.its.csv and h21.ots.csv
  absolute_error_margin: 0.01
  period: 2021
  input:
    people:
      person1:
        age: 30
        employment_income: 130010
        taxable_interest_income: 5505.0
        short_term_capital_gains: 4505.0
        rental_income: 5005.0
        rent: 2000
        ssi: 0  # not in TAXSIM35
        state_supplement: 0  # not in TAXSIM35
        wic: 0  # not in TAXSIM35
        ma_covid_19_essential_employee_premium_pay_program: 0  # not in TAXSIM35
      person2:
        age: 30
        employment_income: 148010
        taxable_interest_income: 5505.0
        short_term_capital_gains: 4505.0
        rental_income: 5005.0
        ssi: 0  # not in TAXSIM35
        state_supplement: 0  # not in TAXSIM35
        wic: 0  # not in TAXSIM35
        ma_covid_19_essential_employee_premium_pay_program: 0  # not in TAXSIM35
    tax_units:
      tax_unit:
        members: [person1, person2]
        premium_tax_credit: 0  # not in TAXSIM35
        local_income_tax: 0  # not in TAXSIM35
        state_sales_tax: 0  # not in TAXSIM35
        ca_use_tax: 0  # not in TAXSIM35
        il_use_tax: 0  # not in TAXSIM35
        nm_2021_income_rebate: 0  # not in TAXSIM35
        nm_additional_2021_income_rebate: 0  # not in TAXSIM35
        nm_supplemental_2021_income_rebate: 0  # not in TAXSIM35
        ny_supplemental_eitc: 0  # not in TAXSIM35
        ok_use_tax: 0  # not in TAXSIM35
        pa_use_tax: 0  # not in TAXSIM35
        vt_renter_credit: 0  # not in TAXSIM35
    spm_units:
      spm_unit:
        members: [person1, person2]
        snap: 0  # not in TAXSIM35
        tanf: 0  # not in TAXSIM35
    households:
      household:
        members: [person1, person2]
        state_fips: 50  # VT
  output:  # expected results from patched TAXSIM35 2024-02-09 version
    vt_income_tax: 17870.45

[PavelMakarchuk]

@martinholmer This difference here is caused by a $5,000 difference in vt_agi. This is likely a result of TAXSIM calculating the Capital gain exclusion while PE does not.
The reason for that is the usage of the adjusted_net_capital_gain in the calculation which are net capital gains (lt - st) including some adjustments. However in the case above net capital gains should be 0 as it is long terms gains (0) reduced by short term gains.
Form

Is this a reasonable assumption or I am missing something here?
This is most reasonable explanation I could find based on the law

[martinholmer]

@PavelMakarchuk said in discussion topic #3836:

This difference here is caused by a $5,000 difference in vt_agi. This is likely a result of TAXSIM calculating the Capital gain exclusion while PE does not.

Agreed.

The reason for that is the usage of the adjusted_net_capital_gain [variable] in the [PEUS] calculation

But is that correct? I don't see how it can be correct because the adjusted_net_capital_gain variable includes qualified dividends, while VT explicitly rules those out as your screenshot shows.

There may well be problems with TAXSIM35, but this needs to be fixed in Policyengine-US.

[PavelMakarchuk]

You are right, I have created PR #3901 to address this issue.

[martinholmer]

Thanks.