Mtr e26270

[codykallen]

Adds calculation of MTR on e26270 (partnership and S-corp income)

[codecov-io]

Current coverage is 98.27% (diff: 100%)

Merging #987 into master will increase coverage by <.01%

@@             master       #987   diff @@
==========================================
  Files            35         35          
  Lines          2085       2089     +4   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits           2049       2053     +4   
  Misses           36         36          
  Partials          0          0          

Powered by Codecov. Last update afd3d4b...60e298a

[codykallen]

I tested this with a reform of pass-through rates:

reform_pt = { 2017: {
'_PT_rt5': [0.3],
'_PT_rt6': [0.3,
'_PT_rt7': [0.3]
}
}

I calculated average MTRs on e02000 and e26270 under the baseline and reform, weighting by the absolute value of the income type.
Baseline MTR, e02000: 32.77%
Baseline MTR, e26270: 30.60%
Reform MTR, e02000: 29.08%
Reform MTR, e26270: 25.34%

[jdebacker]

@codykallen Why add finite_diff to e02000 instead of e26270 here? Seems like adding to e26270 is what would help this MTR enter functions.SchXYZ() in a way that accounts for potential differences between rates on ordinary income and pass-through income.

[codykallen]

finite_diff is already being added to e26270 in the line:
setattr(self.records, variable_str, variable + finite_diff)

However, e26270 is included in e02000, so you have to increase e02000 as well. If not, then you would be calculating the MTR from adding finite_diff to e26270 but substracting it from rental income.

[jdebacker]

Ah, I see. Thanks @codykallen, LGTM then.

[jdebacker]

@codykallen - thanks for opening this, it'll be helpful give the updated to pass-through income logic in taxcalc.

[martinholmer]

@codkallen, Your pull request #987 looks fine except for one thing: you need to add a test to prevent a decline in code coverage. Adding something like this at the end of the test_Calculator_mtr function (in the taxcalc/tests/test_calculate.py file) would be the minimal test:

    (_, _, mtr_combined) = calc.mtr(variable_str='e26270')
    assert type(mtr_combined) == np.ndarray

@MattHJensen @Amy-Xu @GoFroggyRun @andersonfrailey

[martinholmer]

@codykallen, Thanks for adding the test of the new mtr logic in pull request #987.

However, I don't understand why you have added the extra variables to the Records.VALID_READ_VARS set. You're adding four variables: 'FDED', 'e11550', 'e11070', 'e11580'. Why? None of these four are in the current puf.csv file.

[martinholmer]

@codykallen, Thanks so much for pull request #987, which has been merged into master branch.