Add tax years 2011 and 2012 to policy_current_law.json

[jdebacker]

This PR adds 2011 and 2012 tax law to the policy_current_law.json file.

The PR is currently incomplete, but I am opening it so the others can feel free to review changes as I make them.

[jdebacker]

This PR will be ready for review following the resolution of unit testing issues noted in Issue #2525.

[jdebacker]

Note that Travis CI doesn't seem to be reported in the checks here, but it is passing on 3.6 and failing on 3.7 and 3.8. The test that fails is one noted in Issue #2525, tests/test_cpscsv.py::test_agg.

[MattHJensen]

@jdebacker, thanks a lot for this pull request. It will be a big help to some of Tax-Calculator's users.

Are you open to adding an internet-taxsim validation exercise for 2011 and 2012 as a prereq for merging this?

Tax-Calculator's current validation materials are here and @chusloj has a related PR here.

[jdebacker]

@MattHJensen Yes, I'm glad to try to add some validation exercises. I think those would be very important as a check on updates to the policy parameters. Thanks for pointing me towards those resources.

[jdebacker]

I'm not getting zero diffs for 2011 and 2012. I've been able to diagnose a couple errors from the validation process, but have been spinning my wheels with another. In 2011, for filers who are old (>65) and with dependents, I'm getting small (<$20) diffs in iitax that are mirrored in the c62100 output variable. I've searched all kinds of age related parameters but can't find an issue. Any suggestions on the parameters to look for discrepancy in @MattHJensen @Peter-Metz @chusloj?

[jdebacker]

@feenberg I'm trying to validate these additions of 2011 and 2012 tax law against TAXSIM27. I've spent a fair bit of time and can't find the root of some differences that seem related to the AMT in 2011. Would it be possible to get the values of intermediate variables from TAXSIM to help with my validation efforts? The input file I'm using is attached here.

Thanks!

a11.in.txt

[feenberg]

I don't see an attachment. Dan 617-8630343

…

On Thu, 28 Jan 2021, Jason DeBacker wrote: @feenberg I'm trying to validate these additions of 2011 and 2012 tax law against TAXSIM27. I've spent a fair bit of time and can't find the root of some differences that seem related to the AMT in 2011. Would it be possible to get the values of intermediate variables from TAXSIM to help with my validation efforts? The input file I'm using is attached here. Thanks! a11.in.txt ? You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, orunsubscribe.[AB55AVJNZGFJ4BYANY6ZUJ3S4HI4TA5CNFSM4VDN7RX2YY3PNVWWK3TUL52HS4 DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOFXODZZQ.gif]

[jdebacker]

@feenberg I just followed up via email - perhaps that will be easier. Thanks!

[jdebacker]

Update on the validation process: I was able to get the "a" file to match for 2011, but it required setting EITC_ps_MarriedJ to the following:

{
                "year": 2011,
                "EIC": "0kids",
                "value": 5000.0
            },
            {
                "year": 2011,
                "EIC": "1kid",
                "value": 5000.0
            },
            {
                "year": 2011,
                "EIC": "2kids",
                "value": 5000.0
            },
            {
                "year": 2011,
                "EIC": "3+kids",
                "value": 5100.0
            },

These values are inconsistent with the values TPC reports and from what is on the 2011 edition of Publication 596. Both those sources suggest the value should be 5080 for all numbers of qualifying children.

[MattHJensen]

Thanks for the update @jdebacker. It is nice to have a path to test the b file now.

[MattHJensen]

@jdebacker, when you have an opportunity, could you update us on the status of this PR?

[jdebacker]

@MattHJensen This validation is really time consuming and I haven't returned to it since my last comment - which means I still need to validate the "b" and "c" files for 2011 and 2012.

I was also kind of hoping that PR #2453 would be finished in the meantime for two reasons: (1) I think it will greatly improve the validation workflow and (2) TAXSIM 32 has fixed some bugs from TAXSIM27 and given my noticing an issue with the EITC parameters above it'd be really nice to have the most accurate calculator to validate against.