INCORRECT reform file for 2017 (Brown-Khanna) GAIN Act that expands EITC

[martinholmer]

NOTE: this pull request incorrectly specifies the GAIN Act as pointed out below by @MattHJensen.
Therefore, this pull request is being closed without being merged and is replaced by pull request #1555.

This pull request adds the taxcalc/reforms/BrownKhanna.json reform file that implements the EITC reform described in issue #1552.

Using tc puf.csv YEAR --reform taxcalc/reforms/BrownKhanna.json --tables produces the following change-in-tax-liability results for calendar YEAR = 2017 . . . 2026:

2017 -80.5
2018 -82.8
2019 -85.1
2020 -87.4
2021 -89.9
2022 -92.5
2023 -95.3
2024 -98.1
2025 -101.0
2026 -104.0

These static results imply a ten-year total cost (that is, reduction in income tax revenue) of about $917 billion. When leaving the minimum eligibility age for the childless unchanged in the GAIN Act, the ten-year total cost is about $868 billion, which implies that this reform provision (reducing the minimum age from 25 to 21) costs about $49 billion.

These cost estimates are noticeably smaller than the TPC estimates which have the ten-year cost of the GAIN Act at about $1400 billion and the cost of reducing the minimum age at about $100 billion. TPC provides no details on these two estimates; the only detailed model estimates pointed to in their report are for other kinds of EITC reforms that were analyzed in June 2017. Footnote 2 in that June 2017 analysis says "Revenue estimates include the effects of microdynamic responses." There is no further explanation of what those "microdynamic responses" to EITC reforms are, but presumably the Brown-Khanna GAIN Act is simulated with the same assumptions about "microdynamic responses". So, it is possible that the TPC revenue estimates of the GAIN Act are computed assuming reductions in earnings caused by the EITC expansion. But again, there is no way to be certain about this because TPC does not reveal anything about how they estimated the revenue effects of the GAIN Act.

Here are some details for calendar year 2017 of the Tax-Calculator static analysis of the GAIN Act.

Weighted Tax Reform Totals by Baseline Expanded-Income Decile
    Returns    ExpInc    IncTax    PayTax     LSTax    AllTax
       (#m)      ($b)      ($b)      ($b)      ($b)      ($b)
 0    17.00    -119.4      -3.0       6.4       0.0       3.5
 1    17.00     152.4     -13.6      14.4       0.0       0.8
 2    17.01     273.9     -33.6      24.5       0.0      -9.1
 3    17.00     400.6     -37.8      35.8       0.0      -2.0
 4    17.01     558.3     -25.6      52.6       0.0      27.0
 5    17.01     754.0       7.5      72.0       0.0      79.5
 6    17.01    1008.8      55.3      97.0       0.0     152.3
 7    17.00    1379.3     116.0     143.2       0.0     259.1
 8    17.01    2054.4     215.0     233.3       0.0     448.3
 9    17.01    6272.4    1339.3     413.3       0.0    1752.7
 A   170.05   12734.6    1619.5    1092.6       0.0    2712.1

Weighted Tax Differences by Baseline Expanded-Income Decile
    Returns    ExpInc    IncTax    PayTax     LSTax    AllTax
       (#m)      ($b)      ($b)      ($b)      ($b)      ($b)
 0    17.00    -119.4      -0.4       0.0       0.0      -0.4
 1    17.00     152.4      -1.5       0.0       0.0      -1.5
 2    17.01     273.9      -8.7       0.0       0.0      -8.7
 3    17.00     400.6     -20.5       0.0       0.0     -20.5
 4    17.01     558.3     -24.7       0.0       0.0     -24.7
 5    17.01     754.0     -18.9       0.0       0.0     -18.9
 6    17.01    1008.8      -5.6       0.0       0.0      -5.6
 7    17.00    1379.3      -0.1       0.0       0.0      -0.1
 8    17.01    2054.4      -0.0       0.0       0.0      -0.0
 9    17.01    6272.4      -0.0       0.0       0.0      -0.0
 A   170.05   12734.6     -80.5       0.0       0.0     -80.5

[codecov-io]

Codecov Report

Merging #1553 into master will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master   #1553   +/-   ##
======================================
  Coverage     100%    100%           
======================================
  Files          37      37           
  Lines        2593    2593           
======================================
  Hits         2593    2593

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f9a8334...112eb79. Read the comment docs.

[martinholmer]

Under current-law policy the total EITC cost for FY2018 is estimated by TPC to be $67.9 billion according to their report on other kinds of EITC reforms issued in June 2017. Footnote 2 in that June 2017 analysis says "Estimates assume a fiscal split of 10-90 (fiscal year revenue is estimated to be 90% of revenue from the previous calendar year and 10% of revenue for the current calendar year)."

The current-law estimate for total EITC for CY2017 is $64.81 and for CY2018 is $66.32 billion, which produce, using the same rule as used by TPC, a FY2017 estimate of total EITC cost of about $65.0 billion.

The $65.0 is about 4 percent less than the TPC estimate of FY2018 EITC cost under current-law policy.

So, it seems unlikely that differences in simulation input data or "aging" assumptions are the cause of the much larger difference in the GAIN Act reform estimates. As discussed above the 10-year total reform effect is estimated by TPC to be about $1400 billion while Tax-Calculator's static analysis produces an estimate of about $917 billion. This difference is substantial: the Tax-Calculator static estimate is roughly 35% less than the TPC estimate.

This comparison of baseline estimates suggests that most of the difference in the GAIN Act estimates are caused by the TPC use of unspecified "microdynamic responses" to the GAIN Act's EITC expansion.

[MattHJensen]

@martinholmer, thanks very much for tackling issue #1552.

I believe the discrepancy between the results described above and the TPC score is primarily caused by the reform file in this PR not capturing all of the provisions of the GAIN act.

Based on my reading of the legislative text and comparison against existing law, the reform should be:

{
  "policy": {
        "_EITC_c":
            {"2017": [[3000, 6528, 10783, 12131]]},
        "_EITC_rt":
           {"2017": [[.3, .6258, .768, .864]]},
        "_EITC_prt": 
            {2017: [[.1598, .1598, .2106, .2106]]},
        "_EITC_ps":
            {2017: [[18340, 18340, 18340, 18340]]},
        "_EITC_MinEligAge":
            {"2017": [21]}
    }
}

This also conforms to footnote f of the relevant TPC revenue table so far as I can tell given the limited detail provided in that footnote.

[martinholmer]

@MattHJensen thanks for your help.