New version of cps.csv.gz file and associated changes in logic and tests

[martinholmer]

This pull request uses a new version of the cps.csv.gz data file and makes the minimal number of other changes necessary for all the tests to pass.

This new cps.csv.gz data file includes better quality values for two existing variables (e02300 unemployment insurance benefits and e02400 social security benefits) that have been derived from the work of the C-TAM project.

And it also includes six new benefits variables: ssi_ben, vet_ben, snap_ben, mcare_ben, mcaid_ben, and other_ben that have been supplied by the work of the C-TAM project.

And finally, the existing variable n1821 has been renamed n1820 to clarify that it represents the number of individuals in a filing unit who are either 18, 19, or 20 years old. This variable's only use in Tax-Calculator is in the computation of the filing unit's UBI benefit.

As a result of the larger OASDI and UI benefits in the new cps.csv.gz file, aggregate income tax revenue is somewhat higher than it used to be. For example, in 2014, aggregate income tax revenue is now $1063.3 billion up from $1054.0 billion, which is a rise of about nine-tenths of one percent. Of course, there is no change in payroll taxes with this new cps.csv.gz data file.

The new cps.csv.gz file incorporated in this Tax-Calculator pull request has been generated by @andersonfrailey in taxdata pull request 146.

[codecov-io]

Codecov Report

Merging #1853 into master will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master   #1853   +/-   ##
======================================
  Coverage     100%    100%           
======================================
  Files          37      37           
  Lines        3168    3168           
======================================
  Hits         3168    3168

---

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 07e28d1...a87cce9. Read the comment docs.

[martinholmer]

Pull request #1853 changes the name of the Records variable n1821 to n1820 in order to clarify its meaning.

Those using the puf.csv file in their work can accomplish this rename by simply doing the following:

$  mv puf.csv puf.old  ;  sed 's/n1821/n1820/' puf.old > puf.csv

[andersonfrailey]

This looks good @martinholmer. I just have one question. In TaxData PR #146 we found that accuracy of e02300 was being effected by the blowup factors being applied when a new calculator is created. One possible solution was skipping the 2014 blow-up logic when using the CPS file. Did you want to try that or wait until we can compare that to creating blow-up factors specific to the CPS?

[martinholmer]

@andersonfrailey said:

I just have one question. In taxdata PR 146 we found that accuracy of e02300 was being effected by the blowup factors being applied when a new [Records object] is created. One possible solution was skipping the 2014 blow-up logic when using the CPS file. Did you want to try that or wait until we can compare that to creating blow-up factors specific to the CPS?

This skipping of the blowup step in 2014 for the CPS data is implemented in pull request #1854.

Thanks for your astute diagnosis of the bug.

[andersonfrailey]

This skipping of the blowup step in 2014 for the CPS data is implemented in pull request #1854.

Great! This looks good to me now.