Adjustment to January 2020 CBO economic projections due to COVID-19

[chusloj]

This PR updates taxdata to reflect updates to CBO Economic Projects due to COVID-19. The general steps I took were:

1. Follow instructions in CBO_Baseline_Updating_Instructions.md
2. Run make all

Updated data is pulled from the CBO's May 2020 projections for 2020-2021

UPDATE: New data is now pulled from the CBO's July 2020 Economic Projections for 2020-2030. All variables listed under Variables in 10-Year CBO Projections within CBO_Baseline_Updating_Instructions.md were updated and unemployment compensation (UCOMP) was updated with March 2020 data (see this comment).

[MattHJensen]

Thank you for opening this PR @chusloj! Looking forward to following your progress. Please feel free to remove the in progress label when this is ready for review and leave a comment expressing the same.

[chusloj]

UPDATE: make all has been run

The 2019 values for all variables listed under Variables in 10-Year CBO Projections within CBO_Baseline_Updating_Instructions.md were updated to reflect actual (not projected) values.

After running the makefile, test_growfactors.py shows that the 2022 growth rate within growfactors.csv for ADIVS (Dividend Income, Personal) exceeds the growth rate limit of 1.6 set in test_growfactors.py by ~6%. Would like @MattHJensen to comment on this.

Minor updates have been made to the StatMatch README and general README file. Documentation for the cps-matched-puf.csv is currently being constructed.

[chusloj]

The maximum grow factor within test_growfactors.py has been increased from 1.6 to 1.7.

[andersonfrailey]

Thanks for working on this, @chusloj! With the merging of #314, you'll need to recreate the CPS weights. I haven't updated documentation for this yet, but if you just run cps_stage2/stage2.py after pulling all the updates to your local branch you should be good to go.

[chusloj]

UPDATE: The CBO will release new 10-year economic projections on 7/2/2020. This PR will be re-updated after this release.

[chusloj]

Weights have been recalculated using the newest CBO data. The variables listed under Variables in 10-Year CBO Projections have been updated with July 2020 data and unemployment compensation (UCOMP) has been updated with March 2020 data.

updatecbo.py has been updated because of problems recognizing data from the budgetary projections .xlsx file related to multi-indexing in pandas.

@MattHJensen @andersonfrailey @Peter-Metz

[Peter-Metz]

@chusloj thanks a lot for the PR and sorry for the slow review. Things look good to me, but I have a couple of minor questions:

• What was the thinking for removing the README?
• Do the dates in the docs for CGNS data need to be updated? It looks like the CGNS numbers changed but I'm not sure.

For peace of mind, it might be nice to produce tables that use the new CPS/PUF weights to illustrate the aggregate changes in Tax-Calc statistics similar to the ones in #331 (comment)

Also, it might be helpful to add a note to the initial comment in this PR explaining that this PR now updates economic projections per the July CBO report and provide a link to that report.

[chusloj]

What was the thinking for removing the README?

Not sure how that disappeared. README.md from the master branch has been restored to the repo.

Do the dates in the docs for CGNS data need to be updated?

No, the new values are the exact decimal numbers from the most recent CBO budgetary projections listed in the docs. The previous numbers were just the rounded numbers.

[chusloj]

@Peter-Metz Here's the diff tables you suggested above:

Pre-PUF change:

Post-PUF change:

These tables are for changes in the PUF weights only because the number of records in the new CPS weights file is different from the current CPS weights file in Tax Calculator which was compiled before #314 was merged. I'll hold off on making a CPS comparison until that new CPS weight file pre-CBO update is added to Tax Calculator. @andersonfrailey

[Peter-Metz]

Thanks for the tables @chusloj. The magnitude and direction of the changes make sense to me. Think that resolves all of my questions

cc @MattHJensen

[andersonfrailey]

This looks good to me, @chusloj. @Peter-Metz good with me merging this and then updating your tax-calc pr?

[Peter-Metz]

@andersonfrailey yep, good to merge! We might want to keep Tax-Calculator#2444 and the PR for CBO updates separate to be able to separate out how results change for each data modification

[andersonfrailey]

@Peter-Metz I think that's a good idea. Anything you want to add/update with this, @chusloj ? I'll merge if you're all good.

[chusloj]

@andersonfrailey All good, ready to merge!