Update to CBO January 2020 economic projections and extend data to 2030

[Peter-Metz]

This PR updates TaxData to use the latest CBO economic projections. The general steps I took were:

1. Follow instructions in CBO_Baseline_Updating_Instructions.md
2. Run puf_data/StatMatch/runmatch.py after adding asec2016_pubuse_v3.dat to the repo
3. Run Make all

A few notes:

1. The format of the CBO capital gains data changed, so updatecbo.py does not successfully retrieve the CG data. I got around this by manually updating the CG data in CBO_baseline.csv, but we might want to address this issue in a future PR.
2. In running Make all, the new weights files did not successfully overwrite the old files. To get this process to work, I had to delete *_weights.csv.gz before running Make all.
3. I made a slight adjustment to puf_stage2/stage2.py line 68 so that stage3.py is able to read the file.
4. @andersonfrailey confirmed that the 2030 growth rates in benefit_growth_rates.csv should be the same as 2027-2029. If this continues to be the case for future years, we should consider adding a note to CBO_Baseline_Updating_Instructions.md

cc @MattHJensen @andersonfrailey

[Peter-Metz]

Following up on our conversation during yesterday's Tax-Calc call, here's how the CBO update will affect aggregate Tax-Calc statistics using both the PUF and CPS:

I believe the direction of these results makes sense because some factors in growfactors.csv like inflation and wage growth were adjusted downwards, but would appreciate a second pair of eyes on this.

@MattHJensen @andersonfrailey

[andersonfrailey]

Thanks for the PR, @Peter-Metz. At a (very) quick glance, this looks good. I can give it a more thorough review next week if there's no rush to merge it and get a new version of Tax-Calc out

[donboyd5]

The changes look truly de minimis to me, too. Moderately interesting that the longer-run (2029) CPS differences are larger than the longer-run PUF differences but there are a lot of years in between and a lot of potentially plausible explanations and in any event the differences still are small. Don

…

On Wed, Mar 4, 2020 at 10:17 PM andersonfrailey ***@***.***> wrote: Thanks for the PR, @Peter-Metz <https://github.com/Peter-Metz>. At a (very) quick glance, this looks good. I can give it a more thorough review next week if there's no rush to merge it and get a new version of Tax-Calc out — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#331?email_source=notifications&email_token=ABR4JGDUQ7PDXD25YSJJMHLRF4KVTA5CNFSM4K46367KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEN3RNZI#issuecomment-595007205>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABR4JGDHRH4OFQBIUC2CAALRF4KVTANCNFSM4K46367A> .

[Peter-Metz]

Thanks a lot for the review @andersonfrailey and @donboyd5.

Anderson - there's no rush to merge so a closer look whenever you get the chance would be great. Thanks!

[andersonfrailey]

All seems good to me!

[MattHJensen]

@Peter-Metz, thanks a great deal for the PR and @donboyd5 and @andersonfrailey for the reviews. The results look to be as expected to me as well. Merging.