CBO 2026 baseline update

[ashilovtax]

This PR updates the taxdata CBO baseline to February 2026. The corresponding code indicating the end year in cps and puf stages have also been updated. The CBO_baseline.csv file has been manually updated following the instructions for the manual update. The only step that wasn't followed, was for CPI-M, as the projections for Medical Care are available in the CBO Economic Projections.

For reference, the Excel Files that were used in the update:

25projpub6187tables.xlsx
51135-2026-02-Economic-Projections.xlsx
51138-2026-02-Revenue.xlsx
51316-2026-02-unemployment.xlsx
SocSec_manual_calculation_35_36.xlsx
TRTables_TR2025.xlsx

[martinholmer]

@ashilovtax, Thanks for your work on taxdata PR #456.
Who have you asked to review this PR?
Did you do this on your own or did someone as you to do this?

[martinholmer]

@ashilovtax, I have not done a review, but I did notice what looks like a typo in the documentation in the docs/book/content/methods/CBO_Baseline_Updating_Instructions.md
file:

Shouldn't that be "February 2026" instead of "February 2025"?

[ashilovtax]

@ashilovtax, Thanks for your work on taxdata PR #456. Who have you asked to review this PR? Did you do this on your own or did someone as you to do this?

I have been in contact with Jason DeBacker. I did this on my own as part of my work at Tax Foundation - we think it would be useful to have an updated taxdata branch

[ashilovtax]

@ashilovtax, I have not done a review, but I did notice what looks like a typo in the documentation in the docs/book/content/methods/CBO_Baseline_Updating_Instructions.md file:

Shouldn't that be "February 2026" instead of "February 2025"?

Yes, you are right. Should be 2026

[martinholmer]

@ashilovtax said in PR #456:

I have been in contact with Jason DeBacker.
I did this on my own as part of my work at Tax Foundation - we think it would be useful to have an updated taxdata branch.

OK, thanks again for developing this taxdata pull request.
I'll wait for @jdebacker to conduct a review of this pull request.

[jdebacker]

@ashilovtax Thank you for the PR! I've reviewed it and the changes look good. But there are a few things to fix before we can merge this:

1. Merge my PR to your branch. This will help fix the code formatting test failures.
2. Two of the other test failures relate to the fact that the weights files are not updated in this PR and so don't include 2036. Have you run all the stage 2 scripts?
3. There is also a failure of the test tests/test_growfactors.py::test_growfactor_values, although I don't know exactly what is causing that.

Please let me know if you have questions as you address these issues.

[martinholmer]

@ashilovtax and @jdebacker,
In the current version of the growfactors.csv file, all the benefit factors beginning in 2032 seem to be zero.
I'm inferring this because when using the new growfactors.csv file in Tax-Calculator PR 3004, the test_benefits.py fails with these differences:

(taxcalc-dev) tests> diff benefits_actual.csv benefits_expect.csv 
146,169c146,169
< 2032,ssi,0.0,0.0,0.0
< 2032,mcare,0.0,0.0,0.0
< 2032,mcaid,0.0,0.0,0.0
< 2032,snap,0.0,0.0,0.0
< 2032,wic,0.0,0.0,0.0
< 2032,tanf,0.0,0.0,0.0
< 2032,vet,0.0,0.0,0.0
< 2032,housing,0.0,0.0,0.0
< 2033,ssi,0.0,0.0,0.0
< 2033,mcare,0.0,0.0,0.0
< 2033,mcaid,0.0,0.0,0.0
< 2033,snap,0.0,0.0,0.0
< 2033,wic,0.0,0.0,0.0
< 2033,tanf,0.0,0.0,0.0
< 2033,vet,0.0,0.0,0.0
< 2033,housing,0.0,0.0,0.0
< 2034,ssi,0.0,0.0,0.0
< 2034,mcare,0.0,0.0,0.0
< 2034,mcaid,0.0,0.0,0.0
< 2034,snap,0.0,0.0,0.0
< 2034,wic,0.0,0.0,0.0
< 2034,tanf,0.0,0.0,0.0
< 2034,vet,0.0,0.0,0.0
< 2034,housing,0.0,0.0,0.0
---
> 2032,ssi,79.862,17.902,4.5
> 2032,mcare,1553.523,95.617,16.2
> 2032,mcaid,803.732,90.937,8.8
> 2032,snap,160.377,82.593,1.9
> 2032,wic,5.354,23.222,0.2
> 2032,tanf,36.782,9.701,3.8
> 2032,vet,251.045,12.862,19.5
> 2032,housing,84.103,14.431,5.8
> 2033,ssi,80.757,18.094,4.5
> 2033,mcare,1573.033,96.777,16.3
> 2033,mcaid,809.874,91.615,8.8
> 2033,snap,161.978,83.282,1.9
> 2033,wic,5.374,23.33,0.2
> 2033,tanf,36.919,9.743,3.8
> 2033,vet,254.375,13.019,19.5
> 2033,housing,84.96,14.556,5.8
> 2034,ssi,81.748,18.304,4.5
> 2034,mcare,1592.585,97.928,16.3
> 2034,mcaid,816.316,92.304,8.8
> 2034,snap,163.576,83.994,1.9
> 2034,wic,5.409,23.478,0.2
> 2034,tanf,37.246,9.826,3.8
> 2034,vet,257.749,13.182,19.6
> 2034,housing,85.791,14.683,5.8

If benefit growfactors for years after 2031 are not available, can you use the values from the old growfactors.csv file instead of zeros?

[ashilovtax]

@ashilovtax and @jdebacker, In the current version of the growfactors.csv file, all the benefit factors beginning in 2032 seem to be zero. I'm inferring this because when using the new growfactors.csv file in Tax-Calculator PR 3004, the test_benefits.py fails with these differences:

(taxcalc-dev) tests> diff benefits_actual.csv benefits_expect.csv 
146,169c146,169
< 2032,ssi,0.0,0.0,0.0
< 2032,mcare,0.0,0.0,0.0
< 2032,mcaid,0.0,0.0,0.0
< 2032,snap,0.0,0.0,0.0
< 2032,wic,0.0,0.0,0.0
< 2032,tanf,0.0,0.0,0.0
< 2032,vet,0.0,0.0,0.0
< 2032,housing,0.0,0.0,0.0
< 2033,ssi,0.0,0.0,0.0
< 2033,mcare,0.0,0.0,0.0
< 2033,mcaid,0.0,0.0,0.0
< 2033,snap,0.0,0.0,0.0
< 2033,wic,0.0,0.0,0.0
< 2033,tanf,0.0,0.0,0.0
< 2033,vet,0.0,0.0,0.0
< 2033,housing,0.0,0.0,0.0
< 2034,ssi,0.0,0.0,0.0
< 2034,mcare,0.0,0.0,0.0
< 2034,mcaid,0.0,0.0,0.0
< 2034,snap,0.0,0.0,0.0
< 2034,wic,0.0,0.0,0.0
< 2034,tanf,0.0,0.0,0.0
< 2034,vet,0.0,0.0,0.0
< 2034,housing,0.0,0.0,0.0
---
> 2032,ssi,79.862,17.902,4.5
> 2032,mcare,1553.523,95.617,16.2
> 2032,mcaid,803.732,90.937,8.8
> 2032,snap,160.377,82.593,1.9
> 2032,wic,5.354,23.222,0.2
> 2032,tanf,36.782,9.701,3.8
> 2032,vet,251.045,12.862,19.5
> 2032,housing,84.103,14.431,5.8
> 2033,ssi,80.757,18.094,4.5
> 2033,mcare,1573.033,96.777,16.3
> 2033,mcaid,809.874,91.615,8.8
> 2033,snap,161.978,83.282,1.9
> 2033,wic,5.374,23.33,0.2
> 2033,tanf,36.919,9.743,3.8
> 2033,vet,254.375,13.019,19.5
> 2033,housing,84.96,14.556,5.8
> 2034,ssi,81.748,18.304,4.5
> 2034,mcare,1592.585,97.928,16.3
> 2034,mcaid,816.316,92.304,8.8
> 2034,snap,163.576,83.994,1.9
> 2034,wic,5.409,23.478,0.2
> 2034,tanf,37.246,9.826,3.8
> 2034,vet,257.749,13.182,19.6
> 2034,housing,85.791,14.683,5.8

If benefit growfactors for years after 2031 are not available, can you use the values from the old growfactors.csv file instead of zeros?

If I put 1.0 into the empty fields in growfactors for years after 2031, it fixes the tests on my side. Do you think this is a good solution? I believe this is what you meant by values from the old growfactors.csv

[martinholmer]

@ashilovtax said in taxdata PR #456:

If I put 1.0 into the empty fields in growfactors.csv for years after 2031, it fixes the tests on my side.
Do you think this is a good solution? I believe this is what you meant by values from the old growfactors.csv

Yes, this is an excellent fix. Thanks for your quick response!

[martinholmer]

@ashilovtax, It seems as if @jdebacker approved the changes in taxdata PR #456 and then you fixed the 2032+ values of the benefit factors. So, my question is what is left to do on #456? If nothing, what is the plan to merge #456 and develop a new taxcalc PR that incorporates the new 2026 CBO baseline information?

[jdebacker]

@ashilovtax Thanks for the patch to growth rates!

I've got one more PR to your branch that should fix the currently failing tests.