Fix top-coding logic in CPS data preparation

[martinholmer]

In a comment in Tax-Calculator issue 1658 made on 11/16/2017 @andersonfrailey said:

@martinholmer asked:

where in the sequence of scripts you list in this README.md file are the top-coded replicates added?

This occurs in the TopCodingV1.sas script. This is after tax units have been created and adjustments are being made to the final file.

And

what are your views on the thoughts first expressed by @Amy-Xu that the whole top-coded replicates business is unnecessary for our three (post-2011) CPS files?

I talked with John about this a few months ago. His opinion was that there wasn't a huge need to revise/remove the top coding scripts, though he hadn't run the numbers to verify this.

I'm in favor of at least comparing the results that we would get from removing this part of the creation process. There are two problems preventing this from happening immediately though. First, our SAS license has expired and I just checked and the general AEI one has as well. Second, John hasn't sent me the script to recreate the weights file yet and I haven't completed my work on a python version yet so we can't create a new weights file at this time.

[martinholmer]

@andersonfrailey, I'm going to close Tax-Calculator issue 1658 in favor of this taxdata issue #174.

[martinholmer]

@andersonfrailey, Where do we stand on taxdata issue #174?
The problem here seems to be a bug. Do you agree?
Now that you're finished with the UBI/benefits work, I'm hoping this rises to the top of your to-do list.

@Amy-Xu @MattHJensen

[martinholmer]

Around 3/1/2018 @andersonfrailey said in issue #162:

No, I haven't been able to put a significant time into considering how to change our top coding procedures yet. Given that the deadline on the UBI project, I think the next version of the CPS should be available after #161 is merged so that Tax-Calculator and then TaxBrain can be updated accordingly.

After that, in no particular order, here are the major changes I have in mind for the CPS file:

• Create the CPS weights file using our own scripts
• Revisit our top coding procedure to address this issue Fix top-coding logic in CPS data preparation #174
• Finish converting the CPS file creation scripts to Python
• Revisit how n24 is calculated to address issue Why would n24 exceed nu18? #157
• Address missing variables in the CPS pointed out in Missing variables in CPS file #119
• Add testing scripts to the repo

[andersonfrailey]

Closing now that we've switched to python scripts to create the CPS.