Why would n24 exceed nu18?

[MaxGhenis]

My understanding of the Child Tax Credit's definition of child is that it should be a subset of children under age 18. One criterion is being under age 17, and others limit further within that.

2014 CPS data shows that n24 and nu18 have almost identical totals: 78.7M and 78.8M, respectively. Yet n24 exceeds nu18 for 9M tax units, 19% of all tax units where n24>0 (notebook). These 9M tax units have 15M total n24 and 1.3M total nu18

Is there a reason I'm missing here, like different residency requirements? If not, should n24 be capped at nu18?

NB: this will become a slightly starker issue after #131 is resolved.

[andersonfrailey]

@MaxGhenis, I don't have a specific answer for you right now, but I do know we can do a better job calculating n24 in general. I'm wrapping up a few other things right now, but when I finish I'll work on a fix for how we count n24.

[MaxGhenis]

Now that #131 is resolved, n24 is no longer capped at 3, and we see CPS data with it as high as 5. This changes some of the metrics I listed in my first comment, using 2014 CPS data (updated notebook):

• n24 totals 81.6M while nu18 totals 78.8M.
• n24 exceeds nu18 for 9.0M tax units, 19.5% of all tax units where n24>0 (tiny change).
• But, these 9.0M tax units now represent a higher 15.7M total n24 and 1.44M total nu18.

#164 (including 17-year-olds) explains some of this.

[MaxGhenis]

Updating the notebook for taxcalc 0.19.0, where n24 no longer includes 17-year-olds, shows:

• n24 totals 76.3M while nu18 totals 78.6M.
• n24 exceeds nu18 for 8.2M tax units, 18.7% of all tax units where n24>0 (tiny change).
• These 8.2M tax units represent 14.0M total n24 and 1.2M total nu18.

[martinholmer]

@MaxGhenis said in taxdata issue #157:

Updating the notebook for taxcalc 0.19.0, where n24 no longer includes 17-year-olds, shows:

• n24 totals 76.3M while nu18 totals 78.6M.
• n24 exceeds nu18 for 8.2M tax units, 18.7% of all tax units where n24>0 (tiny change).
• These 8.2M tax units represent 14.0M total n24 and 1.2M total nu18.

@andersonfrailey, Why would the CPS values for n24 and nu18 be inconsistent?
Can you see a way to fix this? What would be the time schedule for the fix?

[andersonfrailey]

I can think of a couple of reasons why n24 may be less than nu18.

First, in our scripts, any dependent filers who are under 18 would be counted for nu18, but not n24.

Second, n24 is defined in final prep, not during the tax unit creation process. The SAS scripts export the age of five dependents. That data is then used to form n24. Thus, n24 will be capped at 5.

I can change the SAS scripts so that n24 is counted when we add dependents to a tax unit. This will take me a few days to finish.

[MaxGhenis]

@andersonfrailey to clarify, this is about n24 being higher than nu18. The potential reasons you listed would explain the opposite I think.

Sorry to see this request move logic from Python to SAS. Is a long-term goal to move everything out of SAS? FWIW I'm not too personally concerned with this issue given the totals look OK, in case waiting to move logic to Python is an option.

[andersonfrailey]

Sorry to see this request move logic from Python to SAS. Is a long-term goal to move everything out of SAS?

Yes, I'm working on moving everything from SAS to Python ASAP. That is the next big milestone for this repo.

[martinholmer]

Issue #157 has been resolved by pull request #183.