search

PSLmodels / taxdata

The TaxData project prepares microdata for use with the Tax-Calculator microsimulation project.
http://pslmodels.github.io/taxdata/
Other
20 stars 30 forks source link

Inconsistent child age information #413

Open Thirdhuman opened 2 years ago

Thirdhuman commented 2 years ago

image

I'm getting very inconsistent counts for the number of children within a household hoping someone can let me know whether this is a general bug with taxdata. Or if my own implementation of taxdata for the 2014 PUF is somehow at fault.

martinholmer commented 2 years ago

@Thirdhuman, Is the inconsistent-child-age issue you report in #413 related to the problems reported in issue #408 and/or issue #409?

Thirdhuman commented 2 years ago

@martinholmer - that explanation makes sense. Are you also using the 2014 PUF? And does reversing the following change you identified fix this?

The https://github.com/PSLmodels/taxdata/pull/314 changes included (without offering any rationale) a change from == to >= for CPS data in this code:

I'll also note that population growth for children runs hotter than other estimates (perhaps due to generalized growth parameters not properly accounting for age distribution skew attributable to native fertility vs. immigration?) Given that I'm sure many Tax Calculator practitioners are involved in the CTC-related analyses, ensuring that the child-related stuff accurate should be a priority.

martinholmer commented 2 years ago

@Thirdhuman said in issue #413:

@martinholmer - that explanation makes sense. Are you also using the 2014 PUF?

No, I'm not using any PUF related data. Issues #408 and #409 relate to the CPS input data file used by Tax-Calculator. But my impression is that all ages in each PUF unit in the PUF-derived file used as input by Tax-Calculator are transferred from a CPS unit that is statistically matched to the PUF unit. So, any problems with CPS ages would be transferred to the PUF input data, if my understanding is correct.

And @Thirdhuman also said:

Given that I'm sure many Tax Calculator practitioners are involved in the CTC-related analyses, ensuring that the child-related stuff accurate should be a priority.

I would tend to agree. You should take this matter up with @andersonfrailey and @MattHJensen