Merging property tax imputation results with main ASEC sample - baseline vs full sample

[jo-fleck]

When merging the property tax imputation results with the ASEC sample which contains all the other tax and transfer info, I find a small number of non-matched observations for the baseline sample, but a larger number for the full sample:

baseline, Years: 0506 - households dropped from sample because missing in the matched property tax data: 1
full, Years: 0506 - households dropped from sample because missing in the matched property tax data: 5596
baseline, Years: 1011 - households dropped from sample because missing in the matched property tax data: 2
full, Years: 1011 - households dropped from sample because missing in the matched property tax data: 7276
baseline, Years: 1516 - households dropped from sample because missing in the matched property tax data: 0
full, Years: 1516 - households dropped from sample because missing in the matched property tax data: 5594

Why is this? It it because ACS observations with 0 or negative incomes get dropped?

[jiaxitan]

This is because we drop households with negative gross income in ASEC sample. Refer to line 57 of _ACS_ASEC_datapreparation.jl: filter!(r -> (r[:grossinc] .> 0), df_ASEC_hh); # Innocent This line removes all households in ASEC with negative gross income, and when you merge it with your ASEC sample these households don't exist in my result.

[jo-fleck]

Thanks @jiaxitan ! I'll think about how to address this issue.