Issue on constructing monthly continued claim data

[yzha0]

@kgcsport Right now I have a problem when I construct the outcome variable for state MA. The continued claim data I have is weekly data and is it not right that I just sum the four weekly claim reported data to get the monthly continued claimed data? I tried but the monthly data became very large which is even larger than my total unemployed people(gave me take-up rate>1) monthly so I think there are some problems with how I transform the weekly continued claim data into monthly claim data. My NY take-up rate data looks reasonable since they have the number of beneficiaries(number of continued claim) each month. I will push my code to the repo and you may take a look.

[kgcsport]

Ah that’s a shame. I’d check any documentation you can find. Here are some possible reasons for the discrepancy:

1. Four weeks don’t perfectly align with a month. Generally that probably means four weeks is less than a month, but if there’s a spike in the claims in the week overlap with the other month, you have an issue. (You could plot weekly data to check how often this occurs.)

2. Claims occur with a lag. People don’t immediately claim, but wait to claim. This leads to uneven distribution of claims with respect to unemployment, which is less of a “choice” variable.

3. There are bogus claims driving up numbers.

In general, I suggest you check if/how other literature has measured take up and mimic it. If it continues to throw weird errors like this, show what the potential issue is with plots and make an informed choice on how to deal with the issue.