Using EMPSAME to impute wage information

Would IPUMS advise against using EMPSAME to impute HOURWAGE and EARNWEEK values for months-in-sample 1-3, and 5-7? I’m using linked BMS data to analyze monthly transitions from unemployment to employment. One of the reemployment indicators I’m interested in is wages (even though the BMS doesn’t provide pre-job loss earnings data; I’m also looking at the DWS). But, given the structure of the Earner survey, if I wanted to explore reemployment earnings (again, knowing I’m missing pre-job loss earnings), I’m limited to unemployed who transition to employment between months-in-sample 3 and 4, and 7 and 8. Is this correct? Given this limitation, I thought I would investigate how I could leverage EMPSAME to cautiously impute HOURWAGE and EARNWEEK to earlier months, so I could have earnings information for unemployed who transition to employment in months-in-sample 2 or 3, or 6 or 7.

One significant complication I can think of is that respondents could earn raises/experience wage changes in the time between their first reemployment month and months-in-sample 4 or 8. Also, I understand that EMPSAME has its own issues (e.g., unexplained “niu” responses). So, maybe this isn’t possible; but I thought I would double-check.

IPUMS doesn’t have any particular guidance on this. I don’t see any a priori reason to not do this. Just be cautious that there will be some error due to the things that you mentioned, but it seems likely to be small.

Sounds good. Thank you for the quick response!