Essentially I’m after a data quality flag for INCWAGE, though that does not seem to explicitly exist. Any alternatives gratefully received. I am updating results from this paper, which mentions it excludes non-respondents (e.g. who would have had their wages imputed):
Is there a way to identify individuals who have had their wage income (INCWAGE) imputed (1963-2015)?
For the 1989-present samples, INCWAGE is constructed from OINCWAGE and sometimes includes INCLONGJ (if wages are indicated by SRCEARN as the source of earnings). Both of these variables have corresponding data quality flags that can identify imputed information. Unfortunately, I am unaware of a similar data quality flag for INCWAGE prior to 1988.
I hope this helps.
Thanks. I guess there is a variable in the ASEC cps that allows you to filter out non-respondents (rather than use imputed data for them) as this was done in the paper I’ve linked. Perhaps it exists but isn’t included in IPUMS? In any case I think I can work out rough impact of imputation by using the post 1989 data you suggest.