In the HOURAGE documentation, there is this doozy: The Census Bureau reports that the results in the CPS public use files for this data series included errors for years prior to 1990, so only data from 1990 forward are part of IPUMS-CPS.
Where does the Census Bureau report this?
More broadly: how big are the errors? Because I’ve come across many papers that use the May/ORG data going back well before 1990. And the NBER’s MORG dataset goes back to 1979. Presumably IPUMS discussed the option of including HOURWAGE going back further in time along with a warning, and decided elision was the better option - I’d just like to understand why.
Thanks for this note about the HOURWAGE documentation. Unfortunately, I do not have any sort of source information to provide. This information comes from discussions between individual Census Bureau employees and Unicon (who previously provided the CPS data publicly). The IPUMS CPS Team has been working on updating our documentation to clarify this problematic data from the early ASEC samples. This updated information will be included with the next data release that will likely be in a couple of weeks. Lastly, you are correct that this data is available in the NBER merged outgoing rotation group files back to 1979, but these data are not available from IPUMS CPS. Although the IPUMS CPS Team is working on making HOURWAGE (and other ORG questions) available for basic monthly samples from 1982-1988, they do not have plans to release these data from pre-1982 basic monthly samples or pre-1989 ASEC samples anytime soon. All that being said, we are not aware of any problems with the HOURWAGE data in the non-ASEC samples prior to 1982 if you would like to merge the NBER data onto your IPUMS extract.