HRHHID in 1976-77 in IPUMS and NBER


I’m seeing some small differences in the HRHHID variable in 1976-77 between IPUMS and the NBER files, do you know why this might be happening?

For example, in January 1976, the two individuals in the sample with HUHHNUM = 8 have HRHHID = 167004034020 in IPUMS and HRHHID = 167004034051 in the NBER file (difference in bold). This type of problem seems to be affecting almost all observations in 1976-77. Usually the problem is with the final two digits, but not always.

This problem came up when trying to match the NBER files to IPUMS. I’m matching on HRHHID, HUHHNUM, and LINENO, which matches almost all observations for 1978 onwards.


This is a known issue. The IPUMS CPS team did some work to create a household identifier that uniquely identifies households as the original identifier in the non-supplement file doesn’t do this. It is not currently easy to merge NBER and IPUMS CPS data from 1976 and 1977; however, our IPUMS CPS team is working to document the procedure required to create these unique identifiers (and link them to NBER) more extensively. If the variable you are hoping to get from NBER is an basic monthly variable, you may be able to find it among the unharmonized variables on the IPUMS CPS website.

Thanks for letting me know! Any documentation on this would be very helpful