Help merging NBER to IPUMS March CPS data for 1964 and 1967

This post suggests a way to merge unprocessed NBER March CPS data with IPUMS CPS data. I’ve been able to complete this merge successfully for many years (1966, 1968, 1969, 1971-1975), but there are problems in 1964 and 1967.

The NBER and IPUMS data appear to contain the same observations, but in a different order. E.g., the 13th observation in the 1967 IPUMS data is a 23 year old male who is married to a 24 year old female (obs 14). These people appear to be the 12th and 13th observations in the NBER data. Results for 1964 are similar.

Do you have any information which might save this merge? I know that I can process everything using the NBER data, but would like to fix the merge if this can be easily done.

After looking into the issue, I can not find an easy solution for merging the NBER and IPUMS-CPS March samples for 1964 and 1967. It is clear that both the NBER and IPUMS-CPS files contain the same individuals, however, the sort orders are not shared between the two files. The sort order of these IPUMS-CPS files was changed as part of a reformatting process that only effected a small number of samples and dealt with issues of non-contiguous households (i.e. persons from household A appearing in household F). As a result, the standard sequencial merge is not possible on these files.

I am sorry I could not provide a simple solution to your issue.