Problems linking ASEC CPS data in 1985-1986, 1994-1995 and 1995-1996

[uctpdtz]

I’ve been trying to link ASEC files across years from 1980s onwards using hrhhid for years pre 1989 and cpsid onwards, combined with pernum, and verification using sex, age and race. This seems to work ok with the exception of 1985-1986, 1994-1995 and 1995-1996. I notice these years are mentioned here as being problematic:

https://www.ncbi.nlm.nih.gov/pmc/arti…

Is there any way round this, or do I just need to accept matching is not possible in these years?

[JeffBloem]

Linking ASEC samples is quite a bit more challenging than linking basic monthly samples in the CPS. Due to these challenges, IPUMS CPS has developed a method for linking ASEC samples over time. See this working paper for more details.

[uctpdtz]

Thanks Jeff. This is related to my question here:

Calculating wage growth for individuals over two consecutive months

I’d like to exploit the panel dimension of the monthly files to calculate monthly wage growth for individuals. My naive quick attempt in stata was: gen modate = ym(year, month) format modate %tm xtset cpsidp modate gen wgegrwth=hourwage/L.hourwage-1 However, I don’t get any values generated for this (all data missing). No doubt this is too naive but any thoughts on best practice welcome.

I need to link data before 1989 so tried to construct my own version of MARBASECID. This seems to work well for most years except the ones mentioned above, which seem to have been problematic for other authors. I will have a good read through the paper you suggested and see if I can find the answer there, though of course if anything occurs to you please let me know.