NILF population and WNLOOK/NILFACT

In the basic monthlies, why do NILFACT and WNLOOK not cover the entire nilf population? In 2018, about 12.5 percent of nilf observations do not have an answer for either WNLOOK or NILFACT.

The NILFACT universe are those who are nilf and did not provide a reason. However, there are observations who answered “nilf, other” to EMPSTAT who do not have an answer to NILFACT.

Is there any way to determine why someone is nilf for the entire nilf population? Or is the way to do this a combination of EMPSTAT, NILFACT, and WNLOOK?

Thanks for this note. I am able to replicate the observation you describe here. In particular, the universe statement for NILFACT states that it provides information on the activity of respondents who are not in the labor force and who did not report EMPSTAT==32 “unable to work” or 36 “retired.” This implies that all respondents who have EMPSTAT==34 “other” should have valid values in NILFACT. However, there are a couple hundred respondents in each basic monthly sample who have EMPSTAT==34 and NILFACT==99 “NIU” or “Blank.” This results in a situation where we can’t fully define the activities of those not in the labor force.

After looking into this issue a bit over the last few days and discussing it with several members of the IPUMS CPS Team, we’ve ruled out any sort of error on the part of IPUMS, but do not have a clear explanation for this observation. The only pattern we have been able to find is records with EMPSTAT == 34 and NILFACT == 99 all have non-‘yes’ responses for UH_ANYWK_2 (did work for pay or profit last week). Our best guess is that these people have inconsistent responses regarding their NILF status. For example, some people who respond ‘no’ to UH_ANYWK_2 later respond ‘retired’ to questions about being on layoff, even though ‘retired’ was an available response in UH_ANYWK_2. Based on all of this it seems reasonable to consider these people ‘something else/other’ in NILFACT.

Thanks so much for digging into this, Jeff.