Industry/occupation data attached to those NILF inconsistent

I’m working with the BMS and interested in knowing the industry and occupation of the last job held by individuals NILF. I’d prefer to focus on the marginally attached. But I’m struggling with two things:

  1. According to descriptions of the major industry and occupation variables I’m using (UH_INDMAJ_B2; UH_INDOCC_B2), there should be industry and occupation data for those respondents NILF with any work in the prior 12 months. But it seems like the responses are inconsistent – for instance, a respondent who’s employed in one month, then NILF the next month(s), will have a non-missing industry value in the survey month they’re employed, then have a missing industry value in the next month(s) when they’re NILF. Or a respondent whose labor force status of NILF doesn’t change over the course of four interviews will have a non-missing industry value for just one interview month. Have others raised this as an issue in the past? Is there anything that can be done to resolve (other than cautiously imputing industry/occupation data provided in other interview months)?

  2. How do I isolate the marginally attached in the BMS (IPUMS extract)? I assume they’re a subset of those captured in the “nilf, other” category of EMPSTAT? I feel like others have asked this question here, but haven’t received clear responses.

Thanks so much.

Your first question regarding the changing universe for industry and occupation codes is unfortunately an issue with the original data. The comparability tab for OCC notes that “only one-fourth of CPS respondents who were not currently in the labor force were asked about their past occupation.” This appears to be the same for industry as well as occupation codes and explains cases where a respondent has a missing industry value during a period they are not in the labor force, even though they were employed in the previous month. Additionally, the comparability tab also references cases where “respondents are shown as having valid responses despite not meeting universe requirements.” Unfortunately, there are some inconsistencies with the universe for these variables that we do not yet understand. While you might attempt to impute industry and occupation codes, we leave the choice to researchers on how to best proceed.

The attached spreadsheet explains how to isolate marginally attached workers. Specifically, the variable UH_DSCWK_B2 identifies both discouraged and conditionally interested workers, both of which are included in the set of marginally attached workers.

U-1 through U-6 definitions for IPUMS CPS.xlsx (10.1 KB)

Okay; that’s too bad. Thanks so much.

And thank you for the spreadsheet. So helpful.