I’d like to calculate the number of workers making above a certain wage. When I use the “EARNWT” weight with CLASSWKR, the count for number of wage and salary workers basically lines up with the published BLS numbers. Seems like I’m all good to do the analysis.
But when I restrict the sample to ELIGORG==1, the number of wage and salary workers is about 10M short. The reason is that there is a group of observations with positive EARNWT but have an ELIGORG==0 and don’t have earnings information.
As I’ve dug in, this group has observations where EMPSTAT doesn’t equal 10 or 12 and that’s why they’re not ORG eligible. Screenshot below.
Some are in the armed forces and, from my understanding, those are in the BLS wage and salary worker but not in the ORG sample. All good.
Even more are NILF–“unable to work,” “other,” and “retired.” And even more are “unemployed, experienced worker.” This pattern appears regardless of month so doesn’t look to me like some sort of recent strike-based misclassification.
It doesn’t make sense to me that BLS would be classifying these folks as wage and salary workers, but not count them toward HH employment numbers. Is there something I’m missing? Do we know what’s going on with these people?
I’m trying to figure out what my analysis is doing by excluding them (or perhaps using the published wage and salary worker numbers minus armed forces and then apply the percentage to that number from my analysis).
I believe the answer is that EARNWT is just the ORG weight applicable to any record in the 4th or 8th month in sample, regardless of their labor force status (MISH==4 | MISH==8). ELIGORG meanwhile indicates whether a record is eligible for earnings data, i.e. they are a wage & salary worker. See below:
Thank you for the response!
I think I may have framed the opening of my question a little confusingly though. I’m not confused about there being non-employed people where EARNWT>0 for the reason you mention.
I am confused about why there are people who are wage and salary workers (CLASSWKR between 22 and 28) who are not employed (EMPSTAT isn’t between 10 and 12).
I can replicate the BLS published numbers for number of wage and salary workers using EARNWT, but some of those observations are not employed and don’t have earnings info because they’re not employed.
More generally, I’d like to understand the intuition of how you can be a wage and salary worker and not be employed.
Ah got it, I understand now, That’s because CLASSWKR reports either the class of the current job for employed workers, or the class of the most recent job for nonemployed workers:
Ah, good catch. Always read the codebook. Thank you!
The universe of CLASSWKR is persons 15+ who ever worked. As Primitive_Wallflower pointed out, respondents who were not employed during the week prior to the survey reported the class of their most recent job. Thank you Primitive_Wallfower for helping answer this question.