Is the unionization rate at the state*sector level available?

CPS
1

We are trying to link up workers in a certain state with the economic activity they are engaged in, and whether or not they are members of a union. Is there some way that we can link the industry a worker is in with whether or not they are in a union in a certain state? Perhaps that is not possible, but if not, we definitely want to know that.

2

Yes, that is possible using IPUMS-CPS data. Union membership is recorded in the variable UNION for each currently-employed respondent over the age of 14 that worked a wage/salary job and was in the outgoing rotation group. Since the UNION variable is part of the earner study questions, you will want to use EARNWT as your weight in any analyses that include UNION. Also, you can identify the respondent’s sector/industry and state of residence with the variables IND1990 and STATEFIP, respectively.

Hope this helps.

3

Hi there,
I was searching for an answer to a related question and found this old post. I am doing the same thing, union by industry by state, but I am finding that the cells are rather small in some places. We aggregate the Ind1990 codes into just 7 groups, so those are rather broad, but even aggregating industries we still are having some small cells. For example, there are 0 unionized workers in the Finance and Insurance industry in Arizona. Can CPS be used to get estimates of at this level of disaggregation?

Best,
Elly

4

This level of disaggregation is possible, but your accuracy will depend on your sample size. Only employed individuals in the outgoing rotation groups of the CPS are asked about union status, which translates to about 10% of the sample in any particular month. Selecting the Case-count view on the Codes page for UNION shows that roughly a thousand respondents in any month are members of a union. Therefore, it’s not surprising that you would have empty cells for unionized workers in a particular state/industry combination. My advice would be to add as many monthly surveys (perhaps even several years of data) into your sample as you feel comfortable with while keeping in mind that certain low population or union-averse states and industries may still have zero unionized members.