November 24, 2022
I was very fortunate to be asked by Ellora Derenoncourt to share some of work I’ve done with labordata.bunkum.us to labor economists and union researchers and activists at a conference this fall.
Ellora asked me to describe the opportunities I see with the data, and here’s what I put together. My comments are limited to the public data, as that’s what I have worked the most with.
The core information systems of the agencies that collect data on the labor movement are digital and have been for some time. We have digital case management files for NLRB going back to 1984. Portions of LM10 and LM20 have been digitized for a long time but the Department of Labor finally moved to all digital a couple of years ago.
Previously, the effort of digitizing paper records was so burdensome that it stopped most analyses before they started. Mainly, this is no longer a problem.
Now the two main problems are
With the exception of the Current Population Survey and the NLRB case file data, most of the important federal data sources are published in a form unsuited for analyses. The relatively good condition of the NLRB data is a major reason why it has been so used in research, even though the limitations of that data are well understood.
The FMCS’s F-7 data should be one of the most valuable data sources on the private-sector labor movement. These are statutorily required notices of the intent to bargain for bargaining units covered by the NLRA. Each notice indicates the employer, the union local, the industry, and the size of the bargaining unit at the time of the filing.
It’s data about a stock, so a nice complement to NLRB data. It should include units formed through card checks and voluntary recognitions. It could inform us about the life-cycle of bargaining units: their birth, growth, decline, and disappearance. It can tell us about the opportunity structure for strikes.
There has been some excellent research that took advantage of FMCS data, but it is hard to work with. Ideally, the filings, which are effectively snapshots of a bargaining unit, would be linked over time and standardized into a longitudinal database. That is a large effort in data cleaning.
The tools to make that cleaning more tractable are getting better. I’ve been working on a software package that uses machine learning to standardize messy union local names against the names tracked by the OLMS. If other folks want to help with this lift, please let me know.
If the F-7 data were in reasonable shape, then that would significantly enhance the OLMS data. A great problem with the OLMS data on unions is that there is no consistent measure of membership. There is great variety within and between unions and for the same union across time. If the F-7 data could provide a separate estimate of the number of local-union members currently in a bargaining unit, that would be a great complement to the LM data.
So much for the private sector. The other great problem with public records about the labor movement is federalism.
Over 80% of public sector workers are employed by state and local governments. We have some of our best information about the conditions of this field of organizing from Kate Bronfenbrenner and her colleagues’ attempts to get snapshots by gathering information on filings and elections from every state possible. We do not have continual access to this data however.
Here the solution is likely not technological but organizational.
Without substantial sustained funding, a single organization could not take on that workload every year, but perhaps a consortium of organizations could. In Chicago, we helped build the Chicago Data Collaborative among researchers and journalists that pooled collected public records to build one of the most comprehensive collections of records related to criminal justice in Cook County, and maybe another approach is possible here.
In sum, using some better technology and a lot more cooperation, we can have a much better picture of the dynamics of the labor movement though making much better use of existing public records.
Subscribe to get Notes on Labor Data as an email newsletter.