oth progressives and conservatives have sound reasons to support better data collection on public-sector unions, whose members constitute half of the American labor movement. Progressives believe that public-sector unions can uplift workers, push back against big business, and ensure a more equal society. Conservatives believe that they impose excessive costs on taxpayers and induce bureaucratic sclerosis.
The trouble with these contending views is that, without good data, it’s hard to apply the tools of social science that might help settle the debate—if indeed such ideologically charged issues can be settled. As things stand, much of the data cited in the press to support one side or the other is misleading.
Consider the basic issue of public-sector union membership. We can assess many important questions only if we know how many people belong to unions in state and local government. Do most public workers actually want union representation? How representative are the unions with which government employers negotiate contracts? Does having a large membership cause a union to spend more on political activity? Does high membership increase political influence and lead to higher taxes, lower inequality, or other policy outcomes?
Four years ago, the Supreme Court’s ruling in Janus v. AFSCME stipulated that public-sector unions in 22 states could no longer force those who did not want to be members to pay into union coffers. The ruling stripped these unions of a powerful tool to build membership. With unions no longer able to require nonmembers to pay so-called agency fees, the question became whether existing members would opt out of membership altogether because they could now receive union representation for free.
Answering that question turns out to be harder than one would think. Government does not collect good data on even the simple question of public-union membership. The Bureau of Labor Statistics, a federal agency, conducts the Community Population Survey (CPS), which asks public employees whether they are union members. Looking over the past four years since Janus, the BLS reports relative stability in membership.
But the CPS is not a good instrument for measuring short-term trends in public-sector union membership. The results provide only a total union membership estimate, undifferentiated by public and private sectors. In addition, many respondents don’t know or are unsure about their membership status, which muddies the results.
Breaking down the CPS data by state and sector, as the website UnionStats.com does, yields very small sample sizes. As a result, UnionStats sometimes reports large swings in membership or public employment overall in some states when there’s no evidence that such changes occurred.
Another public source of data comes from union financial reports, called LM-2s, filed annually with the Department of Labor’s Office of Labor-Management Standards (OLMS). These show modest declines in public-sector union membership, but this information source has problems, too. The numbers are not compiled systematically, and it can be hard to sort out which pertain to private- and which to public-sector union membership in the total figures. These reports are also not comprehensive because not all public-sector unions must file—only those with one private-sector employee in their ranks. (This is because federal law governs private-sector unions, while state law governs public-sector ones.) Finally, these reports are only as rel…