Jan 23, 2013 at 9:35 pm
In my earlier post on shrinking union density, I noted the following: “…you don’t see systematic differences in job creation favoring, for example, countries or states with low versus high unionization rates.”
Here’s a simple look at what I mean. The theory among many who disparage unions is that they distort the labor market, creating inefficiencies that may help their members’ paychecks but hurt jobs and growth. If so, you might expect to see a negative correlation between the presence or growth of unions and employment growth at the state level. But you don’t.
The graph plots a scatter of the change in union density by state, 2007-11, against the change in employment rates. Did states where unions decline the most fare better in terms of employment rates over the recession? Not by this simple plot. It pretty much a random scatter, with the regression line sloping “the wrong way”—up—implying states were unions grew most (or fell least) had the best employment growth (or the smallest employment losses). But the slope is statistically insignificant.
The little table below looks at other time periods, including a longer sweep from the 1989 cyclical peak to the 2007 one. It also adds the variable of (log) employment growth. Most coefficients in the table have the “wrong sign” or are insignificant, except the bold one, suggesting, counter to anti-union theory, states where union density grew most had the best employment growth.
By regressing the change in employment rates, 2007-11, on the level of unionization (as opposed to the change) in 2007, one can also test the theory that states with the least union density were in the best position to handle the labor demand shock of the great recession. But the coefficient is positive (wrong sign again) and statistically insignificant. Same if I use (log) job losses, although here, if you remove an outlier—Alaska added a bunch of jobs (energy extraction) and has a 24% unionization rate (see why these regressions go the way they do?)—you do get a significant negative coefficient (-0.002, t-stat: -2.2)—but it takes some work to get there!
This is not anything like a rigorous model—it’s just simple correlations—but you often find in this biz that if the correlations aren’t there are this basic level, you’re not likely to find much without some serious data cooking. Like I said earlier, you want to inveigh against unions? I can’t stop you. But unless you’ve got the evidence, don’t do so based on the notion that they hurt job growth.
[Data note: I used the union density and employment counts by state from the (very excellent) Unionstats.com site, and employment rates by state from the BLS site.]
Thank you for joining the conversation. Comments are limited to 1,500 characters and are subject to approval and moderation. We reserve the right to remove comments that: