A number of folks have asked me about this article claiming that recent minimum wage increases may have “killed as many as 200,000 jobs.” In fact, based on a balanced look at the underlying data, the article could also just as easily have argued that these increases did not kill 200,000 jobs. Let me explain.
The article is based on a review of minimum wage research by economist David Neumark, posted by the SF Fed. Neumark does not estimate whether jobs were lost due to recent increases in the minimum wage. Instead, he estimates how much min wages have gone up in places where increases have occurred, pulls his preferred “negative elasticity” from his lit review, and does this:
Using a −0.1 elasticity and applying it only to teenagers implies that higher minimum wages have reduced employment opportunities by about 18,600 jobs. An elasticity of −0.2 doubles this number to around 37,300 [JB, why not 37,200, as that’s 18,600*2?]. If we instead use the larger 16–24 age group and apply the smaller elasticity to reflect that a smaller share of this group is affected, the crude estimate of missing jobs rises to about 75,600. Moreover, if some very low-skilled older adults also are affected (as suggested by Clemens and Wither 2014), the number could easily be twice as high, although there is much less evidence on older workers.
He thus concludes:
Thus, allowing for the possibility of larger job loss effects, based on other studies, and possible job losses among older low-skilled adults, a reasonable estimate based on the evidence is that current minimum wages have directly reduced the number of jobs nationally by about 100,000 to 200,000, relative to the period just before the Great Recession.
That’s an awfully imprecise treatment of the question, which is unfortunate given the controversy around this issue these days. And it certainly doesn’t justify what you’ll be hearing from opponents of the policy. See the Tweet below where a min wg opponent morphs the finding of “may have cost” to “have killed.” I know Neumark a bit, and I’m sure he’d object to that framing (it would useful for him to do so publically).
That said, his review is worth reading, and while I think he has a thumb on the scale in favor of finding job loss effects, he may be right. Some recent min wg increases have been to levels that take us outside the range of most of the recent research and we simply do not know their impact yet. And that of which we do not know, we should not speak. (Unless, of course, we are DC pundits).
The thumb on the scale is evident in his dismissal of a set of highly rigorous studies that find no significant employment effects, studies designed, I should say, to carefully pick up any job-loss signal from sub-national min wg increases. If he applied those elasticities, the result would be no job loss. That’s not to say those studies are definitively correct either, but his dismissal of them because “[s]ome follow-up studies…suggest that limiting comparisons to geographically proximate areas generates misleading evidence of no job loss effects from minimum wages” is not justified, as suggested by the most comprehensive review to date of the wide range of research on minimum wages: Belman and Wolfson’s book, What Does the Minimum Wage Do?
When Cornell University featured it as their book of the month in November of 2014, they called it “the most comprehensive analytical, and unbiased assessment of the effects of minimum wage increases that has ever been produced.”
Belman and Wolfson provide both a literature review and a more rigorous “meta-analysis” (a technique which combines thousands of results from hundreds of studies) of the research on the minimum wage’s effects on earnings, employment, labor force participation, school enrollment, inequality, and poverty. Their core finding is that moderate:
…increases in the minimum wage raise the hourly wage and earnings of workers in the lower part of the wage distribution and have very modest or no effects on employment, hours, and other labor market outcomes. The minimum wage can then, as originally intended, be used to improve the conditions of those working in the least remunerative sectors of the labor market. While not a full solution to the issues of low-wage work, it is a useful instrument of policy that has low social costs and clear benefits.
Neumark does not even mention this work, choosing instead to focus on his own lit review (with Wascher). However, as economist John Schmitt points out:
The Neumark and Wascher review, however, is considerably more subjective and arguably less relevant to the United States than the two meta-studies discussed earlier. Only 52 of the 102 studies reviewed by Neumark and Wascher analyzed U.S. data. Of these, Neumark and Wascher designated 19 as “most credible,” five of which were their own studies.
It’s inconceivable that Neumark is unaware of Belman/Wolfson, so really, ignoring their exhaustive review of this question of the minimum wage and job loss is more of a fist on the scale than a thumb. To be clear, there are quality studies reviewed by Belman/Wolfson that find job loss estimates of the magnitude Neumark uses here. But they exist alongside equally refined analyses that find zero impact on jobs. A balanced review can’t ignore the latter.
One more point, and it’s one that Neumark, to his credit, makes himself. Re his 200,000 estimate, he writes: “This is a small drop in aggregate employment that should be weighed against increased earnings for still-employed workers because of higher minimum wages.” It’s very important in this research to talk about both sides on the impact: costs (job or hours losses) and benefits (higher hourly pay). Even if Neumark is right about elasticities of -0.1 or -0.2, the fact that those numbers are so much less than -1 implies many more beneficiaries of the increases than those hurt by it.