The debate among economists over whether (higher) minimum wages cause (more) unemployment seems to go on forever. Most published results (and an even greater fraction of methodologically sound published results) find that minimum wages reduce employment. But this doesn't settle the result, and smart, eminent economists like Arindrajit Dube publish papers arguing that with different methodologies and/or correcting for appropriate trends, you cannot find a link. Though this seems to go against our basic model of the microeconomy, economists argue that employers of the low-skilled are monopsonists—single buyers, the converse of monopolies, single sellers—and may even want more labour when its price rises. Those who do believe that minimum wages put people out of work, when faced with results finding they don't, suggest that maybe this just a temporary effect; in the long run workers will be replaced with automation a la modern checkouts, or fewer higher-skilled people once they've had time to search. Or alternatively, less will be spent on working conditions or other benefits.

Given this morass of disagreement, a new paper from Jeffrey Clemens and Michael Wither is unlikely to change any minds, but it's still an interesting result. They use nifty controls to establish that binding minimum wages cause unemployment, lower wages from work, and worse work progression for the low-skilled. This dearth of work experience, the authors say, makes them less likely to join the lower middle class. What's more, because their sample is very broad and spread out across the entire US, they believe their results are uniquely good for extrapolating to the country as a whole.

We estimate the minimum wage's effects on low-skilled workers' employment and income trajectories. Our approach exploits two dimensions of the data we analyze. First, we compare workers in states that were bound by recent increases in the federal minimum wage to workers in states that were not. Second, we use 12 months of baseline data to divide low-skilled workers into a "target" group, whose baseline wage rates were directly affected, and a "within-state control" group with slightly higher baseline wage rates. Over three subsequent years, we find that binding minimum wage increases had significant, negative effects on the employment and income growth of targeted workers. Lost income reflects contributions from employment declines, increased probabilities of working without pay (i.e., an "internship" effect), and lost wage growth associated with reductions in experience accumulation.

Methodologically, we show that our approach identifies targeted workers more precisely than the demographic and industrial proxies used regularly in the literature. Additionally, because we identify targeted workers on a population-wide basis, our approach is relatively well suited for extrapolating to estimates of the minimum wage's effects on aggregate employment. Over the late 2000s, the average effective minimum wage rose by 30 percent across the United States. We estimate that these minimum wage increases reduced the national employment-to-population ratio by 0.7 percentage point.