In the immediate wake of the Great Recession of 2007-2009 the slow pace of job growth and the stubbornly high unemployment rate sparked fears that these problems in the U.S. labor market were structural. One main worry was that  there was a gap between the particular set of skills that employers wanted and the skills that workers actually had. The truth of that narrative at the time was questionable and stronger employment growth since 2014 has alleviated these concerns over time.

But there remains something puzzling in the labor market data. The number of job openings have increased 111 percent since the end of the Great Recession compared to an increase of 41 percent for new hires, according to the Job Openings and Labor Turnover Survey from the U.S. Bureau of Labor Statistics. And the so called Beveridge Curve—a relationship between the unemployment rate and the jobs openings rate—still hasn’t shifted back to its pre-recession position. So is there at least some truth to the skills-gap hypothesis?

Examined through the lens of the Beveridge Curve, research shows that to be unlikely. A paper published by the Federal Reserve Bank of Boston breaks down the relationship between unemployment and job openings. What Rand Ghayard and William Dickens, the authors, find is that the shift in the relationship (the Beveridge Curve) is almost entirely driven by long-term unemployment. The relationship between the short-term unemployment rate and the openings rate hasn’t changed at all. That finding goes against what’d we expect if there were a skills mismatch—namely that recently unemployed workers are just as readily employed as before.

But what would explain the stronger growth in job openings compared to new hires? The answer might be that there is a mismatch between the skills employers ask for and the skills most workers boast. This is not because workers are the ones lacking in the desired skills but rather due to employers raising the bar as more people look for work.

New research by Alicia Sasser Modestino of Northeastern University, Daniel Shoag of the Harvard Kennedy School, and Joshua Ballance of the Federal Reserve of Boston find exactly that development. Their paper looks at what happened to employer requirements for positions during the Great Recession and the resulting recovery. What they find is that an increasing supply of unemployed workers leads to an increase in the requirements for jobs that employers posted. With a larger pool of talent to pick from, employers get to pick the cream of the crop.

To account for macroeconomic factors that could confound their analysis, the authors look at the return of veterans from the Iraq and Afghanistan wars. Their return was an increase in supply of workers that serves as a sort of natural experiment. Modestino, Shoag, and Ballance find that requirements went up more in occupations where veterans were more likely to be employed, exactly what would happen if we think an increase in the supply of workers leads to upskilling. They estimate that the increase in the supply of the job seekers accounts for about 30 percent of the increase in the requirements employers posted from the beginning of the recession in 2007 to the labor market bottom in 2010.

So the mismatch, as it is, really doesn’t come from unmet demand but rather from increased supply. And it’s not about a long-term trend of underinvestment in skills and talents but rather the result of a downswing in the U.S. economy. So when the new data on job openings are released later this morning by the Bureau of Labor Statistics, we’d all do well to remember this story.