At the height of the recession in 2009, the average U.S. job required 13.54 years of education, up from 13.37 in 2005. The increase reflected layoffs in less-education intensive industries such as construction and manufacturing, amid job gains in industries like health care, education, and professional services that demand more education.
The authors use the term "education gap" to describe the difference between school years required for the average job and school years of the average worker. I would advise taking a very cautious and skeptical view of this construct.
1. Beware of priceless economics, by which I mean economics in which "supply" and "demand" are measured without regard to price. In this case, the authors seem to be doing wageless analysis of the labor market (I admit to not having read the whole paper.)
If the relative wage of highly-educated workers gets sufficiently high, you would expect to see a "negative gap," with employers wanting less education than the average worker. To put this another way, one would expect wage adjustment to eliminate any gap of this sort. The existence of a gap tells you something about the (lack of) wage adjustment.
2. Beware of the assumption of zero substitutability. In particular, the authors say that there are areas where the education gap is particular high. What does this say about entrepreneurs in those areas? Why are they fixated on trying to organize their industries around highly-educated workers? Why do other entrepreneurs not move to those areas with capital that is complementary to less-educated workers? For that matter, why do less-educated workers not leave those areas to move to areas with more complementary capital?
I think these issues also matter for the controversial study by Garett Jones and Daniel M. Rothschild, in which they used survey data to suggest that about half the workers hired under ARRA (known popularly as the Obama stimulus) were already employed. One way to read the survey data is that the government raised the demand for (highly) educated workers, who tended to be already employed.
I would not be interested in survey data if I thought that workers were perfectly substitutable. In that case, raising the demand for educated workers will end up raising the demand for all workers. On the other hand, if you believe in the "education gap" construct, then you have to take the survey data as definitive.
There is much more than I have the inclination to discuss here--short run vs. long run, for instance. Suffice to say that there are many ways to go wrong if you don't think things through carefully.