I recently posted on education and the workforce, evaluating the pervasive skills mismatch argument, which I found somewhat wanting, but mostly off point: there’s not a lot of evidence that employers skill demands are going unmet; there is solid evidence of a high wage premium for skill (but it’s not accelerating); and at the end of the day, our economy and society would benefit—I’d say very significantly—from a more highly skilled workforce. Which doesn’t mean everyone needs a four-plus year degree. As I put it then:
The way to understand the nexus of education and the economy/jobs is thus not in the traditional skills mismatch framework. That’s way too vague and disconnected from what’s happening on the ground. Instead, think of an old-fashioned production function where better inputs generate better outputs. Human capital is one of those inputs. The way forward is thus not to just willy-nilly advocate for greater college attainment. It’s to take a clear-eyed look at education and job/career training needs across the life-cycle. The future surely requires kids with STEM training; it also requires health technicians with AA’s who can keep that MRI percolating the way it’s supposed to. And child care workers who thoroughly understand how kids learn, and home health aides who know a lot about gerontology.
But what about kids who go to college but don’t complete? That’s turning out to be a real constraint on boosting attainment, and while I’m on record against “willy-nilly” attainment arguments, we definitely want our share of college graduates, which has been stalled of late, to start rising again.
And if we want to increase the share of kids with at least a four-year college degree, we need to increase the rate of college completion.
The share of college graduates is the product of three sequential probabilities: 1) high-school completion, 2) college acceptance, and 3) college completion.
Roughly speaking, as far as I can tell, those probabilities are ~85% 2: ~70% 3: ~50%
Which means that there’s potentially real traction along the completion margin. As shown in the figure, completion rates have been stuck at 50% since the latter 1990s. (This figure comes from the excellent book “Crossing the Finish Line” by Bill Bowen et al—it’s a must read on the topic.)
Source: Bowen et al, see text.
What factors have been identified as important re college completion?:
Attrition: it’s not just about first two years. Only half of college dropping out occurs in the first two years, but too few schools pay attention to the problem after that.
Undermatching: When students enroll in colleges that don’t match their qualifications and skill levels, they have a higher chance of dropping out, which strikes me as a real tragedy in terms of untapped potential. Note these two slides, showing the percent of 1999 entrants at North Carolina Public Universities that enrolled in less-selective colleges, based on the students’ qualifications. And note the steep gradient by both parental education and family income.
Source: Bowen et al, see text
Smoother transition from 2 yr to 4 yr colleges. I think this may be an area where some smart policy work could really make a big difference. There’s a damaging lack of standardization such that too many four-year schools don’t have much faith in what kids have learned in the two year community colleges. Online programs might be able to help here as well—I know there’s some skepticism re online learning, and you obviously need real human input as well, but some pilot programs have shown this approach to work well and lend itself to standardization in such a way as to help smooth the transition problem.
Overmatching: Interestingly, there’s less evidence for this than there is for undermatching, but I hear lots of anecdotes that too many programs focus solely on access and then leave unprepared students to fend for themselves.
Too Much Work in the Job Market: Working many hours in the paid labor market has been shown to increase attrition. Some of this is students who must make ends meet, suggesting an important role for support with the costs of living. This is particularly germane for community college students who tend to be older and often must support themselves and even their families (tuition is not the issue here—in fact, College Board data show that net of grants and tax credits, community college tuition is net negative). But that’s not the whole story. Research shows that students in computer science find they can make real money before they graduate and many choose not defer gratification, as it were. However, staying in school would often boost their longer term earning power so some financial literacy and net present value education might help here.
Income: While income turns out to be less important than you might think re completion (this was to me one of the most surprising findings in Bowen et al’s work), it does of course matter, especially for kids below the median. And this isn’t just about tuition—typically it’s more about the cost of living to attend school, as noted above.
Based on the probabilities noted above, the fact that completion is the lowest of the three, and the pretty solid research supporting the interventions noted above, I’d argue that our policy eggs should not be solely in the college access basket, but in the completion basket as well.