People with more education don’t just make more money if they have jobs; they’re more likely to have jobs in the first place.  As a result, the earnings premium now greatly exceeds the wage premium.  Consider the following caricature approximation of modern male earnings:

(a) College grads with full-time jobs earn 70% more than high school grads with full-time jobs.

(b) 95% of college grads, but only 70% of high school grads, have full-time jobs.  Everyone else receives $0 of income.

College grads in this scenario enjoy a 70% wage premium but a 131% earnings premium.  This is easy to compute: 1.70*.95/.7=2.31, indicating that college grads earn 2.31 times as much as high school grads.

Ability bias aside, is this computation appropriate?  It depends.  If a college degree really reduced your probability of involuntary unemployment from 25% to 5%, then multiplying the wage effect and the employment effect is fine.  College is your ticket out of a hellish situation.

But what if your unemployment is voluntary – in the sense that, given the market wage for people with your skills, you prefer not to work?  Then multiplying the wage effect by the employment effect seems like double-counting.  Yes, college raises what you actually earn by 131%, but it only raises what you’re able to earn by 70%.

Not convinced?  Suppose college graduates worked more solely because college speeds up your hedonic treadmill.  After four years of high-achievement socialization, college grads feel like losers unless they earn at least $100,000 a year.  High school grads, in contrast, feel perfectly satisfied with half that.  Or suppose that colleges subtly indoctrinate women with the view that material success is much more important than maternal success.  Why should merely shifting priorities from kids to cash count as a personal “benefit”?  This is just the crude materialism economists are often mocked for espousing.

Now recall that official statistics distinguish between the employed, the unemployed, and those who are “out of the labor force.”  The lazy response to my point is to treat the officially unemployed as “involuntarily unemployed” and the officially “out of the labor force” as “voluntarily unemployed.”  So if college grads’ employed/unemployed/out of the labor force breakdown is 95/3/2, and high school grads’ breakdown is 70/10/20, the “correct” college earnings premium is 1.70*(95/98)/(70/80)=94% – more than 70%, but a far cry from 131%.

The diligent response, though, is to adjust official unemployment numbers for the “discouraged worker” and “prideful worker” effects.  At least outside of deep recessions, I tend to think that the latter effect is larger.  But either way, you’ve got to admit: Unless everyone without a job is involuntarily unemployed, counting the full employment gap as a “benefit” of education is not reasonable.