When economists measure the return to education, they usually assume that all years of schooling increase earnings by the same percent.  A standard regression would be:

log (Earnings in dollars)= a + b*(Education in years)

where b is the measured return to education.  If b=.06, this means that an extra year of education boosts your earnings by 6%.

In reality, however, the typical year of education pays very little.  Most of the financial reward of education comes from finishing degrees.  Since diplomas used to be written on sheepskin, this finding is known as the “sheepskin effect.”

Researchers usually interpret sheepskin effects as signaling.  If finishing your last year of college sharply boosts your income, the reason probably isn’t that colleges withhold the financially lucrative material until your senior year.

What I didn’t know until recently is that the sheepskin effect declines with age.  Belman and Heywood’s “Sheepskin Effects by Cohort: Implications of Job Matching in a Signaling Model” (Oxford Economic Papers, 1997) finds that sheepskin effects are large for 24-30 year-olds, moderate for 31-40 year-olds, and small thereafter.  Their earnings regressions for 24-39 year-olds:



The middle school sheepskin effect is “Dummy ed. ≥8,” the high school sheepskin effect is “Dummy ed. ≥12,” and the college sheepskin effect is “Dummy ed.≥16.”  In plain English, finishing high school pays a 7.6 percentage-point bonus in your twenties, and a 6.3 percentage-point bonus in your thirties.  Finishing college, similarity, pays 11.3 extra percentage points in your twenties, and 5.5 percentage-points in your thirties.  This is all over and above the modest per-year payoff.

Now see what happens later in life:


The sheepskin effect for high school graduation falls to 4.4 percentage-points in your forties, 1.9 percentage points in your fifties, and 2.4 percentage points in your sixties.  For college, the analogous bonuses are 2.7%, 1.8%, and a very imprecisely estimated 8.8%.  None of these are statistically significant, but by wage regression standards, they’re still sizable.

What does all this mean?  If you’re eager to dismiss the signaling model, you’ll hail the qualitative result: Signaling effects decline over time.  If you focus on the actual quantities, however, you’ll reach a different conclusion: Large signaling effects endure for decades.

Even more importantly, the fact that the sheepskin effect is front-loaded means that sheepskin effects are a large share of the present value of the return to education.  During your twenties, the return to college is roughly 50% sheepskin, 50% other.  In your thirties, that breakdown falls to about 25% sheepskin, 75% other.  But since you earn the big sheepskin effects first, those years count more.  For example, with a 5% interest rate, a 24-year-old can expect the sheepskin effect to provide 38% of the expected payoff of college during his twenties and thirties. (My calculations).

Does the labor market ever see through mere credentials?  In the long-run, it usually does.  But the glass is half empty: The long-run takes decades to arrive.  By the time the market sees through mere credentials, the merely credentialed have already reaped large rewards.  They never have to pay them back.  And all you need to do to join this gravy train is finish your degree.