Human Capital, Signaling, and Ability Bias for X: A Guest Post by Vipul Naik
By Bryan Caplan
Vipul Naik of Open Borders sent me a very insightful email on the non-pecuniary returns to education. He’s kindly given me permission to reprint it. Vipul speaks:
I’ve been thinking more about your human capital/signaling/ability
bias theories of education. It seems to me that this framework can
apply to things other than just the education -> earnings link. Other
- Education -> Health
- Education -> Self-reported happiness, life satisfaction etc.
- Education -> Marital happiness and success
- Education -> Friend quality
- Education -> Civic virtue/tolerance/progressive social attitudes
For any Education -> X link:
The human capital theory would say that: (i) People learn and retain
stuff from education, and (ii) that stuff helps with X.
Since you’ve essentially debunked (i), this means that the human
capital theory has a strong presumption going against it for Education
-> X linkages for all values of X, not just X = earnings. Thus, for
most of the linkages, we have a contest between ability bias and
The ability bias would say that this is purely correlation — people
who have better X also tend to get more education.
The signaling theory, in contrast, does not make sense for most values
of X. For instance, I’m hard-pressed to think of a signaling theory
for the Education -> Health relationship. In fact, the only areas
other than earnings where I can see signaling playing a role is in
mate selection/marital success and friend quality. I guess you could
argue that education indirectly affects all the others through
signaling via its effect on earnings, but if your view of, say,
health, is Hansonian, then you’re probably skeptical of earnings being
causally implicated in health outcomes.
My guess is that education -> X linkages for most values of X are
purely correlational (ability bias) with a very tiny amount of
causation through the human capital channel. If so, your signaling
theory for earnings is the exception rather than the rule in terms of
education -> X linkage explanations.
P.S. I’ve already made this point for the marriage premium, and shown that the effect of education on job satisfaction appears to be more than 100% due to ability bias (i.e. holding income constant, the well-educated have lower job satisfaction). Note also that Oreopoulos and Salvanes’ “Priceless” in the 2011 Journal of Economics Perspectives tries to do what Vipul suggests, though not very successfully in my view.