This post has two parts: the first, a bit of a rant; the second, some empirical research.
First off, I am becoming increasingly disillusioned with the quality of economic research that appears in even the most distinguished "scientific" journals -- by which I mean journals like SCIENCE and NATURE (and not the heap of neoclassical journals that I have long since assumed were garbage).
Science Magazine recently published a special edition entitled Haves and Have Notes: The Science of Inequality. In this issue is an article by MIT economist David Autor entitled Skills, education, and the rise of earnings inequality among "the other 99 percent".
First off, it should be noted that Autor's article contains some excellent empirical analysis (although much of it is a synthesis of empirical work done by others). Autor documents the increasing pay gap between college graduates vs. high school graduates in the US. The change in the pay gap nicely corresponds with the change in inequality over the same time period.
Autor's proposes that a simple supply and demand mechanism can explain the change in the education pay gap. He suggests that since 1980 there has been a systematic under-supply of college graduates, which has led to an increase in their relative pay.
But crucially, in order to test the existence of any causal force, we must be able to measure it. Autor measures the relative supply of high school/college grads in terms of the number of recent entrants of each group into the workforce. This seems fair enough.
How does Autor measure relative demand? Here is where I become incredulous that this article was published. Autor does not even attempt to measure the relative demand for college vs high school grads. This would involve asking employers highly subjective questions like: how many college graduates would you like to employ? If the majority of employers responded that they would like to employ more college grads than they currently do, we might surmise that there was a supply/demand imbalance.
Instead of attempting to actually measure the relative demand for college grads, Autor simply assumes that it is an exogenously increasing function of time! But this means that the supposed supply/demand imbalance is never actually measured ... it is invented from thin air by the introduction of an arbitrary function for demand.
I expect this sort of academic debauchery from neoclassical journals. I do not expect such debauchery to get published in prestigious journals like Science.
Now to the empirical research. The "production" of highly educated workers is no small feat. Massive amounts of resources must be devoted to the task. For instance, in order to produce a PhD graduate, society must support this person for at least 20 years of schooling. Not only does society subsidize the personal consumption of the potential PhD during this period ... society must also support an enormous institutional structure -- the university system.
How, then, can a society increase the relative production of highly educated workers (graduates as a portion of the population)? The obvious way is by increasing the productivity of the workforce. Increasing output per worker means that each worker can support more people. More resources can be devoted to the production of highly educated people without sacrificing the living standard of the rest of society.
But there is only one way to increase worker productivity -- substitute non-human for human work (by using machines). But this requires an increase in energy consumption.
Thus a reasonable hypothesis is that the production of highly educated workers is fundamentally dependent upon the rate of energy consumption per person. The figure below lends support to this hypothesis. Here I plot the annual number of new PhD gradates in the US (as a portion of the total population) against US energy consumption per capita. The two series are tightly related.
Given a future that almost certainly includes energy constraints, this evidence suggests that the future ability of industrialized nations to produce educated workers will decline.