Sunday, June 01, 2014

The college degree premium and the Lucas critique

Bob Lucas showed us that relying on reduced form historical correlations to predict the effects of a new policy regime is likely to end in tears.

Yet this seems like exactly where we are going with government policy toward higher ed.

Historically, college degree holders earn more money and have lower unemployment rates than to non-degree holders.

So, in an attempt to increase earnings and employment, we are pushing for "everyone" to get a college degree.

Here's David Leonhardt from "The Upshot", after showing that the wage premium for a bachelor's degree has never been higher:

"Those returns underscore the importance of efforts to reduce the college dropout rate"


"At some point, 15 years or 17 years of education will make more sense as a universal goal. That point, in fact, has already arrived."

People, I know this sounds bad, but maybe it's because so many people drop out that the degree premium is so high.

Maybe if twice as many people got a bachelor's degree, the wage premium would fall dramatically.

Degree attainment and wages are outcomes of a complex, simultaneous structural socio-economic model. Making policy recommendations based on one reduced form relationship from that model without an understanding of the deep parameters is very bad science and the recommended policy is extremely unlikely to have the desired results.

It may seem strange for a college professor to be arguing that not everyone should be getting a college degree. But I have seen kids drift along for years, racking up debt and not human capital only to drop out and struggle. I have also seen thousands of kids make an economically unwise choice of major, get their degree and then struggle mightily.

Not all "colleges" produce that big premium and more importantly, not all majors produce a big premium, as I discussed in an earlier post.


J Scheppers said...

According to PEW using Census data college completion has increased from 12% to 31% from 1971 to 2012. I would suggest that moving the cohort from 12% to 31% from no degree to degreed with no change in knowledge would reduce the performance measure of the non-college graduate more than the decrease to the college graduates.

I am not arguing that college has no effect, but a sizable change in the ROI on a degree is simply related to the size of each of the cohorts.

blink said...

Yes, the Lucas-style critique is spot-on. Also missed by Leonhardt and others is that *even if* the difference is reflects skill, a change in the college premium is just as easily (more easily?) explained by a lowering of our standards for a high school diploma. Instead of raising the ceiling, lower the floor.

Trapper_John said...

Well, the world needs ditch diggers, too.