6 Comments
User's avatar
Evan Haldane's avatar

I think the assumption that degree difficulty (ie signalling cost) depends only on individual IQ (i.e. is constant with respect to A) cannot be correct. Degree difficulty must be higher in higher-intelligence societies (holding all else equal). So K is some increasing function of A.

Basically it’s somewhere between “top 30% get a degree regardless” and “whoever is smarter than x gets a degree, where x is constant for all societies”

Expand full comment
T Coddington's avatar

Right, this is the line that led me to the same objection you have...

"If you could increase IQ by sprinkling pixie dust on the population, education levels would rise despite the fact that education has zero effect on productivity."

Presumably if pixie dust made us all 50% more intelligent, universities would change in a way that made obtaining a degree harder (relative to today), so the signalling cost would not go down by 50% but something between 0-50%.

Expand full comment
Bryan's avatar

Narrowly specific and hypothetical models are the best models, I say. Fascinating read!

Expand full comment
Joe Potts's avatar

Yes. The broader the implications of a model, the less credit I give them.

Expand full comment
Emil O. W. Kirkegaard's avatar

IQ is not uniformly distributed, but normally distributed. I don't see why you need this wrong assumption in your model.

Expand full comment
Sebastien's avatar

So, this model would explain rising education as a reaction of to a rising productivity gap?

Expand full comment