I'd love to see Wages = b0 + b1 * Bachelor's degree dummy + b2 * parents income + b3 * IQ + b4 * [thing I can't say] + b5 * [other thing I can't say] + b5 * work ethic measure (HS GPA?) + e There are a million things wrong with doing it this way but it would be an interesting addition to looking at that graph by itself. I half-heartedly tried to Google a decent paper but didn't have much luck with anything newer than the 70's
I propose a businessman whose output cannot be measured very well earns p/r+log(e)+k, where p is the profit the company makes, r is his rank within the company (CEO is 1, CFO/CMO is 2, etc), e is the number of employees beneath him, and k is the base salary at that firm.
The idea here is that upper management at certain companies effectively own the firm and distribute profit accordingly, while middle management usually advances in income by increasing empire.
Measuring the effect of education is one of the most difficult econometrics tasks due to the way that most societies pursue education -- one year at a time until you're done with it. There's not enough variation.... I was about to launch into a detailed technical explanation, but I think I'll just leave it at the intuitive one.
