Looking at cross sectional evidence, namely the 50 states plus D.C., we get evidence supporting the existence of the curve above. First, using regression analysis that controls for some other possible causes of inequality using 2000 Census data, we find that states with high proportions of college graduates on average had also higher amounts of income inequality—more college education was associated with greater disparities between rich and poor. Yet doing the same sort of analysis using the 1970 Census, when only a small proportion of adults had degrees, gives the opposite results: The states with high proportions of college graduates had less income inequality—more “college intensive” states had more equal income distributions.
A bit more sophisticated evidence using the 2000 census data (directly econometrically estimating a non-linear curve like in the graph) also is highly consistent with the curve above. That evidence would suggest the equality maximizing proportion of adult college grads is well below the 30 percent or so we have today. If this is true, “progressives” who are concerned with income equality should oppose efforts such as that of President Obama to expand the proportion of college graduates dramatically.
Why might these results exist? I suspect the law of diminishing returns is at work yet again. When college attendance/graduation is relatively rare, expanding it does bring in a number of able students who graduate and become productive members of the labor force, significantly increasing incomes of persons otherwise not likely to earn as much. But when the proportion of adults gets large and there is a growing disconnect between the pool of graduates and the pool of high paying technical, managerial, and professional jobs, the incremental graduates do not fare as well economically. More of the incremental students are also more ordinary, less able, less bright.
http://chronicle.com/blogs/innovati...e-equality-new-evidence-and-conjectures/30136

