College Returns and Nonexperimental Data
By Arnold Kling
Construction workers, police officers, plumbers, retail salespeople and secretaries, among others, make significantly more with a degree than without one. Why? Education helps people do higher-skilled work, get jobs with better-paying companies or open their own businesses.
Or, perhaps, college filters out people with low cognitive ability, low conscientiousness, and other adverse traits. I want to see an experiment, in which some people are randomly chosen to go to college and others are chosen not to go to college. Then, proceed to compare outcomes. Meanwhile, nonexperimental data is of little or no value.
Various natural experiments — like teenagers’ proximity to a campus, which affects whether they enroll — have shown that people do acquire skills in college.
Except that proximity to college is not necessarily uncorrelated with either the skill distribution of jobs in the area or with unobserved characteristics of the population in the area. I’ve looked at these “natural experiment” studies when I wrote my critique (with John Merrifiedl) of Goldin-Katz. The methods used in the papers were dreadful.
I don’t doubt that the skeptics are well meaning. But, in the end, their case against college is an elitist one — for me and not for thee. And that’s rarely good advice.
My elitism comes from the few years I spent as an adjunct at George Mason. The typical undergrad in my course could not write a paper or solve an algebra problem. I doubt that adding more students at this margin is the way to raise people’s incomes.
Somebody should run the experiment on a small scale, before we embark on a grand social policy on this.