High Variance

Mincer 1958

Jacob Mincer’s 1958 paper in the Journal of Political Economy, “Investment in Human Capital and Personal Income Distribution”, is one of the most well-cited in economics–304 times according to Web of Science. And yet, I would be shocked if more than half the people who’ve cited the paper have even opened it and read the first page. The reason it’s cited so often is that Mincer did indeed write down the famed Mincer equation here, but he wasn’t the first to notice that log wages are an approximately linear function of years of education. What’s special about this paper is that Mincer was the first person to write down a simple utility-maximizing model from which you can derive such an equation.

Wages are not normally distributed. The lower half of the distribution is all clustered together, but the high end is really spread out. And all those one percenters at the high end are making far more than average wages. But physical dimensions like height, head circumference, and forearm length are normally distributed. And though it’s a lot harder to measure, most scholars believe that cognitive ability is normally distributed too. In the early part of the twentieth century, this was seen as something of a paradox–if physical and cognitive ability are the major determinants of productivity then shouldn’t wages (which measure productivity) also be normally distributed?

The average gain in wages that comes from an additional year of schooling increases with the amount of schooling someone has. Expected wages go up by more for the 16th year of schooling than the 12th year of schooling. One of Mincer’s insights was that a big part of the non-normality of the wage distribution is a direct consequence of these non-linear returns to schooling.

Mincer had great intuition. He recognized that a person must eventually be compensated for wages they could have earned during the years they attend school. The more schooling you already have, the more you need to be compensated for getting an additional year of schooling. Mincer shows that in his model, wages must increase multiplicatively for each year and that is exactly what happens when the log wage is a linear function of years of education. In the data, wages are generally associated with about a 15 percent increase for each year of schooling.

Academic economists worship at the altar of the new and technically sophisticated. This means the vast majority of econ papers written before the 1980’s just languish on library shelves. And even though it’s true that most of these older papers aren’t terribly relevant, I think Mincer (1958) is just one of several diamonds in the rough. I’m always on the look out for more, and would welcome any suggestions.