As we all know, Democrats have become less interested in Raj Chetty’s work saying that the way to help poor children is to fire their bad teachers (as identified through Value Added Modeling) and keep the good ones, and become more interested in Raj Chetty’s work saying that the way to help poor children is to move them into more integrated communities. For example, here’s Vox, the voice of the Democratic Party talking to itself:
We also know what sorts of strategies actually promote economic mobility. One of them is reducing economic segregation — that is, when poor people live cut off from wealthier people. This is not a problem that Jeb or Hillary can directly address, but it is one your mayor can alleviate.
I go back and forth with myself about whether Chetty is identifying a true causal effect or the influence of race in this massive study of all of our tax records.
One vote in favor of, “nah, it’s just about race” is that the simple correlations between Chetty’s measures of economic mobility and measures of economic segregation (taken from Richard Florida) don’t go in the direction you’d think. Here, for example, is Florida’s index of economic segregation plotted against Chetty’s index of relative upward mobility for the largest cities. (To quote Demos, “Relative mobility measures how a child’s ranking in the income distribution compares to her parents’. If we use the old cliche of a ladder, it’s how far up, or down, the ladder you move compared to the quintile you were born into. “)
There is a slight positive relationship– the cities measured to have higher economic segregation have slightly higher relative economic mobility, the reverse of the alleged effect.
The pattern is similar for all metros (with the top 100 in green):
On the other hand, let’s look at absolute mobility. Demos again:
“Absolute mobility” measures something entirely different. It measures how your income compares with your parent’s income. So let’s say your parents make $10,000 a year and you are in the bottom quintile. If you stay in the bottom quintile, but make $20,000 a year, you’ve experienced absolute mobility.
Absolute mobility is negatively correlated with economic segregation, the expected direction:
Note, however, that the high economic segregation places tend to be more bustling metropolitan areas that have grown considerably in recent decades (ie, Los Angeles, Washington, San Francisco, Austin, Denver, Houston) while the low economic segregation towns tend to be places like Scranton. How much of this relationship is the fact that if you grow up in Scranton, you tend to leave Scranton, and end up somewhere where they pay you more?
On the third hand, looking at one of Chetty’s calculations that has more intuitive importance– the percent chance that someone born into the lowest 20% of income will grow up to be in the top 20%, there’s even less consistency: for the 30 largest cities, the more economic segregated, the (slightly) better your chance of ending up rich if your parents were poor:
Chetty’s estimates are based on much more robust specifications, that seem to show that the more years you spend in a place like Bucks County, Pennsylvania, the higher your adult outcome, and the more years you spend in a place like Baltimore, Maryland, the lower your adult outcome (if you’re a man.) But “places like Bucks County” aren’t particularly integrated, and “places like Baltimore” aren’t uniquely segregated in economic terms. Instead, living in majority middle-class or wealthy areas and majority white areas seems to be on average economically beneficial for lower-income children, perhaps regardless of those children’s own race. Describing this as economic integration or economic segregation just confuses things.