Economist Tyler Cowen of Marginal Revolution interviews Stanford economist Raj Chetty and borrows a number of his questions from my appreciative 2015 critique in Taki’s Magazine, “Moneyball for Real Estate,” of the flaws in Chetty’s methodology in his huge and much publicized study of how income mobility over the generations varies by county across the United States.
Chetty is trying to find out what are the local policies or customs that make one county a better place to raise your kids than another county. I think that’s a good topic, but he is still a long ways from finding those kind of subtle answers because his results are still clouded by three big methodological problems. As I concluded two years ago:
In summary, Chetty’s data still suffers from crippling problems with:
- Regression toward the mean (especially among races)
- Temporary booms and busts
- Cost of living differences.
Yet, these should not be impossible challenges for him to overcome in future iterations.
Here’s the transcript of the interview (and the podcast if you like listening rather than reading). Excerpts:
… On drivers of upward mobility
COWEN: Let’s go now to some of your research on mobility, which is maybe, at this moment, what you’re best known for. You can identify counties or parts of the United States where mobility for generations is going to be especially high. To what extent do you think that’s picking up that simply some of those regions end up with resource booms or other good events that is, in a sense, just random? It doesn’t per se have to do with the region? Or do you think we can adjust for that?
CHETTY: Yes. Some of what drives upward mobility, of course, is just having a very vibrant economy. To give you an example, parts of North Dakota, with the natural resource boom there, we see are having very high upward mobility. Of course if you discover natural resources, that’s going to help more people move up the income distribution.
But by and large, that is the exception rather than the key driver of the differences in upward mobility that we find across places. I say that for a couple of reasons. First, even if you hold fixed the rate of growth, the rate of economic growth, you find that some places have much higher rates of upward mobility than others.
To give you an example, Atlanta is a city that’s booming in terms of jobs and economic growth overall. But Atlanta’s one of the places with the lowest levels of upward mobility for kids growing up in low-income families there.
Sorry, but an obvious explanation for why the Atlanta metro area has low upward mobility is the classical statistical phenomenon of regression toward the mean. Atlanta, unlike Chetty’s favorite metro area, Salt Lake City, is heavily black. Atlanta has one of the most prosperous and best educated black communities in America, but blacks in America still regress toward a lower mean income than do whites, so it’s almost statistically inevitable that Chetty would find that Atlanta has lower upward income mobility than Salt Lake City.
The second thing you see is these rates of upward mobility, to the extent we have data, they tend to be quite persistent overtime. It’s not like the places that have high upward mobility in one decade, suddenly a very low upward mobility in the next decade. It’s a pretty persistent phenomenon.
A striking example of that is states in the middle of the US, like Iowa for example, which historical data going back to work by Claudia Goldin and Larry Katz has always looked like a place with very good outcomes for kids in lower-income families. And what’s amazing about that is, Iowa suffers from a brain drain phenomenon where the most successful people often end up leaving the state, going to Chicago, going to New York, to get higher-paying jobs. Yet generation after generation, Iowa seems to produce very good outcomes for low-income families. So that again suggests it’s not about natural resources or temporary booms. It’s something more persistent.
Chetty’s single best county in America for blue collar kids’ upward income mobility is Sioux County, Iowa, where iSteve commenter The Last Real Calvinist is from. Somewhat like Mormon Salt Lake City, Sioux County is famously Dutch and socially conservative. Of course, like all the top 25 counties on Chetty’s list, it’s also extremely white.
But there is some evidence from Chetty’s research that social conservatism is good for blue collar kids’ future earnings. Unfortunately, he needs to adjust for the three big problems I identified above to test his hunch.
CHETTY: Yes. Where did that come from? Why does Iowa have good public schools?
CHETTY: One of the strong correlates we find is that places that are more integrated across socioeconomic groups, that have lower segregation, tend to have better outcomes for kids. And that kind of thing in a rural area — you can see why that occurs and why it might lead to better outcomes.
Obviously, Chetty is being silly here. Sioux County, Iowa and his other 24 top counties are not at all what normal people would call “integrated.” They are extremely white.
Instead, he’s using “integrated” as a euphemism for “heavily white and/or Asian,” and “segregated” as a euphemism for “heavily non-Asian minority.” Liberal sociologist Philip N. Cohen pointed out:
Philip N. Cohen pointed out how Chetty’s 2014 paper tries to euphemize the role of blackness behind related factors like de facto “segregation:” in a 2014 blog post entitled “Where Is Race in the Chetty et al Mobility Paper?”
Instead, they drop percent Black for racial segregation. I have no idea why, especially considering … [I]n these normalized correlations, fraction Black has a stronger relationship to mobility than racial segregation or economic segregation! In fact, it’s just about the strongest relationship on the whole long table (except for single mothers, with which it is of course highly correlated).
Back to the interview:
If you live in a big city, it’s very easy to self-segregate in various ways. You live in a gated community, you send your kids to a private school. You essentially don’t interact with people from different socioeconomic classes. If you live in a small town in Iowa, pretty much there’s one place your kids are going to go to school. There’s one set of activities that you can all participate in. And that is likely to lead to more integration.
But living in a small corn-farming town in Iowa will not lead to more racial integration.
COWEN: And you think that’s causal rather than just restating the same fact about the quality of the place?
CHETTY: We don’t have definitive evidence on this, and we’re working on trying to establish clear evidence. But our sense is that integration — actual contact with people from other backgrounds — is a strong predictor and likely a causal determinant of kids’ long-term outcomes. I suspect that that’s one major factor in what’s going on.
This is basically Charles Murray’s Coming Apart view that growing up in Newton, Iowa (home to Maytag) was a healthy social environment because the children of Maytag execs played with the children of local tradesmen. These days, however, Maytag executives mostly live in an upscale suburb of Des Moines and reverse commute 35 miles to Newton. But this doesn’t have much at all to do with racial integration.
It would be perfectly reasonable for Chetty to maintain a stance of agnosticism toward the ultimate causes of the national race gaps in mean income. He could admit that “Until whatever it is that causes the races in America to have different average incomes stops happening, it’s almost statistically inevitable that my methodology will show lower income mobility in heavily non-Asian minority counties than in counties that are heavily white and/or Asian.”
But he doesn’t want to admit that, perhaps because there isn’t much interest in America in how we can influence, say, whites to be more like the whites in Sioux County, Iowa and less like the whites in McDowell County, West Virginia.*
All the excitement and money and prestige is in Closing the Race Gap. Chetty encourages people to assume that his study will find out how to close The Gap between blacks and whites, even though his study is almost hopeless at finding that (because even though he has your IRS returns and mine, he doesn’t have the race of taxpayers due to the IRS not collecting it and the IRS anonymizing the data before giving it to Chetty).
* McDowell County, WV has been notorious since JFK’s famous visit for having the poorest, most backward, most self-destructive white hillbillies in America. And yet, even McDowell Co. does fairly well in Chetty’s measures of upward mobility: such is the power of Regression Toward the Mean. McDowell County comes in at the 46th percentile in Chetty’s rankings, just slightly below the average county in America.
It would be perfectly reasonable for Chetty to adjust for regression toward the racial mean in some fashion so he could look for more subtle drivers of income mobility.
If that’s not feasible, Chetty could simply end up doing what Charles Murray did in Coming Apart and in much of The Bell Curve: just compare highly white counties to other highly white counties. There might still be something interesting to find, although I suspect his white vs. white results would tend to echo David Hackett Fischer’s Albion’s Seed.
Overall, there’s a fair amount of evidence that Chetty’s study really is on to something, which is why I want him to clean up its three big remaining methodological problems of Regression Toward the Mean, Temporary Booms and Busts, and Cost of Living Differences.
If Chetty would take those problems seriously and fix them, he might actually get some results in what he’s been hoping to find out about government policies and social norms that make a difference, pro or con, for the next generation.
Commenter The Last Real Calvinist, who is from heavily Calvinist Sioux County, Iowa, Chetty’s #1 Best County in America explains the trick Chetty is pulling on audiences:
The Last Real Calvinist
I read the whole interview; it’s pretty remarkable stuff.
I think your take (and Steve’s) on this is right; Chetty is referring on one hand, initially, strictly to socio-economic — i.e. class — integration in places such as Sioux County. And he’s right; I went to school with the children of doctors and lawyers and bankers as well as those of farmers and manual workers.
And then he’s saying that this kind of integration is promoted by limitations on options, e.g., presumably, living in a boring, social-capital-scarce setting leads to better outcomes:
If you live in a small town in Iowa, pretty much there’s one place your kids are going to go to school. There’s one set of activities that you can all participate in. And that is likely to lead to more integration.
But then he conflates this class-based integration with — presumably, although he doesn’t come right out and say it — racial/cultural integration:
But our sense is that integration — actual contact with people from other backgrounds — is a strong predictor and likely a causal determinant of kids’ long-term outcomes.
At the least, he’s leaving the door open here to those who would like to interpret his findings in a way that fits The Narrative. That is, successful outcomes are the product not of Sioux County’s cultural homogeneity, but rather of its ‘diversity’.
Surely there are no other identifiable factors that could possibly be leading to the good outcomes for Sioux County kids.
It’s also interesting to juxtapose this class-integration success narrative with Chetty’s own experience in his ‘outstanding college prep school’ as the son of an economist and a medical specialist. How intrepid he must have been to overcome his limited chances to integrate with the proletariat! I wonder how his school managed to compensate for its no doubt shocking deficit in magic class integration opportunities?
Or perhaps I’m assuming too much, and his school instead carefully assembled an alchemically-potent mixture comprising the correct proportions of children from all classes, no doubt mimicking the class breakdown in Sioux County’s schools?
By the way, in my Taki’s Magazine article on Chetty, I speculated:
Fertility is actually a promising avenue for Chetty to pursue in the future. As we’ll see below, his income calculations are stricken with problems, but he appears to have the data to estimate the answers to questions such as: where should you move if you want your child to present you with a legitimate grandchild by the time you are, say, 70? That is the kind of thing you aren’t supposed to discuss in public these days, but I’d be surprised if Mr. and Mrs. Chetty don’t worry about it.
It turns out my speculation was largely on the money:
On geography and gender
COWEN: Yes. Have you thought much about within this country, geographic differences in gender inequality? …
CHETTY: Yeah, that’s a very interesting question. We find sharp differences in outcomes by gender across areas for various reasons. Let me give you a couple of examples. One, we find that areas with more concentrated poverty — take the city of Baltimore, for example — we find very poor outcomes for boys in particular, relative to girls, and we think that that has to do with crime, and getting involved in gangs, and so forth — things that girls are less likely to do.
As a result, growing up in a place like Baltimore turns out to be extremely detrimental for boys. We estimate that you lose something like 30 percent of your earnings relative to if you’ve grown up in an average place in America. Whereas for girls, it’s slightly negative but not nearly as bad. There are a set of urban ghettos, places with concentrated poverty, that tend to have particularly negative outcomes for boys.
There are also other phenomena that are more subtle, related to things like marriage patterns. Relating this back to personal experience, I remember when working on these issues and thinking about our decision to move from Harvard to Stanford. At the time, we actually were expecting our first child, a daughter. And I noticed in our data that, for kids in affluent families in the Bay Area, daughters tend to have very low household earnings. And I found that kind of curious and we spent some time trying to dig into why what was, partly given my personal interest in the issue.
COWEN: So, your own moving decision was influenced by this research.
CHETTY: [laughs] In some ways.
CHETTY: What you find is an interesting explanation, which is, if you look at individual earnings rather than household income, girls growing up here in the Bay Area do extremely well. However, when you look at household income, they don’t do so well, and that’s because they’re much, much less likely to be married than if they grew up somewhere else.
CHETTY: So if you’re in your mid-30s, only something like a quarter or less of girls growing up in the Bay Area are married, and we show in our paper that every extra year you spend growing up in the Bay Area, you’re less likely to get married. I remember telling my wife, “I don’t think we need to worry. Our daughter will be fine in terms of earnings. It’s just that she might not be married if we move to California.”
COWEN: So, you’ve lowered your expectations for grandchildren?
CHETTY: Yes. [laughs]
I’d be interested to know what Mrs. Chetty thinks about not having grandchildren.