The Geography of Family Differences and Intergenerational Mobility

That is the title of a new paper by Robert Kaestner, Ryan Gallagher, and Joseph Persky.  Here is the abstract:

A recent series of studies by the Equality of Opportunity Project has documented substantial geographical differences in intergenerational income mobility. These spatial differences are important because they suggest that place matters more than previously thought in determining economic well-being. In this paper, we show that family characteristics vary widely across areas and simulations indicate that differences these family characteristics can explain a substantial share of the variation in intergenerational income mobility across places documented by the Equality of the Opportunity Project. Additionally, we show that the characteristics of families that move differ substantially from families that do not move, which raise doubts about the external validity of causal inferences based on the Equality of Opportunity Project’s analysis of movers.

And from the paper:

…we find that differences in the income of adult children associated with mother’s race, age, education, marital status and nativity explain 80 to 120 percent of the difference in intergenerational income mobility between the lowest and next lowest quintiles of absolute mobility in Chetty et al.’s (2014) place-based distribution of intergenerational income mobility.

I am wondering to what extent this is a criticism of Chetty et.al., or simply a disaggregation.  I’m still trying to wrap my mind around what exactly are the differences between place-level characteristics and family- or person-level characteristics.  I don’t take Chetty’s original story about places to concern what kind of molecules are in the dirt, or what is the climate, but rather how people in a particular place interact with each other.  In that sense the result always was about family- or person-level characteristics.  Does the ability of family-level characteristics to pick up these interaction effects mean that place-level effects are not operating?

Anyway, regardless of interpretation this paper does seem to me to make some very real progress toward figuring out what is going on in those mobility studies.

Comments

I’m still trying to wrap my mind around what exactly are the differences between place-level characteristics and family- or person-level characteristics. I don’t take Chetty’s original story about places to concern what kind of molecules are in the dirt, or what is the climate, but rather how people in a particular place interact with each other.

Wasn't Chetty's study also (even mainly) about how variations in local governmental policies impact local mobility?

On a different note, this blog post is literally begging Steve Sailer to comment.

I'm thankful for Steve Sailer for his honest curiosity on this topic.

Chetty's county level study has some prima facie validity in that his pick for the single worst county for raising working class kids in America is Oglala Lakota County, South Dakota and the single best is Sioux County, Iowa. Even though they have similar names, they really are plausibly the best and worst places in America to raise a family.

I joke about magic dirt and tragic dirt, but Sioux County, Iowa has extremely fecund soil and thus the highest price for farmland in the Midwest, reaching $20,000 per acre in 2013. It also has a very conservative culture, with lots of Dutch Reformed churchgoers. And it has a couple of colleges and some biotechnology companies. It's kind of the Santa Clara County of farm counties.

Ogalala Lakota County is the newly adopted name for what Chetty calls Shannon County: i.e., the Pine Ridge Sioux Indian reservation, which has seen no end of sad events since the Wounded Knee massacre in 1890. The teen suicide rate there is horrific.

So I think Chetty really is onto something. Sioux County is, long term, a better place to raise kids than Oglala Lakota County, and Chetty's methodology successfully noted that.

However, there are still big problems with his methodology that mean that much of his findings aren't really driven by long-term differences in local government policy or local culture, but by differences in the people (with the most obvious that different races regress over the generations toward different income means) and by temporary booms and busts. Sioux County, IA is a really fine farm county, but the reason the top of Chetty's list is dominated by farm and natural gas counties in the center of the country is because in 2011-2012, while much of the country was still shaking off the aftershocks of 2008, they were booming due to Chinese demand and fracking breakthroughs. Conversely, Horry County, SC was in the bottom 25 because it is the home of the giant Myrtle Beach golf megaresort, and golf real estate was in a historic depression in 2011-12.

The reason I keep pointing out these problems with Chetty's study is because I think he could fix them if he were encouraged to do so by other economists.

Anyway, here is my big 2015 analysis of Chetty's county analysis in Taki's Magazine:

http://takimag.com/article/moneyball_for_real_estate_steve_sailer/print#axzz4jOhdk7jn

It is interesting to see expertise at work. Steve Sailor owns this topic so much that no one else in their right mind would comment on it. Not even p_a is willing to vomit his usual Wiki-spam all over this thread.

Which is sensible. SS should write a book on this topic. I hope he calls it "Magic Dirt".

However it does show the limits of the credentialism that Western universities rely on. I would expect that more than half the people here have a very expensive and time-consuming Ph.D. But I would also expect that SS has no academic background in this subject at all.

Dirt is not actually magic. "Cleanliness is next to godlines", Americans used to believe. International research proved Brazilian are both the people who take more baths (Colombians get the sexond place) and brush their teeth more often (oftener?).

And do they litter less and clean up litter more than others?

Brazilians litter relatively little and clean more little than almost all peoples but the Japanese.

Steve's Hunger Games analogy is apropos. Kentucky is the 12th or 13th colony known for mining and sending our youngsters to die in senseless wars. Nailed it!

Our leaders failed to leverage our resources (coal) for prosperity. Hindsight is 20/20.

In the second half of the 19th century a large chunk of my father's family emigrated to Iowa. Their explanation was simple: cheap land and less competition. When I visited them in the second half of the 20th century they were still thriving.

Magic dirt indeed. It certainly wasn't brain power: as you'd expect it was the mediocre brothers who emigrated. The clever one stayed behind; he had no need of less competition.

"explain 80 to 120 percent of the difference"
I am not an economist or math expert, but what could explaining more than 100% of the difference mean?

I guess it could mean that the difference is actually in the opposite direction of what it appears to be?

It accounts for the whole difference and then some.

If factor A explains 120% of the difference, then the implication is that there is a factor B in the evaluation data set that accounts for -20% of the difference, is correlated with factor A, but is not present in the training set. This explanation assumes a certain sophistication of the social scientist (separate data into training and evaluation data sets), and extremely noisy data.

I do not think it is, but climate could conceivably be a factor.

Yes try living in Equador for 6 months and move to Finland immediately afterwards ;)

To Tyler's question "I am wondering to what extent this is a criticism of Chetty et.al., or simply a disaggregation." These quotes from the conclusion seem pretty critical:

1) We find that much of the differences documented by Chetty et al. (2014) are arguably not place differences at all, but people differences. Indeed, a very limited set of people differences explain most of the place differences in intergenerational income mobility. Specifically, we show that earnings predicted from a relatively few characteristics of low-income parental households generates simulated incomes for adult children that account for 40% to 100% of the interquintile differences reported in Chetty et al. (2014). A large portion of the spatial pattern of upward mobility can be generated without reference to space. It seems reasonable to conclude that differences between places in intergenerational mobility would be even further reduced, perhaps to zero, with the addition of more family characteristics. We also show that low-income movers are a very different group than low-income non-movers, which raises a question about the external validity of the more compelling causal estimates in Chetty and Hendren (2016a, 2016b).

2) The arguably weak premise of the Chetty et al. (2014) study combined with the substantial evidence of significant differences in family characteristics between counties and between movers and non-movers that we presented raises questions about the usefulness and interpretation of the evidence of the research of Chetty and colleagues

Comments for this post are closed