Thanks. I've been thinking about a mechanism behind what you observe.
Looking at the map graphic on p. 9 of the article, it appears that there were far fewer free blacks in the cotton-growing slave states that had opened up in the 19th Century than in the older states, slave or free.
My guess would be that free blacks tended to accumulate in a place generation by generation, due to masters freeing their children, manumissions by humanitarians, and self buy-outs.
But the Cotton Belt was populated by a large scale migration as the Indians were pushed out in the 19th century. This migration from the eastern seaboard west appears to have served as a selection event, with free blacks being left behind. The authors argue that there wasn't selection of slaves who were forced to migrate by their masters' moving west to plant cotton, but there was clearly selection on the black population as a whole, with very few free blacks moving into the Cotton Belt.
Further, there was probably selection among urban slaves being left behind, with relatively fewer city-dwelling masters than plantation-masters, which likely made the Cotton Belt slave population more rural and less urbane than the Eastern Seaboard slave population.
This probably had some effect on the post Civil War trajectory of Jim Crow, with whites in states like Mississippi with black majorities and virtually no bourgeois black pre-Civil War freedmen to act as intermediaries, being especially extremist about rigging the laws to keep blacks down.
The more into cotton you were, the higher a % of the population was black. Including in some cases outright majorities that could enforce their will in democratic elections. Hence the local whites are going to care more about suppresion.
Unfortunately, the will of black voters in the reconstruction south was a lot like the will of black voters in Baltimore, DC, or Detroit today. A corrupt incompetent mess. Rather than a "corrupt bargain" the most likely explanation for why the North stopped trying to enforce reconstruction on the south is that it was an unsustainable failure. Kind of like how Congress stripped DC of all its governing powers because the blacks kept electing crackhead Marion Barry.
> If it’s unclear, the massive convergence between Whites whose ancestors lacked physical or human capital in 1870 and 1940 was due virtually entirely to the first jump from 1870, and that is why it exists in the first place. The gap is definitional: the group without any physical or human capital could only move up, and up they moved, so the line was steeper than it was for the group of Blacks it was being compared to, but only for the unfair comparison in the first period. The comparison is unfair precisely because it was defined as such. If we define individuals in the modern census by having zero assets, we will be unsurprised to see a rapid increase in assets in the next census because unlike picking a more natural group in the population, we have concocted one that can only evolve in one direction
Just a minor note for anybody wondering why many of the graphs lack a data point for 1890. It is because the detailed Census archives for 1890 burned up in a fire.
Tremendous piece of work Cremieux. I enjoyed it even if I do not fully understand the methodological issues.
I would like to add one very minor note. When discussing the ante and post bellum southern plantation owners, we should remember that while they lost their entire investment in slaves, which was a substantial portion of their net worth, they kept their land. The entire society was impoverished by tremendous losses of physical and human capital as a direct result of the war. But the hierarchy embedded in landownership was intact.
Good work, but I'm not convinced that newly-freed slaves didn't have "nowhere to go but up" in exactly the same way as the control group of Whites whose ancestors lacked physical or human capital in 1870.
I didn't say they did. I was explaining why the difference between newly freed slaves and Whites whose ancestors lacked physical or human capital was definitional rather than substantial.
I'm not moved by papers with unintuitive results that differ strongly from similar analyses of the same inaccessible datasets due to differences in pre-analytic data handling.
I understand but heritability of income has not scientific value as well as heritability of IQ. Therefore we should be skeptical. More importantly heritability of IQ is same in every percentile but income is not.
Heritability is a very valuable metric to know. It has a lot of utility for building understanding. Inheritance is not heritability. For bequests to influence heritability, they would have to be systematic with respect to genetic relatedness such that, for example, monozygotic twins received more similar bequests than dizygotic ones.
Sorry; you're right. I was thinking of the slavery study, which looks for a "heritability" of income which combines environment and genes. Whereas the study linked to above tries to discriminate environment from genetics.
That's the mystery. When results sharply differ with the same data and similar methods and all we know is that the data was prepared differently in an unknown way, someone has to explain what's going on and no one can because the data is only available to some Scandinavians.
Thanks. I've been thinking about a mechanism behind what you observe.
Looking at the map graphic on p. 9 of the article, it appears that there were far fewer free blacks in the cotton-growing slave states that had opened up in the 19th Century than in the older states, slave or free.
My guess would be that free blacks tended to accumulate in a place generation by generation, due to masters freeing their children, manumissions by humanitarians, and self buy-outs.
But the Cotton Belt was populated by a large scale migration as the Indians were pushed out in the 19th century. This migration from the eastern seaboard west appears to have served as a selection event, with free blacks being left behind. The authors argue that there wasn't selection of slaves who were forced to migrate by their masters' moving west to plant cotton, but there was clearly selection on the black population as a whole, with very few free blacks moving into the Cotton Belt.
Further, there was probably selection among urban slaves being left behind, with relatively fewer city-dwelling masters than plantation-masters, which likely made the Cotton Belt slave population more rural and less urbane than the Eastern Seaboard slave population.
This probably had some effect on the post Civil War trajectory of Jim Crow, with whites in states like Mississippi with black majorities and virtually no bourgeois black pre-Civil War freedmen to act as intermediaries, being especially extremist about rigging the laws to keep blacks down.
The more into cotton you were, the higher a % of the population was black. Including in some cases outright majorities that could enforce their will in democratic elections. Hence the local whites are going to care more about suppresion.
Unfortunately, the will of black voters in the reconstruction south was a lot like the will of black voters in Baltimore, DC, or Detroit today. A corrupt incompetent mess. Rather than a "corrupt bargain" the most likely explanation for why the North stopped trying to enforce reconstruction on the south is that it was an unsustainable failure. Kind of like how Congress stripped DC of all its governing powers because the blacks kept electing crackhead Marion Barry.
> If it’s unclear, the massive convergence between Whites whose ancestors lacked physical or human capital in 1870 and 1940 was due virtually entirely to the first jump from 1870, and that is why it exists in the first place. The gap is definitional: the group without any physical or human capital could only move up, and up they moved, so the line was steeper than it was for the group of Blacks it was being compared to, but only for the unfair comparison in the first period. The comparison is unfair precisely because it was defined as such. If we define individuals in the modern census by having zero assets, we will be unsurprised to see a rapid increase in assets in the next census because unlike picking a more natural group in the population, we have concocted one that can only evolve in one direction
So, regression to the mean?
Thanks. Great stuff.
Just a minor note for anybody wondering why many of the graphs lack a data point for 1890. It is because the detailed Census archives for 1890 burned up in a fire.
Superb work, though sadly it will do little to stop the propagation of false narratives
Tremendous piece of work Cremieux. I enjoyed it even if I do not fully understand the methodological issues.
I would like to add one very minor note. When discussing the ante and post bellum southern plantation owners, we should remember that while they lost their entire investment in slaves, which was a substantial portion of their net worth, they kept their land. The entire society was impoverished by tremendous losses of physical and human capital as a direct result of the war. But the hierarchy embedded in landownership was intact.
Good work, but I'm not convinced that newly-freed slaves didn't have "nowhere to go but up" in exactly the same way as the control group of Whites whose ancestors lacked physical or human capital in 1870.
I didn't say they did. I was explaining why the difference between newly freed slaves and Whites whose ancestors lacked physical or human capital was definitional rather than substantial.
what is your opinion about this paper?
https://www.nber.org/papers/w21409
I'm not moved by papers with unintuitive results that differ strongly from similar analyses of the same inaccessible datasets due to differences in pre-analytic data handling.
I understand but heritability of income has not scientific value as well as heritability of IQ. Therefore we should be skeptical. More importantly heritability of IQ is same in every percentile but income is not.
What about heritability of wealth? I'd think that would be much more heritable than income, because you can literally inherit wealth non-genetically.
Heritability is a very valuable metric to know. It has a lot of utility for building understanding. Inheritance is not heritability. For bequests to influence heritability, they would have to be systematic with respect to genetic relatedness such that, for example, monozygotic twins received more similar bequests than dizygotic ones.
Sorry; you're right. I was thinking of the slavery study, which looks for a "heritability" of income which combines environment and genes. Whereas the study linked to above tries to discriminate environment from genetics.
What are the important differences in pre-analytic data handling?
That's the mystery. When results sharply differ with the same data and similar methods and all we know is that the data was prepared differently in an unknown way, someone has to explain what's going on and no one can because the data is only available to some Scandinavians.
I commented on this on Sean Last's article: https://seanlast.substack.com/p/on-generational-poverty/comments
"The correlations are negligible and statistically insignificant when looking at adults (see https://www.nber.org/system/files/working_papers/w21409/w21409.pdf#page=54)"
I didnt understand. Do you mean table 5?
Rank Biological Parent=0.128
Rank Adoptive Parent= 0.261
Yes, hence the "#page=54" in the URL.
Hey, do you have any place where I can contact you?