RSS
More from the New York Times’ auto-generator of Amazing Chetty Facts about income mobility:
Manhattan is extremely bad for children in families in the top 1%. It is among the worst counties in the U.S.
Which do you think is more likely?
– That the Masters of the Universe are collective fools about their scions’ financial fates?
– Or that Professor Raj Chetty of Harvard doesn’t fully grasp how regression toward the mean affects his giant database?
It’s a statistical inevitability that, on average, the children of Manhattan’s One Percent — the If I Can Make It There, I Can Make It Anywhere elite — are going to regress somewhat toward a lower mean. Some of them are into organic farming in Vermont, some are ski bums in Boulder, some are writing screenplays in Silver Lake. Heck, some of them are corporate lawyers in Chicago or doctors in Denver or bond traders in New York.
They’re just not going to be quite as high income on average, relatively speaking, as the single richest group in Chetty’s database: the top 1% earners in Manhattan. It doesn’t mean these gilded youths didn’t get a lot of privileges, it’s just how the world has to work, as Francis Galton pointed out.
Conversely, while the top of the One Percent (nationally speaking) in Manhattan goes more into the stratosphere than the One Percent anywhere else, the bottom of the national one percent is probably easier to get into in Manhattan than anywhere else because salaries are higher there than elsewhere because the cost of living is higher.
Mr. Chetty should move someplace where nearly everyone is upwardly mobile. He should move to India.
Oh Steve… the dopey 1%ers stay in Manhattan. The smart ones move to Boone County WV and then their kids will get the ultra double super advantage from moving away.
There’s an IQ bell curve even among the 1%. And the ones on the far right of the curve are moving to the Boone-docks.
Wow. It’s hard to satirize this stuff. There just aren’t that many people smart enough to come up with things that are this stupid.
Maybe if Chetty were at MIT he would figure it out. He’s only a Harvard man.
Steve S. has given the world the gift that keeps on giving: reductio ad absurdum of the liberal Narrative.
There is a dingbat factor with some of the high IQ (off the charts!) foreigners that parachute in here and presume to lecture Americans. Same with that Indian Surgeon General. Did he get the job? I don’t care because Bill Clinton and Obama have turned that position into a joke.
I bet Del Mar is bad for the 1% too. Just think about the sons of the pioneering anesthesiologist, mayor, and author, Dr. Shapiro: a drummer (deceased) and a director of commercials.
If my parents had been rich, I might have considered journalism.
Maybe the financial results for the children of the top 1% are not as great as the top 1% themselves (on average). But I bet that the children of the top 1% are much more attractive on average than the parents whose financial acumen it was that got them into the top 1%.
I think when your parents are rich, you’re less motivated.
If my parents had been rich, I might have considered journalism.
Worst meme ever.
https://imgflip.com/i/l2nt0
So the key for a stable social structure among the 1% is the trust. Look at the Kennedys. Complete regression to the mean in IQ and accomplishment but no worries at all in income and standard of living. In fact their investment advisors have insured an increasing standard of living for them. I am sure the Obamas, like the Clintons (and the Kennedy family before them) are looking into this closely as they are so concerned about the slipping middle class and the loss of the American Dream.
Comparing Obama to the Kennedys is a bit rude: Obama hasn’t drowned a girl.
Although Ted was allowed to turn into a fat and rather degenerate old man, the singular Kennedy "Family" was executed in 1963 and 1968, thus being among the very few politicians to actually pay the ultimate price that politicos so frequently demand from their constituents.
Presidents and high level officials in our country subsequent to that, after committing a series of disastrous deeds and causing much misery, just go on to more terms in office, speaking engagements, and getting their equally evil sons, brothers, and wives elected to the highest positions.
The sad thing is, Raj will be rewarded for this stupendous display of [induced] stupidity.
It’s fun to make fun, but also a little unfair. On the other hand, it’s Chetty’s own fault for setting up this auto-text generator that spits out an answer based on numbers regardless of context. I assume he’s familiar with the concept of regression to the mean, and if you asked him he’d probably admit his model doesn’t work very well in places like Manhattan.
I assumed it wouldn’t work in Fairfield County Connecticut (where all the hedge funds are), but I was wrong. Hmmmmmmmmmmmm.
From the FAQ section of Raj’s website:
Some comments:
(1) There are two straightforward results he admits here. First, race is an important factor in explaining upward mobility. Second, living among blacks is a bad thing.
(2) Raj quickly scoots past the finding that living among blacks is bad to blame “root causes” such as segregation and low public investment.
(3) The analysis of families with two children is a clever way to try to separate out race as an explanatory variable. It lets Raj play up the idea that the lack of black upward mobility is not completely due to blacks being black. But, he is left with the embarrassing result that living among blacks is bad. Hence his quick scoot to blaming segregation and low public investment rather than the more obvious explanations of why living among blacks is bad thing.
Raj’s scoot to “root causes” is kind of funny when you think about what he is carefully trying to avoid noticing and saying. Raj says “Cities with large African-American populations tend to be more segregated and have lower levels of investment in public schools and other public goods. This may explain why these cities generate poorer outcomes for children of all races. ” Nope, nothing else here to notice about what living among blacks is like.
(4) I don’t find it hard to believe that living among blacks is bad for upward mobility. I don’t find it hard to believe that poor people growing up in Detroit do worse than poor people growing up in Salt Lake City.
(5) Since Raj admits blacks have lower upward mobility, any serious analysis would have to include race as an explanatory variable before concluding that anything else is “causal” with regard to upward mobility. This is just standard procedure for serious economic work. I haven’t checked if he controls for this with an explanatory variable such as “probability of being black”. If he did, we could say useful things about how much of the lack of upward mobility is due to factors innate to blacks.
(6) Similarly, since Raj admits that living among blacks is bad, before concluding this is due to segregation and low public investment, he should control for other things that correlate with living among blacks, such as higher crime, lower test scores, etc.
(7) I haven’t read Raj’s papers, so perhaps he courageously confronts the alternate explanations that Steve tends to think of, and perhaps he miraculously finds that his “root causes” dominate the Steve-factors in a rigorous statistical analysis. He doesn’t mention such an analysis in this FAQ, though.
(8) A simple test of the explanatory powers of Steve-factors versus Raj’s “root causes” would be a simple regression of Raj’s upward mobility measure on (a) size of the black population in percentage terms, (b) “segregation”, and (c) “public investment”. Which dominates? Raj’s map of “intergenerational mobility” makes it look like upward mobility is pretty strongly correlated with the size of the black population. http://www.voxeu.org/article/where-land-opportunity-intergenerational-mobility-us
Factsareimportant -- since he is comparing peoples' adult incomes to the adult incomes of their siblings, in almost every case he is comparing blacks to blacks and whites to whites. So pure race effects on income should be more or less washed out.
In section VI.A of that paper, he saysHe used anonymized income tax return data that doesn't include each taxpayer's race. He suggests two possible explanations: (a) there's something about regions with lots of blacks that drags down mobility for everyone and (b) blacks have lower expected income (conditional on parents' income) than whites. Steve's "regression to the mean" idea is, of course, (b). He then looks at (b) and rejects it.
Pause here. This is the big fork in the road. His choice between (a) vs (b) profoundly colors every conclusion he draws in this paper. I think in his final models, he simply does not include %black as a covariate ("regressor"). I think his rejection of (b) is his justification for this politically correct, cowardly move. But every other factor he ultimately includes is ridiculously highly correlated with percentage black. He tosses the elephant in the room (% black) and throws in 5 more politically correct proxies (like segregation and income inequality). I wonder how many of the other covariates (e.g., segregation, 2-parent households) would be greatly attenuated, possibly rendered statistically insignificant, in the final model, if he included %black.
This is probably why the NYT model predicts so many idiotic results: where income inequality and segregation are low (e.g., probably Boone county), the model predicts there are lots of whites and so greater income mobility, and where income inequality and segregation are high (e.g., probably Marin county), the model predicts lots of blacks and so low mobility.
So why's he reject (b)? He does a subsample analysis including only zip codes with 80%+ white. He looks at the resulting map and says, "Hey, this looks similar to the unrestricted map". And they say those economists don't know how to wave their hands.
And beyond this hocus pocus, why does he get to ignore %black even if there are no innate racial differences, as in (a)? Couldn't the presence of large number of blacks, even in nearby zip codes, have profound effects on white income mobility? No, no, don't go there.
No, but he’s droned some.
(1) There are two straightforward results he admits here. First, race is an important factor in explaining upward mobility. Second, living among blacks is a bad thing.
(2) Raj quickly scoots past the finding that living among blacks is bad to blame "root causes" such as segregation and low public investment.
(3) The analysis of families with two children is a clever way to try to separate out race as an explanatory variable. It lets Raj play up the idea that the lack of black upward mobility is not completely due to blacks being black. But, he is left with the embarrassing result that living among blacks is bad. Hence his quick scoot to blaming segregation and low public investment rather than the more obvious explanations of why living among blacks is bad thing.
Raj's scoot to "root causes" is kind of funny when you think about what he is carefully trying to avoid noticing and saying. Raj says "Cities with large African-American populations tend to be more segregated and have lower levels of investment in public schools and other public goods. This may explain why these cities generate poorer outcomes for children of all races. " Nope, nothing else here to notice about what living among blacks is like.
(4) I don't find it hard to believe that living among blacks is bad for upward mobility. I don't find it hard to believe that poor people growing up in Detroit do worse than poor people growing up in Salt Lake City.
(5) Since Raj admits blacks have lower upward mobility, any serious analysis would have to include race as an explanatory variable before concluding that anything else is "causal" with regard to upward mobility. This is just standard procedure for serious economic work. I haven't checked if he controls for this with an explanatory variable such as "probability of being black". If he did, we could say useful things about how much of the lack of upward mobility is due to factors innate to blacks.
(6) Similarly, since Raj admits that living among blacks is bad, before concluding this is due to segregation and low public investment, he should control for other things that correlate with living among blacks, such as higher crime, lower test scores, etc.
(7) I haven't read Raj's papers, so perhaps he courageously confronts the alternate explanations that Steve tends to think of, and perhaps he miraculously finds that his "root causes" dominate the Steve-factors in a rigorous statistical analysis. He doesn't mention such an analysis in this FAQ, though.
(8) A simple test of the explanatory powers of Steve-factors versus Raj's "root causes" would be a simple regression of Raj's upward mobility measure on (a) size of the black population in percentage terms, (b) "segregation", and (c) "public investment". Which dominates? Raj's map of "intergenerational mobility" makes it look like upward mobility is pretty strongly correlated with the size of the black population. http://www.voxeu.org/article/where-land-opportunity-intergenerational-mobility-us
It’s just regression toward different means.
Some black communities enjoy a lot of public investment, such as Baltimore’s and Milwaukee’s, which are in the Top 5 in spending per public school student.
(1) There are two straightforward results he admits here. First, race is an important factor in explaining upward mobility. Second, living among blacks is a bad thing.
(2) Raj quickly scoots past the finding that living among blacks is bad to blame "root causes" such as segregation and low public investment.
(3) The analysis of families with two children is a clever way to try to separate out race as an explanatory variable. It lets Raj play up the idea that the lack of black upward mobility is not completely due to blacks being black. But, he is left with the embarrassing result that living among blacks is bad. Hence his quick scoot to blaming segregation and low public investment rather than the more obvious explanations of why living among blacks is bad thing.
Raj's scoot to "root causes" is kind of funny when you think about what he is carefully trying to avoid noticing and saying. Raj says "Cities with large African-American populations tend to be more segregated and have lower levels of investment in public schools and other public goods. This may explain why these cities generate poorer outcomes for children of all races. " Nope, nothing else here to notice about what living among blacks is like.
(4) I don't find it hard to believe that living among blacks is bad for upward mobility. I don't find it hard to believe that poor people growing up in Detroit do worse than poor people growing up in Salt Lake City.
(5) Since Raj admits blacks have lower upward mobility, any serious analysis would have to include race as an explanatory variable before concluding that anything else is "causal" with regard to upward mobility. This is just standard procedure for serious economic work. I haven't checked if he controls for this with an explanatory variable such as "probability of being black". If he did, we could say useful things about how much of the lack of upward mobility is due to factors innate to blacks.
(6) Similarly, since Raj admits that living among blacks is bad, before concluding this is due to segregation and low public investment, he should control for other things that correlate with living among blacks, such as higher crime, lower test scores, etc.
(7) I haven't read Raj's papers, so perhaps he courageously confronts the alternate explanations that Steve tends to think of, and perhaps he miraculously finds that his "root causes" dominate the Steve-factors in a rigorous statistical analysis. He doesn't mention such an analysis in this FAQ, though.
(8) A simple test of the explanatory powers of Steve-factors versus Raj's "root causes" would be a simple regression of Raj's upward mobility measure on (a) size of the black population in percentage terms, (b) "segregation", and (c) "public investment". Which dominates? Raj's map of "intergenerational mobility" makes it look like upward mobility is pretty strongly correlated with the size of the black population. http://www.voxeu.org/article/where-land-opportunity-intergenerational-mobility-us
Since Raj admits blacks have lower upward mobility, any serious analysis would have to include race as an explanatory variable before concluding that anything else is “causal” with regard to upward mobility. This is just standard procedure for serious economic work. I haven’t checked if he controls for this with an explanatory variable such as “probability of being black”. If he did, we could say useful things about how much of the lack of upward mobility is due to factors innate to blacks.
Factsareimportant — since he is comparing peoples’ adult incomes to the adult incomes of their siblings, in almost every case he is comparing blacks to blacks and whites to whites. So pure race effects on income should be more or less washed out.
Raj finds:So, Raj has found that living among blacks is a bad thing, and it is a bad thing for both blacks and whites.
Update.
To be fair to Raj, in his 2014 paper, he includes the explanatory variable “fraction black” and finds that it has no additional explanatory power against explanatory variables such as high school dropout rates, single motherhood, and social capital. He writes:
So, it isn’t being black per se that reduce mobility, it is the various pathologies that are more often associated with blacks that reduces mobility.
One interpretation of this is that it rules out discrimination. Blacks that don’t exhibit pathologies such as single motherhood and dropping out have the same mobility as non-blacks.
Is this 1% wealth or 1% earnings?
Also, if you are really that rich you don’t actually live in Manhattan — that is why you have second (and third, and fourth) homes.
Also curious on the role of multi-racial declarations on this.
Okay, so five factors associated with being black are ever so slightly more explanatory than one factor (being black).
Which ways does the arrow of causality point? Does being a single mother make you black?
People have spent a lot of effort convincing themselves that race is just skin deep rather than being about your nature and (usually) nurture.
Raj finds that being black isn't what hurts blacks. Rather it is the bad things that blacks do that hurts blacks.
As I mentioned above, this implies that discrimination is not hurting blacks. After you control for the harmful things that blacks do, there is no additional harm from just being black. In Raj's regression analysis, he finds a positive coefficient on "fraction black" after controlling for dropout rates etc. Therefore, he finds that blacks have better mobility than non-blacks with the same pathologies. This is evidence of discrimination in favor of blacks.
And I'm frankly not sure what any of this has to do with causal analysis. Steve's line is great: "Does being a single mother make you black?".
Also: multicollinearity.
A quick factor analysis would help to reveal the nature of this relationship.
Damn.
But from the perspective of statistics, it's even worse than that.
Presume that you've six potential factors, five being relatively "independent," and the sixth being some rough linear combination of the other five (it's unlikely that it is a perfect combination for various practical as well as mathematical reasons). That the five together are "slightly better" at explaining the variance means that the single factor is almost surely the better choice. For one, there is the issue that using five variables comes at the cost of four additional degrees of freedom. For two, because of the linear algebra, the fewer items you must assume are 'related' to the dependent variable, the more likely that your model is going to be over-specific to this population, and thus less transferable (generalisable) to the other populations.
Also, if you are really that rich you don't actually live in Manhattan -- that is why you have second (and third, and fourth) homes.
Also curious on the role of multi-racial declarations on this.
Chetty got his hands on IRS data so it’s income, right? Only colleges know your wealth.
Maybe next he’ll get his hands on FAFSA, although the really proctological wealth data is in that other financial aid application that private colleges demand, CSS(?).
Man, and I’m glad to be done filling out those forms.
(1) There are two straightforward results he admits here. First, race is an important factor in explaining upward mobility. Second, living among blacks is a bad thing.
(2) Raj quickly scoots past the finding that living among blacks is bad to blame "root causes" such as segregation and low public investment.
(3) The analysis of families with two children is a clever way to try to separate out race as an explanatory variable. It lets Raj play up the idea that the lack of black upward mobility is not completely due to blacks being black. But, he is left with the embarrassing result that living among blacks is bad. Hence his quick scoot to blaming segregation and low public investment rather than the more obvious explanations of why living among blacks is bad thing.
Raj's scoot to "root causes" is kind of funny when you think about what he is carefully trying to avoid noticing and saying. Raj says "Cities with large African-American populations tend to be more segregated and have lower levels of investment in public schools and other public goods. This may explain why these cities generate poorer outcomes for children of all races. " Nope, nothing else here to notice about what living among blacks is like.
(4) I don't find it hard to believe that living among blacks is bad for upward mobility. I don't find it hard to believe that poor people growing up in Detroit do worse than poor people growing up in Salt Lake City.
(5) Since Raj admits blacks have lower upward mobility, any serious analysis would have to include race as an explanatory variable before concluding that anything else is "causal" with regard to upward mobility. This is just standard procedure for serious economic work. I haven't checked if he controls for this with an explanatory variable such as "probability of being black". If he did, we could say useful things about how much of the lack of upward mobility is due to factors innate to blacks.
(6) Similarly, since Raj admits that living among blacks is bad, before concluding this is due to segregation and low public investment, he should control for other things that correlate with living among blacks, such as higher crime, lower test scores, etc.
(7) I haven't read Raj's papers, so perhaps he courageously confronts the alternate explanations that Steve tends to think of, and perhaps he miraculously finds that his "root causes" dominate the Steve-factors in a rigorous statistical analysis. He doesn't mention such an analysis in this FAQ, though.
(8) A simple test of the explanatory powers of Steve-factors versus Raj's "root causes" would be a simple regression of Raj's upward mobility measure on (a) size of the black population in percentage terms, (b) "segregation", and (c) "public investment". Which dominates? Raj's map of "intergenerational mobility" makes it look like upward mobility is pretty strongly correlated with the size of the black population. http://www.voxeu.org/article/where-land-opportunity-intergenerational-mobility-us
Chetty’s papers are online. The already published Quarterly Journal of Economics paper is relatively readable. I quickly looked at it. It’s probably not the paper used to generate the NYT’s Amazing Chetty Fact Generator, but it’s along the same lines.
In section VI.A of that paper, he says
He used anonymized income tax return data that doesn’t include each taxpayer’s race. He suggests two possible explanations: (a) there’s something about regions with lots of blacks that drags down mobility for everyone and (b) blacks have lower expected income (conditional on parents’ income) than whites. Steve’s “regression to the mean” idea is, of course, (b). He then looks at (b) and rejects it.
Pause here. This is the big fork in the road. His choice between (a) vs (b) profoundly colors every conclusion he draws in this paper. I think in his final models, he simply does not include %black as a covariate (“regressor”). I think his rejection of (b) is his justification for this politically correct, cowardly move. But every other factor he ultimately includes is ridiculously highly correlated with percentage black. He tosses the elephant in the room (% black) and throws in 5 more politically correct proxies (like segregation and income inequality). I wonder how many of the other covariates (e.g., segregation, 2-parent households) would be greatly attenuated, possibly rendered statistically insignificant, in the final model, if he included %black.
This is probably why the NYT model predicts so many idiotic results: where income inequality and segregation are low (e.g., probably Boone county), the model predicts there are lots of whites and so greater income mobility, and where income inequality and segregation are high (e.g., probably Marin county), the model predicts lots of blacks and so low mobility.
So why’s he reject (b)? He does a subsample analysis including only zip codes with 80%+ white. He looks at the resulting map and says, “Hey, this looks similar to the unrestricted map”. And they say those economists don’t know how to wave their hands.
And beyond this hocus pocus, why does he get to ignore %black even if there are no innate racial differences, as in (a)? Couldn’t the presence of large number of blacks, even in nearby zip codes, have profound effects on white income mobility? No, no, don’t go there.
Your response is to my earliest posting, which is pretty ignorant. I also looked at his papers and found the section you found. In a later post, I conclude that his rejection of regression-to-the-mean is dishonest. Steve made the same point in an earlier post.
Factsareimportant -- since he is comparing peoples' adult incomes to the adult incomes of their siblings, in almost every case he is comparing blacks to blacks and whites to whites. So pure race effects on income should be more or less washed out.
Agreed. As I mentioned above, Raj has done a very clever analysis here.
Raj finds:
So, Raj has found that living among blacks is a bad thing, and it is a bad thing for both blacks and whites.
Agreed.
Raj finds that being black isn’t what hurts blacks. Rather it is the bad things that blacks do that hurts blacks.
As I mentioned above, this implies that discrimination is not hurting blacks. After you control for the harmful things that blacks do, there is no additional harm from just being black. In Raj’s regression analysis, he finds a positive coefficient on “fraction black” after controlling for dropout rates etc. Therefore, he finds that blacks have better mobility than non-blacks with the same pathologies. This is evidence of discrimination in favor of blacks.
It might be even worse than Steve describes. My interpretation of is that they did some type of “model selection” test comparing the 5-factor model to the 5-factor model + race. So there are really 6 factors under consideration and the order of model selection is really all based on “analyst’s judgment”. Why not compare six 1-factor models separately? How about the 5-factor models composed from these 6 factors?
And I’m frankly not sure what any of this has to do with causal analysis. Steve’s line is great: “Does being a single mother make you black?”.
Also: multicollinearity.
Update:
To be fair to Raj, he does confront the fact that his map of mobility is basically a map of black population percentage. His response, though, appears to be less than honest and actually seems to prove the opposite of what he wants.
His response, though, uses slight of hand.
He does a second analysis that looks at areas that are more than 80 percent white and finds a similar pattern. See Figure VI and Figure IX in his paper (look at absolute upward mobility). But, the results for the white sub-sample are much weaker than for the total sample. In the total sample, the worst mobility category is 26.0-37.0 (the deep red color in that chart), while in the white sub-sample, the worst category is 38-40 (misleadingly also shown as deep red). If he had used the same colors in each chart to indicate the same ranges of mobility, it would be more obvious that there is a big difference. There would be no deep red areas in the “white” sub sample chart.
Further, the “white” sub-sample, is only 91% white, so there are still some non-whites, so that could explain why we still see some correlation. (To get technical, the real question is what fraction of the people at the 25th percentile are black. So, even in the mostly white subsample, there is still a significant fraction of the 25th percentile people who are black.)
Further, even if he showed that not all of the pattern is driven by the size of the black population, it would not show that the size of the black population is irrelevant. He is being cute when he pretends that any evidence of the pattern in the white population means that none of the pattern is due to blacks. Indeed, the fact that the pattern is weaker among the “white” sub sample pretty much proves that much of the pattern is due to variation in the size of the black population.
I hope this is helpful.
Right, in other words, it’s easier to make 75k per year….as a janitor in NY. But how far will that 75k take you? Not very far. Probably can’t afford Yankees season tickets on that barely scraping by annual salary.
But what you’ve said regarding regression to the mean, wouldn’t it imply that if you’re the offspring of the top 1% in NY that you’d be at a higher regression when you fell than if you were the offspring of the extremely poor?
In Charles Murray’s 2011 book on White America, he made a point that those who you thought were among the dumbest at your elite school actually had a higher IQ than the average person in a so so ordinary school. In other words, Ralph Lauren’s children: One of them married a Bush; another one is successfully running her own business (candy industry) and the third is “only” a producer at the head a minor league studio. IF that’s regression to the mean then that’s a pretty high loft from which his offspring have fallen. Who wouldn’t want to fall and end up short….by running a successful candy company?
So if you’re going to fall and regress toward your group’s mean, it’s better to fall from a higher level, namely, from the top 1%.
The offspring of the top 1% “only” end up earning 1-2 million per yr so that they’re in the low 1%, but still in the top 1%.
The offspring of the best and brightest to come out of Boone County, WV, is going to fall back to making 40k per yr. or just average.
The offspring of the best and brightest who came from WV’s poorest county, will regress back to….abject poverty.
Nope. If one’s going to regress back, better its from the top 1% so you have a higher plane from which to fall. At least you’re guaranteed the middle class if not your own candy company.
Oldest trick in the linear regression book. Start with a collection of variables that, together, can almost perfectly predict race. And then show that adding race to the mix doesn’t give you a statistically significant F statistic.
There’s a socioligist at Berkeley who used that trick once to “prove” that being black didn’t confer any advantage in getting admitted. But one of his other variables was an indicator for having participated in a special “you can go to college” program for black kids.
There’s a political motivation to what he’s doing: he wants to establish that segregation, ie living in an all-white milieu, is bad for whites. He needs that result to parry the obvious question of, If it’s harmful to move *into* a black area, why isn’t it then harmful to have blacks move into your nice white area?
In fact, if the model has a number of factors that are associated both with being black (a potential factor) and upward mobility (the dependent variable), introducing all into the same analysis runs the risk of multi-colinearity.
A quick factor analysis would help to reveal the nature of this relationship.
Doesn’t this data also cut against Pikkety’s argument about the inevitability of the richest families growing ever richer because r > g? Here we have the top 1%, the Masters of the Universe, living in the locus of financial wealth on Manhattan Island, with a net worth likely in the tens of millions, and their children seem to be downwardly mobile.
Hmmm.
Regarding hereditary wealth, a big variable is going to be the number of children who split the fortune. If all of Sam Walton's wealth went to one heir, that heir would be the richer than Gates.
I just noticed that most of what I said in my comments to this thread was said by you in a post a few days ago, (“Helping Hillary’s Idea Man Raj Chetty Understand America”). … And said better.
Damn.
Steve,
Raj will be giving a talk near me in June.
What one question should I ask him?
“some are ski bums in Boulder”
When they aren’t skiing, you’ll find them in their natural habitat; an outdoor table at a bistro along the Pearl Street mall, drinking bottles of wine over a late lunch on a weekday afternoon. They are far enough out of the orbit of their family back East to avoid damaging their parents standing with their shiftless, trustafarian ways.
Steve, this is just spitballing, but do you suppose that Chetty knows exactly what he’s doing – not legitimate research, per se, but providing a ‘study’ that allows people like Hillary to say “Studies have shown that my kids don’t have as many opportunities as your kids, West Virginia!”
Although no one with half a brain would believe that, it’s technically true, according to Chetty’s research… Rich people certainly like to provide themselves with cover.
Or is that too conspiratorial to even consider?
Then later it turns out my suspicion was correct.
In section VI.A of that paper, he saysHe used anonymized income tax return data that doesn't include each taxpayer's race. He suggests two possible explanations: (a) there's something about regions with lots of blacks that drags down mobility for everyone and (b) blacks have lower expected income (conditional on parents' income) than whites. Steve's "regression to the mean" idea is, of course, (b). He then looks at (b) and rejects it.
Pause here. This is the big fork in the road. His choice between (a) vs (b) profoundly colors every conclusion he draws in this paper. I think in his final models, he simply does not include %black as a covariate ("regressor"). I think his rejection of (b) is his justification for this politically correct, cowardly move. But every other factor he ultimately includes is ridiculously highly correlated with percentage black. He tosses the elephant in the room (% black) and throws in 5 more politically correct proxies (like segregation and income inequality). I wonder how many of the other covariates (e.g., segregation, 2-parent households) would be greatly attenuated, possibly rendered statistically insignificant, in the final model, if he included %black.
This is probably why the NYT model predicts so many idiotic results: where income inequality and segregation are low (e.g., probably Boone county), the model predicts there are lots of whites and so greater income mobility, and where income inequality and segregation are high (e.g., probably Marin county), the model predicts lots of blacks and so low mobility.
So why's he reject (b)? He does a subsample analysis including only zip codes with 80%+ white. He looks at the resulting map and says, "Hey, this looks similar to the unrestricted map". And they say those economists don't know how to wave their hands.
And beyond this hocus pocus, why does he get to ignore %black even if there are no innate racial differences, as in (a)? Couldn't the presence of large number of blacks, even in nearby zip codes, have profound effects on white income mobility? No, no, don't go there.
Agree completely.
Your response is to my earliest posting, which is pretty ignorant. I also looked at his papers and found the section you found. In a later post, I conclude that his rejection of regression-to-the-mean is dishonest. Steve made the same point in an earlier post.
Chetty and Gladwell should team up and co- author an instant best-seller: Schizonomics.
Because nothing makes sense anymore in economics. Not even numbers.
Perhaps you’re right about his agenda. I’ll note that he measures segregation by comparing the “lumpiness” of the geospatial distribution of racial groups within a “commute zone”. Thus, if all the census tracts within a CZ have the same racial composition, segregation is minimized and the Utopian Limit is achieved. Even if, say, it’s 99.9% white and 0.1% black throughout all census tracts. But…. he’s not going to point that out. That’s not in the agenda.
Although no one with half a brain would believe that, it's technically true, according to Chetty's research... Rich people certainly like to provide themselves with cover.
Or is that too conspiratorial to even consider?
You know, every time I have a suspicion like that, I think to myself I am paranoid.
Then later it turns out my suspicion was correct.
Hmmm.
Downwardly mobile in income or wealth?
Regarding hereditary wealth, a big variable is going to be the number of children who split the fortune. If all of Sam Walton’s wealth went to one heir, that heir would be the richer than Gates.
Presumably he's grading on a national income scale, so top 1% of income would be about $400K. That's not a ton of money in Manhattan. But still, only $5K?
When they aren't skiing, you'll find them in their natural habitat; an outdoor table at a bistro along the Pearl Street mall, drinking bottles of wine over a late lunch on a weekday afternoon. They are far enough out of the orbit of their family back East to avoid damaging their parents standing with their shiftless, trustafarian ways.
Saw Boulder for the first time in 25 years last summer. Pretty much dead on, except the trustafarians aren’t as grungy anymore. They seem to have dropped any pretense of “Salt of the Earth Hip”. Before, there was still a lot of dirt and lice around the nails.
One man, Edward Kennedy, was accused of that, not the “Family.” And quite frankly, if I actually cared about EMK, I’d cite the presense of Kennedy-hater E. Howard Hunt on CIA patrole in the area of the fated weekend. But that’s a story for another blog called Zombie Trolls from the Grassy Knoll.
Although Ted was allowed to turn into a fat and rather degenerate old man, the singular Kennedy “Family” was executed in 1963 and 1968, thus being among the very few politicians to actually pay the ultimate price that politicos so frequently demand from their constituents.
Presidents and high level officials in our country subsequent to that, after committing a series of disastrous deeds and causing much misery, just go on to more terms in office, speaking engagements, and getting their equally evil sons, brothers, and wives elected to the highest positions.
--
Sons / Brothers / Wives of ... elected to the highest positions:
1. Woodrow Wilson = o
2. Franklin Roosevelt = 0
3. Harry Truman = 0
4. Lyndon Johnson = 1 (son-in-law, never ran for office while he was alive).
5. Richard Nixon = 0
6. George H.W. Bush = 2
Seems rather de trop to refer to Chuck Robb or any of the Bush trio as 'evil', or perpetrators of 'disastrous deeds'. Then you remember the company you keep on these boards.
--
Might have come as a surprise to Joseph Kennedy, Rose Kennedy, seven of their nine children, and their 28 grandchildren (not to mention Sargent Shriver and Steven Smith) that they were all executed over a five year period.
From Joe Sr. lobotomizing his slutty daughter to the regression to the mean generation's well-documented problems with their prescription meds, they're not a group worthy of much respect.
Although Ted was allowed to turn into a fat and rather degenerate old man, the singular Kennedy "Family" was executed in 1963 and 1968, thus being among the very few politicians to actually pay the ultimate price that politicos so frequently demand from their constituents.
Presidents and high level officials in our country subsequent to that, after committing a series of disastrous deeds and causing much misery, just go on to more terms in office, speaking engagements, and getting their equally evil sons, brothers, and wives elected to the highest positions.
Presidents and high level officials in our country subsequent to that, after committing a series of disastrous deeds and causing much misery, just go on to more terms in office, speaking engagements, and getting their equally evil sons, brothers, and wives elected to the highest positions.
—
Sons / Brothers / Wives of … elected to the highest positions:
1. Woodrow Wilson = o
2. Franklin Roosevelt = 0
3. Harry Truman = 0
4. Lyndon Johnson = 1 (son-in-law, never ran for office while he was alive).
5. Richard Nixon = 0
6. George H.W. Bush = 2
Seems rather de trop to refer to Chuck Robb or any of the Bush trio as ‘evil’, or perpetrators of ‘disastrous deeds’. Then you remember the company you keep on these boards.
Although Ted was allowed to turn into a fat and rather degenerate old man, the singular Kennedy "Family" was executed in 1963 and 1968, thus being among the very few politicians to actually pay the ultimate price that politicos so frequently demand from their constituents.
Presidents and high level officials in our country subsequent to that, after committing a series of disastrous deeds and causing much misery, just go on to more terms in office, speaking engagements, and getting their equally evil sons, brothers, and wives elected to the highest positions.
singular Kennedy “Family” was executed in 1963 and 1968,
—
Might have come as a surprise to Joseph Kennedy, Rose Kennedy, seven of their nine children, and their 28 grandchildren (not to mention Sargent Shriver and Steven Smith) that they were all executed over a five year period.
Regarding hereditary wealth, a big variable is going to be the number of children who split the fortune. If all of Sam Walton's wealth went to one heir, that heir would be the richer than Gates.
I can finally get to the charts (corporate firewall blocking websockets grumble grumble) and the income effects looks shockingly small. They claim a drop in income of $5000 compared to an average county for spending 20 years of childhood in Manhattan, for the children of the top 1% of earners, in the financial capital of the world.
Presumably he’s grading on a national income scale, so top 1% of income would be about $400K. That’s not a ton of money in Manhattan. But still, only $5K?
Although Ted was allowed to turn into a fat and rather degenerate old man, the singular Kennedy "Family" was executed in 1963 and 1968, thus being among the very few politicians to actually pay the ultimate price that politicos so frequently demand from their constituents.
Presidents and high level officials in our country subsequent to that, after committing a series of disastrous deeds and causing much misery, just go on to more terms in office, speaking engagements, and getting their equally evil sons, brothers, and wives elected to the highest positions.
The Kennedy “Family” history is chock full of stories that, while not at the same level as the Chappaquiddick episode, demonstrate an almost genetic lack of judgment.
From Joe Sr. lobotomizing his slutty daughter to the regression to the mean generation’s well-documented problems with their prescription meds, they’re not a group worthy of much respect.
--
She had trouble keeping up in school, may have had schizophreniform episodes, and grew contentious and difficult as she grew older. Not sure why she should be subject to random insults seven decades after the fact.
--
regression to the mean
--
Strange as it may seem to the world's biological determinists, people who grow up in comfortable circumstances are often not very driven even when they suffer no intellectual deficits. See Mark Shriver and Caroline Schlossberg for two examples.
From Joe Sr. lobotomizing his slutty daughter to the regression to the mean generation's well-documented problems with their prescription meds, they're not a group worthy of much respect.
From Joe Sr. lobotomizing his slutty daughter
—
She had trouble keeping up in school, may have had schizophreniform episodes, and grew contentious and difficult as she grew older. Not sure why she should be subject to random insults seven decades after the fact.
—
regression to the mean
—
Strange as it may seem to the world’s biological determinists, people who grow up in comfortable circumstances are often not very driven even when they suffer no intellectual deficits. See Mark Shriver and Caroline Schlossberg for two examples.
We agree on your second point. Why voters want dedicated backbencher dilettantes like Shriver and his cousin, Cong. Joe Da Turd is beyond me, and it makes one pine for the reintroduction of the poll test.
http://standpointmag.co.uk/node/6025/full
Scottish separatists tend to be leftists.
Are leftism and nationalism natural allies? Maybe the right got it all wrong by embracing capitalism, or at least global capitalism.
Think.
Fascism was fusion of leftist economics with nationalism.
National Socialism was fusion of leftist economics and nationalism.
New Deal fused economic leftism with robust American nationalism that crushed the ‘Japs’.
Zionism fused economic leftism with Jewish nationalism.
Ho Chi Minh was fiercely nationalistic and economically leftist.
Castro was a leftist and nationalist.
Arab rulers were nationalist and socialist.
Steve, you’re right of course.
But from the perspective of statistics, it’s even worse than that.
Presume that you’ve six potential factors, five being relatively “independent,” and the sixth being some rough linear combination of the other five (it’s unlikely that it is a perfect combination for various practical as well as mathematical reasons). That the five together are “slightly better” at explaining the variance means that the single factor is almost surely the better choice. For one, there is the issue that using five variables comes at the cost of four additional degrees of freedom. For two, because of the linear algebra, the fewer items you must assume are ‘related’ to the dependent variable, the more likely that your model is going to be over-specific to this population, and thus less transferable (generalisable) to the other populations.
--
Might have come as a surprise to Joseph Kennedy, Rose Kennedy, seven of their nine children, and their 28 grandchildren (not to mention Sargent Shriver and Steven Smith) that they were all executed over a five year period.
Yes Art Deco. I did that deliberately in reply to dearime, who had conflated the whole lot of them. I do not believe in communal guilt or innocence.
Usually its “that poor girl”, as with Mary Jo. When the victim gets called slutty because of her last name, I call bias.
We just know more about them than just about any other Big Pol name you could throw out, and I’ve read a lot about a lot of them. If people don’t read books from a variety of perspectives , read first-hand documents, or genuinely investigative reports, or talk to people who were involved with the goings on of that day (still a lot of them around), where do they get their info? Kennedys? We do know who they are, only too well. I don’t think I would have been too fond of them personally, but there were nice things about them too, like their courage (not seen as much among them today), and their touching patriotism even after 1968, which is more than I can say for the current potus, or, indeed, for most high level U.S. officials and corporate puppet masters in the past 30 years.
All I can say is, JFK is one of the extremely rare politicians/public figures who looks better the closer I look, even starting from very low point. I even began to think some of the sex scandal stuff had to be exaggerated. He just didn’t have the time; no president was as gripped in a vise of enemies as he, regardless of who made them. Really, the job just ain’t worth it. Jury still out on RFK but he seemed to have been sucker punched into some sort of self-awareness before his demise. EMK should have run a pub.
As for their much criticized “liberal” stance, it was the only one for politicians in their position, and from their background, to take in the 1960s. Times and circumstances have changed, but too many act as if the social circumstances that evoked such over-the-top altruism, remain the same. If JFK and/or RFK had survived to later years , they would have been SWPL. There would have been no way out.
no president was as gripped in a vise of enemies as he, regardless of who made them. He did not have any more enemies than any other big time pol and the newspapers treated him with kid gloves. Phillip Graham of the post was an extension of the administration, not its watchdog.
Steve I noticed at least one comment which suggested a somewhat fuzzy understanding of regression to the mean because it didn’t notice that, for example, parents with IQs of about 80 are likely to have children whose average IQ is appreciably higher. But I puzzle over how one determines what population one belongs to for the purpose of calculating the mean toward which one’s measured partly heritable characteristic will regress. Would you care to write something about that conundrum?
BTW do Chetty’s figures show that the top 10 per cent’s, top 20 per cent’s, top 30 per cent’s etc children all have lower incomes on average than their parents? It would be very interesting if they did not!
So, in reality, the President's American grandparents were sort of the black sheep of their own families in terms of academic ambitions and achievements.
So, data on more relatives is better, although there are diminishing returns.
Would blacks moving into a white area be good for blacks at the same time that it was bad for whites? Wasn’t this the original rationale behind segregation? Why is it ethical to demand that one group sacrifice its well-being for another?
http://www.detroit1701.org/SweetHome.htm
All I can say is, JFK is one of the extremely rare politicians/public figures who looks better the closer I look, even starting from very low point.
OK, you’re looking at him with beer goggles.
no president was as gripped in a vise of enemies as he, regardless of who made them.
He did not have any more enemies than any other big time pol and the newspapers treated him with kid gloves. Phillip Graham of the post was an extension of the administration, not its watchdog.
BTW do Chetty's figures show that the top 10 per cent's, top 20 per cent's, top 30 per cent's etc children all have lower incomes on average than their parents? It would be very interesting if they did not!
Consider Obama’s mother Stanley Ann. Her mother never went to college and her father dropped out of Berkeley almost immediately after the War. So her obtaining a Ph.D. would seem anomalous if you looked just at her parents. But her father’s brother Ralph Waldo Emerson Dunham obtained a Berkeley Ph.D. and her mother’s sister earned a Ph.D. and her mother’s younger brother earned a U. of Chicago M.A. and became a professional college librarian.
So, in reality, the President’s American grandparents were sort of the black sheep of their own families in terms of academic ambitions and achievements.
So, data on more relatives is better, although there are diminishing returns.
Somewhat less demanding trade than banking (or furniture sales, when you come right down to it).
Oh yeah, their well being was terribly sacrificed by having a local doctor next door.
http://www.detroit1701.org/SweetHome.htm
So, in reality, the President's American grandparents were sort of the black sheep of their own families in terms of academic ambitions and achievements.
So, data on more relatives is better, although there are diminishing returns.
became a professional college librarian.
Somewhat less demanding trade than banking (or furniture sales, when you come right down to it).
One more example of statistical legerdemain.
http://tinyurl.com/lyvmpes
--
She had trouble keeping up in school, may have had schizophreniform episodes, and grew contentious and difficult as she grew older. Not sure why she should be subject to random insults seven decades after the fact.
--
regression to the mean
--
Strange as it may seem to the world's biological determinists, people who grow up in comfortable circumstances are often not very driven even when they suffer no intellectual deficits. See Mark Shriver and Caroline Schlossberg for two examples.
Tongue was firmly in cheek, Mr. Art Pedant. Rosemary was pretty much a free-spirited carbon copy of her brothers, which just wasn’t acceptable to the execrable Joe Sr. He had his boys’ future to think about, and he demonstrated his familial penchant for treating the fairer sex like so much used toilet paper.
We agree on your second point. Why voters want dedicated backbencher dilettantes like Shriver and his cousin, Cong. Joe Da Turd is beyond me, and it makes one pine for the reintroduction of the poll test.
--
No, she wasn't. She was confined to a convent school and educated by tutorial. She wasn't screwing anything that did not have four legs except in your sicko imagination.
We agree on your second point. Why voters want dedicated backbencher dilettantes like Shriver and his cousin, Cong. Joe Da Turd is beyond me, and it makes one pine for the reintroduction of the poll test.
Rosemary was pretty much a free-spirited carbon copy of her brothers,
—
No, she wasn’t. She was confined to a convent school and educated by tutorial. She wasn’t screwing anything that did not have four legs except in your sicko imagination.
Hmm...
--
No, she wasn't. She was confined to a convent school and educated by tutorial. She wasn't screwing anything that did not have four legs except in your sicko imagination.
Yet per her Wikipedia page, no Kennedy haters they: “…her diaries reveal a woman whose life was filled with outings to the opera, tea dances, dress fittings and other social interests.”
Hmm…
Hmm...
Yeah, I’m sure she was out back fellating men at those tea dances.
“Before her lobotomy, she seemed healthy and attractive. She was never an obstacle, an embarrassment, or anything very negative. It has always been my feeling that her mental condition was borderline, and that the lobotomy her father Joe authorised really messed her up.”–Harvey Rachlin, The Kennedys: A Chronological History 1823-Present
“Joe had two principle concerns about Rosemary. She was not the competition-oriented ideal of Kennedy womanhood, and he thought her sexuality was too intense and untempered by the moral strictures to which his other daughters adhered. Joe destroyed a part of her brain rather than risk what she might become if allowed to follow her own path in life.”–Barbara Gibson and Ted Schwarz, Rose Kennedy and Her Family: The Best and Worst of Their Times and Lives
Got anything to add other than ad hominem idiocies, Artsy?
What ad homs? You’re the one who fantasizes that she was fellating the crew at Buckingham Palace.
I quote researchers and biographers, while you throw out two comments about oral sex at Buckingham Palace, yet I’m the “sicko”?
Uh, OK. Take your meds, kid.