Current Reading

This blog is primarily for me to blog my responses to books that I'm reading. Sometimes I blog about other stuff too, though.

Poverty by America by Matthew Desmond.

Word cloud

Word cloud

Wednesday, December 16, 2020

Final Thoughts on The Son Also Rises

 The last few chapters touch on minority groups and then his final thoughts on the bigger picture. He mostly addresses minorities that have been unusually successful, but also touches on Travelers (Gypsies) in the UK. Naturally, these things attract controversy; Clark is careful, and I will try to do likewise. His points are mostly inoffensive, but since he doesn't have surname data to address the full sweep of history as he would like, he gets speculative. But he does make a few good points:

If a minority group persists unusually long in a successful tier of society (e.g. Christian and Jewish minorities in certain Muslim societies), one thing you should ask is whether unsuccessful members of the group disaffiliate. This doesn't necessarily mean that people leave their religion in frustration if they don't succeed as merchants or whatever. But in many Muslim societies there were historically special taxes levied on non-Muslims. An economist would expect this to lead to conversions among people who cannot easily afford the tax. 

It's not necessarily that a Christian or Jew would say "Screw it, God doesn't pay enough." But if somebody lives among Muslims, likes their neighbors, and sees them as good people who worship the same God of Abraham, somebody might decide to assimilate into their community rather than paying a price to remain distinct. On the other hand, people who feel particularly strong bonds with and benefits from their minority group are going to pay the price to remain in their group. This will not only filter out people based on willingness to pay or ability to pay, it will also foster strong in-group ties that successful merchant classes have leveraged. In eras where people didn't have instant communication and bank transfers, long-distance trade required strong bonds of trust, bonds that are easy to foster in close-knit kin groups. And such bonds will be even stronger if people feel deep connections to far-flung members of their tribe, bonds even stronger than bonds with immediate neighbors. But if somebody does not make their living off those far-flung tribal ties, it's not just that they'll say "Eh, I'm no merchant, let's convert", it's that they'll likely feel stronger ties to their immediate neighbors. Their kids will marry into neighboring families. They'll convert. These are not crass transactions but normal human bonds.

So, as data-sparse as elements of his analysis are, I think they also bring a deeply human insight to groups that haven't always been treated well. Both the Jewish and Armenian diasporas in the Middle East have often been particularly successful in the merchant class, and similar things can be said for Parsis and other groups in India, but Clark argues that this is in part because the people who were most successful stayed in the group, and those who were more like their neighbors married their neighbors, as humans do when they connect to people.

One other point Clark makes is that everyone tries to explain the success of prominent minority groups in terms of some unique cultural factor, but every group has successful and unsuccessful people. If a group seems particularly successful, you need to ask if the people who aren't succeeding select out of the group. (The part about long-range bonds of trust is something I've read elsewhere; you'd expect it to matter less in a world where you can communicate and transfer money instantly across long distances.)

He makes related points about prominently successful diasporas in the US today. If you look at the ranks of medical doctors and certain other professions, you'll see disproportionate numbers from certain immigrant groups. Clark notes that much of this is not because some particular group is particularly good at medicine, but because US immigration law makes it easier for doctors for that group to come in and harder for poor manual laborers. He expects to see regression to the mean over generations, unless selection processes keep it so that only certain members can enter and replenish the ranks, while people who marry out of the group lose their identity.

A few other points:

1) If we all face Markovian dynamics in social mobility, nobody can truly feel secure. Obviously it's better to have more money rather than less, but people are loss-averse, and people with more money make bigger investments that can result in bigger losses or gains. So they face remarkably similar anxiety.

2) Adoption studies are all over the map on nature vs nurture, showing some nurture effects but also real nature effects. One thing he does note is that the nurture effects are subject to statistical range restriction: Adoptive parents go through selection processes that birth parents don't have to go through. Yes, there are always bad adoptive parents, but they'll be comparatively less abundant, so the bottom of the range will be under-represented in adoption studies. Combine that with diminishing returns on parenting beyond the basics, and you wouldn't expect huge nurture effects in adoption studies.

Pinker under-emphasized that point. Clark, an economist, is much more open to the idea that parenting matters but has diminishing returns. Sending your kids to a half-decent school and making them do their homework is important. Sending them to the fanciest school imaginable will not matter nearly as much, because of diminishing returns.

3) He disagrees with Charles Murray on the effects of assortative mating. People select for phenotype, not genotype, and phenotypes are heavily influenced by noise. So regression to the mean will keep happening, and with Markovian dynamics. (His disagreement with Murray made me breathe a sigh of relief.)

4) In the 1830's Georgia had a lottery to give formerly Cherokee land to white settlers. The value of the land was about $150,000 per lot in today's dollars, so effectively Georgia handed out a whole bunch of winning lotto tickets. A generation later, outcomes for descendants of the winners were remarkably unaffected by the win. Some people are bad at business no matter how much of a windfall you give them. (America's most recent head of state provides ample evidence for this.)

Monday, December 14, 2020

The Son Also Rises, Chapters 8-12

 The first few chapters in this section look at India (specifically the state of West Bengal), China, Taiwan, Korea, Japan, and Chile. The specific numbers that he finds when looking at England and certain US demographics don't always hold up; I think he's a bit too wedded to the idea that he's found a universal law of class mobility/persistence. Shocks from big events do matter, to some extent. However, a few patterns do hold:

1) If you look at the persistence of families in elite tiers of society, as opposed to intergenerational correlations of income, you see very slow regression to the mean. Income is just a proxy for status, and is imperfectly correlated 

2) Social reforms don't have nearly the effect that reformers were hoping for. Families that know how to navigate society still do fine, for a while at least. This should be deeply dismaying for social reformers who think that we can remedy inequality in the classroom. We can probably have some effects on the margin, but not only are our students not blank slates, they aren't even erasable slates, and large portions of the slate are off-limits to us.

3) When he's able to look at long time scales there are strong signs of Markovian dynamics in many cases, and these Markovian dynamics are as true for upward regression to the mean as downward regression to the mean. While the pace would be dispiriting to social reformers, the fact that each generation has an element of reset should be encouraging for them.

Chapter 12 is the most interesting chapter for me, as a physicist, because he looks at time reversal. He shows that if, instead of starting with people who were high-status centuries ago in England, you start with people who were high-status more recently, and then look into the past, going backwards in time they also regress to the mean. This is consistent with the idea that microscopic dynamics of Brownian motion look the same in either time direction, even if macroscopic dynamics look very different. This is just such a cool chapter that I would seriously consider discussing it the next time that I teach statistical physics. Seriously.

As to the persistence of the upper classes after putative egalitarian reforms, let me just note that upper-class jerks always write the best Diversity and Inclusion Statements. I don't if it's always been that way, or if they learned it after the French Revolution, but the offspring of the upper classes seem to be absolutely amazing at explaining that they're One Of The Good Ones and it's those other assholes who need to face a guillotine.