Lyman Stone 石來民  🦬 🦬 🦬

This paper includes an extended discussion of the Bradlaugh-Besant trial in England, and cites the paper I mentioned favorably, AND ALSO notes that the coverage of that trial explicitly often described the Malthusian perspective as "French." #NBERday

So these papers are strong together to show that 1) low-birth-rate norms began in France uncorrelated with economic modernization, 2) they spread to other cultures through discrete channels, 3) Bradlaugh-Besant was one very key such channel. Neat! #NBERday

The paper then matches a map of about 300 languages and dialects in 18th-20th c Europe to a historical database of fertility transition, and uses commonly accepted measures of linguistic similarity across languages to estimate cultural connectedness. #NBERday

They then also use a decent measure of physical connectedness (I'd prefer an actual measure of roads/railroads rather than static physical linkage but ymmv), institutional connection (borders changes), urbanization, infant mortality, etc. A nice fancy panel. #NBERday

They specifically also test linguistic distance to English and some other proxies for industrialization, showing that low birth rates ARE PRIMARILY NOT A SIDE EFFECT OF INDUSTRIALIZATION, but are actually a separate cultural movement. #NBERday

Sadly the visuals are pretty boring so no cool screen shots, but the key takeaway is pretty simple and clear. They're doing basically descriptive work which is easy to interpret: fertility transition is a cultural process which often predated industrialization. #NBERday

While industrialization did ALSO lower fertility, there are actually MULTIPLE forces at work here: yes, urbanization and better health reduces birth rates! But SEPARATELY the diffusion of "French fertility behaviors" ALSO reduced fertility by a large amount. #NBERday

Now I'm saying "French" but the authors take pains to note that this isn't actually something inherent to French-ness but just something which began in France. But I note that for much of the 19th century, it was perceived as a French cultural distinctive by outsiders. #NBERday

THAT SAID, we still don't know what "it" was. Was it nationalism? Was it revolution? Was it republicanism? The Napoleonic Code? Liberalism? What was the actual norm? For Bradlaugh-Besant we basically "know" what the broad outlines of the norm were. #NBERday

For this "French linguistic distance effect" measuring cultural transmission.... it isn't clear AT ALL what it was that was being transmitted. Perhaps secularism? Who knows!! That's a topic for another day. #NBERday

Speaking of another day, I will finish #NBERday tomorrow my time, so this evening US eastern time.

This paper looks at the harmattan, a seasonal wind pattern in the Sahel that brings large amounts of dust. Pollution exposure in utero is widely shown to increase infant mortality: does that hold up in west Africa? #NBERday

Yes: it does. Pregnancies exposed to much more intense harmattan winds end up having much higher odds of premature death. They only study infant deaths, not miscarriages, but more miscarriages are probably happening as well. #NBERday

Their effect size is well within the range of what we see in other similar studies. They show that general birth rates don't hugely differ across harmattan wind exposure. And they also show that effects are reduced in areas with better healthcare and higher incomes. #NBERday

In other words, this one is pretty much tied up with a bow on top. All the effects are reasonably sized, have a clear channel, are well-demonstrated, and consistent with prior literature and theoretical expectations. Folks. Air pollution is bad. #NBERday

The nice theoretical improvement this paper makes on other air pollution studies is that it is a major form of air pollution which is not man-made, and which has some plausibly random variation. #NBERday

Now let's be clear what this paper actually claims: If you send parents 3 text messages about their kid's education during the week, it can raise the whole treatment group's standardized test scores by 0.032 standard deviations. #NBERday

Is this big or small? Well it's a bit hard to say. But for one specific metric the authors discuss as an example ("rapid letter naming"), the effect size was to increase student scores by ***16%***. Which to me seems very big. #NBERday

In fact, it seems too big. Let me note that this research design is becoming increasingly common. Stanford has promoted it and it's been used in a number of experimental cases, and they reliably turn up pretty big effects from it. #NBERday

But that is precisely why I have a hard time believing these findings. It boggles the mind that you can increase a kid's alphabet acquisition around ages 3-6 by sending a parent 3 text messages. #NBERday

I know that this critique I'm making is unsophisticated. But I simply do not believe that, if we rolled out a national 3-texts-per-week program, American education would be noticeably better. #NBERday

If this study is TRUE then it reveals an off story things: 1. Parents can have a *huge* influence 2. But most parents *barely lift a finger* on their own to teach their kids 3. But you can *easily motivate* them to teach their kids This is wrong. #NBERday

I'm sorry I can believe parents are influential, but I simply don't believe that there is this huge low-hanging fruit where you can send a parent 3 text messages and BAM they're a better parent. Uber for parenting I guess? #NBERday

Having looked through the study, I don't know what's wrong. So this is making me uncomfortable and it is nudging my priors. But I think strong priors here are reasonable, because the implications of these findings are somewhat absurd. #NBERday

These effects are bigger than a lot of quite expensive educational interventions. If it's really the case that sending a few text messages has this effect then we could close down a bunch of $$$ programs, send some text messages, and be better off. #NBERday

And while on one level as a conservative that appeals to me, the implied coincidence between parents who are highly effective teachers but just can't be bothered to do that unless sent a text just seems TOO GOOD. #NBERday

I hope other researchers more familiar with the research design will explore this method, and the entire family of research. If the stuff coming out of this line of research, not just this paper, is correct, then it has big implications. #NBERday

And to be clear, the implication is a very uncomfortable one for progressives. The biggest benefits were for poor and minority parents, which makes the program sound progressive, but carries a kind of mirror-image implication about pre-treatment parenting. #NBERday