Thread by @lymanstoneky

Some fun new #NBERday papers today!

First up, a paper with very strong @Noahpinion energy:

Q) Does discrimination make it harder to recruit an army?

A) Yes! Very cool evidence from US WWII volunteer data. nber.org/papers/w29482

So first the theory. The argument is that wars are a major test of "state capacity," i.e. the ability of the state to marshal resources to achieve large tasks. A key component of the wartime test of state capacity is the ability to recruit manpower for the army. #NBERday

One possible threat to state capacity is a populace which is deeply divided and regards one another with hostility, such that state-based projects like "the nation needs to win the war against some other nation" may have a hard time motivating pro-state behavior. #NBERday

A classic historical case of this is Quebec anti-draft and anti-enlistment activism during WWI and WWII. Quebecois didn't really feel "part of" Canada, let alone the British empire at large, so when war came, were very opposed to being drafted, and fewer enlisted. #NBERday

But do we have a similar case in the US? It's often been argued this might have been an issue based on essentially historical or interpretive accounts. Racial animus could have caused different groups to not feel that the American cause was *their* cause. #NBERday

The authors use county-level measures of discrimination that are assembled from a mix of indicators: segregationist vote shares, KKK presence, lynchings, residential segregation, etc. I'm satisfied the discrimination measure is a good proxy. #NBERday

Helpfully, they also show their results are robust to any of the components of the index individually. So that's encouraging. One nice extension would be to find some indicator of like counties with sundown laws or something; *formal legal* discrimination. #NBERday

And they measure enlistment with enlistment data by county.

The result is they can compare county-level trends on enlistment rates by racial group compared to a measure of the intensity of discrimination in a county. And they can see how this all changes over time. #NBERday

Their tables are persuasive: they find that after the shock of Pearl Harbor, enlistment rates for white men jump similarly in higher- and lower-discrimination places, but for black men the enlistment bump is suppressed in high-discrimination places. #NBERday

This is a really nice demonstration, then, of an idea hypothesized in the past and demonstrated in other contexts: lack of inclusivity is a threat to state capacity.

I'm listening to a podcast right now on ancient history and I've gotta say, Cyrus the Great, folks. #NBERday

Next up is a paper I first encountered via one of the authors' personal sites a few years ago and I've been waiting to see hit a more citable format.... #NBERday

Q: Which major theory of the demographic transition is more correct: material progress changes economies, or cultural exposures change ideas?

A: get you a theory that can do you both
#NBERday
nber.org/papers/w29480

In the annals of paragraphs that are wildly understated on what they're showing, this one has got to be a winner. #NBERday

Love to see demographers take theory seriously, specify quantitative models to test the theory, then come back and say, hay, demographic theory is pretty sound and well-describes reality! #NBERday

What they find is 1) over the centuries, mortality transition usually comes first and it usually comes around a similar level of real income per capita for a country. 2) fertility transition usually comes later, the timing varies, and is less income-dependent. #NBERday

Their theoretical model which allows for both material-progress factors and ideational contagion via cultural/geographic links explains a lot of quirks in the data: it predicts accelerating transitional speed, for example. #NBERday

What's nice here though is demographers often end up pigeonholed into "it's modernization" or "it's ideation." But in fact, it seems like it's both! Both factors are very predictive of the timing and speed of transition! #NBERday

Since the lag time between mortality and fertility transition appears highly variable, ranging from 0 to like 100 years, we can't argue that it's *necessarily* the case that it's "material progress causes mortality change and ideational change." #NBERday

So they have this very nice chart showing the probability of transition at various combinations of real GDP per capita and "access to transition" (basically geographic/cultural/linguistic exposure to other countries' transition) #NBERday

An important takeaway here is that transitions getting faster is not accidental, but a very real dynamic. Faster transition + stable income-at-transition = More countries getting-old-before-they-get-rich. #NBERday

This is actually a big problem folks! A "short" transition creates a very narrow but large bubble of young workers, an extremely small window of time in which to "develop." Miss that window and you have an already-old population with high health needs, but still poor. #NBERday

Any time I've done interviews on this journalists always decided not to use quotes on it but it's a huge issue: demographic transition in the poorer countries today is going WAY too fast for economies to keep up! They have absolutely no "retirement plan!" #NBERday

The elder crises coming for India and Nigeria are absolutely crazy and the kind of revenue raising necessary to pay for them boggles the mind given their income levels. #NBERday

a modest subhead
#NBERday

When they use a fairly parsimonious model assuming technological diffusion and rising skills premia account for all change, here are some country they get (with fixed "culture" assumptions everywhere): #NBERday

Great Britain, the country they train the data on, is a good fit. Denmark and Spain are decently good fits, though recent Spain fertility is a wide miss. Chile is a not-so-good fit. Malaysia is fine. Chad is lolololol. #NBERday

This isn't a huge problem though, the authors readily grant this is a simplified model not accounting for cultural differences we all know matter. Moreover, they find that *education* is an area their model underpredicts: #NBERday

The actual change in education has been a lot faster than you'd expect from their model parameters, which might explain why on average fertility transitions have been a bit faster than expected, and could explain some ideational differences. #NBERday