Thread by @lymanstoneky

It's #NBERday and I am gonna start with some extremely bad news.

Q: Would pricing groundwater in California yield water sustainability?

A: Errr.... depends on your model.
#NBERday
nber.org/papers/w28706

The authors exploit changes in electricity costs in California AND variation in difficulty of pumping water to guess at the price elasticity of water. Cool! #NBERday

They find that a $10-acre-foot water tax would allow California to hit its water use sustainability targets and also cause the reallocation of 3.9 percent of California's total agricultural acreage. #NBERday

But here's the issue.

Some of that reallocation would be removing annual crops and planting *perennials* like tree nuts. NOT because tree nuts use less water per acre, but because tree nuts are *higher revenue per water gallon*

I feel like the authors misspecified their model somewhere, because an acre of rice or cotton uses TONS less water than an acre of almonds. #NBERday

Tree nuts on average use 2x to 3x as much water per acre as most annual crops, AND they cannot be allowed to die during drought, so they uniquely contribute to water stress during the most sensitive times. #NBERday

So when I saw this paper saying that a $10 acre-foot water fee would both *achieve* California's water goals *and also* lead to more perennial crops, I was perplexed. Those results *can* go together, but it's *weird*. #NBERday

So when we look at their model, we get this little nugget:

Which tells us they did not in fact meaningfully model agriculture in California. #NBERday

Folks, they made a model where each year a farmer chooses to grow either corn or almonds *and can change each year.*

This is not reality! #NBERday

In fact, crop-switching costs are high! Farmers tend to specialize to some extent, equipment is expensive, and of course *perennials* are *perennial*.

Y'all they included *vineyards* in their perennial crops. #NBERday

Vineyards do not just choose to stop being vineyards and instead grow rice for a few years then go back to growing grapes because water prices changed! #NBERday

Farm choices about what crops to plant are *extremely* state-dependent *especially* when you're talking about crops that have different productive lifespans, different crop insurance coverage re: drought, different ability to withstand drought-related stress, etc. #NBERday

Now I suspect that what we have actually learned is that higher water prices *really do* motivate more use of perennials, because they are higher revenue per water input, but that they also take more labor to harvest and maintain, so some land is fallowed. #NBERday

Weirdly enough we've seen the same switch happen in water-scarce central Asia in recent years! #NBERday

But this doesn't mean total water stress is reduced. #NBERday

Let's stay on the environment:

Q: Does pollution kill people?

A: Yes, pollution kills people (in this case, surgery patients).
#NBERday
nber.org/papers/w28708

This paper leveraged the quasi-random assignment of patients to surgical dates in China vs. daily pollution totals and found that having a surgery on a bad AQI day considerably increased the risk of death. #NBERday

Here's the graph. Effects are small because few people die post-surgery, but still pretty interesting! #NBERday

Staying on death...

Q: Do travel bans work?

A: If they're draconian enough, yes (examples here India, China, Kenya, some other developing countries)
#NBERday
nber.org/papers/w28699

Basically, they suggest that *short* travel bans do not do much of anything, *medium-term* travel bans (i.e. ~60-140 days) make COVID cases *worse*, but *long* travel bans (150 days+) are very effective. #NBERday

To beat COVID, you have to commit to LONG interventions. #NBERday

Now look those estimates are bad. Case rates are extremely unreliable.

But the mechanism is plausible: short interventions do little, medium-length interventions may actually create even more population churn, but long interventions create more localized habits. #NBERday

Speaking of death and habits...

Q: What happened when AUstralia banned teen drivers from carrying passengers between 11:00 PM and 5:00 AM?

A: Deaths in that group fell DRAMATICALLY.
#NBERday
nber.org/papers/w28707

And it's not because teens drove safer. They were less likely to appear even in crashes for which they were NOT at fault! It's that this reform simply led to teens driving *less*. #NBERday

These good habits persist for up to 3 years after a driver is no longer subject to the restrictions: Australia actually CREATED A NORM where young drivers learned not to drive late at night with passengers! #NBERday

So I'm just gonna put it out there that "it's illegal for teen drivers to carpool between 11 and 5 AM" actually seems like a really good policy we should just roll out everywhere. #NBERday

Let's talk about a different case of norms, youth, violent deaths, and habits.

Q: What leads young men to join violent militia groups?

A: Seeing their wives and sisters raped.
#NBERday
nber.org/papers/w28714

This study is one in a family exploring violence in Congo using INCREDIBLE datasets of reconstructed violence history. I've tweeted about prior papers. This one is, like the others, truly remarkable in its research. #NBERday

They randomly identified a few hundred villages in North and South Kivu in eastern Congo, and interviewed a random subsample in each village of households. Each household interview TOOK AN ENTIRE DAY. They paid each respondent well above a day's wage. #NBERday

For a whole day a researcher would interview, discuss, tag along, observe, just generally get to know the person's life and story, and work through an extensive but intuitively-designed life history questionnaire. #NBERday

In this area, foreign militias (especially the folks who led the genocide in Rwanda then fled into Congo) have engaged in tons of looting and pillaging, so most victims of violent attacks were victimized by those groups. #NBERday

Being victimized by a foreign group (read: Rwandan genocidaires) increased the odds that a person would end up joining a Congolese militia group BUT ONLY if they were attacked while the official military was ABSENT. #NBERday

This.... is pretty fascinating. Basically, when people perceive that the state cannot protect them because they have been victimized by armed bands, they respond by joining *opposing* armed bands.

Which.... DUH. #NBERday

But it is SO RARE that researchers can actually "see" this process and measure it and understand it carefully! The team in the Congo is doing incredible work merging quantitative and qualitative research. #NBERday

Now to return to the original question, note that the *kind* of attack matters. Sexual violence is most predictive, and sexual violence *against a man's wife* even moreso. #NBERday

The effect of attacking a man's wife or sexual assault on anyone in his household is nearly *double* the effect of some other assault on his household. #NBERday

These effects are extremely robust. They go through tons of different controls, sub-specifications, robustness tests, validation checks... it's the real deal. #NBERday

The authors refer to this behavior as "parochial altruism," i.e. that people will respond to *community* threats but not wider threats; they show that effect sizes diminish when its some other family attacked or a neighboring village. #NBERday

i.e. if your wife is raped, you join a militia; if you hear about a woman raped in the neighboring village, maybe not so much.

Which speaks to the fact that human consciences and capacities have limits, of course. #NBERday

But more broadly, and speaking not professionally here but out of sympathy for the people being studied here, I think we laypeople can refer to what these Congolese men are doing as "being good and honorable husbands and fathers." #NBERday

Would it be better if they were more conscious of someone more distant violence and took proactive measures? Yes, sure.

But the paper shows these men aren't undermining state provision of security. They *only* mobilize when the state *cannot*. #NBERday

They aren't trying to build a rival state and protect the whole province, they're just doing their best to fill in the security gap the state can't manage. So not responding to attacks on other villages doesn't seem unreasonable. That's the state's job! #NBERday

Anyways. That's all I'm gonna cover today. Bunch of other cool papers about education and such! #NBERday