a response to Taleb Continue reading →

It goes to Abhijit Banerjee, Esther Duflo and Michael Kremer for work on field experiments in the economics of (under-) development. Alex Tabarrok at Marginal Revolution has coverage, starting here.

I am currently drafting an essay suggesting Edward Leamer for the Nobel Prize. Last week, I wrote the following paragraph:

The significance of what Angrist and Pischke termed the “credibility revolution in empirical economics” can be seen in the John Bates Clark Medal awards given to researchers who participated in that revolution. Between 1995 and 2015, of the fourteen Clark Medal winners, by my estimate at least seven (Card, Levitt, Duflo, Finkelstein, Chetty, Gentzkow, and Fryer) are known for their empirical work using research designs intended to avoid the problems that Leamer highlighted with the multiple-regression approach.

This year’s Nobel, by including Duflo, would seem to serve to strengthen my case for Leamer.

A commenter asks,

I’m curious what your opinion on Christopher Sims’s econometric work is now.

Sims is another macro-econometrician who was awarded a Nobel Prize for work that I think is of no use.

The problem in macro is causal density–there is a high ratio of plausible causal mechanisms to data. If you have dozens of causal variables and only a relative handful of data points, what do you do?

The conventional approach was for the investigator to impose many constraints on the regression model. This is mathematically equivalent to adding rows of “data” that do not come from the real world but instead are constructed by the investigator to conform exactly to the investigator’s theoretical pet theories. The net result is that you learn what the investigator wanted the data to look like. But other investigators can–and do–produce very different empirical narratives for the same real-world observations.

Sims’ approach was for the investigator to narrow down the number of causal variables, so that the computer can produce a model without the investigator doctoring the data. But that is not a solution to the causal density problem. If there are many important causal variables in the real world, then in a non-experimental setting, restricting yourself to looking at a few variables at a time is pointless.

His paper Let’s Take the Con out of Econometrics was highly influential. It changed the direction of econometric research.

There are several economics Nobels sitting out there, including Koopmans, Klein, and Tinbergen, that would not be considered prizeworthy in light of Leamer’s critique.

April L. Bleske-Rechek writes,
Mayviewed a random sample of poster abstracts that had been accepted for presentation at an annual convention of the premier professional organization in psychology, the Association for Psychological Science. We were disappointed to find that over half of the abstracts that included cause and effect language did so without warrant (i.e., the research was correlational). Of course, poster presentations are held to a less rigorous standard than are formal talks or published journal articles, so in a follow-up study, we reviewed 660 articles from 11 different well-known journals in the discipline. Our findings replicated: over half of the articles with cause and effect language described studies that were actually correlational; in other words, the causal language was not warranted.

In biological families, children resemble their parents in vocabulary and verbal ability; in adoptive families, they do not. The key implication is that Hart and Risley’s finding of a link between parents’ verbal behavior and their children’s verbal ability does not warrant an inference that parents’ verbal behavior influences their children’s verbal ability.

Somebody should put together a YouTube course on “How to be skeptical of statistical studies.” I nominate Russ Roberts.

E. Fuller Torrey writes,

there are now some one million people with serious mental illness living among the general population who, 60 years ago, would have been treated in state mental hospitals. Multiple studies have reported that, at any given time, between 40% and 50% of them are receiving no treatment for their mental illness.

He blames de-institutionalization for the problem of mass murders.

Ordinarily, I try to avoid commenting on the stories that dominate the news for short periods of time. I have a lot of doubts about going ahead with this post, but here goes.

I have a problem with every policy proposal that I have seen for dealing with mass shootings. The problem comes from Bayes’ theorem, which says that the probability of A given B is not the same as the probability of B given A.

When I taught AP statistics, I often used the 9/11 attacks as an example. Nearly all of the terrorists were Saudi nationals. But only a tiny percentage of Saudi nationals are terrorists. So a policy based on the assumption that Saudi nationals are the problem is going to involve a lot of costs relative to potential terrorist acts prevented.

The same thinking applies to guns. Guns account for 100 percent of mass shootings. But only a small percentage of guns are involved in such shootings. If guns provide a benefit to the people who do not use them for mass shootings, then trying to get at mass shootings by going after guns is going to involve a high ratio of costs to benefits.

The same probabilistic reasoning also applies to mental illness. Suppose that people with untreated mental illness account for 100 percent of mass shootings. There are still half a million untreated mentally ill who are not mass shooters. If the cost of mandatory treatment for them is high, then that may not be a good strategy for trying to reduce mass shootings.

Perhaps there are other social benefits to forced institutionalization of those who are mentally ill. That might improve the benefit/cost analysis for such a policy. But even so, it would be difficult to defend from a humanitarian or libertarian viewpoint.

As long as mass shootings remain rare relative the causal factors that are most often cited, it will be hard to come up with a cost-effective solution. This RAND meta-analysis supports my view.

Ben Thompson writes,

it was on 8chan — which was created after complaints that the extremely lightly-moderated anonymous-based forum 4chan was too heavy-handed — that a suspected terrorist gunman posted a rant explaining his actions before killing 20 people in El Paso. This was the third such incident this year: the terrorist gunmen in Christchurch, New Zealand and Poway, California did the same; 8chan celebrated all of them.

Hence, he supports censorship of 8chan. I disagree, although I find it a close call. My thoughts:

1. Correlation does not imply causation. The fact that terrorists were active on 8chan could mean that 8chan attracts individuals who are inclined toward violence, but it does not necessarily increase their propensity toward violence.

2. As a test, you may substitute “radical Islamic preacher” or “Palestinian primary school that teaches kids to hate Jews” for 8chan, and see whether you support the step of absolutely shutting down their right to speak. Maybe you do. I do not.

3. Probably the most effective way to use censorship to reduce mass shootings would be to refuse to allow the media to cover them. As it is, the mainstream media are giving mass shooters the notoriety that seems to be their main motivation. Sometimes there are suggestions that the media voluntarily exercise restraint, for instance in not naming the shooter. But as far as I know nobody wants to impose censorship on the mainstream media, even though they appear to be at least as guilty of aiding and abetting mass shooters as are the dark-web media.

4. Ultimately, censorship gives power to the censors. As time passes, the trend will be for censors to exercise more and more power with less and less wisdom, objectivity, and discretion. I think it is best to stay off the slippery slope altogether.

UPDATE: a commenter points to a very similar post by Craig Pirrong, aka the Streetwise Professor. Note that the first comment on that post repeats my point 3.