Advantageous Selection
I tweeted: Should I be worried or reassured that my taxi driver isn’t wearing a seat belt? An econ puzzle.
Most replies said I should be worried. I think that is correct and it reveals something of importance. First note that there is an incentive and a selection effect. All else equal, a driver without a seat belt should drive more carefully—that’s the rational response to increased personal risk. But drivers who forgo seat belts are probably more risk-loving or less safety-conscious across many dimensions. I think the replies were correct, the second effect, the selection effect, dominates: be worried.
What makes this an economics puzzle is that it reveals a failure of the standard adverse selection story. Adverse selection predicts that if someone wants to buy a lot of life insurance, the seller should be suspicious—fearing the buyer knows something about their own health that the seller doesn’t. Unusually healthy people, by the same logic, should buy less life insurance.
Notice the parallel to the taxi driver: the driver is buying less insurance (by not wearing a seat belt) and so, by adverse selection logic, should be the safer type. But that’s exactly backwards.
In reality, people who buy a lot of life insurance tend to be the kind of people who take care of themselves on many margins—they eat well, exercise, go to the doctor. Insurers know this, which is why the per-unit price of life insurance falls with quantity. Big buyers are the good risk, not the bad one.
The taxi driver puzzle is a clean real-world case where the selection effect runs opposite to what adverse selection theory predicts. Adverse selection theory is correct that information asymmetries can challenge markets but it’s often not obvious which way the asymmetry runs (who know more about your life expectancy, you or an insurance company with millions of data points?). Moreover, preferences and norms can make the selection run the opposite way so be worried about the taxi driver without a seat belt and be happy when someone demands a lot of life insurance.
Are the small tax havens really all that safe?
That is the theme of my latest Free Press column, excerpt:
If you are a Dubai resident, the chance that you will die in this conflict is very small. But you no longer can treat safety as something you do not have to think about. And you may face some uncertainty about when and how you can leave the country, a question that formerly was never in doubt. So two major advantages have vanished, even if the current conflict is settled soon. Another problem is that a substantial part of your supply of desalinated fresh water can be taken out by a well-placed missile.
More generally, the war underlines how tenuous the position of a place like Dubai is in the geopolitical order. I have enjoyed my three trips to Dubai, but I never felt entirely safe there on anything beyond a day-to-day basis. I always knew the place relied on protection from the United States and a certain degree of forbearance from its larger neighbors, including Saudi Arabia. Both Dubai and its larger encompassing unit, the United Arab Emirates, are extremely small.
And:
In most daily life, the small tax havens will feel safer than Cape Town. In the longer run, I am not entirely sure. My longer-run plans might be more robust in Cape Town. Or in Brazil. Or in Mexico. Those are all fairly dangerous places that nonetheless seem to have considerable macro stability in the longer run. South Africa has a pre-1930 history of taking in persecuted Jews from Europe and giving them an environment where they can thrive. Even the coming and going of apartheid, in 1948 and 1994, did not change South Africa’s high degree of security from foreign threats.
Dare I suggest that these larger places are more fun and also have more soul?
Worth a ponder.
I would much rather be exiled to Cape Town than to Dubai, all things considered, even assuming away the current conflict in the Middle East.
The trajectories of science and AI
From my podcast with Nebular:
Cowen: Mainly what they have done is tricked people. The Apollo program was a big trick. It was not intended as a trick. I’m pretty sure almost everyone behind it was quite sincere that it would lead to whatever. It was vague all along, but everyone was truly excited back then. I even remember those times, but it didn’t lead to what we were promised at all.
And you see that when you compare science fiction over time. So I think the norm is that new technology comes and people are tricked. Again, it doesn’t have to be a sinister, devious, conspiracy laden thing, but in fact, they’re tricked. And then it happens anyway. And then we clean up the mess and deal with it and move on to the next set of problems.
And that’s what I think it will be with AI as well.
Murphy: What is the trick with AI?
Cowen: It’s the old paradox. When you add grains of sugar to your coffee. Every extra grain is fine, or it may even taste better, but at some point, you’ve just added too many grains. So that’s the way it is with change. People use ChatGPT. It diagnoses your dog. Do I need to take the dog to the vet? What’s with this rash?
You take the photo…You get a great answer. Everyone’s happy. They’re not actually going to be happy at all the changes that will bring. And here I’m talking about positive ones. I’m not saying, oh, it’s going to kill us all. People just don’t like change that much. So they’ll be sold on the immediate, concrete things and end up seeing things happen where they feel there’s too much change because it will devalue their human capital, and we’ll adjust and get over it and move on to the next set of tricks. That’s my forecast.
Murphy: People don’t like change, but also people are bad at long term planning. Yeah. You’ve spoken before about how faith is a key requirement in terms of being able to plan over the long term. How do you bring that idea to policymakers?
Cowen: I don’t know, I think things will get pushed through for myopic reasons, like we must outpace China, which might even be true, to be clear, but it’s a somewhat myopic reason, and that will be the selling point. You know, I’ve read a lot of texts from the early days of the Industrial Revolution. Adam Smith is one of them, but there’s many others, and a lot of people are for what’s going on, they understand they will be richer, maybe healthier.
They do see the downsides, but they have a pretty decent perspective. But no one from then understood. You’d have this second order fossil fuel revolution, say the 1880s where just things explode and the world is very much different. And whether they would have liked that, you can debate, but they just didn’t see it at all.
We’re probably in a somewhat analogous position. I would say that the Second Industrial Revolution was the more important one. It was a very good thing, even though climate change is a big problem, but it really built the modern world. And with something like AI or any advance, there’s probably some second order version of it that’s coming in our equivalent of 1880 that we just don’t see, and it will be wonderful for us.
But if you told us, we’d be terrified. So how should you feel about myopia? I think as an intellectual, you should be willing to talk about it openly and honestly. But at the end of the day, I think myopia still will rule. And I’m not in a big panic about that.
To recap:
We’ve just published the video on YouTube, X, Spotify, and Apple Podcasts. We also published some extended show notes and the transcript on Substack.
Recent recordings of “big symphonies”
The Marek Janowski box of Bruckner symphonies I find to be the best Bruckner overall. And yes I do know many other versions, even Hermann Abendroth, though I cannot hold a candle to one MR reader I met recently who may know seventy or more versions of Bruckner’s 8th.
Vladimir Jurowski has recorded Maher 1, 2, 4, 8, and with 9 on the way and I read somewhere he will be doing the entire cycle. I expect these will end up as my set of choice.
Both are worthy of your notice, and they put to rest the myth that all the best conductors and orchestras operated in the now somewhat distant past.
On a related note, I flew to Pittsburgh recently to hear Honeck conduct Bruckner’s 8th (it is there I met the MR reader). I was amazed how good the overall performance was, and arguably Pittsburgh is now one of the two or three best orchestras in this country, at least for their favored repertoire. Go hear them if you can, Bruckner being their specialty.
Monday assorted links
1. What people get wrong about women’s rights (Alice Evans, The Economist).
2. The case against liberal interventionism.
3. More government by GPT (NYT).
4. “In 2013, museum management considered introducing a scheme to suction dust off tourists as they walked down the corridor leading to the Sistine Chapel while blasting them with cold air to reduce their body temperature and perspiration. The plan was aborted, presumably for logistical reasons.” (FT)
5. The ongoing migration of Kiwis.
6. Roger Garrison, RIP. And an obituary.
Iran/Venezuela facts of the day
Iran was once one of the key oil suppliers to the world. No longer. Its exports, constrained by sanctions, amount to less than 2 per cent of global supplies, most of which go to China at discounted prices.
A similar change has taken place in Venezuela. Once a star of world oil and one of the founding members of Opec, today it can hardly even be called a petrostate. It produces less oil than the US state of North Dakota and a quarter as much as neighbouring Brazil.
Here is more from Daniel Yergin at the FT.
Country Joe McDonald, RIP
Here is the NYT obituary.
Academic journals and AI bleg
Given the rapid pace of advancement of AI, how should academic journals adapt to these changes? One issue might be an excess of submissions, but what other questions should be considered here? Which reforms should be made?
Your thoughts would be most welcome.
What I’ve been reading
Frank McLynn, Villa and Zapata: A History of the Mexican Revolution. The best book on its topic, and one of the best books on Mexican history flat out. Everything is explained with remarkable clarity. By the way, the central government never really has controlled the entire country, or not for very long anyway.
Sean Mathews, The New Byzantines: The Rise of Greece and Return of the Near East. Anexcellent and original book, somewhere between a history and travel book. Views Greece as part of “the Middle East.” I found every page interesting.
Robert Polito, After the Flood: Inside Bob Dylan’s Memory Palace. An informationally dense, rambling, and frequently insightful and obsessive book about the “late” career period of Bob Dylan. When does his “late” period start? 1990 perhaps? I remember thinking in 1990 that we were well into Dylan’s late career phase. But that was thirty-six years ago!
Muriel Spark, The Driver’s Seat. If you like her at all, you will be entranced by this one. With a radical ending, as you might expect.
Richard Holmes, The Boundless Deep: Young Tennyson, Science, and the Crisis in Belief. A fun new book on Tennyson’s relations with the science of his time, and how he drifted away from religious belief.
Partha Dasgupta, On Natural Capital: The Value of the World Around Us, is a popular summary of some of his thinking on valuing the environment and natural resources.
Davd Epstein, Inside the Box: How Constraints Make Us Better. A good popular look at what the subtitle promises.
José Donoso, The Boom in Spanish American Literature: A Personal History is a good lshort overview, noting that Donoso’s own The Obscene Bird of Night is one of the great underrated works of 20th century literature.
Jazz Samba
Sunday assorted links
2. Adam Smith against hegemony.
3. US median income vs. Europe.
4. The academic papers is a dead format these days (read the whole chain).
A Fly Has Been Uploaded
In 2024, the entire neuronal diagram of the fruit-fly brain–some 140,000 neurons and 50 million connections–was mapped. Later research showed that the map could be used to predict behavior. Now, Eon Systems a firm with some of the scientists involved in the fruit-fly research and with the goal of uploading a human brain has announced that they uploaded the fruit fly brain to a digital environment.
The digital fly appears to behave in the digital environment in reasonably fly like ways–this is not a simulation, the fly’s “sensors” are being activated by the digital environment and the neurons are responding. Some more details here.
N.b. this work is not yet published.
Addendum 1: Of course Robin Hanson is an advisor to Eon Systems.
Addendum 2: In other news, human brain cells on a chip learned to play Doom. No word on whether they were conscious or not.
A market-based officer retention system?
The Army is launching a new Warrant Officer Retention Bonus Auction. This initiative introduces a market‑based approach to retaining senior technical talent while ensuring responsible stewardship of taxpayer dollars. The program represents a shift from traditional, fixed‑rate bonuses to a more flexible, market-driven system.
The structure is designed to make the best strategy straightforward—bid your true value. Eligible warrant officers will submit a confidential bid indicating the minimum monthly bonus they would be satisfied receiving in exchange for a six‑year Active‑Duty Service Obligation. Overbidding increases the risk of missing out on a bonus, while underbidding could result in commitment to a lower rate. Army leadership believes the system rewards transparency and encourages officers to carefully consider the compensation that would make them comfortable with continued service.
“The goal is simple. Reward as many qualified Warrant Officers as possible with the most competitive bonus the budget allows,” said Lt. Col. Tim Justicz, an Army economist who helped design the program.
Once bids are submitted, the Army will determine a single market‑clearing bonus rate that retains the maximum number of qualified warrant officers within the available budget. Every warrant officer whose bid falls at or below that rate will receive the same bonus amount. This means that warrant officers who bid lower than the final rate will still receive the higher, market‑determined bonus.
Here is more, via Charles Klingman.
On social media and parents (from my email)
Fron anonymous:
I personally think social media is pretty bad for people (kids and adults). I got off Facebook around 2009. I never got on Twitter. I had Instagram for a while but only followed my wife to see her posts of our family. This worked great until Instagram started feeding me content beyond the people I was following (really just my wife), so I quit using it. The only social media I currently use is Substack (not sure if that counts?). But the same dynamic may be playing out there as well (the algorithm feeding me stuff I don’t want, and me getting locked into wasting time doom scrolling).
HOWEVER, I completely agree with your point about parents. Our 14-year-old son has an iPhone, but we have locked it down pretty tight. It took some work on our part, to be honest. And we have to be pretty vigilant about enforcing the no-phone-in-your-room rule (which is a source of conflict sometimes). Our son has no social media accounts. He can text and he has access to a few messaging apps that they use at his school. Beyond that, we’ve basically shut down his ability to access the internet on the phone. His Chromebook works perfectly well for any legitimate internet needs.
In principle, any parent can do what we’ve done. So why don’t they? Why are they begging the government to do something they could just do themselves, albeit with a little work? Well, I’ve been struck by how badly many parents desperately need their children’s approval. They find themselves incapable of disappointing or upsetting their children on even the smallest of things. They know they should tell their kids not to use TikTok (or whatever), but they don’t want to make their kids mad. That’s why they want someone else to do it for them.
I don’t get it. Perhaps I’m overly cranky, but I honestly don’t mind it if (when) my kids get mad when I do something I believe is in their best interest. I simply don’t believe my children’s emotional reaction is a very good guide to parenting. Because they’re children. And they don’t know very much. And they especially don’t know what they don’t know and that’s why I’m here. If I won’t tell my kids no when they need to hear it but don’t want to hear it, then what good am I? My wife feels the same way. But we see lots of families that clearly feel differently.
Okay rant over.
See also Arnold Kling on related ideas.