Violent Crime Costs US Cities $7 Trillion | Cremieux

Cremieux (0:00)

So, cities make you richer. Like, if you're in SF, you feel like you want to do more work. You can find capital.

MTS (0:06)

So true.

Cremieux (0:07)

It's just a great, great place to be. We don't have very many places like that in the US, where there are these huge agglomeration effects. These benefits from density.

And a lot of the reason for that is crime. Direct agglomeration loss, meaning like moving out of the Metro core, is on the order of $7 trillion. Which is a large amount, yeah.

MTS (0:26)

That is a large amount.

Cremieux (0:27)

So there are massive, massive losses as a result of the same forces that lead to limited agglomeration effects. And they go and limit it going forward, which just makes America lose out on trillions of dollars.

MTS (0:40)

All right. And we're back. We're live with Cremieux. Once again, to talk about some interesting things. What do we start with? The pizza?

Cremieux (0:52)

Yeah, let's do it.

MTS (0:53)

So what is the pizza exam?

Cremieux (0:56)

So the OECD, the Organization for Economic Cooperation and Development, this big, it's kind of like NATO for economics, except it has a lot more countries in it. They do this test every three years, and now it's going to be every four years, I believe, moving to that, where they go across different countries, and all the countries administer the same test to all their 15-year-olds, or at least a representative subset of their 15-year-olds. That's the hope.

And then once those tests are done, they release all the results, they show us, oh, look, here's how your country does relative to this country. Here's how your country does relative to your country in the previous cycle, it's like three years ago.

Interesting. Basically, they're competency tests. They're, how are they doing in mathematics? How are they doing in reading? How are they doing in science? Stuff like that. They also assess stuff about kids' interests, like are they interested in mathematics? Are they interested in science? Do they want to go on to higher education? Yeah.

MTS (1:51)

So, the reported PISA scores have dropped every three years since 2012, including to 2025 But your argument is that this actually doesn't mean much, because if you adjust for measurement variance, then we see no statistically significant drop.

Cremieux (2:11)

Yeah. So, there's a curious thing. These tests are supposed to be comparable within a year. And they make it so you can compare different countries. And they talk about making them comparable over time. But they don't actually do basically anything to make the scores comparable over time. So, they're not comparable over time. And there are two issues, two really big issues. There are several issues. There are two really big issues. One is that they're not comparable and like the trends and the scores tend to just evaporate with any valid method of correcting for differences in how the scores or the test questions are interpreted over time by like different samples. So, like kids in 2020 might interpret questions differently than kids in 2018

And as a result, they might score differently, even though their underlying levels of competence are really different. Correct for this and basically all the trends go away.

The other issue is that they show a lot of the time aggregate trends, but the sample isn't consistent. So, you might add China in one year and take it out in another year. In 2009, we had India in there and they haven't participated since. We often have countries coming in and out of the sample, and sometimes they exaggerate whatever trend up or down happens in the aggregate. And for some reason, this just doesn't seem to get talked about, even though it's like a very obvious problem.

And when you do actually correct for, you know, differences in sampling and differences in how the tests are interpreted over time, every single wave, the trends go away.

MTS (3:42)

Well, I mean, it does seem like education was harmed by COVID to some extent. There's also this post I saw yesterday with the SAT annual report among public school students in Michigan. And it was like 14% of students did better than 1200, despite the SAT getting somewhat easier in recent years.

And only 3% did better than 1400

Cremieux (4:08)

So, that's kind of a misinterpretation. I mean, the SAT is no easier or harder. It's just that where the midpoint of the range has shifted. It is easier to get a very high score.

That doesn't mean... How does that not mean the test?

We've gotten better at this type of test. It doesn't mean that the kids are smarter, they're less smart, or the test is...