Kevin Drum

There are two types of studies about the association of lead and crime. The first is an ecological study that compares lead levels in a population of children to later crime levels in the same population. The second is an individual study that follows specific children to adulthood and compares each child's lead level to their later criminal history.

Ecological studies are common and show strong evidence that childhood lead poisoning leads to criminal behavior later in life. Individual studies are rarer, and today a paper was published that summarizes the results of every known individual study—17 in all. However, it's not a conventional meta-study since it doesn't attempt to combine the studies into a single result. It's more of a simple qualitative review. The conclusion was pretty clear:

The range of outcomes that were significantly associated with lead exposure [7 of the studies] were primarily related to an arrest, incarceration, or conviction of some type, with increasing blood lead concentrations in childhood prospectively associated with later arrests and convictions in several studies. In addition to this association, 7 studies found strong associations between lead exposure and later delinquent or aggressive behavior irrespective of arrest status....Even in reviewed studies in which statistically significant associations between lead and crime did not exist [2 of the studies], significant relationships between lead and damaging patterns of behavior that are more likely to lead to negative long-term outcomes were still present.

....This review demonstrates an association between exposure to lead and the later development of delinquent, antisocial, and criminal behavior. Although borderline levels of risk are seen in several of our included studies, most are above the null value and estimates of risk are generally precise.

However:

A shift from lower to higher effect estimates as magnitude of lead exposure increased [i.e., a dose-response effect] was a common and not unexpected outcome in our review. This finding is commensurate with existing ecological literature, although the magnitude of the effect size was much lower in our reviewed studies than that seen in ecological ones.

In summary:

• Overall, there is an association between lead and crime in individual studies.
• Even studies that don't show an association with crime show an association with delinquent and antisocial behavior, which themselves are associated with criminal behavior.
• Most studies showed a dose-response relationship. That is, higher lead levels predicted higher crime rates.
• The increase in effect size from higher lead levels is not as dramatic as that shown by ecological studies.

In other words, this is yet more evidence for the link between lead and crime. But the link might be a little weaker than we thought.

I see that Donald Trump has just been indicted for illegally conspiring to overturn the 2020 election. This means we're now up to three Trump indictments, with a fourth from Georgia almost certainly on its way soon. This really is unprecedented, but then again, Trump has engaged in an unprecedented amount of crime.

Will we end up sending a 77-year-old man to prison? That would be unusual. But as the saying goes, don't do the crime if you can't do the time. Nobody forced Trump to do any of this stuff.

Over at National Review, Michael Brendan Dougherty notes a Gallup poll showing that public confidence in the military has declined lately:

As the perception developed that progressives have taken over the top brass in the military, the most important cohort for military recruiting — existing military families — has been pulling away from the U.S. military as an institution.

That's not really a serious response, but it raises the question of what did cause the decline—which is real. Here's a time trend from the General Social Survey:

Confidence in the military is down 15 percentage points since 2018, though some of this is due to normal reversion after an artificial spike in confidence following the 2017 strike on Syria. Given the timeframe, the most obvious possibility is that it's due to some kind of COVID effect, but if that were the case public confidence would have declined in everything, not just the military.

This got me curious, so I took a look at the 2018-22 decline in confidence for everything that GSS asks about. It's worth noting that some institutions had such low confidence already (Congress at 6%, for example) that they really couldn't decline very much. So the only way to measure this properly is to calculate the percentage change from their starting points in 2018. Here it is:

The military doesn't look all that special now. It went from 61% to 46%, a drop of one quarter, or 25%, which puts it right in the middle even though it's coming off an artificial spike.

I don't know why COVID did this. Maybe people just express lower confidence in institutions when they feel frustrated and scared. But whatever the reason, COVID seems like the culprit, not wokeness, and the military doesn't look to have done any worse than average.

Getting a haircut at the American Barber Shop in Santa Ana

Construction spending continues to be flat over the past few months:

The good news is that residential construction isn't falling anymore. The bad news is that nonresidential construction isn't rising anymore. We are just puttering along waiting for either something or nothing to happen later this year.

New hiring dropped by more than 300,000 workers in June, taking it to its lowest point since early 2021:

In better news, job openings and quits were pretty much flat. More evidence of a soft landing?

A new study is out that compares the amount of work completed in a day by office workers vs. home workers. It's based on random assignment to either home or office of data-entry workers in India, so the measure of "work" is pretty objective. The study has two basic conclusions:

• Your best workers are the ones who want to work at home.
• They are also the ones who lose the most productivity when they transition from the office to home.

On average, home workers are 18% less productive than office workers. Those who actively prefer working from home are 27% less productive:

Most of this difference is simply because home workers don't work as hard, but some of it is because home workers also don't learn as fast as office workers.

Roughly speaking, this result is consistent with time use studies in the US, which find that home workers put in about 25% fewer hours than office workers. It's also consistent with other studies and with anecdotal evidence.

The last time I wrote about working from home I got back this comment: Hard workers will work hard no matter where they are. Goof-offs will goof off no matter where they are. This is probably true. Certain occupations—like engineering—probably attract a lot of hard workers who do fine at home. Other occupations don't, and they make up the vast majority of the workforce. For those folks, the evidence is pretty strong that they just don't get as much done at home as they do in the office, and the reason is simply that they take more time off during the day because there's no one around to keep an eye on them.

Conventional wisdom says that, on average, you should earn more from an investment in stocks than an investment in government bonds. This is the equity premium, the reward you get for making a riskier investment.

But the equity premium has been drifting lower for the past year and a half. As of today, it's negative.¹ You can earn more from 100% safe government bonds than you can from the stock market:

If you discount, as you should, the possibility of a red-hot economy that sends corporate earnings soaring, this can correct itself in two ways. First, a recession could cause the stock market to tumble, which would produce higher yields compared to stock prices. Second, bond yields could go down, which will happen if the Fed thinks a recession is imminent and begins to reduce interest rates. Or both.

The equity premium is a pretty core feature of the economic system, and it's likely to correct itself one way or another. Unfortunately, both ways this can happen involve recessions. Stay tuned.

¹This is using trailing 12-month earnings to calculate the earnings yield. It will come out a little differently if you instead calculate it using projected earnings over the next year.

Check this out from JAMA:

Fair Allocation of Scarce CAR T-Cell Therapies for Relapsed/Refractory Multiple Myeloma

I imagine this is of interest to practically nobody, but it's right in my strike zone. So let's see what it says:

A 2023 study of centers offering CAR T-cell therapy for multiple myeloma found that for every allocated slot per month per center, there were 20 patients on the waitlist; patients were waiting a median of 6 months prior to leukapheresis.

That sounds about right. I waited even longer than that. In most places, the earlier you get on the waitlist the better, which the authors think is unfair:

This “first come, first served” allocation system is problematic because the US health care system prioritizes wealthy and well-connected patients for advanced treatments such as CAR T-cell therapy. In the clinical trial setting, there have been socioeconomic and racial disparities in enrollment for CAR T-cell therapy in multiple myeloma, with low representation of Black individuals.

Hey, that describes me! White, affluent, and well insured. The authors believe that the highest priority should instead be given to patients with the earliest date on which they qualified for CAR-T, which is usually only after four previous complete chemotherapy regimens. So even if you signed up later than me, for example, you'd have priority if you had an older case of multiple myeloma and had finished up your fourth round of chemo before I did.

Interestingly, this was something I remember thinking about while I was waiting to get to the top of the waitlist. On the one hand, obviously I thought that someone with a more serious disease profile—i.e., someone who might die if they had to wait very long—should be pushed ahead of me on the waitlist. On the other hand, if you do this too aggressively it means that most people won't receive the treatment until they're so sick that they have little chance of fully recovering. Shouldn't we also include some of the relatively healthier patients who are likely to benefit the most from treatment? The authors also point out that the very sickest patients have a high risk of dying just in the few weeks that it takes to prepare the CAR-T cells. When that happens, the CAR-T slot has been wasted since every cell preparation is unique to the patient and can't be given to someone else.

So what to do? In my case, I believe that City of Hope and Kaiser Permanente have a committee that reviews cases and creates the waitlist—which is based partly on when you signed up but partly on other factors that the committee weighs. That's probably why it took a year before I finally received treatment. I was too damn healthy.

As you may recall from Saturday, the four researchers who wrote "The Proximal Origin of SARS-CoV-2" initially believed that it was quite possible—maybe even likely—that the COVID virus originated in a lab leak. One of the big reasons for this belief was the existence of a furin cleavage site in the virus genome, something that had never before been seen naturally in a SARS-like virus.

However, in late February 2020 they discovered a bat virus that had an insertion in the genome at exactly the point of the furin cleavage site in SARS-CoV-2. This strongly suggested that the furin cleavage site could indeed be created naturally by evolution, which in turn meant it was no longer necessary to think the virus could only have been created artificially.

But the PO paper didn't say exactly that. It said an artificial origin wasn't plausible. But why say that unless there was explicit evidence that the lab leak theory wasn't just unnecessary, but positively unlikely?

Well, there was. I didn't include this in my timeline of the internal Slack messages exchanged by the authors while they were writing PO, but one of the messages outlines exactly the evidence against a lab leak. It's from Eddie Holmes, and was written after publication of PO when Kristian Andersen was having second thoughts:

To me there is too long a series of implausible events to suggest inadvertent escape via lab passage.

(i) The Shi group sequence and publish their bat viruses all the time, but none of these are the obvious progenitor of SARS-CoV-2. It seems improbable to me that the one that escaped was not one that they had sequenced already. And why do lab passage on a virus that you have not sequenced?

(ii) If there had been a lab escape then we would expect an initial outbreak at the [Wuhan lab]. Where's the evidence of the outbreak? How could this be hidden? That group were also well enough to sequence an early genome of SARS-CoV-2 and RaTG13.

(iii) What are the odds that the virus then first appears in the very place — a wildlife market — where we exactly expect a natural species jump to occur? Why not in a far more crowded place in Wuhan of which there are many.

(iv) Why would the Shi group then publish RaTG13 that would only help point the finger at them? Makes no sense.

That was in April 2020. By now, of course, there's far more evidence both in favor of a natural origin of the virus and against a lab leak. But even at the very start, the evidence against a lab leak was pretty strong.