And the obvious missing solution.

The opioids as a class have what is known as a “narrow therapeutic window,” where the “window” is the range between the median effective dose (ED50) – the dose that’s has the desired effect in half the population – and the median lethal dose (LD50). The larger the LD50/ED50 ratio (the wider the “window”) the safer the drug will be in terms of overdose risk.

For the opioids, the ratio (also called the “therapeutic index”) is typically about six, which sounds like a reasonable margin of safety until you remember that individuals differ, that individual vulnerabilities differ from occasion to occasion (especially with the presence of other drugs, notably alcohol), and that people make mistakes, especially when drugs are made and distributed illicitly rather than in pharmaceutical factories and taken by people who are not always operating at their cognitive peak. Given all that, a factor of six is an uncomfortably narrow window.

This is from an excellent analytic piece by illegal-drug analyst Mark Kleiman. The piece is titled “A primer on fentanyl(s),” May 24, 2018. HT2 Tyler Cowen.

This small margin of safety, notes Kleiman, is a major factor in the rising deaths from synthetic opioids. Kleiman writes:

The narrow therapeutic window explains why overdose death is so much more common with the opiods than with the stimulants or the benzos or alcohol. And the smaller the intended dose, the harder it is to measure out precisely. So high potency, which can be an advantage clinically (allowing less painful injections and the use of things like transdermal patches) can be a nightmare on the street. To make things even worse, neither users nor dealers have reliable ways of knowing just what’s in the white powder they’re consuming or selling: someone who injects what he thinks is the right dose of heroin, but has in fact purchased fentanyl, is likely to stop breathing. Even someone who intends to take fentanyl could die if he’s actually been given, say, 3-methylfentanil or some other high-potency analogue.

Kleiman then goes on to discuss why such a dangerous drug is so popular now. It’s the economics, and part of his explanation is what Tyler Cowen highlighted.

Kleiman is also realistic about the difficulty U.S. governments will have in getting this under control:

And I don’t see a snowball’s chance in Hell of stopping the flow. It’s possible that, with adequate urging from the U.S., the Chinese authorities might succeed in cracking down on illicit manufacture and sale. But there’s nothing magical about China. India also has skilled chemists and a huge flow of mail to the U.S. So, for that matter, does Canada. And so does the U.S.; if international sources dry up, the stuff will, once again, be made here.


So I’m tempted to reach a fairly grim conclusion: The “supply side” of current drug policy tries to prevent people from harming themselves by falling into substance use disorder (often harming others in the process) through making abusable drugs more expensive and harder to get by making laws against selling them and enforcing those laws. That becomes a less and less workable policy over time.

Kleiman does propose a partial solution:

In the meantime, all we can do – maybe – is make it easier for people who want to use heroin and not the fentanyls to tell the difference by allowing them to have test kits, and maybe make the penalties for selling much higher for fentanyls than for heroin to give dealers a disincentive. But I wouldn’t bet the farm either that we will do those things or that they would work if we did them.

And he misses the obvious solution: full legalization. If you’re someone inclined to use, which would you prefer: a test kit that you may or may not be competent at using or a powerful drug sold legally sold by a drug company with its reputation on the line?

Side comment: The one analytic mistake in his piece is this:

Falling homicide rates also reduced one major risk of drug-selling.

Co-blogger Scott Sumner says “Never reason from a price change.” Kleiman made a similar error here. Why did homicide rates fall? It’s probably not exogenous. And even if the fall were exogenous, the relevant homicide rate is not the overall rate but the one that pertains to dealing in the illegal drug market. That’s almost certainly endogenous. The question to ask is: What caused the drop in homicides in the illegal drug market? My guess is that it has to do with the other factors that Kleiman mentions in his excellent piece–things like being able to order drugs on your cell phone, for instance. Kleiman mentions this factor himself:

Thirty years ago, illicit retail drug transactions were characteristically carried out either in public locations (parks or street corners) or in dedicated drug-dealing locations (e.g., crack houses). Those locations tended to cluster heavily in low-income, high-crime urban neighborhoods where police had other priorities and neighbors were reluctant to call the police. Having to travel to such a location – risking arrest or robbery – constituted a significant barrier to illicit acquisition. Moreover, for open-air transactions, a buyer had to search for a willing seller-usually, a seller with whom he had an established connection – and that search took time (45 minutes was not uncommon) and sometimes failed entirely. Search time and risk constituted a second kind of “price” of illicit drugs, perhaps as significant (especially to new consumers) as the money price.