Bryan Caplan  

Priors and the Death Penalty

Reminder on Brescia Speech Tod... What About a Travel Ban to Red...
I have long favored the legalization of drugs - and the "opioid crisis" has done nothing to change my mind.  The right to do what you want with your own body is not absolute, but it sacred nonetheless.  Since I oppose any legal punishment for consenting adults who use or sell drugs, I obviously oppose the death penalty for drug offenses.  Still, I was perplexed by Adam Minter's recent piece on the failure of this maximally punitive approach. 

Minter begins:

Unlike in the West, where addiction has long been treated as a medical condition, Asian governments have typically viewed any drug use as a criminal issue. China, for example, has focused on imprisonment and executions since the 1950s. Others followed the same path. Starting in the 1970s, countries ranging from Singapore to Vietnam created criminal codes with low thresholds for executing traffickers, dealers and users. Yet, even as the region's drug enforcement apparatus developed, so did drug addiction. By the early 1990s, 40 years after Mao's eradication campaigns, Chinese officials were forced to concede that entire villages were once again addicted to opiates arriving from Myanmar.

Rather than question their focus on harsh punishments, China and Southeast Asian nations, including Malaysia and Singapore, doubled down through the mid-2000s.
Then Minter makes a series of odd claims:

Yet, evidence that executions serve to deter drug use or crimes in Asia (or anywhere else) is virtually nonexistent. For example, the Chinese government reported that the number of registered Chinese drug addicts increased 6.8 percent in 2016, to 2.51 million (the government concedes such numbers are massive undercounts), up from 901,000 in 2001. The growth has been fueled by new synthetic drugs like methamphetamine, seizures of which surged 106 percent in 2016.

In Singapore, 3,089 "drug abusers" were arrested in 2016 (40 percent of whom were identified as new abusers), compared with 1,127 arrested in 2006. In Malaysia, the number of newly registered drug addicts rose from 10,301 in 2012 to 22,923 in 2016. And in Indonesia, which has unapologetically executed local and foreign drug traffickers in recent years, the number of addicts increased from 3.6 million in 2011 to 5.9 million in 2015, according to the government.

What's so odd here?

First, Minter totally ignores common sense.  In the absence of any specific evidence, we should have extremely high confidence that credibly threatening death for X would sharply reduce X.  Why?  Because almost everyone has a strong desire to stay alive.  If you think that alcohol taxes significantly cut alcohol consumption, how can you not expect the death penalty for drugs to significantly cut drug use?  Yes, it's an empirical question.  But if you don't start with a strong Bayesian prior in favor of the efficacy of the death penalty, you lack good judgment.

Second, the evidence Minter cites is utterly irrelevant.  Suppose the death penalty cut drug use by 90% at every point in time.  We could easily still see enormous shifts in drug use.  Both demand and supply move in response to many factors besides drug policy.  Indeed, you could use exactly the same specious reasoning to argue that treatment programs don't work: "If treatment works, I dare you to explain the doubling of addiction rates."  The reply is straightforward: "If we abolished treatment programs, addiction rates could grow even more."

Minter then makes a slightly better argument:

The most compelling evidence that executions have failed as an anti-drug strategy is the fact that many Asian governments have begun to retreat from them. The trend can take modest form, such as Singapore's 2012 decision to reduce the number of drug crimes eligible for mandatory executions, or China's quiet, decade-long effort to open methadone clinics and voluntary rehabilitation facilities.

Question: If Asian governments were sharply ramping up executions, that would be "compelling evidence" that execution does deter?  Hardly.  If there's any tendency for governments to move toward more effective policies, it's weak.  Politicians often don't know what works.  Without careful social experiments, definitive answers are hard to come by.  More importantly, politicians often don't care what works.  If they seek popularity - and what leader doesn't? - they just have to pander to public opinion.  If the most effective policies horrify the public, leaders will avoid them despite their efficacy.  To quote the murderous Octavian in HBO's Rome, "Agrippa has a point. We should proceed more slowly. We do not want to appear butchers."

To repeat, I'm not advocating the death penalty for drug offenses.  In fact, I consider drug prohibition to be a heinous crime against humanity.  But in the absence of overwhelming contrary evidence, we should still believe that the death penalty heavily deters drug use.  And the contrary evidence that Minter presents is underwhelming indeed.

COMMENTS (13 to date)
mariorossi writes:

I think you commit very similar error you are complaining about. While it's hard to disagree with the prior that the death penalty deters drug use, I see no reason to add the "heavily".

There is no great data to estimate the size of the effect, holding such a strong prior in such situation seems arbitrary to me.

While I agree that most people value their life (and even that is not entirely universal), you seem to argue in main places that people are generally irrational. Such irrationally would lessen the deterrent effect and the net effect (while certain in direction) is very hard to judge in its magnitude.

What do you even mean by heavily? 50%? 90%? 99%?

Miguel Madeira writes:

Remember that we are talking about a population that is not afraid of dying from an overdose - these mean that their desire for drugs is already stronger than their self-preservation instinct.

John Alcorn writes:

There is a misdirected link in Prof. Caplan's blogpost: "drug prohibition to be a heinous crime against humanity" mistakenly links to the HBO series, "Rome."

James writes:

In order for any penalty to influence behavior, the people subject to that penalty need to believe that they have a high likelihood of having to face that penalty. In countries where law enforcement is sloppy and bribery is common, the likelihood of ever receiving the death penalty for drug use may be too small to matter.

A more compelling argument is that many types of drugs come with a "ruin your life and the life of everyone you care about" penalty that drug users were aware of before they started using. At least in the present day US, nearly everyone who starts using heroin or crack or meth knows the horrendous consequences that can come from using those drugs. And yet they still start.

JG writes:

"The right to do what you want with your own body is not absolute, but it sacred nonetheless."

It must be nice to be God like Bryan. When are are God, you can never be wrong. In fact, as he states above - you have the right to do wrong.

In the same post, he states that there is a universal desire to exist. What is the source of this universal desire? A transcendent and eternal God or His creature, Bryan Caplan?

Daniel Gravois writes:

JG, the source of this desire is the same as the source of all other desires of human beings, namely evolution.

Ahmed Fares writes:

"The right to do what you want with your own body is not absolute, but it sacred nonetheless."

It's not out of altruism that we're against drug use. It's because drug users eventually lose the ability to support themselves and their habit and end up turning to crime.

We're really protecting ourselves.

"Yet, evidence that executions serve to deter drug use or crimes in Asia (or anywhere else) is virtually nonexistent."

While executions may not deter drug use in general, they do deter the person being executed on account of the fact that dead people don't do drugs.

Alex writes:


"t's not out of altruism that we're against drug use. It's because drug users eventually lose the ability to support themselves and their habit and end up turning to crime"

Drug users turn to crime to support their addiction, because drugs are illegal and therefore very expensive. Alcoholics rarely turn to crime because alcohol is legal and inexpensive.


"drug prohibition to be a heinous crime against humanity" Well said!!!!

Matt C. writes:
James writes: In order for any penalty to influence behavior, the people subject to that penalty need to believe that they have a high likelihood of having to face that penalty

Gary Becker would disagree. In a Beckerian punishment model, the benefit from the activity needs to exceed the probability of being caught x the pain of the punishment. A low-probability, high pain punishment such as death can be an effective deterrent.

Biggie Smalls writes:

Refreshingly clear thinking here as usual, Bryan. I just wanted to add that this post reminded me in a few parts of this LessWrong post, which I suspect many here have already seen, but which should provide a nice lens for analysis of policy and punditry to those who haven't.

Mark Z writes:


Alcohol is not nearly as addictive as, say, heroin or many other hard drugs. Alcoholics have a much easier time holding jobs than heroin addicts. It's also arguable (I don't know much about the production process) that alcohol would inevitably be much cheaper than heroin as the latter is more expensive to produce (I expect this is true). Lastly, alcohol does cause a decent amount of crime.

Though I favor less stringent drug laws rather than more stringent, I think Ahmed makes an important point: even if drug laws aren't justified to protect people from themselves (they aren't, in my opinion; I agree with Bryan here), they could hypothetically be justified to protect people from each other.

Imagine a drug that causes half the people who take it to subsequently commit murder or rape or arson or another serious violent or property. Would outlawing it be justifiable?

Arthur Lam writes:

At least Minter’s prose is clear. This is so poorly written it’s no wonder it’s stranded on a blog. Catalan presents no evidence for his argument but ‘we should still believe’ the death penalty dyers drug use. Key word: ‘believe.’ This is undergraduate level nonsense.

GregS writes:

Good piece. Some comments above may have said this in a slightly different way: You have to compare the penalty to the cost they are already paying. Heroin users face something like a 3-4% annual mortality rate (crudely dividing the number of deaths, which the CDC estimates, by the number of users, which are estimated based on the SAMHSA survey). From the Econlib article “Risk and Safety”, a Chicago crack dealer supposedly has a 7% annual mortality (higher than a Texas death row inmate, the second highest mortality figure on that table). For a death penalty to even make a dent, the probability of capture/conviction/execution must be pretty high. Of course these figures are high; doing the same division gets me ~0.1% annual mortality for cocaine users. And I’ll bet the “crack dealer in Chicago” is some kind of outlier. Still, we’re contending with already-high mortality, indicating a very high “willingness to pay”. (You can easily deter the use of a drug with a very low mortality, but then why would we want to?)

Fundamentally, this is why prohibition doesn’t work. It may “work” in the sense of deterring some drug use, getting it down to a (large) fraction of what it would be in a legal market. But when you actually count up the costs of the penalties imposed (in this case a pretty gruesome one), it turns into a loser for society as a whole.

To the externalities point, those “external” costs are already partially internalized. If there’s a drug that has a 50% chance of turning you into a murderer (Mark Z’s example), then potential users foresee a “50% chance of facing the consequences of being a murderer”, just like they would anticipate any other cost. We have laws against murder, so anything that will make you irrationally murder someone has a built-in deterrent. Someone could argue that “potential users don’t anticipate costs this way,” but that would be an admission that legal penalties don’t work on potential drug users.


Return to top