Ed West over at the Wrong Side of History has a good post about how, yes, putting people into prisons and keeping them there reduces crime. As near as I can tell, the actual research supports this intuitive finding pretty clearly. The general idea is that most crime is committed by a rather small number of offenders, and once those are in prison they aren’t committing crimes outside of prison, and crime goes down. Even if criminals learn how to better commit crimes while in prison they are not capable of doing so while they are in there; when released whether or not their criminal friends are still in prison or out running around and ready to lead them astray matters; crime is a young man’s game and occupation (for some at least) so many won’t reoffend if released past that point. All sound points, and it is really hard to make a good argument that putting criminals in prison longer does not reduce crime, all else equal.
My quibble is that there is something important that we don’t have to hold equal, and that if we could move that lever just a bit we could do with shorter prison sentences, which would be beneficial. No, not education or pre-k. No, not improved employment for teen boys, although I suspect that would be useful too. No, my radical idea is this:
Actually catch criminals on a regular basis, and upon conviction, put them in jail.
Strangely, that is very far from what we do in the US1. Bryan Caplan notes that in the US approximately “3-5% of rapes, robberies and aggravated assaults- and less than 1% of property crimes- lead to jail time.”2 We are very, very bad at catching and convicting criminals. Partially this is because those listed crimes are committed relatively randomly, as opposed to murders which are almost always committed by someone known to the victim. Yet a percentage that low suggests that we are also just not trying very hard, as quite a few e.g. robberies are committed by people known to the victim and the victim knows who did it.3
The usual response to this (not Ed West’s so far as I know, I hasten to add!) is “Relax, even if catching and punishing the criminal is unlikely, we can just jack up the punishment so that the expected value of the crime is still negative.” This is known as optimal punishment theory, and in short mathematical form is the expected punishment value of a crime is the probability of getting punished multiplied by the punishment. If that punishment outweighs the value of committing the crime, you are deterred and don’t commit the crime. So a 1% chance of getting caught and suffering a $100 dollar fine is the same as a 100% chance of getting caught and suffering a $1 fine.
This theory is very popular, and probably vaguely correct, but only for large and similar probabilities of getting punished. Much like the “Rule of 70,” it breaks down pretty badly outside of an optimal range.
Large probabilities of getting caught are necessary because people are very bad at probabilities, and criminals more so. As evidence of the former I present to you: lottery tickets. Throw in the usual slew of fear of flying vs driving, base rate estimation errors, what “random” results even look like, or any of a thousand different things, and even highly educated humans can’t reliably work out probabilities reliably. Just look at how many doctors can’t do the standard “If a patient tests positive for a disease with a 1% rate in the population, using a test with a 5% false positive rate, what is the probability the patient actually has the disease?” It isn’t pretty.4
My model of how people deal with probabilities in their mind is something like this: >90% is certain to happen all the time, 50-89% will happen, <50% won’t happen, <10% almost never happens, <1% I don’t even know what that means. Yes, I am aware that isn’t really transitive, and that’s part of the point: most people have a really rough and ready sense of probabilities based on some half remembered lessons in school involving flipping coins or rolling dice, combined with rough estimation based on availability of information, with a very strong weight on availability of high magnitude outcomes. If people see lots of stories about shark attacks they revise their estimates of how likely you are to be attacked by a shark up, and this will only be somewhat corrected by showing them actual statistics. A strangely large number of people seem to have very skewed estimates of the probably of dying if you contract COVID, for much the same reason.
So even if people know the probabilities of getting punished change, say 75% to 50%, I don’t think they will grasp that the punishment going from 10 years in prison to 15 years keeps the expected value the same.
But that argues against my point, doesn’t it? They under estimate the probability change relative to the size of the punishment, and that makes them MORE deterred rather than less. True, but we forget the availability aspect. As the probability of punishment goes down, so does the availability of stories about people they know going to jail, while the availability of stories about people committing crimes and getting away with it goes up. If potential criminals are overly weighing the availability, and Dave the successful criminal is around bragging about how he got away with it while Steve went to prison and we don’t hear from him for a few years after that initial incident, we are going to see potential criminals under value the expected punishment. The effect should also be non-linear, drastically increasing as the probability of getting caught goes down. Even if you hang people for stealing car radios, if only 1% get punished the potential criminal is likely to not know a single person who got hanged but know lots of people who never got punished, and so assume it will not happen to him.
Oh, I mentioned that criminals have even greater problems with probability? Yea, they tend to wildly overestimate their chances of success (a trait shared by entrepreneurs, interestingly) and so are much more likely to think “Yea, Steve got caught, but he’s a dumb ass. I won’t get caught.”
So the introduction of the availability bias in reasoning means that the “calculated” expected value isn’t linear, and so adding on a bit more punishment to make up for a bit less probability only approximates the correct amount for small changes in probability. As the probabilities get small the increase in punishment needed likely goes far past what we are able to inflict, even if we had the stomach for it.
All that to say, if we were to get the rate of actual punishment for crime up to, say, double digits, maybe even a lofty 25% incarceration rate, we probably could reduce the sentences quite a bit and get equal or great deterrence. In the US we have gone far, far to long the other way, presumably because it looks a lot cheaper on various decision makers’ balance sheets to not bother with sub-murder level crimes and just throw those dumb enough to get caught5 into prison longer. If our police department actually got enough funding to hire sufficient police to investigate crimes, and had the motivation to spend the money on actually doing so instead of buying tanks or something, we probably could get that double digit punishment probability. I mean, I don’t want to get nuts; my advice to those in charge of policy is “Keep your feet in the gutter, but never stop reaching for that curb!” A >10% non-murder punishment rate would probably do more for crime prevention than any number of…
…well whatever else they are doing to reduce crime…
I guess they are just trying to reduce crime by not stopping it these days? Good lord, but the crime statistics are depressing.
Ed West is in the UK to my understanding, but they hardly seem better at the whole “catch criminals” thing from all accounts I have read.
According to Bryan Caplan’s “The Case Against Education,” citing Lochner and Moretti’s 2004 “The Effect of Education on Crime: Evidence from Prison Inmates, Arrests and Self-Reports.” I recall Caplan mentioning that something around 25% of murderers go to trial, but only as an offhand point in class discussion.
See “Crime as Social Control” by D. Black - American sociological review, 1983
Not that I am great at this either. I certainly can’t do it in my head, and if I need to I look up how to do it properly to check myself. Not many people realize they really need to double check how to do it…
Note that if it really is the dumber than average criminals who get caught, deterrence is even less effective because smarter criminals rightly estimate that their chances of getting caught are smaller still.
> “3-5% of rapes, robberies and aggravated assaults- and less than 1% of property crimes- lead to jail time.”
Gotta be careful with stats like that. The causes are different.
The low rate of property crime clearance is indeed very sad and reflective of the priorities of the police, I think, perhaps also that robbers got better at robbing over time. I recall my parents house was broken into (in the UK) once when I was a kid and the police were useless. They don't usually have much to go on unless they find where the goods are sold.
The low rate of rape clearance is mostly related to the fact that the rate of false accusations and rescinded allegations is so high. I looked into this once some years ago and the average cop will estimate in anonymous interviews that ~50% of all allegations are false. Many of these never make it to court because the accuser admits it before it gets that far. Additionally, "stranger rape" is quite rare. It's far more common that the two people know each other and may well have been in a consensual relationship. So it's trying to litigate what happened between closed doors to the level of "beyond all reasonable doubt", which is inherently difficult for the legal system to handle.
Unfortunately feminist politicians, activists and civil servants usually don't care about any of that. In their minds women never lie, ever, so the rape jail rate should be 100%. These people have a nasty habit of making their own delusional beliefs come true by changing the law any time men find ways to defend themselves against false accusations, for example, in the UK recently there was a run of false accusations that were revealed via phone evidence. The response of the political class was not "wow, men have a serious problem with false rape accusations against them, we'd better fix that", it was, "let's change the law so phone evidence isn't admissible". It's this sort of thing that reveals the lie of feminism - it's a woman's world actually. Women are willing to organize to take control of men in ways modern men would never even dream of doing in return.