More incarceration is not the answer
We need to catch serious criminals, not dole out harsher punishments
One of the more noteworthy political developments of the past 15 years was the successful development of a left-right elite coalition in favor of reducing the American prison population. Donald Trump signing the First Step Act was something of a signature achievement of that movement, but the actual consequences of First Step were modest — in part because the federal government is a decidedly secondary player in crime and punishment.
But the conservative group Right on Crime has had some real state-level policy successes in red states, where de-incarceration is a useful way of keeping taxes low, and of course progressive states in a more straightforward way have been tilting against mass incarceration.
The background context for all that, however, was generally low and falling crime rates. There was a murder spike in 2015-2016, but the murder rate then fell again in 2017 and 2018, so when Tom Cotton and a handful of other right-wingers tried to derail First Step nobody really cared. But the huge murder increase in 2020 and Joe Biden’s accession to the White House give Cotton another chance to try to reboot pro-incarceration politics.
As an anti-anti-police person, I think this is bad.
One of the big tragedies of mass incarceration in the United States is that while high levels of imprisonment are clearly a political and mechanical response to high levels of crime, they are not a very effective response. The basic reason is that soaring prison headcounts are mostly driven by long prison sentences, but the deterrent effect of prison is much more driven by the odds of getting caught. In other words, an 80% chance of needing to serve a two-year sentence and a 20% chance of needing to serve an eight-year sentence have the same expected value. But the former is much more deterring. And since it deters more effectively, it will lead to fewer crimes being committed and less incarceration. And that’s what we actually need — less crime, not harsher punishment for the minority of criminals who get caught.
Gary Becker’s big mistake
The late Gary Becker was a major pioneer in the now-common practice of applying ideas and analytic modes from the field of economics to social issues that are a bit far afield from the core business of “the economy.”
And in one such paper, the 1974 essay “Crime and Punishment, an Economic Approach,” he argued that catching criminals is basically a waste of resources. Having lots of police officers is expensive. Finding and training skilled detectives who can actually solve crimes and crack cases is hard. And if you do catch criminals then you need to do the work of prosecuting them one-by-one. Instead of doing this labor-intensive work of trying to catch lots of criminals, you could just accept that most people will get away with crime most of the time but decrease the expected value of criminal acts by punishing the people who do get caught very harshly.
An increased probability of conviction obviously absorbs public and private resources in the form of more policemen, judges, juries, and so forth. Consequently, a “compensated” reduction in this probability obviously reduces expenditures on combating crime, and, since the expected punishment is unchanged, there is no “obvious” offsetting increase in either the amount of damages or the cost of punishments. The result can easily be continuous political pressure to keep police and other expenditures relatively low and to compensate by meting out strong punishments to those convicted.
Alex Tabarrok has called this Gary Becker’s biggest mistake. I’m not sure I know enough Becker to know if I agree with that, but it’s definitely American crime control politics’ biggest mistake.
For the past 40 years, the policy dialogue has tended to be polarized between leftists who deny that law enforcement and punishment have any useful role to play in reducing crime and the Becker/Cotton idea that extremely severe punishment of offenders is the key to deterring crime. What we need is a return to an older — but more empirically grounded — approach. This approach recognizes that catching criminals is in fact very important. Of course, once you catch criminals you need to punish them. But a sound approach to crime recognizes that the people committing the bulk of the crime are not great long-term planners. Threatening them with consequences that take place in the distant future doesn’t make sense. You want to catch people, punish them, and then help them get their life together combined with creating a situation where they believe breaking the rules is likely to get them caught again.
A sidebar about white-collar crime
We don’t discuss “white-collar” financial crimes that much in the United States, except occasionally from the left as a kind of whataboutist argument.
But to me, one takeaway from the Trump era is that we should take the idea of a “tough on crime” approach and seriously apply them to rich-people crimes. Natasha Sarin is the author of an important line of research that argues that increasing the IRS enforcement budget can turn a huge profit for the government because the “tax gap” currently runs to billions of dollars. She got hired to work in the Treasury Department in mid-March, which I think is great.
That said, I think that a narrow focus on the fiscal benefits of enforcement is a mistake.
If you look at Trump’s career, a prominent aspect of it is that he was constantly getting caught breaking one rule or another. But the normal way regulatory agencies handle these things is they assess a modest fine or they reach a settlement agreement that involves paying money with no admission of guilt. The Obama administration’s handling of the fallout of the financial crisis was full of this kind of thing — cash settlements in lieu of prosecutions.
From the government’s fiscal viewpoint, this makes a lot of sense. A settlement is cheaper than a prosecution, has a higher probability of success, and it puts dollars into play.
The problem is that the reason corporate scofflaws are eager to agree to these deals is that even though they sometimes involve eye-popping headline sums of money, they by definition are not that crippling to the enterprise. That means the deterrent effect is very weak. And it’s precisely these kinds of corporate criminals who most closely resemble Becker’s schematic view of the criminal as a rational actor. The people who make these decisions are successful, middle-aged people who are calmly reflecting on their situations and who have a fiduciary obligation to cut legal corners if doing so has positive expected value once you account for the odds of getting caught and the severity of the fines.
Bringing a much more Beckerian approach to this category of crime would be great. But the kinds of criminals who unleash violent mayhem on the streets aren’t like that.
Most criminals are impulsive, disordered people
The late Mark Kleiman, whose work is the inspiration for this item, noted that even at the time Becker wrote, the cost-benefit calculus of blue-collar crime was bad. Then Becker-inspired policymakers made it worse without doing much to reduce crime:
Even at the nadir of incarceration-to-offense ratios in the late 1970s, blue-collar crime has never paid. For every hour spent behind bars, the average burglar back then “earned” something less than $2; the wages of sin have always been well below the legal minimum. The subsequent increase in burglary-related incarceration and decrease in burglaries have driven wages for serving burglary time to the pennies-per-hour range. The surprise for those who believe that offenders are rational economic actors is how small the decrease in burglary has been even as the terms of trade worsened six-fold. (Crack-dealing wages have also fallen sharply, albeit from a higher level.)
Even these kinds of analyses probably understate what a bad deal crime is because especially in situations where law enforcement is ineffective, various groups of criminals have a marked tendency to commit acts of violence against each other. Jill Leovy in “Ghettoside” describes a situation where low clearance rates and low levels of community cooperation with the police create a cycle of formal impunity for violent criminals, but they are constantly at risk from each other. Violence spirals out of control thanks to cycles of retaliation that are very dangerous for everyone involved.
Murderers are mostly under 30 and overwhelmingly male, which is part of a much larger social pattern of young men engaging in much riskier behaviors than other people.
People are committing violent crimes, as Kleiman writes, because they are “present-oriented and impulsive, with deficient capacities for shaping their current behavior in light of their future goals, and with poor judgment about their actual odds of getting caught: all characteristics, as noted above, likely to be produced by growing up in high-crime neighborhoods.”
Prison and jail inmates suffer from mental health problems at five to seven times the rate of the general population.
Cross-national surveys show that criminals have weaker self-control than average citizens in a wide range of contexts.
Drug and alcohol use disorders are much more common among inmates than non-inmates.
It’s fun to watch TV shows about cops chasing hyper-competent supercriminals (I’m currently watching both Lupin and The Line of Duty), but in the real world, criminals seem to have decidedly below-average IQ. We’re talking about younger men with poor self-control, lots of mental health and addiction problems, and low levels of intelligence. Elongating prison sentences tries to deter them with threats of consequences that occur only in the distant future and it doesn’t work very well.
Severe punishments don’t deter very well
The appeal of making sentences longer is that it’s mechanically easy. All the legislature needs to do is tweak the sentencing guidelines and appropriate the funds for prison construction, and lo and behold you are getting tougher on crime. You might aspire to improve clearance rates for serious offenses, but your legislative initiative might fail. By contrast, if you want people to serve longer sentences you’ll be able to make it work.
But do long sentences deter crime? I’m sure that at some margin they do — like if the sentence for murder was you spend a week in jail, then making the punishment more severe would be a good idea.
On the margin that American policy actually operates on, the answer seems to be no. Here’s a National Institute of Justice review if you’re interested, or a writeup from the Sentencing Project for an advocate’s view. The most famous meta-analysis paper that people point to is Daniel Nagin’s “Deterrence in the Twenty-First Century: A Review of the Evidence.” David Roodman’s roundup of the research for the Open Philanthropy Project is even better and says the same thing.
Again, I would urge against adopting a super-counterintuitive reading of this evidence. It’s not that “prison doesn’t deter crime;” it’s that the marginal impact of an additional year of prison time is a weak deterrent. Very few people who are undeterred by an X% chance of being caught and serving five years in prison will be deterred by an X% chance of a 10-year sentence.
A separate issue is whether prison reduces crime by “incapacitating” criminals (i.e., putting them in prison rather than out on the streets). The answer here is pretty clearly yes. But the cautionary note at the margin is again that older people do not commit a lot of crimes. Here’s a chart from Marc Mauer and Ashley Nellis looking at robbery arrests.
This is what motivated German Lopez to write a smart case for capping prison sentences at 20 years.
But regardless of your position on that, the point is that you get diminishing marginal returns from incapacitating a criminal one extra year because he’s getting older over time. By contrast, the cost — both the direct financial cost, the economic cost of not having the prisoner out and about, and the cruelty cost of keeping people locked up — just scales linearly. Indeed in some ways, the 35th year is costlier than the 15th because you’re increasingly dealing with health problems and other issues as your prisoners get older.
Longer sentences tie up a lot of resources to little benefit.
But it’s a difficult political conversation to have because the people serving long sentences normally have, in fact, committed serious crimes. Advocates sometimes talk as if mass incarceration is all teenage joyriders and the occasional pot dealer. But criminologist John Pfaff, who’s an advocate for less incarceration but also a very rigorous thinker, keeps reminding us this isn’t true. There may be a good number of people in jail on petty charges on any given evening, but the prison population is full of people who’ve committed serious violent crimes.
And this in turn is why I think controlling crime is so central to continuing the progress that’s been made toward a more humane system of punishment. Long prison sentences are not an effective way to fight crime. But if the murder rate is rising year after year, nobody is going to be so politically suicidal as to reduce sentences for violent criminals. You need to actually show there’s a better way to keep people safe.
Effective crime control — better policing
The point that Kleiman and Nagin and Tabarrok and all the smart researchers make about this is that if you think of the people who commit crimes as short-sighted, impulsive, and poor evaluators of risk, the really important thing is swift and certain sanctions.
In other words, you want people to think that if they break the rules they will get caught.
This is why having a lot of police officers walking around on the streets reduces crime, and the reduction does not seem to depend on them taking any particular aggressive measures. People don’t do crime when there’s a cop on the block. And if the cops cruise around frequently, people worry they might be around the corner at any moment. “Hot spot” policing where you just literally make sure the police are frequently present in the places where the crime is happening seems to work almost surprisingly well.
Then the other thing you can do is try to catch people who commit serious crimes. As this chart from the Murder Accountability Project shows, it is disturbingly easy to get away with killing someone.
And of course murder is not the only crime! The odds of a non-fatal assault or robbery being solved are even lower since murders are the top investigative priority.
Unfortunately, to Becker’s point, it is more challenging to solve crimes than it is to simply double the sentence length for the criminals you have already caught.
If you want to make this point in a left-wing way that says mean things about police officers, here’s Ryan Cooper at The Week:
The logic of the police unions is that if you punish murderers, there will be fewer murders. And it turns out that a great many criminals are escaping with impunity — it's just the fault of the police. On the raw numbers, any Philly murderer has a better-than-even chance of evading the cops. In fact, it's worse than that. Typically something like a third of murders basically solve themselves because the culprit is found at the scene, or there is very obvious evidence. Philly cops are doing barely better than that — meaning that if you kill someone and take any steps at all to cover your tracks, you're all but guaranteed to get away with it. Krasner is more than willing to prosecute violent offenders, but Philly cops are too lazy or incompetent to catch most of them.
If you want to make this point in a right-wing way that says defunding the police is bad, here’s a study showing that when you dedicate as many detective hours to the typical non-fatal shooting as a fatal shooting gets, then the clearance rate for non-fatal shootings goes up.
My view is that it’s unfortunate that so much of politics consists of this kind of position-taking and mood affiliation since I think everyone is saying the same thing under the hood. Bureaucratic institutions in general tend to be a bit lazy and, more to the point, they are far more likely to respond to metrics they are actually being evaluated on. If you tally up raw arrest counts, you will get a lot of low-level arrests. If you measure call response time, a lot of person-hours will be deployed on responding rapidly to calls. If you want to solve more crimes, you need to make a point of investing in and rewarding that activity.
The higher synthesis I want everyone to aim for is simply that catching more criminals is not the same as “more incarceration.” An effective system of law and order with lots of police visibility and high clearance rates would lead to relatively low levels of incarceration (because people aren’t committing crimes) and needn’t rely on especially severe sentences (because the odds of apprehension are high).
Punishment without cruelty
I doubt it will be politically viable to do further de-incarceration until we can find a way to make violent crime go back down again.
But one of Kleiman’s big points that’s stuck with me was that we could achieve most of the incapacitation impact of long prison sentences in much more humane ways by letting people out of prison sooner and then subjecting them to heightened surveillance. For example, you could be released into the community but with an ankle monitor. The monitor doesn’t need to be used to keep you under house arrest. But it could be a low-cost way of intensively tracking your compliance with parole terms (are you showing up for work) and also of ensuring that your odds of being caught if you commit new crimes are very high because there will be clear evidence that you were present at the scene.
This kind of monitoring can sound dystopian. But it’s obviously much more humane to let a person out of prison and make him wear an ankle monitor than to say we need to keep him locked up for years and years because we’re worried he’ll commit new crimes.
Another Kleiman idea is mandatory sobriety for people whose offenses were linked to drug abuse. You need to undergo frequent testing, and if you flunk you get locked up for one to two days. The idea is that this is better than the standard parole approach of infrequently delivering severe sanctions. This originated in a Hawaii program called HOPE that seemed very successful. Then some subsequent studies suggested maybe it couldn’t be scaled. I think the most recent look at it from Jennifer Doleac suggests these programs probably are in fact good.
And of course if you had fewer people in prison, the prisons themselves could be less cruel and more focused on safely incapacitating dangerous people while delivering treatment and vocational training designed to eventually keep people out of prison.
At the end of the day, being locked up in a relatively humane prison for “only” a few years is still pretty bad. Even a very short-sighted person is going to try to avoid that outcome. What you want to do is focus your policy efforts on making that person feel that if he breaks the law, the odds of being punished — even slightly — are high. But this is very contrary to how the system works.
Federal officials have worked diligently to identify and arrest tons of people who participated in the January 6 mob, bringing charges against them even if the charges aren’t so serious. Politico has reported this as a situation that “could embarrass the Biden administration” because these cases have “clogged D.C.’s federal court,” but in most cases they won’t generate harsh penalties. To me, this shouldn’t embarrass anyone. This is how the system should work. Everyone who stormed the Capitol broke the rules. It’s important to catch them all and punish them all so in the future, people don’t think participating in a little light rioting is a fun time (and yes, I said this about summertime looting). The point is made and the sanctions don’t need to be harsh. But the normal functioning of the system rebels against this. Actually apprehending and prosecuting people is annoyingly labor-intensive, which is how Becker’s idea initially got off the ground.
But it doesn’t work, and as we confront a rising tide of crime, we owe it to ourselves to do better in the future.
In DC, it was reported that one of the children who murdered the Uber driver had recently also been caught committing another carjacking. I haven't done the statistical analysis on this, but that's something I see a fair amount in violent crimes here - that the perpetrator has recently been caught for illegal gun possession, assault, etc., and they're out on the streets again quickly (in some cases, immediately and while awaiting trial, in some cases after getting a deal involving no or very little jail time.) I don't know to what extent that's a national or merely local phenomenon. That seems bad. Like, extremely bad, both because murder is terrible but also because the government seems to me to be more responsible for a murder committed by someone they just saw for another violent crime and decided to let go. As for doing monitoring instead, that requires a level of competence that may not exist.
I have been intrigued by several recent popular press articles about detectives settling murder cold-cases via DNA, sometimes through reconstructing family trees in order to reduce the number of leads to a manageable size.
I think that's a great way to clear cases. Here's an even better one:
Put a few billion dollars into testing and banking every rape kit in the country. There are thousands of them lying around, untested. That right there is a fast clearance-method, lying to hand.
"But rapes are hard to prosecute and convict!" you say. And so they are, which is a topic for another post (it's a toxic stew of misogyny and patriarchy, and the fault lies with jurors as much as prosecutors).
But serial rapes are relatively easy to get convictions on. And I am not alone in suspecting that serial rape is fairly common, and would be revealed to be so by a national inventory of the DNA from rape kits.
This would be an excellent use of time and money by the Biden administration. Making people less likely to get raped makes them less afraid to work and travel freely. Looks like smart infrastructure spending to me.