The American Interest
Policy, Politics & Culture
The Outpatient Prison

How to lower both the prison population and crime—at the same time.

Published on March 1, 2010

The United States has both the highest homicide rate and the highest incarceration rate in the developed world; no country comes close on either score. One reason: We do a lousy job of managing offenders when they aren’t in prison or jail.

This is a solvable problem; we already know how to do a lot better than we’re currently doing. But to solve any public policy problem takes more than technical knowledge; it takes political will—which tends to be in short supply when a given issue cruises along below the public’s radar, as the crime issue now does—and skilled and energetic public management, which is always scarce.

While it has been said that “Worthwhile Canadian Initiative” is the most boring headline ever written, “Improving Community Corrections” must run a close second. Nonetheless, there’s no way to escape our current crime-incarceration-crime cycle without fixing the three-part mechanism that supervises offenders when they’re not behind bars: probation (for those not sent to prison); parole (also known as “supervised release”, for those let out of prison before the end of their sentences); and bail or release on recognizance (for those awaiting trial).

This is not a small matter. There are more than twice as many offenders under community supervision (4.3 million on probation, 700,000 on parole) as in prison (1.7 million) and jail (700,000) combined. No one seems to have even estimated the number on pretrial release; the population isn’t tracked because pretrial releasees are almost entirely unsupervised, but they must number in the hundreds of thousands at any one time. Offenders under some form of criminal justice jurisdiction account for about 40 percent of new felony arrests. More people are sent to prison each year for violating probation or parole conditions than as a result of conviction for new crimes. Of those released from prison, about two-thirds cycle back within three years; the failure rate for felony probation is lower, but still very bad: about 40 percent.

If we were better at supervising offenders when we’re not paying for their room and board, fewer of them would wind up back inside, and they would commit fewer crimes when on the outside. If probation and parole worked better, judges, parole boards and legislators might also be more willing to let some non-violent prisoners out of prison, or not send them in at all. And that would be good, because prison time tends to beget more prison time. The propensity for recidivism resides not only in the personalities of criminals, but also, especially for younger people, in the experience of being in prison.

The current community corrections system fails because it ignores the elementary principles of using punishment to change behavior. These principles are as old as Cesare Beccaria’s 18th-century work on criminology, and as new as the latest research on behavioral economics. They have been demonstrated in psychology labs—on subjects ranging from pigeons to sophomores—and are well known, in a practical sense, to anyone who has ever successfully raised a child, trained a puppy, or managed a team or an office. The key insight is that the deterrent impact of a punishment depends only weakly on its severity, but strongly on its swiftness and certainty. It also depends—though this is harder to study experimentally—on the perceived fairness of the punishment in both substance and process.

But instead of quick and consistent enforcement using the minimum effective dose of punishment, the American community-corrections system reproduces the basic flaw in the larger criminal justice system: It tries to substitute severity for everything else, leading to an approach that might be called “stochastic draconianism.” Even when a probation or parole violation is detected—and most aren’t—it is more likely than not to lead to nothing more painful than a rebuke from a probation officer or a parole agent, coupled with a vague threat that continued violations will eventually result in revocation of community-corrections status and a spell behind the walls.

That threat, though vague, isn’t a bluff, as the depressing rates of revocation demonstrate. The problem is the offender never knows which “last chance” is actually the last “last chance.” Too often, he keeps ignoring the rules until, inevitably but nevertheless unexpectedly, the boom is lowered.

Randomness focuses the offender’s attention outward, not inward. Whether he gets punished or not seems to depend less on what he does than on the apparently random choices of probation officers, parole agents and judges. If you wanted to design a system to maximize the number of revocations, you could hardly do a better job.

Punishment for violations is random not because officials act on whim—though it may look that way to offenders—but due to the sheer volume of violators. A typical big-city probation officer tries to supervise somewhere between a hundred and two hundred felony probationers. (Misdemeanants get even less attention; parolees get more.) If probationers are required to come in every month for a meeting with the probation officer and a drug test, about three in ten either won’t show up or will come in and test “dirty”, either of which is a violation of probation rules. No probation officer has the time to “write up” that many violations for action by the courts, and if all those reports did get written the judges wouldn’t be able to hold all those hearings. Therefore, most violations result in warnings rather than any actual sanction. But of course the failure to sanction helps maintain the high violation rate. It’s a vicious circle, a social trap for offenders and officials alike, with no obvious way out.

States as Laboratories

Judge Steven Alm in Hawaii found the non-obvious way out. Frustrated by seeing probation-revocation motions reporting seven or ten or a dozen violations, he demanded that the probation department start reporting the first violation instead. When the probation officers explained the arithmetic—they couldn’t possibly do what the judge wanted do, and if they did he couldn’t possibly hear all the resulting motions—he proposed a compromise: that the probation officers make up a list of probationers who had been so non-compliant that the next violation would result in a revocation motion.

Judge Alm then took that list of (as it turned out) 35 probationers—most of them chronic methamphetamine users—and called them all in for what he called a “warning hearing.” At that hearing, he read the assembled scofflaws the riot act. He told them that he and everyone else in the system wanted them to succeed on probation, but that they were flagrantly not keeping up their end of the bargain. And he promised them that, from then on, each detected violation of the rules would result in immediate jail time. He called the new system Hawaii Opportunity Probation with Enforcement: HOPE. (A proposal for Federal grants to spread the idea, offered by Reps. Adam Schiff (D-CA) and Ted Poe (R-TX) maintains the acronym by substituting “Honest” for “Hawaii.”)

Judge Alm then braced himself for a flood of violation reports. But the dog did not bark in the night. The warning was credible enough—Alm had done his homework, lining up not only the probation department but also the police to serve bench warrants on probationers who failed to show up for testing—that only a handful of the 35 decided to risk the judge’s wrath. Indeed, over the course of the next year, fewer than half of them ever earned a jail stay, even though the judge substituted random testing, with each probationer calling in every day to find out whether his number had come up, for the once-a-month scheduled testing that applied to normal probationers. Of those who violated once and were punished for it, only half violated a second time. The HOPE group, which started out behaving much worse than other probationers, wound up behaving much better, with half as many arrests, a third as many probation revocations and prison terms for new offenses, and one-fifth as many missed or “dirty” drug tests.

As a result, HOPE probationers, who were selected for their bad behavior and who faced a hearing and a sanction for each violation, didn’t rack up any more hearing time or jail time than offenders on routine probation. The hearings themselves were short, because, compared to a probation-revocation hearing covering many violations going back a year or more, the issue was simple (“Did you test positive for methamphetamine on Tuesday?”) and the stakes were days rather than months behind bars. Each HOPE probationer uses up only about twenty courtroom minutes per year, which means that a single judge could handle up to 3,000 cases at a time.

Thus all the improvements in probationer behavior—more compliance with the rules, less drug use, less crime—were achieved without putting any more burden on the court or the jail, and while reducing rather than increasing both total days behind bars and total cost. The system had broken out of the social trap, replacing the old equilibrium of sporadic sanctioning and high violation rates with a new equilibrium of consistent sanctioning and low violation rates.

The formula is straightforward in principle, though putting it into practice demands managerial skill: Concentration on a limited number of offenders makes it possible to achieve high certainty of sanctions and deliver those sanctions swiftly; swiftness and certainty, following a clear warning, quickly reduces the rate of violations; and the low violation rate makes the process so economical that the group of offenders under concentrated attention can grow without requiring the addition of new probation officers or courtrooms. (HOPE is more expensive than routine probation, but the difference is in additional drug-testing capacity and in increased use of high-intensity residential or daily outpatient drug treatment, and that cost is paid back several times over in reduced prison spending.)1

To convert Judge Alm’s proven innovation into a revolution, the program needs to grow in four dimensions: geography, type of supervision, range of behavior monitored and age-range covered. Geographically, we need to find out whether other jurisdictions can match Hawaii’s performance. The question is not so much whether offenders elsewhere will respond as Hawaii’s offenders did (they will), but whether officials elsewhere can re-create the swiftness and certainty of Judge Alm’s sanctions. We need to extend the new practice from probation to parole and pre-trial release.2 And, having demonstrated that it’s possible to manage as difficult a behavior change as desistance from methamphetamine as long as you can monitor the behavior and sanction non-compliance, we need to consider what other aspects of offenders’ behavior are both easy to monitor and closely linked to recidivism. Then that same system needs to be extended to the juvenile offenders who account for a big slice of current crime and an even bigger slice of future crime.

The obvious candidate to help make all this happen is position-monitoring. Monitoring aided by global positioning systems (GPS) is already in use within the criminal justice system, mostly for enforcing home confinement on sex offenders. That process, which demands immediate response when the subject strays from his assigned location, costs thousands of dollars per year, though that is still far less expensive than incarceration. Something far less elaborate would be more than adequate for most offenders under community supervision. For less than five dollars per day, we could equip an offender with a GPS monitor on a tamper-evident anklet. So equipped, the offender would find it very hard to commit new crimes without getting caught, since his location could be compared automatically with the times and places of crimes reported to the 911 system. The anklet would also make him unwelcome as a participant in group crimes or gang activity.

But that’s just a start. If a probationer or parolee is wearing an anklet, the probation officer or parole agent can know not just whether he was at a crime scene, but whether he was where he was supposed to be at any time during the day. Is he at work during working hours? At his drug treatment or anger-management class when scheduled? Out of the jurisdiction without permission? Near the street drug market, gang hangout or victim’s address?

Once position monitoring is established, then position restriction can be used as a control, or as a sanction. Why shouldn’t a felony probation sentence start out with three months of a nine p.m. curfew? Or how about a parole term? We know that the first few days and weeks are the most dangerous in terms of going back to prison. Could we cut down on that risk by keeping parolees off the streets late at night?

For those offenders who will wear the GPS device rather than attempting to take it off, position monitoring could provide a large share of the incapacitation of a prison cell at a trivial fraction of the cost. Combine that with drug testing, and you have what might be called “outpatient incarceration.”

Moreover, by linking a relaxed curfew to finding and keeping a job, the authorities could enhance offenders’ motivation to anchor themselves in the licit economy. As a bonus, the combination of position monitoring and drug testing—if communicated to employers by probation and parole agencies—could greatly increase the employability of ex-offenders, and thereby greatly increase their chances of remaining ex-offenders rather than returning to crime. If a job developer working for the probation or parole department could promise employers certified drug-free employees guaranteed to show up for work every day, the current employer bias against ex-convicts might diminish rapidly.

There’s no way to guess how big a reduction in probationer and parolee crime rates this system could accomplish. But if drug testing with HOPE sanctioning reduced new arrests among HOPE probationers by 50 percent, then a 75 percent reduction doesn’t seem out of reach for a system that added position monitoring. That would allow us to continue to push crime rates down while drastically decreasing the prison population.

Getting such a program started the right way takes time; it needs to start small, work the bugs out of the system, and establish the credibility of its sanctions before being taken to scale. It’s been five years since Judge Alm started HOPE, which currently handles about one out of five felony probationers on Oahu and is planned to handle about forty percent by the end of this year. Replications on the mainland are just starting, and it will take more time to get those programs to speed and their results analyzed. But a goal of covering three-quarters of the nation’s probationers and parolees with HOPE-style programs within ten years doesn’t seem out of reach.

Some of those now in prison and jail are worth keeping there to prevent the number and severity of the crimes they would otherwise commit, but many would not be worth the cost of incarceration. If we can make community corrections into outpatient incarceration, that ought to influence judges, parole boards and legislatures to cut back on the use of bricks-and-mortar prisons. Ideally, there would be only three kinds of people in prison:

    Those who had done something so appalling that the community’s sense of justice demands it: the Unabombers, Charles Mansons and Bernie Madoffs of the world for the long haul, and some less-serious offenders for shorter terms.
    Those whose crimes are so numerous or severe that the risk of release, even under tight supervision, is too great to bear. Some sex offenders and violent criminals, but few property-only offenders, will fit this category.
    Those who, instead of complying with outpatient incarceration, take off the anklet and go on the lam. How many such cases would there be? Again, there’s no way to know without trying, but the abscond rate on HOPE has been less than 10 percent.

If we can make this work—a big “if”—we ought to be able to cut the crime rate and the incarceration rate in half by the end of that ten-year period. That would still give us more than twice as many prisoners per capita as any country to which we’d like to compare ourselves, so the eventual rate might be even lower, but cutting the problem in half wouldn’t be a bad start.

The Kids Are Alright

Everything said thus far about the general prison population applies with equal or more force to juvenile offenders. The special challenge for juveniles is finding an appropriate sanction.

For excellent reasons, we are much less willing to lock up juveniles than we are to lock up adults. The last thing we want to do to a young offender is to take him out of school and mix him in with other young offenders. The current juvenile-justice system succeeds insofar as something like two-thirds of juveniles who get caught doing something that would count as a felony if done by an adult never get charged with an adult felony. We don’t want to change that number for the worse.

To punish a juvenile we need to do something to him that he won’t want to experience again, but do it in a way that doesn’t wreck his prospects for going straight. That is, we want something that’s aversive but not damaging. Both “boot camps” and “flash incarceration” (sometimes known as “Scared Straight”) have been tried, evaluated and failed, and I’m not aware of anything that has succeeded, though Kansas reportedly is doing very well with smaller, service-intensive juvenile corrections facilities.

Here’s a speculative idea, offered as something to try rather than something I’m convinced will work. Buy an abandoned motel (of which there are many). Take the televisions, radios and telephones out of the rooms. Put the juvenile you’re trying to punish in the room after school on Friday, with nothing but his schoolbooks, a toothbrush, and two and a half days’ worth of Meals Ready to Eat, and promise to pick him up in time for school on Monday. Tell him the door is not locked but is alarmed, and that if he leaves the room before time is up there’s a solitary-confinement cell in a juvenile jail waiting for him. And tell him he will be drug-tested Monday morning.

My guess is that most juveniles would find sixty hours of social isolation with no electronic entertainment quite unpleasant, but I doubt that it would leave any lasting psychological scars. Who knows? Some of the offenders might even get some studying done, faute de mieux. A motel-room sentence wouldn’t even be dramatic enough to make a good schoolyard story, the way a stay in juvey might. So I predict that most of the people subjected to such a sanction would be highly averse to repeating the experience.

If that’s right, then the threat of the Motel Room can be used to enforce conditions of juvenile probation, including abstinence from drug use, community service and curfews. Like any other threat, if it works, you don’t have to carry it out very often.

It’s not hard to imagine all the ways these ideas could go wrong. Most obviously, they could fail to deliver the promised consistent sanctions, either because some officials don’t want to put in the required work, because they fancy their clinical judgment and don’t want to sanction by formula, or because an over-ambitious manager starts the program off with more offenders than it can effectively handle. Judge Alm started small, shook the program down while it was still in the pilot phase, and then grew it slowly, taking advantage of the fact that new offenders added to the HOPE group had already seen the program in action on existing participants.

The other big risk is that the demands of law-and-order politics will lead officials to pile on more and more restrictions, making outpatient incarceration so onerous and unpleasant that offenders are driven to abscond in large numbers. I shudder to imagine this sort of program in the hands of someone like Sheriff Joe Arpaio of Maricopa County, Arizona, who calls himself “America’s toughest sherrif.” The key is to keep our eyes on the prize: not demonstrating “toughness” but moving offenders back into ordered lives.

Improving community corrections is not the only game in town in terms of reducing crime and incarceration. Smarter policing, much of which relies on the same sort of focused deterrence approach, has much to offer. So does a drug enforcement approach that concentrates on shutting down violent and disorderly flagrant retail markets rather than fantasizing about eliminating the supply of illicit drugs. So do a variety of social programs. While it’s not true that generalized social uplift is either a necessary or a sufficient condition for reducing crime, there are specific interventions that demonstrably make a big difference: reducing lead exposure; improving classroom discipline; extending the school day for middle schools and high schools; or Assertive Community Therapy for chronic offenders with psychiatric problems. Most of those effects are delayed. Improving classroom discipline in the first grade, for example, lowers crime rates ten years from now. This doesn’t mean such programs aren’t worth it, only that we need to be patient about getting and measuring results.

The bottom line is that drastically reducing both crime rates and the number of people behind bars is technically feasible. Whether it is politically and organizationally feasible to achieve this remains an open question. It would be tragic if the politics proved prohibitive, but it would be genuinely criminal if we didn’t even try.

1All of this has now been confirmed in a randomized controlled trial, where some probationers were chosen by lot for HOPE supervision while an otherwise similar group remained on routine probation. Nor was success merely the result of a charismatic judge; in the trial, Judge Alm’s probationers did no better, and no worse, than those assigned to other judges. See “The Impact of Hawaii’s HOPE Program on Drug Use, Crime and Recidivism”, Pew Center on the States (January 2010).
2Pre-trial release poses a special set of problems, since those subject to it are legally innocent. Pre-trial supervision must not be designed to punish, but only to ensure that the defendant will show up for trial and will refrain from lawbreaking in the meantime.

Mark A.R. Kleiman is professor of public policy at UCLA. He is author of the newly published When Brute Force Fails: How to Have Less Crime and Less Punishment (Princeton University Press) and editor of the Journal of Drug Policy Analysis. He blogs at The Reality-Based Community (