The Incentives of Private Prisons

John F. Pfaff*

Full Article.


Few institutions in our deeply flawed and troubled criminal justice system draw as much immediate ire as private prisons. In his 2016 campaign for the Democratic presidential nomination, for example, Senator Bernie Sanders’s first stab at a criminal justice reform platform was to sponsor a (surely unconstitutional) bill banning private prisons in the state and federal systems alike.[1] The 2020 campaign for the Democratic presidential nomination similarly saw multiple candidates make pledges about private prisons, while none gave any real attention to the specifics of publicly run institutions.[2]

This persistent focus on private prisons by politicians and the public alike is misguided for at least two reasons. First, it significantly overstates the role that privatization plays in the U.S. prison system.[3] All told, only about 8% of prisoners in the United States are held in privately run facilities. At least fifteen states do not rely on private prisons at all, and even among those states that use private prisons, a majority have no more than 10% of their prison populations in private facilities.[4] Mass incarceration is a public sector affair in the United States.

The more important misperception, and the one that motivates this piece, is that the distinctions many draw between public and private prisons are ephemeral, if not nonexistent. The usual criticism of private prisons runs something like this: given that private prisons make more money when more people are held in their facilities, these firms have a strong incentive to resist reforms, and in fact will push for tougher laws to keep their prisons full. Implicit in this argument, though tellingly often left unsaid, is that public prisons are somehow . . . different.

In many ways, however, they are not. To start, public prisons have a strong incentive to keep prisons full as well. They may not profit as explicitly as private facilities do, but as we will quickly see, those who work in prisons and the legislators who have prisons in their districts do profit from confining more and more people, both financially and—unlike private firms—politically.

More critically, that private prisons focus on maximizing populations is not inherent to privatization but results from how their contracts are written. If states wrote different contracts—ones that did not pay per prisoner per diem rates but instead based funding on goals such as reduced recidivism risks—then the private prisons would focus on things other than just warehousing people. Importantly, some places, most notably Australia and New Zealand, but also Pennsylvania, have started to experiment with exactly this sort of idea.[5]

Once we appreciate that the issue is not “public versus private” but “what incentives do penal institutions face,” the discussion about private prisons can take some interesting twists and turns. It is quite possible, for example, that it may be easier to incentivize private prisons to try to cut recidivism than public prisons. The private prison firms’ clear focus on profit maximization gives policymakers a direct tool that is absent with the more nebulous goals of public prison administrators. There are some intriguing longer-run issues that arise as well. If the goal is to substantially scale back the scope of incarceration, it may be easier to do that in a system dominated by private facilities. Oversimplifying somewhat, closing private prisons simply requires the state department of corrections to decline to renew a contract; closing public facilities can be much trickier, at least politically.

The intriguing role that private prisons might be able to play both in terms of changing the incentives of prisons as well as scaling back the overall scope of the prison system is uniquely relevant to Arizona, which is something of an exception to my earlier claim that private prisons are of mostly little import. As of 2018, almost 20% of the state’s prisoners were held in private facilities. Only the federal government, Florida, and Texas had more people held in private institutions, and all three of those have much bigger prison populations (and thus much smaller percentages held in private facilities); only Hawaii, Montana, New Mexico, Oklahoma, and Tennessee had larger percentages in private prisons, but all of those have substantially smaller total prison populations. Arizona’s outsized reliance on private institutions means that it could stand to benefit from designing better mechanisms to run them. And Arizona’s large prison size means that it could stand to benefit—should the political will ever arise—from the ability to close them more quickly.[6]

I have three goals here. First, I want to lay out the problems with the conventional “public versus private” framing to illuminate the potential malleability of private prison incentives. My second goal is then to draw attention to what are, as far as I can tell, the only three private prisons currently operating under incentive provisions, both to see the potential they represent but also to acknowledge the very real challenges they have faced. The early lessons coming out of Australia and New Zealand suggest that writing such contracts is tricky, but in a way that should encourage us to experiment more, not to give up on the project.

My third goal is to examine how to think about incentive contracts in particular, and private prisons and privatization in general, at a time when people are pushing to dramatically scale back, if not in some way abolish, prisons. Incentive contracts aimed at cutting recidivism raise some tricky issues for those who wish to radically reduce punishment in the United States. At first blush, such contracts would seem consistent with significant reforms: less recidivism should, perhaps, translate into fewer people in prison.[7] Moreover, prison conditions would likely improve in the process, since harsher conditions seem to lead to higher risks of recidivism.[8] By bringing about these changes, however, such contracts could increase the political legitimacy of prisons and thus actually impede more radical change. This is a well-known challenge that those pushing for abolition or dramatic reductions in prisons regularly consider: how to make conditions as humane as possible in the short- to medium-term, but in such a way that still advances the long-run goal of profound change. Perhaps somewhat surprisingly, incentive contracts may actually be a viable response to this challenge.

I. More Similar than Different

In many ways, the private prison can be traced back to the convict leasing systems that arose in the postbellum South to effectively recreate slavery-like conditions, if not all the way back to the first prisons, such as New York’s Auburn Prison, which were publicly run but deeply entwined with local business in hope that they could be financially self-sustaining.[9] For our purposes here, however, the modern private prison was born in 1983 when Tom Beasley, Doctor Robert Crants, and T. Don Hutto founded Corrections Corporation of America (CCA, now called CoreCivic) and opened a private immigration detention facility in a former hotel in Houston, Texas.[10] By the early 2010s, about 130,000 to 140,000 people were held in private prisons nationwide, making up about 8% of the total number of people serving time behind bars.[11] As prison populations fell over the 2010s, so too did the number in private prisons; in 2018, private facilities held about 120,000 people, which remained about 8% of the U.S. prison population.[12] Even at their peak, then, private facilities held under 10% of people confined in U.S. prisons.

Moreover, most of the people confined in private prisons are held in just a handful of states. Just five jurisdictions—the federal government, and then Texas, Florida, Arizona, and Georgia—are responsible for over half the people detained in private prisons, and three-fourths are held in just nine jurisdictions. Nearly twenty states do not have any private prisons at all, and eleven states each have under 500 people held in private institutions.[13]

Yet despite this relatively minor impact, private prisons face intense criticism. And I sincerely understand why. There is clearly something particularly jarring about the idea of firms financially profiting off locking other people up in cages, all the more so given the racial composition of those being locked up and the legacy of slavery and racism that explains much of it. Yet in the end, what upsets us about private prisons should upset us equally about public ones, since public prisons ultimately profit from putting people in cages as well—and not just financially, but politically too.

In other words, shifting away from private prisons does not shift us away in any real sense from “profit-driven” policies. The nature of the profit might change a bit, but as we will see, even public prisons are often sources of monetary profiting. In fact, if we take a holistic view of “profiting,” public prisons profit more from locking people up than private prisons, and changing that profit motive may be harder for the public prisons.

The private profit incentive, and its deleterious implications, are easy to see. Private prisons are paid a per diem for each prisoner they hold,[14] which encourages them to maximize population counts. Even worse, a per diem encourages them to cut programming and training and staffing, since the only way to make a profit is to push costs below what the per diems bring in. And then on top of that, the logical outcome of those cuts—harsher conditions that in turn lead to higher rates of recidivism—actually boosts the bottom line, since more people returning to prison keeps the beds filled, and thus the per diem payments coming in.

On top of this, private prison firms do not feel any of the costs of recidivism as well. The two largest private prison firms, CoreCivic and the GEO Group, are both publicly traded; their fiduciary obligations are to their widespread shareholders, not the communities impacted by inadequate prison programming and staffing. The remaining firms, such as Management & Training Corporation (MTC), are privately held, but like with CoreCivic and the GEO Group, their owners often live far from the communities they affect.[15] And to the extent that these private prisons are located in rural communities but hold mostly people from urban counties, then even the staff are socially and emotionally disconnected from many of the costs of recidivism as well.

Now, to be clear, these skewed incentives are deeply problematic. My argument here is not that these distorted incentives don’t actually exist, or that they are not that bad. It is that we see the exact same ones in the public sphere. I am not trying to elevate our view of private prisons here so much as I want to make sure we don’t give public prisons a pass they do not deserve.

In some cases, the public prisons act like private prisons because we literally impose the exact same incentive structures on them. In Louisiana, for example, the state paid (public) county sheriffs per diems for each state prisoner that the sheriffs held in their (public) county jails in an effort to manage overcrowding in the state prison system. Unsurprisingly, the sheriffs acted just like the private prisons described above: they fought to make sure their jail beds were full, and instead of taking the per diem and reinvesting it in the jail, they often skimped on services to try to use as much of it as possible on non-jail expenses. And like the private firms, these sheriffs did not have to worry about the harms of their bad decisions, since the people they held were generally returned to communities in other counties, and thus the people who elected the sheriff were unlikely to be harmed as a result of the sheriffs’ decisions.[16]

A case as explicit as this one is rare, but it is a telling example. It lays bare the idea that there is something fundamentally different between the public and private sectors and makes it clear that what matters are the incentives we give them. Give public sector officials the exact same incentives as you give the private sector officials, and they will act just like the private sector ones.

The more significant issue is that public sector prisons operating under conventional public sector arrangements face powerful, if better hidden, profit incentives as well. Take, for example, the wages and benefits paid to correctional officers and prison staff, which take up about two-thirds or more of all prison spending: over $30 billion, maybe even closer to $40 billion, of the ~$50 billion states spend annually on prisons.[17] To put that in perspective, that is more than ten times what states pay private prison firms, and perhaps one hundred times the profit those firms earn running prisons.[18] It makes sense to think of those wages as a form of profit akin to what private prison firms earn from their per diems, and we should expect the correctional officer unions to fight to protect them. And the biggest threat to those wages are layoffs in the wake of prison closures: so, like private prisons, correctional officer unions have a strong incentive to fight to keep prisons full. It’s true that unlike private prison revenue, officer wages do not vary with each and every incremental change to the prison population; but the underlying concept—prison beds must remain sufficiently full to keep state dollars coming in—is for all intents and purposes the same.

These wages are the most direct benefit, but they in turn create other knock-on financial gains. Prisons, at least those in rural areas, are often one of the few (perhaps only moderately) well-paying jobs in the communities in which they are based. The money flowing into those prisons in the form of wages thus help support all sorts of local businesses, many of which could flounder in the absence of the prisons.[19] Legislators with prisons thus have strong incentives to ensure their prisons remain open in order to protect the broader economic stimulus that the prisons are providing their communities.

Another form of “profit” that public prisons generate is the so-called “prison gerrymander,” which arises when states confront the question of where to count people in prison as “living” for the purpose of the census, which in turn shapes how the legislatures draw political district maps. Forty-one states count the people in prison as living in the prison, not at their most recent address prior to incarceration; nine count them as living at their last known addresses.[20]

In the forty-one that count people as living in the prison, that decision effectively transfers political power—via population—out of more urban, more ethnically and racially diverse, and more Democratic areas from which those held in prison disproportionately come and into more rural, white, and conservative areas. Even worse, in thirty-nine of those forty-one states (the exceptions being Maine and Vermont), people in prison count for districting but are denied the right to vote.[21]

This is—and I use this term intentionally—a five-fifths compromise, one that inflates state-level Republican legislative power. One study of Pennsylvania found that reversing the gerrymander would likely move several legislative seats back to Philadelphia and Pittsburgh, in ways that would likely lead to more Democratic legislators and fewer Republican ones.[22] Tellingly, the nine states that have reversed the gerrymander have done so only when the Democratic Party controlled both houses of the legislature and the governor’s mansion; even Republican legislators without prisons in their districts will feel pressure to block reforms that threaten their party’s overall state-level political strength.

And unlike the $30-plus billion spent on wages, which is somewhat insensitive to minor changes in prison populations—states rarely lay off correctional officers absent the closure of an entire facility, and even then they often find ways to avoid that[23]—the power of the gerrymander rises directly with the number of people confined in the facility. It is perhaps more of a per decennium than a per diem, since it matters mostly in years that end in a zero, but it is yet another underappreciated way in which public prisons, and the counties and towns in which they are located, “profit” from prisons and the number of people confined in them.

Finally, public facilities often share with private ones the same indifference to the costs of recidivism, and thus are also fairly insensitive to the costs of inadequate programming, health care, or, say, staff training. The reason here is the same one that gives the prison gerrymander its power: the people held in prisons are not returned to the communities in which the prisons are sited (which are, roughly, the communities in which the staff live). The costs of prison failures—higher rates of recidivism,[24] elevated risks of drug overdose death immediately after release,[25] the spread of infectious diseases and STDs to those communities[26]—are borne . . . elsewhere. The geographic gap is sufficient itself; the economic, cultural, urban/rural, and, most centrally, racial divides make this indifference all the greater.

Tl;dr?[27] The similarities between public and private prisons far exceed their differences. Public prisons, like private prisons, have strong incentives to keep prison populations high. And public prisons, like private prisons, do not have strong incentives to focus on programming and treatment. The two institutions are not identical, and in some situations, the specific differences—per diem payments, say, versus annualized wage bills—will make a difference. But by and large, the distinctions are much more of form than function.

An Obama-era report by the Department of Justice, for example, received significant attention when it reported that federal private prisons had higher rates of violence and other shortcomings.[28] Yet that is not the only study comparing public to private outcomes—although there are fairly few studies, a surprising result given how much attention the public–private issue has received. Collectively, the studies fail to paint any sort of clear picture. The results are muddy and seem to suggest, more than anything else, that there’s little clear difference between the two types of prisons.[29] There are terribly run public prisons and better-managed private ones, and vice versa. Hopefully by now, this should not necessarily be surprising. If they face roughly similar incentives, we should expect roughly similar outcomes.

Yet one key difference remains: it may be much easier to change the incentives of private prisons than of public ones.[30] What the private prisons are most criticized for—their blind focus on the bottom fiscal line—actually creates unique opportunities unavailable to public prisons.

This is not some sort of abstract idea: in a handful of places, governments have started to create private prison contracts that directly incentivize these facilities to focus on reducing recidivism, rather than just warehousing people until their terms expire. There are not many such places, and there have been some birthing pains, but their mere existence is an important “proof of concept” that demands our attention.

II. A Tale of Two Contracts

A.    An Overview

Once we cut through the legalese, conventional private prison contracts are ultimately fairly straightforward. Take, for example, the contract that the Arizona Department of Corrections signed with the GEO Group for GEO to manage Arizona’s Phoenix West prison.[31] The contract itself is immense—the initial agreement is nearly 150 pages, and there are forty-one amendments and additional contracts that follow, but the heart of the contract is easy to summarize. On page 25, Section simply states, “The per diem rate shall be $61.50 per inmate, per day” (a rate that has risen to $69.57 since the contract was first signed in 2005)—just eleven words, but ones that set the entire problematic set of incentives in motion. This simple payment scheme encourages firms to skimp on training and programming and eliminates any reason for them to be concerned with recidivism rates.

Now, to be clear, there are other requirements in the contract, some of which are surely intended to ensure a certain baseline of quality—but which do not appear to be particularly restrictive. When staffing the prison, for example, half the supervisory staff and one-third of the correctional officers must have at least one—just one—year of correctional experience.[32] The contract also requires that all staff be trained similarly to that in the public facilities, but as Shane Bauer’s book American Prison makes clear, such training in private institutions can often be slapdash at best and nonexistent at worst.[33]

There are also some requirements for rehabilitative programs, but these too are quite open-ended. The contract, for example, requires the private facility to provide literacy programs, as required by state law, but those state law requirements essentially boil down to two things: (1) the facility must provide 120 days of education, and (2) the person receiving that education is not eligible for parole if he is not reading at the required grade level.[34] (Of course, that parole denial isn’t a sanction for the prison contractor, which continues to profit off the person denied parole, but it does punish the person in prison, who has absolutely no control over that education.) Moreover, there are no immediate sanctions for failing to comply with these requirements—nor rewards for successfully doing so—outside of the brute-force option of terminating the contract.

Yet such terms are not unavoidable, as several private prison contracts in the Antipodes make clear. Australia now has two prisons with contracts that include incentive provisions—Ravenhall in Victoria and Parklea in New South Wales—and New Zealand has one, the Auckland South (Wiri) Corrections Facility.[35] Although the specific contract terms in all three prisons unsurprisingly differ, they share a basic common design, which is that they condition payment on the prison’s recidivism rate and on the extent to which the prison (or at least the management company) provides job training, drug treatment, and other interventions that are correlated with reduced recidivism. We can think of these goals as “direct” and “intermediate,” respectively.

Here, I will focus on Wiri, which opened in 2015, and Ravenhall, which opened in 2017.[36] Both facilities are relatively new, which makes it hard to measure the long-run effectiveness of their contracts. Moreover, given their newness, it is hard to determine if their initial spotty assessments reflect fundamental problems with the concept of incentive provisions or just the growing pains of a new project. Nonetheless, the contracts at Ravenhall and Wiri make it clear that such approaches are feasible, and their challenges to date will certainly help improve similar contracts in the future.

Now, as a general matter, I am wary of looking to other countries for guidance or inspiration for repairing our criminal justice system. Not only are institutional designs much different elsewhere, but the politics and racial history of others’ criminal justice systems often differ significantly from those in the United States. In this case, however, there are some important similarities between the United States on the one hand and Australia and New Zealand on the other. Now, there is no avoiding the fact that, on average, the United States in general—and Arizona in particular—is far more punitive than the other two countries. The U.S. incarceration rate is ~430 per 100,000, and Arizona’s is ~560, while the rate in Australia is ~115, and New Zealand’s is ~125.[37] Yet when it comes to the incarceration of racial and ethnic minorities, Arizona, Australia, and New Zealand have much more in common. Arizona’s incarceration rate of ~560 rises to ~1650 for Black people;[38] in Australia, the rate of ~115 rises to ~1000 for Aborigines and Torres Strait Islanders,[39] and in New Zealand the rate of ~125 quadruples to ~586 for Māori people.[40] Arizona, along with Australia and New Zealand, also faces similar challenges when it comes to recidivism rates. About 30% of those released from Arizona prisons return within two years, which is actually slightly lower than the rates of 40% in New Zealand and 45% in Australia.[41]

Importantly for our purposes here, Australia and New Zealand designed their private prison contracts to directly target both the high recidivism rates and the racial imbalances. The contracts at both Ravenhall and Wiri introduce several features absent from the Arizonan contract. First, both include bonus payments if the prisons manage to cut recidivism rates for their populations as a whole, and especially for their ethnic minority populations. The Wiri contract, for example, awards more favorable bonuses for cutting Māori recidivism rates than those of non-Māori, and Ravenhall’s sets a greater reduction target for Indigenous recidivism than for non-Indigenous recidivism.[42]

Both contracts also impose financial penalties when certain bad events take place. The contract for Wiri, for example, imposes a fine of $600,000 per escape, “unnatural” death of a person held in the prison, riot, or hostage-taking.[43] There are an additional fifty-two “key performance indicators” that can result in various amounts of (lesser) financial sanctions, for things ranging from assault of a person in prison by staff and assault of a person in prison by another such person in prison (which results in less of a sanction than the attack by staff) to failure to properly investigate allegations of staff misconduct.[44]

Finally, the two contracts also include terms that aim at intermediate outcomes: ones that help reduce recidivism but are not direct measures of it, such as education, housing, drug treatment, and so on. Wiri’s contract frames the incentive as a potential penalty (the management company is penalized if targets aren’t met), while Ravenhall’s does so as a reward, which could have some bearing on incentives.[45] More significantly, Wiri’s contract focuses on what takes place while the person is held in the prison, while Ravenhall’s extends its focus to examine what happens post-release. Wiri, for example, assesses the fraction of people with impending release who have secured someplace to live post-release, while Ravenhall’s contract looks into how many maintain housing after release.[46]

In many ways, these intermediate factors may be more important to focus on than the raw recidivism rate, even though the latter may be the main metric of interest, especially for a department of corrections. In some ways, these intermediate factors may actually track recidivism better than our actual recidivism data, and at least in the United States, a focus on intermediate factors may provide a way to avoid some legal issues that could otherwise arise. To see why, it perhaps makes sense to start with the issue of recidivism.

B.    A Most Misunderstood Word

“Recidivism” is a peculiar term, because it is perhaps the most important metric for success that we have while simultaneously being one of the most misunderstood and misused. When people talk about “recidivism,” they are almost always thinking about reoffending—but that is not what our “recidivism” statistics actually measure.

The core challenge is that we cannot actually observe “offending” in any of our data. People do not regularly come forward to explain what crimes they have committed over the past several months or years. “Recidivism”—the statistic—measures contact with the criminal justice system. We do not see if someone commits a crime, only if they are arrested or charged or convicted or readmitted to prison. The relationship between measures of these contacts and underlying offending are not as straightforward as it may initially appear.

To start, by measuring recidivism as criminal justice contacts, we are effectively (if implicitly) defining success as cessation, not desistance: as the complete rejection of criminal acts, not as the gradual and imperfect process of moving away from them. Consider two people, Bob and Mike. Prior to prison, Bob committed one robbery per day, but after release he commits one per week; Mike commits one per day, before and after his time in prison. Both will eventually be rearrested for robbery, and in our “recidivism” data, both will almost always be indistinguishable.[47] But it certainly seems like Bob is a much bigger success story than Mike, even if an imperfect one: he hasn’t ceased committing crimes, but he is certainly desisting from doing so.

Defining success as “no future contacts” isn’t just a blunt metric, but one that fails to properly wrestle with how people transition away from violent and antisocial behavior. Doing so is difficult in its own right, and it is made all the more difficult by the host of formal and informal barriers to success that people who have come in contact with the criminal justice system persistently face (from informal discrimination to, say, formal bans on employment and public housing). An incentive program that uses a crude measure of “recidivism” will likely under-reward many successful interventions.

Another obvious limitation with our official metrics of “recidivism” is that they are strongly shaped by how law enforcement is deployed. It is harder, for example, to “recidivate” in communities that have less of a police presence, since crimes may be more likely to go undetected or unaddressed. In other words, assume that both Mike and Bob reduced the number of robberies they committed to one per week, but Mike lives in a highly policed community and Bob in a less-patrolled one.[48] Mike is more likely to “recidivate” because his risk of arrest is greater, even if the levels of offending are the same (and Bob’s seeming “cessation” may be less a reflection of his behavior and more one of his risk of detection).

The intermediate goals, however, may be able to avoid some of these problems. There is an extensive literature pointing out how many of the intermediate factors in the Wiri and Ravenhall contracts, like stable housing and employment, can be important pathways toward desistance, and eventually cessation.[49] In fact, some criminologists have recently started to argue that we should reframe the way we think about recidivism in just this way: less measurement of failure, more measurement of these sorts of intermediate successes that we know often lead to (unobserved, and mostly unobservable) desistance.[50]

In other words, while Wiri and Ravenhall have received attention mostly for their direct recidivism incentives, their provisions focusing on these intermediate factors may actually be the more important and consequential of their innovations.

C.    Some Mixed Results

In the abstract, “incentivize better programming” is a straightforward concept. Actual implementation, however, can be a bit harder. The early results from both Ravenhall and Wiri, which have been mixed at best, illuminate some important challenges such approaches face.

In both countries, for example, prison officials have struggled to determine whether or not the prisons are actually causally reducing recidivism. One challenge, of course, is separating effectiveness from luck. Wiri’s management company, Serco, received a $1.1 million bonus at roughly the same time it received sharp criticism from the Office of the Ombudsman for failing to provide adequate services.[51] On the one hand, Wiri’s bonus came in no small part from reducing Māori recidivism rates better than elsewhere. On the other hand, the Ombudsman’s report noted that its inspectors “found little evidence that the Prison was addressing Māori reoffending.”[52] In other words, it certainly seems plausible that the prison’s success was due more to luck than specific programming—which also means that there could be years in which the prison is denied bonuses due to bad luck as well. Perhaps this is simply an unavoidable aspect of incentives such as this, and perhaps such errors will “net out” over time. But it does highlight the concerns of relying simply on recidivism as the core metric, which in turn again highlights the gains from looking at intermediate factors.

Ravenhall’s experience with its recidivism metric pointed to another, more structural challenge. A report by the Victorian Auditor-General’s Office (VAGO) noted several limitations. To start, people held in prisons are often moved from facility to facility, so someone “released” from Ravenhall may have only spent the last short period of his time detained there, and someone “released” from a state-run prison may have spent most of his time at Ravenhall, perhaps benefiting from superior programming.[53] Assuming Ravenhall’s programming really is superior, the first transfer issue artificially inflates Ravenhall’s failure rate, and the second artificially inflates the state system’s success rate; both errors lead to underestimates of Ravenhall’s relative performance.

VAGO went on to acknowledge that the only way to practically assess recidivism was to develop a rigorous causal model to estimate the comparative impact of Ravenhall’s programming. Such modeling is challenging, however. To start, it requires more data and more careful analysis, and the VAGO report depressingly notes that Corrections Victoria “does not currently have plans to evaluate Ravenhall’s outcomes beyond the [current recidivism] measures,” even though “[b]etter indicators and a strong research and evaluation project is required to meaningfully compare Ravenhall’s performance to that of other prisons.” Moreover, even if Corrections Victoria were willing to undertake more rigorous analyses, such modeling would almost inevitably lead to significant disagreement about how they should be designed, given the sorts of methodological challenges such models face.[54]

This is not to say that assessing the intermediate outcomes was without its challenges, but the issues that have arisen here seem more likely to be shorter-run problems that can be ironed out as the contracts progress. One issue is that some provisions were defined too specifically and in a way that perhaps failed to properly account for the challenges that those released from prison realistically face. The Ravenhall contract, for example, classified someone released from Ravenhall as successfully employed only if he secured a steady job of twenty or more hours per week, a target that ultimately counted as “failures” men who were working hard but simply unable to find such employment, an all-too-common problem those released from prison face. Especially if the goal is to emphasize gradual, imperfect desistance and reintegration, not cessation and immediate success, intermediate metrics should be flexible and graduated in a way that meshes with the challenges those released will inevitably face.

Another challenge that the GEO Group and Corrections Victoria have faced with the intermediate factors is gathering data. The recidivism data comes from the public sector, but much of the intermediate data—such as employment, housing, or treatment—has to come from third-party private providers, many of whom apparently have been unwilling or uninterested in providing the necessary data to the GEO Group.[55]

Perhaps the biggest defect with the contracts that neither report addresses is that the incentives simply are not strong enough. The bonuses in the Ravenhall contract come to about 1% of the contract’s payments.[56] Wiri’s $1.1 million bonus comes to what looks like about 2.5% of the payments to Serco.[57] One study of private prison incentive contracts suggested that the contracts should be written so that the private firms cannot break even unless they at least significantly achieve their recidivism targets.[58] Otherwise, the incentives get skewed for both the private prison and the government. The private prison does not have a strong incentive to achieve the targets, but the government also does not have a strong incentive to police what the private firm is doing or to resist its claims for bonuses.[59] Forcing the private firms to have more skin in the reduce-recidivism game would force them to take the issue much more seriously and would encourage the government to monitor them much more closely.

III. A Few Lessons for Arizona (and Beyond)

The contracts for Ravenhall and Wiri are, if nothing else, proof of concept: while they have encountered challenges ranging from the minor to the daunting, they show that it is possible to incentivize prisons to think more carefully about recidivism and re-entry. Cutting recidivism and reoffending is a huge benefit in and of itself, of course, not just to the community but to the people leaving prisons, whose lives are better off as well. Moreover, cutting recidivism would help scale back incarceration. In Arizona for several decades now, about 40% of all people released from prison return within three years, accounting for about 13% of annual admissions.[60] A greater emphasis on recidivism will thus help Arizona and other states alleviate overcrowding in their prisons—Arizona’s prison system, like almost all other state systems, is operating at over 100% of its rated capacity—and will also cut arrests and caseloads for police and prosecutors.[61] Improved contracts will thus cut harms and save city, county, and (ideally) state governments money that can be reinvested into more socially productive efforts.

There are a few lessons that states like Arizona can take away from the contracts already in the field. Perhaps the most important one is that the intermediate factors may be the ones that deserve the most attention, for both empirical and legal reasons. Empirically, as noted above, determining whether someone in fact recidivated is difficult, since people have an incentive to avoid self-reporting criminal acts. It may be easier to assess if they are in school, or have stable housing, or are attending drug or mental health treatment. Now, to be clear, Ravenhall has struggled even here, with some third-party providers of those services either failing or refusing to share data. But it still seems easier to verify these pro-social outcomes than reoffending. Moreover, there could be data problems with the more general recidivism provisions. It is quite possible that state departments of corrections, fearing how they might compare to better-incentivized private providers, may balk at providing the necessary data to draw comparisons.

Finally, these intermediate factors are successes in and of themselves. Far too often we view “recidivism” as the alpha and omega of prison success. Someone who ultimately returns to prison but before doing so has a more fulfilling life—rewarding employment, housing that allows him to feel secure, the chance to build better and stronger connections with family and friends—has certainly “succeeded” more than someone who returns to prison after experiencing fewer (or none!) of those things. Yes, all those things and not reoffending is better still, but the outcomes measured by the intermediate factors have worth, in both moral and consequentialist ways.

Also, at least stateside, the intermediate factors may be able to accomplish goals that the direct recidivism factors cannot. The approaches used in Australia and New Zealand to confront racial bias—rewarding contractors more for cutting the recidivism rates of ethnic and racial minorities—would likely not pass constitutional muster over here. If states determined that certain intermediate goals would benefit different groups differently, however, they could conceivably reward certain factors more than others, in a way that could target racial disparities in prison while being sufficiently race-blind to avoid strict scrutiny.

One obvious concern with emphasizing intermediate goals is that the contract almost starts to look like a private parole contract than a private prison one: a contract like Ravenhall’s that focuses on post-release outcomes in effect encourages the prison manager to also provide and maintain post-release services.[62] Such an idea is certainly not without controversy. While private parole is not something that really exists, private probation is, and it is quite often subjected to harsh criticisms.[63] Note, though, that the criticisms of private probation, like those of private prisons, blame the institution for the contract terms. Private probation as practiced in the United States is quite flawed because current private probation systems are basically designed to make those on probation pay for their probation, and the private companies make more money the longer someone is on probation.[64] These are terrible incentives. Ideally, an intermediate-focused prison contract would, in effect, be a private parole system that pays based on people getting off parole.

A related benefit of focusing on post-release intermediate factors is that it addresses the perhaps surprising fact about the time served in Arizona’s prisons. Like with Ravenhall, many of the people incarcerated in Arizona (and elsewhere) are held in prison for only a short period of time and thus may not qualify for certain programming options, or even if they do, may not be there long enough to take much advantage of them. About one-third of all people held in Arizona prisons are released within six months, and between 40% to 50% in just one year.[65] The share of such short-serving detainees will likely be even larger in the state’s private prisons, which are mostly minimum-security facilities.[66] Post-release incentives are all the more important in this context.

Moreover, a state like Arizona that relies heavily on private prisons may be able to avoid or mitigate at least one of the major problems that Ravenhall and Wiri faced, which was the confusion created by transferring people from one prison to another. For example, the GEO Group manages four of the six private facilities in Arizona, accounting for about 5,800 of the state’s nearly 8,000 privately run beds.[67] Rather than writing contracts that focus on specific prisons, the state could incentivize the companies, by trying—with some (non-trivial) adjustments—to keep transfers within the network of GEO Group facilities.[68]

Even if transfers within a private prison firm’s set of prisons are impractical, the state could design contracts that reward all three private prison companies operating in the state bonuses based on the collective performance of the private prisons, perhaps prorated based on the amount of time someone spent in each firm’s facilities. Outside of the GEO Group’s four private prisons, there is also a small minimum-security prison focused on people with substance use disorders run by the Management and Training Corporation, and a larger general-population minimum-security prison managed by CoreCivic.[69] Building incentives around all six facilities simultaneously could open up more options, even if it forecloses the ability to pit the private prisons against each other to encourage innovation in treatment—and humaneness.

Finally, one lesson from both Ravenhall and Wiri that states like Arizona should note is that the contracts will work best if they ensure that firms have real skin in the game. The incentive provisions at both Ravenhall and Wiri were only minor fractions of the total payments those firms would receive, which significantly mutes the need to focus on them. To ensure that the firms truly emphasize cutting recidivism and advancing the intermediate goals, the contracts should be written such that the firms will lose money on the contracts unless the goals are substantially met.

Obviously, a pivot toward incentive contracts would require Arizona to effectively tear up all the old contracts and start anew with incentive-based ones. But this is not without precedent—Pennsylvania did this with its halfway house contracts in 2013, and the program appears to be working.[70] And doing so has the potential to make prisons more humane places while also potentially cutting back on the scale of incarceration in states that use private prisons.

IV. Incentive Contracts and Abolition

As I write this in the summer of 2020, the politics of criminal justice appears to be in the midst of a significant realignment, with abolitionist goals receiving significantly more attention than they had even just a few months ago. Anyone seriously thinking about police and prison reform has had to wrestle with the implications of abolitionism for a while now, but the protests prompted by the murder of George Floyd by police officer Derek Chauvin have pushed the issue to the forefront of our national debate about what the goals of criminal justice reform should be, and about whether incremental reforms work . . . or if they might even be harmful. It is a question that is essential to raise here as well: if one is committed to dramatically scaling back the scope of incarceration in the United States, are incentive contracts helpful or harmful? To my own surprise, I found myself thinking that a shift toward private prisons with incentive contracts may actually be consistent with long-run abolitionist goals. The argument here is still tentative and conditional, but I also think intriguing.

This is not the place to provide a lengthy discussion of what “abolition” refers to, in part because like any broad concept, the term itself is often contested. But, oversimplifying (perhaps significantly), the basic idea is that instead of investing in police and prosecutors and prisons, we should invest in the sorts of institutions, like schools and jobs, that help communities self-regulate.[71] And to the extent that there should be enforcement, it should be managed by local community organizations, not the formal criminal justice system, given the fraught history with that system that poor communities of color, Black communities in particular, have had.[72] When it comes to prisons, the goal is to replace prisons entirely as institutions: to invest in alternate methods of achieving public safety and accountability.[73]

The underlying intuition here is not as extreme as it might seem at first to those unfamiliar with it. Most communities that experience low crime do not do so because of the constant looming threat of arrest or punishment; if anything, safe neighborhoods are defined by the absence of police and formal state punishments, at least within their borders. These communities have the resources and capacity to regulate themselves, with some but minimal state involvement. The goal of abolition is to explore how to achieve these outcomes everywhere.

Over the years, I have become increasingly sympathetic to the abolitionist goals when it comes to prisons. The data increasingly shows, fairly unambiguously, that prison is a highly ineffective, blunt-force tool for dealing with crime: while it incapacitates, it also causally increases the risk of reoffending upon release, and its deterrent impact is often significantly overstated.[74] Moreover, the (unmeasured) social costs—the years of life lost, the families disrupted, diseases spread, drug overdoses caused, and on and on—are surely staggering.[75] And there is a growing menu of options that can replace the prison, in ways that not only promote public safety, but better address the needs of those harmed and those who cause harm alike.[76]

Abolition, of course, is clearly a long-run goal, and that actually raises some tricky policy issues when it comes to reform efforts. It is not always clear that short-run reforms translate into long-run radical change. In fact, the more effective the reform, perhaps the more risks it poses to fundamental realignment. Consider the case of, say, incentive contracts for private prisons. If incentive contracts actually work, and prisons operating under them perform better, then such contracts may actually increase the political legitimacy of prisons and thus make future, more transformative change harder.[77]The flip side, of course, is that if those contracts work, then prison will be a less oppressive place for those inside it and fewer people will return to it, both huge improvements for those who experience them.[78]

Thus, abolitionists argue that it is important to focus on short-run reforms that also advance, or at least do not hinder, longer-run transformative change.[79] Perhaps surprisingly—I surprised myself in reaching this conclusion—it seems that private prisons with incentive contracts could satisfy both these goals: improving short-run conditions while not impeding, and perhaps in a way even facilitating, future long-run change. I don’t mean for this to sound contentiously contrarian, but the argument at least merits thinking through.

To see why private prison incentive contracts may be consistent with long-run abolitionist goals, it is essential to think about two aspects of the economics and politics of public prisons discussed above. To start, public prisons are often important sources of employment and political power in poorer, more rural parts of the country.[80] And unlike in private prisons, the workforces in public prisons are unionized, and these public sector unions often wield significant political power of their own.[81]

It is easy to see why these two factors mean that public prisons will aggressively resist any sort of significant reform, and in fact will do so far more aggressively than private prison firms. First, private prison firms are much better able to adapt to fundamental changes than public sector corrections officers and the legislators with prisons in their districts. The push for decarceration has often emphasized that funding should be moved into areas such as drug treatment and other social services[82]—services that will be mostly provided closer to where those who need them live. In other words, they will transfer resources from more rural prisons to more urban treatment facilities. To correctional officers and the legislators who rely on prisons for economic and political benefits, this is a direct and significant loss, one that they will resist aggressively. The officers are not trained to work in such centers, nor can they simply relocate halfway across the state for such a job; the politicians are similarly incapable of easily adapting.

But it’s not the same for the private prison firms, which can easily adapt, and in fact are already doing so. While the correctional officer cannot simply move from, say, Kingman to Phoenix, the GEO Group could easily shut down its Kingman facility and open up a Phoenix-based treatment center. Tellingly, firms such as CoreCivic and the GEO Group are already doing this.[83] Of course, this means that these firms are now involved in programs such as drug treatment, but unlike prisons, many social services are already highly privatized. Nearly 90% of all drug treatment facilities, for example, are run by for- or non-profits (with non-profits making up an overall majority of such facilities).[84] Obviously, the proposal here would aggravate any concerns people have with privatization in the treatment or other social services sector, but given the extensiveness of privatization in these areas already, the marginal impact would likely be slight at most.

Moreover, the workforces in private prisons tend not to be unionized,[85] which means they are less able to coordinate their opposition to closures. Such non-unionized correctional officers also seem likely to have weaker ties with other law enforcement unions and lobbying groups, such as police unions and state-level district attorney associations, who could advocate on their behalf as well. As a result, closing private prisons seems far easier to accomplish than closing public ones. The private firms will be far more willing than local politicians to accept a shift from more rural prisons to more urban treatment facilities, and the private prison workforce is less able to resist such a move.[86] Thus, it is not implausible that adopting incentive contracts—which seem easier to impose on private prisons than on public ones—could simultaneously improve conditions in the short run while at least not impeding (and perhaps even facilitating) longer-run change.

In other words, yes, it is true that incentive contracts may make prisons “better” (not good, but better), and in doing so, risk increasing their political legitimacy. But there are valid reasons to think that this sort of entrenchment risk is less when we are dealing with private prisons. While I’ve often argued that public and private prisons are more similar than different, here there is a key difference: public prisons are geographically fixed, while private prison firms are not. Thus, public prisons cannot change their mission as easily as private firms can. In this case, these distinctions—immaterial when considered within the world of prisons, but quite relevant when thinking about moving away from it—suggest that private prison firms will resist transitions away from imprisonment far less than public prisons. Thus, incentive contracts may allow us to improve short-run conditions in prisons in a way that, perhaps surprisingly, at least does not hinder pushes for more transformative change.

In fact, incentive contracts could be uniquely effective in conservative and fairly punitive states like Arizona. If they work, they demonstrate that more humane approaches are consistent with public safety—but do so by taking advantage of capitalistic incentives, which may be more politically palatable to a more conservative electorate. This could open up political options that had earlier been closed off.


Traditionally, public and private prisons alike were effectively paid per diem rates for the people they confined. The form of such payments may have differed, but in substance, the incentives were relatively identical. Recent experiments in Australia and New Zealand suggest that we can, in fact, write contracts that encourage private prisons, at least, to focus more on cutting recidivism. Such contracts remain in the infancy right now, but a state like Arizona, which relies on private prisons more heavily than perhaps any other state in the country, seems like the ideal laboratory for working through how to use such contracts far more extensively.

     *      Professor of Law, Fordham University School of Law.

     [1].     Press Release, Bernie Sanders, Sen., U.S. Senate, House Leaders Introduce Bill To Ban Private Prisons (Sept. 17, 2015), []. While Congress could forbid the Federal Bureau of Prisons from signing contracts with for-profit prison firms, the Tenth Amendment bars it from telling the states what they can do in this situation. It’s true that Sanders’s bill was almost certainly intended to be a symbolic gesture far more than a viable piece of legislation, but it is telling that prison privatization is what he chose to focus on.

     [2].     The Issues: Private Prisons, Politico, [] (Dec. 19, 2019).

     [3].     For the purposes of this paper, when talking about privatization, I will be referring to privately managed prisons. There is a second sort of privatization, which receives far less attention but may in fact be more consequential, which is the privatization of services provided in public prisons, such as food services, medical care, and telephones. Yet it is easy to overstate the scope of privatization here as well. See, e.g., Peter Wagner & Bernadette Rabuy, Following the Money of Mass Incarceration, Prison Pol’y Initiative (Jan. 25, 2017), []; John Pfaff (@JohnFPfaff), Twitter (Feb. 24, 2020, 9:19 AM) (citing The Pew Charitable Trs., State Prisons and the Delivery of Hospital Care (2018), []), []. Nonetheless, these private actors likely impact the lives of far more people in far more significant ways.

     [4].     E. Ann Carson, U.S. Dep’t of Just., Bureau of Just. Stat., Prisoners in 2018, at 27 (2020), [].

     [5].     Lauren-Brooke Eisen & Rebecca Autrey, A Critical Look at Private Prisons Overseas, Brennan Ctr. for Just. (May 13, 2019), [].

     [6].     Carson, supra note 4, at 27. That is not to say that states with private prisons currently want to cut back on incarceration. Perhaps unsurprisingly, the states that tend to rely on private prisons tend to be more politically conservative, and thus more punitive. Except for Hawaii, whose situation is somewhat idiosyncratic, and New Mexico, none of the states listed above with more people or a greater fraction of people in private prisons would be seen as Democratic-leaning states. In recent years, however, even states that were traditionally thought of as “punitive” have sought to scale back their reliance on prisons. Oklahoma and Louisiana have both recently pushed through impactful reforms, and even conservative politicians in Texas now boast about how many prisons it is closing, not opening. See, e.g., James White (@James_E_White), Twitter (June 26, 2020, 7:29 AM) (citing Jolie McCullough, As the Texas Prison Population Shrinks, the State Is Closing Two More Lockups, Tex. Trib. (Feb. 20, 2020, 1:00 PM), []), []. That said, I am assuming here that there is some political will to scale back prisons or at the very least to make them function in more “productive” ways. If that political will is simply absent—if politicians in Arizona (or elsewhere) are content to simply warehouse people, or perhaps even view that as the morally correct response to criminal behavior—then it is unlikely that the proposals here would be viable.

    [7].     Of course, reality is a bit trickier. With almost all prison systems operating at or above capacity, reducing admissions due to recidivism could simply mean that states become more willing to admit people for their first conviction, or that they simply detain the people in prisons longer, thus keeping prison populations stable as admissions fall. This latter effect, at least, has been seen in some states. See, e.g., John Pfaff (@JohnFPfaff), Twitter (Apr. 30, 2020, 12:23 PM) (citing Carson, supra note 4, at 4–5, 13–14), [].

    [8].     See, e.g., M. Keith Chen & Jesse M. Shapiro, Do Harsher Prison Conditions Reduce Recidivism? A Discontinuity-Based Approach, 9 Am. L. & Econ. Rev. 1, 16 (2007).

    [9].     See, e.g., Shane Bauer, American Prison 19 (2018); Philip Goodman, Joshua Page & Michelle Phelps, Breaking the Pendulum: The Long Struggle over Criminal Justice 32 (2017).

     [10].   The CCA Story: Our Company History, Corr. Corp. of Am., [].

     [11].   E. Ann Carson & William J. Sabol, U.S. Dep’t of Just., Bureau of Just. Stat., Prisoners in 2011, at 32 (2012), []; E. Ann Carson, U.S. Dep’t of Just., Bureau of Just. Stat., Prisoners in 2013, at 14 (2014), [].

     [12].   Carson, supra note 4, at 27–28.

     [13].   Id. The federal government alone is responsible for nearly 25% of all people held in private prisons. Id. If we exclude the federal government, the story remains roughly the same: five states hold over half the people in state private prisons (the four above, plus Tennessee), and nine states hold over 75% of such people. Id.

     [14].   Stephanie Leacock, Private Prison Contracts and Minimum Occupancy Clauses, Am. U. Bus. L. Rev.: BLR Buzz Blog (Nov. 19, 2017), [].

     [15].   Dymond Green & Tala Hadavi, Why Big Banks Could Be Killing Private Prisons, CNBC (Jan. 2, 2020, 1:29 PM), []; Management & Training Corporation, LinkedIn, [].

     [16].   See Cindy Chang, Louisiana Is the World’s Prison Capital, (May 13, 2012, 3:00 PM), [].

     [17].   Chris Mai & Ram Subramanian, Vera Inst. of Just., The Price of Prisons: Examining State Spending Trends, 2010–2015, at 8, 9 (2017), []; U.S. Census Bureau, 2018 State and Local Government: US Summary & Alabama–Mississippi (2018), [].

     [18].   Wagner & Rabuy, supra note 3.

     [19].   Recent work by John Major Eason casts doubt on earlier studies that suggested such collateral economic benefits were scant. Eason’s key insight is to note that prisons are not built in random places: the towns with prisons tend to seek them out because they are struggling to land other businesses. Prison communities, in other words, are already disproportionately disadvantaged. Controlling for this selection effect seems to indicate that these sorts of collateral economic benefits are real. John M. Eason, Big House on the Prairie: Rise of the Rural Ghetto and Prison Proliferation 14, 16 (2017).

     [20].   The nine states that do not engage in the gerrymander are California, Delaware, Maryland, Nevada, New Jersey, New York, Oregon, Virginia, and Washington. Tristiaña Hinton, Virginia Becoming 9th State To End ‘Prison Gerrymanders,Fulcrum (Apr. 24, 2020), [].

     [21].   Felon Voting Rights, Nat’l Conf. of St. Legislatures (Oct. 1, 2020), [].

     [22].   Brianna Remster & Rory Kramer, Shifting Power: The Impact of Incarceration on Political Representation, 15 Du Bois Rev. 417, 438 (2018). Pennsylvania’s incarceration rate in 2018 was 369 per 100,000, which was just slightly below the state institution average of 392 per 100,000, so it is not a distinctly punitive state when it comes to incarceration; harsher states should expect to see even bigger effects from the gerrymander. Corrections Statistical Analysis Tool (CSAT)—Prisoners: Imprisonment Rates of Total Jurisdiction Population (Pennsylvania), Bureau of Just. Stat., [] (choose “Custom Tables” from the toolbar; then select “State Institutions (Total)” and “Pennsylvania” for “Jurisdiction”; click “2018” for “Years”; “Year-End Population” under “Population”; then “Total Jurisdiction Population” for “First Variable”; then click “Generate Rates Table”).

     [23].   When Michigan closed the Pugsley Correctional Facility, which employed about 230 people, it laid off only 51, all of whom had declined or refused transfers (and thus had the ability to avoid losing their jobs). John Agar, Michigan Prison Closing Brings 51 Layoffs, MLive (Apr. 2, 2019), [].

     [24].   See, e.g., Mariel Alper et al., U.S. Dep’t of Just., Bureau of Just. Stat., NCJ 250975, 2018 Update on Prisoner Recidivism: A 9-Year Follow-up Period (2005–2014), at 1 (2018), [].

     [25].   See Ingrid A. Binswanger et al., Release from Prison—a High Risk of Death for Former Inmates, 356 New Eng. J. Med. 157, 161 (2007).

     [26].   See Susan Okie, Sex, Drugs, Prisons, and HIV, 356 New Eng. J. Med. 105, 106 (2007).

     [27].   Tl;dr has been in dictionaries since 2018, which seems to suggest its usage is common enough to need no explanation. But it appears to have been used outside of a quote only once in law reviews prior to this one, so in order to be safe, here is a good overview of its ever-evolving etymology and usage: Andrew Heinzman, What Does “TLDR” Mean, and How Do You Use It?, How-To-Geek (Aug. 2, 2019, 8:00 AM), [].

     [28].   Off. of the Inspector Gen., U.S. Dep’t of Just., Review of the Federal Bureau of Prisons’ Monitoring of Contract Prisons 14 (2016), []. The report did not examine whether the private prisons performed better or worse when it came to recidivism, but to be fair, almost all private prisons in the federal system are used to house foreign nationals who will almost all be deported at the end of their sentences. Id. at 25.

     [29].   Sasha Volokh, Are Private Prisons Better or Worse than Public Prisons?, Wash. Post (Feb. 25, 2014, 9:17 AM), [].

     [30].   Again: assuming the political will to do so is there. Perhaps the reason that private prisons have functioned like public ones is because legislators had little to no actual desire to take advantage of the opportunities that the private option presented.

     [31].   Ariz. Dep’t of Corr., Rehab. & Reentry, ADC Contract No. 040176DC (2012), []. Note that the contract is initially between ADOC and Correctional Services Corporation (CSC). The GEO Group purchased CSC in 2005, closing the deal just a few months after this contract was signed. GEO Group Buys Out Correctional Services Corporation, Prison Legal News (Jan. 15, 2006), [].

     [32].   Ariz. Dep’t of Corr., supra note 31, § 8.4.2, at 17.

     [33].   See Bauer, supra note 9. To be fair, Bauer’s experience was in a single facility in another state (Louisiana), and one that is operated by a more regional private firm, LaSalle Corrections, than the international firms that run the facilities in Arizona.

     [34].   Ariz. Dep’t of Corr., supra note 31, § 11.10, at 53–54; see also Ariz. Rev. Stat. Ann. §§ 31-229 to 229.02 (2020).

     [35].   There had been three private prisons with incentive contracts in Australia, but now there are only two. As of 2020, the Western Australia government has taken back control of the Melaleuca Correctional Facility, which had paid private prison manager Sodexho a $15,000 bonus per non-returning detainee. Andrew Bushnell, Institute of Public Affairs, Cutting Costs and Reducing Reoffending (Report, September 2019) 11 (Austl.) <>, archived at <>. The decision came amidst concerns that Sodexho was not properly managing the prison. Hamish Hastie, ‘Then There Was One: WA Left with Just One Private Prison with Melaleuca To Change Hands’, WAtoday (online, 23 December 2019) (Austl.) <>, archived at <>.

     [36].   ‘Ravenhall Correctional Centre’, Corrections, Prisons & Parole Victoria, (Web Page) (Austl.) <>, archived at <>; “Auckland South Corrections Facility” Department of Corrections (N.Z.) <> []. While the Parklea facility has been open for many years, the incentive provisions are quite new, taking effect in 2019. Minister of Corrections, Parklea Correctional Complex – Management Deed (Report, 2018) 54 (Austl.)<>, archived at <>. It is thus far too early to assess how they have worked.

     [37].   For the United States’ and Arizona’s rates, see Carson, supra note 4, at 9, 11. For Australia’s prison population, see ‘Prisoners in Australia’, Australian Bureau of Statistics (Web Page, 12 May 2019) <>, archived at <>. For New Zealand’s prison population, see “Prison facts and statistics – March 2020” (31 March 2020) Department of Corrections (N.Z.) <> [].

     [38].   See Ariz. Dep’t of Corr., Rehab. & Reentry, Corrections at a Glance: April 2020 (2020), [], for the incarceration rate in Arizona, and QuickFacts: Arizona, U.S. Census Bureau, [], for the Black incarceration rate in Arizona.

     [39].   See Australian Bureau of Statistics, supra note 37, and ‘Profile of Indigenous Australians’, Australian Institute of Health and Welfare (Web Page, 11 September 2019) <>, archived at <>, for Australian data on incarceration rates for Aborigines and Torres Strait Islanders.

     [40].   “NZ.Stat” Statistics New Zealand <> [] (for the first table, choose “Corrections,” then select “Calendar Year,” click “Annual Sentenced Prisoners Populations,” then choose “Maori under the “Ethnicity” tab; for the second table, choose “Population Estimates,” then click “Population Estimates Tables,” then select “Estimated Resident Population, national population by ethnic group”).

     [41].   John R. Lott, Jr. & Rujun Wang, Prisoners in Arizona: A 2017 Update on Selected Topics 77 fig.95 (2018),—A-2017-Update-on-Selected-Topics-2017.pdf []; Marcus Boomen “Where New Zealand stands internationally: A comparison of offence profiles and recidivism rates” (2018) 6(1) Practice 87 (N.Z.), <> [] at 93; ‘Released Prisoners Returning to Prison’, Sentencing Advisory Council (Web Page, 2020) (Austl.), <>, archived at <>. Note that because Australia and New Zealand operate “unified” systems, with both people awaiting trial and those already convicted held in the same facilities, the Australian and New Zealand statistics may be higher in part because they include the people held on remand, who are not included in the Arizona data.

     [42].   See “Agreement Relating to the PPP at Wiri Men’s Prison Project” (10 September 2012) Department of Corrections (N.Z.) <> [] at sched. 16 § 16; ‘Ravenhall Prison Project Agreement’, Department of Justice and Community Safety (Web Page, 2014) [§ 13.3] (Austl.) <>, archived at <>.

     [43].   See Department of Corrections, supra note 42, at sched. 16 § 1.2.

     [44].   See sources cited supra note 42. The Ravenhall contract has similar terms, but their specifics are redacted in the public version of the contract.

     [45].   The psychology here is tricky, but a behavioralist insight called Prospect Theory has made it clear that framing identical payoffs as “wins” or “losses” changes how people balance risks. See Bushnell, supra note 35, at 17, for a discussion of Prospect Theory in the context of private prison incentive grants (arguing for bonuses, not penalties).

     [46].   Department of Corrections, supra note 42, §§ 4.05–4.06; Victorian Auditor-General’s Office, Parliament of Victoria, Ravenhall Prison: Rehabilitating and Reintegrating Prisoners (Parliamentary Paper No 118, March 2020) (Austl.) <>, archived at <>.

     [47].   Recidivism studies almost always define recidivism as “at least one” subsequent contact (arrest, conviction, admission). See, e.g., Alper et al., supra note 24, at 1. One could conceivably adopt a more nuanced definition, but such are rarely if ever seen in official government reports on the issue.

     [48].   This dynamic is why using prior criminal history in sentencing is often controversial among reformers, who fear that a person’s history often reflects police presence in his community, which is often shaped by the racial composition of that neighborhood.

     [49].   See, e.g., David S. Kirk et al., The Impact of Residential Change and Housing Stability on Recidivism: Pilot Results from the Maryland Opportunities Through Vouchers Experiment (MOVE), 14 J. Experimental Criminology 213, 213–26 (2018).

     [50].   See, e.g., Cecelia Klingele, Measuring Change: From Rates of Recidivism to Markers of Desistance, 109 J. Crim. L. & Criminology 769, 801–06 (2019).

     [51].   Andrea Vance “$1.1m bonus for private prison company Serco, for keeping people out of jail” (20 February 2019) Stuff (N.Z.) <> []; Peter Boshier Report on an announced inspection of Auckland South Corrections Facility under the Crimes of Torture Act 1989 (Office of the Ombudsman, OPCAT Report, 20 February 2019) (N.Z.) <> [] at 7–8.

     [52].   Boshier, supra note 51, at 29.

     [53].   Victorian Auditor-General’s Office, supra note 46, at 48.

     [54].   The VAGO report also noted that nearly 60% of Ravenhall’s sentenced prisoners spent fewer than ninety days in prison, which rendered them ineligible for Ravenhall’s programming and thus hard to include in the comparative metrics. 29.

     [55].   Id. at 46. In the specific case of Ravenhall, such disputes may be unlikely given how little is relatively at stake—the incentive parts of Ravenhall’s contract come to about 1% of the contract’s annual value. See Bushnell, supra note 35, at 14. In fact, this could be one reason why Corrections Victoria isn’t pushing for more complex approaches. But if, as I will discuss shortly, the incentive part becomes a bigger portion of the contract, the risk of disputes will rise accordingly.

     [56].   See Lauren-Brooke Eisen, The Private Prison Experiments: Is There Any Positive in For-Profit Imprisonment?, Salon (Feb. 25, 2019, 12:00 PM), []; Bushnell, supra note 35, at 11, 14.

     [57].   The twenty-five-year contract has a value of about $1 billion, or $40 million per year. So $1.1 million is about 2.5% of that $40 million. See Vance, supra note 51.

     [58].   Bushnell, supra note 35, at 15.

     [59].   Of course, these defects could in reality be seen as features of both the firms and the government—both get to appear to take recidivism seriously without including terms that force them to do so. Which is to say that pointing out that the contracts should be written in a way that make the incentives parts important is not the same thing as saying that such contracts are easily written as a political matter.

     [60].   See Lott & Wang, supra note 41, at 79 fig.96. In 2010, for example, Arizona released about 13,500 people from prison, nearly 40% of whom returned to prison within three years. Over those three years, Arizona admitted about 40,000 total people to prison; the 40% returning from the 2010 release group made up about 13% of those 40,000 admissions. See id. See generally Corrections Statistical Analysis Tool (CSAT)—Prisoners: Number of Sentenced Prisoners Admitted to State or Federal Prisons, 1978–2018, Bureau of Just. Stat., [] (choose “Quick Tables”; then select “Number of Sentenced Prisoners Admitted to State or Federal Prisons, 1978-2018” under “Prison Admissions”).

     [61].   See Carson, supra note 4, at 25–26 tbl.17. Arizona’s own most recent data from May 2020 has its capacity usage at over 105% of its rated capacity. See Ariz. Dep’t of Corr., Rehab. & Reentry, Corrections at a Glance: May 2020 (2020), [].

     [62].   In fact, one reason the GEO Group received the contract for Ravenhall was that it already had relationships with reentry services through its Bridge Centre, which has provided extended assistance to men released from Ravenhall. SeeVictorian Auditor-General’s Office, supra note 46, at 41 (discussing the “unique features” of Ravenhall, including the Bridge Centre).

     [63].   See, e.g., Radley Balko, Opinion, Georgia’s Privatized Probation and Parole System Isn’t Working, Wash. Post (Oct. 5, 2018, 11:23 AM), [].

     [64].   “Set Up To Fail”: The Impact of Offender-Funded Private Probation on the Poor, Hum. Rts. Watch (Feb. 20, 2018), [].

     [65].   See Lott & Wang, supra note 41, at 23 figs.19–20.

     [66].   The GEO Group’s 3,500-bed Kingman facility is minimum-medium-security, and its 1,300-bed Central Arizona facility is also medium-security but (currently) intended for those convicted of sex offenses. See Kingman, Ariz. Dep’t of Corr., Rehab. & Reentry, []; Central Arizona Correctional Facility (CACF), Ariz. Dep’t of Corr., Rehab. & Reentry, []. All remaining facilities, either run by the GEO Group or other firms, are minimum-security. See, e.g., Marana, Ariz. Dep’t of Corr., Rehab. & Reentry, [].

     [67].   See The GEO Grp., Inc., 2018 Annual Report 118 (2018), []; see also E. Ann Carson, U.S. Dep’t of Just., Bureau of Just. Stat., Prisoners in 2016, at 22 tbl.17 (2018), [].

     [68].   The adjustments may be somewhat significant. One of the GEO Group’s facilities, the minimum-medium-security facility at Kingman, is a general-population institution. See Kingman, supra note 66. The remaining three are smaller, more specialized prisons (Central Arizona focuses on sex offenses, Phoenix West on people with substance use disorders, and Florence West on people with DUIs and other nonviolent offenses). See Central Arizona Correctional Facility (CACF), supranote 66; Phoenix West, Ariz. Dep’t of Corr., Rehab. & Reentry, []; Florence West, Ariz. Dep’t of Corr., Rehab. & Reentry, []. If all four were general-population facilities, the idea of keeping transfers within the GEO Group’s network would be straightforward, at least in theory. Such a proposal at this point may thus require reconfiguring how some facilities are used or the mix of people held in them.

     [69].   See Marana, supra note 66; Red Rock Correctional Center, Ariz. Dep’t of Corr., Rehab. & Reentry, [].

     [70].   See Leonard Gilroy, Pay for Success Contracting Reducing Recidivism in Pennsylvania, Reason Found. (Aug. 31, 2015), [].

     [71].   See Aaron Ross Coleman, Police Reform, Defunding, and Abolition, Explained, Vox (July 16, 2020, 8:00 AM), [].

     [72].   See, e.g., Community Accountability,, [].

     [73].   See, e.g., Rachel Kushner, Is Prison Necessary? Ruth Wilson Gilmore Might Change Your Mind, N.Y. Times (Apr. 17, 2019), [].

     [74].   See David Roodman, Open Philanthropy Project, The Impacts of Incarceration on Crime 7–8 (2017), [].

     [75].   See John Pfaff, The Incalculable Costs of Mass Incarceration, Appeal (Sept. 20, 2018), [].

     [76].   See, e.g., Danielle Sered, Until We Reckon: Violence, Mass Incarceration, and a Road to Repair (2019).

     [77].   Of course, for those who are non-abolitionist reformers—who accept the legitimacy of prisons as an institution but want to make them better—the contracts would have accomplished exactly what they were intended to do. “Prison reform” is often spoken of as if it is, for all intents and purposes, a fairly monolithic movement, but like any broad social movement, it contains factions that often disagree in quite fundamental ways.

     [78].   See, e.g., Lauren-Brooke Eisen, How To Create More Humane Private Prisons, Brennan Ctr. for Just. (Nov. 14, 2018), [].

     [79].   See, e.g., Critical Resistance, Reformist Reforms vs. Abolitionist Steps in Policing, [].

     [80].   See generally Tracy Huling, Building a Prison Economy in Rural America, in Invisible Punishment: The Collateral Consequences of Mass Imprisonment (Marc Mauer & Meda Chesney-Lind eds., 2002).

     [81].   See Suzanne M. Kirchhoff, Cong. Rsch. Serv., R41177, Economic Impacts of Prison Growth 14 (2010), [].


     [83].   See generally Kara Gotsch & Vinay Basti, The Sent’g Project, Capitalizing on Mass Incarceration: U.S. Growth in Private Prisons 1 (2018), [].

     [84].   Eighty-nine percent of all facilities are privately run, and 51% (of the total, not just the non-government facilities) are run by non-profit private firms. Substance Abuse & Mental Health Servs. Admin., National Survey of Substance Abuse Treatment Services (N-SSATS) 13 (2018), [].

     [85].   See Gotsch & Basti, supra note 83, at 6.

     [86].   It should be noted that the economic implications of any prison closure, public or private, deserve our attention, even though that is a too-far tangent for this paper. Prison closures will come with real economic costs that will hurt real people. That is not an argument for keeping prisons open—there are better ways to support economically vulnerable rural communities than putting people in cages—but we should still think about how to assist those communities that will be hurt by closures. See, e.g., John Pfaff, Cory Booker Has a Plan To “Reverse” Mass Incarceration. It Won’t Work., Vox (Sept. 26, 2017, 8:40 AM),[].