Public Safety Policing, Crime Control
December 1st, 2022 60 Minute Read Report by Charles Fain Lehman

Is 3,300 Enough? Why the Borough-Based Jails Are Too Small to Keep NYC Safe

In 2019, then-mayor Bill de Blasio introduced, and the New York City Council approved, plans to close the jail complex on Rikers Island and replace it with four jails in Manhattan, Brooklyn, the Bronx, and Queens. Construction on these borough-based jails is expected to be completed in 2027, at which time the city is expected to shutter Rikers entirely. In so doing, it will replace a system with a maximum capacity of nearly 15,000 beds with one that can hold just 3,300 detainees on a given day.

That capacity is, by any measure, extremely small, representing a daily population almost never seen on Rikers since its opening a century ago. Even after more than half a decade of deliberate decarceration, the jail’s daily population today sits between 5,500 and 6,000, far greater than the borough-based jails’ allotment. Can New York City operate a jail system with just 3,300 beds without either: a) dangerous, likely illegal, overcrowding? or b) making the city less safe?

This report argues that the answer is no. To reach this conclusion, it recounts how the 3,300-bed figure was arrived at, which had more to do with politics than with any reasonable projection of required capacity. It then details the research on the effects of pretrial detention, investigates who is currently on Rikers and who could safely be released, and estimates the relationship between crime rates and jail population. The bottom line: under almost no conceivable scenario can the city expect to safely and sustainably reduce daily jail population to 3,300—never mind, to reduce it below that figure.

Given the city’s commitment to closing Rikers, this report concludes by looking at potential sources of alternative capacity, including refurbishing or repurchasing closed jails; constructing small additional borough jails; and “boarding out” detainees to Long Island and Westchester County. These solutions could buy additional capacity for the system but not enough to provide adequate and safe housing for even the current, much-reduced population. In light of this, the report briefly revisits the case for keeping some of Rikers open.

New York City’s leaders appear dead-set on proceeding with Rikers’ closure. If so, they must reckon seriously with where they will put criminals afterward. The borough-based jails simply are not big enough.



New York City is closing the jails on Rikers Island. The administrations of both Mayors Bill de Blasio and Eric Adams have promised that the century-old detention complex will be shuttered by 2027.[1] In its place, a long-standing dream will finally be realized: four jails erected in the boroughs, so that detainees, their lawyers, and families, as well as correctional officers, will no longer have to make the perilous trek to Rikers Island. Rikers’ dangerous conditions—including a high rate of violence that has only risen in the past year—will be replaced with safe and humane detention facilities.[2]

That is one theory, at any rate. In another telling, the one popular among members of the communities where the new facilities will be erected, the borough-based jails will bring blight, crime, and decay—in some cases, to already-struggling neighborhoods.[3] These “mega jails” have no place in New York, opponents argue—better to refurbish Rikers and keep dangerous offenders there.[4]

The borough-based-jails debate is, in many ways, about the problem of scale. On one side are those who think that the Rikers’ detention complex, in addition to being decrepit, is too large; NYC would have a more humane system if its jails were smaller. On the other side are those who think that the borough-based jails themselves are too big; better to fix Rikers because these hulking buildings will ruin the neighborhoods that they’ve been dropped in.

But there is one question that neither side has really engaged with, one that Mayor Adams raised only in August 2022: What if the borough-based jails aren’t big enough?[5]

Once erected, the four borough-based jails will house 825 detainees each, for a total maximum daily population of 3,300.[6] That figure is 40% lower than the average population recorded at Rikers throughout 2022 (roughly 5,500 through mid-September). It is lower than the lowest daily population (3,809) recorded during the Covid-19 pandemic, when jail decarceration was a nationwide priority. It is smaller, in fact, than almost any figure recorded since the Rikers jail complex opened.[7] Reducing the city’s daily jail population to 3,300, the Mayor’s Office of Criminal Justice (MOCJ) estimated in 2019, would mean a jail-to-population ratio 3 times smaller than Chicago’s, 4.5 times smaller than Los Angeles’s, and 8.5 times smaller than Philadelphia’s.[8] Can NYC possibly reduce its jail population to this point without sacrificing public safety and while still ensuring that criminals show up for trial?

The answer, this report argues, is no. The function of jail detention is to keep dangerous offenders off the street before their trials and to ensure that flight-prone offenders do not fail to appear for their trials. But most individuals currently detained on Rikers Island are either serious, often violent, felony offenders, or at risk of absconding. Given even extraordinarily conservative estimates of which detainees are either a public safety or flight risk, there is almost no point in the last six years when the jail population could have been below 3,300. Any hope of reaching that point, furthermore, depends on crime falling well below the historical lows that New Yorkers have recently enjoyed—an unlikely feat, given that crime has risen over the past three years, ending the decades-long crime drop.

Charging ahead with the borough-based jails plan is a recipe for disaster. It will create a system incapable of detaining everyone it needs to detain. The result will either be severe overcrowding, creating conditions that may be even worse than those on Rikers, or the release of many hundreds of dangerous or flight-prone offenders onto the streets, exacerbating the city’s crime and disorder problems.

Policymakers have cavalierly proceeded under assumptions that are, bluntly, out of touch with reality. This report proposes several possible ways to expand the city’s jail capacity without using Rikers: reopening two unused jails in Manhattan, erecting small additional jails in the Bronx and Staten Island, and sending those serving criminal sentences to other counties. But even these are half-measures. As it stands, the borough-based jails do not contain enough capacity to keep NYC safe; its leaders ignore this fact at their and their city’s peril.

Where Did 3,300 Come From, Anyway?

New York City has been detaining criminal offenders on Rikers Island since the 1930s, when it first began constructing facilities there to replace the decaying jail infrastructure on Blackwell’s Island.[9] (“Jail” generally refers to facilities that hold those not yet convicted of a crime or who are serving sentences of less than a year; “prison” refers to facilities that hold those serving a longer sentence.) Today, the island houses 10 major facilities, with a maximum capacity of nearly 15,000, approximately enough space to house the entire jail population of New York State. Some of these facilities have been closed, however, resulting in an active capacity of 11,300.[10] A further 2,300 can be housed off the island in four current borough jails, court pens, and specialized units in Elmhurst and Bellevue hospitals.[11] As Figure 1 shows, even this capacity has, at times, not been enough to avoid overcrowding, a perennial problem through the heights of crime in the 1980s and 1990s.

Figure 1

The Rise and Fall of New York City’s Jail Population

New York City’s leaders have been considering closing Rikers for nearly half a century. In the early 1970s, a series of riots, guard protests, and lawsuits led federal Judge Morris Lasker to find that the jail’s unsanitary and unsafe conditions—many of the facilities were decrepit even then—violated detainees’ constitutional rights. In response, the city proposed to lease the island to the state and use $200 million in revenue to construct eight new jails in the boroughs. The plan was eventually mothballed, though, after community opposition and the discovery that the city had substantially underestimated capital costs.[12]

The idea of closing Rikers did not, however, go away. In her 2016 State of the City address, then-City Council speaker Melissa Mark-Viverito set the stage for its revival. Invoking Kalief Browder—the 16-year-old who spent three years on Rikers awaiting trial, including a cumulative two in solitary, before taking his own life two years after release—Mark-Viverito condemned “our current Rikers-centric system.” Rikers, she argued, was not only inefficient and “plagued” by violence, but it imposed unnecessary transportation costs on the city and on the families of detainees, and otherwise kept “those who have been accused of committing crimes out of sight and out of mind.” She announced the formation of what would eventually be called the Independent Commission on New York City Criminal Justice and Incarceration Reform—more commonly, the “Lippman Commission,” after its chair, former New York State chief judge Jonathan Lippman.[13]

A little more than a year later, the Lippman Commission announced its expected conclusion: “Rikers Island must be closed.” To replace Rikers, the commission called for five jails, one in each borough, with a capacity of 5,500 beds to hold an average daily population (ADP) of 5,000, “with each facility proportional in size to the number of people held from that borough.” Placing jails in the boroughs, the commission noted, would facilitate ease of access to the city’s courts, as well as making it easier for family members, lawyers, and Department of Correction (DOC) employees to reach the facilities via public transportation, rather than relying on the single bridge to Rikers Island. The commission laid out a series of reforms that, it argued, could cut NYC’s jail population—then roughly 9,700—in half within a decade.[14] Many of these reforms have since been enacted as state laws: bail reform, discovery reform, and “Raise the Age” are all contemplated in the Lippman Commission’s initial report. Indeed, the desire to close Rikers was a major impetus for the criminal justice reforms passed in NYS over the past five years.

The de Blasio administration embraced Rikers closure as a policy goal. Just a few months after the Lippman Commission released its report, the administration promised to reduce the daily jail population to 7,000 within five years and created a task force to identify strategies to reduce population to 5,000. “Once the jail population reaches 5,000,” it noted, “the City will be in a position to close Rikers Island for good.”[15]

A total of 5,500 beds to serve an ADP of 5,000 remained the goal, in fact, as late as September 2019, when the administration went before the City Planning Commission (CPC) to officially present its borough-based jail construction plan. It looked slightly different from the Lippman Commission’s original proposal: four jails instead of five, with none on Staten Island, and with the beds divided evenly (1,437 per jail, for a total of 5,748) among the boroughs, rather than in proportion to their share of the offender population. Still, the basic capacity remained the same, and CPC approved the 5,500+-bed jail plan on September 3, 2019, passing it along to the city council.[16]

The jails faced substantial community opposition even before CPC’s vote, with their size being a leading source of conflict. As proposed, the buildings were quite large—Brooklyn’s would have been 40 stories and Manhattan’s 45—and some were situated in areas without similarly tall buildings. This was a major reason that community boards and borough presidents reported to CPC that they either disapproved of the plan or offered approval conditional on shrinking the size of the jails by 30%–40%.[17]

But could the jails be safely reduced to this size? MOJC had been maintaining for months that its original estimates were high and that, given current trends, ADP would be 4,000 by 2027, requiring a systemwide capacity of just 4,600.[18] But that was still too big. Then, in a shockingly convenient turn of events just three days before the final vote, MOCJ, then helmed by Liz Glazer, announced that it was revising down its estimate of the 2026 ADP to 3,300. It provided scant reason for the drop, except that it was made possible by bail reform and continued declines in city sentences, detention for technical parole violations, and crime.[19] The next day, a group of councilmembers—including Margaret Chin, Stephen Levin, Karen Koslowitz, and Diana Ayala, members from the areas where the new jails were slated to be built—announced that the height of the jails would be slashed to a maximum of 295 feet, with capacity reduced proportionally. Then-City Council Speaker Corey Johnson credited the de Blasio administration for helping to facilitate the changes.[20] On October 17, 2019, the council approved the plan.[21]

The result was that by the time the plan was passed through the city council, the borough jails had been shrunk by nearly 40%, to 886 beds each, a total of 3,544 beds for an ADP of just 3,300—a number, the New York Times noted, “that the city has not recorded in nearly 100 years.”[22] In addition to the smaller jails, the de Blasio administration committed to nearly $400 million in new spending, a large percentage of which would go directly to the neighborhoods where the jails would be placed.[23]

It is impossible to review this story and not conclude that the 3,300 number was reached for political reasons, not because it aligned with plausible projections of how much jail capacity is needed to keep New Yorkers safe. The special commission chartered to close Rikers asked for 5,500, a figure that persisted all the way through CPC’s review. Even the de Blasio administration, faced with mounting political challenges, publicly claimed that the lowest it could go was 4,600. It strains credulity to believe that the sudden revision to 3,300, coming, as it did, mere hours before the city council was set to vote, was a product of some shocking new modeling insight, rather than an effort to ram through what could be rammed through.

That the final figure seems to bear more relation to politics than to capacity needs does not necessarily mean that it is inadequate. Perhaps 3,300 is enough beds to hold everyone who ought to be detained. But to assess that requires judging which offenders on Rikers on a given day really ought to be there—which requires us to ask: Who should be in jail? What, in other words, is jail for?

What Is Jail For?

If jail serves no legitimate purpose, then the optimal capacity of NYC’s jails is zero. But except for the most radical advocates of closing Rikers, most agree that there are some people who ought to be detained at any given time. If we can articulate the standards by which we judge whether someone should be jailed, we can assess what the optimal jail population is on any given day and determine how much the observed jail population deviates from it.

To begin to think about the uses of jail, Table 1 breaks out the total DOC population in May 2022 by the procedural reason for the inmate’s incarceration.[24] The overwhelming majority are detained pending trial either for a new crime (85%), on a violation of their parole (<1%), or both (4%).[25] A further 6% are serving out a brief sentence in city jails, and the remaining 4% are awaiting transfer to a state facility. In other words, today NYC’s jails are primarily for detaining individuals before trial, with a handful of posttrial detainees serving a brief sentence or stopping there before going to prison upstate.

Table 1

Average Daily Population of NYC Jails by Inmate Status, May 1, 2022–May 31, 2022

Pretrial detention serves two functions. First, it ensures a defendant’s appearance in court and prevents flight to another jurisdiction.[26] The second function is public safety: detaining certain people pretrial can, in theory, reduce their risk of re-offending, by keeping them off the street (incapacitation) or by discouraging them from offending in the future (deterrence). While current NYS law permits detention only on the basis of flight risk, almost everyone recognizes this second function as a de facto goal of jail, and many advocate for New York to follow every other state in permitting its consideration.[27] Even the Lippman Commission acknowledged the public safety function of the jail system in its original report, stating that “any new system should not compromise public safety” and that its approach was meant to reserve “incarceration for those who pose a real, cognizable danger to the public,” implying that such a population can legitimately be incarcerated.[28]

Does jail actually serve these functions? Are the 90% of DOC detainees held pretrial being stopped from failing to appear, and being stopped from committing crimes, either during the course of their detention, or in the future?[29] The answer, research shows, is more complicated than either proponents or opponents of jail detention make it out to be.

Table 2 summarizes the results of studies that recover the causal effect of pretrial detention on risk of failure to appear and of re-offending, both before and after disposition.[30] All estimates are reported as the effect of the listed treatment on the listed outcome, in terms of added percentage-point probability of the outcome, given the treatment. The following subsections interpret this research in more detail, including a discussion of how these studies work, their limits, and what we can conclude generally about the efficacy of pretrial detention.

Table 2

Research on the Effects of Pretrial Detention

Flight Risk

Intuitively, being detained pretrial means that there is zero risk that a given offender will abscond or otherwise fail to appear for trial. But how much of a reduction is that, relative to all offenders being allowed to walk free? In 2011, about 14% of NYC detainees who were released subsequently failed to appear in court.[31] Does that mean that if all pretrial detainees were released, 86% would show up to court? Not necessarily, because the decision about whether to set bail is partly based on an offender’s risk of failing to appear. In other words, those who are detained present the greatest flight risk. To isolate the causal effect of jail on failure to appear, we need to compare otherwise similar groups of offenders who are the same except for whether they are detained.

Of the four studies, two use designs that can identify the causal effect of pretrial detention on flight. Gupta et al. estimate that being assigned money bail rather than being released on recognizance has no statistically significant effect on risk of failure to appear, “or, if anything, increases failures to appear.”[32] Their estimate, however, is based on two measures of failing to appear that they admit are flawed—one because it is inconsistently coded in the data, and another because it likely picks up false positives. Both measures, they indicate, likely “underreport the true number of failures to appear.” In addition, their estimates, strictly speaking, measure not the impact of detention per se but of being asked to post bail—a different effect from the one we are interested in.[33]

A better estimate comes from Dobbie et al., who observe actual detention status, as well as failure-to-appear rates, in Philadelphia. They find that the marginal released defendant is a statistically significant 15.6 percentage points more likely to fail to appear, which is a 128.9% increase relative to the detained defendant mean. They also find that being released increases the risk of flight from the jurisdiction by 0.5 percentage points, a large (250%) but nonsignificant increase from the mean.[34] The effects are larger for felons (21.2 pp) than misdemeanants (12.3 pp) but of similar magnitude across type of crime and type of release condition.[35]

In other words, the research confirms that jail significantly reduces the risk of failure to appear. It is thus reasonable to detain individuals who cannot otherwise be stopped from failing to appear for subsequent hearings.

Incapacitation and Deterrence

Pretrial detention can affect crime through two channels: incapacitating the offender; and deterring him from re-offending. All else equal, one would expect incapacitation to reduce crime because people are severely limited in their ability to commit crimes in jail relative to on the street. The deterrent effects of jail are more ambiguous. Jail could discourage future offending, e.g., by “scaring straight” a detainee. But it could also be criminogenic, by making it harder to hold down a job once released, or by exposing a detainee to other criminals who push him toward criminality (the “school of crime” hypothesis). Or it could have no effect at all. Distinguishing between these possibilities is important in understanding jail’s effect on crime.

All four of the reviewed studies address this question:

  • Mueller-Smith looks at detainees in Harris County, TX (i.e., Houston). It finds that a jail stay reduces a felony offender’s risk of offending by 6 percentage points during detention but also that it raises the probability of rearrest within one year of release by 5.6 percentage points.[36]
  • Gupta, Hansman, and Frenchman, previously discussed, do not break out pre- and post-disposition effects but find that being assigned any money bail causes a nonsignificant 0.7-percentage-point increase in re-offense risk within one year, which may be because the incapacitative and deterrent effects net out (as in Mueller-Smith).[37]
  • Dobbie et al., also previously discussed, similarly find that the overall effect of being detained is a statistically insignificant 0.1% increase in rearrest risk. Decomposing that outcome, however, they find that being released increases the risk of offending by 18.9 percentage points pre-disposition but reduces risk of re-offense within the two years after disposition by 12.1 percentage points. They translate this into the actual number of new crimes, finding that “the marginal released defendant is arrested for 1.09 more counts prior to disposition, but 0.73 fewer counts after case disposition.”[38]
  • Leslie and Pope study felony offenders in the NYC jail system. They find that detention reduces the risk of offending by 12.2 percentage points but heightens the risk of offending within the two years following disposition by 7.5 percentage points.[39]

The balance of evidence suggests that incapacitation in jail reduces crime but that the reduction is counterbalanced by a criminogenic effect, where jail detention makes it more likely that someone will re-offend following disposition. At best, jail, on average, has no effect on crime; at worst, i.e., if the criminogenic effect persists beyond the window of the studies, jail is criminogenic.

Does this mean that there is no crime-reducing effect of NYC’s jail system? Not necessarily. The study with the strongest evidence that jail reduces crime is based on NYC’s jails. The effects of jail may not be the same everywhere because different places have different jail populations, and different jail populations likely respond differently to the balance of incapacitation and deterrence. If a place’s jail population is composed exclusively of Charles Manson, incapacitation likely outweighs criminogenesis; if it is composed of kittens, probably the opposite is true.

All four studies, however, have in common a significant limitation—they focus only on offenders on the margins of incarceration versus release. This fact has significant implications when we think about the cost-benefit balance of detaining the most serious offenders.

Beyond the Marginal Offender

To this point, there has been no discussion of how studies of the effects of pretrial detention actually operate. But a brief methodological detour will help frame the significance of the findings above.

Much as with failure to appear, isolating the causal effect of detention on subsequent offending is not so easy. Historically, simple panel regressions found that going to jail was correlated with an increased risk of subsequent offenses.[40] But this may simply be driven by selection: we would expect a judge to be more likely to jail someone who was, for other reasons, also more likely to commit another crime. A judge is more likely, for example, to jail the serial offender than the first-time misdemeanant, meaning that the relationship between jail and future offense is confounded by offender characteristics. To some extent, we can account for this with statistical controls. But controls do not let us account for unobserved characteristics, which may continue to confound the relationship.

In response to similar identification problems across their field, economists have developed several tools for isolating the relationship between two variables. All the aforementioned studies use an “instrumental variables” (I.V.) approach, in which variation in some third variable that is related to the independent variable (in this case, whether a person is jailed) but not the dependent variable (whether he re-offends) is used to isolate the effect of the former on the latter.[41] In particular, the studies exploit the random assignment of defendants to judges. Because judge assignment affects re-offense risk only through the channel of the judge’s propensity to incarcerate, it can be used as an instrument to randomly vary risk of detention, which, in turn, allows for the isolation of the causal relationship between detention and re-offense.[42]

Here is where things get complicated (if they have not already). Almost by definition, detention will affect different people differently—recall Charles Manson vs. the kittens. The analyses used above recover the average effect of a treatment on a population, and what holds true for the average does not necessarily hold true for every individual case. More important, they recover only the average effect for those individuals for whom counterfactual assignment to a different value of the I.V.—i.e., a different judge—would elicit a different outcome. Some offenders would always be released, regardless of the judge they get, while others would always be detained. The estimated effect does not speak to the effects of pretrial detention on these people. Rather, the effect is the average for those on the margins of detention or release, individuals about whose detention judges would (at least according to the model) disagree.[43]

The necessary focus on the marginal offender has important implications for how these data should be interpreted. Criminal justice reform advocates often use the aforementioned studies to argue that jail is universally criminogenic. In reality, these studies report the average effect of detention (or release, or cash bail) for those offenders who are on the margin of detention versus release—in statistical lingo, the “compliers.” In Dobbie et al., “13 percent of our sample are compliers, 36 percent are never [released], and 51 percent are always [released],” while in Leslie and Pope, data in the appendix suggest that compliers make up roughly 28% of the sample.[44]

Because those always or never assigned to detention cannot, by definition, be analyzed with these data, it is hard to identify the effects of detention on them. But it stands to reason that if the net effect of incarceration is about neutral among marginal defendants, then incapacitation benefits outweigh criminogenic costs among those whom all judges would detain, and vice versa among those whom all judges would release. In the former case, it is easy to see how judges would identify a hardened criminal—at high risk of offending pretrial and unlikely to have his behavior altered by jail—and incapacitate him. Similarly, it is easy to imagine how judges would set free a first-time misdemeanant who they believe is more likely to be hurt by jail in the long run than stopped in the short run.

Some suggestive evidence from the research above supports this theory. Dobbie et al. (the only study to provide the relevant breakouts in its public appendices) find that being released makes misdemeanor offenders 15.6 percentage points more likely to be rearrested prior to disposition but 14.5 percentage points less likely to be arrested following disposition. Releasing felony offenders, by contrast, makes them 21.2 percentage points more likely to be rearrested pre-disposition, and only 9.4 percentage points less likely to be rearrested post-disposition, with the last estimate failing to achieve even marginal statistical significance.[45] I.e., even among marginal offenders, there is a gradient by which more serious ones are more incapacitated and less induced to postdisposition re-offense by pretrial incapacitation.

The net null effect of jail on future crime does not mean that jail serves no legitimate public safety function. Rather, it means that jail is not optimal for marginal offenders. It still stands to reason—and, in fact, the previous suggestive evidence seems to support the idea—that jailing the most serious offenders is a net positive for public safety, relative to the counterfactual.

If the preceding argument is true, it seems likely that two groups—which may overlap—should be detained pretrial. One is offenders at some significant risk of flight. The second is individuals for whom the benefits of incapacitation likely outweigh a minimized anti-deterrent effect of their jail stay. Those are most likely the most serious offenders, who are likely to continue to offend if out and whose post-disposition offending is unlikely to be affected by yet another jail stay. Although it is hard to identify this population precisely, we can reasonably suggest that serious criminal offenders pose a significant enough risk to the public that detaining them pretrial is the crime-minimizing approach.

Can New York Plausibly Detain Just 3,300?

Having framed our problem, we now turn to a deceptively simple question: Can NYC plausibly fulfill the twin functions of jail—minimizing the risks of flight and re-offense—with just 3,300 jail beds?

It is worth emphasizing that NYC has already undergone a historical emptying of its jails. Figure 2 shows the daily population of city jails from June 2, 2016—just four months after Mark-Viverito announced the formation of the Lippman Commission—through mid-October 2022. In that time, Rikers’ daily population fell from 9,837 to 3,809 in the depths of Covid, before rebounding to 5,919 in October 2022.

Those trends, as Figure 2 suggests, were a function of three phenomena: criminal justice reform, Covid, and the recent increase in crime. Dramatic changes to NYS’s criminal law, particularly the 2020 reforms that ended the use of cash bail for all but the most serious offenses, were already driving daily population down before the onset of the pandemic.[46] Conspicuously, in fact, the pre-Covid daily population was roughly on par with the level reached after the temporary, dramatic dip in decarceration during the pandemic. Since grand jury proceedings fully restarted in early 2021, daily population has hovered at roughly the same level, even with parole reform (Less Is More Act) effectively ending the detention of most parole violators.

Figure 2

New York City’s Jail Population, June 2016–October 2022

The city has already picked much of the low-hanging fruit of decarceration, diverting thousands of people from jail to alternatives to detention or denying the courts the power to detain them at all. Despite both those changes and a once-in-a-century pandemic, daily population has never once fallen below 3,300. Is there obvious low-hanging fruit left to pick? Can NYC continue to decarcerate while still jailing those who should be jailed?

To address this question, two analyses are presented. The first simply asks whether it is plausible that many offenders still in DOC custody—given their observed characteristics—can be released safely while ensuring their court appearance. It presents a variety of possible scenarios, essentially all of which are incompatible with a maximum daily population of 3,300. The second analysis turns to the Lippman Commission’s proposal for getting to 3,300. It concludes that the commission relied on a too-optimistic-by-half set of assumptions about the viability of decarceration.

Who’s on Rikers Island, and Can They Be Released?

How do we judge a given detainee’s risk to public safety? Data on detainees from DOC offer two theoretically plausible indicators. One is the top charge associated with the offender, which tells us whether he is charged with a felony, a violent offense, etc. The other is whether the offender is in a Security Risk Group, i.e., a prison gang. The following analysis filters out all but the most serious offenders, using the following sequence:

  • Removing all but felons and gang members 
  • Removing all but felons 
  • Removing all but violent and gun felons 
  • Removing all offenders but those charged with felonies for gun possession, murder, robbery, assault, rape, and burglary[47]

In addition to dangerousness, we also must consider which offenders present a flight risk. By definition, everyone in pretrial detention has been found to present a flight risk by at least one arraignment judge; otherwise, they would not have been detained.[48] At the same time, NYC’s jail population has in the past declined without an appreciable increase in the failure-to-appear rate, so we cannot simply claim that all current detainees are a flight risk. The estimates below retain a fixed percentage of those not otherwise identified as dangerous, estimating three separate rates:

  • The failure-to-appear rate of all offenders released on their own recognizance in NYC, January 2020 through December 2021, 11.6%[49
  • The failure-to-appear rate of all offenders released on nonmonetary supervision in NYC, January 2020 through December 2021, 28%[50
  • The lowest risk of failure to appear after which the NYC Criminal Justice Agency (CJA) recommends not releasing an offender on his own recognizance, 35%[51]

These are, in essence, three rosy scenarios for what would happen if people detained in DOC custody were released. Under scenario 1, they would fail to appear at the same rate as those released on their own recognizance. Under scenario 2, they would fail to appear at the same rate as other individuals released at arraignment with some nonmonetary restriction, e.g., maintaining employment or being subject to electronic monitoring. Under scenario 3, they would fail to appear at the same rate as offenders at the lowest risk threshold above which CJA recommends not releasing on recognizance. In reality, it is likely that the true failure-to-appear risk of Rikers residents is higher than all these values—after all, they were judged flight risks—and therefore, these are conservative estimates.

Figure 3 applies each of these filters to the data on the past six years of DOC custody. Each chart shows what the population would have been under each set of assumptions, both with the varied failure-to-appear risks and the varied releases.

Figure 3

Modeling Decarceration Approaches

What Figure 3 shows is that a daily population of 3,300 is unlikely under all but the most extreme conditions. Under certain scenarios, the daily jail population dipped below 3,300 during the early days of Covid-19. But if NYC’s decarceration plan depends on pandemic-level releases as the norm, it is doomed to failure. Even assuming a similar situation, NYC cannot reach that line without releasing all but the most serious offenders—including many serial offenders who are incarcerated for pettier offenses—and likely accepting a much higher failure-to-appear rate.

There is a simple reason for this: years of criminal justice reform have freed or diverted essentially all the easy-to-release cases. Many of those remaining on Rikers are serious flight risks, dangerous, or both. And no law of the universe requires that the population that actually requires incarceration be less than 3,300, no matter what city government might wish.

To give readers a sense of how challenging further decarceration will be, the above applet allows them to explore the most recently available data about the population of the NYC jail system. As should be apparent, it is extremely challenging to do so without releasing violent or otherwise dangerous offenders.

Neither this analysis nor other projections account for the fact that space needs are not merely a function of the size of the population, but the concentration of potentially violent people within it. A jail unit that contains 50 low-level drug offenders is much easier to control than a unit that contains 50 prison gang members from different gangs. If, in fact, Rikers detains only the most serious and violent offenders, then more space may be necessary to ensure that tensions do not escalate.

The Lippman Commission’s Road Map

In July 2021, the Lippman Commission and the Center for Court Innovation released a report detailing their plan to bring ADP below 3,300.[52] The report freely admits that “the majority (79%) [of people detained pretrial] were charged with a violent felony, with nonviolent felonies and misdemeanor or lesser charges making up 16% and 5%, respectively,” adding: “This reality underscores that further decarceration depends, in part, on a willingness to try a different approach with people whose alleged crimes involve violence.”[53]

Rather than carving out certain groups for release or detention, the commission’s road map lays out a series of changes that, it argues, will bring ADP down from 5,753 on June 1, 2021, to between 2,700 and 3,150—conveniently just below the 3,300 cap. (By way of comparison, on May 31, 2022, the population was 5,515, roughly the same.) The remainder of this section reviews the commission’s proposals and the approach used to estimate their impact on ADP. As will become apparent, the proposed changes are variously no longer applicable, disconnected from concerns about public safety or the current political environment, based on unjustifiably optimistic estimates of the share of the population that can be released, and, in at least one case, simply inexplicable.

The “Covid-19 Backlog”

The commission’s report identifies a significant increase in the length of stay of detainees on Rikers, rising from an average of 261 days on March 18, 2020, to 341 days on June 1, 2021. This increased length of stay, the commission claims, is due to Covid-related delays in the court system, particularly the pause on convening grand juries. The commission estimates that there would be 740 fewer people but for Covid-19’s effects.

Even if we grant this premise, the effect of Covid on the jail population has almost certainly, at least in part, attenuated. Grand jury hearings were paused between March 2020 and August 2020, and again between November 2020 and April 2021.[54] But they have since resumed, so it is hard to assert that the level of delay has remained constant.

Moreover, it is not necessarily the case that an increased length of stay is caused exclusively, or even primarily, by Covid-related delays. The early days of the pandemic saw a massive reduction in Rikers’ population, caused by releases and diversions done in the name of public health.[55] This release almost certainly selected for offenders who would have been in jail for a short time, thereby increasing the share of the offender population held for a long period and therefore the mean length of stay.[56] I.e., we would expect the mean length of stay to go up mechanically, even if it did not reflect an actual change in time to disposition.

Pretrial Decision-Making

The most direct route to population reduction that the commission proposes is to simply send fewer people to jail during the arraignment process. This, they argue, is justifiable because the absolute risk of rearrest or flight is low: “Few New Yorkers released before trial pose a genuine flight risk … [and] re-arrest rates while people are released before trial are low, including for people charged with violent felonies.” But, of course, these statistics represent the flight and rearrest risks for individuals released on their own recognizance precisely because a court considers them trustworthy; they are consequently not a good model of what would happen if people currently detained were released.

The commission proposes many changes to pretrial decision-making but provides estimates of the effects of only two of them: a return to the bail-setting practices of the first quarter of 2020, immediately after the passage of bail reform; and greater adherence to the recommendations of CJA’s risk-assessment tool. These would reduce jail population by 750–1,100 people, according to the commission.

These estimates are plausible as far as back-of-the-envelope calculations go, but they run afoul of basic realities of criminal justice in NYC in 2022. The commission notes: “New York City’s judges reduced their bail-setting when the state’s bail reforms first went into effect in 2020—but then reverted to greater bail-setting in the second half of the year on otherwise comparable cases.” That decision was not random: it represented a response to the dramatic increase in violent crime that the city experienced in the latter half of 2020, as well as to the overwhelming backlash against the excesses of bail reform. That backlash saw calls for changes from NYC district attorneys, Mayor Eric Adams, and even Governor Kathy Hochul, culminating in several rounds of reforms (the more recent of which is not included in the commission’s estimate).[57]

The commission notes elsewhere in the report that “achieving the projected reductions in jail will depend on political will”—clearly, there is not the political will to return to the sentencing practices of the first quarter of 2020. The commission takes as given that judges should not deviate from CJA’s risk-assessment tool. It notes that judges set bail or remand in 72% of cases in which the instrument recommends “consider[ing] all options,” approximately the same rate for when the instrument does not recommend release on recognizance. In the commission’s preferred configuration, judges would assign bail to violent felony offenders in 20% of cases where the risk assessment recommended release on recognizance, and in 40% of cases where it recommended “consider all options.” It is not apparent where these figures come from, but they imply that a significant percentage of violent felony offenders are neither a public safety risk nor a flight risk—a claim that strains belief.

Case Processing Delays

The commission projects that 500–550 people could be removed from ADP by addressing “case processing delays,” i.e., any procedural hurdles that lengthen the time between arraignment and disposition. The commission’s estimate assumes that the city can achieve a 25% reduction in case delays (and thus a 25% reduction in the pretrial population) by implementing the practices in a recent Brooklyn pilot program, which reduced case delays by 28%. That program, overseen by several authors of the commission’s road map, imposed a series of procedural changes on one Brooklyn courtroom in an effort to reduce the time between arraignment and disposition.[58]

The 28% reduction in the pilot program represents how long it took to dispose at least half of cases in the pilot (181 days), compared with the control group (231 days).[59] But this is not necessarily the same as a uniform 25% reduction in time to disposition—and the commission acknowledges that the effects of the intervention do not uniformly reduce the length of stay.[60] In fact, the effect is driven by a reduction in the time to disposition of otherwise very long cases. For example, the effect of participation in the pilot is most strong among violent felony offenders and those detained. (Notably, the pilot omitted homicide and sex offenders, for comparability reasons; these make up about a third of DOC population as of May 31, 2022.) And there is no discernible difference in the share disposed between the pilot and control prior to about 110 days into the case. By way of comparison, as of May 31, 2022, about 41% of pretrial detainees had been there for fewer than 110 days. I.e., the effect of expediting cases is heterogeneous regarding the underlying offender population, making a uniform assumption of a 25% reduction in population not necessarily reliable.

Expediting cases has already proved a substantial challenge to the court system at large. The January 2020 discovery reform, which requires prosecutors to share evidence with the defense much earlier than in the past, dramatically accelerated the statutorily required pace of case processing. But that did not necessarily translate into an actual faster pace: as of October 2021, prosecutors have indicated that they met the new discovery requirements in just 21% of cases; and in as few as 10% in the Bronx.[61] That rate is for just one phase of the overall prosecution process. Though the commission does not emphasize it, generalizing expediency will require a massive and dramatic investment in the criminal justice system’s capacity—an investment that seems unlikely to come from the same system that routinely attacks the system as racist and illegitimate.

Sentencing Options

On any given day, several hundred people in DOC custody are serving post-conviction sentences of up to a year. These sentences, the commission argues, are counterproductive, insofar as “[r]esearch in New York City, mirroring results elsewhere, has found that jail sentences make it more likely that people will be rearrested in the future, when compared to similarly situated people who did not receive a jail sentence.” This is not, though, a comprehensive account of the evidence: while pretrial detention is, on average, either not crime-reducing or even criminogenic (see above), many studies of the effect of posttrial incarceration in prison find no effect on risk of indicators of re-offense.[62]

The claim that short sentences serve little function is further belied by the scholarly consensus that swiftness and certainty matter more than severity in deterrence, implying that a shortened sentence is efficacious so long as it is meted out reliably and efficiently.[63] Indeed, a system that relied more on short sentences would, ceteris paribus, be less resource-intensive and reduce aggregate incarceration—priorities that, in other contexts, the commission claims to endorse.

Regardless, the commission estimates that ADP could be reduced by 75–125 by changes to city sentencing practices. To reach this estimate, it observes 219 individuals serving city sentences on June 1, 2021. Of those, it drops 26 charged with “violent felony homicide, weapons, sex offense, and domestic violence charges.” Of the remainder, it assumes that 50% could be “diverted from jail to alternatives to incarceration, restorative justice programs, or other interventions, or would have their sentences shortened due to conditional release or earned merit credits.”

On May 31, 2022, 311 people were serving a city sentence, including two who appear to be technical parole violators. About 108 of those were incarcerated for the offenses named by the commission, identified conservatively.[64] Half the remaining 203 would be 102 sentences, approximately within the commission’s range.

That said, the commission provides little justification for its theory that 50% of the remainder could be diverted or have their sentences shortened. Many of the remaining offenders are sentenced for serious crimes, including assault and robbery. The city already diverts thousands of people through its robust Alternatives to Incarceration programs. It is not established that more could be further diverted without imperiling legitimate incapacitative, deterrent, and retributive ends.[65]

Parole Detention

One June 1, 2021, the commission estimated that there were 799 people held in DOC custody on a parole violation because they had committed a new offense or because they had violated some other term of their community supervision (a “technical” violation). “Nearly half (49%),” the commission noted, “faced a nonviolent, misdemeanor, or lesser charge” and thus would almost certainly not be detained under the provisions of the Less Is More Act, the parole reform that had passed in the state legislature but was not signed into law until September 2021.[66] With Less Is More implemented, the commission predicted that ADP would fall by 400–500 people.

In fact, the effects of Less Is More are visible in Figure 2. While the bill’s passage led to a notable decline in daily population, that decline was smaller than projected and seems to have eroded in recent months.

On May 31, 2022, 239 parole detainees were in DOC custody: 16 for technical violations and 223 for new offenses. Of those with new charges, 93 were charged with a new violent crime, including 34 charged with murder or manslaughter. I.e., the predicted reduction in the parole population happened. Yet, conspicuously, overall population fell by only about 250, 50%–63% of the commission’s projections (assuming the entirety of the decline was attributable to the reduction in parolees, which is unlikely). And further gains are no longer possible, except by releasing many serious offenders detained on parole holds.

Priority Groups

The commission identifies three “priority groups” who merit a presumption of non-incarceration: women/“gender-nonconforming” people, people over 55, and mentally ill people. The former two groups, it contends, are at low risk for re-offense, while the latter might be better served by treatment in the community. They estimate that reducing incarceration of the former two groups would reduce ADP by 100 people, specifically by applying the aforementioned reforms to these groups, and then further reducing by 65% the number of pretrial detainees in those who were not facing murder, sex offense, or weapons charges.

Similar estimates obtain with the May 31, 2022, data: 617 detainees are women, have no listed gender, or are 55 and over. Of those, 159 are detained for murder, sex offense, or weapons charges; 65% of the remainder is 297. As with changes to sentencing, the underlying assumptions are questionable. There is no reason to conclude that 65% of the detained populations can be safely released; the true rate may be higher or lower.

Indeed, as should be apparent, many of the commission’s estimates depend on questionable or unfounded assumptions. This is significant because the commission maintains that it is possible to reliably decarcerate to a population below the 3,300-maximum line. If their estimates are even slightly off, daily population will reliably exceed this line.

Even granting all the commission’s premises, current population would still be near the 3,300 threshold. The commission estimated that under its reforms, there would be roughly 2,700 people in DOC custody. At least one of those estimates, parole reform, did not have the promised effects; re-adding the projected reduction of 400 people gives a lower bound of 3,100. If the Covid-19 backlog has even slightly attenuated—which it almost certainly has—the reduction from addressing it will also be attenuated, pushing daily population close to or over 3,300. Even if the commission was right about everything else, these two errors would lead daily population to persistently run up against the arbitrary threshold set for it by the borough-based jails’ capacity.

The other, unaddressed assumption in the commission’s analysis—indeed, in essentially all analyses of the jail population reduction—is that crime will not exceed current levels. If under the near-historically low crime rates that NYC currently enjoys, DOC jails would be filled near to capacity, what happens if crime goes up?

What if Crime Goes Up?

In some senses, the daily jail population in NYC is purely a function of policy. If, for example, the city decided to divert 100% of offenders, jail population would be zero, regardless of the crime rate. But in reality, the level of crime in the city has at least some relationship to the number of people detained in the city’s jails. Indeed, the city’s jail population has declined roughly in tandem with crime since the latter’s peak in the 1990s.[67] Advocates of closing Rikers have had no trouble acknowledging this connection: in their 2019 filing to CPC, for example, MOCJ and DOC estimated that “every one percent drop in crime results in 60 fewer people in jail on any given day.”[68]

That jail population has fallen in line with crime is unsurprising: fewer crimes mean fewer people who need to be detained pretrial.[69] Projections of future jail population have generally relied on the assumption that crime will continue to fall indefinitely.[70] But by the same token, if the crime rate were to rise, we would expect the jail population also to rise. This is, to a first approximation, what happened in 2021 and 2022: as crime rose, jail populations did, too. Policy changes—releasing more offenders—could keep jail population stable as crime rises, but those changes would likely, in turn, increase re-offense and failure-to-appear rates. In other words, policymakers could continue to reduce the jail population as crime rises, but only at the expense of public safety.

What is the relationship between crime and the number of people in jail? Table 3 reports the results of several simple modeling exercises. Specifically, it uses data on daily crime complaints (published by NYPD) and daily populations (as obtained from DOC) between June 2, 2016, and May 31, 2022, aggregated at both the weekly (columns 1, 2, and 3) and monthly (columns 4, 5, and 6) levels. Columns 1 and 4 estimate the relationship between the total number of criminal complaints in a given period and the average jail population over that period; columns 2 and 5 break out complaints by several major categories of offense. Columns 3 and 6 add in controls for the year and week or month (categorical, rather than continuous), to account for annual and seasonal trends in ADP. These models, of course, do not estimate the causal relationship between crime and detained population, but they give us a benchmark of how a change in one corresponds to a change in the other.

Table 3

Estimated Relationship Between Weekly/Monthly Crime Complaints and Average Daily Jail Population

To summarize: in a given week, there is approximately one criminal complaint reported to NYPD for every person held in jail, on average; in a given month, there are about five criminal complaints reported to NYPD for every person held in jail, on average (columns 1 and 4). But that total model explains less of the variation in ADP than do the models that break out counts by crime categories (columns 2 and 5). Before controls, all types of offense except for “property” and “other” offenses are associated with significant increases in ADP at the weekly and monthly levels. After controlling for period trends (columns 3 and 6), there is a roughly one-to-one correspondence between violent and weapons offenses and jail population (and a two-to-one correspondence between weekly sex offenses and weekly average jail population). At the same time, there is a much weaker and nonsignificant relationship between drug and the broad swath of “other” complaints. And property offenses are significantly and negatively correlated with weekly and monthly average population. I.e., after controlling for time period, the relationship between crime and jail population is driven by serious offenses—not by the pettier crimes that advocates often claim produce most incarceration.

One question we can answer with the aforementioned estimates (using models 3 and 6) is: What change in crime would correspond to an ADP of 3,300 in a given week or month? For example, in May 2022, ADP of city jails was 5,481, and there were a total of 44,771 crime complaints. Using the model from column 3, for an ADP of 3,300, there would need to be only 21,888 complaints.[71] I.e., a 40% reduction in jail population corresponds to a 51% reduction in crime.

Does that mean that a 51% reduction in crime is necessary to achieve the maximum population level detainable by the borough-based jails? Not necessarily—again, ADP is a function of policy as much as crime. But it does suggest that—holding policy, as well as public safety priorities, constant—a large reduction in crime is necessary to reach that threshold. Furthermore, it implies that if crime rises above that threshold, the jail population will exceed 3,300.

Put bluntly, there is little reason to believe that crime will never reach such a level again. New York has, of course, enjoyed a historical crime decline over the past three decades. But that decline followed a dramatic increase in crime in the 1970s and 1980s, relative to earlier periods. A return to crime levels reached within living memory—which is not impossible—would swamp a smaller jail system (see Figure 1). Even a modest increase in crime could push an average jail population over the 3,300 threshold.

Advocates of the borough-based jail system are betting on a virtual impossibility: that crime will remain indefinitely below the point at which a much smaller jail system is all NYC needs. This is simply implausible—particularly because crime is at a level in NYC right now where the jailed population exceeds the 3,300 line. A sober analysis of the facts requires us to admit that 3,300 simply is not enough.

Where Else Can New York Find Jail Capacity?

The goal of the transition to the borough-based jails, of course, was always the full closure of DOC facilities on Rikers Island. To this point, this report has argued that the total capacity of the borough-based jails, 3,300 beds, is simply not enough if NYC’s leaders want to minimize the risk of criminal offense or failure to appear. Some may argue that shrinking the jail system and forcing it to run at full capacity all the time will, in turn, limit the abuses of the criminal justice system by constraining whom it can detain. But if history is a guide, the more likely outcome will be either overcrowding or mass release beyond what public safety permits.

Slack capacity in the detention system is not only necessary but an affirmative good. Overcrowding, as advocates of prison reform are quick to remind us, “has always been a serious problem, correlated with increased violence, lack of adequate health care, limited programming and educational opportunities, and reduced visitation.”[72] One solution to this problem is to reduce incarceration, but another is to build more jail capacity. Additionally, larger jails are harder to manage, which can, in turn, empower prison gangs, who provide brutal “governance” in the absence of control on the part of formal authorities.[73] Building more jails will better serve the goal of allowing more people to be detained in their home borough. It is both bizarre and contrary to the Lippman Commission’s original vision that more populated boroughs will, under the final plan, have the same jail capacity as less populated ones. A simple solution to this problem is to erect more jails in the boroughs where more people live.

But where, other than Rikers Island, can that slack capacity be found? The remainder of this section contemplates how, and whether, that goal can be accomplished.

Table 4

Possible Additional Facilities for NYC Jail System

Unused Jails

Currently, New York City’s jail capacity comprises: a) the facilities on Rikers Island; and b) the detention centers in three of the five boroughs. There are, however, several unused jails or prisons that could conceivably be adapted as supplements to the borough-based jail system. (Table 4.)

The two most obvious candidates are Lincoln Correctional Facility and Bayview Correctional Facility, both located in Manhattan. Bayview, which previously housed women detainees, was closed following extensive damage by Hurricane Sandy. Lincoln, a minimum-security facility for men, was closed by Governor Andrew Cuomo as part of his prison-closing initiative.[74] Neither building appears to have been sold, although their sale has been discussed extensively in the past; Bayview would likely need substantial renovations to address the water damage from Sandy. If obtained by the city, Lincoln could house up to 408 additional people; and Bayview, 323.[75]

Harder to obtain are two facilities that have been sold and would need to be repurchased and substantially renovated. Staten Island’s Arthur Kill Correctional Facility operated as a 931-bed prison until 2011, when Governor Cuomo closed it.[76] The site was subsequently sold for $7 million to Broadway Stages, which poured $20 million into renovating it into a film set, now appearing in shows like Orange Is the New Black.[77] Repurchasing the site and restoring it as a jail would likely carry a hefty price tag, but it would return much-needed jail space to Staten Island, which currently has no jail. The other facility, the Bronx’s 900-man Fulton Correctional Facility, was transferred to the Osborne Association in 2013, and will soon operate as a reentry-services center for people returning from prison.[78] Repurchasing this facility would be expensive, would require Osborne to be willing to part with it, and would reduce the availability of reentry services in the city. None of this is optimal or likely, but the facility is nonetheless included here for completeness’ sake.

Build Several Additional, Smaller Borough-Based Jails

If sufficient physical plant does not exist in the boroughs proper, and the city no longer wishes to house people on Rikers, the city will need to build more jails. This is obviously an unpalatable proposition. The borough-based jails were, as discussed, shrunk down considerably from their original size, mostly to suit community objections and political considerations. Since the beginning of construction, community sentiment has remained hostile.[79] Setting up more jails in the boroughs seems, at the very least, a fraught proposition.

One solution would be to erect several smaller borough-based jails, of roughly similar size to Lincoln or Bayview (i.e., 300–400 beds). These would fit more readily into the surrounding environment, and thus draw fewer objections from locals. Doing so would also address the fact that while one goal of the borough-based system is the housing of detainees in their home borough, the jails are currently of equal size, necessarily meaning that some will be housed outside their home borough. (For a sense of the mismatch, see Table 5.)

Table 5

Borough Allocations of Jail Cells and Criminals

The most obvious site for an additional jail is Staten Island, detainees from which are and would otherwise continue to be housed in Brooklyn. When proposing the borough-based system, the de Blasio administration projected that fewer than 200 people from Staten Island would be detained—too few, they believed, to merit a separate facility.[80] But this is faulty logic on several counts. Even if Staten Island continues to play home to just 5% of the offender population, it would still have 300 residents in jail if total daily population hit as little as 6,000. In addition, smaller, community-based jails have all the benefits discussed earlier. It would obviously be useful to have a small facility, for example, so that Staten Island residents’ families would not need to travel to Brooklyn to see them in jail. (If, somehow, the city repurchases the Arthur Kill Correctional Facility, this jail would obviously be less useful.)

Another possible site for an additional small facility would be directly adjacent to the Bronx Hall of Justice (HOJ). The borough jail currently under construction in the Bronx will be erected on the former site of NYPD’s Bronx tow pound, approximately two miles from HOJ. During discussion of the borough-based jail plan, then-Bronx Borough President Rubén Díaz (among others) advocated shrinking the jail from 1,437 beds to roughly its current proposed size and erecting it adjacent to HOJ. CPC rejected this proposal, on the grounds that the proposed layout would have a “long, noncontiguous parcel and a zig-zag floorplan” because not all the proposed land was owned by the city and because a smaller footprint would mean doubling the building’s height. But all these concerns would be addressed by building a 300–400-bed jail on the site, which would add additional capacity in the Bronx and give pretrial detainees easier access to HOJ.

The two other obvious boroughs to receive additional jail capacity are Manhattan and Brooklyn, which hold larger shares of the offender population than they are currently allocated in the borough-based system. Manhattan could offset this with the above-proposed reopening of Bayview and Lincoln; if one or both cannot be reopened, one or more similar-size jails should be erected there instead. An additional small jail in Brooklyn would serve a similar purpose.

Boarding Out

Another solution, which could be done in addition to those already discussed, is boarding NYC detainees in the jails of adjoining counties. Such “boarding out” is not an uncommon practice in NYS. In September 2021, in fact, Governor Hochul and DOC reached an agreement to transfer city-sentenced individuals serving more than 90 days from Rikers to state facilities.[81] As of the most recent figures available, about 259 individuals were boarded out across all of NYS, about 1.5% of the total jail population.[82]

The counties adjoining NYC appear to have room. As Table 6 shows, Nassau, Suffolk, and Westchester Counties collectively are significantly below the capacity for which their jail facilities are rated. In total, something like 3,000 NYC detainees could conceivably be held in these counties on an ongoing basis.

Table 6

Slack Capacity in County Jails Adjoining New York City

There are, of course, pros and cons to this approach. If we prioritize minimizing jail plant in NYC proper and keeping jails off Rikers Island, then jailing people in other boroughs seems like the next least-bad solution. And if the city can negotiate a fixed rate for boarding out, the cost to taxpayers will fluctuate with the number of boarded detainees, rather than requiring New Yorkers to pay for plant upkeep.

On the other hand, sending thousands of people to Long Island and Westchester County, hours from their homes and their hearings, plainly contradicts the goal of detaining offenders in their communities and minimizing the costs and risks associated with transporting them to and from court proceedings. It would also require attorneys, families, and some DOC employees to travel even farther to work. If Rikers is too far away from the city’s courts and communities, then White Plains and Riverhead certainly are.

If NYC refurbished, purchased, or constructed all the additional capacity recommended above, it would net roughly 4,200 additional jail beds, bringing the total capacity of the system to 7,500. Of course, there is little chance of the city repurchasing either the Fulton or Arthur Kill Correctional Facilities and very little chance of constructing four additional borough jails.

In a still unlikely but conceivable scenario, the city could obtain funding to refurbish Lincoln and Bayview Correctional Facilities, erect a much-needed 400-person jail on Staten Island (which could handle overflow from Brooklyn, as happens in reverse now), and place a second 300-person jail adjacent to the Bronx HOJ. One additional strategy would be to expand the Hochul/DOC agreement’s terms, housing every posttrial detainee on Long Island or in Westchester County. That would likely add 500–600 slots to the system (see Table 6), bringing the total to 5,300—just shy of the Lippman Commission’s original 5,500 goal.

The costs of that proposal, of course, are substantial. They are costly in monetary terms: refurbishing Lincoln and Bayview, constructing two more jails, and paying Nassau, Suffolk, or Westchester Counties for the use of their jails. Boarding-out also involves compromising the vision of the borough-based jail system. And for all those costs, it is simply not apparent if 5,300 beds provide enough capacity to detain everyone who merits detention. That would be below where daily population hovered for most of 2022, for example.

At the same time, this is the absolute minimum that NYC must do if it wants to close Rikers. Even that may not be enough, but if the city envisions a future in which reform does not require sacrificing safety and certainty, that is a starting point.

But What About Rikers?

To this point, this report has taken as a given that all the facilities on Rikers Island will be closed within the next five years. The merits of actually closing Rikers are mostly beyond the scope of this report. That said, having established the total inadequacy of the 3,300-bed figure for the borough-based jails and having shown how challenging it will be to find additional capacity, it is worth rehearsing the case for refurbishing and retaining some or all of the Rikers Island jail complex, as a complement to, or instead of, the borough-based jails.

Many arguments for closing Rikers are, at best, tenuous. Advocates contend that the physical plant is decayed beyond use and unsafe—Rikers inmates have even reportedly used parts of the jail to maim or kill one another.[83] But this does not explain why NYC should erect new jails, rather than spending an equivalent amount of money refurbishing Rikers.[84] Advocates argue that Rikers guards are chronically absent or failing in their duties. But this seems more to do with the power of the prison guards’ union than with Rikers’ location per se.[85] Rikers’ distance from courthouses and services might be a problem but not one for which the most intuitive solution is to shutter the entire island and erect new jails with suboptimal capacity.

Further, there are real questions about the viability of actually closing Rikers on time and under budget. As Manhattan Institute senior fellow Nicole Gelinas noted two years ago, the borough-based jails are “the city’s largest complex infrastructure project in modern history,” similar in scope to the reconstruction of the World Trade Center. “New York City’s Department of Design and Construction, which will oversee the jails plan,” Gelinas wrote, “has no experience in such a large-scale project and has performed poorly on smaller projects.”[86] Indeed, NYC comptroller Brad Landers, an advocate of closure, said in August that the city was not on pace to close Rikers by 2027, “either in the trajectory of the number of people being held there or in the speed with which the new facilities are being built.”[87]

If the city needs more jail capacity—if the borough-based jails are not enough—it is irresponsible not to consider refurbishing and continuing to operate at least some of the Rikers jail complex. Indeed, there are already high-quality plans for restoring the island’s buildings as a modern, humane jail, at costs competitive with constructing the borough-based jails.[88]

The law, as Mayor Adams noted, is that Rikers will be closed. But if the law is hostile to public safety, the law should be changed.[89]

Conclusion: 3,300 Is Not Enough

Though the numbers rehearsed above sound like abstractions, the consequences of a too-small jail system are all too real. Either it will mean overcrowding, with the attendant suffering—suffering that can routinely reach the level of unconstitutionally cruel and unusual—or it will require the mass release of dangerous, violent criminals onto NYC’s streets. For civilians or prisoners, adequate jail capacity is a life-or-death matter.

Yet many of NYC’s policymakers do not seem to regard it as such. The Lippman Commission’s beyond-unreasonable projections appear to be a textbook case of policy-based evidence-making: in search of justification for a number, the commission found it. This whole ethos infuses the entire project of closing Rikers. The city will barrel ahead, no matter how unrealistic the project becomes.

As discussed in the preceding section, the city could take intermediary steps to shore up capacity—indeed, it must take them if it remains committed to closing Rikers. It can refurbish and reopen Bayview and Lincoln Correctional Facilities; erect additional small jails on Staten Island and in the Bronx; and commit to sending all posttrial detainees to out-of-city facilities in Westchester County or Long Island. But even these proposals are unlikely to be enough, as they bring total capacity only below where the daily population has been for most of the year. Further procedural reforms may help, but at some point, those will run up against the hard core of serious, violent offenders whose cases take time and whose detention is of paramount importance while they play out.

It is conceptually possible to continue that project in light of the realities above, to build a 21st-century jail system somewhere other than Rikers Island. But there is little reason to believe that NYC’s leadership understands what that would actually entail. Until they do, the project of closing Rikers is doomed to disaster.

About the Author

Charles Fain Lehman is a fellow at the Manhattan Institute for Policy Research, working primarily on the Policing and Public Safety Initiative, and a contributing editor of City Journal. He has addressed public safety policy before the House of Representatives, at universities including Cornell and Carnegie Mellon, and in the Wall Street Journal, Dallas Morning News, New York Post, National Review, and elsewhere. He was previously a staff writer with the Washington Free Beacon, where he covered domestic policy from a data-driven perspective. Lehman graduated from Yale in 2016 with a BA in history.


The Manhattan Institute thanks the Klinsky Leadership Series for its support in the publication of this paper.


Please see Endnotes in PDF


Are you interested in supporting the Manhattan Institute’s public-interest research and journalism? As a 501(c)(3) nonprofit, donations in support of MI and its scholars’ work are fully tax-deductible as provided by law (EIN #13-2912529).