Among the economic disparities brought to light by protests in Baltimore last week, one was a number: $500,000. That was the bail set for a protester who turned himself in, and his family had no way to pay it (an online fundraising drive has raised just over $6,000). Meanwhile, all six police officers charged in the death of Freddie Gray on Friday faced bail amounts of $350,000 or less, which they were able to post for their release.
It’s a familiar story in Baltimore, where 87 percent of inmates at the jail are there pretrial. Twenty-nine percent of those have actually been deemed likely to show up in court. Nationally, the Bureau of Justice Statistics estimates that 62 percent of the U.S. jail population — there are about 750,000 inmates at any given time — has not been convicted of a crime. Many of them aren’t a danger to society. They just can’t afford to post bail. And those weeks or months of incarceration can wreak havoc on their lives, as jobs are lost, children neglected and cars repossessed.
Cutting down on the pretrial population can take the pressure off overcrowded jails. The key is to keep in custody only those who either won’t show up for court without posting bail or who are deemed likely to commit more crimes if released. And in fact, there are now sophisticated ways of figuring that out, using a defendant’s criminal history and community ties to predict future behavior: a risk assessment, which the judge uses as a guide in deciding whether to set bail, and for how much.
As politicians have begun to look critically at prison systems that absorb huge chunks of state budgets, often because of ballooning caseloads of minor offenses, the concept of risk assessment is catching on across the country. Since 2012, 20 states have passed laws creating or improving the use of risk assessments for pretrial release. Most notably, New Jersey adopted sweeping reforms to its bail system last year, with Gov. Chris Christie (R) working alongside civil rights groups to push them through.
Some counties in Maryland use risk assessments, as well, but without a statewide system, data is spotty and pretrial risk assessments often have little bearing on the size of a bond. Maryland released an extensive report late last year recommending improvements to the system, but the legislature hasn’t yet taken action to implement them. With a new focus on the role of jails on the condition of Baltimore’s poorest neighborhoods, the story next legislative session might be different.
"Jail populations have doubled since 1983, and lengths of stay are driving them up. Research shows we have the wrong people in and the wrong people out,” says Cherise Fanno Burdeen, executive director of the Pretrial Justice Institute, who decried the disparity in bail bonds set in Baltimore. Recent episodes of police brutality, she adds, have intensified attention to the issue. “We have seen a dramatic shift in the public discourse around jail use and pretrial detention of those who don’t need to be detained.”
The unlikely front-runner in progressive bail policy: Kentucky, which started its pretrial system nearly four decades ago and remains the only one in the country to operate statewide. (New Jersey will be second.) Kentucky's system has been widely recognized as the best way to detain people not based on whether they can afford to pay their way out, but whether they’re likely to come back.
How to predict whether a criminal will come back
Sunlight streams through the high, small windows in the Jefferson County Jail, down onto the floor of the booking room, where sullen inmates wait to be called. At one booth, a nonuniformed officer questions a tall, rangy man about the domestic violence incident for which the police had picked him up — and asks what might happen if he went back home.
Why did your ex-wife show up at your house? If you return, will she be there? Do you have any debts? Do you have a place to live? A way to get around? A job? Kids? How about a cellphone?
“It would be really stupid for me to do anything,” says the man, quietly, his cheeks shiny with slow tears. He’s on probation until 2017, he explains. Just trying to stay straight.
At the end of the interview, the pretrial officer types the man's name into a computer terminal in a secure case in a back room — the National Crime Information Center, a repository of all criminal records on file in every state — and then staples everything into a packet, along with data from Kentucky’s own statewide criminal database. The man is assessed: low, medium or high risk, which describes how likely he is to skip his court date.
Everyone who is arrested must be evaluated within 24 hours of being booked. Every few hours, when the staff has a pile of packets, they call a judge, who decides whom to let go and whom to keep overnight for arraignments the next morning. That’s when another judge looks at a room full of bleary-eyed inmates, who come up to the window one by one, and answer questions until the judge feels she has enough information to set bail or let them go.
The stapled packet is key to this process. It contains a data analysis that essentially predicts whether any given offender will need prodding to come back to court, and whether he’s likely to reoffend if he’s released, allowing the judge to make better decisions about how to use that precious space in the jail.
Because in the back of the judge’s mind, she knows the jail doesn’t have many beds to spare. Judges each get a summary every day of how many people are in jail, and for what reason. Though overcrowding has eased somewhat since Kentucky passed sentencing reforms in 2011, the tide of drug offenders keeps rising -- heroin addicts are detoxing in large dorm rooms upstairs, often with overflow beds on the floor.
"The safest thing to do might be to put a million-dollar bail on everyone. I do that, and our jail explodes. It absolutely explodes.”— Judge Ann Bailey Smith
If the offenders haven’t hired a private lawyer to swoop in and assist, they’ll have a public defender, who confers quietly as the interrogation proceeds. “They’re low-risk, your honor,” the defender might protest if the judge makes the conditions of release severe. Then, she might relent, and just require the accused to submit to home-based monitoring rather than setting a bond too high for him to pay.
"The safest thing to do might be to put a million-dollar bail on everyone,” Judge Ann Bailey Smith says. "I do that, and our jail explodes. It absolutely explodes.” She tries to release as many of the “low risk” defendants as possible. And that’s made a difference: While more than six in 10 jail residents nationally are there pretrial, in Kentucky the number is more like 43 percent.
But releasing low-risk defendants isn’t just good for jail populations. It’s also better for the people who are picked up on minor offenses and then face days away from their work and family. That’s what Smith thinks about when she meets people at morning arraignments.
"I’m not trying to get someone to lose their employment. And some of them will come in and look pretty desperate and say, ‘Oh my god, I’ve been waiting for this job for months,” says Smith, who used to be a public defender herself. “They lose their financial backing for their family. Just spending a short time in jail could be devastating to a person. They could lose their car. They could lose their home.”
Kentucky’s road to making pretrial smart
The concept of using risk assessments to filter the people kept in prison before they’re charged emerged with the Manhattan Bail Project, an early-1960s pilot program in New York City, which found that many people would return for their court date even without posting a surety bond. A couple of decades later, a new federal law required judges to take into account several factors in determining whom to release and under what circumstances. But in most states, judges didn’t have much information to go on in making those decisions.
Enter pretrial evaluation. Kentucky had a head start on its system, which began in 1976, for two reasons. First, it’s one of the only states in the country that has a unified court system, which means that county courts share data on offenders that allows every booking and failure to appear in court to be tracked throughout the state. And second, it’s always banned the practice of allowing private businesses to lend people money to post bail, which means it never had to fight an entrenched interest that profits from the status quo.
“As long as there's a private entity that makes money off a system, reform efforts usually don't work, because they have lobbyists and want to protect their income." — Tara Klute, director of Kentucky's Pretrial Services program
“As long as there's a private entity that makes money off a system, reform efforts usually don't work, because they have lobbyists and want to protect their income,” says Tara Klute, who has run Kentucky’s pretrial services agency since 1995.
Still, for decades, Kentucky’s data collection was hit and miss. It was hard to organize that much data -- from the smallest misdemeanor offenses as well as felonies -- when it often hadn’t been entered correctly in the first place. Moreover, judges often didn’t trust the predictions it generated and wouldn’t factor them into their bail decisions or sentencing.
Through the 2000s, the Kentucky prison population just kept getting bigger and more expensive — the corrections budget was $440 million in 2010, up 214 percent over 20 years — to the point where governors started to release people early because the state couldn’t afford to keep them in jail.
Finally, in 2011, state lawmakers passed HB 463, a comprehensive package of criminal justice reforms including reductions in sentences for minor drug crimes and improvements to the parole system to prevent recidivism. One of the key elements was a requirement that judges consider risk assessments in deciding what to do with a defendant before trial, and provide a written justification if they don’t go along with its recommendation.
Meanwhile, the Kentucky pretrial services agency — which now employs nearly 300 people — was cleaning up its act. Decades of data needed to be organized and verified. The risk assessment needed to be upgraded and retested to make sure it still worked. And in just the first year after Kentucky’s reforms passed, release rates for low- and moderate-risk inmates increased by between nine and seven percent. Judges also started releasing more people without imposing bail, and the pretrial jail population declined markedly.
But there’s more room for improvement -- because at the end of the day, the decision to let a prisoner go free before trial is still ultimately up to a judge. And sometimes, judges need convincing that statistics can be better at protecting public safety than they are.
Even in Kentucky, it’s still a compromise. The pretrial services agency will say it doesn’t believe that setting bail is ever really effective — either a person’s going to come back to court or they’re not, and the agency is able to predict who will or won’t, based on the behavior of everyone gone before. Perhaps counterintuitively, the data show that people who commit the most serious crimes are the most likely to return for their court date, no intervention necessary. But even some of the judges who fully support using the risk assessment will balk at what it says in those circumstances.
"I think you’d be hard put to find a judge anywhere in the country who’s gonna release a murder suspect,” Jefferson County Judge David Bowles says. “If someone is charged with murder, even though he might be assessed at low risk, I think the fact that he’s charged with murder at a minimum makes them a flight risk.”
While judges like Smith and Bowles might be sympathetic to prisoners’ cases, Kentucky's county judges are elected, and their client is still the public. “What the public understands is that so-and-so got arrested, and they’re in or out of jail,” Bowles says. “It’s up to us to help people understand that bail is not a punishment.”
“What the public understands is that so-and-so got arrested, and they’re in or out of jail. It’s up to us to help people understand that bail is not a punishment.” — Judge David Bowles
Some local judges still see their role as guardians of morality and public order and will impose draconian monitoring requirements on pretrial defendants out of a sense that the person needs to be taught a lesson.
"Some of my colleagues believe that no matter what you're charged with, your bail has to include drug testing,” says Karen Thomas, a judge in Campbell County who trains others on using the risk assessment. “'No matter what, I care that you're out on bond. And I care whether you're using drugs or not, so I'm going to make you take drug tests over and over again.' And it hits the indigent. That's not your job at this point."
To try to help bolster judges’ confidence in the risk assessment, the Kentucky pretrial services agency worked with the Laura and John Arnold Foundation — which has also attracted attention for its benefactors’ efforts to curtail public pensions — to test-drive a whole new risk assessment, starting in 2013. In just the first six months, release rates increased, and the crime rate among defendants on pretrial release declined by 15 percent.
Now, the Arnold Foundation wants to roll out the risk assessment across the country. And despite some officials’ reluctance to adopt a one-size-fits-all tool, Anne Milgram, the group’s vice president for criminal justice, says most jail populations are actually pretty similar when it comes to pretrial behavior. As New Jersey’s attorney general, she had started the process of knitting the counties' data into a statewide system.
“Every single county was like, ‘No, we’re different.' But when I looked, they were different in the same way,” Milgram says. “The goal, and we’re closer to being there, is to make it available nationally, no cost."
After almost a decade of overhauling and refining its pretrial system, the Jefferson County Courthouse is still a chaotic place, with people milling around outside courtrooms and waiting in lines to pay fines. But fewer of them make long stops in jail on the way, because data predicted they would come back on their own.