The data prove that other forces drove crime higher in 2022.
No criminal legal issue has generated as much controversy in New York as its 2019 bail reform law. Alarmed by highly salient tragedies like the death by suicide of Kalief Browder and broader concerns about the racial and economic biases of pre-trial detention, the state legislature sought to limit judges’ ability to impose money bail on a wide range of defendants. Since the 1970s, judges in New York have had the option to avoid imposing cash bail on almost all defendants; the reforms (since amended twice) required them to avoid doing so for almost everyone charged with misdemeanors, non-violent felony crimes, or even a few violent felony offenses.
Law enforcement, conservative politicians and numerous media outlets began to attack the law aggressively the moment it was passed as part of the state budget in April 2019. They went so far as to blame the reform for crimes that happened in 2019, even though the law did not go into effect until Jan. 1, 2020. And the attacks were effective: While around 55% of New Yorkers supported the reforms in April 2019, that support had fallen to about 35% by the middle of January 2020, barely two weeks after the reforms officially started. Then-Gov. Andrew Cuomo leveraged the COVID crisis to force the legislature to roll back some of the reforms in April 2020, and Gov. Kathy Hochul pushed through more roll-backs in 2022. In her recent State of the State speech, Hochul promised to try to roll them back even more in 2023.
Judges were already releasing a lot of people without bail at the start of 2019, and those rates rose over that year, well before any change in the law.
In the early days of the reforms, there was little to no systematic data, so the debate centered entirely on salacious anecdotes. The April 2020 rollbacks required the Office of Court Administration to provide regular data, but only starting from Jan. 1, 2020 — making it impossible to use to understand what bail reform changed.
In the fall of 2020, however, the state’s Department of Criminal Justice Services released detailed data on case outcomes from 2019 to 2020 (and now extended to 2021). At last, we had data that could shed light on what, exactly, bail reform had actually wrought.
Three important points jump out from this data, which taken together push back strongly against the claims that bail reform drove crime increases in 2020, suggest that the law has likely been a real net social positive, and thus that the on-going efforts to keep rolling back the laws are misplaced.
1. The impact on releases
The first, and perhaps most surprising point, is that for most cases, the impact of bail reform was actually fairly slight. Judges were already releasing a lot of people without bail at the start of 2019, and those rates rose over that year, well before any change in the law. This upends a common critique of bail reform, which attempts to blame the reforms for any crime committed by someone awaiting trial who had been released without bail. “Without the reforms,” the argument goes, “this person would have been locked up!”
As Figure 1 makes clear, this is simply not true, at least for New York City. (I focus on New York City, not the state as a whole, simply because the NYC data is the most reliable, for reasons too needlessly arcane to dive into.) For misdemeanors, which make up about 70% to 75% of all New York City cases each year, judges were already releasing over 85% without bail at the start of 2019, which stabilized around 95% after the rollbacks.
In other words, even if all that increase from 2019 to 2021 is due to bail reform — and given the trend lines, that seems unlikely — then only about 10% of all misdemeanants released without bail in 2020 or 2021 would have faced bail.
It’s more likely than not that someone released without bail in 2020 or 2021 would have been released without bail even without any change in the law.
It's true that the post-reform jump is bigger for felonies, but even for those, judges in 2019 were consistently releasing nearly half of those charged without bail. If we zero in on just the felonies that became ineligible for bail under the reforms, judges in 2019 were releasing perhaps as many as 60% to 70% without bail in early 2019, rising to nearly 70% to 80% by the middle of the year. (It’s somewhat hard to say with precision, because it can sometimes be a bit messy to determine whether some cases are rendered ineligible for bail under the reforms.)
Intriguingly, over 2020 and 2021, there is also a steady increase in the percentage of those who remain eligible for bail who are released without bail. In other words, non-bail releases rose even for those unaffected by reforms. Obviously, other factors were at play.
All of which is to say: It’s more likely than not that someone released without bail in 2020 or 2021 would have been released without bail even without any change in the law. Which means simply blaming reforms each time someone on non-bail release commits a crime is fundamentally dishonest.
2. The impact on crime
While the effect of reform on non-bail releases was slight, the rate at which people on non-bail release were rearrested rose sharply in 2020. In 2019, about 24% of all people released without bail were rearrested for some new offense, about 5.5% specifically for a crime of violence; in 2020, those rates rose to about 30% and 9%, respectively. Those are substantial increases.
But they likely had little to do with bail reform.
After all, something else happened in 2020 that affected crime trends in New York City (and everywhere else): COVID. Figure 2 plots the percent of people released without bail who are rearrested while awaiting trial (the month is the month they are arraigned, not the month rearrested), either for any crime or specifically for a violent felony offense.
The pattern is clear in both graphs: There was a sharp rise in rearrests leading up to April 2020, when COVID started peaking in the city, and then a sharp decline thereafter. Tellingly, the percentage of people on non-bail release in 2021 who were rearrested returned to 2019 levels: about 23% for any crime and 5.6% for a crime of violence.
It’s true that rearrests rose sharply in the first quarter of 2020, before COVID and the lockdowns. Perhaps, one might argue, in the absence of the lockdowns and the resulting drop in non-lethal crime, rearrests would have stayed high or even gotten worse over 2020. But the reversion in 2021 — as COVID faded but the reforms remained in effect — casts significant doubt on this idea.
Also, note that what bail reform data measures is rearrests, not convictions. For violent felonies, these are two very different numbers. In the state DCJS data, over 75% of the violent felony arrests arraigned in 2020 have been resolved, and of these only about 10% have resulted in a violent felony conviction, and only about 25% in any sort of felony conviction. More strikingly, nearly half (around 45%) have either been dismissed, acquitted, or failed to make it out of the grand jury. (Of course, plea bargaining makes it tricky to know what a conviction means — a plea to a non-violent felony or misdemeanor, or even a dismissal, doesn’t always mean that the initial arrest charge overstated what happened.)
3. Costs (without) benefits analysis
Finally, it’s critical to note that the DCJS bail reform data actually provides no data on the benefits of reform. The only thing the data measures is failure: How many people on release get rearrested? Debate that relies solely on this data is thus just a debate over whether the failures are “too big” — which is a debate inherently stacked against reform.
Prosecutors and police are free to point to anyone arrested while on pre-trial release to argue against reforms. But defense attorneys are not allowed to publicize clients who succeed while on pre-trial release, unless their client consents to that — and few people want to be the ‘face’ of bail reform’s success.
The justification for bail reform, of course, is that avoiding pre-trial detention allows people to maintain ties to the community and family they'll return to, to find or keep jobs, to start treatment, and so on, all while avoiding the very real harms that come from being locked up in places like Rikers Island. Yet none of those benefits are in the bail reform data, and so the empirical assessments of bail reform tend to just gloss over them.
There’s an ethical-political challenge here as well. Prosecutors and police are free to point to anyone arrested while on pre-trial release to argue against reforms. But defense attorneys are not allowed to publicize clients who succeed while on pre-trial release, unless their client consents to that — and few people want to be the “face” of bail reform’s success. The costs are easy to broadcast, the benefits are not.
To be clear, the data suggests that bail reform is performing well even when we just look at rearrests. But bail reform that causally leads to a bit more crime could still be defensible if enough social benefits flow from it — especially because many of these (unmeasured) benefits, like starting effective drug treatment, can themselves lead to less offending and victimization in the future. Which are, of course, other sorts of benefits that bail reform data cannot see, since it cannot measure the crimes that end up not happening.
In short, it seems unlikely that bail reform contributed meaningfully to the rise in violence that New York experienced in 2020, and certainly not to the degree that its detractors have suggested. The ongoing attacks on reform rest on claims that are, by and large, empirically unfounded. That’s not to say that attacking bail reform is not good politics. After all, the costs of punitive policies are disproportionately borne by small, politically marginalized communities of color (Black communities in particular), whose pain is not all that politically salient. But it seems pretty clear that attacking bail reform is bad policy.