By René Ropac
New York’s bail reform, put into effect in January 2020 and amended in July 2020, eliminated money bail for most misdemeanors and nonviolent felonies. The law also reduced the use of bail even in cases remaining
legally eligible for it through such measures as providing judges with a universal option to order supervised release and an option to order electronic monitoring for certain charges. The mid-2020 amendments
reinstated bail eligibility for certain cases. Later amendments implemented in 2022 and 2023 were modest. As a result, the July 2020 framework essentially remains the basis of New York’s bail reform today. This study is the first to evaluate the recidivism impacts of the bail law as it currently operates. Using statewide court data and quasi-experimental methods, we tracked people for more than four years (50 months) after arraignment—far longer than prior evaluations—to test the impact of reform following the July 2020 amendments. Data Source, Key Measures, and Methods Drawing on non-public court data, we compared people who had bail set or were remanded at arraignment from July 2 to October 31, 2019 (pre-reform), with similar people released during the same dates in 2020 (post-reform). Each person was tracked for 50 months, encompassing both pretrial and post-disposition periods. This design allowed us to examine the long-term public safety consequences of pretrial release in lieu of bail or remand. To test the effects of bail reform on recidivism, we applied a quasi-experimental approach using propensity score weighting. Specifically, we first used logistic regression models to estimate propensity scores and applied inverse probability of treatment
weighting (IPTW) to create an analytic sample in which the distribution of observed baseline characteristics is comparable between treatment and comparison groups. The models included a large number of covariates, such as demographic factors, county and month of arraignment, charge variables, and criminal history measures. For the main analyses, we then employed a doubly robust approach that combined IPTW with covariate adjustment, reducing potential bias in our results. Analyses were conducted separately for: (1) cases seeing the elimination of bail under reform; and (2) cases remaining legally eligible for bail but seeing reduced use of it due to provisions such as the expansion of pretrial supervision. Overall results are reported separately for New York City and the rest of the State. We also examined whether impacts varied for people with or without recent criminal histories (within the past two and a half years) across all counties combined. Impact of Eliminating Bail and Pretrial Detention for Select Cases Effects in New York City: Bail reform's provisions eliminating the option to set bail or remand led to substantial recidivism reductions in New York City. Over the 50-month follow-up period, people released under reform had significantly lower rates of
overall re-arrest (57% vs. 66%), felony re-arrest (33% vs. 40%), and violent felony (VFO) re-arrest (20% vs. 25%), as well as averaged fewer numbers of overall, felony, and VFO re-arrests, compared to people who had bail set or were remanded pre-reform.
Effects in the Rest of the State: The prohibition of bail and detention had no effect in any direction on any of six recidivism outcomes in the suburban and upstate regions (encompassing rates and prevalence of overall, felony, and VFO re-arrest). Effects Among Different Subgroups: Among two "low-risk" subgroups—those with no recent criminal history and no recent prior violent felony arrests—bail elimination yielded statewide reductions in both re-arrest rates and average numbers of re-arrests. The largest benefits appeared among those with no recent criminal history at all. In other words, for those with no recent criminal history, prohibiting the use of bail substantially reduced re-arrest. Yet even people who may have had a prior record, so long as it did not involve a recent violent felony arrest, experienced meaningful recidivism reductions. In contrast, findings for "high-risk" subgroups were more mixed but generally indicated increases in recidivism. Among
those with any recent criminal history, re-arrest rates were unaffected, though the frequency of overall and felony re-arrests increased. For the smaller fraction of people (under 15% in all regions) with recent prior violent felony arrests, release without bail was consistently associated with elevated rearrest rates and more
re-arrest incidents across all six outcomes. Impact of Reducing the Use of Bail in Legally Eligible Cases Effects in New York City: The reduced use of money bail in eligible cases led to modest reductions in rates of overall re-arrest (53% vs. 56%) and felony re-arrest (30% vs. 33%), while violent felony re-arrest rates and the frequency of re-arrest incidents were unaffected. This suggests that reform reduced the likelihood of being re-arrested at least once without altering the overall frequency of re-arrests. Effects in the Rest of the State:
Results were essentially neutral in the suburban and upstate regions. Re-arrest rates and average numbers of re-arrests were nearly all the same for people released without bail and those in the comparison group.
Effects Among Different Subgroups: "Low-risk" individuals with no recent criminal history experienced notable statewide recidivism reductions, while those with no recent prior violent felony arrests showed no significant changes. For "high-risk" individuals with recent criminal history, on the other hand, the reduced use of bail was consistently associated with worse outcomes. Those with recent criminal histories or prior violent felony arrests had higher re-arrest rates and more rearrest incidents across all outcome measures. These increases were larger among people with recent prior violent felony arrests. Study Results in Context Bail reform aimed to reduce reliance on money bail and pretrial detention while maintaining public safety. By these measures, the current law can be considered a success. However, the recidivism increases among a relatively small subgroup of people with recent criminal histories should not be overlooked. The amendments to “initial” bail reform put into effect in July 2020 sought to allay public safety concerns primarily by reinstating bail for a range of cases. Yet, the present study shows that the detrimental impacts among subgroups of “high-risk” individuals defined by serious recent criminal history—previously observed in DCJ’s studies of initial bail reform—remain evident under the current law as well. Another key finding is the contrast between
New York City and the State’s suburban and upstate regions; in New York City, the prohibition of money bail (or "mandatory release") was associated with declines in both the likelihood and frequency of re-arrest, while rearrest outcomes in the rest of the State remained unchanged. Several factors may explain this divergence: possible baseline differences in pre-reform release rates (courts released fewer people prior to reform outside New York City); differences in the populations processed in each region; or potential gaps in the implementation and effectiveness of pretrial supervision and other support programs outside New York City