The impact of bail reform in the United States

In recent years, many jurisdictions throughout the United States have taken steps to reform their cash bail systems.  Proponents of bail reform advocate for reducing or eliminating the use of monetary bail to reduce jail populations and reduce income disparities regarding how bail is applied. However, opponents of bail reform argue that reforms have been extreme, countering that they could result in more defendants failing to appear for court hearings and more defendants committing crimes while on pretrial release.

Among the many states implementing bail reform include California, New York, New Jersey, Vermont, Illinois, and Alaska. In addition, many individual jurisdictions have implemented their own bail reform policies, such as Atlanta, Philadelphia, Houston, and Chicago, among others. There is no consensus about what should replace cash bail, though some options include implementing strategies to expand pretrial supervision and services, and requiring pretrial risk assessment instruments to inform pretrial release decisions. However, most research has not found decreases in the rates of missed court dates or rearrests associated with either strategy.

An Overview of Bail Reform in the United States

A working paper published by Harvard University in December 2021 gives an overview of what bail reform looks like for 50 states as well as 12 jurisdictions across the nation (all 12 jurisdictions had high quality and publicly available impact evaluations of their reforms). Authors begin by stating the four main categories of actors who implement bail reforms in the U.S. (i.e., courts, prosecutors, city and county governments, and state legislatures), finding dozens of potential permutations of bail reforms. When examining trends over time across the jurisdictions, the authors saw a significant change in the re-arrest rate among people on pretrial release associated with a decrease in the number of people being detained pretrial. The reforms these jurisdictions have adopted vary significantly and there are different levels of bail reform: county-/city-led; court-led; prosecutor-led; and state-led.

Courts at the state and local level can amend court rules or issue opinions that impact cash bail. Some examples include establishing a presumption of release on one’s own recognizance for certain crimes (e.g., Indiana and Ohio), considering a defendant’s ability to pay when setting bail (e.g., California, Maryland, Missouri, and Fort Bend County TX), reducing bond requirements (e.g., Cook County IL and Travis County TX), or requiring risk assessments as part of the pretrial decision-making process (e.g., Virginia, Kentucky, Colorado, San Francisco, and Los Angeles County).

Prosecutors commonly enact bail reforms by refusing to request cash bail for people charged with certain crimes, such as nonviolent misdemeanors (e.g., Cook County IL, Prince George’s County MD, and Manhattan NY). Another example of a prosecutor-led bail policy change occurred in Philadelphia in response to the COVID-19 pandemic, when the District Attorney announced that the jurisdiction would only release on one’s own recognizance or $999,999 in cash bail.

Players in city and county governments may enact policies that can be separated into two main types. First, they can change local laws that call for things like the elimination or reduction of cash bail/bond for certain crimes (e.g., New Orleans and Atlanta, respectively). Second, localities can expand pretrial services to manage people in the community while they are awaiting trial (e.g., Yamhill County OR and Washington, D.C.).

The impact of state legislatures varies greatly state by state. For example, a state can create guidelines for what judges should consider when imposing bail, such as flight risk and dangerousness to the community (e.g., South Carolina). Other states have adopted multiple reforms that range from establishing a presumption of release, to adoption of a risk assessment, and expansion of pretrial services (e.g., New Jersey).

Pretrial Services and Supervision

As jurisdictions have moved away from cash bail systems, more people are eligible for release on their own recognizance or via a bail bond. These jurisdictions often also expand pretrial services to enhance monitoring capacity and increase access to services to encourage people who are released pretrial to desist from crime. The conditions vary by jurisdiction and include things like drug testing, electronic monitoring, court-ordered supervision, court date reminders, and behavioral health supports. Many people think pretrial supervision is essential to bail reform because judges will release more people pretrial if services are available to ensure community safety and court appearances. However, pretrial supervision has come under scrutiny due to concerns about the financial costs and time burden it imposes, as well as the lack of research supporting its effectiveness.

Justice Reinvestment Initiatives (JRIs) focus specifically on diverting funds to community-based providers instead of detaining people pretrial. JRIs tend to use computerized risk algorithms, or risk assessment tools, to predict which defendants are most likely to “succeed” on community supervision.

Risk Assessment Instruments

The use of risk assessment tools is one of the most common bail reform measures discussed, and dozens of jurisdictions have adopted risk assessment tools.  Risk assessments rely on an algorithmic approach to combine item ratings into risk scores, the results of which should inform pretrial decisions.  In theory, those rated as low-risk can be released without bail or any other conditions, and those rated as higher-risk may benefit from extra interventions. Risk ratings are are usually based on similar factors such as prior convictions, incarceration history, failures to appear, violent offenses, pending cases at the time of arrest, and age. The risk scores and accompanying recommendations do not replace judicial discretion, but are used to guide their decision as to whether to release or detain a defendant pretrial.

Sometimes state courts order the statewide adoption of a risk assessment tool (e.g., Arizona), but cities and counties have also taken the lead on implementing risk assessments in some places. For example, while California has not adopted a statewide risk assessment, at least 49 counties in California use a pretrial risk assessment tool to inform pretrial release decisions. For example, San Francisco, and Los Angeles County are two of many jurisdictions that rely on a risk assessment tool called the Public Safety Assessment (PSA), which guides judicial decision-making when deciding to release someone pretrial. Some states/jurisdictions, such as Colorado, Virginia, and Washington, D.C. have created their own risk assessment tools, while others have adopted already established tools. The PSA is one of the most widely used risk assessment instruments. So far, research on risk assessments does suggest that they have strong predictive validity in the aggregate, but it is still unclear which instrument is the ‘best.’ Further, it is hard to know how accurate these instruments are for individual subgroups.

Arguments for Bail Reform

Proponents of bail reform argue that reducing or eliminating bail does not increase the risk of reoffending. Proponents commonly cite a large-scale study published in 2015 that spanned 15 years that found no effect of bail reform at the aggregate-level. The study examined over 100,000 pretrial felony state defendants from the 75 largest counties, 70,000 of whom were released. The average pretrial release period was 122 days, or about four months. Researchers found that a few different factors (e.g., criminal history, initial charge, age) were reliable in predicting future failures to appear, the probability of pretrial re-arrest, and the percentage of defendants who can be released safely before trial.

Of those released, 16% were re-arrested for any reason, 11% were re-arrested for a new felony, and 1.9% were rearrested for a violent felony. While these percentages may appear low overall, it is important to understand the caveats behind these findings. For example, the data show that those charged with violent crimes were not more likely to be rearrested pretrial than their non-violent counterparts, but when arrested, they were more likely to be re-arrested for violent crimes. Further, when the authors looked at differences in violent re-arrest probabilities for those released conditional on the initial charge, they found that murder defendants had the highest violent crime re-arrest rate at 6.4%, followed by robbery (5.8%), rape (3.2%), and assault (2.9%).

Results showed that criminal history (i.e., prior convictions and re-arrests), initial charge, and younger age were key predictors of re-arrest or re-arrest for a violent  crime. Thus, there are identifiable groups of people arrested whose probability of committing a new violent crime is far higher than others. Past failure to appear also predicted probability of being rearrested for a nonviolent crime, though it was not a good predictor of future violent crime. Considering failure to appear, the authors found that flight risk varied by age, but this effect was less pronounced when considering violent crime. Prior record and prior failures to appear were both consistently related with flight risk, and this was evident throughout all offense and age categories.

Arguments Against Bail Reform

Bail reform has come under scrutiny due to concerns about public safety and unintended consequences well as the lack of research supporting the effectiveness of risk assessment tools and pretrial services. A primary argument against bail reform is the potential impact on public safety and increases in crime, particularly violent crime. For example, findings from Los Angeles County showed that those being released on citation-only were the most likely to recidivate as well as fail to appear for court. Conversely, it appears that people being released on bail were the least likely to recidivate and also the most likely to appear in court (followed closely by those released on bond). While limited, the findings suggest that bail might be more effective than ROR and even more effective than citation-only releases. Unfortunately, risk assessments are used in dozens of jurisdictions where there is no high-quality data about the assessments’ impacts, so it is unclear whether risk assessment tools are sufficient (especially on their own) in achieving the intended goals of reducing reliance on cash bail.

While some research suggests that bail reforms do not meaningfully impact crime, others found small increases in pretrial re-arrest rates. For example, in the 2015 study cited above, aggregate trends suggested that high bail had no effect on pretrial crime. However, there are important caveats to this finding. While pretrial re-arrest rates were low overall, a close look at the article revealed that some categories of offenders may re-offend at higher rates than others, and the authors suggest that there are likely certain groups of offenders who are more likely to recidivate than others. The data showed that when arrested, those charged with violent crimes were (not surprisingly) more likely to be re-arrested for violent crimes. Further, when the authors looked at differences in violent re-arrest probabilities for those released conditional on the initial charge, they found that murder defendants had the highest violent crime re-arrest rate at 6.4%, followed by robbery (5.8%), rape (3.2%), and assault (2.9%).

The study’s conclusions are more complex than policymakers may make it seem. In particular, the authors mention that judges could perhaps release very specific groups of offenders (not all offenders, and especially not violent offenders); these might include nonviolent older defendants, people with clean prior records, and people who commit fraud and public order violations, without increasing danger to the public. In contrast, the authors state that others ought to be held pretrial based on dangerousness, particularly violent offenders and those who are young. Additionally, there have been vast changes in bail reform particularly after the COVID-19 pandemic and it is unclear whether results are still applicable to present day. Specifically, this study was done prior to the more recent changes in bail policies, such as those implemented in California in 2018 and New York in 2020.

Takeaways

There is large variation across jurisdictions regarding what bail reform actually is and how it is implemented, and reform efforts also change over time. Thus, it is a challenging endeavor to isolate the causal impacts of bail reform, and effectiveness of reforms can vary based on many factors such as specific guidelines imposed, the actors implementing them, and jurisdiction-specific differences. At present, more research is needed to understand the overall effectiveness and unintended consequences of bail reform, specifically as it relates to the most recent policy changes. Unfortunately, one barrier to advancing research on the topic is that of the many jurisdictions making reforms to their cash bail systems, few seem to be collecting meaningful data on the impacts of these reforms.

Despite the limited number of rigorous studies available on the topic, jurisdictions across the country continue to implement bail reforms. Jurisdictions are encouraged to collect meaningful data related to reforms, such as the percentage of people released on recognizance vs. those released with conditions. Future research should continue to leverage this data as it becomes available to guide policy decisions on this topic.