Realignment and Recidivism Revisited: A Closer Look at the Effects of California’s Historic Correctional Reform on Recidivism Outcomes

Abstract

California’s 2011 Public Safety Realignment has received considerable attention nationally as a watershed moment in the movement to downsize prisons. The present study leverages data collected in 12 California counties to provide the most comprehensive examination to date of how Realignment has impacted recidivism for the key offender groups targeted in the reform. We find small to modest increases in rearrest in three of four groups targeted in the reform. The fourth group experienced moderate decreases in rearrest. Moreover, all groups experienced decreases in reconviction, which gives credence to the idea that a significant reprioritization of who should be in prison can positively affect public safety. These findings point to the complex ways that reforms like Realignment can affect custodial and community-based supervision systems by changing incentives for law enforcement and the people who supervise offenders. Our conclusions discuss the implications for other states and systems considering similar reforms.

Keywords

Public Safety Realignment correctional reform recidivism prison downsizing

Introduction

California’s Public Safety Realignment, also known as AB 109, was a historic reform aimed at downsizing prisons in the largest prison system in the United States. The immediate stimulant for the reform was a U.S. Supreme Court ruling in two cases dealing with the constitutionality of the system’s health care delivery to prisoners, which upheld an order by a lower court to reduce the prison population to 137.5% of prison design capacity. However, the broader context of chronic overcrowding, soaring costs, and persistently high recidivism also provided incentives to policymakers to reform the state’s sentencing and correctional policy.

Public Safety Realignment (Realignment) reduced the prison population by transferring responsibility for managing lower level felony offenders from the state to the counties. The prison population declined dramatically, by more than 27,000 inmates, during the first year following Realignment. Although jail populations increased by about 9,000 inmates following Realignment, the overall level of incarceration in the state declined (Lofstrom & Raphael, 2016). The overall decline in incarceration associated with Realignment resulted from changes in both front-end sentencing policies and back-end revocation policies (Grattet & Bird, 2018). The front-end changes included requiring that certain types of lower level offenders who had no history of serious, violent, or sexual offending serve their time locally. The “back-end” changes included a requirement that all revocations for supervision violations for people released from prison be served in county jails rather than state prison. Together, these changes mean that California counties now shoulder a greater burden for incarcerating and supervising felony offenders in the state. Realignment has generated significant national attention because it resulted in large declines in the state’s prison population without dramatic increases in crime (Lofstrom & Martin, 2015; Lofstrom & Raphael, 2015) and because it prioritizes the use of prison—the most costly form of correctional control—for serious and violent offenders (Grattet et al., 2016).

Dubbed “The Great California Prison Experiment” (VanSickle & Villa, 2018), a small body of research has begun to accumulate about Realignment’s impacts on crime (Lofstrom & Martin, 2015; Lofstrom & Raphael, 2015), local correctional systems (Grattet & Bird, 2017; Grattet et al., 2017, 2016), and recidivism (Bird & Grattet, 2016). To date, other than a preliminary report by the present researchers (Bird et al., 2017), research on recidivism has been limited to only a segment of the population affected by Realignment, people released from prisons. This focus neglects the larger group of offenders who, after Realignment, were required to serve time locally in county jails and/or be supervised by county probation departments.

In addition, prior work is limited in that it only covered the initial months after the reform, when many counties were contending with managing a group of offenders with more serious criminal histories than the populations they were used to supervising. Nonetheless, this early research focusing solely on prison releases shows that recidivism rates remained high in the aftermath of the reform; for some groups of offenders, recidivism rates were somewhat better and for others they were somewhat worse than prior to the reform. It remains to be seen whether the findings of research on prison releases and the findings from the early days of Realignment hold for later release cohorts and for offenders who were realigned to local correctional systems. After all, a key premise of Realignment was that “the counties can do it better” (Lofstrom et al., 2012) when it comes to lower level offenders and thus far the story is incomplete.

The present study evaluates the recidivism outcomes for all of the offender groups affected by Realignment, each of which faced transformed in-custody and in-community conditions under Realignment. We first provide an overview of the policy context and the research findings to date. Then, we describe the unique source of data that allows us to extend prior research by elongating the post-Realignment time frame and by including locally managed subpopulations that have not previously been investigated. We present our findings and conclude with a consideration of how the current research contributes to a retrospective assessment of California’s prison downsizing experiment.

Policy Context of Realignment

In 2011, the state of California passed one of the most far-reaching criminal justice policy reforms in recent U.S. history. Known as Public Safety Realignment, this change marked a new era for corrections and rehabilitation in the state. Realignment has been referred to as “revolutionary and sudden” (Weisberg, 2011), “the most significant correctional reform in decades” (Misczynski, 2011), and “the biggest penal experiment in modern history” (Santos, 2013). The reform was motivated by a U.S. Supreme Court decision upholding an earlier ruling by a three-judge panel that California’s prison system could not adequately meet the health care needs of its inmate population due to overcrowding. The court ordered the state to reduce the prison population by tens of thousands of inmates. Realignment “realigned” responsibilities between state and local correctional systems, making counties responsible for incarcerating and supervising certain kinds of lower level offenders who previously were eligible to be sent into the state prison and parole systems.

There are several reasons to be optimistic about the impact of Realignment on recidivism rates. Under Realignment, the state provides substantial funds to counties to support the management of the new groups of offenders—2.3 billion dollars to counties from 2012 to 2014. By 2015, the annual funding package was over 1 billion annually (California State Association of Counties, 2015). The precise wording of AB 109 encouraged counties to use those funds to develop initiatives to reduce recidivism through the implementation evidence-based practices (EBP) and alternatives to incarceration (Bird & Hayes, 2013). To render county policy choices transparent, the state requires counties to develop annual Realignment implementation plans documenting how they plan to use the state funds. Along with budget information indicating what agencies and programs will be funded, the plans are published annually by the California Board of State and Community Corrections (BSCC).¹ Although counties have wide discretion about how they can spend the state’s money, overall the funding has contributed to a growth in services, both in custody and in the community within county correctional systems. In 2013, the BSCC conducted a survey of the state’s jails and found more than 200 programs designated as EBP in 40 of the state’s 57 county jails.² In addition, California has made available multiple bonds, first in 2007 and again in 2012, to finance counties’ efforts to provide adequate facilities for inmates, including building program space, medical and mental health treatment facilities, and space to support education and reentry programming.³ Finally, the Judicial Council of California publishes annual reports on California probation departments’ services and supervision strategies showing continuing progress in the implementation of EBP in community supervision across the state. These developments, some of which predate Realignment but were substantially expanded by it, offer some evidence that the localization of corrections has come with expanded capacity of jail and probation systems to provide programs and services to offenders and, hopefully, reduced recidivism.

Offender Groups Impacted by Realignment

Realignment potentially impacted four groups of offenders, each of which is subject to different treatment after the reform was passed. First, Realignment restricts parole to individuals whose commitment offense is serious or violent or to people designated as “mentally disordered offenders” or “high risk sex offenders.” These offenders experienced a parole system that was dramatically smaller in the months and years after Realignment and increasingly comprised a more homogeneous group of serious, violent, sexual, and mentally challenged offenders. Parole policies also changed the sanctioning tools available for responding to parole violations, reducing the length of return to custody for revocations from 12 to 6 months and requiring that all revocations be served in county jails rather than state prisons.

The second group created by Realignment comprised individuals released from state prison and placed on postrelease community supervision (PRCS) by county probation departments. This group is restricted to individuals released from prison commitments for nonserious and nonviolent offenses. In addition, such individuals cannot be designated as a high-risk sex offender or as mentally disordered by the California Department of Corrections Rehabilitation (CDCR). This group can and often do have serious, violent, or sexual offenses in their past and yet still qualify for PRCS because the determination of eligibility is based upon the conviction crimes they were serving prison time for at the time of release. Prior to Realignment, these individuals were supervised by state parole and, if they violated the conditions of their supervision, they could be revoked to state prison. Under Realignment, they are supervised by county probation and can be revoked to jail for up to 6 months for supervision violations.

The third and fourth groups are individuals sentenced under PC § 1170(h) whose current and prior offenses are nonserious, nonviolent, and nonsexual. Prior to Realignment, a segment of this group—usually those whose sentence is a year or longer—was eligible to be sent to state prison and, upon release, supervised by state parole. Under Realignment, these people are required to serve their time, which can be considerably longer than 1 year, in local jail. Some of these people are given “straight” sentences, which means that after serving time in jail they are released without a supervision “tail.” For these people, their punishment consists of the length of time they are required to serve and any rehabilitative treatment they receive in jail custody. Upon release, they are no longer under correctional control, unless they happen to have a concurrent sentence that requires supervision in the community. The other group of 1170(h) offenders is given “split” sentences, which divides their sentence into a custodial phase in county jail followed by “mandatory supervision” by probation in the community. These individuals receive a period of incapacitation and in-custody programming as well as services and sanctions as part of their community supervision by probation.

These four different conditions reflect different kinds of experiences for offenders, mixing, in varying degrees, the type of custodial environment and the nature of community supervision (including no supervision at all). Realignment offers the opportunity to compare recidivism outcomes relative to the more costly state prison and parole option that existed prior to Realignment. More generally, the four options that resulted from Realignment reflect a continuum of correctional control strategies that allows us to examine how the custodial environment (prison vs. jail) combined with the community supervision context (parole, supervision by probation, or no supervision) affect recidivism outcomes.

Differences Between State and Local Custodial and Community Correctional Systems

Although prisons and jails are similar in the sense that both house inmates in cells and dorms, provide a structured daily regime, have a large influence of gangs and racial sorting, offer some level of programs and services, and subject inmates to a rigorous social control system (Dolovich, 2012), the two institutions differ in important ways. For example, prior to Realignment, prisons incarcerated a wide range of offenders in terms of the seriousness and the extent of their current and past offending. Jails, even under Realignment, tend to house a more narrow range of offenders in terms of past and present offending types. Moreover, a large segment of the jail population is unsentenced and awaiting adjudication and thus only held in custody a short time (Grattet et al., 2017). As a result, jail populations tend to turn over more quickly, whereas prisons have greater stability in their populations. Historically, prisons have been more violent than jails, with homicide rates substantially higher in the former (Mumola, 2005). Under Realignment, jail populations increased, although not in direct proportion to the declines in the prison population (Lofstrom & Raphael, 2016). As Realigned offenders began to serve time locally, the portion of sentenced inmates increased along with lengths of stay. Jails now have a segment of the population serving sentences beyond 1 year. As a result, the population characteristics of jails in California have become more like prisons under Realignment and, for example, research shows that violence in jails increased under the population changes caused by Realignment (Caudill et al., 2014). All of this suggests that the population and environmental differences between prisons and jails decreased after Realignment and may indicate that recidivism outcomes might be similar.

However, under the Realignment funding arrangement, the state provided counties with resources to support alternatives to incarceration and evidence-based rehabilitative services, although counties varied in how they used these funds (Bird & Grattet, 2017). Perhaps most importantly, jails tend to house inmates closer to home, which some research suggests allows inmates to maintain prosocial connections to family and community in a way that imprisonment in a facility far from the county of commitment does not (Cochran, 2014). Thus, these features of Realignment may help to improve recidivism outcomes for realigned offenders.

Parole and probation also have some similarities and differences in population characteristics, environment, and supervision strategies. People under both forms of community supervision meet with their supervising agents frequently, depending upon their risk classification. Both systems employ supervision technologies, like global positioning system (GPS) monitoring and urinalysis, and both have broadly similar types of sanctions. Moreover, in California, both systems have moved toward adoption of evidence-based programming, often delivered through day reporting centers and community-based organizations (Bird & Grattet, 2020; Braithwaite et al., 2016). However, probation and parole have some important differences. In California, parole agents are armed and probation officers, with some exceptions and some variation across counties, are not. However, in the aftermath of Realignment, the number of armed probation officers increased, which agencies justified based upon the more serious types of offenders the reform placed under their supervision.

The composition of parole and probation caseloads are also different. Historically, probation caseloads generally consisted of people with less serious and violent offense histories. However, Realignment delivered to county probation departments two groups of offenders, the PRCS and split-sentenced individuals, who previously were supervised by state parole. This means that probation caseloads now comprised more serious offenders than prior to Realignment. While Realignment gave a wider range of offenders to probation in terms of seriousness, parole has become more narrowly focused on serious, violent, sexual, and mentally disordered orders. Under Realignment, the parole population declined (Grattet & Hayes, 2013) and the probation population has remained stable (Grattet & Martin, 2015). It may be that the changes in probation composition, the fact that they have taken on more serious offenders than previously, means that probation and parole have become more similar under Realignment, especially with how the realigned offenders are supervised and sanctioned, and thus recidivism outcomes may be roughly similar. However, with the parole population decreasing, which might allow for better supervision of those who remain under parole supervision, there may be improved recidivism outcomes for parolees. On the contrary, Realigned offenders are more likely to receive more attention within probation because they represent the most serious segment of population supervised, whereas under parole supervision, they were the least serious part of the population. Many county probation departments created specialized caseloads for realigned offenders containing fewer supervisees. As a result, it may be that probation is providing more attention and assistance than realigned offenders would have received by parole prior to Realignment. In addition, as part of Realignment, counties received funds from the state to support EBP and many counties used these funds to create Day Reporting Centers and enhance the availability of services and programs. Realigned offenders would likely have been a high priority for those services and it seems plausible that overall realigned offenders would receive more services than they previously would have under state parole. If that is the case, then recidivism outcomes among realigned offenders may improve relative to their pre-Realignment peers on parole.

Prior Research on Effects of Realignment on Recidivism

There have been a small number of studies of Realignment’s impact on recidivism (see also Bird & Grattet, 2016). Gerlinger and Turner (2015) found that based upon an simulation of pre-Realignment prison releases, the individuals most likely to be realigned to counties for PRCS had higher rates of rearrest, reconviction, and returns to prison than individuals who would likely continue to be released to state parole (Gerlinger & Turner, 2015). As they point out, many of the offenders realigned to counties were actually of high risk to recidivate, even if their offense histories were less serious. The implication is that Realignment’s focus on defining groups of offenders based upon current offense characteristics, rather than risk, would result in counties assuming responsibilities for the individuals most responsible for the “churning” via revocations and reconvictions that existed prior to Realignment. A study of the initial cohorts released from prison under Realignment to both PRCS and parole showed that recidivism patterns remained high, but that rearrest actually declined slightly (2.0 percentage points)⁴; reconvictions slightly increased as the option of revoking parole violators to prison was eliminated under Realignment; and returns to prison dropped sharply post-Realignment (Lofstrom et al., 2014). Additional work on Realignment and recidivism focuses on a subset of prisoners released to PRCS to county probation departments. Bird and Grattet (2017) found the PRCS were more likely to be rearrested and reconvicted for felony offenses compared with a pre-Realignment comparison group. However, these findings were from an initial 6-month post-Realignment cohort followed for only 6 months and, as such, Bird and Grattet (2017) cautioned the interpretation of their findings. Together, these findings represent an early look and only a segment of the populations affected by Realignment. In the long term, the realigned population is increasingly composed of individuals who previously were eligible for state prison sentences, but under Realignment are required to serve their sentences locally. One report offered a preliminary look at recidivism among these individuals (Bird et al., 2017), but was, like the research on prison releases, confined only to the first 2 years of Realignment and lacked data on returns to jail custody as a component of recidivism. As a result, the story of the reform’s effect on recidivism remains incomplete. In the next section, we describe a unique data collection effort that affords a more comprehensive and current examination of Realignment’s impact on recidivism. Given the variation in offender groups, release cohorts, and recidivism windows used in prior work, we present this study to provide a more comprehensive and long-term assessment of the effects of Realignment on recidivism.

Data and Measures

Our data come from the PPIC–BSCC Multicounty Study (MCS), which was a collaborative effort between the Public Policy Institute of California (PPIC) and the BSCC. The MCS was established in the wake of Realignment with the goal of bringing together the data needed to rigorously evaluate the statewide effects of this policy reform and to identify the most effective recidivism-reduction interventions at the local level. To achieve these goals, the MCS project team identified a group of counties to represent the state and partnered with these counties to bring together data capturing individuals moving through local jail and probation systems in the wake of Realignment. The 12 MCS counties represent 60% of the state and include Alameda, Contra Costa, Humboldt, Fresno, Kern, Los Angeles, Orange, Sacramento, San Bernardino, San Francisco, Shasta, and Stanislaus. Data from these counties were linked to California Department of Justice (DOJ) and California Department of Corrections and Rehabilitation (CDCR) data sources by the Criminal Identification and Information number. In addition to data on locally held or supervised populations, CDCR also provided data on people released from prison to parole or PRCS.

These data capture the four groups impacted by Realignment—post-Realignment parolees, PRCS, 1170(h) straight-sentenced, and 1170(h) split-sentenced. We draw on pre-Realignment releases to parole to form comparison groups for these post-Realignment groups. Figure 1 presents the release patterns over time for the pre-Realignment parole group and for the four post-Realignment groups. Although our data are drawn from the 12 MCS counties, the temporal patterns reflect trends found in other research, some of which uses statewide data. For example, in the months prior to Realignment, releases to parole were declining statewide (Grattet & Hayes, 2013). In the months after Realignment took effect, a large segment of prison releases were released to PRCS. As the offenders eligible for PRCS drained out of the prison system, the initial surge tapered off in the early months of 2013 (Nguyen et al., 2017). Also, there was steady growth in jail population of people sentenced under 1170(h) (both straights and splits) over the first year of Realignment (Grattet et al., 2017). Below, we describe the construction of the offender groups, comparison groups, and recidivism outcome measures.

Figure 1.

Releases by group over the study period.

Data on Offender Groups: Parole, PRCS, Straights, and Splits

The parole group includes individuals released from California state prisons and received for supervision by state parole between October 2011 and September 2015, whose recidivism outcomes are tracked for 2 years. The PRCS group includes individuals released from prison custody to county probation supervision between October 2011 and September 2015, or during the first 4 years of Realignment. The straight and split groups include individuals released from jail terms for 1170(h) offenses within the MCS counties during the first 4 years following the implementation of Realignment. These individuals must have also been convicted during the Realignment period (October 2011 or later) to be eligible to serve their sentences in local jail rather than state prison. Prior to Realignment, individuals sentenced to more than 1 year in custody were generally sent to prison to serve those sentences. If individuals were sentenced to less than 1 year, they served their time in local jail.

The construction of the 1170(h) straight- and split-sentenced groups is less straightforward than the construction of the PRCS and Parole groups. To identify the effect of Realignment on recidivism, we must isolate those individuals convicted for 1170(h)-eligible offenses and serving jail terms who would have otherwise gone to prison to prior to Realignment. Therefore, we limit the 1170(h) group to those who received a jail sentence of 1 year or more for an 1170(h) conviction and had a corresponding jail spell in an MCS county jail. It is important to note that given that our purpose in constructing these 1170(h) groups is to produce recidivism analyses that appropriately compares groups before and after Realignment, our estimates are not directly comparable with conviction or caseload counts for individuals receiving straight or split sentences over time. First, an individual enters our 1170(h) recidivism cohort when they are released from their first straight or split custody spell. Second, we limit our 1170(h) groups to those who were convicted and released from custody within our 4-year study window (releases from jail custody between October 2011 and September 2015). Third, we impose sentence length restrictions, as described above. Finally, we only include 1170(h) convictions where jail sentences or probation terms were served inside the MCS counties.

Pre-Realignment Comparison Groups

Using CDCR releases in the 2 years prior to Realignment, we constructed pre-Realignment comparison groups that are similar in characteristics to the post-Realignment parole, PRCS, straight, and split groups using a propensity score matching technique. The pre- and post-straight- and split-sentenced groups were matched on 22 variables⁵ and the pre- and post-PRCS and parole groups were matched on 26 variables.⁶ The matching variables were selected to adjust for key differences between the pre- and post-Realignment samples in terms of underlying risk to recidivate to isolate the effect of Realignment on recidivism within each of the offender groups. These measures capture recognized indicators of risk, including the onset, frequency, and type of prior offending, current offense, and demographic characteristics.

Table 1 reports selected matching statistics for each of the four groups. Significant differences in variable means were present in the unmatched post-Realignment groups and pre-Realignment control groups. After matching, the p-score and nearly all of the covariates improve in terms of balance. After prematching, we then apply regression analysis to address any remaining differences between each Realignment group and its corresponding control group.⁷ This strategy improves on the traditional regression approach by reducing reliance on the regression model to adjust for differences in observable characteristics. We leverage exogenous variation in the correctional treatment induced by Realignment and draw on a rich set of individual-level characteristics to estimate the effect of policy change on recidivism outcomes. However, as is the case in all observational studies, selection bias could still play a role to the extent that there remain unobserved differences between the post-Realignment offender and control groups and that any such differences lead to higher or lower recidivism rates.

Table 1.

Pre- and Postmatching Difference in Means for a Sample of Characteristics.

Variable	ParolePrematch	ParolePostmatch	PRCSPrematch	PRCSPostmatch	StraightPrematch	Straight Postmatch	SplitPrematch	SplitPostmatch
p-score	−.153***	.000	−.095***	.000	−.177***	.000	−.075***	.000
Age	3.098***	0.183**	−1.655***	0.022	−1.168***	0.185	0.940***	0.033
Black	−0.015***	−0.003	−0.008**	0.006*	0.082***	0.003	0.082***	−0.005
Age at first conviction	0.528***	0.020	0.836***	0.086*	−0.624***	0.070	−0.905***	−0.051
Violent count	−0.185***	0.000	0.006**	−0.018***	0.173***	0.000	0.173***	0.000
Total drug arrests	1.655***	−0.006	−0.921***	0.001	−2.415***	−0.067	−0.391***	0.009
Total drug convictions	0.644***	−0.004	0.000***	0.458	−1.115***	−0.008	−0.189***	0.001

Source. MCS data 2009–2015.

Note. Matching was undertaken on a total of 26 variables for the parole and PRCS groups and 22 variables for the straight and split groups (full list of matching variables is included in Notes 5 and 6). The variables reported here reflect the range of different types of demographic and offense history characteristics on which matching was done. Each column indicates the difference between the pre- and post-Realignment groups and indicates the significance level of this difference. PRCS = postrelease community supervision; MCS = multicounty study.

p < .10. **p < .05. ***p < .01.

Recidivism Outcome Measures

We use 2-year rearrest and reconviction outcomes to measure recidivism and include felonies and misdemeanors in these measures. In our findings, we present estimates of the effects of Realignment on rearrest outcomes excluding violations and on conviction outcomes excluding revocations. However, we also present—and prefer—a measure of rearrest that included violations and a measure of conviction that is adjusted for revocations. We argue it is particularly important to adjust the conviction measure for revocations because, along with potential effects on recidivism, Realignment-induced changes in how local criminal justice systems responded to new offending and supervision violations.

Under Realignment, most individuals could no longer be revoked for long terms to state prison and, instead, would serve shorter revocations terms in local jails. While revocations were used frequently to reincarcerate individuals suspected of new offending prior to Realignment, local justice systems were more likely to pursue formal reconviction via court adjudication after Realignment. Therefore, a measure of reconviction that fails to account for revocations would bias our estimates of the effect of Realignment by potentially leaving out a larger share reoffending among the pre-Realignment group than among the post-Realignment group. For this reason, we focus our findings on an outcomes measure that captures both formal reconviction and revocations.

Findings

Figure 2 provides a visual summary of the results of the regressions designed to estimate the Realignment effect size on the 2-year adjusted rearrest and adjusted reconviction rates for each of the four offender groups. All of the policy effects reported are statistically significant at the .001 level. These findings are based on the propensity score matched samples for each group and reflect regression adjusted estimates with fixed effects for unobserved county-level factors. Full regression results are reported in Appendix Tables A1 and A2. In addition to the results presented in Figure 2, we also report results estimated without the state’s largest county, Los Angeles, and results are similar in direction and significance, although the magnitude of the effect varies somewhat. We note below where we found differences when Los Angeles is not included. We also estimated findings for rearrest with and without the inclusion of arrests for supervision violations and reconviction with and without revocations to provide additional insights into how the measurement of recidivism influences the findings. Below, our summary of the findings focuses on the most inclusive measures rearrest and reconviction—those that include violations in the measure of rearrest and revocations in the measure of reconviction. Results with all of the different recidivism measures and results with and without Los Angeles are presented in Table 2.

Figure 2.

Estimates of the effects of Realignment on 2-year recidivism rates by treatment group.

Table 2.

Comparison of Coefficient Estimates From Models Including and Excluding Los Angeles County.

Variable	Parole	PRCS	1170(h) straight	1170(h) split
All counties included
Rearrest for felony or misdemeanor	0.017*** (0.004)	0.048*** (0.003)	0.003 (0.005)	0.076*** (0.007)
Rearrest for felony, misdemeanor, or supervision violation	0.012*** (0.004)	0.023*** (0.003)	−0.037*** (0.005)	0.049*** (0.007)
Reconviction for felony or misdemeanor	0.039*** (0.373)	0.064*** (0.003)	−0.015*** (0.005)	−0.011 (0.007)
Reconviction for felony, misdemeanor, or revocation	−0.043*** (0.004)	−0.017*** (0.003)	−0.084*** (0.005)	−0.030*** (0.007)
All counties, except Los Angeles
Rearrest for felony or misdemeanor	0.032*** (0.005)	0.061*** (0.005)	0.038*** (0.008)	0.081*** (0.009)
Rearrest for felony, misdemeanor, or supervision violation	0.026*** (0.515)	0.034*** (0.004)	−0.004 (0.008)	0.059*** (0.008)
Reconviction for felony or misdemeanor	0.060*** (0.513)	0.095*** (0.005)	0.017** (0.008)	−0.011 (0.009)
Reconviction for felony, misdemeanor, or revocation	−0.050*** (0.523)	−0.017*** (0.005)	−0.084*** (0.008)	−0.040*** (0.009)

Source. MCS data 2009–2017.

Note. PRCS = postrelease community supervision; MCS = multicounty study.

p < .10. **p < .05. ***p < .01.

The results displayed in Figure 2 show that three of the four groups experienced increases in rearrest. Parolees released from prison under Realignment had a 1.2 percentage point increase in their rearrest rate—inclusive of violations—compared with their pre-Realignment counterparts. Realignment also resulted in a 2.3 percentage point increase in rearrest for the PRCS and a 4.9 percentage point increase in the rearrest rate for the split-sentenced group. The only departure from this pattern of increased rearrests is the straight-sentenced group, which experienced a 3.7 percentage point decrease in rearrests. However, with respect to the straight-sentenced group, Los Angeles appears to drive the declines in their rearrest rates. When Los Angeles is removed from the analysis, the differences between the straight-sentenced group and its pre-Realignment comparison group is not significant (Table 2).

All four groups experienced decreases in reconviction, including revocations, under Realignment. Because most parolees could no longer be revoked to prison, we expected a shift from the use of revocations to the pursuit of formal reconvictions to return individuals to custody. We see strong evidence of this shift in our analysis. When we consider only misdemeanor and felony reconvictions for parolees, we find an increase of 3.9 percentage points in reconviction rates (Table 2). However, when we adjust for revocations, the direction of the effect of Realignment on recidivism changes. We find that Realignment reduced reconvictions—adjusted for revocations—by 4.3 percentage points for post-Realignment parolees relative to their pre-Realignment counterparts. Like parolees, the PRCS could no longer be revoked to prison under Realignment. We find the reconviction rate—excluding revocations—for this group is 6.4 percentage points higher than prior to Realignment. However, when we adjust for revocations, we actually find a 1.7 percentage point decrease in the reconviction rate for the PRCS relative to their pre-Realignment counterparts. For the split-sentenced group, we find a 3.0 percentage point decrease in reconviction rates relative to similar individuals released pre-Realignment. The straight-sentenced group experienced an 8.4 percentage points reduction in reconviction in comparison with their pre-Realignment counterparts.

The finding of higher rearrest and lower reconvictions is consistent across groups, with the exception of the straight-sentenced group. One possible interpretation is that Realignment lowered deterrence and, therefore, elicited greater misconduct, particularly among the groups with higher rearrest rates. However, if that were the case, we would expect that there would also be increases in revocations and reconvictions. Given that we find evidence that Realignment reduced reconvictions, adjusted for revocations, for all groups, we conclude that the rearrest pattern is more reflective of a change in frontline law enforcement and supervision responses than change in behavior that is serious enough to result in revocation or reconviction.

A second possible interpretation is that both the state and counties improved in the delivery of rehabilitative services and programs and that those interventions in custody and under supervision improved outcomes for each offense group. But here again, if that were the case, it should logically follow that such improvements should be reflected in declines in both rearrest and reconviction rates because such interventions should be preventive of misconduct. That does not appear to be the case. In three of the four groups, rearrests were higher under Realignment, and in all four groups reconvictions were lower. The one case that had both lower rearrests and reconvictions—the straight-sentenced group—received no supervision in the community at all, again suggesting that it may not have been that behavior changed, only the detection of it.

In addition, it could be that parole and probation supervision was intensified under Realignment, leading to increased likelihood that supervision violations and criminal conduct were detected. Prior research on supervision intensity has found this can happen. For example, Petersilia and Turner’s experimental studies of “Intensive Supervision Programs” in various states showed consistent evidence that increasing supervision intensity, achieved through increased contact and drug testing as well as smaller caseloads, led to increased technical violations without affecting other measures of recidivism (Petersilia & Turner, 1991, 1993).⁸ The parole population declined under Realignment, possibly allowing for greater concentration of supervision on the parolees that remained, which could have increased detection and reporting of rearrests. Also, realigned offenders—the PRCS and the split-sentenced individuals—were new responsibilities to county probation agencies and represented more serious offenders among their caseloads. Many counties responded by creating specialized “AB 109” caseloads to provide more intensive supervision of PRCS and split-sentenced offenders. Conversely, straight-sentenced offenders were provided no supervision upon release from custody and they had largest decreases in rearrests under Realignment. Increasing supervision intensity for three of the offender groups and decreasing supervision intensity for the fourth provide further indirect evidence that the changing supervision conditions likely led to the patterns we observe.

Additional information about how Realignment affected rearrest is provided by contrasting the measure of rearrest that includes violations with the measure that focuses only on rearrests for felonies or misdemeanors (Table 2). Here, we see that for each of the four groups, the measure that includes rearrests for supervision violations has a smaller coefficient compared with the coefficients for the measure that only captures rearrests for felonies or misdemeanors. This implies that although rearrests overall increased, rearrests for supervision violations declined—a finding which runs counter to the work of Petersilia and Turner, who only found increases in technical violations. Nonetheless, we interpret the increased rearrests overall as resulting from the changes to the revocation process that disincentivized the initiation of the revocation process via violations and incentivized the use of felony and misdemeanor rearrests to sanction offenders. Thus, at least part of recidivism changes we observe likely resulted from changing system responses to the four groups of offenders.

If rearrest patterns were a product of changes in how the state and county supervision systems responded to the each group of offenders, then it seems likely that the uniform decreases in reconviction might be similarly shaped by changes in prosecution. Considerable research has focused on prosecutorial discretion (Bushway & Forst, 2013; Spohn, 2018). Some of this work has documented the effect of overly punitive uses of prosecutorial discretion that led to the growth of prison populations from 1980 onward (Pfaff, 2017). In the context of decarcerative reforms in California, at least some prosecutors have advocated for alternatives to incarceration, diversion, and a more restrained approach to charging. In recent years, a group of county prosecutors have been at the forefront of supporting criminal justice reform, including Georg Gascon, former San Francisco County District Attorney (DA) (and the current DA in Los Angeles County; Willis, 2020).⁹ These developments serve as a reminder that prosecutorial philosophy is not static over time and perhaps the broader reform ethos set off by Realignment and subsequent reforms are reshaping how at least some prosecutors work, resulting in marginally lower reconvictions.

Limitations

Recidivism research that contrasts groups of individuals over time rather than groups constructed through contemporaneous random assignment to a treatment and control condition is limited in the inability to know with complete confidence what drives changes. Even with an extensive list of covariates used to match groups over time and to control in a regression model, it is possible that unobserved factors may generate the differences in our findings. It is also important to recognize the challenge associated with determining the mechanisms of the effect of a policy change on recidivism, even if the groups were perfectly matched and no unobserved factors were responsible for the differences. The policy change in this case altered not only level of deterrence and access to rehabilitative programs, it also shifted incentives for officials in the system and likely affected how they respond to offenders. This appears to be what we see in our findings. Nonetheless, our study shows that it is critical to use multiple measures of recidivism outcomes to illuminate the mechanism by which a policy change affects outcomes. The ability to use both rearrest and reconviction and the ability to include and exclude supervision violations and revocations into these measures allow us to parse the findings to provide evidence that the changes we observed—the decreases in rearrest across three of four groups and the decreases in reconviction across all groups—are most likely due to changes in system response rather than offender behavior.

This is not say that decreases in deterrence did not partially lead to more rearrests and increases in rehabilitative services did not partially lead to decreases in reconviction. But those effects, if they were dominant causal processes, should have resulted in similar patterns in both rearrest and reconviction. Thus, the mismatch in outcomes between rearrest and reconviction is the central finding that needs to be explained. The most parsimonious and cautious explanation is that this finding resulted from the intended and unintended consequences of the reform that changed how system actors responded to offenders whose trajectory through custody and community supervision phases of their punishment were altered by it.

Conclusion

Taking place in the largest correctional system in the nation, California’s Public Safety Realignment is a significant reform in its own right. A comprehensive study of the impacts of Realignment on recidivism is essential as state policymakers move forward with further legislation intended to downsize prison populations and as county practitioners navigate the management of these different populations. However, evaluating the effects of this prison downsizing effort on recidivism may also provide some clues as to what other states can expect if they embrace reforms designed to narrow the function of prisons to focus on serious and violent offenders, leaving property and drug offenders to be punished and supervised in local jails and community supervision systems. Several states are engaged in efforts that are similar (Eisen, 2020; Porter, 2020, 2021), although not identical, to California and could reasonably be expected to see similar system responses that change incentives for local actors and that lead to a mismatch between rearrest and reconviction patterns.

Nonetheless, California’s experience shows that it is possible to reduce prison population without experiencing dramatic shifts in reoffending. It also shows that such changes alter incentives within law enforcement and supervision systems, which can have varied effects on different groups affected by the reform and which can be reflected differently in different measures of recidivism. In California, the reform essentially created four different kinds of offender groups. Each of these groups experienced different circumstances of custodial-community supervision as a result of the reform. We saw increases in rearrests, often somewhat minimal, among those whose supervision circumstances changed in such a way to increase closer supervision and a decrease in rearrests among those that experienced a reduction in supervision. However, all groups experienced decreases in our reconviction measure, which should give policymakers in other states the greatest hope that reoffending that is serious enough to warrant a reconviction or a revocation can be reduced through reforms that similarly realign selected groups of offenders from state prisons to local correctional systems. Doing so would reduce the costs and collateral consequences of prison incarceration and lower the burdens on local systems by reducing the costs associated with prosecution and reconviction.

California’s policy experiment demonstrates the power of aligning incentives with decision-making. Under Realignment, counties now hold the correctional responsibility for lower level offenders. As a result, decisions about arrest, charging, and sentencing are made by the same local systems that will bear the associated correctional costs. This shift in incentives not only led to dramatic reductions in prison incarceration rates, but in reduced overall levels of incarceration. Even if improvements are the result of system rather than behavioral changes, similar changes are likely to be worthwhile.

Footnotes

Appendix

Table A2.

Full Regression Results: Estimates of the Effects of Realignment on 2-Year Reconviction Rates.

Variable	Parole	PRCS	1170(h) straight	1170(h) split
Intercept	0.647*** (0.030)	0.720*** (0.071)	0.608*** (0.022)	0.570*** (0.026)
Realigned group	−0.043*** (0.004)	−0.017*** (0.003)	−0.084*** (0.005)	−0.030*** (0.007)
Age	−0.012*** (0.000)	−0.013*** (0.000)	−0.017*** (0.000)	−0.015*** (0.001)
Male	0.066*** (0.007)	0.054*** (0.006)	0.074*** (0.007)	0.048*** (0.009)
White	0.060*** (0.008)	0.052*** (0.008)	0.056*** (0.016)	0.049** (0.020)
Black	0.034*** (0.008)	−0.003 (0.008)	−0.002 (0.017)	−0.007 (0.021)
Hispanic/Latinx	−0.001 (0.008)	−0.005 (0.008)	−0.005 (0.016)	−0.033* (0.020)
Asian	−0.005 (0.018)	0.015 (0.017)	0.028 (0.024)	0.024 (0.027)
Person offense	−0.027*** (0.006)	−0.020*** (0.006)	0.021 (0.018)	−0.036** (0.017)
Property offense	0.019*** (0.006)	0.038*** (0.005)	0.063*** (0.009)	0.071*** (0.010)
Drug offense	−0.008 (0.008)	−0.012** (0.005)	−0.015* (0.009)	0.004 (0.010)
Age at first conviction	0.005*** (0.000)	0.005*** (0.000)	0.006*** (0.001)	0.004*** (0.001)
Prior serious offenses	−0.008** (0.004)	−0.002 (0.004)
Prior violent offenses	0.018*** (0.004)	0.017*** (0.004)
Prior arrests	0.011*** (0.001)	0.009*** (0.000)	0.011*** (0.001)	0.013*** (0.001)
Prior felony arrests	0.000(0.001)	0.000(0.001)	0.001(0.001)	0.001(0.001)
Prior person offense arrests	−0.001(0.001)	0.001(0.001)	0.000(0.002)	0.001(0.002)
Prior property offense arrests	0.001(0.001)	0.000(0.001)	0.000(0.001)	0.001(0.002)
Prior drug offense arrests	−0.005*** (0.001)	−0.002** (0.001)	−0.001(0.001)	−0.004** (0.002)
Prior convictions	−0.003** (0.001)	−0.002** (0.001)	0.000(0.002)	−0.002(0.003)
Prior felony convictions	0.019*** (0.002)	0.019*** (0.001)	0.028*** (0.002)	0.025*** (0.003)
Prior person offense convictions	0.001(0.002)	0.003(0.002)	0.002(0.004)	0.003(0.005)
Prior property offense convictions	−0.001(0.002)	0.000(0.002)	0.003(0.003)	0.000(0.004)
Prior drug offense convictions	0.009*** (0.003)	0.008*** (0.002)	0.004(0.003)	0.008** (0.004)
High-risk for person offense	−0.030*** (0.006)	−0.025*** (0.006)
High-risk for property offense	−0.055*** (0.008)	−0.044*** (0.006)
High-risk for drug offense	−0.052*** (0.011)	−0.062*** (0.007)
Moderate risk	−0.113*** (0.006)	−0.126*** (0.005)
Low risk	−0.150*** (0.007)	−0.162*** (0.006)
Mental health needs	0.063*** (0.005)	0.044*** (0.004)
Second-strike sentence	−0.058*** (0.015)	−0.031(0.070)
Determinate sentence	−0.056*** (0.016)	−0.007(0.070)
Committed for new offense	−0.065*** (0.022)	−0.120*** (0.008)
Parole violator with new term	0.016(0.022)	−0.027*** (0.008)
Parole violator/revoked to prison	0.054** (0.022)	0.010* (0.005)
Length of stay in prison	0.000*** (0.000)	0.000*** (0.000)	0.000*** (0.000)	0.000*** (0.000)
County fixed effects	Y	Y	Y	Y
No. of observations	67,414	85,668	35,339	20,850

Source. MCS data 2009–2017.

Note. PRCS = postrelease community supervision; MCS = multicounty study. Significance of bolded variables.

p < .10. **p < .05. ***p < .01.

Acknowledgements

This project was made possible by the willingness of 12 county sheriff and probation departments in California as well as the California Department of Justice and the California Department of Corrections and Rehabilitation to share offender-level data. Support and encouragement for the project was provided by the California Board of State and Community Corrections, California State Sheriffs Association, Chief Probation Officers of California, California State Association of Counties, and the County Administrative Officers Association of California. Research was undertaken with financial support of the Public Policy Institute of California. All inferences, interpretations, and conclusions as well as any errors are solely attributable to the authors.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Ryken Grattet

Notes

Author Biographies

Mia Bird is an Assistant Adjunct Professor at the Goldman School of Public Policy. Her research focuses on criminal justice reform and the intersections of criminal justice, health, and social safety net policies. Her previous work has been published in Criminology & Public Policy, Federal Probation, and Criminal Justice Policy Review. She also serves as a research consultant for the California Committee on the Revision of the Penal Code, the California Council on Criminal Justice and Behavioral Health, and the California State Association for Counties.

Viet Nguyen is a doctoral student in the Department of Criminology at the University of Pennsylvania. His research interests include corrections policy and criminal justice reform. His research has been published in Criminology & Public Policy, Federal Probation, and the Journal of Experimental Criminology. He is currently Managing Editor of the Journal of Quantitative Criminology.

Ryken Grattet is Professor of Sociology at the University of California, Davis. He has published on a range of topics relating to law and policy implementation, including studies of the enforcement of hate crime law, the reform of the California correctional system, and the governance of climate change. His scholarly work has appeared in the Law & Society Review, Criminology, Criminology & Public Policy, Social Forces, and Social Problems.

References

Bird

Grattet

(2016). Realignment and recidivism. The ANNALS of the American Academy of Political and Social Science, 664(1), 176–195.

Bird

Grattet

(2017). Policy change and recidivism the effects of California’s realignment and local implementation strategies on rearrest and reconviction. Criminal Justice Policy Review, 28, 601–623.

Bird

Grattet

(2020). SB 678: Incentive-based funding and evidence-based practices enacted by California probation are associated with lower recidivism rates and improved public safety. California Probation Resource Institute.

Bird

Grattet

Nguyen

(2017). Realignment and recidivism in California. Public Policy Institute of California.

Bird

Hayes

(2013). Funding public safety realignment. Public Policy Institute of California.

Braithwaite

Lavery

Westfall

Turner

(2016). Parole agent workload study. https://cpb-us-e2.wpmucdn.com/sites.uci.edu/dist/0/1149/files/2017/04/Parole-Agent-Workload-Study.pdf

Bushway

S. D.

Forst

(2013). Studying discretion in the processes that generate criminal justice sanctions. Justice Quarterly, 30(2), 199–222.

California State Association of Counties. (2015). Distribution of AB 109 funds: Community corrections and district attorney/public defender subaccounts: 2014-14 and beyond.

Caudill

J. W.

Trulson

C. R.

Marquart

J. W.

Patten

Thomas

M. O.

Anderson

(2014). Correctional destabilization and jail violence: The consequences of prison depopulation legislation. Journal of Criminal Justice, 42(6), 500–506.

10.

Cochran

J. C.

(2014). Breaches in the wall: Imprisonment, social support, and recidivism. Journal of Research in Crime & Delinquency, 51(2), 200–229.

11.

Dolovich

(2012). Two models of the prison: Accidental humanity and hypermasculinity in the LA County Jail. Journal of Criminal Law & Criminology, 102, 965.

12.

Eisen

L.-B.

(2020). Criminal justice reform at the state level. Brennan Center for Justice. https://www.brennancenter.org/our-work/research-reports/criminal-justice-reform-state-level

13.

Gerlinger

Turner

S. F.

(2015). California’s public safety realignment: Correctional policy based on stakes rather than risk. Criminal Justice Policy Review, 26(8), 805–827.

14.

Grattet

Bird

(2017). California probation in the era of reform. https://www.ppic.org/wp-content/uploads/r_0817vnr.pdf

15.

Grattet

Bird

(2018). Next steps in jail and prison downsizing. Criminology & Public Policy, 17, 717–726. https://doi.org/10.1111/1745-9133.12383

16.

Grattet

Bird

Nguyen

Tafoya

(2017). California Jails under realignment and proposition 47. California Journal of Politics and Policy, 9(3). https://escholarship.org/uc/item/6qx3q548

17.

Grattet

Hayes

(2013). California’s changing parole population. Public Policy Institute of California.

18.

Grattet

Lin

(2016). Supervision intensity and parole outcomes: A competing risks approach to criminal and technical parole violations. Justice Quarterly, 33(4), 565–583.

19.

Grattet

Lin

Petersilia

(2011). Supervision regimes, risk, and official reactions to parolee deviance. Criminology, 49(2), 371–399.

20.

Grattet

Martin

(2015). Probation in California [Just the Facts]. Public Policy Institute of California.

21.

Grattet

Tafoya

Bird

Nguyen

(2016). California’s county jails in the era of reform. Public Policy Institute of California.

22.

Lofstrom

Martin

(2015). Public safety realignment: Impacts so far. Public Policy Institute of California.

23.

Lofstrom

Petersilia

Raphael

(2012). Evaluating the effects of California’s corrections realignment on public safety. Public Policy Institute of California.

24.

Lofstrom

Raphael

(2015). Realignment, incarceration, and crime trends in California. Public Policy Institute of California.

25.

Lofstrom

Raphael

(2016). Prison downsizing and public safety: Evidence from California. Criminology & Public Policy, 15, 349–365.

26.

Lofstrom

Raphael

Grattet

(2014). Is public safety realignment reducing recidivism in California? Public Policy Institute of California.

27.

Misczynski

(2011). Rethinking the state-local relationship: Corrections. Public Policy Institute of California.

28.

Mumola

C. J.

(2005). Suicide and homicide in state prisons and local jails. U.S. Department of Justice, Office of Justice Programs.

29.

Nguyen

Grattet

Bird

(2017). Changing characteristics of California’s probation population: Revelations of new data. Perspectives, Fall, 45–57.

30.

Petersilia

Turner

(1991). An evaluation of intensive probation in California. Journal of Criminal Law & Criminology, 82, 610–658.

31.

Petersilia

Turner

(1993). Evaluating intensive supervision probation/parole: Results of a nationwide experiment. U.S. Department of Justice, Office of Justice Programs, National Institute of Justice.

32.

Pfaff

(2017). Locked in: The true causes of mass incarceration-and how to achieve real reform. Basic Books.

33.

Porter

N. D.

(2020). Top trends in state criminal justice reform, 2019. The Sentencing Project. https://www.sentencingproject.org/publications/top-trends-in-state-criminal-justice-reform-2019/

34.

Porter

N. D.

(2021). Top trends in state criminal justice reform, 2020. The Sentencing Project. https://www.sentencingproject.org/publications/top-trends-in-state-criminal-justice-reform-2020/

35.

Santos

(2013, March 11). California’s realignment: Real prison reform or shell game? Huffington Post. https://www.huffpost.com/

36.

Spohn

(2018). Reflections on the exercise of prosecutorial discretion 50 years after publication of the challenge of crime in a free society. Criminology & Public Policy, 17(2), 321–340.

37.

VanSickle

Villa

(2018, December 20). The great California prison experiment. The Marshall Project. https://www.themarshallproject.org/2018/12/20/the-great-california-prison-experiment

38.

Weisberg

(2011). California’s de facto sentencing commissions. Stanford Law Review, 64, 1.

39.

Willis

. (2020, September 15). A new association of progressive D.A.s could overhaul California’s reform movement. The Appeal. https://theappeal.org/politicalreport/california-prosecutors-progressive-association/