This appendix reviews what is known about the recidivism rates of incarcerated individuals from national studies, evidence suggesting that correctional authorities can use information at hand to identify individuals at high risk of recidivism, and potential factors that might mitigate that risk through supportive reentry planning. The key implications of this literature for the current pandemic are discussed in Chapter 3.
OVERALL RECIDIVISM LEVELS
A very large body of research examines the likelihood of future interactions with the criminal justice system among those placed under correctional supervision in the community (e.g., those on probation or parole) or released from the custody of a state or federal prison or local jail. Recidivism is commonly measured using recorded incidents of future interactions with the criminal justice system, including being arrested, being arraigned for a new criminal charge, being convicted, and being placed in custody (often being returned to custody). Yet there is considerable debate regarding what is being measured using these official markers of criminal justice involvement. For example, people under community corrections supervision can be and often are arrested for behaviors that would not otherwise result in arrest, such as missing appointments, not reporting to a central authority, drinking or using drugs, leaving a specific county, and other such technical violations. Individuals on parole may be returned to prison custody with or without conviction for a new criminal offense, and in some states may be returned to custody directly by a parole
officer. In some instances, people arrested for new crimes may be returned to custody through parole revocation with no formal charging and adjudication for a new offense. Jurisdictions vary in the relative frequency with which people on parole are returned to custody for less serious violations. In jurisdictions with so-called swift-and-certain community corrections systems, brief returns to custody for minor infractions may routinely be used as a behavior management strategy. In other jurisdictions, returns to custody may be reserved for those who repeatedly violate terms of parole or engage in new crimes. Hence, it is often difficult to gauge whether cross-jurisdictional differences or changes over time in formal gauges of recidivism reflect actual changes in criminal behavior or differences in policies and practice. Arguably, a new conviction provides a less ambiguous gauge of subsequent criminal activity.
Overall, much of what is known about overall recidivism rates among released prison inmates comes from two national recidivism projects conducted by the Bureau of Justice Statistics (BJS). Langan and Levin (2002) report the results of an analysis of recidivism among individuals released from 15 state prisons in 1994. The study linked a random sample of releases from these states to criminal history records from the releasing states’ criminal history repositories, as well as data from the cross-state repository maintained by the Federal Bureau of Investigation (the III database). The study found that within 3 years of release, nearly 70 percent of released individuals had been rearrested, roughly 47 percent convicted of a new crime, and 25 percent returned to prison with a new conviction. Among those released from prison, over half were returned without a new conviction. This result, however, likely reflects the outsized influence of California and the parole practice in place at the time, a factor explored in detail below. Langan and Levin (2002) also tabulate the proportion of all arrests for serious offenses committed by the people in this release cohort. While released individuals accounted for a relatively small percentage of all arrests for serious offenses, they conclude that the rate at which these formerly incarcerated individuals committed serious violent offenses was quite high relative to the general population.
Durose, Cooper, and Snyder (2014) and Alper, Durose, and Markman (2018) present comparable analyses of individuals released from prison in 2005. In these studies, the authors analyze people released from 30 state prison systems and measure recidivism outcomes over a 5-year period (the 2014 study) and a 9-year period (the 2018 study); the latter study focused solely on arrests. As with the 1994 release cohort (Langan and Levin, 2002), a large percentage of released inmates had been rearrested—nearly 67 percent within 3 years, 75 percent within 5 years, and 83 percent within 9 years. Within 5 years, 28 percent had been returned to prison because of
a conviction for a new offense, and 51 percent had been returned either for a new offense or for a parole violation without a new conviction, figures quite close to those from the 1994 study.
For at least one of the key recidivism outcomes—return to prison custody—the outcomes documented by these studies may paint an excessively negative picture of the reentry prospects of formerly incarcerated individuals. First, both studies rely on a subset of states, with California contributing the largest number of prison releases to the weighted recidivism estimates. To be specific, for the 1994 release cohort, California releases account for approximately 35 percent of the weighted analysis sample of releases, while for the 2005 release cohort, California accounts for nearly 26 percent of the weighted sample. Prior to a 2011 reform that limited returns to custody without a new conviction for individuals on parole, California had by far the highest rate of readmission from parole to its state prison system, for two reasons: (1) nearly all individuals released from prison in California were placed on parole, with a postrelease term of 5 years; and (2) California made heavy use of returns to custody for technical parole violations, with the returns often resulting in very short prison stays (Lofstrom, Raphael, and Grattet, 2014).
Figure A-1 illustrates how unusual California’s practices were relative to the rest of the nation. The figure presents the number of prison admissions during calendar year 2010 that involved individuals on parole divided by the size of the parole population at the beginning of the year. States are ranked in the figure from highest to lowest values of this ratio. The value for California (0.68) is the highest in the nation, and is 24 percentage points higher than the value for the next-highest state (Connecticut, at 0.44). Moreover, 77 percent of the California admissions involved returns to custody without a new prison sentence.
While subsequent reports from the BJS cannot be used to construct a comparable figure post-2011, recidivism estimates published by the California Department of Corrections and Rehabilitation show a sharp decline in return-to-custody rates with the 2011 reforms. Figure A-2 reports several 3-year recidivism measures for individuals released from California state prisons during fiscal years 2002 through 2014. For 2002 through 2008, more than 61 percent of releases were returned to custody within 3 years. This rate begins to show a sharp decline in 2009 as the 3-year observation window for each subsequent cohort begins to overlap and then entirely overlaps with the post-2011 period, when parole practices had changed. By 2011, the 3-year return-to-custody rate had declined to below 25 percent. Declines in rearrest rates, as well as reconviction rates, can also be observed following this reform. While these recidivism outcomes can be attributed in part to change in the composition of each release cohort, a careful analysis
Given the disproportionate contribution of California to the above widely cited release cohort studies, the sharp decline in California’s return-to-custody rates, and the large drop in admissions in the state resulting from this reform, a comparable analysis of a post-2011 release cohort would likely reveal that the majority of released prison inmates do not return within 3 years.
Beyond the outsized influence of California, several researchers have raised concerns about the focus on release cohorts rather than on cohorts defined by people who have ever served time in prison. Rosenfeld (2008) points out that release cohorts defined by a specific time period (say, during a given calendar year) disproportionately comprise individuals who have been returned to prison and who will, in turn, be returned to prison in the future. For an entire year’s worth of releases, there are indeed some individuals who will have been released and readmitted multiple times over the course of the year. Hence, the probability that a randomly chosen release over the course of a year will result in a return to custody is likely higher
1 Specifically, inmates released following the reform were older and more likely to be released from the first term in prison. Adjusting for these characteristics still reveals sharp declines in return-to-custody rates of a magnitude comparable to that observed in Figure 3-2 (in Chapter 3). In addition, Lofstrom, Raphael, and Grattet (2014) find little evidence of an offsetting increase in prosecution and a very small increase in the likelihood of a new criminal conviction among released inmates as a result of the reform.
than the likelihood that someone who eventually serves a prison sentence will be returned for a second prison spell. Moreover, this observation may apply to other recidivism outcomes, such as arrests and convictions that do not result in a return to custody.
Rhodes and colleagues (2016) provide a recidivism analysis using annual release cohorts and data from the National Corrections Reporting Program for the years 2000 through 2014. The authors estimate recidivism in terms of both rearrest and return to custody for each annual release cohort during this period for observations of varying length. They present two sets of estimates: those in which each release receives equal weight in tabulating recidivism outcomes (what they refer to as the event-based sample) and those in which each release is inversely weighted by the number of times the individual in question was released during the year. For example, a release observation for someone released three times during a given calendar year would receive a weight of 1/3. Hence, the average of the releases of a person released more than once receives the same weight as the single release of someone released only once. For the 2000 cohort, the authors conclude that while 51 percent of releases would have resulted in return to prison custody over the subsequent 12 years, the individual-based analysis indicates that 33 percent of the unique individuals released over the course of the year would have been returned to prison over the same time period.
A careful reading of the methodology sections of two BJS analyses suggests that the approach of Rhodes and colleagues (2016) would yield
lower recidivism rates for at least the earlier study.2 Both studies would certainly yield lower return-to-custody rates if they were implemented today, given that California would contribute proportionately fewer releases and would have much lower return-to-custody rates. Regardless, the change in policy in California, coupled with analysis focused on individuals rather than events, strongly indicates that a return to prison custody among those released is far from certain and that most who do prison time and are released, under current correctional practices, will not return to prison.
INCARCERATION RATES AND CRIME
Reducing prison population as a response to COVID-19 raises questions concerning whether such a change would cause an increase in crime rates. There is a large body of literature studying the relationship among incarceration levels, the risk of incarceration, and crime. While there is evidence that prisons incapacitate some individuals who would otherwise be criminally active, research also shows that such effects are heterogeneous across incarcerated people and that in societies that rely heavily on incarceration, the crime-prevention effects of incarceration for many are negligible (NRC, 2014). Moreover, recent reforms in California that have drawn down the combined prison and jail population by nearly a quarter over a relatively short time period have had little effect on crime rates.
Theoretically, incarceration may impact crime through several potential channels. First, incarcerating a criminally active person incapacitates that person, preventing the individual from committing new crimes within non-institutionalized society (though criminal offending may clearly still occur within institutions and this remains unmeasured in the research). The effects on crime of physically removing someone from society are likely to vary from person to person, often in predictable ways. Second, the threat of incarceration (or the severity of a sentence) may deter some individuals from committing a crime (a pathway referred to as general deterrence). If individuals consider the costs and benefits of their actions, stiffer sentences may increase costs above benefits and tip the decision-making scales away from committing a crime. Finally, serving a prison sentence may alter the future offending trajectories of former prison inmates. A prison spell may
2 In relation to the two national release cohort studies, both had a complex sampling structure whereby they selected stratified random samples of releases from the included states, based on controlling by offense groupings. Langan and Levin (2002) appear to draw random samples of all releases from each controlling offense stratum, and thus must have sampled individuals who had been released more than once over the course of the year. In contrast, Alper, Durose, and Markman (2018) and the earlier study by Durose, Cooper, and Snyder (2014) indicate in their methods section that for individuals released more than once they select the first release.
reduce offending if the experience itself deters (a factor often referred to as “specific deterrence”). Alternatively, education and treatment services while incarcerated may have rehabilitative effects on those who pass through prison and reduce future offending as a result. On the other hand, a prison spell may erode connections to the workforce, lead to the erosion of marketable skills, and perhaps increase the likelihood of a future offense.
Beginning with incapacitation, there is a sizable body of research estimating the amount of officially recorded crime prevented by a year of detention. The methods employed by these studies range from surveys of prison inmates regarding past offending (reviewed in Spelman, 1994, 2000), state panel data analyses (Johnson and Raphael, 2012; Levitt, 1996; Liedke, Piehl, and Useem, 2006), studies that exploit sentencing reforms that either enhance (Vollaard, 2012) or reduce (Owens, 2009) sentences for admittedly criminally active people, to studies that evaluate the effects of sudden, discrete, and policy-induced changes in correctional populations (Barbarino and Mastrobuoni, 2014; Buonanno and Raphael, 2013; Lofstrom and Raphael, 2016). The general findings from this research are the following.
First, most analyses find average incapacitation effects associated with prison incarceration and to a lesser extent jail incarceration. Second, these incapacitation effects tend to be much higher in low incarceration settings. That is to say, the marginal effect on crime of a one-person increase in the incarceration rate tends to be much higher in countries with low incarceration rates or in periods of time in the United States when the incarceration rate was low, a pattern suggestive of diminishing crime-fighting returns to scale. For example, Lofstrom and Raphael (2016) find no impact on violent crime and a modest effect on property crime of a very large decline in California’s incarceration that resulted from the 2011 reform that greatly reduced the likelihood of a parole revocation. It is noteworthy that the decline in incarceration caused by realignment occurred in a state with a total incarceration rate (prison plus jail) that exceeded 700 per 100,000. By contrast, evaluation of a similarly sized prison decline that resulted from Italy’s 2006 collective clemency, reported in Buonanno and Raphael (2013), revealed considerably larger incapacitation effect. Italy’s total incarceration rate on the eve of the clemency was roughly one-seventh that of California’s. Interestingly, the study also found smaller effects of prison releases on crime rates in Italian provinces with relatively high incarceration rates, despite the generally low Italian incarceration rate (suggesting diminishing marginal crime fighting effects even in a low incarceration rate setting). Liedke, Piehl, and Useem (2006) find evidence in state panel data indicating that incapacitation effects in the United States diminish with the incarceration rate, as do Johnson and Raphael (2012). The first of these two studies (Liedke, Piehl, and Useem, 2006) finds that at suf-
ficiently high incarceration rates the marginal effects of incarceration on crime may turn positive (suggesting that a criminogenic effect may swamp incapacitation at some point). These findings suggest that incapacitation effects are inherently heterogeneous—for example, likely to be larger for the young than the old, likely to vary with prior criminal history, and likely to vary on average with the extensiveness with which a given society deploys incarceration in an attempt to control crime.
Beyond incapacitation, results from research on whether changes in sanction severity deter criminal offending tend to be mixed. Among the studies finding evidence of general deterrence, Drago, Galbiati, and Vertova (2009) find that individuals released from Italian prisons under the 2006 Italian Collective Clemency who faced larger sentence enhancements for re-offending tended to recidivate at a lower rate. Similarly, Helland and Tabarrok (2007) find that individuals facing the prospect of a sentence enhancement due to state three-strikes laws recidivate at relatively lower rates (though the effect of even a very severe sentence enhancement is relatively small). In contrast, research exploiting the discontinuous increase in sentencing severity at the age of majority tends to find little evidence of an impact of the stiffer sentencing on offending (see Hjalmarsson, 2009; Lee and McCrary, 2009; for a comparable analysis with contrary findings, see Levitt, 1998) as does much of the research on the deterrent effects of capital punishment (NRC, 1978, 2012). A thorough review of the deterrence effects of sanction severity concludes that general deterrence effects associated with stiffer penalties tend to be small (Nagin, 2013).
While this existing body of research is instructive, the recent experiences of California are particularly relevant to the question at hand, since the state enacted several policies that led to discrete and large declines in both prison and jail populations. California’s incarceration rate on the eve of these reforms was quite close to the average for the nation and thus presents an abrupt change in incarceration rates for a high-incarceration rate system (a factor that the literature reviewed above that would suggest relatively small effects on crime from reducing the prison incarceration rate). By way of background, two broad factors converged to generate these policy changes. First, decades of litigation pertaining to conditions of confinement and the availability of health and mental health services in the state prison system culminated in a federal court order to reduce state prison overcrowding. Second, public opinion pertaining to sentencing severity and the use of incarceration in particular softened, resulting in several notable ballot measures aimed at undoing many of the stringent sentencing practices introduced in past decades.
To address the court order, California enacted broad corrections reform legislation under the banner of corrections realignment (passed in April
2011 and implemented on October 5, 2011).3 The legislation eliminated the practice of returning parolees to state prison custody for technical parole violations for all but a small set of the most serious offenders. The legislation also defined a group of nonserious, nonsexual, nonviolent offenders who serve their sentences in county jails. The act generated an immediate reduction in weekly prison admissions from roughly 2,100 per week to 600 per week and a steady, permanent decline in the prison population.
Regarding the change in public opinion, in recent years California voters passed several state ballot initiatives aimed at reducing the use of prison along both the intensive and extensive margins. In 2012, voters approved a ballot measure that narrowed the definition of felonies that would qualify for second- and third-strike sentence enhancements, limiting these felonies to serious and violent offenses (Proposition 36). More recently, voters passed a proposition that incentivizes prison inmates to engage in rehabilitative programming and refrain from institutional misconduct in exchange for shorter prison terms (Proposition 57 passed in November 2016).
The passage of Proposition 47 in November 2014, however, is perhaps one of the most far-reaching sentencing reforms passed by way of ballot initiative and had immediate impacts on the operations and practices of several different arms of the state’s criminal justice system. Put simply, the proposition redefined a subset of “wobbler” offenses (offenses that can be charged as either a misdemeanor or felony) as straight misdemeanor of-
3 The legislation was prompted by the state’s need to reduce its prisoner population in the wake of Brown v. Plata, 563 U.S. 493 (2011), in which the U.S. Supreme Court upheld a population reduction order entered by a three-judge 9th Circuit panel against the California Department of Corrections and Rehabilitation (CDCR). Plata consolidated two decades-long federal class actions—Plata and Coleman—brought on behalf of people incarcerated in the California prisons. Coleman concerned the constitutionality of mental health care delivery in the state’s prisons, and Plata challenged as unconstitutional the quality of the medical care. In both cases, federal district courts sided with the plaintiffs on the merits, finding that conditions in each context violated the Eighth Amendment prohibition on cruel and unusual punishment. After years of failure by the CDCR to remedy the constitutional violations found by the two courts, the cases were consolidated, and a three-judge panel was struck to consider whether a population reduction order was necessary. The three-judge panel sided with the plaintiffs on all counts, including a finding—required under 18 U.S.C.(a)(3)(E)—that prison overcrowding was the primary cause of the unconstitutional medical and mental health delivery in CDCR facilities and that no other less-intrusive relief would remedy the violation. The three-judge panel issued an order requiring the CDCR to reduce the population density of its facilities to 137.5 percent capacity (down from a high of roughly 200 percent capacity in 2006), and the U.S. Supreme Court upheld the order. The order left it up to the state to determine the best approach to achieving this reduction. Rather than releasing people from CDCR custody, Governor Brown instead spearheaded the passage of Assembly Bill 109 (referred to in the state as “corrections realignment”) which shifted to the counties responsibility for those people convicted of the lowest level crimes (a group known as the “non-non-nons,” for those convicted of nonserious, nonviolent, nonsexual offenses).
fenses. Regarding property offenses, the proposition redefined shoplifting, forgery, crimes involving insufficient funds, petty theft, and receiving stolen property offenses where the value of the property theft falls below $950 as misdemeanors. The proposition also eliminated the offense of petty theft with a prior. Regarding drug offenses, a subset of possession offenses were redefined as misdemeanors. These new charging protocols apply to all new cases with the exception of instances where the individual in question has certain prior convictions. The proposition also included a provision for individuals currently serving sentences for reclassified offenses to file a resentencing petition, as well as a provision for those convicted in the past to file a petition to have the prior conviction reclassified as a misdemeanor (California Judicial Council, 2016).
The cumulative effects of these reforms have been a large reduction in the state’s prison incarceration rate with slight overall reductions in the size of the jail population. Figure A-3 presents the prison incarceration rates for California and for the United States from the late 1970s through 2016. Both series exhibit pronounced increases during the last two decades of the 20th century. From the early 2000s on, however, there are notable departures with large relative decreases in California’s incarceration rates post 2010.
Figure A-4 presents long-term trends for overall California violent and property crime rates. Similar to national trends, California’s violent crime
rate peaks in the early 1990s before declining to current historical lows. While the historical peak for property crime occurs in the early 1980s, the largest declines in property crime occur post 1990, with the rate declining by roughly 50 percent over the subsequent 26 years. In both figures, the years 2010 (the last pre-realignment year) and 2014 (a year mostly preceding the implementation of Proposition 47) are marked with vertical lines. Notably, these reforms reduced the state’s prison incarceration rate to early 1990s levels while crime rates have remained at historical lows. Lofstrom and Raphael (2016) present a thorough analysis of the 2011 reforms and find no evidence of an impact of the large reduction in the prison population on violent crime rates, but evidence of a small effect on part I property crime (auto theft). Bartos and Kubrin (2018) conduct a comparable analysis of the 2014 passage and implementation of Proposition 47 and similarly conclude that there is no evidence of an impact on violent crime and mixed though not statistically significant indicators of a small effect on larceny theft.
Overall, the research on these changes basically confirms what is visibly evident in the figures. Despite a sizable contraction in the overall incarcerated population in the state, crime rates remain at historically low levels. At face value, this suggests that it is certainly possible to reduce incarceration in a high incarceration setting without jeopardizing public safety.
This page intentionally left blank.