17 January 2011

Does Recidivist Sentencing Work?

Making the length of a prison sentence conditional on an individual’s offense history is shown to be a powerful way of preventing crime. Under a law adopted in the Netherlands in 2001, prolific offenders could be sentenced to a prison term that was some ten times longer than usual. We exploit quasi-experimental variation in the moment of introduction and the frequency of application across 12 urban areas to identify the effect. We find the sentence enhancements to have dramatically reduced theft rates. The size of the crime-reducing effect is found to be subject to sharply diminishing returns.


From the abstract to Preventing Crime Through Selective Incapacitation by Ben Vollaard.

Some highlights:

[J]udges almost exclusively sentenced drug-using, older individuals under the law for whom there was thought to be no hope of preventing high-rate offending by any other means than incapacitation. Most of the convicted offenders were not able to maintain a normal life style. They were out of work and did not have stable housing. They committed theft for a living, collecting a daily income of some 50 to 100 euro ($70-130) to be able to maintain their habit, which implies stealing property valuing some 300 to 600 euro ($400-800) on a daily basis. By 2001, many of these highly prolific offenders were aged 40 or over: they had fallen victim of the heroin epidemic that swept Europe back in the 1980s. The offenders spent some three to four months in prison each year in absence of the new law, and some had as many as 300 offenses on their criminal record. On average, offenders had been convicted 31 times prior to being sentenced under the habitual offender law. The enhanced prison sentence was not only meant to reduce crime through incapacitation but to provide a window for coercive treatment as well. Incarceration was often combined with drug treatment and other rehabilitative services, such as social skills training. Evaluations of the law suggest the treatment programs had little effect on recidivism. . . .

The offense data show that 85 percent of all offenses known to the police were committed within the urban area an offender has been assigned to. The other 15 percent of offenses were mostly committed in smaller communities directly bordering the urban area. . . .

[T]he rate of theft is some 30 to 40 percent lower as a result of selective incapacitation of prolific offenders. The size of the drop in crime corresponds with the results of some back-of-the-envelope calculations. If 1,200 offenders are responsible for 70 percent of crime, as we argued above, then the close to 700 offenders serving time under the law by mid-2007 are responsible for 40 percent of crime. . . . The crime-reducing effect of the law is smaller than this percentage share, however, since some of the offenders would have been doing time also in absence of the law. Assuming 8 additional months of incarceration per year as a result of the law, the drop in crime can be put at some 30 percent, which is close to what we find. . . . under the assumption that the affected offenders spend 8 additional months in prison per year, the law prevents some 80 thefts from car and 9 domestic burglaries annually per long term incapacitated offender. That implies that the costs per crime prevented are equal to some 600 euros. In other words, if the social costs of a domestic burglary and a theft from car are higher than 600 euros, then the policy is welfare improving. Estimates of the costs of crime are surrounded by controversy. Ex post approaches estimate the cost of crime that has already occurred to identifiable victims. Based on jury awards, Roman (2009) estimates the average costs of a burglary to be $4,444 (3,300 euro). Accounting for all of the costs that are known to be related to crime, including damage and the costs of use of the criminal justice system, the Home Office produces a somewhat higher estimate of the costs of a burglary of 4,600 euro ($6,000) per incident. A separate estimate for the cost of a theft from car is only available from the Home Office study, which puts it at 1,200 euro ($1,600) per incident. . . .

A habitual offender law adopted in the Netherlands in 2001 allowed for a two to three year prison sentence for offenders with ten or more offenses on their criminal record. Although the group of offenders sentenced under the law accounted for only 5 percent of the prison population six years after its introduction, the sentencing policy lowered the rate of burglary and theft from car by an estimated 40 percent through the incapacitation effect alone. The estimated impact of the law is large, but in line with self-reported crime. In addition, police counts of active prolific offenders are found to go down proportionally with the number of prolific offenders serving extra time in prison as a result of the law. . . .

Even for this highly selective sentencing policy that only affected 1,400 offenders in the period 2001-2007 we find evidence for rapidly decreasing returns to scale. The marginal crime-reducing effect of incapacitating another prolific offender declines by more than half from the lowest to the highest rate of application of the law. The benefit-cost ratio drops sharply when more offenders are serving time under the habitual offender law. The social returns to selective incarceration remain positive over the whole range of application of the policy, however. . . .

The incapacitation effect may be particularly large in the case of the Netherlands as the habitual offender law primarily affected offenders that were addicted to drugs, heroin in particular. These offenders tend to have an age-crime curve that is flatter than that of other groups of offenders – even other prolific offenders. Possible negative effects of longer prison sentences on the life of offenders such as disruption of employment, relationships and housing were limited as most of the affected offenders were out of work and did not have stable housing.

The Dutch policy of selective incapacitation started from a low base. The rate of incarceration in the Netherlands around 2001 was similar to the rate in the beginning of the 1970s in the US, for instance. Enhancing prison sentences of a few weeks or months to three years is likely to have a greater payoff in terms of preventing crime than enhancing prison sentences that are already many years long. To compare: an enhanced prison sentence for burglary of 2 to 3 years based on the Dutch habitual offender law is comparable to the default sentence for burglary in the United States. Our finding that the habitual offender law adopted in the Netherlands had a large incapacitation effect should therefore not be interpreted as evidence that all policies of selective incapacitation are likely to have a similarly favorable cost-benefit ratio. Given the rapidly diminishing returns to incarceration, the high costs of the enhanced prison sentences may soon exceed the benefits of crime prevented.


The delicacy with which the Dutch approach a very modest habitual offender law that applies only to those with ten previous offenses (although many of those would be misdemeanors under U.S. law and generally not within habitual offender provisions except under laws that allow aggravation of multiple theft amounts into a single felony theft prosecution), is itself fascinating. Habitual non-violent felony property crime defendants in Colorado and most U.S. states receive sentences many times as long.

Also, the fact that someone in the Dutch public policy process is even seriously trying to quantify the cost-benefit ratio of incarcerating habitual thieves for long periods of time, and considering even the economic impact of the sentence on the thieves themselves, is itself interesting. So too is the fact that somebody in the Dutch criminal justice policy making world finds bicycle thefts, larcenies from cars, and non-violent burglaries of unoccupied dwellings worth of interest at all. By comparison, most American criminal justice policymakers are so interested in reducing violent crime that any other objective is virtually irrelevant.

Felony Sentencing In The United States

A link to the latest data on felony sentencing in the United State can be found here (almost no state or national level statistics are kept on misdemeanor sentencing apart from overall jail incarceration rates, records of court cases commenced, and sometimes a distinction between the share of those in jail who are awaiting trial and those who are actually convicted). This year's highlights:

* In 2006 an estimated 69% of all persons convicted of a felony in state courts were sentenced to a period of confinement--41% to state prison and 28% to local jails.
* State prison sentences averaged 4 years and 11 months in 2006.
* Men (83%) accounted for a larger percentage of persons convicted of a felony, compared to their percentage (49%) of the adult population.
* Most (94%) felony offenders sentenced in 2006 pleaded guilty.


Jury trial rates do not exceed 5% for any category of crimes other than serious violent crimes. Even 61% of murder convictions are the result of guilty pleas.

The average burglary sentence in the U.S. in 2006 was 44 months of incarceration where incarceration is imposed (median 24 months), and the average larceny sentence was 22 months (median 12 months) of incarceration where incarceration is imposed. Of course, many and probably most of those sentences don't involve recidivist offenders.

Life sentences are imposed in the U.S. in about 25% of murders, 5% of rapes, 1.5% of robberies, 0.8% of non-rape sexual assault cases, 0.6% of aggravated assaults, 0.4% of other violent crimes, and 0.1% of non-violent crimes for which incarceration in prison is imposed as a sentence.

Surprisingly, no fine is imposed in 55% or more of cases for every particular kind of felony. Overall, fines are not imposed 62% of the time.

Colorado Sentencing in 2009

In Colorado, there were 42 habitual offender commitments to prison in fiscal year 2009 (see page 23 of the pdf), compared to 26-66 per year in years since 2005. These included a 40 year sentence for forgery, a 25 year sentence for burglary, four cases with an average 36 year sentence for burglary, three theft sentences of an average of 18 years each, four for trespassing or criminal mischief for an average of 11 years each, a 6 year forgery sentence, a 6 year perjury sentence, and an 11 year trespassing/criminal mischief sentence. Three habitual offender sentences were for escape, fourteen were for violent or weapon related crimes, six were for drug crimes, and two were for a felony traffic offense. All but one of the 42 habitual offenders were men and the trigger for habitual offender sentencing is two or three prior felonies of the relevant type under each habitual offender statute (there are several). While these terms are long, they are far more lenient in most cases than under California's controversial three strikes law. The average Colorado prisoner serves about half their full sentence due to good time and other considerations.

About 3% of offenders in Colorado prisons were sentenced under habitual criminal statutes. In all there were 22,961 people in Colorado prisons as of June 30, 2009. Another 12,773 were on parole.

The "need assessments" of Colorado prisoners is always notable.

Educationally, just 1% of those admitted to prison had an associates degree or more education although about 11% have some college, while 37% lacked a high school diploma with 36% being at least functionally illiterates who needed adult basic education instruction, rather than high school level GED instruction which would be too advanced for them. About two-thirds of those with either a high school diploma or GED had a GED rather than a high school diploma. So, less than a quarter of Colorado prison inmates graduated from high school in the ordinary course. In Colorado as a whole, 11% lack a high school diploma or GED, 89% of the age 25+ population has at least a high school diploma or GED, 65% have at least some college, 43% have an associates degree or higher degree, and 33% have a bachelor's degree.

About 8% had an IQ of under 81. A moderate to severe mental health problem is an issue for 30%. A moderate to severe substance abuse problem is an issue for 79%. A moderate to severe medical problem is present in 15%. Sex offenders make up 11% with another 5% suspected of having sex offense histories who are not convicted. An absence of adequate skills to get a job is a factor for 42%. Mental health needs differed considerably based on gender. A moderate to severe mental health problem was an issue for 22% of men and 55% of women.

The DOC doesn't include crosstabs in its annual report or relate needs data to recidivism data, although some data along that line are collected in a separate report and here. Offenders with mental health issues are slightly more likely to lack of high school diploma or GED (31%-32% v. 28%), to lack job skills (94% v. 91%), to be sex offenders (22%-24% v. 18%), to have substance abuse problems (80%-83% v. 78%) and to have anger issues (40%-41% v. 39%) than other inmates. They are much more likely to have medical problems (25%-28% depending on severity v. 16%), to have IQ below 81 (about 8% v. 4%), and to have suicidality issues (about 21% to 30% depending on severity v. 9%). Only 30% of inmates without a substance abuse problem have a high school diploma and 24% have neither that nor a GED.

Some mental health data don't make much sense. Those who were classified as having mental health issues often had prior psychiatric hospitalization (18%-24% depending on severity) and out patient mental health treatment (42%-47% depending on severity), but among those not classified as having mental health issues, 5% had prior psychiatriic hospitalization and 27% had prior outpatient mental health treatment, suggesting significant underdiagnosis of mental health issues by the DOC. Among those with mental health issues 23%-34% had a history of psychotropic medications, but so did 4% of those not so classified. Notably, less than 1% of inmates with mental health issues had a prior not guilty by reason of insanity case.

The most common mental health conditions were drug addiction, depression, bipolar disorder, anxiety disorders, alcoholism, schitzophrenia and psychotic disorders, dsythmic disorders, "disorders usually diagnosed in childhood" like ADHD, and "sexual and gender identity disorders" 1%. In all 34% of disciplinary violations were attributed to the 25% of inmates classified as having mental health issues in the detailed study on the issue, and these inmates were much more likely to be in solitary confinement or "close" supervision than other inmmates (23%-24% v. 11%), despite generally similar offense severity.

The overall percentage of inmates with moderate to severe needs in some category other than job skills (which almost all inmates seem to lack) is probably in excess of 90%, and once job skills are considered is probably in excess of 95%.

The DOC also doesn't detail good time forfeitures or gang crime connections in its annual report, although it tracks both. About 7% of Colorado inmates are eligible for deportation upon release because they are not U.S. citizens. About 9% are foreign born (the same as the 9% of the general Colorado population that is foreign born foreign born), but the remainder are U.S. citizens not eligible for deportation. Colorado's inmates are 45% Anglo, 32% Hispanic, 20% African American, 3% Native American and 1% Asian. Colorado as a whole is 71% Anglo, 20% Hispanic, 4% African American, 1% Native American and 3% Asian.

For prisoners released in 2005 and for prisoners released in 2006, the three year return to prison rate was 53.2% (a little higher for men, a little lower for women). The largest proportions of returns to prison are for technical violations of parole. For example, looking at the cohort released in 2004, the percentage of the cohort commiting ofirst time technical violations or new crimes by number of years from release is as follows:

---------------New Crime----------Technical Violation
Year 1----------8.8%------------------24.1%
Year 2----------6.2%-------------------7.8%
Year 3----------3.6%-------------------1.9%
Year 4----------2.3%-------------------0.4%
Year 5----------1.8%-------------------0.3%

Even a 1.8% crime conviction rate per year is huge compared to the general population. But, inmates in Colorado prisons are highly atypical of the general population in a variety of "needs" categories as well as in having a history of having seriously violated the law. About 43%, however, will go five years from release from prison without a single technical violation forcing their return to prison or new crime, and about 77% will not be returned to prison for a new crime in that five year time period.

Recidivism rates are higher for those with moderate mental health issues than those with none or those with severe mental health issues.

The average annual cost of incarcerating an inmate in Colorado was $32,338 in fiscal year 2009. The state has 23 prisons. None of them was under 100% of capacity. Twelve were over 100% of capacity, in one case at 186% of capacity (Skyline Correctional Center). The state prison system employs more than 6,000 people full time. State prison populations did almost hold steady in 2009 (increasing by only 189 inmates), however, after many years of steady and rapid prison population growth due to both decreased admissions and increased releases. New crime convictions leading to prison sentences declined, while technical parole violations rose.

No comments: