VoxEU Column Frontiers of economic research

How to reduce high incarceration rates

How to reduce incarceration rates without fuelling a crime boom? This column argues that by being more selective over whom to lock up and for how long, scarce public funds can be put to better use.

Incarceration is costly – easily €100 to €200 per night per prisoner, depending on the country and the prison regime. That makes €36,500 to €73,000 per prisoner per year, excluding fixed costs of building prisons, and all other costs such as time not spent at work or with the family.

Given the pressure on government budgets, many states and countries are looking into reductions in prison expenditures. The short-term solution is to reduce the number of prisoners by way of early release. The longer-term solution is to change sentencing policy. This is all easier said than done, however, given public concerns about the effect of lower incarceration rates on crime. Incarceration also provides important benefits to society after all, including deterrence (Durlauf and Nagin 2011) and incapacitation (Owens 2009) of offenders.

We argue that a reduction in the number of inmates taxes these benefits of incarceration the least if it happens selectively. Being more selective in whom to incarcerate for how long puts scarce public resources to their best use.

Selective incapacitation

The idea of selective incapacitation is to make a distinction between offenders with a high and with a low propensity to commit crime. Those of the high propensity type – the prolific offenders – are responsible for a large share of violent and property crime (Tracy et al. 1990). To them, the default penalties have little deterrent effect. By making the length of a prison sentence conditional upon an offender’s criminal record, enhanced prison sentences can be targeted at this population. After all, by repeatedly breaking the law, these offenders reveal themselves to be of the prolific type (Polinsky and Rubinfeld 1991). Once the harsher sentences apply, the penalties may begin to make a difference, if not through deterrence, then by way of incapacitation in prison.

Distorted views

Selective incapacitation has received a lot of bad press. The main reason is the not-so-sensible application of the idea in the state of California. In the 1990s, this state adopted a sentencing policy commonly known as the ‘three strikes law’.

What makes this law so problematic? First, the three strikes law added sentence enhancements for repeat offenders to prison sentences that were already long. As incidence of criminal activity decreases as the offender ages, enhancing long prison sentences may 'incapacitate' offenders who would have refrained from crime for most of the extra time that they spent in prison anyway. Second, California's law is not very selective. One or two prior convictions for a wide range of crimes can trigger prison terms of up to 25 years or life. Casting such a wide net will catch many prolific offenders, but also many people who are already at the end of their criminal careers. Incarcerating these marginal offenders will lower the crime-reducing effect of prison and the net social benefit of this policy.

Even though California's three strikes law is not particularly selective, virtually all evaluations of selective sentencing are based on this particular law. It has been found to have a deterrent effect, albeit at high cost (Helland and Tabarrok 2007). How much of an incapacitation effect the California law had remains unclear (see Vollaard 2012 for a review). Notwithstanding the absence of evidence on the overall crime-reducing effect of the policy, commentators and scholars alike are quick in condemning a policy of selective incapacitation as part of California's ill-guided strategy of 'mass incarceration' (see Chen 2008).

Experiences outside California

Given the limited evidence outside the California context, experiences with a policy of (truly) selective incapacitation elsewhere should be particularly informative. Policies of similar nature have been adopted in a number of countries, including Great Britain and Australia in 1997, the Netherlands in 2001, Hungary in 2009, and New Zealand in 2010.

Recently, we published the results of a study into the costs and benefits of a habitual offender law adopted in the Netherlands in 2001 (Vollaard 2012). The law was highly selective. Only offenders with ten or more offenses on their criminal record and a history of being resistant to any short-term rehabilitative programme faced the enhanced prison-terms. Between 2001 and 2007, 1,400 mostly non-violent, relatively old and invariably drug-addicted offenders were sentenced under the law. They accounted for 5% of the prison population. The law implied sentence enhancements of some 1,000%, typically a two-year rather than a two-month sentence for the affected offender population. While this is a huge increase in sentence length, the overall rate of incarceration was hardly affected because of the small number of offenders involved. To identify the crime-reducing effect of the law, we exploited the fact that its adoption at the local level happened under quasi-experimental circumstances and that the criminal activities of the affected offender population were strongly tied to a specific locality.

Figure 1. Average rate of theft from car and domestic burglary – pre and post-introduction of the Dutch habitual offender law

Note: Plotted coefficients show the average crime rate relative to the month preceding introduction of the habitual offender law. The bars show the 95% confidence intervals. Based on monthly data for 31 cities during 1998-2007.

Source: Vollaard (2012).

We find that, on average, the sentence enhancements resulted in a 25% drop in acquisitive crime – exactly the types of crimes that the affected offenders committed. Figure 1 illustrates the average impact of the policy in the years after its introduction. We show that the law did not have an impact on violent and sexual crimes, offenses that were rarely committed by the affected offenders. We do not know of any crime policy in the history of the Netherlands that was similarly effective.

In addition, we find the benefits of the policy to exceed the costs by a large margin. These benefits go down rapidly with a more intensive use of the law, however. The marginal crime-reducing effect of convicting another prolific offender to an enhanced prison sentence declines by some 25% when going from low to high use of the law. During 2001-2007, the benefits of the policy remained higher than the costs, even for the cities which used the law most intensively.

Lessons for policy

These findings suggest that making the length of prison sentences more dependent on someone’s prior criminal record can be a very cost-effective crime policy. The Dutch policy affected only 5% of the prison population, but reduced property crime rates by 25% to 40%. How effective such sentencing policies are in other contexts depends on existing policies and how selective the law is.

What does that teach us for states like California that are struggling with high incarceration rates and accompanying high costs of its prison system?
First, when prisons are overcrowding, the default prison sentences for low-rate offenders can be reduced (instead of enhancing sentences for high-rate offenders, a strategy for regions with short prison sentences). That also results in a greater distinction in prison terms between high and low-rate offenders. At the same time, in states like California and Florida, the additional years in prison for repeat offenders can be greatly limited – at limited cost in terms of higher crime rates. The additional prison years that extend into ages at which very few people still show criminal inclinations could be scrapped.

Second, the enhanced sentences can be limited to only the most prolific of offenders. That would mean going from two to three strikes as the threshold for the harsher sentences in California to, for instance, five to ten strikes.

Finally, in line with the experiences in the Netherlands, additional criteria for receiving an enhanced sentence can be formulated, such as a history of denying drug treatment or a failure to successfully complete other rehabilitative programmes. That would make the sentence enhancements even more selective, making sure that our criminal justice euros are spent on the most harmful criminals.


Chen, EY (2008), “Impacts of ‘Three strikes and you’re out’ on crime trends in California and throughout the United States”, Journal of Contemporary Criminal Justice, 24(4):345-370.

Durlauf, S and D Nagin (2011), “The deterrent effect of imprisonment”, 43-94, in PhJ Cook, J Ludwig, J McCrary (eds.), Controlling crime: Strategies and Tradeoffs, University of Chicago Press.

Helland, E and A Tabarrok (2007), “Does three strikes deter? A nonparametric estimation”, Journal of Human Resources, 42(2):309-330.

Owens, Emily (2009), “More Time, Less Crime? Estimating the Incapacitative Effect of Sentence Enhancements”, Journal of Law and Economics, 52(3):551-579.

Polinksy, AM and DL Rubinfeld (1991), “A model of optimal fines for repeat offenders”, Journal of Public Economics, 46(3):291-306.

Tracy, PE, ME Wolfgang, and RM Figlio (1990), Deliquency careers in two birth cohorts, Plenum Press.

Vollaard, Ben (2012), “Preventing crime through selective incapacitation”, The Economic Journal, forthcoming. (a copy is available upon request from the author)

2,730 Reads