DOES PRISON WORK? - OVERSEAS EVIDENCE (2003)
The Government has set out to reduce crime, but the evidence from a study comparing the policies pursued in the USA with those in England and Wales suggests it has adopted the wrong policies. From the early 1980s until the mid-1990s the risk of imprisonment increased in the USA and the crime rate fell; while in England and Wales the opposite happened: the risk of imprisonment fell and the crime rate increased. Then, from 1993, policy in England and Wales was reversed and the risk of imprisonment increased, though it remained historically low. Even this relatively small increase in the use of prison was followed by a reduction in crime.
Two-Country Comparison: USA and England and Wales, 1981-1996
A study carried out by one of Britain's most distinguished criminologists, Professor David Farrington of the University of Cambridge, in conjunction with Patrick Langan of the US Department of Justice, compared the USA with England and Wales between 1981 and 1996. Crime rates based on crime surveys were available in both countries up to 1995, and crime figures based on police records to 1996. The study investigated the possibility that increasing the risk of punishment would lead to falling crime. Two measures of the risk of punishment were used: the conviction rate per 1,000 alleged offenders; and the incarceration rate per 1,000 alleged offenders. The method of calculation is explained in the Appendix. Click here to check the original report (PDF).
The original study by Langan and Farrington was published in 1998 and gives imprisonment rates and crime rates in England and Wales based on the British Crime Survey up to 1995. The 1997 BCS results were not available and so the study does not fully capture the impact of the policy reversal in 1993. However, Professor Farrington and Darrick Jolliffe have subsequently updated the estimates for England and Wales to 1999, allowing us to appraise the impact of the policy reversal. This study is not yet published and we are very grateful to them for permitting us to quote from it.
The original 1998 comparison showed that the USA and England and Wales pursued very different policies and produced very different results. The study found that the chances of being imprisoned increased in the USA between 1981 and 1995 and fell in England and Wales. During the same period, crime fell in the USA and increased in England and Wales.
From 1981 to 1995 (1994 for the USA), an offender's risk of being caught, convicted, and sentenced to custody increased in the United States for all six crimes in the study (murder, rape, robbery, assault, burglary, and motor vehicle theft) but fell in England and Wales for all except murder. For example, in the US in 1981 there were 13 imprisoned robbers for every 1,000 alleged robbers. By 1994 there were 17. In England and Wales, there were 7 imprisoned robbers for every 1,000 alleged robbers in 1981, and only 4 in 1995. There were 5.5 imprisoned burglars for every 1,000 alleged burglars in the US, increasing to 8.4 in 1994. In England and Wales there were 7.8 in 1981 and only 2.2 in 1995.
The first set of graphs show the situation from 1981 until 1995, based on the Langan and Farrington study.
The Imprisonment Rate 1981-1995
Incarcerated robbers per 1,000 alleged robbers
Incarcerated assaulters per 1,000 alleged assaulters
Incarcerated burglars per 1,000 alleged burglars
Incarcerated vehicle thieves per 1,000 alleged vehicle thieves
What happened to the crime rate during this period? According to the 1995 victim surveys, rates of robbery, assault, burglary, and motor vehicle theft were all higher in England and Wales than the United States. According to 1996 police statistics, crime rates were higher in England and Wales for three crimes: assault, burglary, and motor vehicle theft. In the US in 1996 there were 9.4 burglaries for every 1,000 population, compared with 22.4 in England and Wales; and there were 5.3 car thefts per 1,000 population in the US compared with 9.5 in England and Wales. The 1996 recorded crime rate for robbery would have been higher in England and Wales than the United States had the police in England and Wales recorded the same proportion of robberies reported to them by the public as did the American police. In 1996 the US police recorded 78% of robberies reported to them, whereas the English and Welsh police recorded only 35%.
The Crime Rate 1981-1995
Robbery: Victim survey rate per 1,000 population
Assault: Victim survey rate per 1,000 population
Burglary: Victim survey rate per 1,000 households
Motor vehicle theft: Victim survey rate per 1,000 households
The major exception to the trend is the murder rate. The 1996 US murder rate was nearly six times higher than the rate in England and Wales, although the difference between the two countries narrowed from 1981 to 1996. Guns were more frequently used in violent crimes in the United States than in England and Wales. According to 1996 police statistics, firearms were used in 68% of US murders and in 7% of English and Welsh murders, and in 41% of US robberies but only 5% of English and Welsh robberies.
The overall US crime rate -- whether measured by surveys of crime victims or by police statistics -- was lower in 1996 than in 1981. In the US, for assault, burglary, and motor vehicle theft (according to victim surveys) and murder, robbery, and burglary (according to police records) the rates in 1996 were the lowest recorded in the 16-year period from 1981 to 1996. By comparison, English and Welsh crime rates in 1995 (as measured by the BCS) and 1996 (based on police statistics) were higher than they had been in 1981.
How large were the differences between 1981 and 1996? And how did they change during the period studied?
Public Policy Changes in the Two Countries
Langan and Farrington found that in England and Wales in the early-1990s, criminals faced a lower risk of punishment compared with the USA. Moreover, the risk had fallen between 1981 and 1995. Why did the risk of punishment fall in England and Wales and increase in the US? The study suggests three causes of diminishing conviction rates in England and Wales. First, there was an increased use of cautions and unrecorded warnings.(3) (This policy has subsequently been changed for young offenders.) Second, the Police and Criminal Evidence Act of 1984 increased the procedural safeguards for the accused. And third, the Crown Prosecution Service was established in 1986, leading to an increased tendency to drop cases.(4)
Two special factors caused the decreasing risk of prison from 1987-1991. First, official Home Office advice encouraged judges to make less use of prison and second, theft of a motor vehicle was downgraded in 1988 to a non-indictable offence encouraging the use of non-custodial sentences.
From 1993, however, Home Office policy changed and the use of prison was encouraged, especially for repeat offenders, although the rate of imprisonment remained low compared with the US. In the US, however, during the same period, the police made more arrests as a percentage of total alleged offenders and prosecutors obtained more convictions. And, after 1986, US prisoners served a longer percentage of their sentences.
The Risk of Punishment or the Severity of Punishment
In addition to examining the impact of changes in the risk of punishment on the crime rate, Langan and Farrington also looked at the impact of changes in the severity of punishment. Four measures of severity were used: the proportion of those convicted who were sent to prison; length of the sentence; actual time served; and the percentage of the sentence served. They also used a measure called 'days of incarceration at risk of serving', which combines elements of both risk and severity.(5)
A negative correlation between the risk of punishment and the rate of crime was taken as support for the theory that an increased risk of punishment leads to a fall in crime. In England and Wales they found strong support for the theory that 'links falling risk of punishment to rising crime'.(6) After 1981 the conviction rate in England and Wales fell and the crime rate (whether based on victim surveys or police records) rose. Similarly, the incarceration rate fell and the crime rate rose. However, the correlations between the severity of punishment and the crime rate were mixed. There was, however, a strong link between the severity of punishment of car thieves and the rate of vehicle theft. After 1981, the proportion of car thieves sentenced to prison, their average sentence, the time served and the percentage of sentence served, as well as the number of days of actual incarceration, all fell. During this time, vehicle theft rose according to both the British Crime Survey and police records.
Was it the greater risk of punishment that explains the difference in crime rates in England and Wales at the end of the period? Or was it the severity of sentencing? Sentences were likely to be longer in the US. For all offences (murder, rape, robbery, assault, burglary, motor vehicle theft), courts in the United States sentenced convicted offenders to longer periods of incarceration than courts in England and Wales, and the length of time actually served before being released was also longer in the United States than in England and Wales. However, over the period, sentences for serious crimes generally did not increase in length in the United States, while in England and Wales sentences generally did get somewhat longer for the three violent crimes, murder, rape and robbery. Overall, this evidence is consistent with the theory that the important factor in reducing crime is the risk of imprisonment rather than the severity of the sentence as such.
Historical Comparison: England and Wales Before and After 1993
Click here for a discussion of the impact of the 1993 reversal of policy.
Crime and Punishment in England and Wales 1950-2000
The Langan and Farrington study provides powerful evidence, but only looks at the period since 1981 because the British Crime Survey began that year. By then, however, crime (based on police records) had increased sharply compared with the 1950s. To understand changes since 1950 we have used police records. We looked at the changing risk of punishment in England and Wales for all indictable offences, and also took a separate look at burglary, robbery and theft and handling stolen goods.
We have used the number of offences as a baseline and measured the risk of punishment in three ways: the risk of detection (the clear-up rate); the risk of conviction (cautions and convictions); and the risk of imprisonment (immediate custody). Each is presented as a proportion of the total number of offences.
In 1950 there were just over 1,000 crimes for every 100,000 population. The number rose to a peak of 10,900 in 1992, falling back to 9,700 in 2000, nearly ten times the 1950 rate. During the same period the risk of being caught, convicted and punished fell.
Convictions or cautions
The graphs show the long-term trends.
Recorded Crime per 100,000 Population, 1950-2000
Detection, Conviction and Imprisonment, 1950-2000
The Clear-Up Rate, 1950-2000
The Conviction Rate, 1950-2000
The Imprisonment Rate, 1950-2000
Is the Blair Government pursuing the right policies? The Government is ambiguous about prison. In its 2002 white paper, Justice For All, it says that it wants to send the 'strongest possible message' to criminals that the system will be effective in 'detecting, convicting and properly punishing them'. So far so good: after many years of being opposed to prison and favouring community sentences, the Government now recognises that prison protects the public more effectively. But prison is to be reserved for 'dangerous, serious and seriously persistent offenders and those who have consistently breached community sentences'. For the bulk of criminals, the Government still hopes to find alternatives to prison that combine community and custodial sentences, including weekend prison and more intensive supervision by the Probation Service. Moreover, the use of the term 'seriously persistent offender' rather than just plain 'persistent offender' suggests that criminals serving community sentences will be allowed to commit more than one offence before being sent to jail.
Comparison of the policies pursued in the US and England and Wales suggests that the strategy outlined in Justice for All is unlikely to achieve its intended purpose. For crimes such as robbery, burglary, car theft and assault, increasing the risk of imprisonment has produced a fall in crime in the USA. It appears to be less effective for murder and rape and we may conjecture that this is because the motives or emotional drives leading to these offences are less subject to rational calculation. Where the crimes are calculated to acquire material possessions, potential offenders may be more likely to weigh up the risk of being punished. People addicted to drugs are an exception - but it is a truism to say that they are not likely to be thinking clearly, precisely because they have fallen under the sway of a narcotic substance.
The Blair Government is actively promoting alternatives to prison, a strategy that may well be seen as weakening the risk of punishment. Two effects led to the fall in crime in America. First, there was a deterrent effect and second, an incapacitation effect. It is clear from the evidence set out so far that the Government's proposals are likely to weaken the deterrent effect, because criminals will be aware that they face a lower risk of imprisonment. But, in addition, the proposals will also weaken the incapacitation effect, because more offenders will be at large rather than in custody. A criminal cannot commit offences in the wider society while he is in prison, but a person under a community sentence can. The Government partly acknowledges this weakness and plans to overcome it by increasing the level of supervision of offenders in the community - under intensive supervision and surveillance programmes. However, the greater the time spent unsupervised, the longer the time offenders will have available for crime. And, however tough sounding the language used to describe community sentences, it is an inescapable fact that the Government is increasing the amount of time known offenders will be unsupervised. They may be required to attend a training course or to go to work during the working day, and they may be tagged from 7.00 pm until 7.00 am, but this leaves them free to carry out crimes on the way home from work or college.
Prison works as a method of protecting the public and deterring criminals, but some commentators are reluctant to accept the truth of this conclusion because they feel that punishment and the rehabilitation of offenders are mutually exclusive alternatives. Prison is certainly a punishment, but it is not only a punishment. It is also a means of protecting the public from known offenders and of deterring others. It is also - as the Prison Service fully acknowledges - an opportunity to reform criminals in the hope of encouraging a law-abiding lifestyle on release. And, as the Prison Service freely admits, efforts to reform prisoners are much in need of improvement. Measures intended to rehabilitate offenders while they are in custody need to be quickly and substantially stepped up. Much has been learnt about how to discourage anti-social attitudes, to create a stronger moral sense, and to encourage a more positive attitude in convicted criminals. But while their offending continues, it is preferable to rehabilitate them in prison, so that members of the public continue to be protected.
The evidence produced by this comparison of America and England and Wales suggests that, to the extent that it is reducing the risk of punishment and decreasing the supervision of known offenders, the Blair Government is doing the opposite of what is necessary to reduce crime.
Langan and Farrington's Method
Langan and Farrington began with the total number of offences, drawing on both victim surveys and police records. They used 'offenders' rather than 'offences' because many crimes are committed by more than one person. Robberies in England and Wales, for example, are carried out by an average of 2.4 people, vehicle thefts by 2.1 people and burglaries by 1.8. They multiplied this average by the number of offences to calculate the total number of criminals at risk of punishment (called 'alleged offenders' in the text). For example, in America in 1994 there were 5,482,720 domestic burglaries. This figure was multiplied by the US average number of offenders in burglary cases (1.4) to arrive at 7,675,808 offenders at risk of conviction. They are not, of course, all different people.
Langan and Farrington then calculated how many alleged offenders were detected, convicted and imprisoned. For England and Wales, the definition of 'immediate imprisonment' is defined in Criminal Statistics England and Wales 2000.(7) The main categories are sentences with no part suspended, referred to as 'unsuspended imprisonment', and for persons aged 21 and over, 'immediate custody'. For persons aged under 21, 'immediate custody' is equivalent to detention in a young offender institution from October 1988.
England and Wales 1950-2000
This information uses the same definitions as the Home Office. Therefore, Burglary includes acts 27, 28, 29, 30, 31, 32, and Theft and Handling Stolen Goods includes acts 37, 37.2, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 54, and 126. The definitions of some acts were revised over the time period, and some of these acts were only recently introduced.
Data for all indictable offences use a combined figure of those found guilty and those cautioned by the police. Data for theft and handling of stolen goods, burglary and robbery use a combined figure for those found guilty and those cautioned, apart from years 1950-1965 which do not include figures for cautions. Sentences classified as immediate custody are explained in the previous note.
1. Secretary of State for the Home Department, the Lord Chancellor and the Attorney General, Justice for All, Cm 5563, London: 2000, p. 87.
2. Justice for All, p. 87.
3. Langan, P. and Farrington, D., Crime and Justice in the United States and in England and Wales, 1981-96, Washington: US Department of Justice, 1998, p. 43.
4. Home Office, Criminal Statistics 1993, Table 6.2.
5. Langan and Farrington, p. 38.
6. Langan and Farrington, p. 38.
7. Home Office, Criminal Statistics 2000, Cm 5312, p. 249.