
The Government is currently trying to reduce the number of prisoners in our jails. Recently it persuaded Lord Woolf to lean on judges to stop sending so many offenders to jail because of Treasury concern about the cost. But this view is very short sighted. If we take into account the full social and economic cost of allowing persistent offenders to roam free, prison is a bargain. Even on very cautious assumptions, for every £1 spent on prison, we would save at least £1.07 and it could be over £12.
There has been a long debate among US academics about how to calculate the costs and benefits of prison. One study by Professor John DiIulio estimated that the annual cost of keeping a criminal in jail is $25,000 and the total social and economic cost to society (including policing, insurance, injuries, replacing stolen property, and household expenditure on security measures) of allowing the median offender to remain at large is $70,098. The resulting costbenefit ratio is 25,000:70,098 or 2.80. On these assumptions it is worth jailing about 75% of criminals, but would be a waste of money to imprison lowrate offenders, who impose a cost on their fellow citizens of less than $25,000.
Based on a similar estimate of the social and economic costs of crime, how many more prison places should be provided in England and Wales? In the 2001 document, Criminal Justice: The Way Ahead the Home Office said that there were about 100,000 persistent offenders who carried out about half of all crime and that about 20,000 were in jail at any one time. What would it cost to incarcerate the remaining 80,000 and would it be worth it?
According to a Parliamentary answer given in December 2000, 12,265 additional prison places were provided between 1995/96 and 2000/01 at a total cost of £1.287 billion. This produces an average cost per place of £105,000.
The total cost of 80,000 places at £105,000 each would be £8.4 billion. The total annual cost of crime has been estimated by the Home Office to be £60 billion. The building programme would need to proceed in stages, perhaps at a rate of 5,000 places per year, or £525 million, an easily manageable figure.
What would the running costs be? The average cost of a prison place in 2002 was £38,753 per year. An additional 5,000 prisoners would therefore cost only £194m per year. On these figures, imprisoning the most serious and persistent offenders would be highly costeffective. If 100,000 offenders commit half of all crime, then they impose costs on society of £30 billion, or £300,000 each for every year they are free. Even if the building costs are charged to a single financial year and added to the running costs we arrive at a total of £143,753, a saving of £156,247.
But all such estimates are based on assumptions and everyone knows that if you tweak the assumptions you can get the answer you want. How can we produce a more reliable estimate? Rather than opting for a single set of assumptions, let's explore a range. The first relies on a Home Office selfreport survey of prisoners in 2000. The second is ultraconservative, the third emulates Home Office calculations used to work out the crimereducing effects of offending behaviour programmes, and the fourth is based on a study by Cambridge University's Professor Farrington to discover the costeffectiveness of youth custody. The cost estimates were made in 2000 at 1999 prices and should be compared with prison costs for a similar period. According to the Parliamentary answer given in June 2001, the average cost per prisoner place in a male closed Young Offender Institution (where the most persistent offenders might find themselves) was £23,063.
Prisoners SelfReports: The Home Office report, Making Punishments Work, reported the results of a survey of prisoners in 2000, which found that the average offender carried out 140 offences per year. The variation was large, and offenders who admitted to a drug problem were committing an average of 257 crimes per year.
If we were to jail 5,000 criminals who would otherwise have committed 140 offences, then 700,000 offences against the public would be prevented by 12 months in jail. If they were highrate offenders (257 crimes), the effect would be 1.3 million offences. According to a Home Office estimate in 2000, the average cost of crimes against individuals and households (excluding commercial crime) was £2,000. An offender committing 140 crimes per year would, therefore, impose costs on society of £280,000. If true, for every £1 spent on prison, we would save £12.14.
But, if a majority of crimes are carried out by a minority of offenders, doesn't the average exaggerate the rate of offending? Another method of calculating crimereducing effects has been used by Professor Farrington, also in a Home Office study. In 2002 he monitored for two years the reconvictions of a 'control group' of young offenders released from custody in 1997 and 1998. He then calculated the average cost to society of their crimes. It was £9,903 each, based on an average of 5.1 convictions each over the two years. This figure should be halved to produce an annual cost of £4,951, with the average cost per crime, £1,923, close to the Home Office estimate above.
No serious scholar believes that convictions measure the actual rate of offending, but if most crimes go undetected how can we estimate the true figure?
Ultraconservative assumptions: Professor Farrington found that offenders were convicted on average 2.57 times per year and that the average cost of each crime was £1,923 each, a total cost of £4,942. He thought that to estimate the real rate of offending, this figure should be multiplied by at least five, producing 12.85 crimes at a total cost of £24,710. In this case, for every £1 spent we save £1.07 (£23,063:£24,710)
Home Office assumptions: However, when the Home Office made a similar calculation of the crimereducing effects of one of its programmes it multiplied the number of convictions by five to correspond to police recorded crime and then multiplied the result by a further 4.2 to correspond to the British Crime Survey, which has consistently revealed more crime than police records. This method produces a total number of 54 offences per year, which in turn produces a total cost of £103,842. For every £1 spent we save £4.50.
Professor Farrington's preferred assumptions: Professor Farrington has expressed doubts about the 'five' multiplier used by the Home Office and cited his own earlier study of 18 yearolds in South London, which had found that for six types of crime (burglary, taking vehicles, stealing from vehicles, shoplifting, theft from automatic machines and vandalism) only about one in 30 led to conviction. If there are 2.57 convictions, they should be multiplied by 30 to arrive at the number of offences, 77. This produces a total social cost of £148,071. For every £1 spent we save £6.42.
Even if we make the highly unrealistic assumption that criminals only commit the offences for which they are caught and convicted, prison is good value for money. When we make more accurate assumptions, based on official crime figures and Home Office estimates of the social and economic costs of crime, incarcerating persistent offenders is not only good value for money, it's a bargain.

