Police, prisons and crime rates

Richard Garside
Wednesday, 11 July 2012

The man with a hammer tends to see every problem as a nail. By the same token, the think tank in search of prison as the answer to crime will surely find it.

So it is that Civitas, a think tank that has long argued in favour of higher imprisonment rates, has published a paper arguing that tougher prison sentences result in lower crime.

The paper, written by the University of Birmingham economist Siddhartha Bandyopadhyay, has a suitably academic feel to it. Tables with linear and quadratic regressions mix it with an impressively incomprehensible graph and knowing references to the ‘Sargan Hansen test’.

Those committed to the 'more prison equals less crime' proposition will seize on the paper, mostly without reading or understanding it. Those who have their doubts might struggle through the first couple of pages before turning to something more interesting.

So what does it argue? Those who did not get beyond the paper’s breathless media coverage might be surprised at its most striking finding: the more young adults - 15 - 24 year olds - living in a given area, the fewer are the burglaries, thefts, robberies and frauds recorded by the police.

The paper also finds a relationship between increases in the number of crimes the police successfully clear up - the so-called detection rate - and average sentence lengths on the one hand, and the numbers of burglaries, thefts, robberies and frauds recorded by the police on the other hand. Higher detections and longer sentences are related to some lower recorded crime rates, though not, apparently, robbery in the case of prison sentences.

If this all sounds like a rather flaky argument, that is because it is built on hopelessly shaky foundations.

It relies on police recorded crime data, a measure the author himself concedes is ‘a far from perfect indicator’. I have written elsewhere about the problems of using police recorded crime data. Relying on them to assess levels of crime in an area is akin to assessing the health problems of a neighbourhood through a Friday night visit to the local A & E department.

Statistical modelling, however impressive, tells us nothing of value when the data upon which it is built is stretched to snapping point.

At best, the paper demonstrates that certain factors - demography, income and employment levels, police and sentencing activity - are correlated, when different areas of England and Wales are compared, with the propensity of the police to record certain crime types.

These are interesting observations about bureaucratic processes. They do not prove that tougher sentences and higher police detections results in lower crime in the real world. Indeed the number of police detections fell at a faster rate between 2002 and 2010 than did police recorded crime.

The argument in favour of improved police detection and tougher sentencing also relies on a naive view of human nature, peculiar to economists, that individuals are largely rational calculators who calmly weigh up the benefits and disbenefits of any given action: stealing an iPhone for instance versus a long spell in prison if caught.

The idea that a career burglar might decide to ditch his trade in favour of running a flower stall because of marginal increases in prison sentence lengths strikes me as one of the more absurd implicit propositions of this paper. The obverse - that were it not for the threat of imprisonment I would happily mug my fellow passengers on the train home - is equally ridiculous.

Rational calculation might inform consumers' decisions to switch electricity suppliers. Applying the same logic to the complex social relationship that is crime is a mystifying simplification.

What is missing in all this dessicated number crunching is anything approaching a coherent analysis of human nature and motivations, the effects of social relationships and the potential for policy change that transcends the prison-punishment dead end.

The lesson of all of this: impressive graphs and complex tables do not of themselves a good argument make.