Last week I posted an analysis of trends in police data over the past decade, as part of an argument that talk of 'overall crime' was best left to crime involving overalls. I concluded by arguing that we might see a rise in homicide in the coming years as a result of the current economic depression.
This piece takes the analysis in a slightly different direction, taking a look at data in the Crime Survey for England and Wales (CSEW) and the British Crime Survey (BCS) over a longer time-frame: 1981 to 2012. The underlying frame of reference of this piece is the early 1980s recession. As Danny Dorling pointed out in his important study of homicide trends, published here, the 1981 recession had a significant and lasting impact. Those young men in particular who entered the jobs market in the early 1980s, when jobs were few and far between, experienced an elevated risk of homicide victimisation that was still measurable in the late 1990s, nearly 20 years on.
To put the point differently, one way of looking at the BCS/CSEW dataset for the 1981 to 2011/12 period is as a chronicle of the effects and after effects of the early 1980s recession. The rise in 'overall crime' from 1981 to 1995 can then be seen as a long-wave social reaction to the economic shock of the 1980s. The decline since the mid 1990s would then be seen as the long-wave social recovery from this shock.
So, what do the BCS/CSEW data show? This first graph shows indexed trends for some property offences: burglary, theft from the person, vehicle theft, robbery and bicycle theft, with 1981 as 100.
The orange line with triangle data points shows the all BCS/CSEW trend. As can be seen, 'overall' BCS/CSEW victimisation episodes rose from 1981 through to 1995, before decreasing to their starting point. Burglary (in red) and vehicle crime (in green) followed a similar trajectory, though rising more quickly and declining more sharply. Theft from the person rose to 1995 before stabilising at this higher level. Robbery followed a similar, if more exaggerated, pattern. Bicycle theft had its own pattern. It rose dramatically to 1995. It then declined before rising again from the early 2000s.
To summarise what this graph shows, it reflects the interaction of at least two factors. First there are the underlying political economic factors related to economic recession and growth. The start of the trend line coincided with the early 1980s recession, which had a major effect up and down the country right into the 1990s. During that period property crime rose as more resorted to various forms of theft to make ends meet. The improving economic position from the mid 1990s on offers some explanation for the generally declining levels of these property crimes.
The second factor is more contingent, related to improved security on the one hand and increased availability on the other. Some of the dramatic declines in burglary and vehicle theft are likely related to improvements in home and vehicle security. This might help to explain why we have not witnessed an increase in these offences in the current depression.
People and bicycles are less easy to secure than cars and homes of course. Moreover, people carrying expensive electronic gadgets make attractive targets. The growth in bicycle use means there are more to steal. These all help to explain why robbery, theft from the person and bicycle theft either stabilised or rose while burglary and vehicle crime were falling. Why go through the risk and hassle of breaking into a house when you can steal a smartphone or a laptop or pinch a bike with a cheap pair of bolt cutters?
As an aside, the ongoing and widespread anxiety about street crime and 'mugging' would, on the basis of this graph, be quite rational. Relatively speaking, theft and robbery is more common than 20 years ago. The fear of street crime trumps the decline in 'overall crime' precisely because the 'overall' trend masks what is happening. We don't need elaborate theories about public fear being out of kilter with 'real' crime trends.
What of the other main victimisation category covered by the BCS/CSEW dataset: violence? The graph below gives indexed trends for three categories of interpersonal violence: domestic, acquaintance and stranger violence.
This very striking graph tells us something very important I think. First there is the all BCS/CSEW trend, this time in turquoise. As before, it rises from 1981 to 1995 before declining back to its starting point. The green 'stranger violence' stays stable throughout, failing to rise at all. If the early 1980s recession had an impact on interpersonal violence, it wasn't reflected in stranger violence.
Acquaintances and intimate partners, on the other hand, were a different matter. Domestic violence grew particularly sharply during this period, staying at a relatively higher level for longer. The growth in acquaintance violence was also notable, though it declined at an earlier point.
It is widely recognised that most people, and in particular women and children, are far more at risk of harm from those they know and love than from strangers. This graph indicates that recessions might significantly intensify this known risk.
Recessions and depressions are harmful in all sorts of ways. People lose their jobs; families struggle to get by; individual self-esteem takes a battering; physical and mental health problems increase. Some people kill themselves as a result. Suicide is again on the rise in the UK. Some take out their anger on others: their partners, children and families notably.
Against the backdrop of the current depression the BCS/CSEW dataset offers a sense of what happened in the last major recession and what may be round the corner. It shows how misguided smugness about the current fall in 'overall crime' might be.
There is no inevitability in the rise in homicide, domestic and acquaintance violence in the coming year. Sadly, though, it would be more surprising if they did not increase than if they did.
The methodology of the BCS and CSEW has changed over the years. Scotland was covered in the BCS until the late 1980s, but was dropped thereafter. In the early 2000s the BCS moved from an annual survey of around 5,000 people to a rolling survey of around 50,000. There are now plans to scale it back to around 30,000 people. In recent years the experiences of under 16s has started to be included, though they weren't before. The name was also changed from BCS to CSEW.
The BCS/CSEW surveys also only cover a small number of victimisation experiences: mainly burglary, vehicle and other kinds of thefts and property crimes, and certain violence offences. Homicide is not included, nor are sexual, though data on the latter are published in a separate report.
In summary, there are all sorts of data irregularities and omissions with the BCS/CSEW dataset. It is not the 'gold standard' measure of crime that some claim. That said, it does have at least one distinct advantage over police recorded crime data. It is based on interviews with members of the general public about their victimisation experience. In that sense it gives a better guide to the certain crime victimisations than police data, which are reliant on the public reporting offences and/or the police recording them.