Under the rational choice model of crime, police reduces crime by increasing the expected probabilities of arrest and punishment.
But having a larger police force is not the only way to achieve this. Surveillance technology, such as street surveillance cameras (CCTVs) and dashboard cameras, can also deter crime by increasing the probabilities of arrest and punishment, at a fraction of costs of hiring more officers. Can cameras be a more cost-effective alternative in reducing crime?
In light of a rapid growth in surveillance cameras in recent years, many researchers investigated whether the use of surveillance cameras reduces crime. However, there is no clear consensus yet. Criminologists Brandon Welsh and David Farrington reviewed 44 existing studies on the effect of surveillance cameras on crime and found that that the evidence is mixed: 15 found that cameras significantly reduce crime, 3 found that cameras significantly increase crime, and 23 found no significant effect of cameras on crime.
However, they note a few interesting patterns from existing findings.
First, most studies that find that surveillance cameras reduce crime (14 out of 15) are based on the U.K. data. Incidentally, the U.K. is one of the heaviest users of the surveillance cameras in the world (with an estimated number of cameras between 4 million and 6 million), and it may be that the number of surveillance cameras must be sufficiently high in order to achieve their full crime-reducing potential.
Secondly, existing studies show that the effect of surveillance cameras widely varies across different empirical settings. Specifically, research consistently shows that introducing surveillance cameras in (previously unattended) car parks leads to a large and significant crime reduction, while its effect on city and town centers, public transport, and public housing is more muted.
While these studies are certainly informative, there are several conceptual problems that limit our understanding of the causal effect of surveillance cameras on crime. One problem is that most policy interventions that bring new surveillance cameras to car parks, city centers, and public housing also bring other changes. For example, when a city government sets up new surveillance cameras in a large car park, this change is often accompanied by fencing, additional security staff, and improved street lighting. As a result, it becomes difficult to separate the effect of surveillance cameras from the effects of these other crime-prevention measures.
Perhaps a more fundamental challenge is that there are just so many different ways to set up and use security cameras. For example, suppose a researcher runs a regression analysis on how the number of surveillance cameras is related with crime rates, and concludes that increasing the number of cameras by 20 percent leads to a 5 percent reduction in crime. But how should these additional cameras be set up? Should they be close to each other so that the camera coverage would be complete, or should they be set up more sparsely to cover a larger area? Should they be set in places with high crimes or high traffic? A simple regression result cannot answer such questions.
One possible remedy to this problem is to use alternative measures of surveillance camera presence (other than the mere number of cameras) in regression analyses. For example, several studies use the coverage rate of the surveillance cameras (that is, how much of the target area is covered by surveillance cameras) in their empirical analyses. This is probably a better measure to assess the relationship between surveillance cameras and crime, but computing the rate of surveillance camera coverage is a lot more difficult than counting the number of cameras. (To compute the coverage rate, one would have to know exact views of each surveillance camera.)
Furthermore, many modern surveillance camera systems are equipped with advanced features, such as turning automatically toward the sound of gunshots and the ability to recognize people’s faces and other objects, and the effect of these newer surveillance cameras are likely to be very different from the effect of older analog cameras. How can a simple regression analysis pick up this difference?
Can Surveillance Cameras be a More Cost-effective Alternative to Hiring More Police?
Theoretically, the presence of more surveillance cameras on the streets should increase the probabilities of detection and arrest, and deter some potential criminals from offending. However, the effectiveness of surveillance cameras will be greatly compromised if there is no matching increase in the number of police officers and security managers who analyze the images and actually go out and catch criminals.
Thus, it is probably more appropriate to view surveillance cameras and police as complements instead of substitutes, and look for an optimal combination of cameras and officers which achieves the maximum crime-reduction given the available budget. Investments on other areas of crime prevention are also likely to be relevant. How effective would a surveillance camera be at night without proper lighting?