How crime worsens sprawl

And by creating sprawl, creates big costs for society as a whole

In a recent post I looked into how bad crime is. I think the evidence says that crime is really bad – so bad that governments should be willing to spend several million dollars to avert one extra murder. This means that crime might be one of the USA’s biggest problems, with its direct effects reducing social welfare by something like a quarter of GDP – with the burden heavily concentrated on already disadvantaged groups.

A key failing of this post is that I only looked at the more direct costs of crime: the harm, sadness, pain and fear involved in actual crimes (and attempted crimes) that occur. I didn’t look at the indirect costs of crime being high.

For example, in many cities across the West women tend to avoid walking around alone after dark. If we only count the costs of crimes that are actually committed, then we don’t capture the huge cost to women’s freedom that this represents. I intend to look into quantifying this cost in a later post.

Some of the other behaviours people use to deal with crime also have indirect costs – external spillover costs on other people and society at large.

For example, one study by Julie Cullen and Steven Levitt finds that when crime rates across the city rise ten percent, city centre populations fall one percent – with people generally moving to the suburbs. One crime tends to push one person out of the city centre, on average. 

Quantifying this in terms of a real world city, the roughly 400 percent increase in New York City’s murders from 1955 to 1975 (from around 300 to over 1,500 per year) would have been expected to empty the densest parts of the city out by about 40 percent, assuming that other crimes rose in line with murder. And indeed, the population of the centre city – Manhattan – fell about 35 percent over that period, while the population and physical extent of the suburbs grew rapidly.

Murders in New York City peaked in 1990 at over 2,000 per year, roughly as population reached its nadir in the city centre. They have cratered by over three quarters, to about 300. This would have likely driven city centre population up massively, much moreso than it actually did recover, but building restrictions have prevented this happening anywhere near as much as it might, meaning that it has driven up prices instead.

So this story implies that crime in city cores drives people to the suburbs, creating urban sprawl. If so, then crime’s costs are even higher than we thought.

In NYC, let’s say an individual murder represents a 0.3% increase in crime. This is obviously not quite right – it worked in the previous comparison because murder correlates well with other crimes – but let’s use it as a starting point. This means an individual murder would shift 0.03% of the city centre population to the suburbs. Again, let’s be extremely approximate by calling Manhattan the city centre population (this is obviously a massive undercount because clearly elements of Brooklyn, Queens, and the Bronx, and possibly parts of Hoboken, Jersey City, and Newark should count as ‘centre city’). Two million (the current population of Manhattan) by 0.03% is 600 people.

What’s the marginal external cost of 600 people moving from Manhattan to the New York City suburbs? These 600 people will do more driving – meaning more noise, carbon, and particulate pollution for others, and more city centre space used up for roads and parking, and if insufficient can be provided, more traffic congestion. They will use more energy heating and cooling their homes.

How much does this list of externality costs add up to for each of these individuals? Calculating some of them can be pretty difficult – if any readers know of a straightforward methodology then do tell me. I will focus on just one, greenhouse gas emissions, where I can pretty easily compute a reasonable estimate.

New York suburbanites tend to produce about 50 extra tonnes of carbon per year than city dwellers. Six hundred people, for let’s say thirty years, times 50 extra tonnes per year, times a $50 social cost of carbon would be a $45m lifetime cost of one murder in climate change externalities due to sprawl.

This is just a first pass estimate. To improve it we might look back to the estimates in my last post and think about how much of the cost of crime comes from murder, expand our definition of the centre city, and add in estimates for the other externality costs of sprawl like noise, air pollution, and traffic congestion.

But even this first pass estimate suggests that my original estimate of the costs of crime – making them one of America’s largest problems in total welfare terms, and a substantial problem across the West – was an underestimate. If crime is a major contributor to sprawl, and if sprawl is, as many credible estimates suggests, the largest contributor to carbon emissions, which in turn are causing another of America and the world’s largest problems – climate change – then it may be several times more serious than I suggested.

P.S. We have just released a new issue of Works in Progress that I am very proud of. See in particular the highly related article ‘Reclaiming the Roads’ by Carlton Reid.