The Cost of Murder

A team of researchers at Iowa State University has just published a paper that estimates the cost to society of several violent crimes. Led by Matt DeLisi, an associate professor of sociology and director of the criminal justice program, the team came up with the following numbers:

Crime	Social Cost 2008 US$
Murder	17,252,656
Rape	448,532
Armed robbery	335,733
Aggravated assault	145,379
Burglary	41,288

Scott Greenfield at Simple Justice stumbled onto this study through a New York Times op-ed by Charles Blow, and he isn’t buying any of it:

it’s studies like this, and op-eds that assume some level of credibility, that make sound policy sound absurd.  Want to kill a good idea?  Wrap it up in crap.
…

I’m not calling the Iowans liars or morons.  I’ve no doubt that in some parallel academic universe, applying logic that defies any semblance of reality and only by suspension of reality can link together theoretical costs and values that would never otherwise exist, and then only if you hop up and down on your left foot while turning your head to the left and squinting.  No, I’m sure they have a basis for their numbers.

He further elaborates in a comment:

Do you credit that every murder costs $17.25 million?  If not, then the argument relying on this incredible number is rendered similarly incredible.

Sigh. I can’t really defend this paper, because I don’t nearly enough about it or the economics behind it, but I have a vague idea of what the paper is trying to do, and I think it’s a good idea. So consider what follows a defense of this kind of research, if not this research in particular.

I think there are several things going on here, and it’s not as crazy as Scott thinks it is. That’s because economists use an expansive and all-inclusive definition of cost.

Over the last couple of decades, academic economists have been engaged in what I’ve heard called “economic imperialism”–trying to apply the techniques of economics to fields not traditionally thought of as economic in nature.  Some of this is quite experimental, and the results should be taken with a huge grain of salt. Often, the real purpose of the study is not so much to learn about it’s subject, but to test the methods of the study itself to see if they return results that are robust and meaningful.

I don’t know enough about economics to judge the quality of the Iowa State study. Are the studies it’s based on considered reliable? Are they cited often in other studies? How are academic economists reacting to this study? Perhaps they’re excited. Or perhaps they’re all going “Bah, it’s just DeLisi again!” I haven’t got a clue.

That said, an average cost to society of $17.25 million per murder doesn’t strike me as outrageously wrong, at least not by the standards of these kinds of studies, meaning I doubt it’s more than twice the correct figure.

Keep in mind that this isn’t an estimate of just the out-of-pocket costs, it also includes all the intangible costs. Acacemic economists insist that just because intangible costs are very difficult to measure doesn’t make them any less real. We may have our doubts about the value of the intangible costs, but that doesn’t mean we can arbitrarily assume them away. Intellectual honesty compells us to make an estimate, or at least admit to our assumptions.

So, right from the start, when someone is murdered, society loses the value of that person’s life. That seems like an impossible thing to put a number to, but as I discussed in a previous post about the death penalty, economists have ways of figuring these things out. I can’t follow the details of their calculations, but their methodology makes sense to me. Obviously, the number they come up with is going to be a very rough, but I’ve read that $8 million is a common estimate. This accounts for almost half of DeLisi’s $17.25 million figure.

Most of the rest of the figure is due to what the researchers are calling “willingness to pay.” This is an estimate of the cost our society is willing to pay to prevent murders. As such, it doesn’t quite represent the cost of a murder. Rather, it represents the cost of living in a society that has murderers among us.

That cost is rather high, because it includes literally everything we do to avoid murder. To start with, there’s the cost of our police departments and our justice system. I don’t mean the direct cost of catching murderers, but the entire cost of police patrols, the 911 system, and the having detectives standing by in case a murder occurs. Preventing murder isn’t the only thing the police do, but some fraction of their budget has to be apportioned to murder prevention.

We also have personal costs of preventing murder, such as security hardware and alarms for our businesses and homes. If you pay more for a house in a low-crime neighborhood, some portion of the annualized cost of the house is attributable to avoiding being a murder victim. If you quit your job as a liquor store clerk because you’re afraid you’ll get shot in a holdup, that’s part of the cost too. Every door lock, every cell phone, every personal firearm, all of these things–they’re all part of the cost of living with murderers.

Finally, we pay a cost in opportunities not taken. Every time you compromise the quality of your life in order to reduce your risk of death at the hands of another, you are paying the cost of avoiding murder. Every man who foregoes a nighttime jog through the park, every woman who decides not to leave the bar with an interesting stranger, every child who turns down a friendly driver’s offer of a ride–they’re all paying the murder cost.

These things add up, and they add up to a hell of a lot. All of which brings me back to Charles Blow’s op-ed, which contains this interesting paragraph:

By their estimates, more than 18,000 homicides that the Centers for Disease Control and Prevention recorded in 2007 alone will cost us roughly $300 billion. That’s about as much as we’ve spent over nine years fighting the war in Afghanistan. That’s more than the 2010 federal budget for the Departments of Education, Justice, Housing and Urban Development, Health and Human Services, Labor and Homeland Security combined. Does anyone else see a problem here?

It’s a huge figure, but you have to put it in some perspective. It amounts to about $1000 per year for every man, woman, and child. That’s a lot of money, but it’s less than we spend on a lot of other things such as food, transportation, and entertainment. Personally, my wife and I spend more than that on telecommunications.

Here’s one quick factoid I’ve been wanting to work into a blog post for months: Economists have used this same approach to benefit-cost analysis to estimate the cost in the U.S. of traffic congestion, including motor fuel, the value of driver and passenger time, pollution, and the cost of avoiding congestion by taking different routes or driving at different times or simply not making trips. It works out to about $50 billion per year. In other words, the social cost of murder is about six times the social cost of traffic jams. It doesn’t sound so bad when you put it that way, does it?

(Obviously, murder is much worse than a traffic jam, but traffic jams are a far more common problem, and the aggregate misery should not be ignored. Think of it this way: Suppose someone invented a technology that allowed you to completely avoid traffic jams but slightly increased your chance of a fatal accident. How much more dangerous would it have to be for you to not take advantage of it?)

Here’s another way of looking at the cost of murder: The United States gross domestic product for 2009 was $14.119 trillion, so the $300 billion cost of murder was about 2.1 percent of GDP. The Iowa State paper cites studies that suggest that the total social cost of all crime is a little over three times higher the cost of murder, or roughly $1 trillion. That means the cost of crime amounts to about a 7 percent drag on our economy. This doesn’t strike me as an outrageous number.

I should mention that parts of the Iowa study strike me a bit odd. For one thing, the study assigns the victim’s life a value of less than $5 million, which is lower than the $8 million estimate I’ve seen elsewhere. I think this is because the $8 million economists’ estimate includes a willingness-to-pay component which the study’s authors have lumped into their own willingness-to-pay figure, but I’m not smart enough to be sure.

It’s also a little odd that the numbers are presented with absurd precision: $17,252,656 for a murder. If this is like most of these kinds of estimates, I’d be surprised if it was considered accurate to even two significant figures. In fact, many of the numbers in the paper are presented without any characterization of the amount of error. I assume this is because the numbers are pulled from other studies, and the error is described there, but then shouldn’t the numbers be rounded off for presentation?

More confusing to me, however, is that one of the components the authors include in the cost of murder is the lost productivity of the murderer while he is in prison. I don’t understand the reason for this. The lost productivity of the murderer is a cost that is paid solely by the murderer himself. If we’re going to include the murderer’s costs, shouldn’t we also include all his intangible costs as well? After all, a prisoner is essentially undergoing the same ordeal as a kidnap victim. I’m guessing that his willingness-to-pay to avoid spending years in prison is pretty high.

And if you’re going to consider the costs paid by the murderer, shouldn’t you also consider the benefits? After all, murderers kill for a reason. They kill for the same reason anybody does anything: To improve the quality of their lives. The victim of a murder is worse off, but the murderer himself is better off for having gotten rid of a cheating spouse or a rival drug dealer. He may not be better off in prison, but prison is a risk, not a certain cost. 

That’s probably not a path we want to go down. It would be better to eliminate the murderer’s benefit-cost calculation by making the technical assumption that the murder’s decision to kill included an evaluation of the risk of incarceration. Thus, while we don’t know all the details, we can assume that the kill was a risk-adjusted good thing for the killer. Then we just leave all the murderer’s costs and benefits out of the final equation.

I suspect that the authors included the murder’s lost productivity because of they were following the reprehensible but common practice of counting offender productivity losses as a loss to society. I see this a lot in arguments about the war on drugs: Prohibitionists like to argue that drug use leads to lost productivity–as if our personal productivity somehow belonged to society or our employers instead of to us as individuals.

One final point, and then really, I will stop writing: The same expansive benefit-cost ratio that gives us the $17.25 million cost of murder can also be used to estimate the costs of overcriminalization. These are probably also surprisingly huge if you’re not used to economic benefit-cost analysis. Not only do you have to count the cost running the prisons, you also have to count the cost of imprisonment to the prisoners themselves.

In a previous post, I used the average salary of $11.40 per hour to calculate that it would cost $23,712 per year to hire a high-school dropout for 40 hours a week for a full year. I used this to set a floor on the cost of imprisonment. I think I could improve on this estimate by using the pay figures for jobs that require the worker to be on-site for 24 hours a day like a prison does. For example, unlicensed deckhands on oceangoing vessels make $25,738 – $34,376 a year. Of course, unlike prisoners, deck hands aren’t required to be onsite for the entire year. They also receive benefits, including pensions. And shipboard living conditions are much nicer than in prisons.

Perhaps a better way to look at the real cost of a single year in prison from the point of view of the prisoner is to ask how much you would be willing to pay to avoid prison for a year. $20,000? $50,000? $250,000? Add to this all the costs of abusive police officers and overzealous prosecutors–every hispanic man who is hassled by the cops for sitting on the stoop of his own home, every black man who avoids driving through certain neighborhoods because he doesn’t want to get pulled over, every businessman who has to hire lawyers to help him avoid breaking detailed regulations.

As a libertarian, I would also add the cost of the vast and pointless enforcement against all manner of consensual crimes, including drugs, gambling, pornography, prostitution, and all the related non-crimes such as money laundering. It’s all pure deadweight loss, with no significant benefits.

I don’t have any good estimates for these costs, but I’ll bet incarceration, even of the guilty, is hideously costly to society. Incarceration of the innocent is even worse because it brings no benefits, it allows the true offender to roam free to offend again, and it chips away at the value our institutions of justice. I like to mock our justice system, but on an international scale and a historic scale, our trustworthy institutions are a valuable asset. We should be careful to treat them that way.