I don't blog about the death penalty very much, mostly because I don't care about the death penalty very much. I'm not saying it's unimportant, but I just don't have the urge to write about it. My position on it has been pretty bland: Some crimes are so evil that the perpetrators have forfeitted their right to live, but my libertarian leanings prevent me from trusting the government to execute the right people for the right reasons.

The extremists on both sides bother me. There's a difference between believing the death penalty produces a net benefit to society, and the lusty desire for executions that seems to animate some of its supporters. Conversely, no matter how much you disagree with the death penalty, it's wrong-headed to hold a candlelight vigil for John Wayne Gacy.

Over at a public defender, Gideon has come out of his blogging hiatus, and he's posted about the Death Penalty Information Center's new report on the cost of the death penalty. It's a mixed bag, but I think its economic theory has solidified my opinion against the death penalty.

Parts of the report don't pass the smell test, starting with the subtitle---"Reconsidering the Death Penalty in a Time of Economic Crisis"---which comes across as a desperate attempt to make the death penalty relevant to the hot topic of the moment.

Another dubious feature of the report is a survey of police chiefs that seems to reach some surprising conclusions:

• When asked to name one area as "most important for reducing violent crime," greater use of the death penalty ranked last among the police chiefs, with only 1% listing it as the best way to reduce violence...
• The death penalty was considered the least efficient use of taxpayers' money...

These results aren't as impressive as they may seem at first glance. It's too easy to game the results through a careful choice of alternative items.

For example, if I wanted to argue for eliminating the exclusionary rule for evidence in a criminal trial, it would be great for my side if criminal defense attorneys said it was unimportant. No problem. Just ask them a question like this:

When it comes to giving defendants a fair trial, which of the following features of our justice system is most important:

• The presumption of innocence.
• The right to trial by an impartial jury.
• The protection against double jeopardy.
• The right to confront witnesses.
• The exclusionary rule. 

I think there's a pretty good chance that if you asked a thousand criminal defense attorneys that question, the exclusionary rule would be at the bottom of their rankings. Then I could put out a report boldly claiming that "defense attorneys rate the exclusionary rule as the least important requirement for a fair trial."

(If they didn't rate it the way I wanted, I could just change the questions and try again.)

On the other hand, when you ignore the hyperbole and fake(ish) survey, the report makes an interesting economic case, starting with the estimated cost of a death sentence:

The high costs to the state per execution reflect the following reality: For a single death penalty trial, the state may pay $1 million more than for a non-death penalty trial. But only one in every three capital trials may result in a death sentence, so the true cost of that death sentence is $3 million. Further down the road, only one in ten of the death sentences handed down may result in an execution. Hence, the cost to the state to reach that one execution is $30 million.

That's a lot to pay to execute somebody, and it may be worse than that:

In 2008, the California Commission on the Fair Administration of Justice released an exhaustive report on the state's capital punishment system... The report found that the state was spending $137 million per year on the death penalty... Since the number of executions in California has averaged less than one every two years since the death penalty was reinstated in 1977, the cost for each execution is over $250 million.

So is an execution worth $30 million? Or $250 million?

That depends, of course, on what the benefits of an execution are. Death penalty blawger Tom McKenna has referred to evidence that each execution has a deterrent effect that results in 18 fewer murders for every execution. (I think he's talking about Is Capital Punishment Morally Required? Acts, Omissions, and Life-Life Tradeoffs by Cass R. Sunstein and Adrian Vermeule, which gets the figure of 18 from Does Capital Punishment Have a Deterrent Effect? New Evidence from Postmoratorium Panel Data by Hashem Dezhbakhsh, Paul H. Rubin and Joanna M. Shepherd.) This figure is highly controversial, but for the sake of discussion, let's accept it.

That leads us to the next question in our quest to determine the value of an execution: How much is a human life worth? It's easy to say that human life is priceless, but as economists have shown, that's not how we behave.

Economists have sifted through large data sets to uncover how people make decisions that affect their own mortality. How much will people spend on a safer car to reduce their chance of dying in an accident by 1% over the life of the car? How much more do you have to pay people to take dangerous jobs? How much more do houses cost in safe neighborhoods?

These studies typically show that people value their own lives at between three and ten million dollars. I've heard a number of economists use a value of $8 million per life, so that's what I'm going to use.

Finally, we can plug in all the numbers. According to the paper, executing someone could cost around $30 million dollars. Add to this the value of the executed criminal's life (he counts too) and an execution has a cost to society of $38 million. On the other hand, the 18 lives saved are worth $144 million. Therefore, under our assumptions, the death penalty provides a net benefit to society of $106 million per execution.

That's great for the death penalty, but only if it's true. For example, the benefit goes away completely if we use the alternate cost estimate obtained in California. In that case, the $144 million value of the averted murder victims pales in comparison to the $258 million cost of the execution.

There's also the problem that these studies are difficult to perform and not terribly robust. The Death Penalty Information Center's report refers to a summary paper (Uses and Abuses of Empirical Evidence in the Death Penalty Debate by John J Donohue and Justin Wolfers) which argues for an interpretation of the data that finds no significant deterrent effect. For one thing, all of the benefits seem to be concentrated in Texas. Drop that one state from the data, and executions have no significant effect on the murder rate. Or keep Texas and simply weight the data by population, and again the benefits vanish.

(If this data is meaningful, it tends to confirm some criminologists' opinions that that the deterrent effect does not emerge until you start to execute a lot of people. Killing one or two criminals a year just won't get people's attention. Not enough to rise above the noise.)

Basic economic theory says that people respond to incentives, and that's pretty much what we find whenever we study human behavior, so I'm confident that executions must have some deterrent effect. But if we believe Donohue and Wolfers, then it seems unlikely the effect is very large. I've heard of other studies which conclude that each execution only deters 3 murders. That's $24 million worth of human lives saved at a cost of $38 million---almost $13 million each---which is a poor bargain. And the true deterrent factor could be much smaller.

Understand that I'm not saying we shouldn't spend tens or hundreds of millions of dollars to save lives. What I'm saying is that resources are scarce (economic crisis or not), so if we decide to spend money saving lives, we should be careful to spend it in a way that will save the most lives possible, which probably isn't on executions.

In California, for example, the death penalty costs $137 million per year, but with one execution every other year, even using the highest estimates of the deterrence effect, they are preventing only 9 deaths per year. I'm having trouble finding figures, but that sounds like enough to hire hundreds of additional police officers, firefighters, and EMTs, or to cover the unreimbursed operating costs of maybe a dozen Level I trama centers (it depends on the patient mix), or to fund better infection control protocols in the state's hospitals.

There are dozens of things that might change the benefit-cost calculation for executions, and I can't possibly consider them all here, but if saving lives is the reason for executing criminals, I think there are better ways to spend our money.