MORE EXECUTIONS - Fewer deaths : LUSENET : Current News - Homefront Preparations : One Thread

Hudson Institute

More Executions, Fewer Deaths?

A new study suggests that the death penalty deters many more murders than most people thought plausible. A death-penalty opponent analyzes the evidence.

by Iain Murray

Europe assails us as barbaric for embracing it. Churchmen worry for our immortal souls because we think that it is just. Governors lose sleep over the issue. It is the death penalty, and the debate over its imposition is now more intense than at any time since its brief suspension as unconstitutional in the 1970s. The federal executions of terrorist Timothy McVeigh and drug kingpin Juan Raul Garza, the first uses of the punishment by federal authorities since 1963, have intensified the arguments on both sides. European governments used the issue as a stick with which to beat President Bush during his first visit to the Continent, and in Ohio, a group of religious-conservative lawmakers have stood up to oppose the penalty based on their religious faith.

In this atmosphere, death penalty proponents have found their arguments tested as never before. The contention that it is a just punishment is countered by the possibility that innocents have been executed. Although there is no proof that such a calamity has occurred since the restoration of the death penalty in 1976, its mere potential has been enough for some state governors to impose moratoria on executions. The argument that the penalty at least incapacitates the murderer himself and prevents him from murdering again has been attacked by life-imprisonment advocates as an overreaction. Murderers are the least likely of all criminals to repeat their crime, but it does occur. One notable recent case occurred in June 1999, when Leroy Schmitz, who served eleven years in a Massachusetts prison for strangling his girlfriend, murdered his wife in similar fashion in Montana. But for the most part, murderers who kill again have not been found guilty of capital murder and have never faced the death sentence.

The deterrent effect of execution, the argument that might serve death penalty proponents best, has had the worst time of all. A USA Today poll carried out in early June, around the time of McVeigh's execution, found that 66 percent of respondents did not think that his death would serve as "a deterrent to future acts of violence and murder." Major figures in the debate, such as former governor of New York Mario Cuomo, point to cases such as that of Andrea Yates, who admitted slaying her five children, as evidence that many murders are irrational acts and therefore cannot be deterred. Others have argued that there are too few executions to have any deterrent effect. The Chicago Tribune editorialized on June 10th, "If we wanted the death penalty to deter future criminals, we would impose it the way China does: far more often, and with far less evidence. Or we would apply it to all kinds of offenses the way we did in low-crime colonial times, when Virginians could be executed for stealing grapes or killing chickens."

New Data

Until now, believers in the deterrence effect of executions have had little hard evidence with which to counter such straw man debating techniques. The work of economist Isaac Ehrlich of the State University of New York in the 1970s, which found a significant deterrent effect, had been diluted by constant reinvestigation and criticism. In the end, it suffered most from being out of date, as it was based on evidence from before the suspension of the death penalty in 1972. Its relevance to the modern debate was therefore questionable.

But there now come impressive new findings from a trio of economists at Emory University in Georgia. Hashem Dezhbakhsh, Paul Rubin, and Joanna Mehlhop Shepherd released their paper "Does Capital Punishment Have a Deterrent Effect? New Evidence from Post-Moratorium Panel Data" in January 2001 (a working paper online at the National Bureau of Economic Research). Its findings are striking. The authors conclude that each execution deters other murders to the extent of saving between eight and twenty-eight innocent lives, with a best-estimate average of eighteen lives saved per execution.

The researchers reached this conclusion scientifically, by expressing the murder rate mathematically. They calculated the effect on the murder rate of a number of factors including, specifically, the likelihood of being arrested, the chance of being sentenced to death after arrest, and the chance of being executed after sentence. They were then able to work out how significant the chance of being executed is to the murder rate. They found that executions themselves are a very significant factor, certainly much more so than the simple removal of the murderer from the pool of potential killers. And their findings pass all the statistical tests that show that it's not just by chance that the math works that way.

This "econometric" method of looking at crime and other social phenomena is gaining popularity among researchers and has led to some controversial claims. Foremost among the econometricians is John Lott Jr., a research scholar at Yale, whose pioneering work on the effects on crime of issuing permits to carry concealed handguns has provoked a storm of controversy. Perhaps equally famous is the claim by the University of Chicago's Stephen Levitt and Stanford's John Donohue that the legalization of abortion in the early 1970s led to the drop in crime some twenty years later. Criminologist James Alan Fox of Northeastern University characterized Levitt and Donohue's work as "voodoo economics," while Lott's work has come under repeated fire for failing to control for numerous outside factors or for being circular in some way. Interestingly, Lott has answered all of his critics by continually refining his model, and has always obtained the same basic results. Nevertheless, his work and methodology continue to infuriate those who think they know better.

With these precedents in mind, it is important to look critically at the new work from Emory. The most obvious objection to the research is that it might fail to capture all the outside factors that feed the murder rate besides the criminal's rational assessment of his chances of getting caught. Factors such as drug trafficking, gun availability, and the overall supply of potentially violent young males are all recognized as important contributors to murder rates.

The researchers attempted to include these factors by constructing another element in their equation, taking account of crime rates for assault and robbery (which sometimes lead to murder), income levels, welfare levels, population density, six demographic categories for race and gender, and the state-level membership rate for the National Rifle Association (NRA), to serve as a proxy for gun ownership rates. They also took into account national-level trends such as the increasing amount of violence in America's popular culture. Finally, they added a variable to account for completely random factors. They then measured the results of their equations at both state and county levels, to give them as detailed a picture as possible. No earlier research, it is important to note, had ever gone into this level of detail.

Some Puzzles

Most of the results they obtained were as expected. The murder rate increased as assault and robbery increased, and it also varied with the number of males in a county and with the proportion of African-Americans. It decreased according to the size of the non-African-American minority population. It also decreased with higher population density, which may at first sight seem strange, but it should be borne in mind that rural areas, with low population density, often have higher murder rates than the peaceful suburbs.

Three results do, however, appear rather odd. The first is that the murder rate appears to increase with per capita income. The researchers explain this by suggesting that drug consumption, which is heavily linked to murder, may increase along with income. This is, however, speculation. The researchers did not include a true measure of drug consumption or trafficking in their equation, which is probably the biggest single mark against it. If, however, the researchers' assumption is true, it appears that the metropolitan elite's habit of purchasing drugs as a recreational luxury is contributing to the murder rate. This is not an argument that is often made in considering how the war against drugs should be fought, but it does provide food for thought.

Another seemingly odd result is that a higher percentage of the population being teenage seems to lower the murder rate. This is again surprising, as the teen murder rate is the one that showed the biggest increase during the 1990s. It may be, however, that higher teenage populations involve proportionally more teenage girls, who are much less likely to murder than boys. Furthermore, the prime age for murder remains the immediate post-teen years, and it may be the size of that category that is most important. Unfortunately, the demographic category the researchers used was of ages twenty through twenty-nine, which includes large numbers who are putting their risky pasts behind them.

The final odd result was that the size of the state's NRA membership seems to increase the murder rate. This cannot be an effect of the murder rate rather than a cause, because the researchers took time lags into account in their models. NRA membership may, however, be a good indicator of the violence potential in a state, as states that have been previously more inclined to violence during earlier crime cycles retain high membership rates. States that have no real history of murder, however, may have fewer NRA members, as there is less need to join for self-protective purposes.

Deadly Implications

Despite all these reasonable explanations for the few odd results, the latter create enough doubt to cause one to worry about the robustness, if not the direction, of the authors' overall conclusion about deterrence. As mentioned earlier, one of the strengths of John Lott's work on guns is that the continued attacks by opponents of his view have resulted in continuous reassessment that has confirmed and thereby strengthened his original conclusions. His original paper was reworked into a book, More Guns, Less Crime, in 1998, which was updated again with a second edition, in 2000. Perhaps the best thing to happen to this research on capital punishment would be for the opponents of the death penalty to attack the assumptions and modeling techniques on which the findings rely. The researchers would then be forced to refine their model to rebut the attacks, by, for instance, including a measure of drug trafficking offenses or the like in the equation. If the conclusions still held true after such refinements, then the argument would be bolstered further.

In a way, this research is already a refinement and updating of Ehrlich's earlier work. It seeks, for instance, to answer the Cuomo argument that not all murders are rational, by estimating the number of murders that are unpreventable by deterrence and controlling for that factor. Similarly, the use of data collected since the restoration of the death penalty makes the research reasonably up-to-date. The use of county-level data avoids the problem that arises from using only national-level data, that of being unable to assess the true effects of individual states' policies. By any measure, this study is already a hugely important contribution to the debate.

Which makes one wonder why it has not received the publicity it should have. A search of the Lexis-Nexis news database reveals that the story has only been covered in the Washington insiders' magazine The National Journal and in a brief news item on Fox News' Special Report with Brit Hume. Without wishing to speculate, one might surmise that death penalty opponents saw how quickly John Lott's work entered the national consciousness once gun control proponents attacked it, and hoped to avoid such a boomerang effect. (A similar process occurred around the Levitt-Donohue study when anti-abortion activists went after it.) Moreover, in contrast to the cases of gun control and abortion, which have strong advocacy groups such as the NRA and the various pro-choice groups arguing over them, there is no big pro-death penalty organization eager to spread the word about this new research.

But the implications are huge. By the study's estimate, the two recent federal executions will save approximately thirty-six lives. On the final day of 1999 (the last day for which we have accurate figures), there were 3,527 prisoners under sentence of death in American prisons. This study suggests that if all those sentences were carried out 63,000 lives would be saved. There were approximately 15,000 homicides in America in 1999, meaning that the deterrence effect could be the equivalent of four years free from murder. Even the most committed opponents of the death penalty should take notice of that figure.

Iain Murray is a senior research analyst specializing in crime and justice issues at the Statistical Assessment Service—a Washington, D.C.-based nonprofit, nonpartisan public policy organization dedicated to improving public understanding of scientific and statistical information. He is personally opposed to the death penalty for religious reasons.

To respond to this article, please send an email to

-- Anonymous, October 21, 2001

Moderation questions? read the FAQ