A Punishing Debate
Does the death penalty deter homicide? New economic studies seek the answer to an age-old question.
Published June 1, 2002 | June 2002 issue
In 1764, a young Italian economist, Cesare Bonesana, the Marquis of Beccaria, wrote An Essay on Crimes and Punishments. The slim volume, almost unknown today, held ideas that were revolutionaryso radical, in fact, that Beccaria initially published An Essay without attaching his name to it.
With time, however, many of Beccaria's ideas about criminal justice would become widely accepted in Europe and the United States, incorporated whole cloth into penal codes and constitutions. That defendants should be given a speedy trial, heard by a jury of peers, judged on evidence rather than gossip, and without coerced testimony, were among the innovations proposed by Beccaria. But a central proposition of Beccaria's essaythat the death penalty be abolishedwas then, and remains still, a highly contentious idea.
At the heart of his argument was the concept of deterrence, the linchpin of economic theories of crime. Beccaria's theory of punishment explicitly assumed that people make rational economic choices about whether to violate laws based on their comparison of the costs and benefits of doing so. Punishments therefore should be designed so as to minimize the net benefit of committing crimes. "That a punishment may produce the effect required," he wrote, "it is sufficient that the evil it occasions should exceed the good expected from the crime, including in the calculation the certainty of the punishment, and the privation of the expected advantage. All severity beyond this is superfluous, and therefore tyrannical."
Execution is a less effective deterrent than life imprisonment, according to Beccaria, because it exacts a lower cost from a lawbreaker. "The death of a criminal is a terrible but momentary spectacle,"he wrote at a time when public hangings were a matter of course. "And therefore [it is] a less efficacious method of deterring others than the continued example of a man deprived of his liberty, condemned as a beast of burden."
Even a cold-blooded murderer will be deterred by the prospect of a life without freedom; the price is just too high. "There is no man who, upon the least reflection, would put in competition the total and perpetual loss of his liberty, with the greatest advantages he could possibly obtain in consequence of crime," Beccaria argued. And unending incarceration, he felt, was more costly than death itself. "Perpetual slavery, then, has in it all that is necessary to deter the most hardened and determined, as much as the punishment of death. Indeed, I say it has more. There are many who can look upon death with intrepidity and firmness, some through fanaticism and others through vanity, which attends us even to the grave."
Was Beccaria right? Is the death penalty comparatively powerless? Is life imprisonment more effective? These are questions that seem empirically testable, open to verification or refutation through careful analysis of data. By looking at statistics on capital crimes, death row executions and whatever other variables might be thought to factor into crime levels in a given society, answers could be found. And indeed, social scientists have been engaged for decades in efforts to measure the deterrent effect of capital punishment. But as revealed in the conflicting results of two studies released this past year by economists from Harvard, the University of Chicago and the University of Colorado, the questions remain stubbornly resistant to resolution.
Modern analysis of capital punishment
Though much of Beccaria's legal theory found widespread acceptance, the economic analysis of crime and punishment on which it rested remained in obscurity for centuries. Jeremy Bentham elaborated on Beccaria's economic ideas in the early 1800s, but most later discussions of crime, and of murder in particular, relied on psychological, sociological and even phrenological theories. By and large, these theories dismissed the idea that threat of execution might effectively dissuade a potential murderer: most murders were impulsive, not planned. And if other social scientists were dismissive of capital punishment, the term evoked nothing for economists but concern over capital gains taxation.
But Gary Becker's 1968 analysis, "Crime and Punishment: An Economic Approach," resurrected the lost idea that criminals might rationally weigh the costs and benefits of committing a crime. "Lest the reader be repelled by the apparent novelty of an 'economic' framework for illegal behavior," wrote Becker in the conclusion of his seminal article, "let him recall that two important contributors to criminology during the eighteenth and nineteenth centuries, Beccaria and Bentham, explicitly applied an economic calculus." (See interview with Gary Becker in this issue of The Region.)
While Becker provided the first modern economic framework for the supply and demand of crime, echoing Beccaria's thoughts in many passages, his work didn't examine Beccaria's claim about the ineffectiveness of the death penalty. It fell to Isaac Ehrlich, a colleague of Becker at the University of Chicago, to put the question to a careful empirical test.
Ehrlich's 1975 analysis, "The Deterrent Effect of Capital Punishment: A Question of Life and Death," published soon after the U.S. Supreme Court's 1972 ruling that prohibited capital punishment, used Becker's general model to derive a specific "murder supply function." Like any conventional production function, Ehrlich's equation said that the number of murders "supplied" would be a function of numerous cost variables, especially those related to the direct cost (to the criminal) of committing a murder (that is, the probabilities of arrest, conviction and execution).
Also included were variables to measure the resources allocated by society to deter crime (e.g., per capita expenditures on police). Still other variables covered economic variables that might measure the opportunity costs of committing a crime (e.g., unemployment rate, per capita income) as well as demographic variables (e.g. racial composition, age distribution). Using regression analysis on U.S. data from 1933 to 1967, Ehrlich was able to estimate the independent effect of each of these variables on the number of murders committed.
The results were stunning. "On the average," wrote Ehrlich, "the tradeoff between the execution of an offender and the lives of potential victims it might have saved was on the order of magnitude of 1 for 8." Eight lives saved by one deatha result Beccaria would never have imagined.
Ehrlich cautioned that the results didn't imply capital punishment is necessarily a desirable form of punishment. Levels of alternative punishments for murder may not be optimal, he noted, and murders "may also be reduced through increased employment and earning opportunities." But death penalty proponents were elated by the analysis. And Robert Bork, then solicitor general, brought Ehrlich's study to the attention of the Supreme Court in arguing that the death penalty should be reinstated. In 1976, the Court so ruled.
But Ehrlich's findings were quickly challenged. Several economists said Ehrlich's sample had too little variation in arrest, conviction and execution rates to truly measure their effect on murder rates. Many argued that Ehrlich had excluded important variables (e.g., length of sentence, gun ownership, deaths at the hands of police), which negated the deterrent effect of prisoner executions once they were incorporated into the model.
Economists Peter Passell and John Taylor, then at Columbia University, were two of the more prominent critics of Ehrlich's work. They pointed out that his results were highly sensitive to how different variables were defined and to the statistical form of equation used to relate variables. Without prior guidance from theory on which format or definitions to use, there was no way to judge which was most valid, and therefore no reason to believe in one set of results or another. "The results are sensitive to specification of the variables and transformation of the data," wrote Passell and Taylor. "This sensitivity raises grave (and in our own opinion, overwhelming) doubt about the utility of Ehrlich's time-series estimates." Even more crucially, perhaps, Passell and Taylor showed that Ehrlich's deterrent results disappeared if just the last seven years of the 34-year sample period were excluded from analysis.
Though not entirely devastating, the flood of critiques was enough to throw Ehrlich's seemingly unambiguous findings into significant doubt. The many studies that followed have clarified, corrected and elaborated the impact of various factors on murder rates, but none has provided a definitive answer to the question of the deterrent impact of capital punishment.
"By the end of the century," wrote Stuart Banner, professor of law at Washington University, in his 2002 book, The Death Penalty: An American History, "there was an abundant literature in journals of academic law and economics. A few studies found a deterrent effect, but most did not." Battles raged over methodology and there remained "a nagging suspicion" that researcher beliefs affected research outcomes. "This diversity in academic opinion translated poorly into the public policy arena, where proponents of each side tended to ascribe validity only to those studies which supported their own view."
A fresh look
Into this clouded arena have stepped two new studies. The first, by economists Lawrence Katz of Harvard University, Steven Levitt at the University of Chicago and Ellen Shustorovich of the Monitor Co., examines crime rates in relation to execution rates and prison conditions. The second, by H. Naci Mocan, chair of the economics department at the University of Colorado at Denver, and R. Kaj Gittings, a graduate student in the department, looks at homicide rates relative to executions and pardons of prisoners on death row.
In "Prison Conditions, Capital Punishment, and Deterrence," Katz, Levitt and Shustorovich start from the perspective that prison conditions are more likely than executions to have a deterrent effect on crime, including homicide. In the second half of the 20th century, executions were relatively rare and they occurred only after a long lag from the actual commission of the crime; the murders themselves were often perpetrated under the influence of alcohol or drugs. The impact on criminal decision-making of such a low probability, highly time-discounted outcome, suggest the authors, was likely to be minimal.
After all, they point out, the execution rate on death row (in 1997, 74 prisoners were executed, about 2 percent of the roughly 3,400 on death row) is only twice the death rate from accidents and violence for all American men, and only slightly greater than the rate of accidental and violent death for all black males between the ages of 15 and 34. "Based on these figures," they wrote, "it is hard to believe that in modern America the fear of execution would be a driving force in a rational criminal's calculus."
On the other hand, bad prison conditionswell known, pervasive and immediateare likely to play a significant deterrent role. "If I'm thinking about committing a crime," said Levitt in an interview, "and I know what happens in prison, and I know that there are lots of people with AIDS [there] that are going to rape me, you know, if there's anything that could deter me, that might be it."
The data bear out the hypotheses. Their study, forthcoming in the American Law and Economics Review, analyzed state-level data from 1950 to 1990, controlling for a variety of socioeconomic variables, and found that prison conditions (as measured by rates of prison death, mostly caused by illness) have a large and significant impact on crime. "In terms of crimes reduced per prison death, the estimated effects are quite large: 30-100 violent crimes and a similar number of property crimes." The number of homicides prevented is lower but still strong: 10 prison deaths are associated with one to eight fewer murders.
In contrast, execution rates have only a small, statistically insignificant effect on murder rates. "We find little systematic evidence supporting a deterrent effect of capital punishment," wrote the authors. "There simply does not appear to be enough information in the data on capital punishment to reliably estimate a deterrent effect."
Though the authors make no reference to Beccaria's theory of crime and punishment, their study seems an almost perfect confirmation of it. The prospect of being sentenced to years of horrendous prison life appears to be a strong deterrent to criminal activity. As Beccaria described the potential criminal's internal monologue: "If I commit such a crime, ... I shall be reduced to that miserable condition for the rest of my life." Execution, on the other hand, appears too unlikely, too abstract to be a significant factor in the crime supply function. "The fear of death," wrote Beccaria, "men always behold in distant obscurity."
A second look
The second paper by Naci Mocan and Kaj Gittings, starts from the same fundamental theory of deterrence. "Economists consider criminal activity as a reaction to costs and incentives," said Mocan. "So the application of this framework to capital punishment is just another test of whether the increased costs or decreased costs of punishment are significant enough to impact the reaction of individuals, and therefore the homicide rate."
In "Pardons, Executions and Homicide," Mocan and Gittings look at homicide rates between 1977 and 1997, using regression analysis to control for a variety of state characteristics like age distribution, unemployment rate, racial composition and even beer consumption. But the variables of principal interest to them are the execution rate andin a novel innovationthe rate of death row commutations. "According to economic theory," explained Mocan, "an increase in executions should send a signal pertaining to an increase in the cost of crime, and an increase in clemency or commutations is a signal indicating the price of criminal activity is actually getting lower."
Again, the results turned out as the authors expected. "We find that one extra execution reduces the homicides by five to six," said Mocan. "And three extra commutations generate one to one and a half extra homicides." The findings, said the authors, are robust to variations in data sets, other variables and mathematical transformations.
Mocan and Gittings also decided to simulate the earlier paper by Katz, Levitt and Shustorovich. "It's an important paper, so we tried to make sure that we were operating in the same framework," said Mocan. Among other things, they added prison deaths to their initial equation to see if it had a significant impact. "We tried hard to make these two papers as comparable as possible. But in our framework, the prison deaths did not turn out to be significant determinants. The execution rates and clemency rates turned out to be more important."
So what does it mean?
The conflicting results are puzzling. Respected economists with similar theories and methods arrive at opposite conclusions about capital punishment. One possibility is simply that things have changeda structural change in deterrence such that executions didn't impact homicide during the time period Katz, Levitt and Shustorovich examined (1950-1990) but did have an effect in the more recent years (1977-1997) that Mocan and Gittings studied.
"Structural change is possible," said Mocan. "We cannot rule out that there was a structural change in the way that executions influence the homicide rate after the mid-1980s or so." In future work, he plans to look at that possibility. But "my gut feeling is that the reason [for different results] is that there is more variation in the data in later years."
Indeed, Katz, Levitt and Shustorovich did find somewhat different results in the first 20 years of their data than in the second 20. So it's possible that adding data from another seven years during which executions rose and murders declined could bolster the strength of the execution deterrent effect, while possibly diminishing the effect of prison conditions.
Levitt, though, is skeptical. The number of executions is just too low to expect it to have either a theoretical or statistical impact of any real significance. Even if executions do have a substantial effect, the annual number of executions is so low and the variability in murder rates so wide from year to year, that it is difficult to detect "the execution-related signal from the noise in homicide rates."
"If all I knew were the coefficients you get from [1977 to 1997], the years Mocan looked at, then I'd say, yeah, it looks like execution works," said Levitt. "But given that we've got an earlier 40 years of experience where we don't see much evidence that it works, and where there's a lot more variation, I guess that would make me a lot more suspicious of drawing any strong conclusions."
Gary Becker, a University of Chicago colleague of Levitt, points out that policy measures can alter deterrence. "The disincentive or deterrent effects through the death penalty will be higher if prisons are made more comfortable ... because that means death [would be] a much worse alternative to imprisonment," he noted, commenting on this Region article. "On the other hand, it's hard to believe that death penalty [deterrence] can be very high under present conditions, when few get it, it's often very delayed and its imposition is highly uncertain." But under different conditions of penalty imposition and prison treatment, capital punishment deterrence could be greater, he noted. "These are political and public policy issues one has to resolve."
Both Levitt and Mocan emphasize that it is not their desire to campaign for or against the death penalty. "The economist's contribution," said Mocan, "is just to provide a piece of objective evidence on whether or not there's a deterrent effect, and how big it is, without taking a position whether it is good or bad." But they recognize that their results will inevitably be used or dismissed by advocates of one position or the other.
Still, they take pains to stress other considerations. Katz, Levitt and Shustorovich caution against drawing policy prescriptions from their results on prison conditions. While the effect on crime of prison deaths is strong in one sense, the elasticities imply that a doubling of the prison death rate would reduce the crime rate by just a few percentage points. "Given the limited efficiency gains implied by these estimates, the moral and ethical considerations surrounding these issues would appear to dominate any economic arguments," they concluded.
And Mocan points out that even if the death penalty is a very strong deterrent, "that still doesn't mean that one needs to support it, because there [are many other] ethical and religious considerations that need to be taken into account." Mocan added that much recent research has found substantial flaws in the application of the death penalty, including incompetent legal counsel, racial disparities and unfair discretion in commutations. The focus of public debate on the issue has begun to shift, said a recent Economist editorial, "from the question of whether the death penalty is right in theory to whether it is fair in practice."
That question hasn't gone unnoticed by the Supreme Court. "Serious questions are being raised about whether the death penalty is being fairly administered in this country," said Justice Sandra Day O'Connor last July. "If statistics are any indication, the system may well be allowing some innocent defendants to be executed." O'Connor noted that over 90 inmates have been exonerated and set free since 1973.
Is deterrence what we seek?
It may well be that the American publicand their policymakerssee capital punishment as a means to ends other than crime deterrence. Over the last 25 years, the gap between those who believe in the death penalty and those who believe it has a deterrent effect has grown ever wider, according to public opinion polls by Harris Interactive. Sixty-seven percent of Americans surveyed in July 2001 said they believe in the death penaltyidentical to the fraction that felt that way in 1976but belief in its deterrent effect has slipped from 59 percent to 42 percent during that same quarter century (see chart).
Banner, the death penalty historian, reports that surveys conducted between 1983 and 1991 indicated that a large majority of death penalty supporters would still favor it even if it had no effect at all on the murder rate. And he suggests that the death penalty has come to serve two other goals for Americans: retribution and collective expression of condemnation. "It is this expressive quality that best accounts for the renewed popularity of capital punishment at the end of the twentieth century." Or as Frank Keating, the governor of Oklahoma, said last year, "Capital punishment is a statement of moral outrage and justice sought and restored."
Beccaria would not have approved. In his mind, the purpose of punishment was not vengeance or venting, but prevention and deterrence. "The intent of punishments is not to torment a sensible being, nor to undo a crime already committed," he wrote. "The end of punishment is no other than to prevent the criminal from doing further injury to society and to prevent others from committing the like offence."
But whether Beccaria was right about the minimal deterrence of capital punishment, we may never know. The best efforts of economists over the last quarter century have been nothing if not inconclusive. Theory itself provides no definitive guidance. "I don't think the economic approach to crime requires or necessarily implies that the death penalty itself has a big deterrent effect," said Becker. "It doesn't rule it out, but it doesn't imply it."
And it seems unlikely that empirical analysis will soon settle the question. Indeed, a draft paper released early this year by economists at Emory University says the deterrent effect of capital punishment is at least three times stronger than the Mocan-Gittings estimate. The latter study, intimate the Emory economists, has "econometric problems."
"What's interesting about this is that it mirrors so closely the Ehrlich debate of the '70s," said Chicago's Levitt, "which basically all came down to if you tweak his specification at all, you get numbers that are totally different." And reaching a definitive answer about deterrence could well be impossible since current execution rates may be too low to provide sufficient empirical data. "I really think not that the answer is 'yes' or 'no,'" said Levitt, "but that there's not enough information to figure it out. There may never be enough. It may just be a question that can't be answered."