Two Views on Imprisonment Policies: Lethal Violence and the Overreach of American Imprisonment Supply Side Imprisonment Policy.

MENU TITLE: Two Views on Imprisonment Policies:
Lethal Violence and the Overreach of American
Imprisonment
Supply Side Imprisonment Policy         
Series: NIJ Research Report
Published: July 1997
32 pages
66,489 bytes

Figures, charts, forms, and tables are not 
included in this ASCII plain-text file. To view 
this document in its entirety, download the Adobe 
Acrobat graphic file available from this Web site 
or order a print copy from NCJRS at 800-851-3420.

U.S. Department of Justice
Office of Justice Programs
National Institute of Justice

Research Report

Two Views on Imprisonment Policies:

Lethal Violence and the Overreach 
of American Imprisonment

Supply Side Imprisonment Policy
 
Presentations From the 
1996 Annual Research and 
Evaluation Conference
Washington, D.C.

------------------------------

From the Director


Each year, the National Institute of Justice
sponsors--in collaboration with other bureaus of
the Office of Justice Programs--a national
conference on research and evaluation. NIJ invites
to the conference respected researchers who are
asked to share their assessment of the knowledge
base on core topics; NIJ then publishes the
presentations to make the debates widely available
to policymakers, practitioners, and researchers
nationwide. The 1996 conference, jointly sponsored
by NIJ, the Bureau of Justice Assistance, and the
Office of Juvenile Justice and Delinquency
Prevention, focused on two themes that reflect
crucial national concerns: crime prevention and
sentencing policies. 

Because of the far-reaching consequences of
sentencing policies on the Nation's correctional
systems, NIJ sought the views of two prominent
scholars on the merits of current sentencing and
imprisonment policies. Their lectures, presented
here, reflect the intense national debate over
these issues and offer readers carefully reasoned
arguments for opposing interpretations of what
research and experience indicate for the future of
public policy in this arena.

This report presents a debate on a topic of
considerable consequence--philosophical, practical,
and fiscal--for public policy. NIJ anticipates that
the views presented here will spark further debate
as part of our national dialogue on effective crime
prevention and control policies for the coming
years.

Jeremy Travis
Director 
National Institute of Justice

------------------------------

Two Views on Imprisonment Policies:

Lethal Violence and the Overreach 
of American Imprisonment

by Franklin E. Zimring
With the Collaboration of Gordon Hawkins

and
 
Supply Side Imprisonment Policy

by Michael K. Block

Presentations From the 
1996 Annual Research and 
Evaluation Conference
Washington, D.C.

July 1997

------------------------------

U.S. Department of Justice
Office of Justice Programs
633 Indiana Avenue N.W.
Washington, DC 20531

Janet Reno
Attorney General
U.S. Department of Justice

John C. Dwyer
Acting Associate Attorney General


Laurie Robinson
Assistant Attorney General

Jeremy Travis
Director, National Institute of Justice

Justice Information Center
World Wide Web Site
http://www.ncjrs.org

Franklin E. Zimring is Professor of Law and
Director of the Earl Warren Legal Institute at 
the University of California. Michael K. Block 
is Professor of Economics and Law, University of
Arizona; President of the Goldwater Institute; 
and Senior Research Fellow, Progress and Freedom
Foundation, Washington, D.C. 

Opinions or points of view expressed in this
document are those of the authors and do not
necessarily reflect the official position of the
U.S. Department of Justice.

------------------------------
 
The National Institute of Justice is a component of
the Office of Justice Programs, which also includes
the Bureau of Justice Assistance, Bureau of Justice
Statistics, Office of Juvenile Justice and
Delinquency Prevention, and the Office for Victims
of Crime.
 

NCJ 165702

------------------------------

Contents

Lethal Violence and the Overreach of American
Imprisonment 

by Franklin E. Zimring

With the Collaboration of Gordon Hawkins

Supply Side Imprisonment Policy

by Michael K. Block

Appendix: Data Sources and Estimates for Tables 
1-1, 1-3, 1-5, and 3-1

------------------------------

Lethal Violence and the Overreach 
of American Imprisonment

by Franklin E. Zimring
With the Collaboration of Gordon Hawkins

The purpose of this paper is to provide a brief
summary of my views on two of what [NIJ Director]
Jeremy Travis has called "core topics" for the
plenary session:

1. What are the differential merits of
incarceration as a sanction with respect to
property, violent, and drug offenses?

2. Does current sentencing policy confine too many
or too few offenders in the United States?

My answers to these two questions are linked. The
serious problem in the United States is not crime
but lethal violence. We have about the same rate of
theft and burglary as other developed nations in
the 1990s but very high rates of those offenses
that put life in jeopardy. At current levels,
property crime is not a serious problem in a
wealthy nation that has insurance and other loss-
spreading devices. Life-threatening violence
creates fear because government and insurance can
do little to ameliorate the harm that life-
threatening assault and rape produce. Life-
threatening violence should be a special priority
for criminal sanctions, and prison should be mainly
reserved for these most serious threats.

Instead, the huge escalation in rates of
imprisonment during the past two decades has
reduced the share of prison population represented
by persons convicted of violent crimes. No matter
what the question in American criminal justice,
prison has been the answer. From an incapacitation
standpoint, this does not reduce the prevention of
violence. But the system's sense of proportionality
and the moral and educative emphasis on violence as
particularly disapproved have suffered as a wider
variety of offenses are routinely sanctioned with
prison. And deterrent incentives to avoid life-
threatening violence are reduced when property
crime routinely leads to prison.

So American criminal justice systems are
overimprisoning in a quantitative and a qualitative
sense. Restoring the proper priority to life-
threatening violence will mostly require nonprison
sanctions for auto thieves and housebreakers.

The following two sections outline the evidence on
crime and lethal violence and show how broadening
the range of prison offenses has shifted the focus
of penal policy away from appropriate priority.

Most Americans believe they are the victims of
crime far more often than the citizens of other
developed nations. With one important exception,
this is not the case. Figure 1 shows police
statistics for Los Angeles and Sydney, Australia.

The two cities have similar levels of nonviolent
property crimes but vastly different levels of
robbery and criminal homicide. In Sydney, victims
are angry about housebreaking but not terrified.
Fear is the more common reaction in Los Angeles.
Living in a city with 20 times the homicide rate,
the citizens of Los Angeles have a specific worry
the citizens of Sydney do not have.
 
Figure 2 shows crime rate comparisons for New York
City and London.

Rates of nonviolent theft offenses were higher in
London than in New York in 1990. Because larceny
and burglary are high-volume offenses, this meant
that the aggregate rate of what we call "index
felonies" was higher in London in 1990 than in New
York. There were more crimes and more criminals in
London, but the homicide rate was more than 10
times higher in New York.

Gordon Hawkins and I sought out data on the
homicides that resulted from victims of robbery 
and burglary being killed by the offender in these
two cities in 1992. A total of 212,000 incidents in
London caused 7 deaths. But the 191,000 burglaries
and robberies in New York caused 378 victim deaths,
a total 54 times as great as in London.

A series of new victimization surveys asking the
same question in many different countries confirms
that U.S. crime rates are not much higher than
crime rates in other industrial nations. Rates of
theft, burglary, and even nonaggravated assault
reported in the United States are rarely more than
30 percent above those of many Western nations
(Zimring and Hawkins, 1997, at chapter 3). But the
rates of killing are 4 to 18 times those of other
developed countries (Zimring and Hawkins, 1997, at
figure 1.5 and chapter 3).

Figures 3 and 4 show comparative rates of property
and violent crimes for several nations surveyed, as
reported by van Dijk and Mayhew.

The United States is clustered with five other
nations in property offense victimization, ranging
from Spain to Poland. Further, 11 of the 17 other
countries in figure 3 are within 30 percent of the
U.S. property crime rate.

Figure 4 provides parallel data for offenses of
violence.
 
For offenses of violence, the United States ranks
second to Australia but has a rate close to the
rates reported by Canada, Poland, and New Zealand.
Eight of the seventeen nations have violent crime
rates within 30 percent of the U.S. rate. The U.S.
rate for lethal violence is a high multiple of
rates in other developed countries.

Why Fear Lethal Violence?

The offender who kills his victim takes something
from that victim and the victim's family that
cannot be given back. By contrast, a compensation
program can make the victim whole in the theft of
the person's BMW with insurance or other loss-
spreading devices. So interests in life and bodily
security are not just more important than property
interests; they also cannot be compensated in a
truly commensurate fashion. When modern states make
property hard to steal and when they facilitate
insurance, they all but guarantee that lethal
violence will be the citizen's major worry about
crime, even in settings where homicide is low. In
the United States, however, where rates are quite
high, the fear of lethal violence is debilitating.

The rational response of a criminal justice system
to the special importance of lethal violence would
be to make the prevention of serious violence the
dominant priority in criminal justice. Because
prison is our most serious available criminal
sanction, such a focus would create a strong link
between life-threatening violence and the available
sanction of greatest seriousness. That has not,
however, been the trend in American criminal
justice lately, as the next section demonstrates.

U.S. Imprisonment Rates

We begin our discussion of U.S imprisonment rates
in familiar statistical territory, with data about
rates of imprisonment per 100,000 population.
Figure 5 comes from The Economist (June 8, 1996)
and shows levels of American imprisonment and
trends over time.

The data in figure 5 support the now familiar
litany of the grandiose levels of imprisonment in
America. As The Economist puts it, "America now
imprisons seven times as many people
(proportionately) as does the average European
country largely as a result of get-tough-on-crime
laws."

Even more impressive than the current level of
American
imprisonment is the rapidity with which levels of
imprisonment have risen in recent years. There were
more than five times as many people in prison in
the United States in 1996 as there were in 1973.
The growth in American imprisonment has been
greater and more sustained during the past two
decades than in any prior period (Zimring and
Hawkins, 1991).

This discussion adds two new twists to the familiar
complaints about high rates of imprisonment in the
United States. The first additional item we would
add to the data on imprisonment rates and
comparison countries is the information about
reported crime rates provided in figures 3 and 4.
It would be one thing if America's sevenfold
advantage in rates of imprisonment paralleled a
sevenfold higher crime rate. But this is clearly
not the case for the most common forms of index
crime. It appears that the English have just as
many thieves as can be found in American cities and
a larger number of burglaries. Thus, rates of
common crime do not begin to explain the large
difference in American rates of imprisonment. 

But what about the large difference among American
rates of homicide, life-threatening assault, and
robbery? The much larger rate of lethal violence in
the United States probably explains one-quarter to
one-third of the difference in rates of
imprisonment between the United States and other
countries. The reason lethal violence differences
do not explain a larger proportion of the
difference in imprisonment is that serious-violence
offenders represent a modest and declining share of
the American prison population. Prior to the huge
expansion in American imprisonment, the
concentration of violent offenders in American
prisons was much more substantial than in current
circumstances.
 
In California, for example, about 60 percent of all
prisoners in 1979 reached the prison system because
they were convicted of violent crimes. That was the
distribution of offenders in a prison system with
an inmate population under 25,000. For the 80,000
prison spaces added to the penal system during the
next decade, the concentration of violent offenders
was much smaller. If the 1979 population is held
constant and if the criminal records of the
offenders occupying additional capacity are
separately analyzed, only 27 percent of the new
space has been allocated to persons convicted of
offenses of violence.

So if lethal violence is the problem, the mechanism
of imprisonment has become less efficient as the
prison population has expanded. This was to be
expected because the most serious offenses produce
imprisonment decisions even when prison populations
are relatively low. The system is not inclined to
put Charles Manson or Willie Horton on probation in
any event. Adding dramatically to prison numbers
means pushing into prison many more people at the
margin between prison and nonprison sanctions.
Expanding prison numbers means imprisoning larger
numbers of housebreakers and car thieves to join
the armed robbers already incarcerated under prior
policy. Sure enough, as the rates of imprisonment
mushroomed in California during the 1980s, the
number of burglars in prison grew more than three
times as fast as the number of convicted robbers,
and the number of car thieves grew six times as
fast (Zimring and Hawkins, 1992:39). There can be
no doubt that one natural result of sharp increases
in imprisonment is the watering of the stock by
increasing the proportionate share of less serious
offenders behind bars.

Does this massive addition to a prison population
of offenders not convicted of violent crimes reduce
the capacity of the system to prevent violence?
There are three functions of imprisonment as a
criminal sanction to be considered separately in
addressing this question: (1) incapacitation, (2)
the educative or moralizing function of the threat
of imprisonment, and (3) general deterrence.
Considering only incapacitation, the imprisonment
of large numbers of nonviolent offenders does not
reduce the amount of violence prevented by
incapacitation as long as there are no practical
upper limits on the amount of prison space
available in the criminal justice system. If extra
nonviolent offenders are imprisoned but no violent
offenders avoid imprisonment as a result, the net
amount of violence prevented by imprisonment will
probably be larger as a result of the prison
expansion. Thus, the prison system in California,
which held more than 100,000 prisoners in 1991,
probably prevented a larger number of life-
threatening acts of violence than the prison system
of fewer than 25,000 a decade before. How many
extra acts of life-threatening violence were
prevented is not known, however, and there are
indications that the number was relatively small
(Zimring and Hawkins, 1995, at chapter 6; Cohen and
Canela-Cacho, 1994). But even if the marginal
returns from extra imprisonment are very low, the
absolute number of violent offenses prevented
should increase. 

If the amount of prison capacity available for
incapacitation is limited, using that limited
capacity for those convicted of nonviolent offenses
can reduce the amount of life-threatening violence
prevented if future life-threatening violence is
more common among persons already convicted of
violent offenses. So a huge increase in
imprisonment risk for thieves and drug sellers
would reduce the net amount of violence prevented
by imprisonment if nonviolent and violent offenders
are in a zero-sum competition for scarce prison
space. But if the scale of imprisonment can be
expanded indefinitely, even a large increase in the
proportion of nonviolent offenders imprisoned will
not lead to a palpable reduction in deaths and
injuries prevented.

The period 1980-94 appears to have been a time when
additional prison space for nonviolent and drug
offenders could be provided without diminution of
the penal resources available for persons convicted
of life-threatening violence. In the more crowded
and resource-competitive conditions of the late
1990s, however, the prospects for prison expansion
may not be unlimited, and locking up a large
proportion of those without previous records of
violence could reduce the amount of violent crime
restrained by penal confinement.

The American criminal justice system may already be
shooting itself in the foot with respect to the
educative and moralizing influence of criminal
punishments generally and imprisonment
specifically. When 60 percent of a prison
population is serving time because of a conviction
for violent crime, the criminal justice system is
drawing a clear boundary between serious violence
and other types of crime. That is an environment
where armed robbery and housebreaking are regarded
as very different types of crime. Vastly expanding
the imprisonment of housebreakers and car thieves
blurs the distinction between crimes that involve
the risk of injury to the victim and those that do
not.

There are a number of indicators that the educative
and moralizing influence of criminal sanctions can
be a significant influence on the behavior of
potential criminals. In areas as diverse as drunk
driving and domestic violence, changes in criminal
justice policy that emphasized the seriousness of
driving after drinking and willful domestic injury
played a major role in reducing death rates from
automobile accidents and intimate homicide.
Blurring the distinction between robbery and
burglary in American criminal justice is unwise
because the risk of death in the average robbery is
50 times as great as in the average burglary in the
United States (Zimring and Hawkins, 1997, at
chapter 4).

And an imprisonment policy that treats the robber
and the burglar in a similar fashion mutes more
than the educative and moralizing influence of the
criminal law. Reducing the difference in the threat
of punishment for armed robbery and housebreaking
reduces the incentive for the potential criminal to
choose the less serious of the two methods of
obtaining property. Thus, a large increase in
imprisonment that reduces the gap between burglary
and robbery may produce more armed robbery as a
result of the smaller gap between the burglary and
robbery punishment, even if the threatened
punishment for robbers is expanded. To the extent
that the relative magnitude of punishment threats
influences the choice of crime, general crackdowns
on criminality could produce higher rates of lethal
violence. Indeed, if a general crackdown on crime
has to deter 50 additional burglaries for every
burglar it converts to robbery to break even on
crime victim deaths, the impact of such a policy on
the death rate from crime may well be undesirable.

There is one other casualty worthy of mention when
a war on crime greatly extends the range of
offenses to be punished with imprisonment, and 
that is the sense that the punishments meted out by
the system are proportionate to the seriousness of
the crimes committed (Allen, 1996:43-46). The
diverse offenses found in a modern penal code
involve not only many different kinds of harm but
also substantially different degrees of social
cost. Policies that tend to homogenize punishments
and spread them evenly over offending populations
are not only problematic because potential
offenders may miss important differences in
culpability--this is the earlier point about the
educative and moralizing influence of punishment--
but also misleading because citizens and those who
enforce the law may regard all criminal harms as
morally indistinguishable.
 
The principal recent misadventure in American
criminal justice that tended in this direction was
the "War on Drugs" in the mid-1980s. Making prison
the presumptive punishment for an ever-widening
list of offenses is a significant symptom of a
diminished sense of penal proportion. In this
sense, too, large increases in imprisonment obscure
the moral differences between different types of
crime.

References

Allen, Francis A. (1996). The Habits of Legality.
New York: Oxford University Press.

Cohen, Jacqueline, and Jos‚ Canela-Cacho. (1994).
"Incapacitation and Violent Crime." In Albert J.
Reiss, Jr., and Jeffrey A. Roth, eds.,
Understanding and Preventing Violence. Volume 4,
pp. 296-388. Washington, D.C.: National Academy of
Sciences.

"Crime in America: Violent and Irrational--and
That's Just the Policy." The Economist (June 8,
1996), pp. 23-24.

Jones, Michael A., and James Austin. (1995). The
1995 NCCD National Prison Population Forecast.
Washington, D.C.: National Council on Crime and
Delinquency.

Mauer, Marc. (1994). Americans Behind Bars: The
International Use of Incarceration, 1992-1994.
Washington, D.C.: The Sentencing Project.

New South Wales Bureau of Criminal Statistics.
(1992). Personal communication.

United Kingdom Home Office. (1990). Criminal
Statistics in England and Wales. London.

U.S. Department of Justice, Federal Bureau of
Investigation. (1993). Uniform Crime Reports, 1992.
Washington, D.C.: U.S. Government Printing Office.
 
U.S. Department of Justice, Federal Bureau of
Investigation (1991). Uniform Crime Reports, 1990.
Washington, D.C.: U.S. Government Printing Office.

van Dijk, Jan, and Pat Mayhew. (1992). Criminal
Victimization in the Industrialized World. The
Hague: Ministry of Justice.

Zimring, Franklin E., and Gordon Hawkins. (1997).
Crime Is Not the Problem: Lethal Violence in
America. New York: Oxford University Press.

______. (1995). Incapacitation: Penal Confinement
and the Restraint of Crime. New York: Oxford
University Press.

______. (1992). Prison Population and Criminal
Justice Policy in California. Berkeley, California:
Institute of Governmental Studies.

______. (1991). The Scale of Imprisonment. Chicago:
University of Chicago Press.

------------------------------

Supply Side Imprisonment Policy

by Michael K. Block

My hypothesis about how imprisonment affects public
safety is quite straightforward. I am one of those
few but growing number of academics who believe
that threatening and, not too infrequently,
actually imposing noticeable prison sentences helps
to control crime. Moreover, I think the empirical
evidence, such as it is, supports this position.

I am convinced that there are too many prisoners
and prisons in the United States today not because
we overuse imprisonment but, quite the contrary,
because in major part in the past we have not been
willing enough to imprison serious offenders. There
are too many prisoners because there are too many
criminals committing too many crimes, and we find
ourselves in this predicament at the turn of the
century because, for most of the last half of the
20th century, sentencing practices have not been
harsh enough.

This is by no means the conventional wisdom among
my colleagues. More common is the assertion that we
overuse imprisonment. The unflattering comparison
of the ratio of prisoners to population in the
United States with the ratios in other major
industrial powers is adduced as evidence of our
wrongheadedness on this score.[1] Even if one
ignored the empirical evidence on the disincentive
effects of imprisonment, the ratio of prisoners to
population would not be particularly informative.
Only if the propensity to commit crime and the
costs of controlling crime were similar in all of
the industrial nations would this comparison of per
capita imprisonment be relevant. Casual empiricism
suggests that this precondition is not even roughly
satisfied. Why this is the case is not well
understood, but even a brief treatment of the issue
is well beyond the scope of this paper.

It is also quite conventional to argue that we have
tried getting tough on criminals and have failed.
The refrain here is that the ratio of prisoners to
population has increased dramatically since 1980,
but the crime rate, particularly the violent crime
rate, has not fallen appreciably. In actuality,
violent crime is down quite substantially of late,
but, again, this comparison of ratios of prisoners
to population only makes sense to the extent that
the propensity to commit crime--or, in the
vernacular, the "criminal element"--has remained
relatively stable over time. The empirical evidence
suggests that this is not the case. Moreover, if we
focus on the more relevant ratio of imprisonment 
to crime, the trends since 1980 are not
unambiguously upward.[2]

Instead of comparing per capita imprisonment across
countries or even over time, a more reliable
comparison to judge imprisonment policy would be
one between the costs and benefits of changing that
policy. As shown in the work that follows, such a
comparison generally indicates that our
imprisonment policy has not been harsh enough and
that the threat of additional imprisonment would
reduce crime and might actually reduce the number
of individuals in prison. 

Crime and Punishment in Modern America

Table 1-1 shows that it is possible to gain some
appreciation for the actual magnitude of the crime
problem and the scale of our response in terms of
arrests and imprisonment. According to the most
recent National Crime Victimization Survey, almost
35 million index crimes (homicide, rape, robbery,
assault, burglary, auto theft, and larceny) were
committed in 1994.[3] According to the Federal
Bureau of Investigation (FBI), about 14 million of
these crimes were reported to them by police
departments around the country. In terms of
arrests, these same police departments made almost
3 million arrests for index crimes in 1994. Table
1-2 reveals that only about 1 in 5 reported crimes
(or about 1 in 12 actual victimizations) are
cleared by an arrest.

What is perhaps most striking about the data
presented in tables 1-1 and 1-2 is the enormous
falloff between arrests and convictions. While
there were almost 3 million arrests for index
crimes in 1994, there were only about 375,000
felony convictions for such crimes in State courts
during the same year. Only about 13 percent of the
arrests for index crimes actually ended in a
conviction for a felony. Even for crimes like
murder and rape, the ratio of convictions to
arrests was remarkably low--only slightly over 55
percent in both cases. Undoubtedly, plea agreements
muddy the water here. A number of those arrested
for one crime were convicted of or pled guilty to
another. When we look at the violent crime category
as a whole, fewer than 1 in 5 of those arrested for
a violent crime were actually convicted of a
violent crime.

To account for convictions and pleas to a lesser
offense and to obtain a more precise idea of
conviction probabilities, data from the National
Pretrial Reporting Program were used to estimate
the likelihood that an offender arrested for a
specific crime was convicted of any crime (table
1-3). According to table 1-3, when we use this
additional information, the conviction risk for
murder reached a more respectable 71.5 percent.
(The conviction risk for rape in this table is
quite unreliable because of the differences in
definitions of the crime between data sources.)
However, even when all conviction offenses are
accounted for, the overall conviction risk for
violent crimes as a group is still quite modest
(27.6 percent) and for all index crimes as a group
it is even less impressive (17.3 percent). If an
offender is arrested, and this is by no means a
high-probability event, the likelihood of actually
being convicted of the crime that he/she was
arrested for (or, for that matter, any crime) is
remarkably low.

Given the fact that the likelihood of being
arrested and convicted for an index crime serious
enough to be reported to and by the police is only
about 3 percent or 4 percent, the consequences
after conviction are enormously important in
determining the expected punishment.[4] The
"Imprisonment Risk" column in table 1-2 contains my
estimate of the likelihood of being sent to prison
after conviction for an index crime. The likelihood
of an offender convicted of an index crime being
sentenced to prison is about 52 percent. For an
offender convicted of a violent crime, the
offender's chance of being sent to prison is
estimated to be about 63 percent. Of course, these
ratios misstate the likelihood of imprisonment for
an offender arrested for a specific crime to the
extent they ignore the real possibility that he/she
may be convicted of a less serious crime than that
for which he/she was arrested. As is clear in table
1-3, this is a common occurrence. In table 1-4 the
estimates of imprisonment risk are recalculated to
reflect the conviction outcomes in table 1-3. This
recalculation, as expected, substantially lowers
all of the imprisonment risks.

Using my estimates of arrest, conviction, and
imprisonment risks, the probability of imprisonment
was calculated for the various index crimes. This
is presented in the last columns of tables 1-2 and
1-4. What is obvious from the calculations in the
two tables is just how hard it is to get into
prison in the United States. Only about 1 in 100
index crimes reported to the police actually result
in imprisonment for an index crime. Even for
violent crimes, only about 5 in 100 such crimes
result in imprisonment. The ratio varies from 1 
in 3 for murder to about 3 in 100 for aggravated
assault. Overall, the likelihood of going to prison
for committing an index crime is surprisingly low.
And this does not change appreciably when
convictions for lesser offenses are considered
(table 1-4). As I will argue below, increasing the
probability of imprisonment for these crimes by
even modest amounts will pay handsome dividends.

Although it is true that imprisonment is not the
only possibility for incarceration after conviction
(the individual could be sent to a county jail),
and that incarceration is not the only form of
punishment, imprisonment is clearly the most severe
sanction. Even when jail time is given for an index
offense, it tends to be much shorter than a prison
term. And although probation, especially the more
recent intensive probation programs, imposes some
punishment, it is usually not comparable to prison.
Because prison is still by far the most severe form
of punishment that is used on a regular basis, it
is of special interest to those interested in the
disincentive effect of punishment.[5]

Up to this point, I have concentrated on the
likelihood of being punished for committing a
serious crime. However, in analyzing the
disincentive effect of imprisonment policy, it is
also important to consider the severity of the
punishment. As noted above, the most severe
punishment regularly imposed is prison. As a
practical matter, the only method of increasing the
severity of a prison sentence is to increase its
length. Given the propensity of most offenders to
think only over the short term, this turn of events
is quite unfortunate. Recent jurisprudence on the
rights of prisoners and the standards of
confinement appear to leave policymakers with only
the weakest instrument (sentence length) for
increasing the severity of punishment. Policy
initiatives aimed at increasing the unpleasantness
of prison life would likely be a cost-effective
method of fighting crime.
 
In table 1-5 average time served and "to be served"
is displayed for index crimes in 1992. The time-
served data in the first column are the average or
mean time served by offenders convicted of the
specific index crime and released in 1992. For
example, the 71.5 months shown for homicide in the
first column is the average time served by all
offenders (including parole violators) who were
released in 1992. The "to be served" column is the
Bureau of Justice Statistics' estimate of the
average time that will be served to first release
of those convicted of these specific index offenses
in 1992. (I did not attempt to adjust time-to-be-
served estimates for any expected "add on" for
parole violations because of the changes in
imprisonment policy, including elimination of
parole, that have been sweeping the country.)
Columns 3 and 4 show the same information adjusted
for the fact that not all those charged with a
specific crime actually get convicted of that
crime. These entries reflect the average time
served of those charged with the specific index
crime and convicted and sent to prison for some
crime.


What is quite apparent in table 1-5 is how the
length of prison sentences has been increasing in
recent years, especially for some violent crimes.
The average prison time just to first release for
those convicted of murder in 1992 is almost 70
percent longer than the time served by all those
released in 1992 who were convicted of murder. For
index crimes as a whole, time to be served in 1992
was 31 percent longer than the time actually served
for those released in 1992.
 
Notwithstanding the recent trend of increasing
sentence length, imprisonment is still very rare
relative to crime. This is illustrated in tables
1-6 and 1-7. The "probability of imprisonment"
(table 1-2) and the average "time to be served"
estimates (table 1-5) are combined to get expected
months per criminal and reported crime.

Reported index crimes, which are arguably the most
serious crimes, have an expected sentence of 18
prison days. In other words, if all of the prison
time given out in a year for index crime is
totaled, it averages out to 18 prison days for
every reported index crime. What this means is that
a criminal committing a number of index crimes can
expect to spend 18 days in prison for every crime
he/she commits. This overall average for all index
crimes hides a good deal of variation. Performing
the expected prison time calculation by type of
crime produces averages that vary from nearly 40
months for every murderer to 3 days for every petty
thief. Nevertheless, it is clear from the results
reported in tables 1-6 and 1-7 that very little
prison time is delivered relative to the volume of
serious crime.

Is Imprisonment an Effective Crime 
Control Strategy?

Over the past several decades, a number of efforts
have been made to quantify the deterrent effects of
both the severity and certainty of punishment. In
the mid-1980s Lewis (1986) reviewed most of the
econometric evidence on the deterrent effect of
increases in the length of imprisonment published
up to that time. His review and analysis of the
literature were clearly supportive of the
hypothesis that an increase in the length of prison
terms reduces crime. His results, which are
reproduced in table 2-1, suggest that longer prison
sentences have a deterrent effect and that these
effects are statistically significant at high
levels of confidence for many crimes and categories
of crimes.

The studies that Lewis (1986) summarized all derive
their estimates of the deterrent effect of
imprisonment by attempting to statistically control
for nondeterrence-related changes in the
environment. Most include some control for changes
in demographic factors over time or differences
between areas at a point in time. Some studies
attempt to control for changes (or variations) in
economic conditions over time or between different
areas at the same time. The elasticities that
result from these studies are intended to provide
information on how crime rates would change if only
deterrent variables, such as the likelihood and
severity of punishment, change. To the extent that
other factors are also changing, crime rates will
reflect changes in both deterrent levels and
demographic and economic variables.

Lewis finds that in the studies he reviewed, the
mean elasticity of sentence length (for index
crimes) was -0.468. Translated from "econo-speak,"
this means that, on average, a 10-percent increase
in the sentence length for index crimes is
estimated to result in a 4.68-percent decrease in
such crimes. For homicide he finds that, on
average, the studies he reviewed suggest that a 10-
percent increase in sentence length will result in
a 2-percent decrease in the murder rate, a 7-
percent decrease in the rate for rape, about a 4.7-
percent decline in the robbery rate, and a 6-
percent decrease in the rate for aggravated
assault.

Because there is nothing like the Lewis (1986)
survey currently available for the econometric
estimates of the likelihood of imprisonment, Block
and Herbert (1994) adopted his methodology to
assess the effectiveness of increasing the
certainty of imprisonment. The same approach was
used to prepare a similar but much abridged study
for this report. To produce estimates of the
deterrent effect of the certainty of imprisonment
comparable to Lewis' findings on the deterrent
effect of the length of imprisonment, the
elasticity-averaging technique employed by Lewis
was also used in this study. Most of the articles
included in Lewis' study are also included here.
The exceptions to this rule were two articles in
which it was unclear how the data were included in
Lewis' study. The present study also includes three
additional articles not included in Lewis' survey.


The actual calculation of the certainty elasticity
estimates parallels the method used by Lewis.
First, if any given article contained more than one
estimate of a single elasticity for a particular
crime, an unweighted average of those elasticities
was calculated. Second, across articles, an
unweighted average of elasticities derived from a
common data set was calculated so that there is
only one estimate for each type of crime per data
set. Finally, for each type of crime, the estimated
elasticity of certainty was calculated by taking
the average of the elasticities estimated from each
data set.

Increasing the certainty of punishment may involve
increasing one or a number of different
probabilities: It may invoke increasing the
probability of arrest, conviction, or imprisonment.
Estimates for all these elasticities are included
in Block and Herbert (1994). However, since the aim
of the present analysis is to demonstrate the power
of increasing imprisonment as a method of
controlling future offense rates, subsequent
analysis will focus on the probability of
imprisonment. Estimates for these elasticities
appear in table 2-2. For additional details on the
individual empirical studies used as a base for
table 2-2, see Block and Herbert (1994).[6]

In table 2-3 Lewis' (1986) estimates of sentence
length elasticities were contrasted with my
estimates of the certainty elasticities. Overall,
the results suggest the existence of deterrent
effects for both severity and certainty. However,
with the exception of larceny, it appears that for
most crimes there is a greater deterrent effect for
an increase in the likelihood of imprisonment than
for an increase in sentence length. Moreover, the
results of both analyses (with the exception of
murder in Lewis' analysis) indicate that the
effects of the severity and certainty of punishment
are greater for individual violent crimes than for
individual property crimes.

The findings comport well with conventional wisdom
(it is reassuring that such "wisdom" is not always
wrong) and is buttressed by my experimental
findings on deterrence. In a series of economic
experiments that included monetary payoffs and
penalties, convincing evidence was obtained that
prisoners are much more powerfully deterred from
criminal acts by an increase in the likelihood that
the penalty will be imposed than by an increase in
the severity of the penalty.[7]

More Cops, More Prosecutors, 
More Prisons?

In the last several years, a number of econometric
studies have appeared that concentrate on directly
measuring the impact on crime rates of increasing
the prison population and suppress the distinction
between the certainty and severity of punishment.
The most recent of these studies, "The Effect of
Prison Population Size on Crime Rates," was
authored by Levitt and appeared in the May 1996
Quarterly Journal of Economics. Using a somewhat
different approach than the econometric studies of
deterrence summarized above, Levitt finds that a
10-percent increase in the prison population leads
to between a 3-percent and 4-percent decline in
index crimes. This is roughly comparable to the
findings shown in table 2-3. Levitt's estimates are
significantly higher than the elasticities obtained
by Marvell and Moody (1994) in their study--
summarized in "Prison Population Growth and Crime
Reduction" in the Journal of Quantitative
Criminology--that also directly addressed the
impact of imprisonment on crime. It should be
noted, however, that Marvell and Moody did not
employ as sophisticated an estimating technique as
Levitt, who used information on prison litigation
to obtain more reliable estimates of the impact of
imprisonment on crime.

Although the results of these studies are
comforting because they both suggest that
imprisonment works to control crime, and the
implication of Levitt's estimate is that additional
imprisonment is likely to be warranted, the
blending of the deterrent effects of certainty and
severity has a cost. Specifically, these studies--
relating increases in the prison population to
crime rates--are not very informative about what is
the most effective way to increase the prison
population.

In the short run, prison populations will increase
as a consequence of changes in either the
likelihood of imprisonment or the length of prison
sentences. One relevant question that previous
studies addressed but that more recent studies do
not is whether one of these approaches is better
than the other.

The implication of comparing the empirical
estimates of the elasticities in table 2-3 is that
change in the certainty of imprisonment is a more
effective approach to controlling crime, especially
violent crime. However, when comparing the efficacy
of increases in the certainty of imprisonment with
the increases in the length of prison terms, the
point has traditionally been made that increases in
certainty may be even more expensive than increases
in prison terms. Even if increases in certainty are
more powerful than increases in severity, cost
considerations may be offsetting. Whatever the
merits of this argument in general, and clearly
there is a strong presumption in its favor, in the
current environment it is unlikely to be true.
 
Although increasing arrest rates and conviction
rates may be resource consuming, they are not the
only, or, for that matter, the most direct, method
of increasing the probability of imprisonment. With
the likelihood of being sentenced to prison after
conviction at only about 52 percent for index
crimes and 62 percent for violent crimes, changes
in sentencing practices could deliver a fairly
substantial increase in the likelihood of
imprisonment with costs comparable to simple
increases in sentence length. The quickest and
least expensive way to increase the certainty of
imprisonment is simply to change sentencing policy.

Costs and Benefits of Changes in Imprisonment
Policies

One clear way to illustrate the relative
attractiveness of increases in sentence length
(average time served) and imprisonment risk is to
present these alternatives in a cost-benefit
framework. To that end, tables 3-1, 3-2, 3-3, and
3-4 present a cost-benefit analysis for both a 5-
percent increase in the average time served and a
5-percent increase in imprisonment risk. The
benefits are the crimes averted, and they are
valued at the total costs that each type of crime
imposes on victims. These estimates are those used
by Levitt (1996) and include both estimates of the
pecuniary and nonpecuniary losses resulting from
each of the index crimes. The costs of changing
either of these imprisonment policies (length of
time served and sentencing practices) are basically
the costs associated with adding additional
prisoners. To err on the conservative side, an
estimate of the full cost of imprisonment of $2,765
per month per prisoner was used. 

Because the monthly cost of other common sentencing
outcomes such as probation or jail are not zero,
the appropriate cost for the analysis of increasing
imprisonment risk should be the net cost, or the
difference between the cost of other options such
as probation or jail and prison. Likewise, if an
increase in sentence length comes at the expense of
parole time, the costs of the longer prison terms
should reflect an offset for parole expense
avoided. By ignoring these refinements, the results
are biased against finding a positive benefit-cost
ratio of any policy change that leads to more
prison, and, in particular, an increase in
imprisonment risk.

As is apparent in tables 3-1 and 3-2, increasing
the risk of imprisonment has an unambiguously
favorable benefit-cost ratio in all cases except
burglary and larceny. For victimization data the
case is similar for increases in the length of
prison terms. However, for reported index crimes,
increases in the length of sentences have marginal
benefit-cost ratios for robbery and rape. Overall,
as one would predict from the relative
elasticities, increases in imprisonment risks are
more attractive policy options. Increasing the
likelihood of imprisonment following conviction
produces benefit levels substantially in excess of
costs for all violent crimes and auto theft. The
results in tables 3-3 and 3-4 provide benefit-cost
calculations for crime aggregates. The results for
the aggregates are presented as both a simple sum
of the relevant crimes (total) and an independent
estimate using the aggregate elasticities.

Getting Tough and Getting It Wrong

What this exercise highlights is the finding that
there are substantial gains to revamping the
sentencing structure so that more criminals,
especially violent criminals, are sentenced to
prison rather than to probation or jail. These
benefit-cost results clearly indicate that too few
violent criminals are imprisoned in the United
States.

As noted above, recent attempts at "getting tough"
appear to have taken the form of longer prison
sentences and not higher levels of imprisonment
risk. Contrary to the situation observed in table
1-5 where it was clear that average time served has
been increasing in recent years, no such trend is
discernible for imprisonment risk in table 3-5. In
fact, if any trend is observed, it is a downward
trend. The benefit-cost results imply that this
concentration on increasing sentence length in
recent years is not particularly good public
policy. 

Do a Violent Crime, Do Time: A Supply Side
Imprisonment Policy 

One simple modification of the sentencing structure
that would be consistent with the benefit-cost
analysis above would be to adopt a practice of
sentencing all offenders convicted of violent crime
to prison. This would change the sentencing system
so that probation and/or jail were no longer
sentencing options for offenders convicted of
violent crimes.

To see how a "prison only" approach to sentencing
for violent crime would fare in terms of benefit-
cost analysis and prison populations, a rough cost-
benefit analysis was conducted for the case where
all offenders convicted of homicide, rape, robbery,
and aggravated assault are sentenced to prison for
the current average time served. 

To construct this hypothetical case, the percentage
change in imprisonment risk implied by a policy of
making prison a certainty for all violent crime
convictions was first calculated. Since the
imprisonment risk for violent crime is currently
61.84 percent (table 1-2), it would take an
approximately 62-percent increase in the risk of
imprisonment to make imprisonment a certainty for
all offenders convicted of a violent crime.[8]
 
Using this 62-percent change in the probability of
imprisonment along with the elasticities in table
2-2, the year-by-year path of crimes and prisoners
as well as costs and benefits were estimated for
implementing the prison-only sentencing scheme for
violent crime. In these calculations it was assumed
that the representative construction cost was
$40,000 per bed and that all cells were paid for in
the year they were built. The annual cost of
keeping an inmate, exclusive of interest, was
assumed to be $2,500 per month. The results of this
analysis appear in table 3-6.

Again, to err on the conservative side, it was
assumed in each of these examples that the
deterrent effect did not begin until the fifth year
and that it did not reach its full effect until the
eighth year after the sentencing reform was
instituted. Even with these unfavorable behavioral
assumptions and the fact that the prisons had to be
fully paid for in the year they went into service,
the policy had a positive benefit-cost ratio by the
sixth year, and every violent crime had a positive
total benefit-cost ratio by the eighth year. This
is without considering the reduction in
expenditures for arrests and convictions that will
also result from this imprisonment policy. 

Not only does this example have favorable total
discounted benefit-cost ratios, it will also
eventually induce a lower prison population.
Although the benefit-cost ratio for the policy
turns positive before the prison population
stabilizes, it will stabilize at a level lower than
the level it would have been had there been no
change in imprisonment policy. For all types of
violent crime, the initial increase in imprisonment
rates results in a lower prison population as well
as a lower crime rate. It will also eventually
result in fewer arrests and convictions.
 
A Prison-Only Approach in a Less Accommodating
World

There is, however, one factor discussed above that
may act to reduce the power of imprisonment and,
thus, the benefit-cost ratio reported in table 3-6.
Because the elasticities in table 2-3 were
calculated by averaging studies conducted using
data between 1940 and 1970--to the extent that
there is a downward trend in these elasticities,
possibly resulting from the increase in the level
of social pathology since 1960--the elasticities in
table 2-3 may overstate the impact of increases in
the likelihood of imprisonment on crime rates.[9]
Although there is some evidence that the
elasticities, at least for violent crimes, have
declined in recent years, there are not enough
studies to adequately explore this issue
empirically. 

We can, nevertheless, provide some perspective on
this potentially important point by asking what the
benefit-cost pattern of the prison-only policy for
violent crime would be if the elasticities were
only half of those reported in table 2-3. The
results for this much less powerful variant appear
in table 3-7. What is observed is that, if the
elasticity with respect to the likelihood of
imprisonment is now substantially less than an
estimate based on historical data, the benefits of
a prison-only policy for violent crime are still
likely to outweigh the costs of such a policy.
However, in this case the eventual prison
population, while lower than the population level
immediately after the intervention, will not dip
below its original level.

Real Offense Sentencing for Violent Crime

An important aspect of the implementation in a
prison-only policy for violent crime not explicitly
accounted for in the analysis reported in tables
3-6 and 3-7 is the change in plea practices such a
policy might induce. Mandatory prison sentences for
any serious violent crime convictions might, by
their effect on plea practices, increase the number
of violent charges that result in nonviolent
convictions. This would reduce somewhat the
deterrence yield of the imprisonment-only policy
for violent crime, but the existing elasticity
estimates are not up to the task of modeling this
type of feedback. One way to avoid some of this
undercutting of the deterrent effect of a prison-
only approach is to adopt "real offense"
sentencing, i.e., sentencing based on offense
behavior and not on the offense of conviction.

Conclusion

Although the benefit-cost calculations in this
report, especially those involving hefty changes in
imprisonment risks, stretch the empirical estimates
of deterrence elasticities to their breaking point,
the examples do make a point. Specifically, there
is a credible argument that the excessive levels of
crime and imprisonment in society have as their
cause a sentencing structure that is not harsh
enough. There is evidence that instituting a more
draconian sentencing structure, i.e., one that
involves imprisonment for essentially all
convictions for violent crimes, is likely to be
cost justified and may actually reduce the prison
population. Imprisoning additional offenders is
expensive, and society has to sacrifice current
consumption to build and operate prisons. However,
the evidence suggests that the costs are worth it,
at least for violent crimes. Making imprisonment a
certainty for violent crimes will make society much
better off.

Notes

1. "[T]he evidence is that we are quite harsh, in
general harsher than other modern industrial
nations, and that we have gotten strikingly harsher
in recent years.... In 1988, the U.S. rate of
incarceration (in jails and prisons) was 388
prisoners per 100,000 people.... For the same year,
the rate for the United Kingdom was 97.4 per
100,000, for West Germany it was 84.9, for Belgium
it was 65.4. The rate for Canada, with a society in
many ways much like our own, was about 110 persons
for every 100,000 inhabitants." (Jeffrey Reiman,
The Rich Get Richer and the Poor Get Prison:
Ideology, Class, and Criminal Justice, 4th ed.,
Boston, Massachusetts: Allyn & Bacon, 1995, p. 18.)

2. In particular, if we consider the ratio of new
prison commitments to crime, we find that in 1980
the ratio was approximately 1 in 100 for index
crimes and 4 in 100 for violent crimes. Both ratios
were essentially unchanged in 1992. It is true that
the average time served has been increasing of
late, and, hence, there have been some increases in
the ratio of prison population to crime. But the
willingness to use prison as a sanction, which is
probably best measured by the new commitment rate,
has not changed appreciably in recent years.

3. Here, and in the work that follows, the term
index crime is used to refer to the crimes of
homicide, rape, robbery, aggravated assault,
burglary, auto theft, and larceny. In actuality,
the FBI uses the term "modified index crime" to
refer to this group of crimes because "index crime"
now also includes the crime of arson.

4. If we ignore the punishment that results simply
from being arrested and assume that imprisonment is
the only substantial punishment, the expected
punishment (EF) can be expressed as  
[EF = pA* pC * pI * F] 
where pA is the probability of arrest, pC is the
probability of conviction, pI is the probability of
imprisonment given conviction (imprisonment risk),
and F is the average time served.

5. While conviction itself imposes a reputational
penalty, this is most relevant for offenders who
commit white-collar crime, and such crimes are not
a very significant proportion of the index crime
figures used in this report.

6. The elasticities of the likelihood of
imprisonment can vary substantially. It is
important to bear in mind when interpreting the
elasticities in table 2-2 that these estimates are
themselves based on often disparate estimates.

7. See Block and Gerety (1995).

8. This is equivalent to increasing the probability
of imprisonment for violent crime from 0.053 (table
1-2) to 0.086.

9. The relevant elasticities may also be smaller
than the average based on historical evidence
because we are analyzing such a large change in the
imprisonment rate.

Bibliography

Avio, K., and C. Clark. (1978). "The Supply of
Property Offenses in Ontario: Evidence of the
Deterrent Effect of Punishment." Canadian Journal
of Economics, Vol. 11, pp. 1-19.

Bartel, A. (1979). "Women and Crime: An 
Economic Analysis." Economic Inquiry, Vol. 17, pp.
29-51.

Becker, G.S. (1968). "Crime and Punishment: An
Economic Approach." Journal of Political Economy,
Vol. 76 (March/April), pp. 169-217.

Block, M.K., and V.G. Gerety. (1995). "Some
Experimental Evidence on Differences Between
Students and Prisoners Reactions to Monetary
Penalties and Risk." Journal of Legal Studies
(January). 

Block, M.K., and C.L. Herbert. (1994). "Deterrence:
What We Know." University of Arizona. Unpublished
working paper.

Blumstein, A., J. Cohen, and D. Nagin. (1978).
Deterrence and Incapacitation: Estimating the
Effect of Criminal Sanctions. Washington, D.C.:
National Academy of Sciences.

Chamlin, M.B., H.G. Grasmick, R.J. Bursik, Jr., and
J.K. Cochran. (1992). "Time Aggregation and Time
Lag in Macro/Level Deterrence Research."
Criminology, Vol. 30(3), pp. 377-395. 

Cohen, M.A. (1990). "A Note on the Cost of Crime to
Victims." Urban Studies, Vol. 27(1), pp. 139-146.

___________. (1988). "Pain, Suffering, and Jury
Awards: A Study of the Cost of Crime to Victims."
Law & Society Review, Vol. 22(3), pp. 537-555.

Ehrlich, I. (1981)."On the Usefulness of
Controlling Individuals: An Economic Analysis of
Rehabilitation." American Economic Review, Vol. 71,
pp. 307-322.

________. (1973). "Participation in Illegitimate
Activities: A Theoretical and Empirical
Investigation." Journal of Political Economy, Vol.
82, pp. 521-565.

Ehrlich, I., and R. Mark. (1977). "Fear of
Deterrence: A Critical Evaluation for the Report of
the Panel of Research on Deterrent and
Incapacitation Effects." Journal of Legal Studies,
Vol. 6, pp. 293-316.

Grogger, J. (1991). "Certainty vs. Severity of
Punishment." Economic Inquiry, Vol. 29 (April), pp.
297-309.

Levitt, S.D. (1996). "The Effect of Prison
Population Size on Crime Rates: Evidence From
Prison Overcrowding Litigation." Quarterly Journal
of Economics, Vol. CXI (May), pp. 319-351.

Lewis, D.E. (1986). "The General Deterrent Effect
of Longer Sentences." British Journal of
Criminology, Vol. 26(1), pp. 47-62. 

Marvell, T.B., and C.E. Moody, Jr. (1994). "Prison
Population Growth and Crime Reduction." Journal of
Quantitative Criminology, Vol. 1(2), pp. 109-140.

Mathur, V. (1978). "Economics of Crime: An
Investigation of the Deterrent Hypothesis for Urban
Crime." Review of Economics and Statistics, Vol.
60, pp. 446-459.

Miller, T.R., M.A. Cohen, and S.B. Rossman. (1993).
"Victim Costs of Violent Crime and Resulting
Injuries." Health Affairs (Data Watch) Winter, pp.
186-197.

Myers, S. (1982). "Crime in Urban Areas: New
Evidence and Results." Journal of Urban Economics,
Vol. 11, pp. 148-158.

_________. (1980). "Why Are Crimes Underreported?
What Is the Crime Rate? Does It Really Matter?"
Social Science Quarterly, Vol. 61(1), pp. 23-43.

Nagin, D. (1981). "Methodological Issues in
Estimating the Deterrent Effect of Sanctions." In
Models in Quantitative Criminology, J. Fox (ed.).
New York: Academic Press, pp. 95-139. 

Sjoquist, D. (1973). "Property Crime and Economic
Behavior: Some Empirical Results." American
Economic Review, Vol. 63, pp. 439-446.

Wahlroos, B. (1981). "On Finnish Property
Criminality: An Empirical Analysis of the Post War
Era Using an Ehrlich Model." Scandinavian Journal
of Economics, Vol. 83, pp. 553-562.

Wolpin, K. (1980). "A Time Series-Cross Section
Analysis of International Variations in Crime and
Punishment." Review of Economics and Statistics,
Vol. 62, pp. 417-423.

__________. (1978). "An Economic Analysis of Crime
and Punishment in England and Wales, 1894/1967."
Journal of Political Economy, Vol. 86, pp. 815-840.

Zedlewski, E.W. (1985). "When Have We Punished
Enough?" Public Administration Review, Vol. 45, pp.
771-779.

------------------------------

Appendix:

Data Sources and Estimates for Tables 
1-1, 1-3, 1-5, and 3-1

Table 1-1

Victimization 1994
 
Data Source: U.S. Department of Justice, Bureau of
Justice Statistics, Criminal Victimization 1994,
Bulletin (Washington, D.C., April 1996), table 1:
"Criminal Victimizations and Victimization Rates,
1993-94: Estimates From the Redesigned National
Crime Victimization Survey." NCJ 158022.

Reported Crimes 1994

Data Source: U.S. Department of Justice, Federal
Bureau of Investigation, Crime in the United
States, 1994 (Washington, D.C., 1995), table 1:
"Index of Crime, United States, 1975-1994." NCJ
153022.

Arrests 1992
 
Data Source: U.S. Department of Justice, Federal
Bureau of Investigation, Crime in the United
States, 1992 (Washington, D.C., 1993), table 29:
"Total Estimated Arrests, United States, 1992." NCJ
144768.

Arrests and Cleared by Arrest 1994

Data Source: U.S. Department of Justice, Federal
Bureau of Investigation, Crime in the United
States, 1994 (Washington, D.C., 1995), table 29:
"Total Estimated Arrests, United States, 1994," and
table 25: "Offenses Known and Percent Cleared by
Arrest, Population Group, 1994." NCJ 158553.

Convictions (Felony) 1992
 
Data Source: U.S. Department of Justice, Bureau of
Justice Statistics, Felony Sentences in the United
States, 1992, Bulletin (Washington, D.C.: U.S.
Department of Justice, 1996), table 2: "Number of
Felony Convictions in State and Federal Courts,
1992"; column 2: "Felony Convictions--State." NCJ
153257.

Convictions (Felony) 1994
 
Derivation of Estimate: Estimated Convictions
(Felony) 1994 = Convictions (Felony) 1992
multiplied by the result of dividing Total Index
Arrests 1994 by Total Index Arrests 1992.

Prisoners 1994
 
Data Source: U.S. Department of Justice, Bureau of
Justice Statistics, Felony Sentences in the United
States, 1992, Bulletin (Washington, D.C., 1995),
table 4: "Type of State and Federal Felony
Sentences, by Violent Offenses, 1992," column 3:
"Percent of Felons Sentenced to Incarceration--
Prison." NCJ 153257.

Derivation of Estimate: Prisoners 1994 =
Convictions (Felony) 1994 multiplied by the
percentage of felons sentenced to prison in 1992.

Table 1-3

Charged With Index Crime
 
Data Source: U.S. Department of Justice, Bureau of
Justice Statistics, Felony Defendants in Large
Urban Counties, 1992 (Washington, D.C., 1995),
table 22: "Conviction Offense of
Defendants Arrested for a Violent Offense and
Subsequently Convicted, by Most Serious Arrest
Charge, 1992," and table 23: "Conviction Offense of
Defendants Arrested for a Nonviolent Offense and
Subsequently Convicted, by Most Serious Arrest
Charge, 1992." NCJ 148826.

Table 1-5

Average Time Served in Prison

Data Source: U.S. Department of Justice, Bureau of
Justice Statistics, National Corrections Reporting
Program, 1992, NCJ 145862 (Washington, D.C., 1994),
table 2-4: "State Prison Releases, 1992: Time
Served in Prison, by Offense and Release Type."

Derivation of Estimate: Average Time Served in
Prison = weighted average of mean time served by
first releases and subsequent releases for given
offense. NCJ 145862.

Average Time To Be Served in Prison 

Data Source: U.S. Department of Justice, Bureau of
Justice Statistics, Felony Sentences in the United
States, 1992, Bulletin (Washington, D.C., 1996),
table 11: "Corresponding Time To Be Served in State
and Federal Prison, by Offense, 1992," column 9:
"Corresponding Time To Be Served in Prison--State."
NCJ 153257. 

Table 3-1

Cost per Crime
 
Data Source: Levitt (1996), table VIII: "Estimated
Impact on Crime From Adding One Additional
Prisoner" and "Cost per Crime, Adjusted to 1995
Dollars (CPI95/CPI92 = 1.082)."

Note: NCJ numbers listed above identify documents
available from the National Criminal Justice
Reference Service (see inside back cover).

------------------------------

For more information on the National Institute of
Justice, please contact:

National Criminal Justice Reference Service
P.O. Box 6000
Rockville, MD 20849-6000
800-851-3420
e-mail: askncjrs@ncjrs.org

You can view or obtain an electronic version of
this document from  the NCJRS Justice Information
Center World Wide Web site at http://www.ncjrs.org

If you have any questions, call or e-mail NCJRS.