Federal and State Prisons: Inmate Populations, Costs, and Projection
Models (Letter Report, 11/25/96, GAO/GGD-97-15).
GAO reviewed the trends in U.S. prison inmate populations and operating
and capital costs since 1980, including projections for 2000 and beyond
and the reasons for the trends and the models and methodologies used by
federal and state corrections agencies and nongovernmental forecasting
organizations to make these projections.
GAO found that: (1) the total U.S. prison population grew from about
329,800 inmates in 1980 to about 1.1 million inmates in 1995, which is
an increase of about 242 percent; (2) during this period, the federal
inmate population grew about 311 percent, and the inmate populations
under the jurisdiction of state prisons grew about 237 percent; (3) the
corresponding average annual growth rates were 9.9 percent of federal
populations and 8.4 percent for state populations; (4) in June 1996, the
Bureau of Prisons (BOP) projected that the federal prison population
could reach about 125,000 inmates by 2000, an increase of 25 percent
over the 1995 level; (5) in July 1995, the National Council on Crime and
Delinquency (NCCD) projected that the total federal and state prison
population under sentencing policies in effect in 1994 could reach 1.4
million inmates by 2000, representing an increase of about 24 percent
over the 1995 level; (6) in recent years, inmate population growth can
be traced in large part to major legislative initiatives that are
intended to get tough on crime, particularly on drug offenders; (7) U.S.
prison annual operating costs grew from about $3.1 billion in fiscal
year (FY) 1980 to about $17.7 billion in current dollars in FY 1994; (8)
BOP projected that its capital costs for new federal prisons scheduled
to begin operations during fiscal years 1996 to 2006 could total about
$4 billion; (9) BOP, NCCD, California, and Texas each use a form of
microsimulation modeling to forecast prison inmate populations; and (10)
according to BOP, its projections of federal prison inmate populations
for 1991 to 1995 were within 1.4 percent, on average, of the actual
populations.
--------------------------- Indexing Terms -----------------------------
REPORTNUM: GGD-97-15
TITLE: Federal and State Prisons: Inmate Populations, Costs, and
Projection Models
DATE: 11/25/96
SUBJECT: Correctional facilities
Prisoners
Administrative costs
Future budget projections
Computer modeling
Convictions
Law enforcement
Statistical data
Statistical methods
Cost analysis
IDENTIFIER: California
Texas
******************************************************************
** This file contains an ASCII representation of the text of a **
** GAO report. Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved. Major **
** divisions and subdivisions of the text, such as Chapters, **
** Sections, and Appendixes, are identified by double and **
** single lines. The numbers on the right end of these lines **
** indicate the position of each of the subsections in the **
** document outline. These numbers do NOT correspond with the **
** page numbers of the printed product. **
** **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced. Tables are included, but **
** may not resemble those in the printed version. **
** **
** Please see the PDF (Portable Document Format) file, when **
** available, for a complete electronic file of the printed **
** document's contents. **
** **
** A printed copy of this report may be obtained from the GAO **
** Document Distribution Center. For further details, please **
** send an e-mail message to: **
** **
** <info@www.gao.gov> **
** **
** with the message 'info' in the body. **
******************************************************************
Cover
================================================================ COVER
Report to the Subcommittee on Crime, Committee on the Judiciary,
House of Representatives
November 1996
FEDERAL AND STATE PRISONS - INMATE
POPULATIONS, COSTS, AND PROJECTION
MODELS
GAO/GGD-97-15
Inmate Populations, Costs, Projection Models
(182831)
Abbreviations
=============================================================== ABBREV
BJS - Bureau of Justice Statistics
BOP - Bureau of Prisons
FEDSIM - Federal Sentencing Simulation model
NCCD - National Council on Crime and Delinquency
JUSTICE - x
Letter
=============================================================== LETTER
B-272244
November 25, 1996
The Honorable Bill McCollum
Chairman
The Honorable Charles E. Schumer
Ranking Minority Member
Subcommittee on Crime
Committee on the Judiciary
House of Representatives
This report focuses on trends in U.S. prison inmate populations and
costs, and it is intended to assist Congress and the administration
in considering the implications of sentencing policies and law
enforcement initiatives. Our specific objectives in initiating this
review were to identify (1) the trends in federal and state prison
inmate populations and operating and capital costs since 1980,
including projections for 2000 and beyond and the reasons for the
trends and (2) the models and methodologies used by federal and state
corrections agencies and nongovernmental forecasting organizations to
make these projections, including whether any validity or reliability
assessments have been conducted. Although you did not request this
review, we addressed this report to you, as agreed, because of your
interest in prison issues, as exemplified by the hearings you held on
June 8, 1995.
To accomplish our objectives, we obtained and reviewed relevant
population, cost, and modeling information from the Federal Bureau of
Prisons (BOP), the National Council on Crime and Delinquency (NCCD),
and corrections agencies in the two states with the largest numbers
of inmates (California and Texas). Also, we relied extensively on
data compiled by the federal Bureau of Justice Statistics (BJS) and
the U.S. Bureau of the Census, and data in private publications,
such as the Corrections Compendium, which is a journal published by
CEGA Publishing in Lincoln, NE. We did not independently verify the
accuracy of the prison inmate population and cost data. Our
objectives, scope, and methodology are discussed in more detail in
appendix I.
We performed our work from May 1996 to September 1996 in accordance
with generally accepted government auditing standards.
RESULTS IN BRIEF
------------------------------------------------------------ Letter :1
From 1980 to 1995, the latest year for which complete data were
available, the total federal and state\1 prison population grew at an
average annual rate of 8.5 percent. As figure 1 shows, the total
prison population grew from about 329,800 inmates in 1980 to about
1.1 million inmates in 1995, which is an increase of about 242
percent. During this period, the federal inmate population grew
about 311 percent, and the inmate populations under the jurisdiction
of state prisons grew about 237 percent. The corresponding average
annual growth rates were 9.9 percent for federal populations and 8.4
percent for state populations.
Figure 1: Trends in Prison
Inmate Populations, 1980
Through 1995
(See figure in printed
edition.)
Source: BJS.
Federal and state corrections agencies--as well as nongovernmental
forecasting organizations, such as NCCD--have projected that the
prison population will continue to grow in future years. For
example, in June 1996, BOP projected that the federal prison
population could reach about 125,000 inmates by 2000, an increase of
25 percent over the 1995 level. In July 1995, NCCD projected that
the total federal and state prison population under sentencing
policies in effect in 1994 could reach 1.4 million inmates by 2000,
representing an increase of about 24 percent over the 1995 level.
The size of the prison population is a function of many factors,
including the nation's crime levels, sentencing laws, and law
enforcement policies. In recent years, inmate population growth can
be traced in large part to major legislative initiatives that are
intended to get tough on crime, particularly on drug offenders.
Examples of this new get-tough policy include mandatory minimum
sentences and repeat offender provisions.
Reflecting the growth in inmate populations, U.S. prison (federal
and state) annual operating costs grew from about $3.1 billion in
fiscal year 1980 to about $17.7 billion in current dollars in fiscal
year 1994. All prison costs (operating and capital costs)
cumulatively totaled about $163 billion\2 during the fiscal years
1980 to 1994 period. The corresponding average annual growth rate
for this period was 9.1 percent in inflation-adjusted terms. In June
1996, BOP projected that its prison operating costs could total about
$3.6 billion in fiscal year 2000, which is an increase of about 88
percent over the fiscal year 1994 level. BOP also projected that its
capital costs for new federal prisons scheduled to begin operations
during fiscal years 1996 to 2006 could total about $4 billion.
Regarding the states' needs, NCCD has estimated that $10.6 billion to
$15.1 billion could be needed to construct additional prisons to
accommodate anticipated inmate population increases from 1995 to 2000
and that $21.9 billion could be needed by the end of the decade to
operate these prisons.
To forecast prison inmate populations, BOP, NCCD, California, and
Texas each use a form of microsimulation modeling. Microsimulation
provides the flexibility to adjust assumptions and data in response
to new sentencing laws or policies and other criminal justice or law
enforcement initiatives that could affect the size of prison
populations.\3 Except for BOP's projection model, on the basis of a
literature search and discussions with federal and state agency
officials, we did not identify any independent assessments of the
various projection models' validity or reliability. Generally,
officials commented that projections beyond 5 years, and perhaps even
beyond 2 years, are usually considered rough estimates. However, BOP
and NCCD have tracked the accuracy of their respective projections.
According to BOP, its projections of federal prison inmate
populations for 1991 to 1995 were within 1.4 percent (on average) of
the actual populations. Also, according to NCCD, its projections for
1991 through 1994 were within 2 percent (on average) of the actual
populations.
--------------------
\1 State data include all 50 states and the District of Columbia.
\2 See tables III.1, III.2, and III.3 in appendix III.
\3 The various agencies or organizations do not use modeling to
project operating and capital costs. Rather, costs generally are
projected linearly using historical cost data (e.g., per-inmate
operating costs and per-bed construction costs) adjusted for
inflation and projected population data.
BACKGROUND
------------------------------------------------------------ Letter :2
In recent years, Gallup opinion polls have indicated that the
American public is concerned about crime and related violence. For
example, in a 1995 poll, 27 percent of the respondents listed crime
and violence as the most important problems facing the country.
Polls also have suggested that tougher anticrime legislation is a top
priority for the public. For instance, 80 percent of the respondents
to a 1996 Gallup Poll supported life sentences for drug dealers.
Congress has authorized grants to the states that support tougher
sentencing policies for criminals and expanded prison construction to
house the growing number of inmates. For example, in the Department
of Justice's 1996 appropriations,\4 Congress authorized about $10.3
billion in grants to states for fiscal years 1996 through 2000 for,
among other things, building or expanding correctional facilities to
house persons convicted of violent crimes.
According to BJS' National Crime Victimization Survey (April 1996),\5
there were 51 violent victimizations\6 per 1,000 U.S. residents in
1994, which was the latest year that complete data were available.
Since its inception in 1973, the survey has determined that crime
rates and levels have fluctuated over extended periods.
Specifically, violent crime rates leveled off between 1992 and 1994,
compared with a 20-percent decline between 1981 and 1986 and a
15-percent rise between 1986 and 1991. Property crime\7 continued a
general 15-year decline. The survey did not provide any reasons for
the fluctuations in crime rates and levels. Even though crime rates
have fluctuated, overall crime rates in the 1990s remain
substantially higher than those in the 1960s. For example, according
to Uniform Crime Reports data compiled by the Federal Bureau of
Investigation, the nation's overall crime rate was about 2,000 crimes
per 100,000 residents in the early 1960s compared with 5,374 crimes
per 100,000 residents in 1994.
Against the backdrop of these higher crime rates, there is a
continuing debate over the use of incarceration as a means of
addressing increasing crime. Both proponents and opponents of
increasing the use of incarceration as a solution to the crime
problem can cite research to support their positions. For example,
proponents of increased incarceration assert that investing in new
prisons will have long-term benefits of crime reduction.\8 On the
other hand, critics of increased incarceration argue that continued
prison-building is wasteful and unaffordable and is unlikely to
affect crime rates.\9
In 1994, RAND issued a study of California's "three strikes" law,
which mandates sentences ranging from 25 years to life for certain
three-time felony offenders.\10 The study, which weighed crime
reduction and cost, concluded that the California law, if fully
implemented, will reduce serious felonies committed by adults in the
state by between 22 and 34 percent below what may have occurred. The
study also concluded that the reduction in crime would be achieved at
an additional cost of between $4.5 billion and $6.5 billion in
current dollars annually. According to the study, most of the cost
increase would result from the need to build and operate additional
prisons to house the inmate population, which could be expected to
double as a result of sentencing under the law. A more recent RAND
study\11 indicates that some preventative measures, such as parent
training and graduation incentives, could potentially reduce crime
rates more cost effectively than incarceration.
--------------------
\4 Public Law 104-134, April 26, 1996.
\5 Conducted in 1994, the survey measured personal and household
offenses, including crimes not reported to police, by interviewing
all occupants age 12 or older in a nationally representative sample
of U.S. households, including persons living in group quarters, such
as dormitories. In total, approximately 120,000 residents in 56,000
housing units were interviewed about the crimes they had experienced
in the previous 6 months.
\6 Violent crimes included rape/sexual assault, robbery, aggravated
and simple assault as measured by the survey, and murder as reported
to the Federal Bureau of Investigation.
\7 The survey defined property crimes as household burglary, motor
vehicle theft, and thefts of other property.
\8 John J. DiIulio, Jr., "Prisons Are A Bargain, by Any Measure,"
The Wall Street Journal, January 16, 1996. Mr. DiIulio, a professor
of politics and public affairs at Princeton University, also is one
of the authors of The State of Violent Crime in America (Council on
Crime in America: Washington, D.C.), January 1996.
\9 For example, see the testimony of Todd R. Clear presented in
hearings on the "Federal Prison Population: Present and Future
Trends" before the House Subcommittee on Intellectual Property and
Judicial Administration, Committee on the Judiciary, 103d Cong., 1st
Sess. 80-98 (1993).
\10 Peter W. Greenwood, et al, RAND, Three Strikes and You're Out:
Estimated Benefits and Costs of California's New Mandatory Sentencing
Law (RAND/MR-509-RC), 1994.
\11 Peter W. Greenwood, et al, RAND, Diverting Children from a Life
of Crime: What Are the Costs and Benefits? (RAND/MR-699-UCB/RC/IF),
1996.
TRENDS IN FEDERAL AND STATE
PRISON INMATE POPULATIONS AND
COSTS
------------------------------------------------------------ Letter :3
Federal and state prison inmate populations have been growing since
1980, reaching about 1.1 million inmates in 1995. Federal and state
corrections agencies and nongovernmental forecasting organizations
project that these populations will continue to grow, potentially
reaching 1.4 million inmates in 2000. Prison operating and capital
costs\12 have also been growing and are projected to continue doing
so in the future. For federal and state prisons, operating and
capital costs cumulatively totaled about $163 billion\13 for fiscal
years 1980 through 1994.
--------------------
\12 As defined by the Census Bureau--our source for cost
data--"costs," or "direct expenditures," are payments to employees,
suppliers, contractors, beneficiaries, and other final recipients of
government payments. Expenditures are net of recoveries and other
correcting transactions and exclude retirement of debt, investment in
securities, extensions of credit, and agency transactions.
\13 See table III.3 in appendix III.
PRISON INMATE POPULATIONS
(1980 THROUGH 1995)
---------------------------------------------------------- Letter :3.1
From 1980 to 1995, which was the latest year that complete data were
available, the total U.S. prison inmate population under federal and
state jurisdiction\14 grew by about 242 percent, from 329,821 to
1,127,132,\15 respectively. The corresponding average annual prison
population growth rate during this period was 8.5 percent (9.9
percent for the federal population and 8.4 percent for the state
populations). The prison population increased at a slower rate--6.8
percent--between 1994 and 1995 than the average growth rate.
Although an August 1996 BJS report\16 on prison and jail inmates--the
source for our prison inmate population data--did not provide
specific reasons for the decrease in the rate of growth, a BJS
official commented that the smaller growth rate may be the result of
the growing population base, currently over 1 million inmates.
Nevertheless, we do not know whether this marks the start of a trend
toward smaller rates of growth.
As previously shown in figure 1, the federal prison inmate population
grew from 24,363 in 1980 to 100,250 in 1995, which is an increase of
about 312 percent. The state prison population grew from 305,458
inmates in 1980 to 1,026,882 inmates in 1995, which is an increase of
about 236 percent. In California, the state prison inmate population
grew from 24,569 in 1980 to 135,646 in 1995, which is an increase of
about 452 percent. In Texas, the inmate population grew from 29,892
in 1980 to 127,766 in 1995, which is an increase of about 327
percent. Not all states exhibited inmate population increases to
such an extent. For example, in Maine, the inmate population grew
from 814 in 1980 to 1,447 in 1995, which is an increase of about 78
percent. In North Carolina, the inmate population grew from 15,513
in 1980 to 29,374 in 1995, which is an increase of about 89 percent.
--------------------
\14 According to BJS, this population includes inmates serving
sentences of more than 1 year ("sentenced prisoners"); those with
sentences of a year or less; and those with no sentences (e.g., those
awaiting trial in states with combined jail/prison systems). These
inmates are under the jurisdiction of correctional authorities in the
50 states, the District of Columbia, and the federal government.
Inmates under jurisdiction include persons under the legal authority
of a prison system held elsewhere or outside its facilities.
\15 According to BJS, this is an advance count of prisoners under
federal and state jurisdiction for 1995, and it may be revised.
\16 BJS, Prison and Jail Inmates, 1995 (NCJ-161132), August 1996.
INMATE INCARCERATION RATES
(1980 THROUGH 1995)
---------------------------------------------------------- Letter :3.2
Corresponding to the growth in prison populations, the incarceration
rates\17 for federal and state prison inmates have also shown steady
growth during the 16-year period of 1980 through 1995. As figure 2
shows, the total incarceration rate grew from 145 inmates in 1980 to
428 inmates in 1995 for every 100,000 U.S. residents, which is an
increase of about 195 percent. Reflecting an even larger percentage
increase (about 245 percent), the incarceration rate for federal
inmates grew from 11 inmates for every 100,000 residents in 1980 to
38 inmates for every 100,000 residents in 1995.
Because most prisoners are under state jurisdiction, the
incarceration rate for state inmates closely follows (and, indeed, is
largely determinative of) the nation's total incarceration rate.
Specifically, the incarceration rate for states grew from 134 inmates
for every 100,000 residents in 1980 to 390 inmates for every 100,000
residents in 1995, which is an increase of about 191 percent. During
this period, the incarceration rate in California increased 312
percent, growing from 104 inmates for every 100,000 residents in 1980
to 428 inmates for every 100,000 residents in 1995. The
incarceration rate in Texas increased 222 percent, growing from 210
inmates to 677 inmates for every 100,000 residents in 1980 and 1995,
respectively.
Figure 2: Trends in
Incarceration Rates, 1980
Through 1995
(See figure in printed
edition.)
Source: BJS.
--------------------
\17 The rate of incarceration is the total number of prisoners in
correctional facilities per 100,000 U.S. resident population.
REASONS FOR PRISON
POPULATION GROWTH
---------------------------------------------------------- Letter :3.3
According to various sources, including BJS and the U.S. Sentencing
Commission, the significant growth in federal and state inmate
populations since the 1980s is largely the result of changes in
criminal behavior, law enforcement practices, sentencing law and
policy, and release practices. For example, according to BJS, during
the 1980s, an increasing number of probation and parole violators
returned to prison, while in the 1990s, declining rates of release
have sustained the growth in inmate populations.
More specifically, regarding federal offenders, under the Sentencing
Reform Act of 1984,\18 parole was abolished, and good-time credits
(time off sentence for good behavior) were limited to 54 days per
year. In 1986, the Anti-Drug Abuse Act\19 established mandatory
minimum sentences for certain drug offenses. In 1988 and 1990,
Congress passed additional sentencing legislation, which increased
mandatory minimum sentences for drug and weapons offenses. As a
result of these statutory changes, the use of probation has been
reduced and the length of prison stays has increased. According to
BJS data, after 1986, the average time served in federal prisons
increased from 15 months to 24 months. For violent offenses, the
time served increased from 50 months to 56 months, and the time
served for drug offenses increased from 22 months to 33 months.\20
Particularly noteworthy has been the trend regarding drug offenders
as a percentage of the total inmate population. According to a 1991
study by the U.S. Sentencing Commission, drug offenders constituted
about 91 percent of all federal defendants sentenced under mandatory
minimum provisions.\21 According to BJS, in 1993, which was the
latest year that complete data were available, drug offenders
constituted 26 percent of all federal and state inmates,\22
whereas these offenders constituted 8 percent of all inmates in 1980.
Also, BJS has reported that the increase in drug offenders accounts
for nearly three-fourths of the total growth in federal prison
inmates since 1980.
The state prison inmate populations have grown as a result of, among
other things, the increased number of arrests, higher probabilities
of incarceration, and more severe sanctions. Specifically, according
to BJS, the number of arrests increased by 41 percent between 1980
and 1993, the latest year that complete data were available. The
rate of sending offenders to prison also increased. For example, the
likelihood of incarceration increased 5-fold for drug violations and
4-fold for weapons offenses. According to the California Department
of Corrections,\23 the prison population in that state has grown
because of court decisions, voter initiatives, and legislation, all
of which have resulted in stronger law enforcement and more severe
criminal sanctions. For example, a California law prohibits the use
of good-time allowances to reduce the sentences of repeat offenders
convicted of certain violent felonies. State corrections officials
expected that the law may result in inmates' serving additional time,
which could lead to an increase in the state's prison population in
future years.
--------------------
\18 Public Law 98-473, October 12, 1993.
\19 Public Law 99-570, October 27, 1986.
\20 BJS, Prisoners in 1994 (NCJ-151654), August 1995.
\21 U.S. Sentencing Commission, Special Report to Congress:
Mandatory Minimum Penalties in the Federal Criminal Justice System,
August 1991.
\22 In 1993, drug offenders constituted 60 percent of all federal
inmates and 22 percent of all state inmates.
\23 The California Department of Corrections is responsible for,
among other things, the incarceration of adult felons and nonfelon
narcotics addicts. The Department also supervises and treats
parolees released to the community.
PRISON POPULATION
PROJECTIONS FOR 2000
---------------------------------------------------------- Letter :3.4
While sources differed somewhat in their projected growth for federal
and state prison inmate populations, they all showed substantial
anticipated increases for these populations in 2000 and beyond. In
June 1996, BOP projected that the federal prison population could
reach about 125,000 inmates by 2000, which is a 25-percent increase
over the 1995 level (see table 1). In July 1995, NCCD projected
that, under sentencing policies in effect in 1994, the total inmate
population for federal and state prisons could reach 1.4 million by
2000, which is an increase of 24 percent over the 1995 level. NCCD
also projected that, if all states were to adopt truth-in-sentencing
statutes, which would require inmates to serve at least 85 percent of
their sentences, the states' prison population could grow by an
additional 190,000 inmates and total about 1.6 million inmates by
2000, which would be an increase of about 42 percent over the 1995
level.
Table 1
Projected Prison Populations in 2000
Total Percentage
projected increase
population from 1995
Prison system Source of projection in 2000 to 2000
---------------------- ---------------------- ---------- ----------
Federal BOP 125,144 25
Federal and states NCCD 1,400,000 24
Federal and states Corrections Compendium 1,341,832 19
(April 1996)
California California Department 203,593 50
of
Corrections
Texas Texas Criminal 143,748 13
Justice Policy
Council
----------------------------------------------------------------------
Sources: As indicated in table.
The April 1996 issue of the Corrections Compendium presented a
compilation of inmate population projections that were based on a
survey of federal and state corrections agencies. The combined
self-reported projections showed that the federal and state prison
population could reach over 1.3 million in 2000, representing an
increase of 19 percent over the 1995 level. However, the survey
summary in the Compendium indicated that this total may be
understated. According to the summary, if the historical growth rate
(8.7 percent per year from 1980 through 1994) continues in future
years, the prison population could actually increase by 95 percent
over the 1994 level, essentially doubling to about 2 million inmates
by 2002.
In July 1995, NCCD projected that the inmate population in California
could reach about 210,000 by 2000, which would be an increase of 55
percent over the 1995 level. Separately, in spring 1996, the
California Department of Corrections projected that the state's
prison population could reach 203,593 inmates in 2000, which would be
an increase of about 50 percent over the 1995 level. For Texas, in
July 1995, NCCD projected that the inmate population could reach
about 149,000 by 2000, which would be an increase of about 17 percent
over the 1995 level. In September 1996, the Texas Criminal Justice
Policy Council projected that the state's prison population could
reach 143,748 in 2000, which would be an increase of about 13 percent
over the 1995 level.\24
Appendix II presents additional information about actual and
projected federal and state prison inmate populations and
incarceration rates.
--------------------
\24 According to the Council's September 1996 correctional population
projection report, this projection was revised downward from earlier
versions to account for the reduced growth in Texas' correctional
population. According to the report, the reduced growth has resulted
from a decline in reported crime, a stabilization in the number of
new convictions, a decline in parole revocations, and a decline in
the number of offenders sentenced to prison.
PRISON OPERATING COSTS
(FISCAL YEARS 1980 THROUGH
1994)
---------------------------------------------------------- Letter :3.5
Prison operating costs\25 grew steadily during fiscal years 1980 to
1994, reflecting in part the growth in prison populations. As figure
3 shows, total U.S. prison operating costs grew from about $3.1
billion in fiscal year 1980 to about $17.7 billion in current dollars
in fiscal year 1994. This is an increase of 224 percent based on
constant or inflation-adjusted dollars.\26 Of this total, federal
prison operating costs grew from about $319 million in fiscal year
1980 to about $1.9 billion in fiscal year 1994, which is an increase
of about 242 percent based on constant dollars. The corresponding
average annual growth rate during this period was 9.9 percent. State
prison operating costs grew from about $2.8 billion in fiscal year
1980 to $15.8 billion in fiscal year 1994, which is an increase of
222 percent based on constant dollars. The corresponding average
annual growth rate during this period was 8.7 percent. In
California, operating costs grew from about $320 million in fiscal
year 1980 to $2.6 billion in fiscal year 1994, which is an increase
of 357 percent based on constant dollars. In Texas, operating costs
grew from about $105 million in fiscal year 1980 to $1.2 billion in
fiscal year 1994, which is an increase of 529 percent based on
constant dollars.
Figure 3: Trends in Prison
Operating Costs, Fiscal Years
1980 Through 1994
(See figure in printed
edition.)
Note: Dollar figures represent actual dollars (no adjustment for
inflation).
Source: Table III.1 in appendix III.
--------------------
\25 As defined by the Census Bureau, prison operating costs include
compensation of officers and employees, supplies, materials,
operating leases, and contractual services. Operating costs also
include repairs to existing works and structures.
\26 In this report, except for projections or future trends,
percentage increases involving costs are based upon the conversion of
current dollars to constant or inflation-adjusted dollars.
Specifically, in calculating percentage increases for operating and
capital costs during past years, we used annual implicit price
deflators (as published in the U.S. Department of Commerce's Survey
of Current Business) to adjust all dollars to 1994 purchasing power.
PRISON CAPITAL COSTS (FISCAL
YEARS 1980 THROUGH 1994)
---------------------------------------------------------- Letter :3.6
Prison capital costs,\27 while growing overall, have actually
fluctuated on almost a year-to-year basis during fiscal years 1980 to
1994. As figure 4 shows, total U.S. prison capital costs grew from
about $538 million in fiscal year 1980 to about $2.3 billion in
current dollars in fiscal year 1994. This is an increase of 141
percent based on constant or inflation-adjusted dollars. Federal
prison capital costs grew from about $22 million in fiscal year 1980
to about $312 million in fiscal year 1994, representing an increase
of about 715 percent (based on constant dollars). The corresponding
average annual growth rate during this period was 87.9 percent.\28
State prison capital costs grew from about $516 million in fiscal
year 1980 to about $2 billion in fiscal year 1994, representing an
increase of about 116 percent based on constant dollars. The
corresponding average annual growth rate during this period was 7.4
percent. In California, capital costs grew from $16 million in
fiscal year 1980 to $413 million in fiscal year 1994, which is an
increase of 1,327 percent based on constant dollars. In Texas,
capital costs grew from $20 million in fiscal year 1980 to about $577
million in fiscal year 1994, which is an increase of 1,531 percent
based on constant dollars.
Figure 4: Trends in Prison
Capital Costs, Fiscal Years
1980 Through 1994
(See figure in printed
edition.)
Note: Dollar figures represent actual dollars (no adjustment for
inflation).
Source: Table III.2 in appendix III.
--------------------
\27 As defined by the Census Bureau, prison capital costs are the
direct expenditures for contract or forced account construction of
buildings, roads, and other improvements, purchase of equipment,
land, and existing structures, and payment of capital leases.
Capital costs also include expenditures for additions, replacements,
and major alterations to fixed works and structures, but exclude
repairs to such works and structures. Repairs are treated as
operating expenditures.
\28 The average annual growth rate is distorted by large increases in
certain fiscal years. For example, in fiscal year 1983, federal
capital costs increased by almost 700 percent over the level in
fiscal year 1982. Also, in fiscal year 1990, capital costs increased
by about 440 percent over the level in fiscal year 1989.
PROJECTIONS OF PRISON
OPERATING AND CAPITAL COSTS
---------------------------------------------------------- Letter :3.7
BOP estimates federal prison operating costs to grow through fiscal
2000, while capital costs are expected to fluctuate on a year-by-year
basis. Specifically, in June 1996, BOP projected that its operating
costs could grow to about $3.6 billion by fiscal year 2000, almost
double the level in fiscal year 1994. BOP also projected that its
capital costs for new federal prisons scheduled to begin operations
during fiscal years 1996 to 2006 could total about $4 billion.
According to Justice officials, these cost increases were projected
on the basis of historically high rates of prison population
increases. According to these officials, since recent BJS statistics
show that the rate of increase in prison populations from 1994 to
1995 was below the average for the preceding 5 years, the BOP cost
projections for 2000 and beyond may be overestimated.
NCCD has estimated that state prison population increases from 1995
to 2000 could result in total additional capital and operating costs
of $32.5 billion to $37 billion for this period. Specifically, NCCD
estimated that $10.6 billion to $15.1 billion could be needed to
construct additional state prisons, and that $21.9 billion could be
needed by the end of the decade to operate these prisons.
Appendix III presents additional information about actual and
projected federal and state operating and capital costs.
MODELS AND METHODOLOGIES USED
TO PROJECT PRISON POPULATIONS
------------------------------------------------------------ Letter :4
BOP, NCCD, California, and Texas each use microsimulation models to
project prison inmate populations.\29 The models are similar in
providing flexibility to adjust assumptions and data in response to
new sentencing laws or policies and other criminal justice or law
enforcement initiatives that could affect the size of prison
populations in the respective jurisdiction. Appendix IV provides
more detailed information about microsimulation and other models and
methodologies used to project inmate populations.
On the basis of a literature search and discussions with federal and
state agency officials, we did not identify any independent
assessments of the various projection models' validity or
reliability, except for BOP's model. This model, according to BOP
officials, has been subjected to various reviews. For example, the
officials made the following comments:
-- In 1993, BOP staff published a paper (which was peer reviewed)
on the projection methodology.\30
-- Justice's budget staff annually reviews BOP's inmate population
projections and often reports to the Attorney General on the
accuracy of the projections.
Some of the forecasting organizations and state corrections agencies
have tracked and self-reported on the accuracy of their respective
projections. For example, according to BOP, its projections of
federal prison inmate populations for 1991 to 1995 were within 1.4
percent (on average) of the actual populations. Also, according to
NCCD, its projections for 1991 through 1994 were within 2 percent (on
average) of the actual populations.
However, BOP officials and a modeling expert said that assessing a
model's reliability by comparing projections with actual populations
is not necessarily the only approach. For instance, the officials
noted that after projections showing potential impacts are presented
or published, legislators or administrators are more likely to modify
or change certain policies or practices, taking the projections into
consideration. Thus, according to these officials, another benefit
of a population simulation is to inform the public policy debate.
The April 1996 issue of the Corrections Compendium presented the
results of a survey that asked respondents to report on the accuracy
of their models' population projections. The survey was originally
sent to federal and state corrections agencies in October 1995, and
the responses with the applicable data were collected through
February 1996. Of the 39 respondents\31 to this question, 54 percent
reported that their past projections were "accurate,"\32 23 percent
reported that their past projections were "too low," and 8 percent
said their past projections were "too high."\33 The other respondents
to the overall survey reported that they either did not project
populations or did not assess the accuracy of their projections.
--------------------
\29 As previously noted, the various agencies or organizations do not
use modeling to project operating and capital costs. Rather, costs
generally are projected linearly using historical cost data adjusted
for inflation and projected population data. For example, to project
its operating costs, BOP adjusts actual per-inmate costs by an annual
inflation factor of 3.1 percent.
\30 Gaes, G.G., Simon, E.S., and Rhodes, W.M., "20/20 Hindsight:
Effectiveness of Simulating the Impact of Federal Sentencing
Legislation on the Future Prison Population," The Prison Journal,
1993, pp. 5-29.
\31 The respondents were BOP, 37 states, and the District of
Columbia.
\32 For example, BOP, California, and Texas reported that their
projections were accurate.
\33 The survey did not define or specify a given time period for
"past" projections.
AGENCY COMMENTS AND OUR
EVALUATION
------------------------------------------------------------ Letter :5
We obtained comments on the draft of this report from Justice
officials, including the Director of Justice's Audit Liaison Office,
BOP's Chiefs of Research and Evaluation and Budget Development, and
BJS' Chief of Corrections Statistics. These officials generally
agreed with the contents of the draft report. However, BOP and BJS
officials provided technical comments and clarifications related to
certain numerical data in the report. Also, the BOP officials
provided revised federal prison inmate data. We have incorporated
these technical comments, clarifications, and revisions where
appropriate in this report.
Regarding prison costs, BJS officials expressed the view that actual
expenditure data compiled and published by the Census Bureau (e.g.,
Census of Government Finances) would be more accurate and complete
than data from The Corrections Yearbook, the source we used for the
draft of this report. Accordingly, from the Census Bureau, we
obtained state prison expenditure data for fiscal years 1980 through
1994 (the latest year that complete data were available), and we
incorporated this information and revised the related analyses in
this report where appropriate.
In our draft report, we noted that we did not identify any
independent assessments regarding the validity or reliability of the
various models used to project federal and state prison inmate
populations. However, in commenting on the draft report, BOP
officials called to our attention examples of various reviews or
evaluations that could be considered assessments of the Bureau's
microsimulation model. We incorporated BOP's comments and examples
in this report.
We also obtained comments on the draft of this report from NCCD's
Executive Vice President and a NCCD Senior Researcher. These
officials agreed with the contents of this report and stated that it
factually represented information and statistical data developed by
and previously published by NCCD. The officials also offered one
technical clarification, which we have incorporated in this report.
---------------------------------------------------------- Letter :5.1
We are providing copies of this report to the Attorney General; the
Assistant Attorney General, the Office of Justice Programs; the
Director, BOP; and other interested parties. Copies also will be
made available to others upon request. The major contributors to
this report are listed in appendix V. Please contact me on (202)
512-8777 if you or your staff have any questions.
Laurie E. Ekstrand
Associate Director, Administration
of Justice Issues
OBJECTIVES, SCOPE, AND METHODOLOGY
=========================================================== Appendix I
We initiated this review to identify (1) the trends in federal and
state prison inmate populations and operating and capital costs since
1980, including projections for 2000 and beyond and the reasons for
these trends\1 and (2) the models and methodologies used by federal
and state\2 corrections agencies and nongovernmental forecasting
organizations to make these projections, including whether any
validity or reliability assessments had been done.
To address these objectives, we initially conducted a literature
search to identify available data sources and to determine to what
extent these issues had received congressional attention. In the
latter regard, we noted that a Subcommittee of the House Committee on
the Judiciary held hearings in 1993\3 that were useful in our
analyses.
More specifically, to identify the trends in prison populations and
costs, we contacted relevant federal agencies, such as the Bureau of
Justice Statistics (BJS), the Federal Bureau of Prisons (BOP), and
the U.S. Bureau of the Census, and corrections agencies in the two
states with the largest prison populations (California and Texas).
BJS compiles and publishes considerable statistical information
covering both federal and state correctional systems. For example,
two relevant series of BJS publications are the Sourcebook of
Criminal Justice Statistics and Correctional Populations in the
United States. The Census Bureau compiles and publishes, among other
things, statistical information about federal and state government
expenditures. For example, relevant series of Census Bureau
publications are the Census of Government Finances and State
Government Finances.
From BOP, we obtained historical as well as projected data covering
both populations and costs for federal prisons. From the state
agencies, we obtained and reviewed historical and projected prison
inmate population data. State agency officials told us that prison
operating and capital costs generally are not projected beyond the
next fiscal year.
Furthermore, in identifying prison population and cost trends, we
also contacted nongovernmental sources, such as the National Council
on Crime and Delinquency (NCCD). As a private organization engaged
in research, training, and advocacy programs to reduce crime and
delinquency, NCCD has published several studies of prison-related
topics, including projections of inmate populations.\4 Also, another
useful nongovernmental source was the Corrections Compendium, which
is a journal from CEGA Publishing.
We discussed the population and cost data we obtained with cognizant
officials at the federal and state agencies and the nongovernmental
organizations. We did not independently verify the accuracy and
quality of the data we obtained.
To identify the models and methodologies used by federal and state
corrections agencies and nongovernmental organizations to make
projections, we obtained and reviewed modeling and methodology
information from BOP, NCCD, the Corrections Compendium, the
California Department of Corrections, and the Texas Criminal Justice
Policy Council.\5 We focused our review on BOP's Federal Sentencing
Simulation model, NCCD's Prophet model (used by 23 states in addition
to NCCD), and Texas' JUSTICE model (Texas has the second largest
prison inmate population).
To identify the extent, if any, to which the forecasting models and
methodologies had been assessed for validity and reliability, we
conducted a literature search.\6 Also, we interviewed officials in
BOP's Office of Research and Evaluation, which is responsible for,
among other things, forecasting federal inmate populations.
Similarly, we interviewed state corrections agency officials in
California and Texas. We discussed issues related to the models and
methodologies and their validity and reliability with cognizant
officials from BOP and NCCD and the author of a 1990 BJS-sponsored
study that reviewed (but did not evaluate) some of the projection
models used by federal and state criminal justice systems.\7
--------------------
\1 We have previously reported on issues related to prison expansion
and crowding. See Federal Prison Expansion: Overcrowding Reduced
but Inmate Population Growth May Raise Issue Again (GAO/GGD-94-48,
Dec. 14, 1993) and Prison Crowding: Issues Facing the Nation's
Prison Systems (GAO/GGD-90-1BR, Nov. 2, 1989).
\2 State data include all 50 states and the District of Columbia.
\3 "Federal Prison Population: Present and Future Trends," hearings
before the House Subcommittee on Intellectual Property and Judicial
Administration, Committee on the Judiciary, 103d Cong., 1st Sess.
80-98 (1993).
\4 For example, see Michael A. Jones and James Austin, "The NCCD
National Prison Population Forecast: The Cost of Truth-in-Sentencing
Laws," NCCD Focus (San Francisco, CA: National Council on Crime and
Delinquency), July 1995.
\5 The Council was created as an independent state agency in 1983 by
the Texas legislature. As mandated by state law, the Council's basic
operational activity includes making projections and impact
simulations regarding criminal justice policy alternatives.
\6 For example, the April 1996 issue of Corrections Compendium
presents the results of a survey on inmate population projections.
The survey asked respondents whether past projections had been
accurate.
\7 BJS, Models of the Criminal Justice System: A Review of Existing
Impact Models (NCJ-124011), prepared by William Rhodes, Abt
Associates, Inc. (Cambridge, MA), June 1990.
TRENDS IN FEDERAL AND STATE PRISON
INMATE POPULATIONS AND
INCARCERATION RATES
========================================================== Appendix II
Federal and state prison inmate populations--and corresponding
incarceration rates--have been growing since 1980. Federal and state
corrections agencies and nongovernmental forecasting organizations
project that these populations will continue to grow through 2000 and
beyond. Populations in the other three correctional
categories--probation, parole, and jail--have also grown since
1980.\1 However, in terms of the percentages of the overall adult
correctional population, the relative distribution of adult offenders
among the four categories were similar in 1994 and 1980.
--------------------
\1 The adult correctional population is the total number of adult
offenders under some form of supervision--prison, probation, parole,
or jail--by the criminal justice system.
PRISON INMATE POPULATIONS AND
INCARCERATION RATES (1980
THROUGH 1994)
-------------------------------------------------------- Appendix II:1
Table II.1 shows that the federal prison inmate population and the
corresponding incarceration rate have grown consistently from 1980 to
1995. By 1995, the prison population had grown 4-fold from the 1980
level, reaching over 100,000 inmates. The incarceration rate had
grown more than 3-fold, reaching 38 inmates for every 100,000 U.S.
residents in 1995.
Table II.1
Federal Prison Inmate Populations and
Incarceration Rates at Year-End, 1980
Through 1995
Inmates per
Popula 100,000 U.S.
Calendar year tion residents
---------------------------------------------- ------ --------------
1980 24,363 11
1981 28,133 12
1982 29,673 13
1983 31,926 14
1984 34,263 14
1985 40,223 17
1986 44,408 18
1987 48,300 20
1988 49,928 20
1989 59,171 24
1990 65,526 27
1991 71,608 28
1992 80,259 31
1993 89,587 34
1994 95,034 36
1995 100,25 38
0
======================================================================
Percentage change, 311.5% 245.4%
1980-1995
----------------------------------------------------------------------
Source: BJS.
Table II.2 shows the federal prison inmate population at fiscal
year-end. According to BOP, the population data presented are
different than BJS' data (presented in table II.1) in that, in
addition to being compiled by fiscal year rather than calendar year,
they represent inmates both in BOP facilities and alternative
confinements, such as contract facilities.
Table II.2
Federal Prison Inmate Populations at
Fiscal Year-End, 1980 Through 1995
Popula
Fiscal year tion
-------------------------------------------------------------- ------
1980 24,640
1981 26,313
1982 30,531
1983 33,216
1984 35,795
1985 40,330
1986 46,055
1987 49,378
1988 50,513
1989 57,762
1990 64,936
1991 71,508
1992 79,678
1993 88,565
1994 95,162
1995 100,95
8
======================================================================
Percentage 309.7%
change,
1980-1995
----------------------------------------------------------------------
Source: BOP.
As table II.3 shows, from 1980 to 1995, the state prison inmate
population and the corresponding incarceration rate grew by about 236
and about 191 percent, respectively. In 1995, the state prison
inmate population reached just over 1 million, compared with just
over 300,000 in 1980. The incarceration rate reached 390 inmates for
every 100,000 U.S. residents in 1995, compared with 134 in 1980.
From 1980 to 1995, the prison populations in California and Texas
grew by well over 400 percent and 300 percent, respectively.
Table II.3
State Prison Inmate Populations and
Incarceration Rates at Year-End, 1980
Through 1995
Inmates per
100,000 U.S.
All Califo residents, all
Calendar year states rnia Texas states
------------------------------ ------ ------ ------ --------------
1980 305,45 24,569 29,892 134
8
1981 341,79 29,202 31,502 148
7
1982 384,13 34,640 36,149 165
3
1983 404,92 39,373 35,259 172
9
1984 427,73 43,197 36,682 180
9
1985 462,28 50,158 37,532 193
4
1986 500,58 59,484 38,534 207
4
1987 536,78 66,975 38,821 218
4
1988 577,67 76,171 40,437 234
2
1989 653,19 87,297 44,022 260
3
1990 708,39 97,309 50,042 280
3
1991 753,95 101,80 51,677 295
1 8
1992 802,24 109,49 60,487 313
1 6
1993 879,71 119,95 92,013 331
4 1
1994 960,03 125,60 118,19 367
9 5 5
1995 1,026, 135,64 127,76 390
882 6 6
======================================================================
Percentage change, 236.2% 452.1% 327.4% 191.0%
1980-1995
----------------------------------------------------------------------
Source: BJS.
PROJECTED PRISON INMATE
POPULATIONS
-------------------------------------------------------- Appendix II:2
Table II.4 shows that the federal prison inmate population is
projected by BOP to continue growing, reaching over 125,000 inmates
in 2000 and over 138,000 inmates in 2006. These projected
populations represent increases of about 25 and about 38 percent,
respectively, over the 1995 level.
Table II.4
Projected Federal Prison Inmate
Populations, 1996 Through 2006
Projec
ted
popula
Year tion\a
-------------------------------------------------------------- ------
1996 105,12
8
1997 111,34
7
1998 116,76
2
1999 120,97
4
2000 125,14
4
2001 128,45
1
2002 130,90
8
2003 133,14
0
2004 135,00
6
2005 136,67
8
2006 138,12
0
----------------------------------------------------------------------
\a These are point estimates generated by BOP's model. The model
does not generate confidence intervals.
Source: BOP.
Table II.5 shows NCCD's prison population projections through 2000
for the 21 states that use its Prophet population projection model,
California (which uses a similar model), and Texas, which provided
its own projections to NCCD. Using data for these 23 states, and
assuming that the sentencing policies in effect in 1994 would
continue, NCCD estimated that the federal and state prison inmate
population could reach 1.4 million in 2000.
Table II.5
Projected Prison Inmate Populations for
23 States, 1996 Through 2000
State 1996 1997 1998 1999 2000
------------------------------ ------ ------ ------ ------ ------
Arkansas 9,079 9,106 9,185 9,264 9,337
California 142,55 159,99 176,01 192,81 210,44
1 2 3 4 2
Colorado 12,261 13,308 14,326 15,455 15,455
Connecticut 12,989 13,301 13,458 13,684 13,999
Florida 72,357 73,999 75,493 76,512 77,896
Hawaii 2,102 2,290 2,415 2,540 2,261
Idaho 2,759 2,836 2,941 2,979 2,989
Illinois 41,726 43,586 46,105 48,561 51,216
Indiana 15,100 15,300 15,500 15,800 16,100
Kansas 7,045 6,997 6,839 6,637 6,637
Kentucky 12,320 13,072 14,518 15,227 15,987
Louisiana 28,000 29,500 31,000 32,500 34,000
Massachusetts 10,632 11,250 11,580 11,911 12,356
Michigan 44,073 45,796 47,580 49,440 51,365
Mississippi 12,194 12,684 13,184 13,664 14,170
Nevada 7,670 8,101 8,530 9,032 9,516
Ohio 43,059 43,915 44,850 46,080 47,215
Oklahoma 18,466 18,918 19,327 19,825 20,112
Oregon 7,472 8,600 9,764 11,440 13,116
Rhode Island 3,181 3,215 3,248 3,248 3,293
Tennessee 18,372 18,758 19,227 19,671 19,886
Texas 140,88 148,86 153,65 151,84 149,44
2 6 4 2 4
Virginia 31,700 34,659 38,050 39,005 40,984
======================================================================
Total 695,99 738,04 776,78 807,17 837,80
0 9 7 6 0
----------------------------------------------------------------------
\a These are point estimates generated by NCCD's model. The model
does not generate confidence intervals.
Source: NCCD.
GROWTH IN OTHER ADULT
CORRECTIONAL POPULATIONS
-------------------------------------------------------- Appendix II:3
As table II.6 shows, the populations in all four adult correctional
categories--prison, probation, parole, and jail--have increased
between 1980 and 1995. The total federal and state prison inmate
population in custody\2 grew by about 237 percent from 1980 to 1995.
In comparison, during the same period, the probation population grew
by 176 percent, the parole population grew by 218 percent, and the
jail population grew by 178 percent.\3 Overall, the total adult
correctional population grew by 192 percent, from 1.8 million in 1980
to about 5.4 million in 1995.\4 During this time, the U.S. adult
population grew by about 19 percent, from 163.5 million to about
194.0 million. Accordingly, the adult correctional population
represented 2.8 percent of the total adult population in 1995, well
over double the 1.1-percent level in 1980.
Table II.6
Adult Correctional Populations by
Category, 1980 and 1995
Percentage
population
increase
from 1980
Population category 1980 1995 to 1995
----------------------------------- ---------- --------- ----------
Prison 319,598 1,078,545 237
Probation 1,118,097 3,090,626 176
Parole 220,436 700,174 218
Jail 182,288 507,234 178
Total adult correctional population 1,840,419 5,376,579 192
U.S. adult population 163,541,00 194,015,0 19
0 00
----------------------------------------------------------------------
Source: GAO analysis of BJS data.
Figure II.1 shows that the populations in the four adult correctional
categories as a percentage of the total adult correctional population
were essentially unchanged between 1995 and 1980. Specifically, in
1995, the prison inmate population represented 20 percent of the
total adult correctional population, compared with 17 percent in
1980. Also, in 1995, the probation population represented 57 percent
(61 percent in 1980), the parole population 13 percent (12 percent in
1980), and the jail population 9 percent (10 percent in 1980) of the
total adult correctional population.
Figure II.1: Distribution of
Adult Correctional Populations,
1980 and 1995
(See figure in printed
edition.)
Source: GAO analysis of BJS data.
--------------------
\2 Populations in custody are those actually held in a state's
correctional facility. According to BJS, for comparison purposes
across correctional populations, custody figures are used for prison
and jail populations to avoid the double-counting of inmates.
\3 Jail population data are as of June 30, 1996.
\4 According to BJS, because some persons may have multiple statuses,
the sum of persons incarcerated (prison or jail) and under community
supervision (probation or parole) overestimates the total
correctional population.
TRENDS IN FEDERAL AND STATE PRISON
OPERATING AND CAPITAL COSTS
========================================================= Appendix III
PRISON OPERATING COSTS (FISCAL
YEARS 1980 THROUGH 1994)
------------------------------------------------------- Appendix III:1
Table III.1 shows that federal and state prison annual operating
costs have grown significantly (a combined 224 percent increase in
inflation-adjusted terms) since fiscal year 1980. These costs
cumulatively totaled about $137.7 billion in current dollars for
fiscal years 1980 through 1994.
Table III.1
Federal and State Prison Operating Costs
in Current Dollars, Fiscal Years 1980
Through 1994
Fiscal year Federal State Total\a
------------------------------- ----------- ----------- -----------
1980 $319,274 $2,787,369 $3,106,643
1981 346,517 3,229,234 3,575,751
1982 368,000 3,794,178 4,162,178
1983 435,000 4,346,273 4,781,273
1984 529,245 5,066,666 5,595,911
1985 500,941 5,934,160 6,435,101
1986 555,097 6,619,534 7,174,631
1987 580,120 7,601,594 8,181,714
1988 878,502 8,586,498 9,465,000
1989 900,334 9,611,020 10,511,354
1990 1,148,678 11,194,236 12,342,914
1991 1,318,741 12,514,171 13,832,912
1992 1,585,498 13,290,202 14,875,700
1993 1,767,019 14,239,710 16,006,729
1994 1,918,067 15,776,174 17,694,241
======================================================================
Total\a $13,151,033 $124,591,01 $137,742,05
9 2
----------------------------------------------------------------------
Note 1: Dollar figures represent actual dollars (no adjustment for
inflation).
Note 2: According to BOP, the federal cost data presented are actual
obligations, adjusted for equipment and other capital item costs.
\a Details may not add to total due to rounding.
Source: BOP and U.S. Bureau of the Census.
PRISON CAPITAL COSTS (FISCAL
YEARS 1980 THROUGH 1994)
------------------------------------------------------- Appendix III:2
As table III.2 shows, federal and state prison capital costs have
also grown significantly from fiscal year 1980 to 1994. Total
capital costs reached about $2.3 billion in fiscal year 1994, an
inflation-adjusted increase of about 141 percent over the level in
fiscal year 1980. Federal and state capital costs cumulatively
totaled about $25.4 billion for fiscal years 1980 through 1994.
Table III.2
Federal and State Prison Capital Costs
in Current Dollars, Fiscal Years 1980
Through 1994
Fiscal year Federal State Total\a
---------------------------------- ---------- ---------- ----------
1980 $21,766 $515,854 $537,620
1981 20,807 541,939 562,746
1982 6,500 604,715 612,215
1983 54,000 589,325 643,325
1984 42,072 792,963 835,035
1985 111,787 992,071 1,103,858
1986 145,382 1,473,544 1,618,926
1987 249,279 1,226,049 1,475,328
1988 544,392 1,715,463 2,259,855
1989 266,994 1,984,518 2,251,512
1990 1,505,953 2,175,823 3,681,776
1991 419,262 2,497,997 2,917,259
1992 475,733 2,057,383 2,533,116
1993 366,144 1,726,171 2,092,315
1994 311,687 1,965,763 2,277,450
======================================================================
Total\a $4,541,758 $20,859,57 $25,401,33
8 6
----------------------------------------------------------------------
Note 1: Dollar figures represent actual dollars (no adjustment for
inflation).
Note 2: According to BOP, the federal cost data presented are actual
obligations, adjusted for equipment and other capital item costs.
\a Details may not add to total due to rounding.
Source: BOP and U.S. Bureau of the Census.
Table III.3 shows that federal and state prison operating and capital
costs cumulatively totaled about $163.1 billion in current dollars
for fiscal years 1980 through 1994. Federal costs totaled about
$17.7 billion, while state costs totaled about $145.5 billion during
this period. Operating costs totaled about $137.7 billion, while
capital costs totaled about $25.4 billion.
Table III.3
Federal and State Prison Total
Cumulative Costs in Current Dollars,
Fiscal Years 1980 Through 1994
Operating Capital
cost cost Total
---------------------------------- ---------- ---------- ----------
Federal $13,151,03 $4,541,758 $17,692,79
3 1
State 124,591,01 20,859,578 145,450,59
9 7
======================================================================
Total $137,742,0 $25,401,33 $163,143,3
52 6 88
----------------------------------------------------------------------
Source: Summary of tables III.1 and III.2.
PROJECTED FEDERAL PRISON
OPERATING AND CAPITAL COSTS
------------------------------------------------------- Appendix III:3
Table III.4 shows BOP's projections for federal prison operating and
capital costs through fiscal year 2006. As shown, BOP projects that
operating costs in fiscal year 2006 could be almost double the 1996
level. The projections also show that capital costs are expected to
fluctuate on a year-by-year basis.
Table III.4
Projected Federal Prison Operating and
Capital Costs, Fiscal Years 1996 Through
2006
Fiscal year Operating Capital Total
---------------------------------- ---------- ---------- ----------
1996 $2,440,394 $445,903 $2,886,297
1997 2,843,292 506,552 3,349,844
1998 3,054,347 553,493 3,607,840
1999 3,365,142 424,136 3,789,278
2000 3,604,601 341,741 3,946,342
2001 3,805,956 303,576 4,110,532
2002 3,958,572 287,608 4,246,180
2003 4,163,304 282,823 4,446,127
2004 4,343,282 283,718 4,627,000
2005 4,528,328 287,555 4,815,883
2006 4,721,072 293,005 5,014,077
----------------------------------------------------------------------
Note: According to BOP, projections of operating and capital budgets
are based on estimated obligations--as presented in the Office of
Management and Budget's fiscal year 1998 budget for the federal
government--adjusted for equipment and other capital item costs.
Source: BOP.
MODELS AND METHODOLOGIES USED FOR
PROJECTING PRISON INMATE
POPULATIONS
========================================================== Appendix IV
Various types of models and methodologies are used to project prison
inmate populations, but microsimulation is the model type most widely
used by federal and state corrections agencies.\1 As used by BOP and
27 states, including California and Texas, microsimulation modeling
can project prison populations by simulating a wide range of
legislative, policy, or administrative changes that affect the
criminal justice system. Other states use flow models or statistical
methods to project populations. Except for BOP's projection model,
we did not identify any independent assessments of the validity or
reliability of the various projection models. However, self-reported
data indicated that the models have been accurate.
--------------------
\1 NCCD, which is a nongovernmental forecasting organization, also
uses a microsimulation model--NCCD Prophet.
MICROSIMULATION MODELING IS
MOST WIDELY USED
-------------------------------------------------------- Appendix IV:1
Microsimulation models replicate the flow of persons through the
criminal justice system, incorporating considerable detail from the
actual records of convicted offenders. As table IV.1 shows,
microsimulation modeling is used by BOP and 27 states. In 1987, BOP
and the U.S. Sentencing Commission jointly developed the Federal
Sentencing Simulation Model (FEDSIM) to comply with a series of
congressional initiatives that required an impact analysis of federal
sentencing guidelines. In January 1995, BOP began using a revised
model (FEDSIM-2), which incorporates different data sets based upon
experience under federal sentencing guidelines. The NCCD Prophet
model is based on a model that the California Department of
Corrections has used since 1976. The Texas Criminal Justice Policy
Council developed the JUSTICE microsimulation model in 1987. Each of
these three models is discussed separately in the following sections.
Table IV.1
Types of Models Used by BOP and States
to Project Prison Populations
Number
of
Model type User (BOP or state)\a users
------------------------------ ------------------------------ ------
Microsimulation
FEDSIM-2 Bureau of Prisons 1
NCCD Prophet Arizona, California, Colorado, 23
Connecticut, Florida, Hawaii,
Idaho, Illinois, Indiana,
Kansas, Kentucky, Louisiana,
Maine, Massachusetts,
Michigan, Mississippi,
Nevada, New Jersey, Ohio,
Oklahoma, Rhode Island,
Tennessee, Virginia
Other\b Georgia, Minnesota, North 4
Carolina, Texas
Flow\c
IMPACT District of Columbia, Montana, 6
Nebraska, New Mexico,
Vermont, Wyoming
Other Pennsylvania, Oregon, Utah 3
Statistical methods Alabama, Alaska, Arkansas, 10
(various)\d Delaware, Iowa, Maryland,
Missouri, New York, South
Dakota, Wisconsin
Other (proprietary)\e New Hampshire, South Carolina, 3
Washington
======================================================================
Total users N/A 50
----------------------------------------------------------------------
Legend: N/A equals not applicable.
\a State data exclude North Dakota and West Virginia because these
states do not project inmate populations.
\b Includes the JUSTICE model used in Texas.
\c Flow models track the movement or "flow" of persons throughout
each stage of the criminal justice system, from arrest to parole.
One flow model used by several states is the Interactive Model for
Projecting Arrest and Corrections Trends (IMPACT).
\d Various regression and time series analyses and techniques.
\e Proprietary models not classifiable as microsimulation, flow
models, or statistical methods.
Source: Developed by GAO from data presented in Corrections
Compendium (April 1996) and NCCD FOCUS (July 1995), and discussions
with BOP, NCCD, and state corrections agency officials.
BOP'S MODEL (FEDSIM-2)
------------------------------------------------------ Appendix IV:1.1
Using convicted offenders' cases, two data sets are used when
FEDSIM-2 is updated annually: (1) the total prison population at the
end of the prior fiscal year and (2) all inmates admitted into
federal prisons during the prior fiscal year. In this model,
prospective release dates for individuals in both groups are
recorded, and sentencing time is distributed into monthly groupings
or "trace elements" to track the total time served for each prisoner.
FEDSIM-2 tracks convicted drug offenders,\2 along with 20 other types
of offenders, to determine the overall trend in the federal prison
population.
--------------------
\2 Since drug offenders constitute over one-half of the total federal
prison population (60 percent in 1993), this category has been called
the "tail that wags the dog."
NCCD'S MODEL (PROPHET)
------------------------------------------------------ Appendix IV:1.2
The Prophet model, which NCCD has customized to accommodate states'
correctional information systems, can predict future population
levels, isolate the effects of specific practices, and predict the
effects of proposed policy changes. This model is conceptually
designed around the movement of offenders into, through, and out of
the prison and parole systems. As shown in table IV.1, 23 states
(including California) use a form of this model. The Prophet model
simulates offender subgroup compositions and lengths of stays within
each stage of the correctional system. Individual cases are then
processed through a series of probability distribution arrays or
matrices, which allows the model to compute prison populations.
Using the model, the total correctional population can be separated
into subgroups, and forecasts for each subgroup can be made on the
basis of the proposed policy changes, without altering the status of
the other subgroups. Prophet requires five data sets to
operate--prison admissions, prison exits, current prison population,
current parole population, and parole exits.
TEXAS' MODEL (JUSTICE)
------------------------------------------------------ Appendix IV:1.3
Texas' JUSTICE microsimulation model uses convicted felony offenders'
records from the state's jail, prison, and parole populations. On a
monthly basis, these data are loaded into or updated in the model,
which has two parts. One part covers prisoners coming into the
correctional system, and the second part covers the policies that
determine movement within the system. Projections are made from the
first part, and impact analyses of proposed policy changes are made
from the second part.
JUSTICE creates future offenders' records by duplicating key
characteristics (e.g., offense and sentence) of the current
admissions and parolees and assessing the probability of these
characteristics being present in future admissions. The model
accounts for the specific months that offenders enter the different
stages of the system and projects a total number of adult felony
arrests on the basis of the at-risk population--i.e., that portion of
the Texas population (aged 18 to 44 years old) considered most likely
to engage in criminal activity. Each offender's key characteristics
determine the flow of the offender through the system by triggering
certain criteria (e.g., parole eligibility) that affect the time and
direction of the offender's movement through the system.
The first part of the JUSTICE model is used to make projections of
those most likely to be sent to prison or placed on probation. The
second part of the model permits simulating the impact of proposed
changes affecting the size of the probation, prison, and parole
populations. Texas' JUSTICE model has considerable flexibility in
simulating changes in the major "rules of movement" through the
state's correctional system. For example, 29 parameters can be
interactively altered to assess the impact of proposed policy
changes.
RELIABILITY OF MICROSIMULATION
MODELS
-------------------------------------------------------- Appendix IV:2
According to experts in the prison modeling field, there are no
standard criteria for assessing or validating the reliability of
microsimulation models used to project prison populations. The NCCD
and state agency officials we contacted said that microsimulation
models are generally considered reliable if the projections come
within 2 percent of the actual populations. These officials also
commented that projections beyond 5 years, and perhaps even beyond 2
years, are usually considered rough estimates.
Notwithstanding that comparing projections with actual prison
populations may be an insufficient gauge of a model's reliability, on
the basis of self-reported assessments, the three major models we
identified (FEDSIM-2, NCCD Prophet, and JUSTICE) are reported to
produce accurate projections. For instance, the April 1996 issue of
the Corrections Compendium presented the results of a survey about
prison population projections. In responding to a question in the
survey related to projection accuracy, BOP, California, and Texas
reported that their respective projections were accurate.
OTHER MODELS AND METHODS
-------------------------------------------------------- Appendix IV:3
As shown in table IV.1, in contrast to the microsimulation models
used by BOP and most states, other states use flow models and
statistical methods to project prison populations. Flow models
separate the characteristics of the various groups or cohorts of
prisoners moving through the system from the aggregate population for
analysis. These models track the offenders through the criminal
justice system by calculating percentages (or branching ratios) of
the offender population that continue through each stage of the
system. For example, of every 100 arrests, perhaps only 30 percent
of the individuals will be indicted; and, of the indictments, perhaps
only 50 percent will be convicted; etc. In other words, flow models
represent continuation into the next stage, with branching ratios
used to "prune" out those offenders who will not become part of the
prison inmate population.
Also shown in table IV.1, 10 states use statistical methods, such as
regression analysis and time series analysis, to project prison
populations. Statistical methods all use data from past patterns to
project future inmate populations. Regression analysis, for example,
is a statistical technique based on equations that functionally
relate one or more independent variables, with coefficients
determined from previous analysis, to a dependent variable.
Statistical methods tend to be nonpolicy sensitive and, therefore,
are not particularly useful for impact analyses. However, reasons
for changes can be deduced retrospectively from these statistical
methods.
Finally, as shown in table IV.1, three states use models or
methodologies that are not classifiable either as microsimulation,
flow models, or statistical methods. For example, one state projects
its future population by extrapolating the previous 5-year growth
trend in the existing population.
MAJOR CONTRIBUTORS TO THIS REPORT
=========================================================== Appendix V
GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C.
--------------------------------------------------------- Appendix V:1
Seto J. Bagdoyan, Evaluator-in-Charge
David Thomas, Intern
David Alexander, Senior Social Science Analyst
Pamela V. Williams, Communications Analyst
Katherine M. Wheeler, Publishing Advisor
OFFICE OF THE GENERAL COUNSEL,
WASHINGTON, D.C.
--------------------------------------------------------- Appendix V:2
Geoffrey R. Hamilton, Senior Attorney
DALLAS FIELD OFFICE
--------------------------------------------------------- Appendix V:3
Danny R. Burton, Assistant Director, Administration
of Justice Issues
Mary K. Muse, Senior Evaluator
*** End of document. ***