_..-

-

GAO

United

States General

May 1992

Office

____ ._

-----_

_-~_._

STATEANDFEDERAL
PRISONS
Factors That Affect
Construction and
Operations Costs
146913

RESTRICTED--Not
to be released outside the
General Accounting
Office unless specifically
appro,ved by the Office of Congressio
Relations.

----------.GAO/GGD-92-73

_^

R&x-t to Congressional Requesters

-.- ..-

I

Accounting

554362

-WY

-

GAS3

Acconndng

United States
General
Wahhgtm,

Of&e
D.C. 20548

General Government Divieion

B-248117
May 19,1992
The Honorable Bob Graham
United States Senate
The Honorable Dennis DeConcini t ,,/’
United States Senate
The Honorable Richard Bryan
United States Senate

,J

The Honorable Joseph Lieberman /
United States Senate
The Honorable Herb Kohl
United States Senate

: .,d

In response to your joint request, we recently issued a report that
compared construction and operations costs for medium security state
and federal prisons opened between 1986and 1989and identified
opportunities for savings in the federal system.1For the purposes of that
report, we aggregateddata for the state and federal prisons in our sample
and, except for a few examples, did not include data for individual prisons.
After the report wss issued, your offices suggestedthat publishing the cost
information for the individual state and federal prisons we sampled and
the major reasons for cost differences might encourage some of the higher
cost jurisdictions to try to reduce costs. We agreed to prepare a report on
the information we obtained for the individual prisons and the factors that
contributed to differences in their construction and operations costs.

Background

The state and federal governments are spending billions for new prison
construction to accommodate continuing increases in inmate populations.
According to the February 1992 Corrections Compendium, 26 state
corrections systems requested a total of $2.3 billion for the 1992-1993fiscal
year. Included were requests for 86 new facilities, which would add over
66,000new prison beds2 Texas alone asked for more than $600 million in
construction funds to add over 26,000new beds. The Federal Bureau of
Prisons (BOP) is in the midst of an unprecedented expansion program that
(GAOfGGD-02-3,Oct.
--,m.....s_,_
,_ _
aA “bed” is a generic unit of measure for a prison’s inmate capacity. For example, a MO-bed prison
would have a rated capacity of 600 inmates.

Page 1

OAOioOD-92-72

Prbon Colt Factore

B-248117

will double its 1989prison capacity by 1996at a cost of about $3 billion, In
reality, construction costs are only the down payment on a prison’s total
cost to society. BOP has estimated that operating a prison over its useful
life costs 16 to 20 times its construction costs.
Prisons can vary widely in size, design, and costs of construction. There is
no universal standard or “cookie cutter” prison design, although some
jurisdictions have adopted their own standard layouts. Many factors can
influence a prison’s ultimate structure, including its intended capacity, the
security level of inmates expected to be housed in the facility, the urgency
of need for prison beds, the jurisdiction’s desire to meet the accreditation
standards of the American Correctional Association (ACA), budget
constraints in the jurisdiction, and the corrections policy and philosophy
of the jurisdiction.

Resultsin Brief

Construction costs varied widely among the medium security state and
federal prisons we sampled.At the 36 medium security prisons included in
our sample (32 state, 4 federal), construction costs ranged from $11,243to
$93,333per bed and averaged$66,374.3The most important factor
contributing to differences in prison construction costs per bed was the
amount of space provided, measuredin terms of gross square feet (GSF)
per inmatem4
This factor accounted for 95 percent of the variability in per
bed construction costs for the 36 prisons in our sample.
Other factors that m ight have contributed to the cost differences were the
type of building structure, the housing area design and layout, whether the
facility was designedfor a m ix of security levels, and geographic location.
We tested alternative combinations of these factors. We found that none of
the combinations explained a significant amount of additional variability
a
in construction costs beyond that explained by the amount of space
provided to each inmate. Although state and federal prison systemsare
revising their design standards to allow for more double celling of inmates,
we believe all of the five factors identified above will continue to affect
differences in prison construction costs after the revisions are fully
implemented.
9’he cost per bed is the total cost of the facility divided by the number of inmates that the facility was
designed to accommodate. The cost per bed includes costs for all areas of the prison, including
housing, recreation, education, and prison industry.
‘Gross square feet is defined by the American Institute of Architects as the sum of the areas of the
several floors of a building, measured from the exterior faces of exterior walls or from the centerline
of walls separating buildings. The areas of covered walkways, porches, and similar space are
multiplied by a factor of 5.

GAO/GGD-92-73 Prieon Cost Factors

Pa2e 2

‘i,:

B-242117

Operations costs also varied widely at the 23 prisons (21 state, 2 federal)
that provided operations cost information, ranging from $22.26to $81.08
per inmate per day (referred to as an inmate day) and averaging$41.93.
The key factors that contributed to the operations cost differences were
personnel salariesand related expenses,inmate-tostsffratios, and the
amount spent on supplies, materials, and food.

Objectives,Scope,
and Methodology

Our objective was to identify the factors that contributed to differences in
prison construction and operations costs. We obtained prison construction
and operations cost information from the questionnairesdeveloped for our
recently issued prison cost report (see footnote 1). The questionnaires
were designedto obtain reliable and comparable data for each state and
federal prison that met the following criteria:
opened between 1986and 1989;
new, independent facilities;
designedto house adult males;
designedfor a population of 200 inmates or more; and
in operation for one full year at or near design capacity (operations costs
only).
We took several steps to ensure that the questionnaireswould obtain
sufficient data to perm it meaningful comparisonsdespite the great number
and diversity of reporting jurisdictions. In designing the questionnaires,we
met with architects, engineers,and cost accountants to identify the key
information that would account for differences in design and costs. To
encourageparticipation in our study and lessen the burden of responding,
we focused the questionnaireson information that (1) was readily
available in the states’departments of correctioti and BOP; (2) was, for the
most part, consistently defined and captured in standard government cost
accounts, and (3) was objective, measurable,and comparable (e.g., size,
populations, number of rooms).
We pretested the questionnairesat three state corrections departments
and BOP to further increase the likelihood that the respondents would
understand how to complete them and provide comparable and reliable
data. We also followed up with respondents that appeared to have
submitted incomplete or erroneous data On the other hand, we did not
make a detailed cost reconciliation for each prison, nor did we assess
what effect, if any, prison design and construction may have had on
enhancing prisoner rehabilitation and the incidence of prison violence.

Page 3

GAWGGD-92-73 Prima Coot Fwtmw

B=242117

We used ACA’S1990Directory of Juvenile and Adult Correctional
Departments,Institutions, Agencies,and Paroling Authorities as our
source for states and prisons to receive the questionnaires.We mailed
questionnairesto BOP, the District of Columbia, and the 37 states that the
Directory identified as building prisons during the target period. This
distribution covered 62 state and 4 federal prisons.
BOP provided construction cost information for all four facilities built
between 1986and 1989which were, for the most part, all designedto
house a majority of medium security inmates. These prisons are the
Federal Correctional Institution (m) Phoenix, Ark; m Marianna, Fla.; F+CI
Sheridan,Oreg,; and FCIMcKean,Pa. BOP’S construction cost information
and our analysisdid not include a 126bed temporary dorm itory built at FCI
EIOP also provided operations cost
Phoenix in 1990at a cost of $6Q8,6QQ.
information for the two prisons that had been in operation for at least one
year at or near their design capacity. These prisons are FCXPhoenix and FCI

Of the 62 questionnairesmailed to state prisons, 11 were not used because
we later found that the projects did not meet one or more of our criteria
Two states voluntarily completed questionnairesfor prisons that met our
selection criteria but that were not listed in ACA’S 1990directory. Of the 63
state prisons we expected to participate, 46 (from 30 states and the
District of Colwnbia) returned the construction portion of our
questionnaire,and 29 (from 21 states and the District of Columbia)
returned the operations cost portion. However, we reduced the operations
cost sample to 28 becauseone jurisdiction did not isolate operations costs
by individual departments, and thus the questionnaire responsewas not
usable. Becausethe four federal prisons built during the defined time
frame were designedto house mostly medium security inmates, we
0
reduced the state sample to include only prisons designedto house a
majority of medium security inmates. Our final tally was construction cost
data from 32 prisons in 20 states and the District of Cohunbia, and
operations cost data from 21 prisons in 16 states. A list of the state prisons
that reported construction cost information is in appendix III. A list of the
state prisons that reported operations cost information is in appendix Iv.
To facilitate our analysisof construction costs, we divided the state and
federal prisons in our sample into three cost groups-low, medium, and
high. When the 36 prisons were arrayed in order of cost per bed from low
to high, natural breakpoints occurred between the low cost and medium
cost groups and between the medium cost and high cost groups.

Pqe 4

GAO/GGD-92-73 Prhon Ckut Flctom

B-240117

Construction costs per bed in the $11,243to $24,679range were classified
as low cost, those in the $46,007to $73,438range were classified as
medium cost, and those in the $83,771to $93,333range were classified as
high cost. Similarly, to analyze operations costs, we divided the prisons
into low, medium, and high cost groups. We used breakpoints that existed
in the daily operations costs per inmate to define the three cost groups.
Daily operations costs per inmate in the $22 to $37 range were classified as
low cost, those in the $42 to $61 range were classified as medium cost, and
those in the $69 to $81 range were classified as high cost. There was no
direct relationship between the operations cost groups and the
construction cost groups.
We used standard statistical techniques to determ ine the relationships
between prison construction costs and the factors for which we obtained
data. These techniques allowed us to determ ine the amount of variability
within different measuresof construction costs that was explainable by
each factor and by various combinations of factors. We were able to
identify the factors that explained at least 96 percent of the variability of
each of the following three measuresof prison construction costs: total
construction costs, costs per bed, and costs per GSF.These factors are
discussedindividually in the report. Other factors were significant for
particular groups of prisons but were not consistent across all of the
prisons included in the analysis.For example, housing area design and
layout proved to be important in explaining construction costs for state
prisons, but not for federal prisons.
The results of our statistical analysis must be considered in light of certain
lim itations inherent in our study. Becausethe 36 prisons included in the
analysis were not randomly selected,we cannot infer that they are
representative of the universe of prisons. If additional ‘or tiother set of
prisons were included in the analysis,the results m ight be’different. It is
also possible that additional factors for which data was not collected may
affect prison construction costs.
We did our work between ,December1991and March 1992in accordance
with generally accepted government auditing standards.

Construction Costs
VbriedW idely ”

Construction costs varied widely among the medium security prisons we
sampled.At the 36 prisons, total construction costs ranged from a low of
$6,464,644(rated capacity of 312 inmates) to a high of $266,066,796(rated
capacity of 2,916inmates). Per bed construction costs ranged from $11,243

Page 6

-D-92-78

P&on

Coat Factmu

B=242117

to $93,333and averaged $50,374.The cost per bed of the high cost prisons
($87,271) averaged almost five times as much as the cost per bed of the
low cost prisons ($17,730).The average per bed cost of the medium cost
prisons was $58,282.See figure 1.
Flgure 1: Avorago Prlron Conatruttlon
Co8ta p8r Bed

100000

Dollan In thousands

60000
60000

70000
60000
60000
40000
20000
2oooo
10000
0

CO81 coet

coift

prisons

prisons

-----

The Amount of Space
Provided to Inmates
Accounted for Most
Construction Cost
Differences

prlronr

Weighted average for 36 prisons, $56,374

Of the factors we examined, the amount of space provided, measured in
terms of GSFper inmate, accounted for most of the differences we found in
prison construction costs per bed. The high cost prisons provided an
overall average of 554 GSFper inmate, over two and one-hslf times the
average of 215 GSFper inmate provided at the low cost prisons. After
testing alternative factors, we found that, when considered independently,
the amount of space provided to inmates accounted for 95 percent of the
variability in cost per bed. Figure 2 illustrates the close relationship
between cost per bed and GSFper inmate.

Page 9

GAO/GGD-92.73 Prison Coot Factom

L

B.248117

Agun 2: Comparlron Botwoen Average per Bad Conrtructlon Cortr and Grou Square Feet per Inmate in State and
Fodenl Prhonr

loo

90

80

70

60

60

40

30

20

10

0

Dollsrs In thousands
Aversgo par brd construction costs

0

100 200 300 400 500 600 700 800 900 1000

Gross squrrrr fort par lnmrtr

Low cost prisons
Medium cost prisons
High cost prisons

‘.Sypeof Structure,
Housing
Configuration,M ix of
Security Levels,and
GeographicLocation
Also Affected
Construction Costs

In addition to the amount of space provided to inmates, we examined
several other factors in terms of per bed construction costs. One of these
factors was the type of building structure. The National Directory of
Corrections Construction, published by the National Institute of Justice
(April 1988),classifiesprisons into several general types of structures,
including an integrated structure (one building); clusters (a number of
individual buildings that are interconnected); and campus style (a number
of individual buildings that are not connected).6Although construction
costs varied for each design style becauseof factors such as size and
housing layout, our analysisfound that integrated structures, on average,
were the most costly of the three types of structures, followed by clusters
and campusstyle. Of the seven high cost prisons in our sample, five were
either single buildings or clusters. In contrast, seven of the eight low cost
prisons were campus style.
The design of prison housing units also contributed to construction cost
differences at our sample prisons. The high cost prisons reported that,
The prteons in our sample were in these three categories. Other types of structurea described in the
directory were hi rise (one building, more than four stories in height); ladder, telephone pole (linear
cell blocks arrange
-@ii- in parallel off a central connecting corridor); wheel, apok e, or radial (linear cell
blocks that emanate from one central contsol area like spokes f&m the hub of a wheel); and courtyard
(linear cell blocks interconnected around a central enclosed courtyard).

Pnge 7

GMMGQD-92-72 Prbon Coet Factors

a

B-242117

overall, 90 percent of their beds were designedto be in single cells, less
than one percent in multiple occupancy cells, and 10 percent in
dorm itories. In contrast, only 4 percent of beds at the low cost prisons
were designedto be in single cells, while 60 percent were in multiple
occupancy cells and 36 percent in dorm itories.
Another factor that contributed to construction cost differences was
whether the prison was built to accommodateinmates &om different
security levels. Construction costs per bed tended to increase as the
percentageof medium security beds declined. Overall, the high cost
prisons classified 76 percent of their beds as medium security, compared
to 89 percent for the medium cost prisons and 90 percent for the low cost
prisons.
The geographic location of the prison also affected construction costs.
According to the National Institute of Justice and the ACA, prison
construction costs tend to be higher in the Northeast and West and lower
in the South and M idwest due to significant differences in the cost of
materials and prevailing labor rates. The prisoxisin our sample reflected
those tendencies.Of the 8 low cost prisons, 6 were in the South, while only
2 of the 21 medium cost and 1 of the 7 high cost prisons were in the South.
Conversely,no Northeast prisons were in the low cost group, while five
Northeast prisons were in the medium cost group and three in the high
cost group.
Another indicator of the importance of geographic location is its effect on
the cost per GsF.The cost per osFis, in effect, the measure of the amount
of space the jurisdiction was able to buy for its money, independent of the
number of inmates the prison was designedto house. It encompassessuch
cost factors ss site acquisition and preparation as well as materials and
labor. To some extent, cost per GSFcould even be a measure of the
economic conditions and contracting environment during the period
leading up to construction. Cost per GsFat the 36 prisons ranged from
$68.06to $216.60and averaged$129.48.We analyzedthe effect of various
factors on the cost per GSFand found that about 96 percent of the
variability in cost per GSFwas explainable by the national construction
cost index. This Index is a surrogate measurefor the state in which the
prison is built.
The factors that contributed to differences in prison construction costs are
discussedin appendix I.

CJMUGGD-92-78 Primon Cht Factma

Page 9

,:;

a

B=248117

Changesin Prison
Design Stand&s
W iU
Affect Future
Construction Costs

For comparative purposes our cost per bed analyseswere based on a
common baseline-the number of inmates the facilities were actually
designedto accommodate(referred to as the design capacity or the rated
capacity) as reported by the participating jurisdictions. The prisons we
sampled were built with design standards that called for housing one
inmate in a single cell or two or more inmates in multiple occupancy cells
or dorm itories.
BOP has recently adopted a lim ited double celling standard (two inmates
per cell) for the design of medium security prisons. The new standard
allows for double celling in up to 60 percent of cells having 76 or more
square feet. This change also increased the rated capacity of existing nor
facilities that met the cell size criterion. In practice, BoPfacilities have
been double celled extensively for some time and without unmanageable
problems.

Prison design standards are being revised at the state level as well. In
August 1991,the ACA revised its accreditation standards for medium
security facilities to perm it double celling and reduced the required space
in multiple occupancy and dorm itory housing areas. Some states will likely
revise their rated capacities based on the new ACA standards. Further, in
January 1992,Attorney General W illiam P. Barr announced an effort to
help states lift some cour%orderedprison population ceilings. These are
believed by some to unreasonablylim it the number of inmates that may be
housed in a prison.
To the extent that the new standards increase rated capacity, new prisons
that incorporate the new standards will have lower per bed construction
costs. Nevertheless,we believe that the factors that affected prison
construction costs at the prisons we sampled will continue to signMcantly
affect construction costa after the revisions are fully implemented.That is,
prison construction costs will continue to be driven in large measureby
the amount of spaceprovided to inmates (GSF per inmate), the type of
building structure, the housing area design and layout, whether the facility
was designedfor a m ix of security levels, and geographic location.

PersonnelExpenses
Accounted for the
Majority of ”
OperationsCosts

Operations costs also varied widely at the 23 prisons (21 state, 2 federal)
that provided operations cost information. Operations costs ranged from
$22.26to $81.08per inmate day and averaged$41.93(see fig. 3). The low
cost prisons averaged$32.37per inmate day, compared to $46.83for the
medium cost prisons and $62.81for the high cost prisons. The single

Page 9

GAOJGGD-92.78 FMaon coat Fu?tora

B-242117

largest operational expensewas personnel compensation+&aries and
related expenses.Personnelcosts ranged from 66 to 93 percent of total
operations costs and averaged76 percent.
Figure 3: Avongo Opmtlonr Co&
p& Inmtio Day _

100.00

Cc4 In dollrn

PO.00
60.00
70.00
60.00
60.00
40.00

--------

30.00
2o.ao
10.00
0
LoW
OOSt

p&on0
-----

Medium - High
cwt
oost
prlaonr
prlrons
Weighted average coat for 23 prieons, $41.83

An important factor in accounting for differences in personnel costs is the
staff@ levels of a prison relative to its inmate population (the
inmate-to-staff ratio). The prisons that employed more staff relative to
a
their inmate populations (i.e., those with lower inmate-tc-staff ratios)
tended to incur higher personnel costs-and, consequently,higher
operations costs. The low cost prisons reported an averageinmate&Maff
ratio of 3.13 to 1, compared to 2.71 to 1 for the medium cost group and 1.76
to 1 for the high cost group. Figure 4 shows that as the inmate-tu-staff ratio
increases,personnel costs per inmate day decrease.

Page 10

GAOAXD-92-72

Prbon Coat Factors

B-248117

Flguro 4: Comparhn Between the
Inmate-tctWafi Ratio and Pwwonnel
Co&to per Inmate Day

Inmate-to-atatf ratio

Pononnol coata par lnmrtl dry
66.0
62.0
46.0

330
3.26
3.00

44.0
40.0

2.75
2.50

36.0

2.26

32.0

2.00
1.76

1.60
20.0

1.25

16.0

1 .oo
0.76

12.0

0.60
0.26
0

6.0
4.0
0
tort
prisons

prlronr

Amounts Spent for
Supplies,Materials,
and Food Contributed
to Operations Cost
Variances

Conclusions

High
cost
prisons

Modlum

Low
COIL
-

Pemonnrl

--

Inmate-to-rtalf

cortt:prr

inmate day

ratio

Other important factors that contributed to differences in operations costs
were expensea for supplies, materials, and food. Although there were
notable differences in the amounta spent by individual prisons, the low
cost prisons spent an average of $4.76 per inmate day for supplies,
materials, and food, compared to $6.24 at the medium cost prisons and
$7.22 at the high cost prisons.
The Eactorsthat contributed to differences in operations costs are
discussed in appendix II.
At the 96 medium security prisons included in our sample, per bed
construction costs varied widely, ranging from $11,243to $93,333.The
amount of space provided, measured in terms of GSF per inmate,
accounted for 96 percent of the variability in per bed construction costs.
Other factors that might have contributed to the differences were the type
of building structure, the housing area design and layout, whether the
facility was designed for a mix of security levels, and geographic location.

Page11

GAWGGD92-78PrbonCortFa&om

B-248117

However, these did not have a sign&ant additional effect when
considered in combination with the amount of space provided to inmates.
Operations costs also varied at the 23 prisons that provided operations
cost information. Per inmate day operations costs ranged from $22.26to
$81.08.The factors that contributed to the difYerencesin operations costs
were personnel salaries and related expenses,inmate-to-stsff ratios, and
the costs of supplies, materials, and food.
Through better understanding of the reasonsfor cost differences in
various prisons, jurisdictions concerned about the high costs of building
and operating prisons can consider less costly alternatives. In designing
new prisons, significant economies can be realized by providing less GSF
per inmate (consistent with acceptable standards), using lower cost
building types, making greater use of dorm itories and multiple occupancy
cells in place of single cells, and, for somejurisdictions, selecting lower
cost geographic locations. Similarly, designing new prisons to operate with
greater inmate-to-staff ratios where appropriate can help hold down
personnel costs-the single largest operations cost at a prison.

AgencyComments
and Our Evaluation

We discussedthe contents of this report with JSOPofTiciah, who have
overall responsibility for prison construction. They generally agreed with
the facts presented. BOP officials informed us that its new design standard
for cells in medium security prisons is 76 square feet, a reduction from the
90 squarefeet required under the old standard. This change is expected to
be incorporated into BOP’S ofEicial policy guidelines in the near future. No
change is anticipated to EIOP’S policy of assumingthat 60 percent of the
cells will be double occupancy for purposes of calculating rated capacity.
At the Suggestionof BOP officials, we includedthis information in our
report, but the revised design standards did not affect our analysis of
construction costs for existing facilities that we sampled.
We also discussedthe contents of the report with an official of the ACA. He
stated that the report presented important information that will be very
useful to prison planners. In addition, he suggestedseveral factors that
contribute to differences in prison costs. He stated that the intended
inmate population, the m ission of the facility, climate, local building
codes, and whether the prisons are in heavily unionized or right-to-work
state3 can all affect prison construction and/or operations costs.

P-e

12

GAWGGD-92-78 Prbon Chat Factora

0

B-248117

In doing our work we took into account most of the factors described by
the ACAoffkial as they affected construction costs. For example, the
national construction estimator index, used in our analysti of construction
costa, was baaed on actual nationwide construction coats and thus
accounted for differences in climates, wage rates, and other construction
cost variables. Also, in developing our selection criteria, we excluded
prisons designedfor less than 200 inmates and prisons with special
m issions becausewe wanted to make prisons in our sample comparable
and reduce cost distortions.
Unless you announce the contents of this report earlier, we plan no further
distribution of it until 30 days from ita issue date. We will then make
copies available to the Attorney General,the Director of BOP, the states
that participated in our study, and other interested parties. Copies will be
made available to others upon request.
The mqjor contributors to this report are listed in appendix V. If you have
any questions about this report, please call me at (202) 6664026.

Harold A. Valentine
Associate Director, Administration
of Justice Issues

P4e18

Contents

Letter
Appendix1
FactorsThat
Contributedto
DifferencesinPrison
Construction Costs
Appendix11
FactorsThat
Contributed to
DifferencesinPrison
OperationsCosts
AppendixIII
The32StatePrisons
Submitting
QuestionnairesUsed
inAnalysisof
ConstructionCosts
AppendixIV
The21StatePrisons
Submitting
QuestionnairesUsed
in Analysis of
Operations Costs
AppendixV
Major Contributorsto
ThisReport

16
16
18
21
24
26

Amount of Space Provided to Inmates
TSpeof Structure
Design and Layout of Housing Units
Mix of Secwlty Levels
Geographical Location

29
31

Differences in Personnel Costa
Inmate-t&StaffRatios
Supplies and Services

34

36

38

39

40

GAWGGD-92-78 P&on

Page 14

:

Coot Factme

content4

Tables

Figures

Table 1.1:Costs per Bed Comparedto Gross SquareFeet
per Inmate
Table 1.2:Per Bed Construction Costs by Type of Structure
Table 1.3:Housing ConfiguxMon
Table 1.4:Distribution of Beda by Security he1
Table 1.5:GeographicalAreas and National Construction
Estimator Index Comparedto Costs per Bed and Coats per
Gross SquareFoot
Table II.1: AverageDaily Costa of Operations per Inmate
Table 11.2:PersonnelCosts
Table 11.3:PersonnelCosts per Inmate Day
Table 11.4:Inu~te-t~-Staf’fRatios
Table II.6: Daily per Inmate Operational ExpenseaOther
Than PersonnelCosts
Figure 1: AveragePrison Construction Costaper Bed
Figure 2: ComparisonBetween Averageper Bed
Construction Costa and Gross SquareFeet per Inmate in
State and Federal Prisons
Figure 3: Average Operations Costs per Inmate Day
Figure 4: ComparisonBetween the Inmate-to-Staff Ratio
and PersonnelCosts per Inmate Day
Figure 1.1:Building Con@urations
Figure 1.2:Typical Prison Housing Layouts

Abbreviations
ACA
BOP

FCI
GSF

American Correctional Association

Bureau of Prisons
Federal Correctional Institution
Gross SquareFoot

17
20
23
26
27
30
32
33
36
36
6
7
10
11
19
22

Appendix I

Factors That Contributed to Differences in
Prison Construction Costs
Officials managingthe acquisition of a new prison can directly influence
its cost through their control over the design of the facility and, to some
extent, where the facility is built. The 36 prisons (32 state, 4 federal) that
participated in our study reported a wide range in construction costs.
Total construction costs ranged from a low of $6,404,644(rated capacity
312) to a high of $266$66,796(rated capacity 2,916).Per bed construction
costs ranged from $11,243to $93,333-more than an eight-fold difference.
This section will provide some insights into the factors that contributed to
these differences.
To facilitate our analysis of the factors that affected construction costs, we
divided the prisons into three cost groups-low, medium, and high.
Natural breakpoints existed between the low and medium groups and
between the medium and high groups. Table I.1 shows the prisons that
comprise each cost group and arrays the prisons in ascending order by
cost per bed. This same ascendingorder will be used for the other tables
presented in this appendix.
Where appropriate, the tables also include totals, weighted averages,’and
medians for each cost group and for all 36 prisons.

Amount of Space
Provided to Inmates

The most important factor contributing to differences in prison
construction costs per bed was the amount of space provided, measured
in terms of gross square feet (GSF) per inmate. Our analysis showed that 96
percent of the variability in the cost per bed was due to the amount of
space provided. Table I.1 shows that as the amount of space provided per
inmate increases,the per bed costs of the prisons also tend to rise. This
increase in costs is especially dramatic when the lowest and highest cost
groups are considered, with the cost per bed of the high cost prisons
averagingalmost five times as much as the low cost group ($87,271vs.
$17,730).The relationship between space and cost is quite striking for
these cost groups, with the high cost prisons providing an averageof 664
GSF per inmate, over two and one-half times the averageof 216 GSF per
inmate provided at the low cost prisons.

‘To compute the weighted averqjes, the value of each item to be avera8ed (cost per bed, for example)
wee multiplIed by its weight (dee@ capacity) and the total of these products divided by the sum of the
weight8 (aggregate design capacity for all 86 prison@. Source for wei@ed average formula:
Fundamental Statistics for Bueineee and Economics, Third Edition, by John Neter and William
Wawem (Boston: Allyn and B

Page 16

GAWGGD-92-78 Pr&on Ckmt Factorm

0

Tablo I.1: Cooto par Bad Comparod to
Qroor Squaw Foe1par Inmato

Prleon name
Low coot prhonr

state

Varner
Calhoun
Chlppewa
McCormick
Evans
Allendale

AR
FL
MI
SC
SC
SC

Craggy

NC

Winslow
Weiahted averaaes

AZ

Coat par
bed

Grow quan foot
nor Inmato

$11,243
13,825
15,625
19,006
19,370
20.277
.
20,720
24,679
$17.730

194
219
185
220
220
220
220
251
215

$45,007
45,424
45,920
47.289
48,793
49,966
50.824
54,206
56,460
58,702
59,013
59,386
62,092
63,411
64,107
64,980
65,517
67,006
67,446
70,188
73,438
$58.282

411
423
260
615
447
597
460
521
413
414
402
540
481
431
588
355
507
404
671
562
627
475

Medium tort prlrone
Danville
Hill
Lorton
Avovelles
Illinois River
FCI Phoenix (BOP)
Western Illinois
Frackvllle
Dayton
Arkansas Valley
Ross
Smithfield
Carson City
Chuckawalla
Correctional Complex
Cayuga
EC. Brooks
Riverfront
FCI Marlanna (BOP)

EIY
FCI Sheridan (BOP)
Weiahted averaaes

IL
IL
DC
LA
IL
AZ
IL
PA
OH
co
OH
PA
Ml
CA
IN
NY
Ml
NJ
FL
NV
OR

(continued)

Page 17

GAO/GGD92-78

Prioon Coat Factarm

.

&w*

1

Fmtoro!llWCoatrlbute~dtoDlffemncah
Prtaonconatnsctloncow
.

Prlron name’
High coat prlaonr

state

Northern
Old Colony
FCI McKean (BOP)
Corcoran
Mule Creek
Eastern
Oshkosh
Weighted averages

NJ
MA
PA
CA
CA
KY
WI

Welahted averiwte). 36 Drl#onr

Cart per
bed

Grorr l quaro feet
Dar Inmate

$83,771
85,203
85,391
87,814
66.277
88,577
93,333
$67,271
556.374

389
565
670
524
624
634
619
554
435

Qtly the “short name”that diStingUlShe8 each facility from others in the same jurisdiction was
used in the tables. For example, Arizona State Prison Complex-Winslow is shown as Winslow,
and Pennsylvania’sState CorrectionalInstitutionat Frackville is shown as Frackville.Also, the 36
responding prisons are listed in ascending order of contruction costs per bed. The order is
retained in the subsequenttables in appendix I.

Another factor that we examined in terms of per bed construction costs
was the type of building structure. The National Directory of Corrections
Construction, published by the National Institute of Justice (April 1988),
classified prisons into the following general types (see fig. I.1 for

Type of Structure

illustration):
l
l
l

l

l

l
l

integrated structure-one building;
building, more than four stories in height;
ladder; telephone pole-linear cell blocks arranged parallel to one another
off a central connecting corridor;
wheel, spoke, or radial-linear cell blocks connected to one central
control area like spokes from the hub of a wheel;
courtyard-linear cell blocks interconnected around a central enclosed
courtyard;
clusters-a number of individual buildings that are interconnected, and
campus style-a number of individual buildings that are not
interconnected.
high rise-one

Y

Page 18

GAWWD-W-78

Prbon Cad Factma

flgun I.1: Bulldlng Conflgumtlonr
Campus

0
Ladder, telephone

I

cl

0

I

pole

Wheel, spoke or radial

L

I

I

Clusters

Courtyard

Page 19

UWGGtD-92-78

Prhon Cast Factors

Appentus I
hctom That ConMbuted to Dtmreneaa
Prtmm coMtrQedon casta

tn

According to the questionnaires,all of the prisons in &r sample were
either clusters, campus style, or integrated structures. Although per bed
construction costs varied for each design style becauseof factors such as
size and housing layout, the single building and cluster styles tended to be
more costly than the campus style. As table I.2 shows, although per bed
construction costs varied for each type of structure, integrated structures,
on average,were the most costly of the three types of structures, followed
by clusters and campus style. Of the seven high cost prisons in our sample,
five were single buildings or clusters. In .contrast, seven of the eight low
cost prisons were campusstyle.
Tablo 1.2:Pot Sod Conrtructlon Coats
by Typo of Structure
Prlron name
Low tort prlsonr

State

Varner
Calhoun
Chippewa
McCormick
Evans
Allendale

AR
FL
MI
SC
SC
SC

Cwxw

NC

Winslow
Weiclhted averages

AZ

Co& by type of structure
Single q Campur
bulldlng
rtylo
Cluatwr
$11,243
$13,825
15,625
19,006
19,370
20,277
20,720
24,679
$18,986

$11,243

Medium co&t prloonr
IL
IL
DC
LA
IL
AZ
IL
PA
OH
co
OH
PA
Ml
CA

Danville
Hill
Lorton
Avovelles
Illinois River
FCI Phoenix (SOP)
Western Illinois
Frackville
Dayton
Arkansas Vallev

Ross
Smithfield
Carson Citv
Chuckawalla

$45,007
45,424
$45,920

1,

$47,289
48,793
49,966
50,824
54,206
56,460
58,702
59,013
59,386
62,092
63,411
(continued)

GAO/GGD-92-78 Primn Co& Factam

Page 20

,^
,‘I’

;

Prloon name
Correctional Complex
Cayuga
EC. Brooks
Riverfront
FCI Marianna (BOP)

EIY
FCI Sherldan (BOP)
Weiahted averaaes

St&
IN
NY
Ml
NJ
FL
NV
OR

Coat8 by type of atructuro
Slnglo
Camp448
bulldlng
Cluoton
8W
64,107
64,980
65,517
67,006
67.446

70,188
73,436

$61.834

$61,140

$53,445

High coot prlronr
Northern
Old Colonv
FCI McKban (BOP)
Corcoran
Mule Creek
Eastern
Oshkosh
Welahted averaaes

Wolghtrd averages, 96 prlronr

Design and Layout of
Housing Units

NJ
MA
PA
CA
CA
KY
WI

$83,771
85.203

$85,391
$87,814
66,277
88,577

wm
$84,187

$86,781

$68,311

$73,55!5

$47,129

$64,012

The design and layout of the housing units is another important factor
affecting prison construction costs. Table I.3 shows that prisons with
higher percentages of cells designed to accommodate a single inmate tend
to cost more to build than prisons designed with multiple occupancy cells
and dormitories. For example, only about 4 percent of the beds in the low
cost prison group are in single cells, compared to about 72 percent for the
medium cost prisons and 99 percent for the high cost prisons. In contrast,
about 96 percent of the beds in the low cost prisons are either in multiple
occupancy cells or dormitories, compared to about 29 percent in the
medium cost prisons and 11 percent in the high cost prisons. Figure 1.2
illustrates typical housing layouts as examples of how prison designs csn
differ.

Page 21

GAOK+GD-92-78 Pttmn Coat Factmu

I,

Faetzm That Ckmtrtbutedw D@raneamin
Pri8on con8tllIedon corta

Pigun 1.2:Typltal Prison Housing Layout8

Linear, with Outside

Cells

Linear, with Inside Cells

Dormitory

Module/Pod

Page 22

MO&GD-92-72

Prbon Coat Factor6

Facton That Contibuted to DlffereneerIn
PllaoncanatnlcttonCosta ~

In August 1991,the Federal Bureau of Prisons (EIOP) adopted a limited
double celling design standard (two inmates per cell), but the design
capacities of the four medium security federal correctional institutions
(RI) included in our review were based on a single celling standard in
effect when the information was provided. Therefore, the percentagesin
the “Single cell” column of table I.3 would be expected to be 100 for each
of the federal prisons. However, three of these projects included an
adjacent minimum security camp which housed inmates in dormitories.2
For the three rc~~,BOP was unable to separate the construction costs of the
medium security prisons from the minimum security camps. Consequently,
we showed the prisons and the camps as single units, resulting in the
housing configuration percentagesshown in table 1.3.
Table 1.3:Housing Conflguratlon (Design)
Prlron name
Low cost prisons

Total bed8
(rated
State capaclty)

Varner
Calhoun
Chippewa
McCormick
Evans
Aliendale

AR
FL
MI
SC
SC
SC

fWxiy

NC

Winslow

AZ

TotaWwelghted averager

Cart per
bed

Slngle cells
Beds Percentage

Multiple occupancy
cells
Beds Percentage

Dormltorlw
Beds Percentage

1,100
768
640
1,104
1,104
1,104
312
650
6,782

$11,243
13,825
15,625
19,006
19,370
20,277
20,720
24,679
$17,730

0
0
0
96
96
96
0
0
268

0
0
0
9
9
9
0
0
4

0
0
640
1,008
1,006
1,008
0
400
4,064

0
0
100
91
91
91
0
62
60

1,100
768
0
0
0
0
312
250
2,430

100
100
0
0
0
0
100
38
36

896
896
400
610
787
518
728
504
498
724

$45,007
45,424
45,920
47,289
48,793
49,966
50,824
54,206
56,460
58,702

896
896
192
78
787
518
726
504
498
724

loo
100
48
13
100
100
100
106
100
100

0
0
0
52
0
0
0
0
0
0

0
0
0
9
0
0
0
0
0
0

0
0
208
480
0
0
0
0
0
0

0
0
52
79
0
0
0
0
0
0
(continued)

Medium cost prlsons
Danville
Hill
Lorton
Avoyelles
Illinois Piver
FCI Phoenix (BOP)
Western Illinois
Frackville
Dayton
Arkansas Valley

IL
IL
DC
LA
IL
AZ
IL
PA
OH
co

?he minimum security camp a&went to FCI Phoenix was built as a separate construction project

GAWGGD-92-72 Prison Coat Factan

Page 22

.,,

1,

Prleon name

wet.

I3089

Smlthfleid
Carson City
Chuckawalla
Correctional Complex
Cayupa
EC. Brooks
Rlverfront
FCI Marlanna (BOP)

w

FCI Sheridan (BOP)

OH
PA
MI
CA
IN
NY
Ml
NJ
FL
NV
OR

TotaWkel~hted . l vemgw

Total bed@
(rated
capaolty)
1,268
448
612
2,ooo
716
756
580
462
698
476
752

Yultlplo~~~pancy
Sing10 oelle
Bedr Peroontago

Coat pr
bed

11,319

59,013
59,386
62.092
63,411
64,107
64,980
65,517
67,006
67,446
70,168
73,438
$58,282

1,047
428
646
2,916
1,700
500
300
7,537
29,638

1,008
448
612
0
716
0
580
462
550
290
496

Bodr Percontago
0
0
0
1,992
0
0
0
0
0
186
0
2,230

0
0
0
~100
0
0
0
0
0
39
0

10,983

80
100
100
0
100
0
1’00
100
79
61
66
72

$83,771
85,203
85,391
07,014
88,277
88,577
93,333
$87,271

1,007
428
496
2,524
1,500
500
300
6,755

96
100
77
67
88
loo
loo
90

$56,374

18,026

61

Dormltorlee
Bedr Percentage
250
0
0
8
0
756
0
0
148
0
266

20
0
0
0
0
100
0
0
21
0
34

lb

2,106

14

40
0
0
0
0
0
0
40

4
0
0
0
0
0
0

0
0
23
13
12
0
0

1

0
0
150
392
200
0
0
742

10

6,334

21

5,270

18

High ooat prloonm
Northern
Old Colony
FCI I&Keen (6OP)
Corcoran
Mule Creek
Eastern
Oshkosh

NJ
MA
PA
CA
CA
KY
WI

TotaWwelghtod avenger
Td$Iw$$ded
P

l voragee,

Note: Percentages may add to more than 100 due to rounding.

lklix of Security Levels

Although each of the prisons in our sample was designed to house a
predominantly medium security population, some were also designed to
accommodate minimum security and/or maximum security inmatea 83s
well. Our analysis found that building a prison to accommodate a mix of
security levels tended to add to construction costs.
The prisons in table I.4 are listed in ascending order of construction costs
per bed. The tab!e shows that construction costa per bed tended to
increase as the percentage of medium security beds declined. Overall, the

GAMWD-92-78

Page 24

,.
>’
,(

“
/,’

Prlaon Cwt Factarm

‘,

prisor~ classified 76 percent of their beds as medium security,
compared to 89 percent for the medium cost prisons and 00 percent for
the low cost prisons.

high cost

Tablo 1.4:Dlatrlbutlon of Bedr by Socurlty Lwol
Tot81 hod,
Mlnlmum
(rated
State capacity) Bodr Porcentaao

Prleon nam3
Low tort prlrono
Varner
Calhoun
Chiopewa
McCormick
Evans
Allendale
Craoav

AR
FL
MI
SC
SC
SC
NC

Winslow

AZ

Total@/
Pclrcentaaoo

1,100
768
640
1,104
1,104
1,104
312
650

400

6.782

400

36

6

Numkr of bode at oath wourlty Iovol
Medium
Maximum
Beds Porcontaao Bodr Porwntaae
700
768
640
1,608
1,008
1,008
312
650

64
100
100
91
91
91
100
100

6.064

90

896
896
208
610
787
518
728
504
466
724
944
448
360
1,992
716
756
360
441
496

100
100
52
100
100
100
100
100
94
100
75
100
59
loo
loo
loo
62
95
71

96
98
Q6

s
9
9

266

4

192

48

Other
Bode Porwntaao

Medium oort prlronr
Danvilie
Hill
Lorton
Avovelles
Illinois River
FCI Phoenix (BOP)
Western llllnols
Frackvllle
Dayton
Arkansas Valley
Ross
Smlthfleld
Carson’Citv
Chuckawalla
Correctional Complex
Cayuga
EC. Brooks
Riverfront
FCI Marianna (BOP)

*

IL
IL
DC
LA
IL
AZ
IL
PA
OH
co
OH
PA
Ml
CA
IN
NY
Ml
NJ
FL

896
896
400
610
787
518
728
504
498
724
1,258
448
612
2,000
716
756
580
462
698

250

20

60
8

10
0

60

10

148

21

192

31

160

28

54

32

6

64

5

21

5

8
(continued)

Page 26

,.

ApQendtx I
Fwtor~ That Contributed to DlfPereneer in
Prloon CoMtnledon co&8

Total beda
M lnlmum
(rated
State capacity) Sedr Percentage
NV
478
OR
752 266
34

Prlron name@
EIY
FCI Sheridan (BOP)

Total@/
prcontagor

18,319

782

1,047

40

Numkr of bodr at aach aecurlty level
Medium
Maxlmum
Other
Bed8 Percentage Bedo Percentage Bedo Percentage
266
60 190
40
496
66

5 13,632

89

788

5

117

1

92

30

3

17

2

51 1,024

35

High coot prloonr
Northern
Old Colony
FCI McKean (BOP)
Corcoran
Mule Creek
Eastern
Oshkosh

NJ
MA
PA
CA
CA
KY
WI

Total@/
percantagrr
Total8/pwcontager, 36
prlrona

428
646

4

150

23

960
428
496

2,916
1.700

392
200

13
12

1,500
1,500

500
300

500
300
7,537

782

29,638 1,964

10

loo

77
88
loo
loo

5,684

75 1,054

14

17

0

7 25,410

86 2,130

7

134

0

‘As pointed out previously,the prisons are arranged in order from lowest constructioncost per
bed to highest constructioncost per bed. See table I.1 for specific cost per bed Informatlon.

Geographical
Lwation

Y

Prison construction costs can also be affected by geographic location.
According to the National Institute of Justice and the American
Correctional Association (ACA), construction costs can vary from one part
of the country to another due to sharp contrasts in the cost of materials
and prevailing labor rates. For example, according to the National
Construction Estimator indexes for m id-1989,construction costs tended to
be higher in the Northeast and West and lower in the South and M idwest.
The prisons in our sample generally reflected those tendencies. Table 1.6
shows that of the 8 low cost prisons, 6 were in the South, while only 3 of
the 21 medium cost and 1 of the 7 high cost prisons were in the South.
Conversely,there were no Northeast prisons in the low cost group, while
four Northeast prisons were in the medium cost group and three in the
high cost group.
Severalcompaniespublish construction cost indexes that allow cost
estimators to adjust for regional differences in the costs of labor, material,
and equipment.Table I.6 shows the 1989“National Construction

Phge 26

GAWGGD-92-78 Primon Co& Fmtorm

Elstimator”index for each of the states in which the sample of state
prisons were located. This index allows interested parties to make cross
jurisdictional comparisons of construction ~osts.~Our analysis showed that
as the estimator index for the state prisons in our sample increased,the
cost per bed also tended to increase.The median index for the low cost
prisons is 31, compared to 1.06for the medium cost group and 1.17for the
high cost group.
Another indication of the importance of geographic location is its effect on
cost per GsF.The cost per GsFis, in effect, the measure of the amount of
space the jurisdiction was able to buy for its money, independent of the
number of inmates the prison was designedto house. Table I.6 shows that
the cost per GSF at the 36 prisons ranged from $68.06to $216.60and
averaged$129.48.Further, the table shows that as costs per GSF increased,
costs per bed also tended to increase. We found that about 96 percent of
the variability in cost per GSF was explained by the national construction
estimator index, which is a surrogate measure for the state in which the
prison is built.
Table 1.5:Geogmphlcal Areas and Natlonsl Constructlon Estlmator Index Compared to Costs per Bed and Costs per Gross
Square Foot
Prison name
State
Costs per bed
U.S. Region
Index Costs per GSF
Low co8t prlrons
Vamer
Calhoun
Chlppewa
McCormick
Evans
Allendale

AR
FL
MI
SC
SC
SC

Cww

NC

Winslow
Weighted average
Median index

AZ

$11,243
13,825
15,625
19,006
19,370
20,277
20,720
24,679
$17,730

South
South
Midwest
South
South
South
South
west

.83
-90
39
.78
-78
.78
.79
1.01

$58.06
63.22
84.45
86.48
88.14
92.27
94.03
98.20
$82.40

.81

Medlum:cost prlrons
Danville
Hill

IL
IL
”

$45,007
45,424

Midwest
Midwest

1,06
1.06

$109.63
107.31
(continued)

The following example Illustrates how the estimator Index works. The Correctional Industrial
Complex in Indiana coat about $46,900,000.If the aame prlaon had been built in California in the same
year, the Index indicates it would have coat about $60,369,0oo(1.211.92 = 1.315 x ~6,900,000 =
$60,363,600).On the other hand, lf the aame facility had been build in South Carolina, the index
indicatea it would have cost about $33,900,000(.78 / .92 = .S43 x $46,900,000= $33,923,000).

Page 27

GAO/GOD-92-73 Prison Coat Factora

l

Prbon name

St&

Lorton
Avwelles
Illinois River
FCI Phoenix (BOP)
Western llllnols
Frackville
Dayton
Arkansas Valley

DC’
IA
IL
AZ
IL
PA
OH
co
OH
PA
MI
CA
IN

Ross
Smlthfleld
Carson City
Chuckawalla
Correctional Complex

Cayuw

NY

EC. Brooks
Riverfront
FCI Marlanna (BOP)

Ml
NJ
FL
NV
OR

EIY
FCI Sheridan (BOP)
Welghted average
Median index

Coot8 per

bed

45,920
47.289
48,793
49,966
50,824
54,206
56,460
58,702
59,013
59,386
62,092
63.411
84,107

64,980
65,517
67,006
67,446
70,188
73,438
$58,282

U.S. Region

South
South
Midwest
west
Midwest
Northeast
Midwest
West
Midwest
Northeast
Midwest
West
Midwest
Northeast
Midwest
Northeast
South
West
West

Index

Co&r per GSF

.92
.a5
1.06
1 .Ol
1.06
1.07
1.10
1.07
1.10
1.07
99
1.21
-92
1.13
99
1.17
90
1.21
1.04

176.35
76.93
109.12
83.73
110.38
104.06
136.86
141.67
146.85
110.01
129.17
146.96
109.09
183.21
129.17
165.81
10059
124.78
117.05
$122.77

1.06

High oort prlronr,
Northern
Old Colony
FCI McKean (BOP)
Corcoran
Mule Creek
Eastern
Oshkosh
Welphted average
Median index

NJ
MA
PA
CA
CA
KY
WI

$83,771
85,203
85,391
87,814
88,277
88,577
93,333
$87,271

Northeast
Northeast
Northeast
west
West
South
Midwest

1.17
1.19
1.07
1.21
1.21
.91
1 .Ol

$21550
150.69
127.42
167.62
141.56
139.71
150.66
$157.64

1.17

Woightod average, 36 prlsono
Mddlrn Index, 36 prlwn8

$56,374

$129.48

1.06
The District of Columbia’s Lorton facility is located in suburban Virginla.

Page 28

WGGD-92-72

Primon ihta

and I&tom

.

rs That Contributed to Differences in
Prison Operations Costs
Operationscostaat our sampleprisons varied signiilcsntly, although not
to the extent of the differences in construction costs discussed in
appendix I. At our sample of 23 prisons (2 federal, 21 state), operations
costs per inmate dsy ranged from $22.26to $81.98,with a weighted
average of $41.93.
The following tables will show that the operations cost differences were
due mostly to differences in salaries and related expenses,staffing levels
relative to inmate population, and amounts paid for supplies, materials,
food, and services.
Because operationscostavaried so widely, we divided the prisons in our
sample into three cost groups for analysis purposes- low, medium, and
high. We used breakpoints that existed in the daily operations costs per
inmate to define the three cost groups. We defined prisons with daily
operations cost per inmate in the $22 to $37 range as low cost, those in the
$42 to $61 range as medium cost, and those in the $69 to $81 range ss high
cost. Table II.1 shows the prisons in each cost group, listed in ascending
order by average daily cost’per inmate. There wss no direct relationship
between the cost groups described in this section and the construction
cost groups discussed in appendix I.

Page 29

G,UM3GD-S2-78 Rbon

Coat Factora

Factmu That Contrtbnted
Priwn Op4mttonm Coota

to Diibmmcea in

Teblo 11.1:Avoraao Dally Coetr of Oueretlonr per Inmato
Prlron name
Low coat Prleonr
Ross

State

Annual l xpeneer

McCormick
Chlppewa
Calhoun
Frackville
FCI Marianna (BOP)

OH
SC
MI
FL
PA
FL

$13,709,314
9,184,304
9,933,378
9,363,992
11,762,OOO
13,230,154

Crams

NC

Hutchinson
Arkansas Valley
FCI Phoenix (BOP)
Hill
Danville

KS
co
AZ
IL
IL

$129,904,910

lotalo/Melghted average,

Avenge dally
population
1,688
1,075
946

Costs per inmate
Per year

Costs per Inmate
per day

$8,122

$22.25

8,544

23.41
28.77

900
1,000

10,500
11,793
13,069
13,230

36.25

30844.809

288

13.350

36.58

5,372,376~

400

13,431

36.80

12,694,OOl
14,781,482
12.947.700
13,081,400

935
1,078
944
946

13,576
13,712
13.716

37.20

13,828

10,994

$11,818

$32.37

470

$15,374

$42.12

300

15,460

42.36

794

32.31
35.81

37.57
37.58
37.89

Medium coot prlronr
Dayton
Al Burruss
Mule Creek
Smithfield
Corcoran
Oshkosh
Old Colony
Cayuna

OH
GA
CA
PA
CA
WI
MA
NY

$7,225,893
4,637,974
50.020.790
7,332,OOO
82,538,576

7,141,779
10,787,163
17,651,991

Totalsiwetlghtedaverage8

$187,33&l 66’

3.204

15.612

42.77

450

16,293

44.64

4,838

17,060

46.74

399

17,899

49.04

589
950

18,314
18,581

50.18
50.91

11,200

$16,726

$45.93
6

Hlah coat prloone
Eastern
Northern
Riverfront

MD
NJ
NJ

TotaWwelahtod avoraclee
ToteWwelghted avrrager, 23
prlronr

$31,189,074
22,461,OOO
13,QO8,500

1,440
1,037

$21,659

21,660

59.34

470

29,593

81.08

S 87.558.574

2.947

$22.925

$82.81

$394,799,650

25,141

$16,309

$41.93

$59.34

% this table, the 23 responding prisons are listed in ascending order of the daily costs of
operations per inmate.This order is used in the subsequenttables in appendix II.
I

Page 80

GAWGGD-92-78 Primon Co& Factora

Appsndlx II
Footon That Cantrlbutad
Prhon oporatl0M cooto

Differencesin
PersonnelCosts

to Milerenasr

Lr

The single largest expenseof operating a prison is the cost of personnel
compensation.As table II.2 shows, personnel costs ranged from 66 percent
to 93 percent of total operations costs, with an an overall averageof 76
percent.

Page 31

GMUGGD-92-72 Prison Co& Factore

Tabla 11.2:Pereonnol Coat0
Prlron nom0
Low cod prl8ono
Ross
McCormick
Chlppewa
Calhoun
Frackviiie
FCI Marianna (BOP)
Craaav
Hutchinson
Arkansas Valley
FCI Phoenix (BOP)
Hill
Danvilie

stmto
OH
SC
MI
FL
PA
FL
NC
KS
co
AZ
IL
IL

Totals/weighted averages

Poraonnrl cost8
$10,293,742
6,163,171
6,360,376
6.659067
7,943,ooo
6,661,206
3,048,585
4,039,103
9,102,164
9,561,593
6.801 .ooo
8,824,900

$91,476,931

Poreonnol cortr a0
percentage of total
cortr
75
67
64
71
68
65
79
75
72
65
66
67
70

Medium tort prlronr
Dayton
Al Burruss
Mule Creek
Smithfield
Corcoran
Oshkosh
Old Colony
Cavuga

OH
GA
CA
PA
CA
WI
MA
NY

$5,559,613
3,790,863
36,301,226
5,275,OOO
64,682,355
5599,232
10,050,670
14,925,770

$14&l 94,749

TotalaIw~lghkd rvera&r

77
62
77
72
76
78
93
65
79
8

Hlah co81 prlrons
Eastern
Northern
Riverfront

Totala/welghted avoragw
TotaNwolghtod avoragw,
23 prlaonr

MD
NJ
NJ

2 50,127,160

71
75
80
74

$289,788,640

75

$22,129,160
16,687,OOO
11,111,OOO

Differences in personnel costs did not account for all of the variances and
were not always consistent with differences in overall costs. For exainple,
table II.2 shows that personnel costs comprised 79 percent of total costs at

Pago a2

GAOIGGD-92.72 Primon Coat Factora

the medium cost prisons, compared to 74 percent at the high cost prisons.
Table II.3, however, shows that daily personnel costs per inmate at the
high cost prisons were about $10 higher than at the medium cost prisons
due to staffing levels relative to inmate populations (see following
section).
Table 11.3:Pwoonnal Coti
av

pof Inmab
Prloon nrmo
Low coat
prbno
Ross
McCormick
Chippewa
Calhoun
Frackviiie
FCI Marianna
WP)
Craaov
Hutchinson
Arkansas Valley
FCI Phoenix
(BOPI
Hill
Danvliie

SUtO

OH
SC
MI
FL
PA
FL
NC
KS
CO
AZ
IL
IL

TotalWelghtod
avorager

PoraonnoI cortr

Avonga dally
populrtlon

Poraonnol coata
per Inmato day

$10,293,742
6,163,171
8,380,378
6658,087
7,943,ooo

1,688
1,075
946
794
900

$16.71
15.71
24.27
22.97
24.18

8661,208
3,046,565
4,039,103
9,102,164

1,000
286
400
935

23.73
29.00
27.67
26.67

9.561.593
8,801,OOO
8,824,900

1,078
944
946

24.30
25.54
25.56

8 91,476,931

10,994

822.88

$5559,613
3.790.883
38,301,226
5,275,OOO
64682,355
5,599,232
10,050,670
14,925,770

470
300
3,204
450
4,836
399
569
950

832.41
34.62
32.75
32.12
36.63
38.45
46.75
43.04

6146,164,749

11,200

836.25

Modlum coot
prlronr
Dayton
Al Burruss
Mule Creek
Smithfield
Corcoran
Oshkosh
Old Colony
Cayuga

TotaWwolghted
avomger

OH
GA
CA
PA
CA
WI
MA
NY

(continued)

Page88

-D-82-78

Prhon ChtFactom

8

lkotmn ‘hat Contributed
Prlum op8miOM co8t8

Prktn nom8

stata

to DIlYorenw

In

Average dally
population

Pamonnel coat8
per lnmata day

$22,129,160
16,887,OOO
11.111,000

!,440
1,037

$42.10
44.61

470

64.77

$60,127,100

2,947

$46.60

5289.788.540

25.141

S31.66

Pomonnal coat8

High coat
wlaona
Eastern
Northern
Rlverfront

TotaWWalghtad
average8
TotaWvelghted
rvaragaa, 23
Drlaona

Inmate-to-StaffRatios

MD
NJ
NJ

The stxifEi level of a prison relative to it9 inmate population
(inmate-to-staff ratio) is an important factor in accounting for differences
in personnel costs. For example, table II.4 shows that the low cost prisons
reported an averageinmate-to-staff ratio of 3.13 to 1, compared to 2.71to 1
for the medium cost prisons and 1.76to 1 for the high cost prisons. This
clearly shows that the prisons in our sample that employed more staff
relative to their inmate populations (i.e., those with lower inmate-to-staff
ratios) tended to incur higher operational costs.

Page a4

6lAWGGD-92-72 Prbon Coat Factarm

Tabla 11.4:Inmate-to-Staff Ratloa
Prlaon name
Low ooat prlaona
Ross

stat8

McCormick
Chlppewa
Calhoun
Frackvilie
FCI Marianna (BOP)

OH
SC
MI
FL
PA
FL

Craggy

NC

Hutchinson
Arkansas Valley
FCI Phoenix (BOP)
Hill
Danvilie

KS
co
AZ
IL
IL

TotaWwalghted avaragaa

Authorized
atatf

Avamga dally Inmate-topopulatlon staff rat108

377
342
217
266
253
295
132
173
301
322
404
428

1,666
1,075
946
794
900
l,ooO
266
400
935
1,078
944
946

4.48
3.14
4.36
2.98
3.56
3.39
2.18
2.31
3.11
3.35
2.34
2.21

3,510

10,994

3.13

220
157
870
267
1,582
188
370
457

470
300
3,204
450
4,838
399
569
950

2.14
1.91
3.68
1.57
3.66
2.12
1.59
2.08

4,131

11,200

2.71

791
566
323

1.82
1.83
1.46

1,682

1,440
1,037
470
2,947

9,323

25,141

Madlum coat prlaona
Dayton
Al Burruss
Mule Creek
Smithfield
Corcoran
Oshkosh
Old Colonv
Cayuga

OH
GA
CA
PA
CA
WI
MA
NY

TotaWweightad average8
Hlgh coat prlaona
Eastern
Northern
Riverfront

TotaWwalghtad avaragea
TotaWwalghtad averagea,
23 prlaona

Supplies and Services

MD
NJ
NJ

most of the prisons in our sample was supplies, material, and food.
Although there were notable differences in the amounts spent by

Pqe a6

1.711
2.70

individual prisons, the medium and high cost prisons tended to spend
more in this category. For example, table II.6 shows that the low cost
prisons spent an averageof $4.76per inmate day for supplies, materials,
and food, compared to $6.24at the medium cost prisons and $7.22at the
high cost prisons.
Table II.6 also shows that the high cost prisons spent more than the other
cost groups on services.1Servicesat the high cost prisons amounted to
$6.47per inmate day, compared to $2.62at the low cost prisons and $.86 at
the medium cost prisons.
Table II.5: Dally par Inmato OperatIonal Expenwa Other Than Petwonnol Coat8
Rent,
Staff communlcatlonr~
Prloon nom0
travel
utllltllt,
Setvlcer
St&

SuptWo,
miierlal;
food Equlpmant

Other

Total
$5.54
7.70
4.50
0.34

Law coat arlaena

McCormick
Chlppewa
Calhoun
Frackville
FCI Marlanna (BOP)

OH
SC
MI
FL
PA
FL

$0.01
0.02
0.09
0.10
0.05
0.43

Cww

NC

0.06

Hutchinson
Arkansas Valley
FCI Phoenix (BOP)
Hill
Danvllle
Weighted averages
P@rcentages

KS
co
AZ
IL
IL

ROSS

3.42
0.00
2.02
6.12
2.01
1.74
0.60
0.39
4.22
4.71
4,66
$2.52
26

$3.72
2.57
3.74
5.06
3.63
6.32
4.14
5.95
7.97
6.06
4.32
4.45
$4.75
50

$0,05
0.03
0.10
0.15
0.10
1.04
0.02
0.05
0.43
0.64
0.19
0.30
$0.28
3

$0.00
0.05
0.00
0.58
0.03
0.07
0.00
0.00
0.09
0.16
0.66
0.74
$0.20
2

$0.56
0.69
0,.46
3.65
0.80
1.49

$4.92
4.74
5.35
6.03
5.46
6.36

$0.11
0.23
0.74
1.52
0.32
0.10

$1.96
0.00
1.03
0.02
1.32
0.11

$1.58

$0.16

0.14
0.03
0.45
0.04
0.04
$0.12
1

1.62
0.57
1.44
1.49
2.65
1.62
2.19
1.61
1.71
2.12
2.13
$1.71
18

$0.06
0.02
0.20
0.12
0.27
006

$2.09
2.06
2.24
1.19
1.94
2.45

11.63
12.52
7.57

9.13
10.52
13.27
12.03
12.33
$B.W

100

Medium coot prlronr
Dayton
Al Burruss
Mule Creek
Smlthfleld
Corcoran
Oshkosh

u

OH
GA
CA
PA
CA
WI

5 9.71
7.74

10.02
12.52

10.11
10.59
(continued)

%ervices include such expenses ae trash dispoeal, laundry and dry cleaning, repair and maintenance of
equipment, and medical treatment from outside sources.

QAWGGD.92.78 Priooa Co& Facton

Page 86

”

6

Prleon name,

state

Old Colonv
Cavuaa
- -. -wWeighted averages
Percentages

MA
NY

Staff
_ .-..
tnvol

Rant,
communicntlona.
__.~...~. ~~~~ ~,
utllltler

supplier,
material.
SOWICOO
tooi Equipment

0.12
0.02
$0.19
2

0.21
1.09
$1.86
19

1,08
0.97
$0.86
9

1.60
5.48
$5.24
55

$0.17
0.02
0.00
$0.09
1

$2.57
0.85
1.22
$1.75
11

$7.42
5.31
6.13
$6.47
40

$0.15

$1.78

$2.24

Other

Total

0816
0.15
$0.45
5

0.26
0.15
$0.98
10

3.43
7.86
$9.53

$6.05
8.10
8.85
$7.22
45

$0.28
0.41
0.03
$0.29
2

$0.75
0.03
0.07
$0.39
2

$17.24

$5.26

$0.35

$0,57

$10.35

100

Hlgh ooet prleonr
Eastern
Northern
RIverfront
Weiohted averaaes
Percentages

WzfLtz

rvoragw,

Parcantaaar

MD
NJ
NJ

14.73
16.31
$16.21

100

23

1

Page 27

WGGD-92-78

Pr&n

Costa and Pactmu

Appendix JII

The 32 State Prisons Submitting
QuestionnairesUsed in Analysis of
Construction Costs
state
Arizona
Arkansas
California

Colorado
District of Columbia
Florida
Illinois

Indiana
Kentucky
Louisiana
Massachusetts
Michigan

Nevada
New Jersey
New York
North Carolina
Ohio
Pennsylvania
South Carolina

Wisconsin

Page 813

Prlmn
Arizona State Prison Complex - Winslow, Winslow
Varner Unit, Grady
California State Prison, Corcoran
Chuckawalla Valley State Prison, Blythe
Mule Creek State Prison. lone
Arkansas Valley Correctional Facility, Crowlsy
Modular Facility, Lorton, Virginia
Calhoun Correctional Institution, Blountstown
Danville Correctional Center, Danville
Hill Correctional Center, Galesburg
Illinois River Correctional Center, Canton
Western Illinois Correctional Center, Mt. Sterling
Correctional Industrial Complex, Pendleton
Eastern Kentucky Correctional Complex, West Liberty
Avoyelles Correctional Center, Cottenport
Old Colonv Correctional Center. Bridnewater
E. C Brooks Regional Facility, Muskegon
Carson City Regional Facility, Carson City
Chippewa Temporary Correctional Facility, Kincheloe
Elv State Prison, Elv
Northern State Prison, Newark
Riverfront Correctional Facility, Camden
Cayuga Correctional Facility, Moravia
Craggy Correctional Center, Asheville
Dayton Correctional Institution, Dayton
Ross Correctional Institution, Chillicothe
State Correctional Institution at Frackville, Frackville
State Correctional Institution at Smithfield, Huntingdon
McCormick Correctional Institution, McCormick
Allendale Correctional Institution, Fairfax
Evans Correctional Institution, Bennettsville
Oshkosh Correctional Institution, Oshkosh

GLWGGD-92-78 Prison Comt Facto-

The 21 State ‘Prisons Submitting
Questionnaires Used in Analysis of
Operations Costs
state
California
Colorado
Florida
Georaia
lllinols
Kansas
Marvland
Massachusetts
Michigan
New Jersey
New York
North Carolina
Ohio
Pennsylvania
South Carolina
Wisconsin

Pype 2B

Prkn
California State Prison, Corcoran
Mule Creek State Prison, lone
Arkansas Valley Correctional Facility, Crowley
Calhoun Correctional Institute, Blountstown
Al Burross Correctional ‘Trainina Center, Forsvth
Danville Correctional Center, Danville
Hill Correctional Center, Galesburg
Hutchinson Correctional Work Facility, Hutchinson
Eastern Correctional Institution. Westover
Old Colony Correctional Center, Bridgewater
Chippewa Temporary Correctional Facility, Kincheloe
Northern State Prison, Newark
Riverfront Correctional Facility, Camden
Cayuga Correctional Facility, Moravia
Craggy Correctional Center, Ashville
Dayton Correctional Institution, Dayton
Ross Correctional Institution, Chillicothe
State Correctional Institution at Frackville, Frackville
State Correctional Institution at Smithfield, Huntingdon
McCormick Correctional Institution. McCormick
Oshkosh Correctional Institution. Oshkosh

GAWGGD-92-72 Prhon Cost Factora

AppendixV

Major Contributors to This Report

General Government
Division, Washington,
D.C.

Rkhard M. Stana,AmWant Director,Administrationof JusticeIsauea
JoanneParker,SeniorSocialScienceAnalyst
BonnieSteller,SeniorStatitician
Kim Wheeler,PublishingAdvisor

Los Angeles Regional
Office

DannyM. Bullock, Evalua~r-in-charge

(lanSl9)

Page 40

Ordering

Information

The first copy of each GAO report and testimony is free. Additional
copies are $2 each. Orders should be sent to the following address,
accompanied by a check or money order made out to the Superintendent of Documents, when necessary. Orders for 100 or more
copies to be mailed to a single address are discounted 25 percent.
I J.S. General Accounting Office
I’.(). 130x 6015
Gaithersburg,
MD 20877
Orders may also be placed by calling

(202) 2756241.

_.. .“, .” “,

United States
General Accounting
Office
Wwhington,
I).(?. 20548
Official
Penalty

_

Husiness
for Private

,,“....”.“_.I.---

Fir-&Cl&s
Mail
Postage 8~ Fees Paid
GAO
Permit No. GlOO

lJse $300

.._..._ .. - _ ___...-.

.~ .-.“. .^-~

---