Defense IRM: Alternatives Should Be Considered in Developing the New
Civilian Personnel System (Letter Report, 01/27/99, GAO/AIMD-99-20).

Pursuant to a congressional request, GAO reviewed the Department of
Defense's (DOD) efforts to reduce the costs associated with civilian
personnel management, focusing on: (1) how DOD determines the number and
locations for civilian personnel regional service centers and why is
there a wide disparity in the number of regional centers among the
services; (2) whether DOD is applying the investment principles of the
Clinger-Cohen Act in overseeing, managing, and developing the Defense
Civilian Personnel Data System (DCPDS); (3) whether DCPDS duplicates the
Office of Personnel Management's (OPM) Employee Express System: (4)
whether DOD leadership is aware of the extent and cost of the needed
modifications to the commercial-off-the-shelf (COTS) software
applications; and (5) whether DOD identified and mitigated the risks
associated with the major COTS modifications.

GAO noted that: (1) DOD's current initiative can potentially improve
civilian personnel operations and achieve cost savings; (2) however,
because the Department has not examined other business process
alternatives that could have potentially achieved even greater savings
and process efficiencies, there is no assurance that this is the best
alternative for civilian personnel operations; (3) before embarking on
its costly initiative to improve personnel management, DOD examined two
alternatives: (a) outsourcing personnel computer operations to the
Department of Agriculture's Finance Center; and (b) regionalizing
personnel centers; (4) DOD determined that it would take the National
Finance Center about 6 years to prepare for transferring computer
operations and that some new functionality built into its legacy system
would be lost; (5) however, DOD did not examine several other
potentially effective alternatives, including: (a) continuing to
centralize all or parts of its personnel management operations to reduce
duplicative layers of oversight at the components and ensure more
consistent operations DOD-wide; (b) integrating its personnel and
payroll management systems; (c) restructuring its regional offices to
serve multiple components rather than perpetuating regional offices
dedicated to only one component; (d) restructuring local personnel
offices to serve multiple bases or installations (they now serve only
one base or installation); and (e) outsourcing all civilian personnel
operations to the private sector; (6) these alternatives are feasible
and may have helped DOD to achieve even greater savings and efficiencies
than the current approach; (7) in addition, the Defense Science Board
determined that integrating payroll and personnel systems was a viable
and cost beneficial option for military personnel; (8) the Civilian
Personnel Management Service (CPMS) officials who were responsible for
the personnel initiative said that they did not consider these business
processing alternatives because: (a) CPMS did not have authority to
require the military services and DOD agencies to adopt such approaches;
(b) DOD did not allow sufficient time to rigorously examine
alternatives; and (c) DOD lacked basic cost and performance data needed
to study the alternatives; (9) after it decided on its approach, DOD did
not follow a sound process for selecting regions; (10) DOD did not
adequately consider a full range of technical options before deciding to
replace its legacy system with the Oracle COTS product; and (11) after
DOD acquired the Oracle system, it did not mitigate critical technical
risks.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  AIMD-99-20
     TITLE:  Defense IRM: Alternatives Should Be Considered in 
             Developing the New Civilian Personnel System
      DATE:  01/27/99
   SUBJECT:  Cost effectiveness analysis
             Personnel management
             Management information systems
             Defense cost control
             Commercial products
             Computer software
             Productivity in government
             Information resources management
             Civilian employees
             Privatization
IDENTIFIER:  DOD Defense Civilian Personnel Data System
             OPM Employee Express System
             Y2K
             
******************************************************************
** 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:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to the Subcommittee on Personnel, Committee on Armed Services,
U.S.  Senate

January 1999

DEFENSE IRM - ALTERNATIVES SHOULD
BE CONSIDERED IN DEVELOPING THE
NEW CIVILIAN PERSONNEL SYSTEM

GAO/AIMD-99-20

Defense Civilian Personnel Management

(511634)


Abbreviations
=============================================================== ABBREV

  AFB - Air Force Base
  AFOTEC - Air Force Operational Test and Evaluation Center
  CDA - Central Design Activity
  COTS - commercial-off-the-shelf
  CFO - chief financial officer
  CPMS - Civilian Personnel Management Service
  DCPDS - Defense Civilian Personnel Data System
  DISA - Defense Information Systems Agency
  DOD - Department of Defense
  FASA - Federal Acquisition Streamlining Act of 1994
  GPRA - Government Performance and Results Act of 1993
  I-CASE - Integrated Computer-Aided Software Engineering
  IDIQ - indefinite delivery, indefinite quantity
  IT - information technology
  MAISRC - Major Automated System Review Council
  OMB - Office of Management and Budget
  OPM - Office of Personnel Management
  OSD - Office of the Secretary of Defense
  PA&E - Program Analysis and Evaluation
  PRA - Paperwork Reduction Act of 1995
  USDA - Department of Agriculture

Letter
=============================================================== LETTER


B-278058

January 27, 1999

The Honorable Wayne Allard
Chairman
The Honorable Max Cleland
Ranking Minority Member
Subcommittee on Personnel
Committee on Armed Services
United States Senate

During the past 5 years, the Department of Defense (DOD) has been
reducing the costs associated with civilian personnel management by
reducing the number of staff working in personnel, consolidating
selected personnel management functions at newly created regional
centers, and attempting to improve personnel management business
processes.  A key part of this initiative is Defense's development of
a new information management system--the Defense Civilian Personnel
Data System (DCPDS)--to support a wide range of personnel management
functions including recruitment, staffing, benefits administration,
and training.  Defense expects to complete deployment of this system
by March 2000.  This letter responds to the request from your
subcommittee that we answer the following questions about this
initiative and recommend corrective actions, where appropriate. 

  -- How did Defense determine the number and locations for civilian
     personnel regional service centers and why is there a wide
     disparity in the number of regional centers among the services? 

  -- In overseeing, managing, and developing DCPDS, is Defense
     applying the investment principles of the Clinger-Cohen Act? 

  -- Does DCPDS duplicate a system that is available through the
     Office of Personnel Management (OPM) called the Employee Express
     System? 

  -- Was Defense leadership aware of the extent and cost of the
     needed modifications to the commercial-off-the-shelf (COTS)
     software application? 

  -- Has Defense identified and mitigated the risks associated with
     the major COTS modifications? 

In conducting our review, we examined Defense requirements on
development, management, and oversight of information systems in
light of relevant legislative and federal requirements, including the
Clinger-Cohen Act of 1996.  We discussed Defense's efforts to develop
and manage DCPDS with officials from (1) Defense's Civilian Personnel
Management Service (CPMS), (2) the Air Force Central Design Activity
(CDA) responsible for managing technical modifications, (3) Oracle
Corporation, the contractor from which Defense acquired the new
system, (4) the military services and Defense agencies that plan to
use the system, and (5) the Office of Personnel Management.  We also
visited and interviewed officials from five of the regional personnel
centers and four of the local or installation-level offices.  We
conducted our review from August 1997 through October 1998 in
accordance with generally accepted government auditing standards.  We
requested comments on a draft of this report from the Department of
Defense.  The Acting Assistant Secretary for Force Management Policy
provided us with written comments.  These comments have been
incorporated where appropriate and are discussed in the Agency
Comment and Our Evaluation section of this letter and appendix I. 
Details on the scope and methodology of our work are provided in
appendix II. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

Defense's current initiative can potentially improve civilian
personnel operations and achieve cost savings.  However, because the
Department has not examined other business process alternatives that
could potentially achieve even greater savings and process
efficiencies, there is no assurance that this is the best alternative
for civilian personnel operations. 

Before embarking on its costly initiative to improve personnel
management, Defense examined two alternatives (1) outsourcing
personnel computer operations to the Department of Agriculture's
(USDA) National Finance Center\1

and (2) regionalizing personnel centers.  It determined that it would
take the National Finance Center about 6 years to prepare for
transferring computer operations and that some new functionality
built into its legacy system would be lost. 

However, Defense did not examine several other potentially effective
alternatives, including (1) continuing to centralize all or parts of
its personnel management operations to reduce duplicative layers of
oversight at the components and ensure more consistent operations
DOD-wide, (2) integrating its personnel and payroll management
systems, (3) restructuring its regional offices to serve multiple
components rather than perpetuating regional offices dedicated to
only one component, (4) restructuring local personnel offices to
serve multiple bases or installations (they now serve only one base
or installation), and (5) outsourcing all civilian personnel
operations to the private sector. 

These alternatives are feasible and may have helped Defense to
achieve even greater savings and efficiencies than the current
approach.  For example, as of June 1998, there were 886 people
performing civilian personnel management and oversight functions at
component headquarters and major command levels at a cost of about
$63 million annually.  By consolidating some or portions of these
component oversight functions, Defense could reduce the number of
staff that perform duplicative overhead functions and decrease
personnel management oversight costs.  In addition, the Defense
Science Board\2 determined that integrating payroll and personnel
systems was a viable and cost beneficial option for military
personnel.  Among other benefits, this alternative might have enabled
the Department to cut system operation and maintenance costs as well
as streamline and dramatically improve both payroll and personnel
business processes.  Furthermore, by having regions serve multiple
services and agencies, Defense could have further consolidated
regional offices and reduced duplicative regional overhead costs. 
The Washington Headquarters Service has already demonstrated the
feasibility of this option by managing personnel services for
numerous smaller Defense agencies. 

CPMS officials who were responsible for the personnel initiative said
that they did not consider these business processing alternatives
because (1) CPMS did not have authority to require the military
services and Defense agencies to adopt such approaches, (2) the
Department did not allow sufficient time to rigorously examine
alternatives, and (3) the Department lacked basic cost and
performance data needed to study the alternatives.  As a result,
Defense selected a business processing alternative which, in the long
run, may not provide the most effective personnel operations at the
lowest cost. 

In addition, after it decided on its approach, Defense did not follow
a sound process for selecting regions.  For example, it did not
require military services and Defense agencies to base their
decisions on data-driven analyses and it allowed only a short time
frame for the selection.  Consequently, the analyses of the services
and agencies were inconsistent, considering different factors in
choosing their regions, and none included a formal cost/benefit
analysis.  As a result, there is a wide disparity in the numbers of
regions selected, and there is no convincing rationale or objective
evidence that any of the selections were optimal. 

Furthermore, Defense did not adequately consider a full range of
technical options before deciding to replace its legacy system with
the Oracle COTS product.  Defense informally surveyed the potential
market of COTS\3 products and selected three COTS packages for
further evaluation.  It then considered functional, technical, and
cost differences among the three but did not rigorously analyze their
costs, benefits, and expected returns-on-investment nor did it assess
the desirability of continuing to use the legacy system.  After the
Oracle product was acquired, Defense performed a limited economic
analysis for the system which did not consider all of the promising
business operation options or all of the technical options and did
not separate the costs and benefits of the selected regionalization
approach from those of the Oracle product.  As a result, there is
still no objective evidence that either element of Defense's approach
(regionalization or the use of the Oracle product) is the best
option. 

Finally, after Defense acquired the Oracle system, it did not
mitigate critical technical risks, as the following examples
illustrate. 

  -- Because the Oracle product did not satisfy many federal and
     Defense-unique requirements, modifying the system would entail a
     significant effort.  Further, there was no guarantee that the
     modifications would be successful or that the system would be
     able to accommodate Defense's large-scale workload.  To mitigate
     this risk, Defense could have first worked with the developer to
     define unique Defense and federal personnel requirements and
     postponed purchasing the product until after it was modified. 
     While Defense worked with the developer to define unique Defense
     and federal requirements, it committed to purchasing the product
     before the software was modified and could be demonstrated to
     perform successfully. 

  -- Defense has not fully mitigated critical security risks for
     either the legacy- or the Oracle-based systems.  Despite the
     fact that these systems contain sensitive privacy data, Defense
     has not established encryption or firewall standards.\4 These
     standards are needed to ensure a consistent level of protection
     for personnel data and to ensure that all DCPDS partners can
     safely and effectively access the system.  In addition, Defense
     has not promoted security awareness among the local offices that
     will be operating the new system. 

  -- Defense has not adequately addressed risks associated with the
     Year 2000 computing problem.  While it has made good progress in
     renovating the legacy system and ensuring the modern system's
     compliance, it has not developed agreements with its data
     exchange partners that specify date format changes, time frames
     for these changes, or processes for resolving interface
     conflicts.  In addition, Defense has not developed adequate
     contingency plans for either of the systems.  Even if systems
     are compliant, civilian personnel business operations are at
     risk of disruptions caused by external interfacing systems and
     the public infrastructure.  As such, detailed contingency plans
     are necessary to ensure that Defense can maintain the basic
     functionality of its core civilian personnel operations. 


--------------------
\1 The National Finance Center provides payroll, personnel,
financial, and other administrative services to USDA agencies as well
as a broad range of federal departments and agencies. 

\2 Report of the Defense Science Board Task Force:  Military
Personnel Information Management, August 31, 1996. 

\3 Over 100 different software products were initially identified. 

\4 Encryption involves the transformation of original text (also
known as plaintext or cleartext) into unintelligible text (also known
as ciphertext).  Firewalls are hardware and software components that
check all incoming network traffic and block unauthorized traffic. 


   BACKGROUND
------------------------------------------------------------ Letter :2

Defense's civilian personnel community provides Defense managers with
the personnel management services and support needed to accomplish
their missions, including recruitment, job classification, position
management, training, career development, and benefits
administration.  Traditionally, the military services and Defense
agencies have managed their civilian personnel service delivery
organizations and systems through local civilian personnel offices
located at or near military bases and installations all over the
world.  During the past 5 years, Defense has been attempting to
reduce personnel management costs through the following actions. 

(1) Reducing the number of civilian personnelists.  Personnelists
provide face-to-face assistance to civilian employees, answering
questions about such issues as life insurance, health insurance, and
position classification.  They process paperwork for new hires,
promotions, awards, and a wide variety of personnel actions and
assist in training, benefits administration, management/employee
relations, recruitment, and staffing.  In 1994, Defense reported that
a single personnelist served about 67 employees.  Defense's goal was
to reduce the number of personnel staff to the point where one
personnelist served 88 employees by the year 2001 and 100 employees
by the year 2003.\5 As of June 30, 1998, Defense reported that it had
cut 1,700 personnelists and had achieved a ratio of 1 personnelist to
77 employees. 

(2) Improving personnel management processes.  To help increase the
personnelist-to-civilian employee ratio, Defense is attempting to
improve and automate its personnel management business processes. 
For example, it has automated and improved processes for (1)
developing, tracking, and monitoring all personnel actions, (2)
handling injury compensation claims, and (3) estimating retirement
eligibility and benefits.  It has acquired an automated tool called
RESUMIX, which helps personnelists analyze resumes of people applying
for a position with Defense.  It is also developing an interactive
voice response system that enables employees to use a Touch-Tone
phone to change selected data in their own personnel records. 

(3) Creating regional centers.  Defense is creating regional centers
that will specialize in selected personnel management functions and
reducing the number and size of local offices.  It anticipates that
specialization of labor within the regions combined with improved
business processes will reduce operating costs.  As of September 30,
1998, the Army had established all 10 of its planned regions, the
Navy had established 7 of 8 planned regions, the Air Force had
established its 1 region, and the Defense agencies participating in
this initiative had established all 3 of their planned regions. 
Table 1 further illustrates the changes in personnel management that
will occur through Defense's improvement initiative. 



                                Table 1
                
                  Differences in Personnel Management

Before personnel improvements       After personnel improvements
----------------------------------  ----------------------------------
Local personnel offices provided    Local personnel offices will still
service to all civilian employees   provide face-to-face service to
and carried out all work            civilian employees. However, 40 to
processes, such as processing       60 percent of the processing of
paperwork for new hires,            personnel-related actions are to
processing promotions, developing   be done at the regional offices.
vacancy announcements, and
assisting in management/employee
relations.

Some personnelists specialized in   Most personnelists at the local
certain work processes while        offices will be generalists.
others provided a broader range of  Specialists will be located at
personnel services.                 regions.

In 1994, there were 389 local       By fiscal year 1999, there are to
offices and no regional offices.    be 311 local offices plus 22
                                    regional offices.

Most work processes were manual     Business process improvement
and paper-oriented.                 efforts are targeted at automating
                                    many work processes, such as
                                    estimating retirement eligibility
                                    and benefits and analyzing
                                    resumes.

Before 1994, only personnelists     Functional managers, civilian
had access to personnel management  employees and personnelists are to
systems.                            have access to the personnel
                                    management information system.
                                    Among other things, civilians can
                                    view their own records and make
                                    prescribed changes to insurance
                                    and thrift savings retirement
                                    data. Functional managers will be
                                    able to initiate personnel actions
                                    on the system.
----------------------------------------------------------------------

--------------------
\5 In 1989, the Army and the Air Force had civilian personnelist
servicing ratios of 1 to 50 and 1 to 48, respectively, while the
Navy's ratio was 1 to 61.  At the time, DOD began efforts to increase
servicing ratios in the other services to at least the Navy's ratio. 
The goal of reaching 1:100 was derived based on recommendations by
the National Performance Review, as well as DOD's own internal
benchmarking study.  DOD's internal study indicated that some DOD
organizations had servicing ratios exceeding 1:100. 


      A COTS PERSONNEL MANAGEMENT
      SYSTEM IS ACQUIRED TO
      SUPPORT INITIATIVE
---------------------------------------------------------- Letter :2.1

At the beginning of this effort, Defense components operated a number
of personnel management information systems that assisted in all
aspects of personnel operations, such as developing position
classification documents; preparing vacancy announcements; and
processing appointments, reinstatements, transfers, promotions,
retirements, and terminations.  These systems were redundant and not
interoperable, and Defense believed that they were antiquated. 

To modernize this environment, Defense eliminated the duplicative
systems and used the Air Force civilian personnel management
information system, located in San Antonio, Texas, to do all
personnel processing.  This legacy system meets Defense-unique
personnel management requirements; is able to process Defense's
large-scale workload successfully; and because it operates in one
location, it can be maintained by CDA personnel with experience in
operating and protecting systems.  However, Defense believed that
there were a number of significant shortfalls with this mainframe
system\6 and, therefore, the system should be replaced with a new
COTS system.  For example, according to Defense

  -- the legacy system relied on outdated technology for its database
     structure, file update, and retrieval;

  -- manpower resources and costs needed to develop and maintain the
     system were extensive;

  -- the system required duplicative data entry;

  -- the system could only be accessed by personnelists--it could not
     be easily modified to provide access to civilian employees so
     that they could review and make prescribed changes to their own
     benefit, insurance, and other personnel-related data;

  -- modifications reflecting improvements in business processes were
     difficult to make; and

  -- the system was not Year 2000 compliant. 

As a result, Defense acquired a COTS product from Oracle Corporation. 
In contrast to the legacy system, which operated on two 1970s era
mainframes, the new system will operate in a distributed, networked
environment\7 at regional and local offices.  According to Defense,
the system

  -- will enable any authorized civilian employee with a personal
     computer to directly access the system and to perform prescribed
     personnel-related operations or management tasks,

  -- can be easily modified to reflect improvements in business
     processes,

  -- will cost less to maintain and operate, and

  -- will be Year 2000 compliant. 

However, because the Oracle product was originally designed for use
in the private sector, it did not satisfy all federal and
Defense-unique requirements for personnel management.  For example,
it could not process federal personnel forms, such as the standard
personnel action form (Form 52).  It did not address the federal
General Schedule for salaries, Defense's demonstration projects for
pay banding, or the Defense-unique salary schedule for tens of
thousands of foreign nationals who work for the Department overseas
but do not get the same salaries or benefits as American employees. 
It did not have DOD-unique data for security and mobilization.  In
addition, it did not directly interface with Defense's existing
payroll system.  As a result, the product needed to be modified
and/or enhanced before it was deployed. 

The Civilian Personnel Management Service (CPMS), which was
established in 1993 to provide departmentwide leadership for the
civilian personnel business area, is responsible for managing the new
system.  CPMS acquired the system using an indefinite delivery,
indefinite quantity (IDIQ) DOD contract\8 under which Oracle
Corporation was a participating vendor.  Defense components are
responsible for purchasing and maintaining hardware to support the
new system.  CPMS has assigned the Air Force Central Design Activity
(CDA) responsibility for managing technical modifications to the
system under the contract.\9 According to CPMS, the system is
currently in the test phase.  Once system qualification tests are
completed, the system will be deployed to four tests sites during
January and February 1999.  The Air Force Operational Test and
Evaluation Center (AFOTEC) will then evaluate the test results to
ensure that the system meets user needs in an operational
environment.  Deployment to the remaining sites is expected to begin
in late 1999 and end by March 2000.  DOD officials stated that this
schedule is likely to slip at least 2 months to ensure that the
system is fully tested and meets user needs before it is fully
deployed. 


--------------------
\6 A mainframe is a very large computer capable of supporting
hundreds or even thousands of users simultaneously.  Mainframes use
smaller computers as front-end processors that connect to
communications networks. 

\7 Rather than processing all applications on a single mainframe,
applications are distributed to run on independent, networked
computers. 

\8 The Integrated Computer-Aided Software Engineering (I-CASE)
contract.  This is an indefinite delivery, indefinite quantity
contract awarded to Logicon in April 1994.  DOD can use this contract
to purchase IT systems, hardware, and software tools from approved
vendors without having to prepare a separate contract. 

\9 There is an integrated team of contractors working for CDA in San
Antonio that includes Oracle staff as well as individuals who work on
a contract basis for CDA.  The Oracle employees work on Oracle's
federal system while the other contract employees are responsible for
developing DOD-unique add-ons to the system. 


      COSTS OF DOD'S PERSONNEL
      INITIATIVE
---------------------------------------------------------- Letter :2.2

The cost of Defense's personnel initiative is estimated to be $1.2
billion over its estimated 15-year life cycle (fiscal years 1995
through 2009), of which Defense reports that over $300 million has
been spent through the end of fiscal year 1998.  These totals are
itemized in table 2. 



                                Table 2
                
                 Estimated Costs of Defense's Personnel
                    Initiative (Dollars in Millions)

                                                                Amount
                                                                 spent
                                                               through
                                                  Estimate      fiscal
Purpose                                             d cost   year 1998
------------------------------------------------  --------  ----------
Cost to develop and deploy the new system.            $177        $142
Cost to establish regional offices.                   $190        $159
Operational and support costs for the new system      $621          $0
 for fiscal years 1999 through 2009.\a
Operational and support costs for regions for         $256         $13
 fiscal years 1995 through 2009.\a
======================================================================
Total                                               $1,244        $314
----------------------------------------------------------------------
\a This includes costs for site operations, replacement software and
hardware, equipment upgrades, program management oversight, and
administration. 


   QUESTION:  HOW DID DEFENSE
   DETERMINE THE NUMBER AND
   LOCATIONS FOR REGIONAL CENTERS
   AND WHY IS THERE A WIDE
   DISPARITY? 
------------------------------------------------------------ Letter :3

Answer:  Defense considered only a narrow range of alternatives for
improving personnel operations before deciding to regionalize
personnel centers.  This left the Department without assurance that
it was pursuing the most cost-effective and beneficial approach. 
After it decided to regionalize, Defense did not follow a sound
process for selecting regions, it did not require services and
agencies to base their decisions on data-driven analyses. 
Consequently, the analyses of the services and agencies were
inconsistent, each considering different factors in choosing regions
and none included a formal cost/benefit analysis.  This process
resulted in the wide disparity in the number of regions chosen, and
it left Defense without the objective data needed to determine
whether any of the choices were optimal. 

Before embarking on a major, costly initiative to improve personnel
management, sound practices call for examining a range of improvement
options, including those that would radically change the current way
of doing business.  For example, in addition to, or instead of
regionalizing, Defense could have considered (1) outsourcing its
personnelist computer operations or all of its civilian personnel
management services, (2) integrating its personnel/payroll management
systems, (3) creating regions that cross-service between agencies and
the military services, (4) consolidating local personnel offices that
are near each other to provide face-to-face services to multiple
bases or installations out of the same office, and/or (5)
centralizing all, or portions of, civilian personnel management in
DOD.  By thoroughly considering these and other choices, Defense
would have ensured that the most cost-effective and beneficial
alternative was chosen before deciding to invest $367 million\10 in
the project and that any systems acquired or developed would support
the most efficient and effective business processes. 

Defense did not examine all of these promising alternatives. 
Instead, it considered only the possibility of outsourcing computer
operations with the National Finance Center.  This option was
determined to be infeasible.\11 Defense did not analyze other
alternatives, including cross-servicing, integrating
payroll/personnel systems, collocating personnel offices, DOD-wide
management of personnel operations, or outsourcing all of its
personnel operations. 

In addition, once it decided on regionalization, Defense did not
follow a sound process for selecting the regions.  For example,
Defense did not require the services and agencies to base their
selections on data-driven analyses.  In fact, the services were
allowed to select whichever and as many regions as they wanted as
long as they achieved at least a 1 to 88 personnelist-to-civilian
employee ratio. 

Consequently, the services considered different factors in choosing
their regions.  However, none based their selections on a thorough
cost/benefit analysis.  This resulted in the wide disparity in the
number of regions chosen, as the following examples illustrate. 

  -- The Army and the Navy considered the distance between regions,
     proximity to the installations they serviced, and coverage
     across time zones as well as some costs associated with
     establishing and operating regions and transferring personnel. 
     After considering these factors, the Army selected 10 regions
     and the Navy selected 8.  It was decided that the regions would
     be responsible for about 60 percent of the work while local
     offices would be responsible for about 40 percent.  Neither the
     Army or the Navy conducted cost/benefit analyses in making their
     decisions.  Nor did they consider the costs of personnel work
     processes or the relationship between per capita servicing costs
     and region size. 

  -- Because it had already demonstrated that it could reduce
     overhead and technology costs and facilitate standardization in
     service and business processes by collocating the civilian
     personnel center with its military center, the Air Force decided
     to use a single Air Force personnel center to serve all of its
     personnel.  The Air Force decided that its local offices would
     continue to be responsible for about 53 percent of the work. 

While Defense allowed the services wide latitude in choosing their
regions, it directed that its agencies be serviced by three regional
offices.\12 The two largest agencies--the Defense Finance and
Accounting Service and the Defense Logistics Agency--were directed to
establish their own regions and the Washington Headquarters Service
was directed to serve as a regional personnel office for the smaller
agencies.  The Defense Finance and Accounting Service selected the
location for its regional center based on the fact that it had
already started to regionalize personnel operations there.  The
Defense Logistics Agency selected the location for its regional
center after considering the location and space availability of its
depots.  However, neither conducted formal cost/benefit analyses in
choosing their regions or considered the cost of personnel work
processes and the relationship between per capita servicing costs and
region size. 

CPMS officials cited several reasons for taking this approach. 
First, they pointed out that CPMS had no authority to require the
services and agencies to base their decisions on thorough,
data-driven analyses or, in fact, to require that they adopt any
standard personnel system or approach at all.  At the same time, they
noted that the military services had a vested interest in maintaining
the status quo and had the independent budget authority to see that
the status quo was preserved.  Second, Defense lacked basic cost and
performance data for examining options, including data on the cost of
personnel work processes and the relationship between per capita
servicing costs and region size.  Third, the agency was directed in
1994 to implement the Office of the Secretary of Defense's (OSD)
recommendations quickly, i.e., to reduce the number of personnelists
to a ratio of one personnelist to every 88 civilian employees by
fiscal year 1998.  CPMS officials held that this did not allow time
to develop objective data and rigorously examine alternatives.  The 1
to 88 goal was later extended to the year 2001.  Fourth, CPMS
officials stated that because most of the costs for performing
personnel functions are for personnelists, and systems, facilities,
and operations constitute relatively smaller costs, as long as it
achieved the 1 to 88 ratio, Defense would accrue significant cost
savings regardless of the number of regions selected.\13

Nevertheless, several of the alternatives Defense ignored offered the
opportunity to achieve far greater savings while streamlining
personnel operations, as the following examples illustrate. 

  -- By consolidating some or all of its personnel management,
     Defense could reduce the numbers of staff that perform
     duplicative overhead functions.  As of June 1998, there were 886
     people performing civilian personnel management and oversight
     functions at component headquarters and major command levels at
     a cost of about $63 million annually.\14 Furthermore, if Defense
     had centralized management of departmentwide personnel
     operations, it could take a departmentwide perspective in
     deciding which local offices and which regions should be
     consolidated. 

  -- Cross-servicing could have enabled Defense to further
     consolidate regional offices and reduce duplicative overhead
     costs.  Some Defense components have already found this
     alternative to be beneficial.  The military services, for
     example, are doing some cross-servicing with employees in remote
     locations and the Washington Headquarters Service is servicing
     the smaller Defense agencies as well as some federal agencies,
     including the Office of Personnel Management.\15 Additionally,
     having local personnel offices service multiple bases or
     installations could further reduce duplicative overhead costs. 

  -- Integrating payroll and personnel systems could have helped
     Defense reduce system operation and maintenance costs as well as
     further streamline and improve personnel and payroll management
     business processes.  In fact, after considering the potential
     benefits of this alternative and its feasibility, the Defense
     Science Board recommended it as a solution for military
     personnel in 1996.\16

While it may have required more time and greater management
commitment to change Defense practices, the potential for
substantially greater savings and efficiencies should have compelled
Defense to perform a rigorous analysis of all alternatives and to
select the one proven most cost effective. 


--------------------
\10 Defense planned to initially invest $177 million to develop and
deploy the new system and $190 million to establish the regional
offices, for a total of $367 million. 

\11 Defense considered the possibility of outsourcing the IRM support
function to the private sector.  It concluded that this option was
not feasible due to the size of Defense's operations.  In exploring
the possibility of outsourcing computer operations with the National
Finance Center, Defense learned that it would take the Center about 6
years to prepare for transfer and that some new functionality built
into its legacy system would be lost. 

\12 DOD has over 20 separate agencies and activities.  Most are small
and in the Washington, D.C.  area.  The intelligence agencies were
excluded from this initiative and allowed to acquire their own
personnel software program (PeopleSoft). 

\13 According to Defense's economic analysis, over 80 percent of the
costs of performing personnelists functions are for personnelists. 

\14 Our estimate is based on DOD/CPMS data on personnelists costs and
numbers. 

\15 Defense does not have information on the savings being derived
from its current cross-servicing activities. 

\16 Report of the Defense Science Board Task Force:  Military
Personnel Information Management, August 31, 1996. 


   QUESTION:  IN DEVELOPING,
   MANAGING, AND OVERSEEING DCPDS,
   IS DEFENSE APPLYING THE
   CLINGER-COHEN ACT? 
------------------------------------------------------------ Letter :4

Answer:  Defense did not adequately apply the three requirements of
the Clinger-Cohen Act of 1996 we reviewed which are designed to
maximize the value of major investments.  While the act was passed
after Defense initiated its development of DCPDS, the act's
requirements reflect basic and widely accepted principles of sound
system acquisition management.  Similar practices are also called for
by Defense's own system acquisition regulations and guidelines,
Office of Management and Budget (OMB) guidance, and other legislative
requirements effective at the time DCPDS decisions were made,
including the Government Performance and Results Act of 1993, the
Federal Acquisition Streamlining Act of 1994, the Paperwork Reduction
Act of 1995, and the Chief Financial Officers Act of 1990. 

The Clinger-Cohen Act requires federal agencies to focus on the
results achieved through information technology investments while
streamlining the federal information technology (IT) procurement
process.  Specifically, this act introduces much more rigor and
structure into how agencies approach the selection and management of
IT projects.  Although the act was passed after Defense decided to
develop a new personnel management system, its principles are based
on practices that are widely considered to be integral to successful
IT investments.\17

We examined whether Defense applied the following three requirements
of Clinger-Cohen, which are designed to maximize the value of a major
investment such as DCPDS. 

(1) Agency heads should analyze the missions of the agency and, based
on the analysis, revise the agency's mission-related and
administrative processes, as appropriate, before making significant
investments in IT supporting those missions. 

(2) Investments should be selected based on objective data, including
quantitatively expressed projected net, risk-adjusted return on
investment, and specific quantitative and qualitative criteria for
comparing and prioritizing alternative information system projects. 

(3) Agency heads should ensure, through the use of performance
measurements, that mission-related benefits are defined and assessed
for all IT investments. 


--------------------
\17 See Executive Guide:  Improving Mission Performance Through
Strategic Information Management and Technology (GAO/AIMD-94-115, May
1994) for an analysis of the management practices of several leading
private and public sector organizations on which the Clinger-Cohen
Act is based and Assessing Risk and Returns:  A Guide for Evaluating
Federal Agencies' IT Investment Decision-making (GAO/AIMD-10.1.13,
February 1997) for an overview of the IT management process
envisioned by the Clinger-Cohen Act. 


      DEFENSE DID NOT REENGINEER
      BUSINESS PROCESSES BEFORE
      INVESTING IN DCPDS
---------------------------------------------------------- Letter :4.1

Defense did not reengineer its personnel processes before investing
in the new system.  Before initiating development, CPMS and the
individual services conducted an extensive effort to identify and
document the preproject business processes at the local offices. 
Most of the improvements they made to these operations were minor. 
For example, they developed automated tools to help personnelists
analyze resumes and to track civilian employee costs.  However, for
the most part, these initiatives did not involve radical or major
changes to existing processes.  As noted in the previous section,
Defense considered only the option for outsourcing computer
operations and failed to consider other alternatives that had the
potential to provide significantly greater benefits, such as
integrating personnel and payroll systems, centralizing personnel
management, or cross-servicing.  Because Defense did not examine
these options, there is no evidence that the personnel management
system acquired will support the most effective way of doing business
or provide optimal return on investment. 


      COSTS, BENEFITS, AND RETURNS
      ON INVESTMENTS NOT
      ADEQUATELY ANALYZED
---------------------------------------------------------- Letter :4.2

Costs, benefits, and returns on investments were not adequately
analyzed before Defense acquired the Oracle package.  Defense
informally surveyed the potential market of COTS products and
selected products from PeopleSoft, Inc., Integral Software Systems,
Inc., and Oracle Corporation for evaluation.  In evaluating these
products, a DOD team considered various characteristics of the
software products, including functionality, technical merit, and
cost. 

However, Defense did not perform a rigorous analysis of costs,
benefits, and returns on investments for these products before
deciding to acquire the Oracle product, nor did it rigorously analyze
the other available commercial products or the possibility of
continuing to use the legacy system.  The importance of developing
complete and accurate analyses of the costs/benefits and returns of
system alternatives is underscored by several governmentwide
requirements in addition to the Clinger-Cohen Act.  For example,
OMB's Circular A-130, Management of Federal Information Resources,
calls on agencies "to conduct benefit-cost analyses to support
ongoing management oversight processes that maximize return on
investment and minimize financial and operating risks for investments
in major information systems and on an agencywide basis." Likewise,
Supplement to OMB's Circular A-11 (July 1997), Part 3, Capital
Programming Guide Version 1.0, and OMB Bulletin No.  95-03, Planning
and Budgeting for the Acquisition of Fixed Assets, state that "the
planning for fixed asset acquisitions should be based on a systematic
analysis of expected benefits and costs." Because Defense did not
perform these analyses, it does not know if it chose the best system. 

Once an alternative is selected, Defense regulations\18 require that
an economic analysis be prepared to compare the selection against the
status quo.  This analysis establishes baseline life cycle costs,
estimates benefits for the new system, and calculates expected return
on investment.  However, Defense did not perform an economic analysis
before acquiring the new system.  In addition, the analysis that
Defense performed after the initiative was underway did not separate
the costs and benefits of the system from costs and benefits
associated with cutting personnel and regionalizing.  As a result,
Defense still does not know if it chose the best business process
alternative. 


--------------------
\18 Economic analyses are required by DOD's Instruction 7041.3,
"Economic Analysis for Decisionmaking" and its "5000" acquisition
regulations. 


      PERFORMANCE MEASURES
      DEVELOPED BUT DATA NEEDED
      FOR COMPARISONS IS LACKING
---------------------------------------------------------- Letter :4.3

To measure how the Oracle product supports its personnel
administration mission, CPMS developed four major mission performance
measure categories to be collected by each service and Defense
agency.  These categories included (1) servicing ratio, (2) customer
satisfaction, (3) process cycle time (e.g., how long it takes to
process a specific personnel action, such as filling an opening or
promoting an employee), and (4) regulatory compliance (i.e., whether
personnel paperwork complies with applicable laws and regulations). 
The military services and Defense agencies then developed several
detailed measures within the categories, and CDA and CPMS developed
several information technology or system-level measures to measure
DCPDS' contribution to the mission area, including process cycle time
and system response time. 

However, because military services have not agreed on two fundamental
definitions, they will not be able to calculate these measures
consistently and compare measures across services.  First, the
military services could not agree on how to define the start and end
date for the process of filling a position or whether certain
personnel actions (rejecting a list of qualified job applicants, for
example) would be considered as part of the process for filling a
position.  Second, they could not agree on a common definition of
"paperwork errors." Because the military services are not using
common definitions, some critical performance measures will not be
comparable across DOD.  In addition, Defense does not have baseline
performance information on how long it takes to fill a position and
the accuracy of personnel paperwork.  As a result, it will not be
able to accurately assess whether the system has improved mission
performance in these areas or by how much. 


   QUESTION:  DOES DCPDS DUPLICATE
   EMPLOYEE EXPRESS? 
------------------------------------------------------------ Letter :5

Answer:  DCPDS is not a duplicate of OPM's Employee Express system. 
OPM's Employee Express system is designed to be used in conjunction
with existing personnel and payroll systems of the agencies.  It does
not perform all basic personnel and payroll functions.  Instead, it
allows employees to interface with the existing personnel and payroll
systems.  For example, Employee Express enables a federal civilian
employee to use a Touch-Tone phone or personal computer connected to
the Internet to make changes to certain data in his/her automated
personnel/payroll records.\19

The new DCPDS system is to eventually replace existing DOD personnel
systems.  It is intended to support the full range of core functional
requirements needed by Defense for an automated human resources
management system, including position management and classification,
recruitment and staffing, personnel action administration, benefits
administration, labor-management and employee relations, work force
development, and retention and reporting.  These requirements are
defined in a November 1997 study by the Human Resources Technology
Council, an inter-agency group associated with the President's
Management Council and chaired by the Office of Personnel Management. 
Although Defense civilian employees will not be able to use the
Employee Express system to make changes to DCPDS data, Defense plans
to add employee express-type features at a later date that will allow
changes to be made using a Touch-Tone phone or personal computer
connected to the Internet. 


--------------------
\19 For example, direct deposit information, financial allotments,
federal and state tax withholding, home or check mailing address,
health benefits, and Thrift Savings Plan contributions. 


   QUESTION:  WAS DEFENSE
   LEADERSHIP AWARE OF EXTENT AND
   COST OF MODIFICATIONS? 
------------------------------------------------------------ Letter :6

Answer:  Defense leadership was aware that the COTS package it
acquired would need to be substantially modified in order to support
federal and Defense-unique personnel requirements although the full
extent of the modification was not known.  According to the
Acquisition Program Manager, Oracle had orally agreed not to charge
Defense for the modifications it was making to the system because it
believed it could market the package to other federal agencies after
it was "federalized."


   QUESTION:  HAS DEFENSE
   IDENTIFIED AND MITIGATED RISKS
   ASSOCIATED WITH THE COTS
   MODIFICATIONS? 
------------------------------------------------------------ Letter :7

Answer:  Defense has not identified and mitigated significant risks
associated with its acquisition.  Specifically, as discussed below,
Defense does not yet know (1) if the modifications will satisfy DOD
needs and provide required functionality and performance, (2) how it
will handle future system modification, (3) how it will maintain the
system, (4) how it will protect sensitive data in the system, and (5)
how it will ensure the continuity of core civilian personnel
operations in the event of Year 2000 failures. 


      DEFENSE DOES NOT KNOW IF
      MODIFICATIONS WILL SATISFY
      REQUIREMENTS
---------------------------------------------------------- Letter :7.1

Defense has no assurance that the modified product being developed by
Oracle will meet all its needs.  It does not know whether Oracle can
provide all required functionality and performance or deliver it on
time.  Although Defense worked closely with Oracle to define
requirements and test the changes that were made to the COTS package,
it acquired the system before these modifications were completed and
before the modified product could be tested.  As a result, Defense
faces the risk that the system it has already acquired may not meet
all its requirements.  This risk could have been avoided by waiting
for Oracle to produce the "federalized" product and thoroughly
testing it before purchasing it. 


      DEFENSE DOES NOT KNOW HOW IT
      WILL HANDLE FUTURE SYSTEM
      MODIFICATION
---------------------------------------------------------- Letter :7.2

Compounding the risk that the system will not meet Defense
requirements is the fact that Defense has not secured the legal right
to modify and upgrade the package it has acquired.  CPMS obtained a
software licensing agreement for 3 years (with an option to extend to
8 years) that provides for Oracle to correct programming errors found
in its product.  However, the agreement does not require Oracle to
provide upgrades to DOD's modified product at the same time and at
the same cost as it provides upgrades to its private sector
commercial product.  As a result, Defense has no assurance that
Oracle will make future versions of the software available to Defense
at a reasonable cost or make future needed modifications at a
reasonable cost, so that its version of Oracle product will not
become obsolete.  In addition, the agreement does not specify whether
Oracle will make DOD-required modifications to its customized
product, or how much Oracle will charge for such work. 


      DOD DOES NOT KNOW HOW IT
      WILL MAINTAIN THE SYSTEM
---------------------------------------------------------- Letter :7.3

CPMS has not taken several actions which are essential to ensuring
that the system is adequately maintained.  First, CPMS has not yet
developed agreements between the DCPDS partners that define each
partner's responsibility for systems, operations, maintenance, and
security.  Whereas the legacy system was centrally maintained, the
military services and Defense agencies will be responsible for
maintaining the new system hardware and related local area networks. 
It is critical that CPMS develop agreements with its DCPDS partners
to ensure effective, efficient, and secure systems operations and
maintenance. 

Second, CPMS has not yet established a configuration control board
comprised of DCPDS users to assist in deciding what changes need to
be made to the system once it is deployed and to prioritize change
requests.  As noted in Defense's Program Manager's Guide to Software
Acquisition Best Practices, configuration management is vital to the
success of any software effort because it prevents uncontrolled,
uncoordinated changes to shared project software and products
(documentation and test results, for example). 

Third, CPMS has not decided who will provide technical assistance to
the personnel sites operating the system.  CDA currently performs
this function; however, CPMS has not decided whether to continue
using CDA after deployment or to outsource this function. 

Fourth, CPMS has not yet developed agreements with DCPDS interface
partners, which include the Office of Personnel Management and DOD
agencies responsible for payroll, security, and manpower systems.  As
noted in Defense's Program Manager's Guide to Software Acquisition
Best Practices, interfaces constitute essential elements of the
system but are not completely controlled by the developer.  As a
result, the guide recommends that explicit written agreements with
interface partners be developed to ensure that the partners clearly
understand their roles and responsibilities. 


      DEFENSE HAS NOT ADEQUATELY
      ADDRESSED SECURITY RISKS
---------------------------------------------------------- Letter :7.4

It is even more difficult to protect the new system and its data than
it is to protect the legacy system and its data.  Whereas the
mainframe-based legacy system operated in one location and was
maintained by CDA personnel with experience in protecting information
systems, the new system will be distributed to 22 centers and many
local offices where staff have little or no experience in providing
the type of security required for DCPDS.  Furthermore, both systems
are vulnerable to outside computer attacks since they use an unsecure
telecommunications network to transmit data.\20

According to our Executive Guide:  Information Security
Management,\21 there are five key principles for managing these types
of risks that were identified by studying private and government
organizations with reputations for having good information security
programs.  First, organizations should assess their risks and
determine their security needs.  Second, they should establish a
central management focal point for security issues.  Third, they
should implement appropriate policies and related controls.  Fourth,
they should promote security awareness.  Fifth, they should
continually monitor and evaluate policy and control effectiveness. 
An important factor in effectively implementing these principles is
linking them in a cycle of activity that helps ensure that
information security policies address current risks on an ongoing
basis. 

A security risk assessment was performed for the new system, a
central security focal point was established, and some effective
measures were implemented, including a software application that can
identify and notify appropriate officials of unauthorized or
suspicious attempts to access personnel data and produce summary
audit reports highlighting unauthorized access attempts.  However,
Defense has not implemented appropriate departmentwide or
DCPDS-specific security policies and related controls nor effectively
promoted security awareness as indicated by the following examples of
identified weaknesses which have increased both the legacy and modern
system's vulnerability to computer attacks. 

  -- Defense officials, including the Deputy Secretary of Defense,
     believe that encryption technology is necessary to maintain the
     secrecy and integrity of data that is transmitted over Defense's
     unsecure networks.  Encryption involves the transformation of
     original text (also known as plaintext or cleartext) into
     unintelligible text (also known as ciphertext).  However, the
     Defense Information Systems Agency (DISA), which is responsible
     for establishing computer security standards for the Department,
     has not established a standard encryption approach for sensitive
     but unclassified Defense data.  In the absence of these
     standards, CPMS is planning to acquire a package for encrypting
     DCPDS data.  As other organizations do the same, DOD may be
     faced with managing multiple, incompatible encryption products
     and approaches. 

  -- The military services and Defense agencies recognize that
     firewalls, which are hardware and software components that check
     all incoming network traffic and block unauthorized traffic, are
     also essential to protecting sensitive data and have begun
     installing them.  However, DISA has not established standards to
     ensure a consistent level of protection and to ensure that
     computer systems protected by firewalls can still communicate
     with each other. 

  -- During our review, we identified several sites that were not
     maintaining adequate physical security over computer resources,
     indicating a lack of security awareness at the local level.  For
     example, at two of the four local personnel offices we visited,
     the door to the computer room was unlocked.  At one of these
     offices, one of the computer room's walls consisted of a row of
     standard metal filing cabinets, offering little obstruction to
     the room even if the door had been locked.  At a third local
     office, the computer room was collocated with the office's paper
     shredder, to which the personnel office staff were given
     unsupervised access.  Also, the network communications room at
     one of the local offices was unlocked and personnel office staff
     were given unsupervised access to the room.  Additionally, at
     one of the four regional offices we visited, the network
     communications room door was unlocked and tied open.  Further,
     our review identified fire protection deficiencies at four
     offices--three local offices and one regional office. 
     Specifically, the four offices did not have automatic fire
     detection equipment in or near the computer room. 

  -- Our review identified problems with disaster recovery procedures
     and planning for the regional and local offices.  For example,
     we observed inadequate data backup and recovery procedures at
     one of the four regions visited.  In this regard, the draft
     DCPDS Trusted Facilities Manual, dated February 2, 1998, noted
     that Defense had not resolved basic disaster recovery planning
     issues for DCPDS such as, "what data to backup, how often that
     data will require backup, the method of backup, and testing to
     ensure the backup has been accomplished successfully."\22
     Additionally, the military services had not completed
     service-level or site-specific disaster recovery plans for their
     regional and local personnel offices.  As of July 1998, CDA had
     drafted guidelines for the services and agencies to use in
     developing disaster recovery plans, but it did not have complete
     data on the number of regional and local offices that had
     finalized and tested site-level disaster recovery plans.  After
     discussions on this issue, CDA began requiring all sites to
     provide these plans before becoming operational.  However,
     neither CPMS nor CDA have determined how the plans will be
     tested or whether CDA will periodically verify that the disaster
     recovery plans are updated. 


--------------------
\20 Defense uses its Non-Secure Internet Protocol Router Network
(NIPRNet) to transmit DCPDS data. 

\21 Executive Guide:  Information Security Management
(GAO/AIMD-98-68, May 1998). 

\22 Final draft of the Trusted Facilities Manual dated February 2,
1998, Section 6.5, Trusted Backup and Recovery Guidance. 


      YEAR 2000 RISKS NOT FULLY
      MITIGATED
---------------------------------------------------------- Letter :7.5

The Year 2000 computing problem is rooted in the way dates are
recorded and computed in automated information systems.  For the past
several decades, systems have typically used two digits to represent
the year, such as "97" to represent 1997, in order to conserve
electronic data storage and reduce operating costs.  With this
two-digit format, however, the Year 2000 is indistinguishable from
1900, or 2001 from 1901, etc.  As we reported earlier this year, the
impact of computer failures resulting from the problem could be
widespread, costly, and potentially disruptive to military
operations.\23 Year 2000 problems could adversely affect Defense's
ability to train civilian personnel, administer benefits, recruit
staff, and handle management/employee disputes.  However, Defense has
not fully mitigated this risk. 

We compared Defense's efforts to correct the Year 2000 problem to
criteria detailed in our Year 2000 Assessment Guide.\24 This guide
advocates a structured approach to planning and managing an effective
Year 2000 program though five phases:  (1) raising awareness of the
problem, (2) assessing the extent and severity of the problem and
identifying and prioritizing remediation efforts, (3) renovating,
retiring, or replacing systems, (4) validating or testing
corrections, and (5) implementing corrected systems.  We and OMB
established a schedule for completing each of the five phases,
including requiring agencies to complete the assessment phase by
August 1997 and the renovation phase by September 1998. 

Our Assessment Guide also identifies other dimensions to solving the
Year 2000 problem, such as identifying interfaces with outside
organizations, specifying how data will be exchanged in the Year 2000
and beyond, and developing contingency plans to ensure that core
business functions can be performed even if systems fail.  As further
detailed in the following sections, while Defense is making good
progress in renovating the legacy system and ensuring that the new
system is compliant, it has not yet ensured that its external
interfaces will be remediated or developed effective contingency
plans. 


--------------------
\23 Defense Computers:  Year 2000 Computer Problems Threaten DOD
Operations (GAO/AIMD-98-72, April 30, 1998). 

\24 Year 2000 Computing Crisis:  An Assessment Guide
(GAO/AIMD-10.1.14, September 1997).  Published as an exposure draft
in February 1997 and finalized in September 1997. 


         ADEQUATE INTERFACE
         AGREEMENTS AND BUSINESS
         CONTINUITY AND
         CONTINGENCY PLANS NOT
         DEVELOPED FOR LEGACY
         SYSTEM
-------------------------------------------------------- Letter :7.5.1

Defense has nearly completed renovation work on its legacy system,
according to the Acquisition Program Manager, and release/deployment
is planned for December 1998.  In addition, in August 1998, Defense
finalized a Year 2000 test plan for the legacy system.  However,
Defense does not yet have interface agreements that specify changes
to date formats and how and when conflicts will be resolved with its
data exchange partners.\25

Because noncompliant interfacing partners can introduce Year
2000-related errors into compliant systems, our Assessment Guide
recommends that agreements with interface partners be established in
the assessment phase in order to allow enough time for resolving
conflicts.  Until these agreements are in place, Defense will not
have assurance that partners are working to correct interfaces
effectively or promptly. 

In addition, Defense has not developed adequate business continuity
and contingency plans for the legacy system.  To mitigate the risk
that Year 2000-related problems will disrupt operations, our guide,
entitled Year 2000 Business Continuity and Contingency Planning,\26
recommends that agencies perform risk assessments and develop and
test realistic contingency plans to ensure the continuity of critical
operations and business processes.  Business continuity and
contingency plans are important because they identify the manual or
other fallback procedures to be employed should systems miss their
Year 2000 deadline or fail unexpectedly in operation.  Business
continuity and contingency plans also define the specific conditions
that will cause their activation. 

In order for these plans to be effective, our guide recommends that,
among other things, agencies analyze business process composition and
priorities, dependencies, cycles, and service levels, and most
important, the business process dependency on mission-critical
information systems.  The results of this analysis should be used to
assess the cost and benefits of contingency alternatives and to
identify and document contingency plans and implementation modes. 
These plans should define roles and responsibilities for contingency
operations and provide a master schedule and milestones. 

Defense recently developed a contingency plan for the legacy system,
but this plan is perfunctory and does not meet the minimum criteria
defined in our Business Continuity and Contingency Planning guidance
which OMB has adopted as a standard for federal agencies. 
Specifically, the plan only states that if the legacy system fails,
critical personnel actions will be prepared using one of three other
commercial software packages.  The plan does not provide a
description of the resources, staff roles, procedures, and timetables
needed for its implementation.  And there is no evidence that Defense
(1) assessed and documented risks posed by external systems and the
public infrastructure, (2) defined the minimum acceptable level of
outputs and services for each core business process, or (3) assessed
the costs and benefits of contingency strategy alternatives. 

The steps detailed in our guide are integral to helping agencies to
manage the risk of potential Year 2000-induced disruptions to their
operations.  For example, the civilian personnel business area
depends on information and data provided by other Defense and federal
agencies whose systems can introduce Year 2000 problems into DCPDS. 
It also relies on services provided by the public infrastructure,
which are susceptible to Year 2000 problems that could disrupt
personnel operations--including power, water, and voice and data
telecommunications.  Until business continuity and contingency plans
are developed that focus on this chain of critical dependencies,
Defense will not be able to ensure that it can maintain the basic
functionality of its core civilian personnel operations. 


--------------------
\25 Defense has interface agreements for the legacy system that
define general interface partner relationships and responsibilities,
but these have not been updated to address these Year 2000 issues. 

\26 Year 2000 Computing Crisis:  Business Continuity and Contingency
Planning (GAO/AIMD-10.1.19).  Published as an exposure draft in March
1998 and finalized in August 1998. 


         NEW SYSTEM FACING SIMILAR
         RISKS
-------------------------------------------------------- Letter :7.5.2

Since the new system already has a four-digit year field, it does not
require renovation.  Defense has obtained certification of Year 2000
compliance on all applications in the new system and completed Year
2000 tests on the system.  However, CPMS has not identified all
system interfaces or developed agreements with its interface
partners.  In addition, while CPMS recently developed a contingency
plan, this plan is cursory.  It only states that if the modern system
fails, Defense will revert to using the legacy system for critical
personnel actions.  It is not based on a business impact analysis nor
does it describe resources, staff roles, procedures, and timetables
needed for its implementation. 

As stressed above, even if the modernized system is compliant,
Defense's civilian personnel management operations are at risk
because of dependencies on external systems and the public
infrastructure.  Therefore, until it develops specific interface
agreements and contingency plans that focus on critical dependencies,
it will have no assurance that it can prevent Year 2000-related
disruptions to critical personnel operations. 


   CONCLUSIONS
------------------------------------------------------------ Letter :8

Because Defense did not consider alternatives, such as centralizing
personnel functions, restructuring its regional and/or local offices
to serve multiple agencies and services, or integrating
payroll/personnel systems, its current regionalization approach may
not be optimal.  Defense lacked cost and performance data to analyze
the options and it faced resistance from Defense components.  While
it may have required more time to develop needed data and greater
management commitment to changing Defense business practices, the
potential for substantially greater savings and efficiencies should
have persuaded Defense to perform a rigorous analysis of all
alternatives and to select the one proven most cost effective. 

Additionally, because Defense did not adequately estimate and
evaluate costs, benefits, and returns, there is not adequate
assurance that its decision to replace the legacy system with the
Oracle COTS package is optimal.  Furthermore, Defense does not know
whether modifications to the Oracle product will satisfy its needs,
how it will maintain the system, how it will protect sensitive data
in the system, or how it will ensure the continuity of core civilian
personnel operations in the event of Year 2000 failures.  Despite
this uncertainty, Defense reports having already spent about $300
million on developing the system and establishing the regional
offices and plans to spend hundreds of millions of dollars more to
operate and support DCPDS and the regions. 


   RECOMMENDATIONS
------------------------------------------------------------ Letter :9

Before Defense starts to deploy the new system beyond test sites, we
recommend that the Secretary of Defense rigorously evaluate all
business and system alternatives to providing personnel services as
envisioned by the Clinger-Cohen Act, and, using this data and the
system test results, select the most cost beneficial business and
system alternative and develop and implement a transition plan for
that alternative. 

Specifically, business alternatives considered should include (1) use
of regions and local offices to serve specific agencies or services,
(2) use of regions or local offices to serve multiple agencies and
services, (3) centralizing all or parts of personnel management
operations that currently operate at component headquarters and major
commands, (4) integrating DOD's civilian personnel and payroll
management systems, (5) outsourcing civilian personnel computer
operations, (6) outsourcing all civilian personnel management
services, and (7) acquiring other commercially available products. 
In analyzing commercially available products, we recommend that
Defense consider the costs, benefits, and returns-on-investment of
all commercially available products that support personnel
management.  We also recommend that the analysis of commercially
available products consider technical risks, including whether each
available product can support Defense's needs and whether each one
can be modified in the future at a reasonable cost.  In evaluating
the range of business alternatives consideration should be given to
the substantial investment that has already been made in the current
approach. 

Regardless of the business and system alternative selected, we
recommend that Defense optimize it by collecting, analyzing and using
reliable cost and performance data and making improvements.  We also
recommend that, regardless of the chosen approach, Defense take the
following actions to mitigate technical, security, and Year 2000
risks. 

  -- To ensure that the system is adequately maintained and that
     modifications are carefully controlled, Defense should (1)
     develop agreements with system partners and interface partners
     to define responsibility for system operations, maintenance, and
     security, (2) establish a configuration control board comprised
     of system users to assist in deciding on which changes need to
     be made to the system, prioritizing change requests, and
     ensuring that changes are correctly made, (3) assign clear
     responsibility for providing technical assistance to Defense
     components. 

  -- To ensure that sensitive personnel data are adequately
     protected, Defense should (1) assess its risks and determine
     security needs, (2) define and implement appropriate policies
     and related controls, including standards for encrypting data
     and firewalls, (3) promote security awareness at all sites
     maintaining the system, and (4) continually monitor and evaluate
     policy and control effectiveness. 

  -- To mitigate Year 2000 risks, Defense should (1) establish
     interface agreements that clearly specify date format changes,
     time frames for these changes, and processes for resolving
     conflicts, (2) refine business continuity and contingency plans
     to ensure that they consider risks posed by external systems and
     infrastructure; assess the costs and benefits of alternative
     contingency strategies; and describe resources, staff roles,
     procedures, and timetables needed for implementation of the
     plan, and (3) test contingency plans to ensure that they are
     capable of providing the desired level of support to the
     agency's core business processes and can be implemented within a
     specified period of time. 


   AGENCY COMMENTS AND OUR
   EVALUATION
----------------------------------------------------------- Letter :10

The Acting Assistant Secretary for Force Management Policy provided
written comments on a draft of this report, which are reprinted in
appendix I.  He concurred with all five of our recommendations and
agreed to evaluate recommended alternatives as Defense proceeds with
its regionalization and modernization efforts. 

In concurring with our recommendations, however, Defense questioned
our use of the Clinger-Cohen Act of 1996 as criteria for evaluating
civilian personnel system decisions since these decisions were made
before the act took effect.  We used the Clinger-Cohen Act to
evaluate Defense's decisions because the act's requirements reflect
basic and widely accepted principles of sound system acquisition
management.  Similar practices are also called for in OMB Circulars
A-11 and A-130, the Chief Financial Officers Act of 1990, the
Government Performance and Results Act of 1993, the Federal
Acquisition Streamlining Act of 1994, and the Paperwork Reduction Act
of 1995--all of which were applicable in some manner to Defense's
decisions in this effort.  Moreover, Defense was required to follow
such practices by its own system acquisition regulations and
guidelines.  Finally, during the course of our review, Defense
officials responsible for DCPDS told us that they were attempting to
follow Clinger-Cohen Act principles in developing the system. 
Appendix I provides our detailed responses to Defense's views on our
recommendations and findings. 


--------------------------------------------------------- Letter :10.1

We are sending copies of this report to the Chairmen and Ranking
Minority Members of the Senate Committee on Armed Services; Senate
Committee on Governmental Affairs; Subcommittee on Defense, Senate
Committee on Appropriations; House Committee on Armed Services;
Subcommittee on Defense, House Committee on Appropriations; and
Senate and House Committees on the Budget; the Secretary of Defense;
the Senior Civilian Official of the Office of the Assistant Secretary
of Defense for Command, Control, Communications and Intelligence; the
Under Secretary of Defense (Comptroller); the Acting Assistant
Secretary of Defense for Force Management Policy; and the Director,
Office of Management and Budget.  Copies will also be made available
to others upon request. 

If you have any questions about this report, please call me or Carl
Urie, Assistant Director at (202) 512-6240.  Other major contributors
of this report are listed in appendix III. 

Jack L.  Brock, Jr.
Director, Governmentwide and Defense
 Information Systems




(See figure in printed edition.)Appendix I
COMMENTS FROM THE DEPARTMENT OF
DEFENSE
============================================================== Letter 



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)


The following are GAO's comments on the Department of Defense's
letter dated January 11, 1999. 

GAO COMMENTS

1.  Although the Clinger-Cohen Act was not in existence when DOD made
the initial decisions in developing the modern DCPDS, it has been in
effect since 1996 and should have been applied to all decisions made
subsequent to its enactment.  Further, OMB Circulars A-11 and A-130
existed prior to the initial decisions related to DCPDS and included
basic principles of sound system acquisition management.  In
addition, several acts that were in effect when the initial decisions
were made contain requirements similar to those outlined in the
Clinger-Cohen Act relating to improved information technology
management in the federal government.  For example (1) the Government
Performance and Results Act of 1993 (GPRA) requires federal agencies
to set strategic goals, measure performance, and report on
accomplishments, (2) the Federal Acquisition Streamlining Act of 1994
(FASA), Title V, requires agencies to define cost, schedule, and
performance goals for federal acquisition programs (including
information technology projects) and to monitor these projects to
ensure that they remain within prescribed tolerances, (3) the
Paperwork Reduction Act of 1995 (PRA) emphasizes achieving program
benefits and meeting agency goals through the effective use of
information technology, and (4) the Chief Financial Officers (CFO)
Act of 1990 focuses on the need to improve financial management and
reporting practices of the federal government, which is critical for
knowing an information technology project's actual costs and for
computing accurate returns on investment.  Finally, Defense's own
system acquisition regulations and guidelines, in existence at the
time Defense made the initial decisions in developing the modern
DCPDS, include requirements similar to those outlined in the
Clinger-Cohen Act related to basic principles of sound system
acquisition management. 

2.  Before embarking on an improvement approach for its civilian
personnel mission area, Defense performed cost and performance
analyses which indicated the Department's civilian personnel
servicing ratios could be improved significantly.  However, because
these analyses did not fully consider the costs and benefits of
numerous alternative business and systems approaches for improving
the servicing ratios, the Department may not have selected the most
cost-effective improvement approach. 

3.  We revised the report to delete specific information on the
scoring criteria used in the DCPDS procurement. 

4.  While Defense reports that it has already consolidated some
civilian personnel functions at component headquarters and major
commands and reduced staff by 23 percent, in June of 1998, there were
still 886 people performing civilian personnel management and
oversight functions at component headquarters and major command
levels at a cost of about $63 million a year.  Given that the
Civilian Personnel Management Service performs the same management
and oversight functions as component headquarters and major commands,
there are substantial opportunities for further consolidation and
staff reduction. 

5.  The A-76 study includes some but not all promising alternatives. 
While it will evaluate outsourcing civilian pay operations, it will
not consider outsourcing personnel operations or integrating
personnel and payroll systems.  Furthermore, while Defense considered
the possibility of outsourcing personnel computer operations in 1994,
it lacked the cost and performance data necessary to sufficiently
analyze this approach. 

6.  While it is important for Defense components to develop
comprehensive metrics to measure the timeliness and value of regional
service center work, they must also standardize these metrics so that
meaningful comparisons can be made across the Department.  The
components must also collect baseline data that define the current
operations so that Defense can determine whether new systems and
business strategies are achieving predicted cost and performance
improvements. 

7.  If implemented effectively, the site-by-site risk assessments and
other actions Defense is taking should help address the security
concerns identified in this report.  However, to maximize protection
over DCPDS data, Defense still needs to establish departmentwide
standards on encryption and firewalls. 

8.  Although CPMS has interface agreements with the owners of major
external interfaces for the legacy DCPDS system, those agreements
have not been adequately updated to include Year 2000 issues. 
Specifically, the agreements do not define agreed upon date formats,
nor describe how problems with data exchanges will be resolved. 
Further, as of the completion of our review, CPMS had not identified
the system interfaces or developed agreements with its interface
partners for the modern DCPDS. 

9.  Defense plans to complete interface agreements by April 1999 and
contingency plans by May 1999 and to begin testing contingency plans
by June 1999.  However, the Office of Management and Budget and GAO's
Year 2000 guidance recommend that agencies develop interface
agreements and realistic contingency plans during the assessment
phase, i.e., by August 1997, in order to minimize the risk of Year
2000 problems. 


SCOPE AND METHODOLOGY
========================================================== Appendix II

To analyze how Defense determined the number and locations for
civilian personnel regional service centers and why there is a wide
disparity in the number of regional centers among the services, we
interviewed Office of the Secretary of Defense, military service, and
Defense agency officials and reviewed guidance mandating
regionalization, the services' and Defense agencies' regionalization
studies, and their rationale for determining the number and location
of regions.  Where appropriate, we interviewed officials from CPMS,
the military services, and the Washington Headquarters Service to
understand perspectives regarding regionalization plans and status of
regionalization actions.  We visited five regional centers, toured
the facilities, and interviewed numerous officials.  These five
centers were Ft.  Riley, Kansas; Aberdeen Proving Ground, Maryland;
Silverdale, Washington; Randolph AFB, Texas; and Washington, D.C. 

To assess whether Defense is applying the Clinger-Cohen Act in
overseeing, managing, and developing DCPDS, we compared Defense's
actions taken on DCPDS to the investment principles included in the
act.  We reviewed GAO, OMB,\1 and Defense best practices guidance\2
for implementing the Clinger-Cohen Act and reviewed other Defense
policies and guidance for developing and implementing information
systems.  We analyzed selected major studies of information
technology and personnel management matters in Defense, including
studies by Coopers & Lybrand, a consulting organization\3 and the
Defense Science Board,\4 prior GAO studies of major defense
information systems projects, and selected Defense Office of
Inspector General reports.  We interviewed appropriate Defense and
OMB representatives familiar with personnel legislative requirements
and officials responsible for the development and oversight of DCPDS,
including officials from CPMS, the Major Automated Information System
Review Council (MAISRC), the Under Secretary of Defense/Comptroller,
the Comptroller's Program Analysis and Evaluation (PA&E) unit, and
service and agency staff responsible for regionalization, and DCPDS
program management. 

To determine whether DCPDS duplicates the Employee Express System
available through the Office of Personnel Management (OPM), we
reviewed documentation Defense prepared justifying the need for DCPDS
and Defense documentation reviewing the Employee Express System.  We
requested that OPM review and comment on Defense's rationale for not
using the Employee Express system; we requested that Defense respond
to OPM's comments; and we analyzed both Defense's and OPM's positions
on this issue.  In addition, we contacted representatives of six
other federal organizations that were developing new civilian
personnel systems and were not using the Employee Express system to
determine their rationale. 

To determine whether (1) Defense's civilian personnel management
requirements are sufficiently different to require extensive
modification of the commercial-off-the-shelf software (COTS)
application which Defense selected as the foundation for developing
DCPDS and (2) Defense leadership was aware of the extent and cost of
modifications that would be needed, we interviewed the Functional and
Acquisition Program managers and their staff as well as
representatives of the Oracle Corporation to solicit information on
the selection, acquisition, and modification of the Oracle COTS
product. 

To assess whether Defense identified and mitigated the risks
associated with the major modifications, we interviewed CDA officials
to determine Defense's actions to date, including those planned, in
process, and completed to address mitigating risks in overseeing,
managing, and developing DCPDS.  We reviewed pertinent regulations,
studies, and documentation, including the technical risk analysis,
configuration management plan, testing plans, and the Department's
Program Manager's Guide to Software Acquisition Best Practices.  As
requested, we determined whether Defense used this guide in
overseeing, managing, and developing DCPDS.  In assessing security
risks, we reviewed Defense's Deployment, Concept of Operations,
Encryption, Security Support, and Contingency Plans.  We reviewed
Defense directives and regulations on computer security, including
Regulation 5000.2-R, dated March 23, 1998, Directive 5200.28, dated
March 21, 1998, and Military Standard 498, dated December 1994.  In
addition, we assessed the physical security threats at four local and
four regional offices, through interviews and observations.  In
assessing Year 2000 risks, we reviewed the Year 2000 plans for the
legacy and modern systems and we compared these plans to our own Year
2000 Assessment Guide.\5 We conducted our review from August 1997
through July 1998 in accordance with generally accepted government
auditing standards. 


--------------------
\1 Office of Management and Budget, Capital Programming Guide,
Version 1.0, Supplement to Office of Management and Budget Circular
A-11, Part 3:  Planning, Budgeting, and Acquisition of Capital, July
1997. 

\2 Department of Defense Software Acquisition Best Practices
Initiative, The Program Manager's Guide to Software Acquisition
Practices, undated. 

\3 Department of Defense, Office of the Comptroller, Civilian
Personnel/Payroll Private Sector Benchmarking Survey, Final Report,
Coopers & Lybrand, September 21, 1994. 

\4 Defense Science Board, Report of the Defense Science Task Force: 
Military Personnel Information Management, August 31, 1996. 

\5 Year 2000 Computing Crisis:  An Assessment Guide
(GAO/AIMD-10.1.14).  Issued as an exposure draft in February 1997 and
finalized in September 1997. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix III

ACCOUNTING AND INFORMATION
MANAGEMENT DIVISION, WASHINGTON,
D.C. 

Dr.  Rona Stillman, Chief Scientist
Carl M.  Urie, Assistant Director
Brian C.  Spencer, Technical Assistant Director
Cristina T.  Chaplain, Communications Analyst
Robert L.  Crocker, Jr., Senior Evaluator

KANSAS CITY FIELD OFFICE

George L.  Jones, Evaluator-in-Charge
David R.  Solenberger, Senior Evaluator
Denise M.  Wempe, Senior Evaluator
Karl G.  Neybert, Staff Evaluator


*** End of document. ***