GAO-09-3SP, GAO Cost Estimating and Assessment Guide: Best Practices for Developing and Managing Capital Program Costs


This is the accessible text file for GAO report number GAO-09-3SP 
entitled 'GAO Cost Estimating And Assessment Guide' which was released 
on March 2, 2009. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as part 
of a longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

United States Government Accountability Office: 
GAO: 

Applied Research and Methods: 

GAO Cost Estimating And Assessment Guide: 
Best Practices For Developing And Managing Capital Program Costs: 

March 2009: 

GAO-09-3SP: 

Preface: 

The U.S. Government Accountability Office is responsible for, among 
other things, assisting the Congress in its oversight of the federal 
government, including agencies’ stewardship of public funds. To use 
public funds effectively, the government must meet the demands of 
today’s changing world by employing effective management practices and 
processes, including the measurement of government program performance. 
In addition, legislators, government officials, and the public want to 
know whether government programs are achieving their goals and what 
their costs are. To make those evaluations, reliable cost information 
is required and federal standards have been issued for the cost 
accounting that is needed to prepare that information.[Footnote 1] We 
developed the Cost Guide in order to establish a consistent methodology 
that is based on best practices and that can be used across the federal 
government for developing, managing, and evaluating capital program 
cost estimates. 

For the purposes of this guide, a cost estimate is the summation of 
individual cost elements, using established methods and valid data, to 
estimate the future costs of a program, based on what is known today. 
[Footnote 2] The management of a cost estimate involves continually 
updating the estimate with actual data as they become available, 
revising the estimate to reflect changes, and analyzing differences 
between estimated and actual costs—for example, using data from a 
reliable earned value management (EVM) system.[Footnote 3] 
 
The ability to generate reliable cost estimates is a critical function, 
necessary to support the Office of Management and Budget’s (OMB) 
capital programming process.[Footnote 4] Without this ability, agencies 
are at risk of experiencing cost overruns, missed deadlines, and 
performance shortfalls—all recurring problems that our program 
assessments too often reveal. Furthermore, cost increases often mean 
that the government cannot fund as many programs as intended or deliver 
them when promised. The methodology outlined in this guide is a 
compilation of best practices that federal cost estimating 
organizations and industry use to develop and maintain reliable cost 
estimates throughout the life of a government acquisition program. By 
default, the guide will also serve as a guiding principle for our 
auditors to evaluate the economy, efficiency, and effectiveness of 
government programs. 

The U.S. Government Accountability Office, the Congressional Budget 
Office (CBO), and others have shown through budget simulations that the 
nation is facing a large and growing structural deficit in the long 
term, primarily because the population is aging and health care costs 
are rising. As Comptroller General David Walker noted, “Continuing on 
this unsustainable path will gradually erode, if not suddenly damage, 
our economy, our standard of living and ultimately our national 
security.”[Footnote 5] New budgetary demands and demographic trends 
will place serious budgetary pressures on federal discretionary 
spending, as well as on other federal policies and programs, in the 
coming years. 

As resources become scarce, competition for them will increase. It is 
imperative, therefore, that government acquisition programs deliver as 
promised, not only because of their value to their users but also 
because every dollar spent on one program will mean one less available 
dollar to fund other efforts. To get better results, programs will need 
higher levels of knowledge when they start and standardized monitoring 
metrics such as EVM so that better estimates can be made of total 
program costs at completion. 

[End of Preface] 

Contents: 

Preface: 

Contents: 

Abbreviations: 

Introduction: 
The Guide’s Case Studies: 
The Cost Guide in Relation to Established Standards: 
The Guide’s Readers: 
Acknowledgments: 

Chapter 1: 
The Characteristics of Credible Cost Estimates and a Reliable Process 
for Creating Them: 
Basic Characteristics of Credible Cost Estimates: 
A Reliable Process for Developing Credible Cost Estimates: 

Chapter 2: 
Why Government Programs Need Cost Estimates and the Challenges in 
Developing Them: 
Cost Estimating Challenges: 
Earned Value Management Challenges: 

Chapter 3: 
Criteria for Cost Estimating, EVM, and Data Reliability: 

Chapter 4: 
Cost Analysis Overview: 
Differentiating Cost Analysis and Cost Estimating: 
Main Cost Estimate Categories: 
The Overall Significance of Cost Estimates: 
The Importance of Cost Estimates in Establishing Budgets: 
Cost Estimates and Affordability: 
Evolutionary Acquisition and Cost Estimation: 

Chapter 5: 
The Cost Estimate’s Purpose, Scope, and Schedule: 
Purpose: 
Scope: 

Chapter 6: 
The Cost Assessment Team: 
Team Composition and Organization: 
Cost Estimating Team Best Practices: 
Certification and Training for Cost Estimating and EVM Analysis: 

Chapter 7: 
Technical Baseline Description Definition and Purpose: 
Process: 
Schedule: 
Contents: 
Key System Characteristics and Performance Parameters: 

Chapter 8: 
Work Breakdown: Structure: 
Best Practice: Product-Oriented WBS: 
Common WBS Elements: 
WBS Development: 
Standardized WBS: 
WBS and Scheduling: 
WBS and EVM: 
WBS and Risk Management: 
WBS Benefits: 

Chapter 9: 
Ground Rules and Assumptions: 
Assumptions: 
Global and Element-Specific Ground Rules and Assumptions: 
Assumptions, Sensitivity, and Risk Analysis: 

Chapter 10: 
Data: 
Data Collection: 
Types of Data: 
Sources of Data: 
Data Applicability: 
Validating and Analyzing the Data: 
EVM Data Reliability: 
Data Normalization: 
Recurring and Nonrecurring Costs: 
Inflation Adjustments: 
Selecting the Proper Indexes: 
Data Documentation: 

Chapter 11: 
Developing a Point Estimate: 
Cost Estimating Methods: 
Production Rate Effects on Learning: 
Pulling the Point Estimate Together: 

Chapter 12: 
Estimating Software Costs: 
Unique Components of Software Estimation: 
Estimating Software Size: 
Estimating Software Development Effort: 
Software Maintenance: 
Parametric Software Estimation: 
Commercial Off-the-Shelf Software: 
Enterprise Resource Planning Software:
Software Costs Must also Account for Information Technology 
Infrastructure and Services: 
Unique Components of IT Estimation: 

Chapter 13: 
Sensitivity Analysis: 
Sensitivity Factors: 
Steps in Performing a Sensitivity Analysis: 
Sensitivity Analysis Benefits: 

Chapter 14: 
Cost Risk and Uncertainty: 
The Difference Between Risk and Uncertainty: 
Point Estimates Alone Are Insufficient for Good Decisions: 
Budgeting to a Realistic Point Estimate: 
Developing A Credible S Curve of Potential Program Costs: 
Risk Management: 

Chapter 15: 
Validating the Estimate: 
The Cost Estimating Community’s Best Practices for Validating 
Estimates: 

Chapter 16: 
Documenting the Estimate: 
Elements of Cost Estimate Documentation: 
Other Considerations: 

Chapter 17: 
Presenting the Estimate to Management: 

Chapter 18: 
Managing Program Costs: Planning: 
The Nature and History of EVM: 
Implementing EVM: 
Federal and Industry Guidelines for Implementing EVM: 
The Thirteen Steps in the EVM Process: 
Integrated Baseline Reviews: 
Award Fees: 
Progress and Performance-Based Payments Under Fixed-Price Contracts: 

Chapter 19: 
Managing Program Costs: 
Execution: 
Contract Performance Reports: 
Monthly EVM Analysis: 
Project Future Performance: 
Provide Analysis to Management: 
Continue EVM Until the Program is Complete: 

Chapter 20: 
Managing Program Costs: 
Updating: 
Incorporating Authorized Changes into the Performance Measurement 
Baseline: 
Using EVM System Surveillance to Keep the Performance Measurement
Baseline Current: 
Overtarget Baselines and Schedules: 
Update the Program Cost Estimate with Actual Costs: 
Keep Management Updated: 

Appendixes: 
 
Appendix 1: Auditing Agencies and Their Web Sites: 

Appendix 2: Case Study Backgrounds: 

Appendix 3: Experts Who Helped Develop This Guide: 

Appendix 4: The Federal Budget Process: 

Appendix 5: Federal Cost Estimating and EVM Legislation, Regulations, 
Policies, and Guidance: 

Appendix 6: Data Collection Instrument: 

Appendix 7: 
Data Collection Instrument: Data Request Rationale: 

Appendix 8: SEI Checklist: 

Appendix 9: Examples of Work Breakdown Structures: 

Appendix 10: Schedule Risk Analysis: 

Appendix 11: Learning Curve Analysis: 

Appendix 12: Technology Readiness Levels: 

Appendix 13: EVM-Related Award Fee Criteria: 

Appendix 14: Integrated Baseline Review Case Study and Other 
Supplemental Tools: 
Exhibit A: 
Exhibit B: 
Exhibit C: 
Exhibit D: 

Appendix 15: Common Risks to Consider in Software Cost Estimating: 

Appendix 16: Contacts and Acknowledgments: 

References: 

Image Sources: 

List of figures: 

Figure 1: The Cost Estimating Process: 

Figure 2: Challenges Cost Estimators Typically Face: 

Figure 3: Life-Cycle Cost Estimate for a Space System: 

Figure 4: Cone of Uncertainty: 

Figure 5: An Affordability Assessment: 

Figure 6: Typical Capital Asset Acquisition Funding Profiles by Phase: 

Figure 7: Evolutionary and Big Bang Acquisition Compared: 

Figure 8: Incremental Development: 

Figure 9: Disciplines and Concepts in Cost Analysis: 

Figure 10: A Product-Oriented Work Breakdown Structure: 

Figure 11: A Work Breakdown Structure with Common Elements: 

Figure 12: A Contract Work Breakdown Structure: 

Figure 13: A Learning Curve: 

Figure 14: A Sensitivity Analysis That Creates a Range around a Point 
Estimate: 

Figure 16: A Cumulative Probability Distribution, or S Curve: 

Figure 17: A Risk Cube Two-Dimensional Matrix: 

Figure 18: The Distribution of Sums from Rolling Two Dice: 

Figure 19: A Point Estimate Probability Distribution Driven by WBS 
Distributions: 

Figure 20: Integrating Cost Estimation, Systems Development Oversight, 
and Risk Management: 

Figure 21: Integrating EVM and Risk Management: 

Figure 22: Inputs and Outputs for Tracking Earned Value: 

Figure 23: WBS Integration of Cost, Schedule, and Technical 
Information: 

Figure 24: Identifying Responsibility for Managing Work at the Control 
Account: 

Figure 25: An Activity Network: 

Figure 26: Activity Durations as a Gantt Chart: 

Figure 27: Earned Value, Using the Percent Complete Method, Compared to
Planned Costs: 

Figure 28: The Genesis of the Performance Measurement Baseline: 

Figure 29: The Time-Phased Cumulative Performance Measurement Baseline: 

Figure 30: A Performance-Based Payments Structured Contract: 

Figure 31: The EVM System Acceptance Process: 

Figure 32: IBR Control Account Manager Discussion Template: 

Figure 33: Monthly Program Assessment Using Earned Value: 

Figure 34: Overall Program View of EVM Data: 

Figure 35: A Contract Performance Report’s Five Formats: 

Figure 36: Understanding Program Cost Growth by Plotting Budget at 
Completion Trends: 

Figure 37: Understanding Program Performance by Plotting Cost and 
Schedule Variances: 

Figure 38: Understanding Expected Cost Overruns by Plotting Estimate at
Completion: 

Figure 39: Rolling Wave Planning: 

Figure 40: The Effect on a Contract of Implementing an Overtarget 
Budget: 

Figure 41: Steps Typically Associated with Implementing an Overtarget 
Budget: 

Figure 42: Establishing a New Baseline with a Single Point Adjustment: 

Figure 43: MasterFormat™ Work Breakdown Structure: 

Figure 44: Network Diagram of a Simple Schedule: 

Figure 45: Example Project Schedule: 

Figure 46: Estimated Durations for Remaining WBS Areas in the Schedule: 

Figure 47: Cumulative Distribution of Project Schedule, Including Risk: 

Figure 48: Identified Risks on a Spacecraft Schedule: An Example: 

Figure 49: A Risk Register for a Spacecraft Schedule: 

Figure 50: Spacecraft Schedule Results from a Monte Carlo Simulation: 

Figure 51: A Schedule Showing Critical Path through Unit 2: 

Figure 52: Results of a Monte Carlo Simulation for a Schedule Showing 
Critical Path through Unit 2: 

Figure 53: Sensitivity Index for Spacecraft Schedule: 

Figure 54: Evaluation of Correlation in Spacecraft Schedule: 

Figure 55: An Example of Probabilistic Branching Contained in the 
Schedule: 

Figure 56: Probability Distribution Results for Probabilistic Branching 
in Test Unit: 

Figure 57: A Project Schedule Highlighting Correlation Effects: 

Figure 58: Risk Results Assuming No Correlation between Activity 
Durations: 

Figure 59: Risk Results Assuming 90 Percent Correlation between 
Activity Durations: 

Figure 60: Schedule Analysis Results with and without Correlation: 

Figure 61: IBR Team’s Program Summary Assessment Results for Program X: 

Figure 62: Program X IBR Team’s Assessment Results by Program Area: 

Figure 63: Program X IBR Team’s Detailed Assessment Results for an 
Individual Program Area: 

List of Tables: 

Table 1: GAO’s 1972 Version of the Basic Characteristics of Credible 
Cost Estimates: 

Table 2: The Twelve Steps of a High-Quality Cost Estimating Process: 

Table 3: Cost Estimating and EVM Criteria for Federal Agencies: 
Legislation, Regulations, Policies, and Guidance: 

Table 4: Life-Cycle Cost Estimates, Types of Business Case Analyses, 
and Other Types of Cost Estimates: 

Table 5: Certification Standards in Business, Cost Estimating, and 
Financial Management in the Defense Acquisition Education, Training, 
and Career Development Program: 

Table 6: Typical Technical Baseline Elements: 

Table 7: General System Characteristics: 

Table 8: Common Elements in Work Breakdown Structures: 

Table 9: Cost Element Structure for a Standard DOD Automated 
Information System: 

Table 10: Basic Primary and Secondary Data Sources: 

Table 11: Three Cost Estimating Methods Compared: 

Table 12: An Example of the Analogy Cost Estimating Method: 

Table 13: An Example of the Engineering Build-Up Cost Estimating 
Method: 

Table 14: An Example of the Parametric Cost Estimating Method: 

Table 15: The Twelve Steps of High-Quality Cost Estimating Summarized: 

Table 16: Sizing Metrics and Commonly Associated Issues: 

Table 17: Common Software Risks That Affect Cost and Schedule: 

Table 18: Best Practices Associated with Risks in Implementing ERP: 

Table 19: Common IT Infrastructure Risks: 

Table 20: Common Labor Categories Described: 

Table 21: Potential Sources of Program Cost Estimate Uncertainty: 

Table 22: A Hardware Risk Scoring Matrix: 

Table 23: A Software Risk Scoring Matrix: 

Table 24: Eight Common Probability Distributions: 

Table 25: The Twelve Steps of High-Quality Cost Estimating, Mapped to 
the Characteristics of a High-Quality Cost Estimate: 

Table 26: Questions for Checking the Accuracy of Estimating Techniques: 

Table 27: Eight Types of Independent Cost Estimate Reviews: 

Table 28: What Cost Estimate Documentation Includes: 

Table 29: Key Benefits of Implementing EVM: 

Table 30: Ten Common Concerns about EVM: 

Table 31: ANSI Guidelines for EVM Systems: 

Table 32: EVM Implementation Guides: 

Table 33: Typical Methods for Measuring Earned Value Performance: 

Table 34: Integrated Baseline Review Risk Categories: 

Table 35: Contract Performance Report Data Elements: Format 1: 

Table 36: EVM Performance Indexes: 

Table 37: Best Predictive EAC Efficiency Factors by Program Completion 
Status: 

Table 38: Basic Program Management Questions That EVM Data Help Answer: 

Table 39: ANSI Guidelines Related to Incorporating Changes in an EVM 
System: 

Table 40: Elements of an Effective Surveillance Organization: 

Table 41: Key EVM Processes across ANSI Guidelines for Surveillance: 

Table 42: Risk Factors That Warrant EVM Surveillance: 

Table 43: A Program Surveillance Selection Matrix: 

Table 44: A Color-Category Rating System for Summarizing Program 
Findings: 

Table 45: Overtarget Budget Funding Implications by Contract Type: 

Table 46: Common Indicators of Poor Program Performance: 

Table 47: Options for Treating Variances in Performing a Single Point 
Adjustment: 

Table 48: Case Studies Drawn from GAO Reports Illustrating This Guide: 

Table 49: Phases of the Budget Process: 

Table 50: The Budget Process: Major Steps and Time Periods: 

Table 51: Aircraft System Work Breakdown Structure: 

Table 52: Electronic/Automated Software System Work Breakdown 
Structure: 

Table 53: Ground Software Work Breakdown Structure: 

Table 54: Missile System Work Breakdown Structure: 

Table 55: Ordnance System Work Breakdown Structure: 

Table 56: Sea System Work Breakdown Structure: 

Table 57: Space System Work Breakdown Structure: 

Table 58: Surface Vehicle System Work Breakdown Structure: 

Table 59: Unmanned Air Vehicle System Work Breakdown Structure: 

Table 60: Department of Energy Project Work Breakdown Structure: 

Table 61: General Services Administration Construction Work Breakdown 
Structure: 

Table 62: Automated Information System: Enterprise Resource Planning 
Program Level Work Breakdown Structure: 

Table 63: Environmental Management Work Breakdown Structure: 

Table 64: Pharmaceutical Work Breakdown Structure: 

Table 65: Process Plant Construction Work Breakdown Structure: 

Table 66: Telecon Work Breakdown Structure: 

Table 67: Software Implementation Project Work Breakdown Structure: 

Table 68: Major Renovation Project Work Breakdown Structure: 

Table 69: Sample IT Infrastructure and Service Work Breakdown 
Structure: 

Table 70: CSI MasterFormat™ 2004 Structure Example: Construction 
Phase: 

Table 71: The Anderlohr Method for the Learning Lost Factor: 

Table 72: IBR Leadership Roles and Responsibilities: 

Case studies: 

Case Study 1: Basic Estimate Characteristics, from NASA, GAO-04-642: 

Case Study 2: Basic Estimate Characteristics, from Customs Service 
Modernization, GAO/AIMD-99-41: 

Case Study 3: Following Cost Estimating Steps, from NASA, GAO-04-642: 

Case Study 4: Cost Analysts’ Skills, from NASA, GAO-04-642: 

Case Study 5: Recognizing Uncertainty, from Customs Service 
Modernization, GAO/AIMD-99-41: 

Case Study 6: Using Realistic Assumptions, from Space Acquisitions, GAO-
07-96: 

Case Study 7: Program Stability Issues, from Combating Nuclear 
Smuggling, GAO-06-389: 

Case Study 8: Program Stability Issues, from Defense Acquisitions, GAO-
05-183: 

Case Study 9: Development Schedules, from Defense Acquisitions, GAO-06-
327: 

Case Study 10: Risk Analysis, from Defense Acquisitions, GAO-05-183: 

Case Study 11: Risk Analysis, from NASA, GAO-04-642: 

Case Study 12: Applying EVM, from Cooperative Threat Reduction, GAO-06-
692: 

Case Study 13: Rebaselining, from NASA, GAO-04-642: 

Case Study 14: Realistic Estimates, from Defense Acquisitions, GAO-05-
183: 

Case Study 15: Importance of Realistic LCCEs, from Combating Nuclear 
Smuggling, GAO-07-133R: 

Case Study 16: Importance of Realistic LCCEs, from Space Acquisitions, 
GAO-07-96: 

Case Study 17: Evolutionary Acquisition and Cost Estimates, from Best 
Practices, GAO-03-645T: 

Case Study 18: Incremental Development, from Customs Service 
Modernization, GAO/AIMD-99-41: 

Case Study 19: The Estimate’s Context, from DOD Systems Modernization, 
GAO-06-215: 

Case Study 20: Defining Requirement, from United States Coast Guard, 
GAO-06-623: 

Case Study 21: Managing Requirements, from DOD Systems Modernization, 
GAO-06-215: 

Case Study 22: Product-Oriented Work Breakdown Structure, from 
Air Traffic Control, GAO-08-756: 

Case Study 23: Developing Work Breakdown Structure, from NASA, 
GAO-04-642: 

Case Study 24: Developing Work Breakdown Structure, from Homeland 
Security, GAO-06-296: 

Case Study 25: The Importance of Assumptions, from Space Acquisitions, 
GAO-07-96: 

Case Study 26: Testing Ground Rules for Risk, from Space Acquisitions, 
GAO-07-96: 

Case Study 27: The Industrial Base, from Defense Acquisition, GAO-05-
183: 

Case Study 28: Technology Maturity, from Defense Acquisitions, GAO-05-
183: 

Case Study 29: Technology Maturity, from Space Acquisitions, GAO-07-96: 

Case Study 30: Informing Management of Changed Assumptions, from 
Customs Service Modernization, GAO/AIMD-99-41: 

Case Study 31: Fitting the Estimating Approach to the Data, from Space 
Acquisitions, GAO-07-96: 

Case Study 32: Data Anomalies, from Cooperative Threat Reduction, GAO-
06-692: 

Case Study 33: Inflation, from Defense Acquisitions, GAO-05-183: 

Case Study 34: Cost Estimating Methods, from Space Acquisitions, GAO-07-
96: 

Case Study 35: Expert Opinion, from Customs Service Modernization, 
GAO/AIMD-99-41: 

Case Study 36: Production Rate, from Defense Acquisitions, GAO-05-183: 

Case Study 37: Underestimating Software, from Space Acquisitions, GAO-
07-96: 

Case Study 38: Sensitivity Analysis, from Defense Acquisitions, GAO-05-
183: 

Case Study 39: Point Estimates, from Space Acquisitions, GAO-07-96: 
 
Case Study 40: Point Estimates, from Defense Acquisitions, GAO-05-183: 

Case Study 41: Validating the Estimate, from Chemical Demilitarization, 
GAO-07-240R: 

Case Study 42: Independent Cost Estimates, from Space Acquisitions, GAO-
07-96: 

Case Study 43: Documenting the Estimate, from Telecommunications, GAO-
07-268: 

Case Study 44: Validating the EVM System, from Cooperative Threat 
Reduction, GAO-06-692: 

Case Study 45: Validating the EVM System, from DOD Systems 
Modernization, GAO-06-215: 

Case Study 46: Cost Performance Reports, from Defense Acquisitions, GAO-
05-183: 

Case Study 47: Maintaining Performance Measurement Baseline Data, from 
National Airspace System, GAO-03-343: 

Case Study 48: Maintaining Realistic Baselines, from Uncertainties 
Remain, GAO-04-643R: 

Best Practices Checklist: 

1. The Estimate: 

2. Purpose, Scope, and Schedule: 

3. Cost Assessment Team: 

4. Technical Baseline Description: 

5. Work Breakdown Structure: 

6. Ground Rules and Assumptions: 

7. Data: 

8. Developing a Point Estimate: 

9. Estimating Software Costs: 

10. Sensitivity Analysis: 

11. Cost Risk and Uncertainty: 

12. Validating the Estimate: 

13. Documenting the Estimate: 

14. Presenting the Estimate to Management: 

15. Managing Program Costs: Planning: 

16. Managing Program Costs: Execution: 

17. Managing Program Costs: Updating: 

Abbreviations: 

ACWP: actual cost of work performed: 

ANSI: American National Standards Institute: 

AOA: analysis of alternatives: 

BAC: budget at completion: 

BCA: business case analysis: 

BCWP: budgeted cost for work performed: 

BCWS: budgeted cost for work scheduled: 

CAIG: Cost Analysis Improvement Group: 

CBO: Congressional Budget Office: 

CEA: cost-effectiveness analysis: 

CER: cost estimating relationship: 

COSMIC: Common Software Measurement International Consortium: 

CPI: cost performance index: 

CPR: contract performance report: 

C/SCSC: Cost/Schedule and Control System: 

CSDR: cost and software data report: 

DAU: Defense Acquisition University: 

DCAA: Defense Contract Audit Agency: 

DCMA: Defense Contract Management: 

DOD: Department of Defense: 

EA: economic analysis: 

EAC: estimate at completion: 

EIA: Electronic Industries Alliance: 

ERP: enterprise resource planning: 

EVM: earned value management: 

FAR: Federal Acquisition Regulation: 

GR&A: ground rules and assumptions: 

IBR: integrated baseline review: 

ICA: independent cost assessment: 

ICE: independent cost estimate: 

IGCE: independent government cost estimate: 

IMS: integrated master schedule: 

IT: information technology: 

LCCE: life-cycle cost estimate: 

NAR: nonadvocate review: 

NASA: National Aeronautics and Space Administration: 

NDIA: National Defense Industrial Association: 

OMB: Office of Management and Budget Criteria: 

OTB: overtarget baseline: 

OTS: overtarget schedule: 

PMB: performance measurement baseline: 

PMI: Project Management Institute: 

SCEA: Society of Cost Estimating and Agency Analysis: 

SEI: Software Engineering Institute: 

SLOC: source line of code: 

SPI: schedule performance index: 

TCPI: to complete performance index: 

WBS: work breakdown structure: 

[End of section] 

Introduction: 

Because federal guidelines are limited on processes, procedures, and 
practices for ensuring credible cost estimates, the Cost Guide is 
intended to fill that gap. Its purpose is twofold—to address generally 
accepted best practices for ensuring credible program cost estimates 
(applicable across government and industry) and to provide a detailed 
link between cost estimating and EVM. Providing that link is especially 
critical, because it demonstrates how both elements are needed for 
setting realistic program baselines and managing risk. 

As a result, government managers and auditors should find in the Cost 
Guide principles to guide them as they assess (1) the credibility of a 
program’s cost estimate for budget and decision making purposes and (2) 
the program’s status using EVM. Throughout this guide, we refer to 
program cost estimates that encompass major system acquisitions, as 
well as government in-house development efforts for which a cost 
estimate must be developed to support a budget request. 

The basic information in the Cost Guide includes the purpose, scope, 
and schedule of a cost estimate; a technical baseline description; a 
work breakdown structure (WBS); ground rules and assumptions; how to 
collect data; estimation methodologies; software cost estimating; 
sensitivity and risk analysis; validating a cost estimate; documenting 
and briefing results; updating estimates with actual costs; EVM; and 
the composition of a competent cost estimating team.[Footnote 6] The 
guide discusses pitfalls associated with cost estimating and EVM that 
can lead government agencies to accept unrealistic budget requests—as 
when risks are embedded in an otherwise logical approach to estimating 
costs. Since the Department of Defense (DOD) is considered the leader 
in government cost estimating, the guide relies heavily on DOD for 
terminology and examples that may not be used by, or even apply to, 
other federal agencies. 

Chapters 1–17 of the Cost Guide discuss the importance of cost 
estimating and best practices associated with creating credible cost 
estimates. They describe how cost estimates predict, analyze, and 
evaluate a program’s cost and schedule and serve as a critical program 
control planning tool. Once cost estimates have been presented to and 
approved by management, the chapters also establish the basis for 
measuring actual performance against the approved baseline plan using 
an EVM system. 

Those chapters explain how EVM, if it is to work, must have a cost 
estimate that identifies the effort that is needed—the work breakdown 
structure—and the period of time over which the work is to be 
performed—the program schedule.[Footnote 7] In essence, the cost 
estimate is the basis for establishing the program’s detailed schedule, 
and it identifies the bounds for how much program costs can be expected 
to vary, depending on the uncertainty analysis. When all these tasks 
are complete, the cost estimate can be used to lay the foundation for 
the performance measurement baseline (PMB), which will measure actual 
program performance. 

Since sound acquisition management requires more than just a reliable 
cost estimate at a project’s outset, chapters 18–20 provide guidance on 
converting the cost estimate into an executable program and a means for 
managing program costs. Our program assessments have too often revealed 
that not integrating cost estimation, system development oversight, and 
risk management—three key disciplines, interrelated and essential to 
effective acquisition management—has resulted in programs costing more 
than planned and delivering less than promised. Therefore, chapters 
18–20 address best practices in implementing and integrating these 
disciplines and using them to manage costs throughout the life of a 
program. 

OMB has set the expectation that programs will maintain current 
estimates of cost. This requires rigorous performance-based program 
management, which can be satisfied with EVM. Chapters 18–20 address the 
details of EVM, which is designed to integrate cost estimation, system 
development oversight, and risk management. Additionally, for programs 
classified as major acquisitions—regardless of whether the development 
work is completed in-house or under contract—the use of EVM is a 
requirement for development, as specified by OMB.[Footnote 8] The 
government may also require the use of EVM for other acquisitions, in 
accordance with agency procedures. 

Since linking cost estimating and EVM results in a better view of a 
program and allows for greater understanding of program risks, cost 
estimators and EVM analysts who join forces can use each other’s data 
to update program costs and examine differences between estimated and 
actual costs. This way, scope changes, risks, and other opportunities 
can be presented to management in time to plan for and mitigate their 
impact. In addition, program status can be compared to historical data 
to better understand variances. Finally, cost estimators can help EVM 
analysts calculate a cumulative probability distribution to determine 
the level of confidence in the baseline. 

But bringing a program to successful completion requires knowing 
potential risks and identifying ways to respond to them before they 
happen—using risk management to identify, mitigate, and assign 
resources to manage risks so that their impact can be minimized. This 
requires the support of many program management and engineering staff 
and it results in better performance and more reliable predictions of 
program outcomes. By integrating EVM data and risk management, program 
managers can develop current estimates at completion (EAC) for all 
levels of management, including OMB reporting requirements. Therefore, 
chapters 18–20 expand on these concepts by examining program cost 
planning, execution, and updating. 

The Guide’s Case Studies: 

The Cost Guide contains a number of case studies drawn from GAO program 
reviews. The case studies highlight problems typically associated with 
cost estimates and augment the key points and lessons learned that the 
chapters discuss. For example, GAO has found that in many programs cost 
growth results from optimistic assumptions about technological 
enhancements. Experts on cost estimating have also found that many 
program managers believe they can deliver state-of-the-art technology 
upgrades within a constrained budget before proof is available that the 
requirements are feasible. Studies have shown that it costs more to 
develop technology from scratch than to develop it incrementally over 
time.[Footnote 9] Appendix II gives some background information for 
each program used in the case studies. (Appendix I is a list of 
auditing agencies.) 

The Cost Guide In Relation To Established Standards: 

Our intent is to use this Cost Guide in conjunction with Government 
Auditing Standards and Standards for Internal Control in the Federal 
Government, commonly referred to as the yellow book and the green book, 
respectively.[Footnote 10] If auditors cite compliance with these 
standards and internal controls and find inconsistencies between them 
and the Cost Guide, they should defer to the yellow and green books for 
the prevailing rules. 

This guide’s reference list identifies cost estimating guides and 
sources available from other government agencies and organizations that 
we relied on to determine the processes, practices, and procedures most 
commonly recommended in the cost estimating community. Users of the 
guide may wish to refer to those references for more information. In 
addition, we relied on information from the Society of Cost Estimating 
and Analysis (SCEA), which provides standards for cost estimating, and 
the Project Management Institute (PMI), which provides EVM standards. 
[Footnote 11] 
 
The Guide’s Readers: 

The federal audit community is the primary audience for this guide. In 
addition, agencies that do not have a formal policy for conducting or 
reviewing cost estimates will benefit from it, because it will inform 
them of the criteria GAO uses in assessing a cost estimate’s 
credibility. Besides GAO, auditing agencies include Inspectors General 
and audit services such as the Naval Audit Service and the Army Audit 
Agency. Appendix I lists other auditing agencies that GAO may contact 
at the start of an audit. The list may help ease the burden on agencies 
as they work to meet the needs of various oversight offices and should 
help speed up delivery of data request items. 

We intend to periodically update the Cost Guide. Comments and 
suggestions from experienced users are always welcome, as are 
recommendations from experts in the cost estimating and EVM 
disciplines. 

Acknowledgments: 

The Cost Guide team thanks the many members of the cost community who 
helped make the guide a reality. After we discussed our plans for 
developing the guide with members of the cost community, several 
experts expressed interest in working with us. The number of experts 
who helped us create this guide grew over time, beginning with our 
first meeting in June 2005. Their contributions were invaluable. 

Together with these experts, GAO has developed a guide that clearly 
outlines its criteria for assessing cost estimates and EVM data during 
audits and that we believe will benefit all agencies in the federal 
government. We would like to thank everyone who gave their time by 
attending meetings, giving us valuable documentation, and providing 
comments. Those who worked with us on this guide are listed in appendix 
III. Additional acknowledgments are in appendix XVI. 

Chapter 1: The Characteristics Of Credible Cost Estimates And A 
Reliable Process For Creating Them: 

More than 30 years ago, we reported that realistic cost estimating was 
imperative when making wise decisions in acquiring new systems. In 
1972, we published a report called Theory and Practice of Cost 
Estimating for Major Acquisitions, in which we stated that estimates of 
the cost to develop and produce weapon systems were frequently 
understated, with cost increases on the order of $15.6 billion from 
early development estimates.[Footnote 12] In that report, we identified 
factors in the cost estimating function that were causing this problem 
and offered suggestions for solving or abating the problem of 
unexpected cost growth. 

We found that uniform guidance on cost estimating practices and 
procedures that would be the basis for formulating valid, consistent, 
and comparable estimates was lacking within DOD. In fact, evidence 
showed that each military service issued its own guidance for creating 
cost estimates and that the guidance ranged from a detailed estimating 
manual to a few general statements. In addition, we reported that cost 
estimators often ignored this guidance.[Footnote 13] 

In that 1972 report, we also stated that cost estimates for specific 
systems were frequently revisions of previously developed estimates and 
that accurate revisions of both the original and updated cost estimates 
required documentation showing data sources, assumptions, methods, and 
decisions basic to the estimates. However, we discovered that in 
virtually every system we reviewed for the report, documentation 
supplying such information was inaccurate or lacking. Among the 
resulting difficulties were that: 
 
* known costs had been excluded without adequate or valid 
justification; 
 
* historical cost data used for computing estimates were sometimes 
invalid, unreliable, or unrepresentative;
 
* inflation was not always included or was not uniformly treated when 
it was included; and; 

* understanding the proper use of the estimates was hindered because 
the estimates were too low.[Footnote 14] 
 
Another finding was that readily retrievable cost data that could serve 
in computing cost estimates for new weapon systems were generally 
lacking. Additionally, organized and systematic efforts were not made 
to gather actual cost information to achieve comparability between data 
collected on various weapon systems or to see whether the cost data the 
contractors reported were accurate and consistent.[Footnote 15] 

Our conclusion was that without realism and objectivity in the cost 
estimating process, bias and overoptimism creep into estimates that 
advocates of weapon systems prepare, and the estimates tend to be too 
low. Therefore, staff not influenced by the military organization’s 
determination to field a weapon system, or by the contractor’s 
intention to develop and produce the system, should review every weapon 
system at major decision points in the acquisition.[Footnote 16] 
 
Basic Characteristics Of Credible Cost Estimates: 
 
The basic characteristics of effective estimating have been studied and 
highlighted many times. Their summary, in table 1, is from our 1972 
report, Theory and Practice of Cost Estimating for Major Acquisitions. 
These characteristics are still valid today and should be found in all 
sound cost analyses. 

Table 1: GAO’s 1972 Version of the Basic Characteristics of Credible 
Cost Estimates: 
 
Characteristic: Clear identification of task; 
Description: Estimator must be provided with the system description, 
ground rules and assumptions, and technical and performance 
characteristics; Estimate’s constraints and conditions must be clearly 
identified to ensure the preparation of a well-documented estimate. 
 
Characteristic: Broad participation in preparing estimates; 
Description: All stakeholders should be involved in deciding mission 
need and requirements and in defining system parameters and other 
characteristics Data should be independently verified for accuracy, 
completeness, and reliability. 
 
Characteristic: Availability of valid data; 
Description: Numerous sources of suitable, relevant, and available data 
should be used; Relevant, historical data should be used from similar 
systems to project costs of new systems; these data should be directly 
related to the system’s performance characteristics. 
 
Characteristic: Standardized structure for the estimate; 
Description: A standard work breakdown structure, as detailed as 
possible, should be used, refining it as the cost estimate matures and 
the system becomes more defined; The work breakdown structure ensures 
that no portions of the estimate are omitted and makes it easier to 
make comparisons to similar systems and programs. 

Characteristic: Provision for program uncertainties; 
Description: Uncertainties should be identified and allowance developed 
to cover the cost effect; Known costs should be included and unknown 
costs should be allowed for. 
 
Characteristic: Recognition of inflation; 
Description: The estimator should ensure that economic changes, such as 
inflation, are properly and realistically reflected in the life-cycle 
cost estimate. 

Characteristic: Recognition of excluded costs; 
Description: All costs associated with a system should be included; any 
excluded costs should be disclosed and given a rationale. 
 
Characteristic: Independent review of estimates; 
Description: Conducting an independent review of an estimate is crucial 
to establishing confidence in the estimate; the independent reviewer 
should verify, modify, and correct an estimate to ensure realism, 
completeness, and consistency. 
 
Characteristic: Revision of estimates for significant program changes; 
Description: Estimates should be updated to reflect changes in a 
system’s design requirements. Large changes that affect costs can 
significantly influence program decisions. 
 
Source: GAO. 

[End of table] 

In a 2006 survey to identify the characteristics of a good estimate, 
participants from a wide variety of industries—aerospace, automotive, 
energy—as well as consulting firms and the U.S. Navy and Marine Corps 
corroborated the continuing validity of the characteristics in table 1. 
Despite the fact that these basic characteristics have been published 
and known for decades, we find that many agencies still lack the 
ability to develop cost estimates that can satisfy them. Case studies 1 
and 2, drawn from GAO reports, show the kind of cross-cutting findings 
we have reported in the past. Because of findings like those in case 
studies 1 and 2, the Cost Guide provides best practice processes, 
standards, and procedures for developing, implementing, and evaluating 
cost estimates and EVM systems and data. By satisfying these criteria, 
agencies should be able to better manage their programs and inform 
decision makers of the risks involved. 

Case Study 1: Basic Estimate Characteristics, from NASA, GAO-04-642. 
GAO found that the National Aeronautics and Space Administration’s 
(NASA) basic cost estimating processes—an important tool for managing 
programs—lacked the discipline needed to ensure that program estimates 
were reasonable. Specifically, none of the 10 NASA programs GAO 
reviewed in detail met all GAO’s cost estimating criteria, which are 
based on criteria Carnegie Mellon University’s Software Engineering 
Institute developed. Moreover, none of the 10 programs fully met 
certain key criteria—including clearly defining the program’s life 
cycle to establish program commitment and manage program costs, as 
required by NASA. 

In addition, only 3 programs provided a breakdown of the work to be 
performed. Without this knowledge, the programs’ estimated costs could 
be understated and thereby subject to underfunding and cost overruns, 
putting programs at risk of being reduced in scope or requiring 
additional funding to meet their objectives. Finally, only 2 programs 
had a process in place for measuring cost and performance to identify 
risks. 

Source: GAO, NASA: Lack of Disciplined Cost-Estimating Processes 
Hinders Effective Program Management, [hyperlink, 
http://www.gao.gov/cgi-bin/getrpt?GAO-04-642] (Washington, D.C.: May 
28, 2004) 

[End of case study] 

Case Study 2: Basic Estimate Characteristics, from Customs Service 
Modernization, GAO/AIMD-99-41. GAO analyzed the U.S. Customs Service 
approach to deriving its $1.05 billion Automated Commercial Environment 
life-cycle cost estimate with Software Engineering Institute (SEI) 
criteria. SEI had seven questions for decision makers to use in 
assessing the reliability of a project’s cost estimate and detailed 
criteria to help evaluate how well a project satisfies each question. 
Among the criteria were several very significant and closely 
intertwined requirements that are at the core of effective cost 
estimating. Specifically, embedded in several of the questions were 
requirements for using (1) formal cost models; (2) structured and 
documented processes for determining the software size and reuse inputs 
to the models; and (3) relevant, measured, and normalized historical 
cost data (estimated and actual) to calibrate the models. 

GAO found that Customs did not satisfy any of these requirements. 
Instead of using a cost model, it used an unsophisticated spreadsheet 
to extrapolate the cost of each Automated Commercial Environment 
increment. Its approach to determining software size and reuse was not 
documented and was not well supported or convincing. Customs had no 
historical project cost data when it developed the $1.05 billion 
estimate and did not account for relevant, measured, and normalized 
differences in the increments. Clearly, such fundamental changes can 
dramatically affect system costs and should have been addressed 
explicitly in Customs’ cost estimates. 

Source: GAO, Customs Service Modernization: Serious Management and 
Technical Weaknesses Must Be Corrected, [hyperlink, 
http://www.gao.gov/cgi-bin/getrpt?GAO/AMD-99-41] Washington, D.C.: Feb. 
26, 1999. 

[End of case study] 

A Reliable Process For Developing Credible Cost Estimates: 
 
Certain best practices should be followed if accurate and credible cost 
estimates are to be developed. These best practices represent an 
overall process of established, repeatable methods that result in high-
quality cost estimates that are comprehensive and accurate and that can 
be easily and clearly traced, replicated, and updated. Figure 1 shows 
the cost estimating process. 

Figure 1: The Cost Estimating Process: 

[Refer to PDF for image: illustration] 

Initiation and research: 
Your audience, what you are estimating, and why you are estimating it 
are of the utmost importance. 

Assessment: 
Cost assessment steps are iterative and can be accomplished in varying 
order or concurrently. 

Analysis: 
The confidence in the point or range of the estimate is crucial to the 
decision maker. 

Presentation: 
Documentation and presentation make or break a cost estimating decision 
outcome. 

Define the estimate's purpose; 
Develop the estimating plan; 
- Define the program; 
- Determine the estimating structure; 
- Identify ground rules and assumptions; 
- Obtain the data; 
- Develop the point estimate and compare it to an independent cost 
estimate; 
Conduct sensitivity; 
Conduct a risk and uncertainty analysis; 
Document the estimate; 
Present estimate to management for approval; 
Update the estimate to reflect actual costs/changes. 

Analysis, presentation, and updating the estimate steps can lead to 
repeating previous assessment steps. 

Source: GAO. 

[End of figure] 

We have identified 12 steps that, followed correctly, should result in 
reliable and valid cost estimates that management can use for making 
informed decisions. Table 2 identifies all 12 steps and links each one 
to the chapter in this guide where it is discussed. 

Table 2: The Twelve Steps of a High-Quality Cost Estimating Process: 
 
Step: 1; 
Description: Define estimate’s purpose; 
Associated task: 
* Determine estimate’s purpose, required level of detail, and overall 
scope; 
* Determine who will receive the estimate; 
Chapter: 5. 

Step: 2; 
Description: Develop estimating plan; 
Associated task: 
* Determine the cost estimating team and develop its master schedule; 
* Determine who will do the independent cost estimate; 
* Outline the cost estimating approach; 
* Develop the estimate timeline; 
Chapter: 5 and 6. 

Step: 3; 
Description: Define program characteristics; 
Associated task: 
* In a technical baseline description document, identify the program’s 
purpose and its system and performance characteristics and all system 
configurations; 
* Any technology implications; 
* Its program acquisition schedule and acquisition strategy; 
* Its relationship to other existing systems, including predecessor or 
similar legacy systems; 
* Support (manpower, training, etc.) and security needs and risk items; 
* System quantities for development, test, and production; 
* Deployment and maintenance plans; 
Chapter: 7. 

Step: 4; 
Description: Determine estimating structure 
Associated task: 
* Define a work breakdown structure (WBS) and describe each element in 
a WBS dictionary (a major automated information system may have only a 
cost element structure); 
* Choose the best estimating method for each WBS element; 
* Identify potential cross-checks for likely cost and schedule drivers; 
* Develop a cost estimating checklist; 
Chapter: 8.

Step: 5; 
Description: Identify ground rules and assumptions; 
Associated task: 
* Clearly define what the estimate includes and excludes; 
* Identify global and program-specific assumptions, such as: 
- the estimate’s base year, including time-phasing and life cycle; 
* Identify program schedule information by phase and program 
acquisition strategy; 
* Identify any schedule or budget constraints, inflation assumptions, 
and travel costs; 
* Specify equipment the government is to furnish as well as the use of 
existing facilities or new modification or development; 
* Identify prime contractor and major subcontractors; 
* Determine technology refresh cycles, technology assumptions, and new 
technology to be developed; 
* Define commonality with legacy systems and assumed heritage savings; 
Describe effects of new ways of doing business; 
Chapter: 9. 

Step: 6; 
Description: Obtain data; 
Associated task: 
* Create a data collection plan with emphasis on collecting current and 
relevant technical, programmatic, cost, and risk data; 
* Investigate possible data sources; 
* Collect data and normalize them for cost accounting, inflation, 
learning, and quantity adjustments; 
* Analyze the data for cost drivers, trends, and outliers and compare 
results against rules of thumb and standard factors derived from 
historical data; 
* Interview data sources and document all pertinent information, 
including an assessment of data reliability and accuracy; 
* Store data for future estimates; 
Chapter: 10. 

Step: 7; 
Description: Develop point estimate and compare it to an independent 
cost estimate; 
Associated task: 
* Develop the cost model, estimating each WBS element, using the best 
methodology from the data collected, and including all estimating 
assumptions[A]; 
* Express costs in constant year dollars; 
* Time-phase the results by spreading costs in the years they are 
expected to occur, based on the program schedule; 
* Sum the WBS elements to develop the overall point estimate; 
* Validate the estimate by looking for errors like double counting and 
omitted costs; 
* Compare estimate against the independent cost estimate and examine 
where and why there are differences; 
* Perform cross-checks on cost drivers to see if results are similar; 
* Update the model as more data become available or as changes occur 
and compare results against previous estimates; 
Chapter: 11, 12, and 15. 

Step: 8; 
Description: Conduct sensitivity analysis; 
Associated task: 
* Test the sensitivity of cost elements to changes in estimating input 
values and key assumptions; 
* Identify effects on the overall estimate of changing the program 
schedule or quantities; 
* Determine which assumptions are key cost drivers and which cost 
elements are affected most by changes; 
Chapter: 13. 

Step: 9; 
Description: Conduct risk and uncertainty analysis; 
Associated task: 
* Determine and discuss with technical experts the level of cost, 
schedule, and technical risk associated with each WBS element; 
* Analyze each risk for its severity and probability; 
* Develop minimum, most likely, and maximum ranges for each risk 
element; 
* Determine type of risk distributions and reason for their use; 
* Ensure that risks are correlated; 
* Use an acceptable statistical analysis method (e.g., Monte Carlo 
simulation) to develop a confidence interval around the point estimate; 
* Identify the confidence level of the point estimate; 
* Identify the amount of contingency funding and add this to the point 
estimate to determine the risk-adjusted cost estimate; 
* Recommend that the project or program office develop a risk 
management plan to track and mitigate risks; 
Chapter: 14.

Step: 10; 
Description: Document the estimate; 
Associated task: 
* Document all steps used to develop the estimate so that a cost 
analyst unfamiliar with the program can recreate it quickly and produce 
the same result; 
* Document the purpose of the estimate, the team that prepared it, and 
who approved the estimate and on what date; 
* Describe the program, its schedule, and the technical baseline used 
to create the estimate; 
* Present the program’s time-phased life-cycle cost; 
* Discuss all ground rules and assumptions; 
* Include auditable and traceable data sources for each cost element 
and document for all data sources how the data were normalized; 
* Describe in detail the estimating methodology and rationale used to 
derive each WBS element’s cost (prefer more detail over less); 
* Describe the results of the risk, uncertainty, and sensitivity 
analyses and whether any contingency funds were identified; 
* Document how the estimate compares to the funding profile; 
* Track how this estimate compares to any previous estimates; 
Chapter: 16.

Step: 11; 
Description: Present estimate to management for approval; 
Associated task: 
* Develop a briefing that presents the documented life-cycle cost 
estimate; 
* Include an explanation of the technical and programmatic 
baseline and any uncertainties; 
* Compare the estimate to an independent cost estimate (ICE) and 
explain any differences; 
* Compare the estimate (life-cycle cost estimate (LCCE)) or independent 
cost estimate to the budget with enough detail to easily defend it by 
showing how it is accurate, complete, and high in quality; 
* Focus in a logical manner on the largest cost elements and cost 
drivers; 
* Make the content clear and complete so that those who are unfamiliar 
with it can easily comprehend the competence that underlies the 
estimate results; 
* Make backup slides available for more probing questions; 
* Act on and document feedback from management; 
* Request acceptance of the estimate; 
Chapter: 17. 

Step: 12; 
Description: Update the estimate to reflect actual costs and changes 
Associated task: 
* Update the estimate to reflect changes in technical or program 
assumptions or keep it current as the program passes through new phases 
or milestones; 
* Replace estimates with EVM and independent estimate at completion 
(EAC) from the integrated EVM system; 
* Report progress on meeting cost and schedule estimates; 
* Perform a post mortem and document lessons learned for elements whose 
actual costs or schedules differ from the estimate; 
* Document all changes to the program and how they affect the cost 
estimate 
Chapter: 16, 18, 19, and 20. 

Source: GAO, DHS, DOD, DOE, NASA, SCEA, and industry. 

[A] In a data-rich environment, the estimating approach should precede 
the investigation of data sources; in reality, a lack of data often 
determines the approach. 

[End of table] 

Each of the 12 steps is important for ensuring that high-quality cost 
estimates are developed and delivered in time to support important 
decisions.[Footnote 17] Unfortunately, we have found that some agencies 
do not incorporate all the steps and, as a result, their estimates are 
unreliable. For example, in 2003, we completed a cross-cutting review 
at the National Aeronautics and Space Administration (NASA) that showed 
that the lack of an overall process affected NASA’s ability to create 
credible cost estimates (case study 3). 

Case Study 3: Following Cost Estimating Steps, from NASA, GAO-04-642: 
NASA’s lack of a quality estimating process resulted in unreliable cost 
estimates throughout each program’s life cycle. As of April 2003, the 
baseline development cost estimates for 27 NASA programs varied 
considerably from their initial baseline estimates. More than half 
the programs’ development cost estimates increased. For some of these 
programs, the increase was as much as 94 percent. In addition, the 
baseline development estimates for 10 programs that GAO reviewed in 
detail were rebaselined—some as many as four times. 

The Checkout and Launch Control System (CLCS) program—whose baseline 
had increased from $206 million in fiscal year 1998 to $399 million by 
fiscal year 2003—was ultimately terminated. CLCS’ cost increases 
resulted from poorly defined requirements and design and fundamental 
changes in the contractors’ approach to the work. GAO also found that 
 
* the description of the program objectives and overview in the program 
commitment agreement was not the description used to generate the cost 
estimate; 

* the total life cycle and WBS were not defined in the program’s life-
cycle cost estimate; 

* the 1997 nonadvocate review identified the analogy to be used as well 
as six different projects for parametric estimating, but no details on 
the cost model parameters were documented; and; 

* no evidence was given to explain how the schedule slip, from June 
2001 to June 2005, affected the cost estimate. 

GAO recommended that NASA establish a framework for developing life-
cycle cost estimates that would require each program to base its cost 
estimates on a WBS that encompassed both in-house and contractor 
efforts and also to prepare a description of cost analysis 
requirements. NASA concurred with the recommendation; it intended 
to revise its processes and its procedural requirements document and 
cost estimating handbook accordingly. 

Source: GAO, NASA: Lack of Disciplined Cost-Estimating Processes 
Hinders Effective Program Management, [hyperlink, 
http://www.gao.gov/products/GAO-04-642], Washington, D.C.: May 28, 
2004. 

[End of case study] 

NASA has since developed a cost estimating handbook that reflects a 
“renewed appreciation within the Agency for the importance of cost 
estimating as a critical part of project formulation and execution.” It 
has also stated that “There are newly formed or regenerated cost 
organizations at NASA Headquarters The field centers cost organizations 
have been strengthened, reversing a discouraging trend of decline.” 

Finally, NASA reported in its cost handbook that “Agency management, 
from the Administrator and Comptroller on down, is visibly supportive 
of the cost estimating function.”[Footnote 18]

While these are admirable improvements, even an estimate that meets all 
these steps may be of little use or may be overcome by events if it is 
not ready when needed. Timeliness is just as important as quality. In 
fact, the quality of a cost estimate may be hampered if the time to 
develop it is compressed. When this happens, there may not be enough 
time to collect historical data. Since data are the key drivers of an 
estimate’s quality, their lack increases the risk that the estimate may 
not be reliable. In addition, when time is a factor, an independent 
cost estimate (ICE) may not be developed, further adding to the risk 
that the estimate may be overly optimistic. This is not an issue for 
DOD’s major defense acquisition programs, because an ICE is required 
for certain milestones. 

Relying on a standard process that emphasizes pinning down the 
technical scope of the work, communicating the basis on which the 
estimate is built, identifying the quality of the data, determining the 
level of risk, and thoroughly documenting the effort should result in 
cost estimates that are defensible, consistent, and trustworthy. 
Furthermore, this process emphasizes the idea that a cost estimate 
should be a “living document,” meaning that it will be continually 
updated as actual costs begin to replace the original estimates. This 
last step links cost estimating with data that are collected by an EVM 
system, so that lessons learned can be examined for differences and 
their reasons. It also provides valuable information for strengthening 
the credibility of future cost estimates, allowing for continuous 
process improvement. 

[End of chapter 1] 

Chapter 2: Why Government Programs Need Cost Estimates And The 
Challenges In Developing Them: 

Cost estimates are necessary for government acquisition programs for 
many reasons: to support decisions about funding one program over 
another, to develop annual budget requests, to evaluate resource 
requirements at key decision points, and to develop performance 
measurement baselines. Moreover, having a realistic estimate of 
projected costs makes for effective resource allocation, and it 
increases the probability of a program’s success. Government programs, 
as identified here, include both in-house and contract efforts. 

For capital acquisitions, OMB’s Capital Programming Guide helps 
agencies use funds wisely in achieving their missions and serving the 
public. The Capital Programming Guide stresses the need for agencies to 
develop processes for making investment decisions that deliver the 
right amount of funds to the right projects. It also highlights the 
need for agencies to identify risks associated with acquiring capital 
assets that can lead to cost overruns, schedule delays, and assets that 
fail to perform as expected. 

OMB’s guide has made developing accurate life-cycle cost estimates a 
priority for agencies in properly managing their portfolios of capital 
assets that have an estimated life of 2 years or more. Examples of 
capital assets are land; structures such as office buildings, 
laboratories, dams, and power plants; equipment like motor vehicles, 
airplanes, ships, satellites, and information technology hardware; and 
intellectual property, including software. 

Developing reliable cost estimates has been difficult for agencies 
across the federal government. Too often, programs cost more than 
expected and deliver results that do not satisfy all requirements. 
According to the 2002 President’s Management Agenda: 

Everyone agrees that scarce federal resources should be allocated to 
programs and managers that deliver results. Yet in practice, this is 
seldom done because agencies rarely offer convincing accounts of the 
results their allocations will purchase. There is little reward, in 
budgets or in compensation, for running programs efficiently. And once 
money is allocated to a program, there is no requirement to revisit the 
question of whether the results obtained are solving problems the 
American people care about.[Footnote 19] 

The need for reliable cost estimates is at the heart of two of the five 
governmentwide initiatives in that agenda: improved financial 
performance and budget and performance integration. These initiatives 
are aimed at ensuring that federal financial systems produce accurate 
and timely information to support operating, budget, and policy 
decisions and that budgets are based on performance. With respect to 
these initiatives, President Bush called for changes to the budget 
process to better measure the real cost and performance of programs. 

In response to the 2002 President’s Management Agenda, OMB’s Capital 
Programming Guide requires agencies to have a disciplined capital 
programming process that sets priorities between new and existing 
assets.[Footnote 20] It also requires agencies to perform risk 
management and develop cost estimates to improve the accuracy of cost, 
schedule, and performance management. These activities should help 
mitigate difficult challenges associated with asset management and 
acquisition. In addition, the Capital Programming Guide requires an 
agency to develop a baseline assessment for each major program it plans 
to acquire. As part of this baseline, a full accounting of life-cycle 
cost estimates, including all direct and indirect costs for planning, 
procurement, operations and maintenance, and disposal is expected. 

The capital programming process, as promulgated in OMB’s Capital 
Programming Guide, outlines how agencies should use long-range planning 
and a disciplined budget process to effectively manage a portfolio of 
capital assets that achieves program goals with the least life-cycle 
costs and risks. It outlines three phases: (1) planning and budgeting, 
(2) acquisition, and (3) management in use, often referred to as 
operations and maintenance. For each phase, reliable cost estimates are 
essential and necessary to establish realistic baselines from which to 
measure future progress. 

Regarding the planning and budgeting phase, the federal budget process 
is a cyclical event. Each year in January or early February, the 
president submits budget proposals for the year that begins October 1. 
They include data for the most recently completed year, the current 
year, the budget year, and at least the 4 years following the budget 
year. The budget process has four phases: 

1. executive budget formulation, 
2. congressional budget process, 
3. budget execution and control, and, 
4. audit and evaluation. 

Budget cycles overlap—the formulation of one budget begins before 
action has been completed on the previous one. (Appendix IV gives an 
overview of the federal budget process, describing its phases and the 
major steps and time periods for each phase.) 

For the acquisition and management in use phases, reliable cost 
estimates are also important for program approval and for the continued 
receipt of annual funding. However, cost estimating is difficult. To 
develop a sound cost estimate, estimators must possess a variety of 
skills and have access to high-quality data. Moreover, credible cost 
estimates take time to develop; they cannot be rushed. Their many 
challenges increase the possibility that estimates will fall short of 
cost, schedule, and performance goals. If cost analysts recognize these 
challenges and plan for them early, this can help organizations 
mitigate these risks. 

Cost Estimating Challenges: 

Developing a good cost estimate requires stable program requirements, 
access to detailed documentation and historical data, well-trained and 
experienced cost analysts, a risk and uncertainty analysis, the 
identification of a range of confidence levels, and adequate 
contingency and management reserves.[Footnote 21] Even with the best of 
these circumstances, cost estimating is difficult. It requires both 
science and judgment. And, since answers are seldom if ever precise, 
the goal is to find a “reasonable” answer. However, the cost estimator 
typically faces many challenges. These challenges often lead to bad 
estimates—that is, estimates that contain poorly defined assumptions, 
have no supporting documentation, are accompanied by no comparisons to 
similar programs, are characterized by inadequate data collection and 
inappropriate estimating methodologies, are sustained by irrelevant or 
out-of-date data, provide no basis or rationale for the estimate, and 
can show no defined process for generating the estimate. Figure 2 
illustrates some of the challenges a cost estimator faces and some of 
the ways to mitigate them. 

Figure 2: Challenges Cost Estimators Typically Face: 

[Refer to PDF for image: illustration] 

Detailed documentation available; 
Adequate cost reserve; 
Well defined; 
Risk analysis conducted; 
Stable program; 
Adequate budget; 
Historical data available; 
Well trained and experienced analysts. 

Versus: 

Inexperienced analyst; 
Unreliable data; 
Unrealistic assumptions; 
Historical cost databases not available; 
Data not normalized; 
Unreasonable program baselines; 
Overoptimism; 
New processes; 
First-time integration; 
Cutting edge technology; 
Obtaining data; 
Program instability; 
Complex technology; 
Diminishing industrial base; 
Unrealistic projected savings. 

Source: GAO. 

[End of figure] 

Some cost estimating challenges are widespread. Deriving high-quality 
cost estimates depends on the quality of, for example, historical 
databases. It is often not possible for the cost analyst to collect the 
kinds of data needed to develop cost estimating relationships (CERs), 
analysis of development software cost, engineering build-up, and many 
other practices. In most cases, the better the data are, the better the 
resulting estimate will be. Since much of a cost analyst’s time is 
spent obtaining and normalizing data, experienced and well-trained cost 
analysts are necessary. Too often, individuals without these skills are 
thrown into performing a cost analysis to meet a pressing need (see 
case study 4). In addition, limited program resources (funds and time) 
often constrain broad participation in cost estimation processes and 
force the analyst (or cost team) to reduce the extent to which trade-
off, sensitivity, and even uncertainty analyses are performed. 

Case Study 4: Cost Analysts’ Skills, from NASA, GAO-04-642: 
GAO found that NASA’s efforts to improve its cost estimating processes 
were undermined by ineffective use of its limited number of cost-
estimating analysts. For example, headquarters officials stated that as 
projects entered the formulation phase, they typically relied on 
program control and budget specialists—not cost analysts—to provide the 
financial services to manage projects. Yet budget specialists were 
generally responsible for obligating and spending funds—not for 
conducting cost analyses that underlay the budget or ensuring that 
budgets were based on reasonable cost estimates—and, therefore, they 
tended to assume that the budget was realistic. 

Source: GAO, NASA: Lack of Disciplined Cost-Estimating Processes 
Hinders Effective Program Management, [hyperlink, 
http://www.gao.gov/products/GAO-04-642], Washington, D.C.: May 28, 
2004. 

[End of case study] 

Many cost estimating challenges can be traced to overoptimism. Cost 
analysts typically develop their estimates from technical baselines 
that program offices provide. Since program technical baselines come 
with uncertainty, recognizing this uncertainty can help form a better 
understanding of where problems will occur in the execution phase. For 
example, if a program baseline states that its total source lines of 
code will be 100,000 but the eventual total is 200,000, the cost will 
be underestimated. Or if the baseline states that the new program will 
reuse 80,000 from a legacy system but can eventually reuse only 10,000, 
the cost will be underestimated. This is illustrated in case study 5. 

Case Study 5: Recognizing Uncertainty, from Customs Service 
Modernization, GAO/AIMD-99-41: 
 
Software and systems development experts agree that early project 
estimates are imprecise by definition and that their inherent 
imprecision decreases during a project’s life cycle as more information 
becomes known. The experts emphasize that to be useful, each cost 
estimate should indicate its degree of uncertainty, possibly as an 
estimated range or qualified by some factor of confidence. The U.S. 
Customs Service did not reveal the degree of uncertainty of its cost 
estimate for the Automated Commercial Environment (ACE) program to 
managers involved in investment decisions. For example, Customs did not 
disclose that it made the estimate before fully defining ACE 
functionality. Instead, Customs presented its $1.05 billion ACE life-
cycle cost estimate as an unqualified point estimate. This suggests an 
element of precision that cannot exist for such an undefined system, 
and it obscures the investment risk remaining in the project. 
 
Source: GAO, Customs Service Modernization: Serious Management and 
Technical Weaknesses Must Be Corrected, [hyperlink, 
http://www.gao.gov/products/GAO/AMD-99-41], Washington, D.C.: Feb. 26, 
1999. 

[End of case study] 

Program proponents often postulate the availability of a new 
technology, only to discover that it is not ready when needed and 
program costs have increased. Proponents also often make assumptions 
about the complexity or difficulty of new processes, such as first-time 
integration efforts, which may end up to be unrealistic. More time and 
effort lead directly to greater costs, as case study 6 demonstrates. 

Case Study 6: Using Realistic Assumptions, from Space Acquisitions, GAO-
07-96: 
 
In five of six space system acquisition programs GAO reviewed, program 
officials and cost estimators assumed when cost estimates were 
developed that critical technologies would be mature and available. 
They made this assumption even though the programs had begun without 
complete understanding of how long they would run or how much it would 
cost to ensure that the technologies could work as intended. After the 
programs began, and as their development continued, the technology 
issues ended up being more complex than initially believed. 

For example, for the National Polar-orbiting Operational Satellite 
System (NPOESS), DOD and the U.S. Department of Commerce committed 
funds for developing and producing satellites before the technology was 
mature. Only 1 of 14 critical technologies was mature at program 
initiation, and it was found that 1 technology was less mature after 
the contractor conducted more verification testing. 

GAO found that the program was later beset by significant cost 
increases and schedule delays, partly because of technical problems 
such as the development of key sensors. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Acton to 
Address Unrealistic Initial Cost Estimates of Space Systems, 
[hyperlink, http://www.gao.gov/products/GAO-07-96], Washington, D.C.: 
Nov. 17, 2006. 

[End of case study] 

Collecting historical data and dedicating the time needed to do this 
continuously is another challenge facing cost estimators. Certain 
acquisition policy changes and pressured scheduling have had the 
unintended consequence of curtailing the generation of a great deal of 
historical data used for cost estimating. Outside of highly specific 
technology areas, it is often difficult for the cost analyst to collect 
the kinds of data needed to develop software cost estimates, valid 
CERs, and detailed engineering build-up estimates. 

In addition, limited program resources in terms of both funds and time 
often constrain broad participation in cost estimation processes and 
force the analyst or cost team to reduce the extent to which trade-off, 
sensitivity, and even uncertainty analyses are performed. Addressing 
these critical shortfalls is important and requires policy and cultural 
adjustments to fix. 

Program stability presents another serious challenge to cost analysts. 
A risk to the program also arises when the contractor knows the 
program’s budget. The contractor is pressured into presenting a cost 
estimate that fits the budget instead of a realistic estimate. Budget 
decisions drive program schedules and procurement quantities. If 
development funding is reduced, the schedule can stretch and costs can 
increase; if production funding is reduced, the number of quantities to 
be bought will typically decrease, causing unit procurement costs to 
increase. For example, projected savings from initiatives such as 
multiyear procurement—contracting for purchase of supplies or services 
for more than one program year—may disappear, as can be seen in case 
study 7. 

Case Study 7: Program Stability Issues, from Combating Nuclear 
Smuggling, GAO-06-389: 
 
According to officials of Customs and Border Protection (CBP) and the 
Pacific Northwest National Laboratory (PNNL), recurrent difficulties 
with project funding were the most important explanations of schedule 
delays. Specifically, according to Department of Homeland Security and 
PNNL officials, CBP had been chronically late in providing appropriated 
funds to PNNL, hindering its ability to meet program deployment goals. 
For example, PNNL did not receive its fiscal year 2005 funding until 
September 2005, the last month of the fiscal year. According to PNNL 
officials, because of this delay, some contracting activities in all 
deployment phases had had to be delayed or halted; the adverse effects 
on seaports were especially severe. For example, PNNL reported in 
August 2005 that site preparation work at 13 seaports had ceased 
because PNNL had not received its fiscal year 2005 funding allocation. 

Source: GAO, Combating Nuclear Smuggling: DHS Has Made Progress 
Deploying Radiation Detection Equipment at U.S. Ports-of-Entry, but 
Concerns Remain, [hyperlink, http://www.gao.gov/products/GAO-06-389], 
Washington, D.C.: Mar. 22, 2006. 

[End of case study] 

Stability issues can also arise when expected funding is cut. For 
example, if budget pressures cause breaks in production, highly 
specialized vendors may no longer be available or may have to 
restructure their prices to cover their risks. When this happens, 
unexpected schedule delays and cost increases usually result. A 
quantity change, even if it does not result in a production break, is a 
stability issue that can increase costs by affecting workload. Case 
study 8, from a GAO report on Navy shipbuilding, illustrates this 
point. 

Case Study 8: Program Stability Issues, from Defense Acquisitions, GAO-
05-183: 
 
Price increases contributed to growth in materials costs. For example, 
the price of array equipment on Virginia class submarines rose by $33 
million above the original price estimate. In addition to inflation, a 
limited supplier base for highly specialized and unique materials made 
ship materials susceptible to price increases. According to the 
shipbuilders, the low rate of ship production affected the stability of 
the supplier base. Some businesses closed or merged, leading to reduced 
competition for their services and higher prices. In some cases, the 
Navy lost its position as a preferred customer and the shipbuilder had 
to wait longer to receive materials. With a declining number of 
suppliers, more ship materials contracts went to single and sole source 
vendors. Over 75 percent of the materials for Virginia class 
submarines—reduced from 14 ships to 9 over a 10-year period—were 
produced by single source vendors. 
 
Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, [hyperlink, 
http://www.gao.gov/products/GAO-05-183], Washington, D.C.: Feb. 28, 
2005. 

[End of case study] 

Significantly accelerating (sometimes called crashing) development 
schedules also present risks. In such cases, technology tends to be 
incorporated before it is ready, tests are reduced or eliminated, or 
logistics support is not in place. As case study 9 shows, the result 
can be a reduction in costs in the short term but significantly 
increased long-term costs as problems are discovered, technology is 
back-fit, or logistics support is developed after the system is in the 
field. 

Case Study 9: Development Schedules, from Defense Acquisitions, GAO-06-
327: 

Time pressures caused the Missile Defense Agency (MDA) to stray from a 
knowledge-based acquisition strategy. Key aspects of product knowledge, 
such as technology maturity, are proven in a knowledge-based strategy 
before committing to more development. MDA followed a knowledge-based 
strategy without fielding elements such as the Airborne Laser and 
Kinetic Energy Interceptor. But it allowed the Ground-Based Midcourse 
Defense program to concurrently become mature in its technology, 
complete design activities, and produce and field assets before end-to-
end system testing—all at the expense of cost, quantity, and 
performance goals. For example, the performance of some program 
interceptors was questionable because the program was inattentive to 
quality assurance. If the block approach continued to feature 
concurrent activity as a means of acceleration, MDA’s approach might 
not be affordable for the considerable amount of capability that 
was yet to be developed and fielded. 

Source: GAO, Defense Acquisitions: Missile Defense Agency Fields 
Initial Capability but Falls Short of Original Goals, [hyperlink, 
http://www.gao.gov/products/GAO-06-327], Washington, D.C.: Mar. 15, 
2006. 

[End of case study] 

In developing cost estimates, analysts often fail to adequately address 
risk, especially risks that are outside the estimator’s control or that 
were never conceived to be possible. This can result in point estimates 
that give decision makers no information about their likelihood of 
success or give them meaningless confidence intervals. A risk analysis 
should be part of every cost estimate, but it should be performed by 
experienced analysts who understand the process and know how to use the 
appropriate tools. On numerous occasions, GAO has encountered cost 
estimates with meaningless confidence intervals because the analysts 
did not understand the underlying mathematics or tools. An example is 
given in case study 10. 

Case Study 10: Risk Analysis, from Defense Acquisitions, GAO-05-183: 
 
In developing cost estimates for eight case study ships, U.S. Navy cost 
analysts did not conduct uncertainty analyses to measure the 
probability of cost growth. Uncertainty analyses are particularly 
important, given uncertainties inherent in ship acquisition, such as 
the introduction of new technologies and the volatility of overhead 
rates. Despite the uncertainties, the Navy did not test the validity of 
the cost analysts’ assumptions in estimating construction costs for the 
eight case study ships, and it did not identify a confidence level for 
estimates. 

Specifically, it did not conduct uncertainty analyses, which generate 
values for parameters that are less than precisely known around a 
specific set of ranges. For example, if the number of hours to 
integrate a component into a ship is not precisely known, analysts may 
put in low and high values. The estimate will generate costs for these 
variables, along with other variables such as weight, experience, and 
degree of rework. The result will be a range of estimates that enables 
cost analysts to make better decisions on likely costs. Instead, the 
Navy presented its cost estimates as unqualified point estimates, 
suggesting an element of precision that cannot exist early in the 
process. Other military services qualify their cost estimates by 
determining a confidence level of 50 percent. 

Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, [hyperlink, 
http://www.gao.gov/products/GAO-05-183], Washington, D.C.: Feb. 28, 2005

[End of case study] 

A risk analysis should be used to determine a program’s contingency 
funding. All development programs should have contingency funding 
because it is simply unreasonable to expect a program not to encounter 
problems. Problems always occur, and program managers need ready access 
to funding in order to resolve them without adversely affecting 
programs (for example, stretching the schedule). Unfortunately, budget 
cuts often target contingency funding, and in some cases such funding 
is not allowed by policy. Decision makers and budget analysts should 
understand that eliminating contingency funding is counterproductive. 
(See case study 11.) 

Case Study 11: Risk Analysis, from NASA, GAO-04-642: 

Only by quantifying cost risk can management make informed decisions 
about risk mitigation strategies. Quantifying cost risk also provides a 
benchmark for measuring future progress. Without this knowledge, NASA 
may have little specific basis for determining adequate financial 
reserves, schedule margins, and technical performance margins. Managers 
may thus not have the flexibility they need to address program, 
technical, cost, and schedule risks, as NASA policy requires. 

Source: GAO, NASA: Lack of Disciplined Cost-Estimating Processes 
Hinders Effective Program Management, [hyperlink, 
http://www.gao.gov/products/GAO-04-642], Washington, D.C.: May 28, 
2004. 

[End of case study] 

Too often, organizations encourage goals that are unattainable because 
there is overoptimism that their organizations can reach them. These 
decisions follow a thought process that accentuates the positive 
without truly understanding the pitfalls being faced—in other words, 
the decision makers are avoiding risk. Recognizing and understanding 
risk is an important program management discipline, but most program 
managers believe they are dealing with risks when in fact they have 
created risk by their assumptions. History shows that program managers 
tend to be too optimistic. They believe that lessons learned from past 
programs will apply to their program and everything will work out fine. 
But a plan is by its nature meant to be optimistic, to ensure that the 
results will be successful. While program managers believe they build 
risk into their plan, they often do not put in enough. This is because 
they believe in the original estimates for the plan without allowing 
for additional changes in scope, schedule delays, or other elements of 
risk. In addition, in today’s competitive environment, contractor 
program managers may overestimate what their company can do compared to 
their competition, since they want to win. 

Since most organizations have a limited amount of money for addressing 
these issues, optimism is prevalent. To properly overcome this 
optimism, it is important to have an independent view. Through the 
program planning process, overoptimism can be tempered by challenging 
the assumptions the plan was based on. This can be done by 
independently assessing the outcomes, by using comparative data or 
experts in accomplishing the efforts planned. While this function can 
be performed by either inside or outside analysts, if the organization 
is not willing to address and understand the risks its program faces, 
it will have little hope of effectively managing and mitigating them. 
Having this “honest broker” approach to working these programs helps 
bring to light actions that can potentially limit the organization’s 
ability to succeed. Therefore, program managers and their organizations 
must understand the value and need for risk management by addressing 
risk proactively and having a plan should risks be realized. Doing so 
will enable the program management team to use this information to 
succeed in the future. 

Earned Value Management Challenges: 

OMB recommends that programs manage risk by applying EVM, among other 
ways. Reliable EVM data usually indicate monthly how well a program is 
performing in terms of cost, schedule, and technical matters. This 
information is necessary for proactive program management and risk 
mitigation. Such systems represent a best practice if implemented 
correctly, but qualified analytic staff are needed to validate and 
interpret the data. (See case study 12.) 

Case Study 12: Applying EVM, from Cooperative Threat Reduction, GAO-06-
692: 

In December 2005, a contractor’s self-evaluation stated that the EVM 
system for the chemical weapons destruction facility at Shchuch’ye, 
Russia, was fully implemented. DOD characterized the contractor’s EVM 
implementation as a “management failure,” citing a lack of experienced 
and qualified contractor staff. DOD withheld approximately $162,000 of 
the contractor’s award fee because of its concern about the EVM system. 
In March 2006, DOD officials stated that EVM was not yet a usable tool 
in managing the Shchuch’ye project. They stated that the contractor 
needed to demonstrate that it had incorporated EVM into project 
management rather than simply fulfilling contractual requirements. DOD 
expected the contractor to use EVM to estimate cost and schedule 
effects and their causes and, most importantly, to help eliminate or 
mitigate identified risks. The contractor’s EVM staff stated that they 
underestimated the effort needed to incorporate EVM data into the 
system, train staff, and develop EVM procedures. The contractor’s 
officials were also surprised by the number of man-hours required to 
accomplish these tasks, citing high staff turnover as contributing to 
the problem. According to the officials, working in a remote and 
isolated area caused many of the non-Russian employees to leave the 
program rather than extend their initial tour of duty. 

Source: GAO, Cooperative Threat Reduction: DOD Needs More Reliable Data 
to Better Estimate the Cost and Schedule of the Shchuch’ye Facility, 
[hyperlink, http://www.gao.gov/products/GAO-06-692], Washington, D.C.: 
May 31, 2006. 

[End of case study] 

Case study 12 shows that using EVM requires a cultural change. As with 
any initiative, an agency’s management must show an interest in EVM if 
its use is to be sustained. Executive personnel should understand EVM 
terms and analysis products if they expect program managers and teams 
to use them. Additionally, at the program level, EVM requires qualified 
staff to independently assess what was accomplished. EVM training 
should be provided and tracked at all levels of personnel. This does 
not always happen, and government agencies struggle with how to obtain 
qualified and experienced personnel. Perhaps the biggest challenge in 
using EVM is the trend to rebaseline programs. This happens when the 
current baseline is not adequate to complete all the work, causing a 
program to fall behind schedule or run over cost (see case study 13). A 
new baseline serves an important management control purpose when 
program goals can no longer be achieved: it gives perspective on the 
program’s current status. However, auditors should be aware that 
comparing the latest cost estimate with the most recent approved 
baseline provides an incomplete perspective on a program’s performance, 
because a rebaseline shortens the period of performance reported and 
resets the measurement of cost growth to zero. 

Case Study 13: Rebaselining, from NASA, GAO-04-642: 

Baseline development cost estimates for the programs GAO reviewed 
varied considerably from the programs’ initial baseline estimates. 
Development cost estimates of more than half the programs increased; 
for some programs, the increase was significant. The baseline 
development cost estimates for the 10 programs GAO reviewed in detail 
were rebaselined—that is, recalculated to reflect new costs, time 
periods, or resources associated with changes in program objectives, 
deliverables, or scope and plans. Although NASA provided specific 
reasons for the increased cost estimates and rebaselinings—such as 
delays in development or delivery of key system components and funding 
shortages—it did not have guidance for determining when rebaselinings 
were justified. Such criteria are important for instilling discipline 
in the cost estimating process. 

Source: GAO, NASA: Lack of Disciplined Cost-Estimating Processes 
Hinders Effective Program Management, [hyperlink, 
http://www.gao.gov/products/GAO-04-642], Washington, D.C.: May 28, 
2004. 

[End of case study] 

These challenges make it difficult for cost estimators to develop 
accurate estimates. Therefore, it is very important that agencies’ cost 
estimators have adequate guidance and training to help mitigate these 
challenges. In chapter 3, we discuss audit criteria related to cost 
estimating and EVM. We also identify some of the guidance we relied on 
to develop this guide. 

[End of Chapter 2] 

Chapter 3: Criteria For Cost Estimating, EVM, And Data Reliability: 

Government auditors use criteria as benchmarks for how well a program 
is performing. Criteria provide auditors with a context for what is 
required, what the program’s state should be, or what it was expected 
to accomplish. Criteria are the laws, regulations, policies, 
procedures, standards, measures, expert opinions, or expectations that 
define what should exist. When auditors conduct an audit, they should 
select criteria by whether they are reasonable, attainable, and 
relevant to the program’s objectives. 

Criteria include the: 

* purpose or goals that statutes or regulations have prescribed or that 
the audited entity’s officials have set, 

* policies and procedures the audited entity’s officials have 
established, 

* technically developed norms or standards, 

* expert opinions, 

* earlier performance, 

* performance in the private sector, and, 
 
* leading organizations’ best practices. 

In developing this guide, we researched legislation, regulations, 
policy, and guidance for the criteria that most pertained to cost 
estimating and EVM. Our research showed that while DOD has by far the 
most guidance on cost estimating and EVM in relation to civil agencies, 
other agencies are starting to develop policies and guidance. 
Therefore, we intend this guide as a starting point for auditors to 
identify criteria. 

For each new engagement, however, GAO auditors should exercise 
diligence to see what, if any, new legislation, regulation, policy, and 
guidance exists. Auditors also need to decide whether criteria are 
valid. Circumstances may have changed since they were established and 
may no longer conform to sound management principles or reflect current 
conditions. In such cases, GAO needs to select or develop criteria that 
are appropriate for the engagement’s objectives. Table 3 lists criteria 
related to cost estimating and EVM. Each criterion is described in more 
detail in appendix V. 

Table 3: Cost Estimating and EVM Criteria for Federal Agencies: 
Legislation, Regulations, Policies, and Guidance: 

Type: Legislation or regulation: 

Date: 1968; 
Title: SAR: Selected Acquisition Reports, 10 U.S.C. § 2432 (2006); 
Applicable agency: DOD; 
Notes: Became permanent law in 1982; applies only to DOD’s major 
defense acquisition programs. 

Date: 1982; 
Title: Unit Cost Reports (“Nunn-McCurdy”); 10 U.S.C. § 2433 (2006); 
Applicable agency: DOD; 
Notes: Applies only to DOD’s major defense acquisition programs. 

Date: 1983; 
Title: Independent Cost Estimates; Operational Manpower Requirements, 
10 U.S.C. § 2434 (2006); 
Applicable agency: DOD; 
Notes: Applies only to DOD’s major defense acquisition programs. 

Date: 1993; 
Title: GPRA: Government Performance and Results Act, Pub. L. No. 103-62 
(1993); 
Applicable agency: All; 
Notes: Requires agencies to prepare (1) multiyear strategic plans 
describing mission goals and methods for reaching them and (2) annual 
program performance reports to review progress toward annual 
performance goals. 

Date: 1994; 
Title: The Federal Acquisition Streamlining Act of 1994, § 5051(a), 41 
U.S.C. § 263 (2000); 
Applicable agency: All civilian agencies; 
Notes: Established congressional policy that agencies should achieve, 
on average, 90 percent of cost, performance, and schedule goals 
established for their major acquisition programs; requires an agency to 
approve or define cost, performance, and schedule goals and to 
determine whether there is a continuing need for programs that are 
significantly behind schedule, over budget, or not in compliance with 
performance or capability requirements and to identify suitable 
actions to be taken. 

Date: 1996; 
Title: CCA: Clinger-Cohen Act of 1996, 40 U.S.C. §§ 11101–11704 (Supp. 
V 2005); 
Applicable agency: All; 
Notes: Requires agencies to base decisions about information technology 
investments on quantitative and qualitative factors associated with 
their costs, benefits, and risks and to use performance data to 
demonstrate how well expenditures support program improvements. 

Date: 2006; 
Title: Major Automated Information System Programs, 10 U.S.C. §§ 2445a 
– 2445d (2006); 
Applicable agency: DOD; 
Notes: Oversight requirements for DOD’s major automated information 
system (MAIS) programs, including estimates of development costs and 
full life-cycle costs as well as program baseline and variance 
reporting requirements. 

Date: 2006; 
Title: Federal Acquisition Regulation (FAR), Major Systems Acquisition, 
48 C.F.R. part 34, subpart 34.2, Earned Value Management System; 
Applicable agency: All; 
Notes: Earned Value Management System policy was added by Federal 
Acquisition Circular 2005-11, July 5, 2006, Item I—Earned Value 
Management System (EVMS) (FAR Case 2004-019). 

Date: 2008; 
Title: Defense Federal Acquisition Regulation Supplement; Earned Value 
Management Systems (DFARS Case 2005–D006), 73 Fed. Reg. 21,846 (April 
23, 2008), primarily codified at 48 C.F.R. subpart 234.2, and part 252 
(sections 252.234-7001 and 7002); 
Applicable agency: DOD; 
Notes: DOD’s final rule (1) amending the Defense Federal Acquisition 
Regulation Supplement (DFARS) to update requirements for DOD 
contractors to establish and maintain EVM systems and (2) eliminating 
requirements for DOD contractors to submit cost/schedule status 
reports. 

Policy: 

Date: 1976; 
Title: OMB, Major Systems Acquisitions, Circular A-109 (Washington, 
D.C.: Apr. 5, 1976); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 1992; 
Title: OMB, Guidelines and Discount Rates for Benefit-Cost Analysis of 
Federal Programs, Circular No. A-94 Revised (Washington, D.C.: Oct. 29, 
1992); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 1995; 
Title: DOD, Economic Analysis for Decisionmaking, Instruction No. 
7041.3 (Washington, D.C.: USD, Nov. 7, 1995); 
Applicable agency: DOD; 
Notes: [Empty]. 

Date: 2003; 
Title: DOD, The Defense Acquisition System, Directive No. 5000.1 
(Washington, D.C.: USD, May 12, 2003). Redesignated 5000.01 and 
certified current as of Nov. 20, 2007. 
Applicable agency: DOD; 
Notes: States that every program manager must establish program goals 
for the minimum number of cost, schedule, and performance parameters 
that describe the program over its life cycle and identify any 
deviations. 

Date: 2003; 
Title: DOD, Operation of the Defense Acquisition System, Instruction 
No. 5000.2 (Washington, D.C.: USD, May 12, 2003). Cancelled and 
reissued by Instruction No. 5000.02 on Dec. 8, 2008. 
Applicable agency: DOD; 
Notes: Describes the standard framework for defense acquisition 
systems: defining the concept, analyzing alternatives, developing 
technology, developing the system and demonstrating that it works, 
producing and deploying the system, and operating and supporting it 
throughout its useful life. 

Date: 2004; 
Title: National Security Space Acquisition Policy, Number 03-01, 
Guidance for DOD Space System Acquisition Process (Washington, D.C.: 
revised Dec. 27, 2004); 
Applicable agency: DOD; 
Notes: [Empty]. 

Date: 2005; 
Title: DOD, “Revision to DOD Earned Value Management Policy,” 
memorandum, Under Secretary of Defense, Acquisition, Technology, and 
Logistics (Washington, D.C.: Mar. 7, 2005); 
Applicable agency: DOD; 
Notes: [Empty]. 

Date: 2005 
Title: OMB, “Improving Information Technology (IT) Project Planning and 
Execution,” memorandum for Chief Information Officers No. M-05-23 
(Washington, D.C.: Aug. 4, 2005); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 2006 
Title: DOD, The Program Manager’s Guide to DOD OMB, Capital Programming 
Guide, Supplement to Circular A-11, Part 7, Preparation, Submission, 
and Execution of the Budget (Washington, D.C.: Executive Office of the 
President, June 2006) 
Applicable agency: All; 
Notes: [Empty]. 

Date: 2006 
Title: DOD, Cost Analysis Improvement Group (CAIG), Directive No. 
5000.04 (Washington, D.C.: Aug. 16, 2006) 
Applicable agency: DOD; 
Notes: [Empty]. 

Guidance: 

Date: 1992; 
Title: DOD, The Program Manager’s Guide to 
CAIG, Operating and Support Cost-Estimating Guide (Washington, D.C.: 
DOD, Office of the Secretary, May 1992); 
Applicable agency: DOD; 
Notes: [Empty]. 

Date: 1992; 
Title: DOD, Cost Analysis Guidance and Procedures, DOD Directive 5000.4-
M (Washington, D.C.: OSD, Dec. 11, 1992); 
Applicable agency: DOD; 
Notes: [Empty]. 

Date: 2003; 
Title: DOD, The Program Manager’s Guide to the Integrated Baseline 
Review Process (Washington, D.C.: OSD, April 2003); 
Applicable agency: DOD; 
Notes: [Empty]. 

Date: 2004; 
Title: NDIA, National Defense Industrial Association (NDIA) Program 
Management Systems Committee (PMSC) Surveillance Guide (Arlington, Va.: 
October 2004); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 2005 
Title: NDIA, National Defense Industrial All Association (NDIA) Program 
Management Systems Committee (PMSC) Earned Value Management Systems 
Intent Guide (Arlington, Va.: January 2005); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 2006; 
Title: Defense Contract Management Agency, Department of Defense Earned 
Value Management Implementation Guide (Alexandria, Va.: October 2006); 
Applicable agency: DOD, FAA, NASA; 
Notes: [Empty]. 

Date: 2006; 
Title: National Defense Industrial Association, Program Management 
Systems Committee, “NDIA PMSC ANSI/EIA 748 Earned Value Management 
System Acceptance Guide,” draft, working release for user comment 
(Arlington, Va.: November 2006); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 2007 
Title: American National Standards Institute, Information Technology 
Association of America, Earned Value Management Systems (ANSI/EIA 748-
B) (Arlington, Va.: July 9, 2007); 
Applicable agency: All; 
Notes: [Empty]. 

Date: 2007 
Title: National Defense Industrial Association, Program Management 
Systems Committee, “NDIA PMSC Earned Value Management Systems 
Application Guide,” draft, working release for user comment (Arlington, 
Va.: March 2007); 
Applicable agency: All; 
Notes: [Empty]. 

Source: GAO, DOD, and OMB. 

[End of table] 

Determining Data Reliability: 
 
Auditors need to collect data produced from both a program’s cost 
estimate and its EVM system. They can collect these data by 
questionnaires, structured interviews, direct observations, or 
computations, among other methods. (Appendix VI is a sample data 
collection instrument; appendix VII gives reasons why auditors need the 
information.) After auditors have collected their data, they must judge 
the data for integrity as well as for quality in terms of validity, 
reliability, and consistency with fact. 

For cost estimates, auditors must confirm that, at minimum, internal 
quality control checks show that the data are reliable and valid. To do 
this, they must have source data and must estimate the rationale for 
each cost element, to verify that: 

* the parameters (or input data) used to create the estimate are valid 
and applicable,[Footnote 22] 

* labor costs include a time-phased breakdown of labor hours and rates, 

* the calculations for each cost element are correct and the results 
make sense, 

* the program cost estimate is an accurate total of subelement costs, 
and; 

* escalation was properly applied to account for differences in the 
price of goods and services over time. 

Auditors should clarify with cost estimators issues about data and 
methodology. For example, they might ask what adjustments were made to 
account for differences between the new and existing systems with 
respect to design, manufacturing processes, and types of materials. In 
addition, auditors should look for multiple sources of data that 
converge toward the same number, in order to gain confidence in the 
data used to create the estimate. 

It is particularly important that auditors understand problems 
associated with the historical data—such as program redesign, schedule 
slips, and budget cuts—and whether the cost estimators “cleansed the 
data” to remove their effects. According to experts in the cost 
community, program inefficiencies should not be removed from historical 
data, since the development of most complex systems usually encounters 
problems. The experts stress that removing data associated with past 
problems is naïve and introduces unnecessary risk. (This topic is 
discussed in chapter 10.) 

With regard to EVM, auditors should request a copy of the system 
compliance or validation letter that shows the contractor’s ability to 
satisfy the 32 EVM guidelines (discussed in chapter 18).[Footnote 23] 
These guidelines are test points to determine the quality of a 
contractor’s EVM system. Contract performance reports (CPR) formally 
submitted to the agency should be examined for reasonableness, 
accuracy, and consistency with other program status reports as a 
continuous measure of the EVM system quality and robustness. Auditors 
should also request a copy of the integrated baseline review (IBR) 
results (also discussed in chapter 18) to see what risks were 
identified and whether they were mitigated. Auditors should request 
copies of internal management documents or reports that use EVM data to 
ensure that EVM is being used for management, not just for external 
reporting. Finally, to ensure that EVM data are valid and accurate, 
auditors should look for evidence that EVM analysis and surveillance 
are performed regularly by staff trained in this specialty. 

[End of Chapter 3] 

Chapter 4: 

Cost Analysis Overview: 

Although “cost estimating” and “cost analysis” are often used 
interchangeably, cost estimating is a specific activity within cost 
analysis. Cost analysis is a powerful tool, because it requires a 
rigorous and systematic analysis that results in a better understanding 
of the program being acquired. This understanding, in turn, leads to 
improved program management in applying resources and mitigating 
program risks. 

Differentiating Cost Analysis And Cost Estimating: 
 
Cost analysis, used to develop cost estimates for such things as 
hardware systems, automated information systems, civil projects, 
manpower, and training, can be defined as: 
 
* the effort to develop, analyze, and document cost estimates with 
analytical approaches and techniques;

* the process of analyzing, interpreting, and estimating the 
incremental and total resources required to support past, present, and 
future systems—an integral step in selecting alternatives; and; 

* a tool for evaluating resource requirements at key milestones and 
decision points in the acquisition process. 

Cost estimating involves collecting and analyzing historical data and 
applying quantitative models, techniques, tools, and databases to 
predict a program’s future cost. More simply, cost estimating combines 
science and art to predict the future cost of something based on known 
historical data that are adjusted to reflect new materials, technology, 
software languages, and development teams. 

Because cost estimating is complex, sophisticated cost analysts should 
combine concepts from such disciplines as accounting, budgeting, 
computer science, economics, engineering, mathematics, and statistics 
and should even employ concepts from marketing and public affairs. And 
because cost estimating requires such a wide range of disciplines, it 
is important that the cost analyst either be familiar with these 
disciplines or have access to an expert in these fields. 

Main Cost Estimate Categories: 

Auditors are likely to encounter two main cost estimate categories: 

* a life-cycle cost estimate (LCCE) that may include independent cost 
estimates, independent cost assessments, or total ownership costs, and, 
 
* a business case analysis (BCA) that may include an analysis of 
alternatives or economic analyses. 

Auditors may also review other types of cost estimates, such as 
independent cost assessments (ICA), nonadvocate reviews (NAR), and 
independent government cost estimates (IGCE). These types of estimates 
are commonly developed by civilian agencies. 

Life-Cycle Cost Estimate: 
 
A life-cycle cost estimate provides an exhaustive and structured 
accounting of all resources and associated cost elements required to 
develop, produce, deploy, and sustain a particular program. Life cycle 
can be thought of as a “cradle to grave” approach to managing a program 
throughout its useful life. This entails identifying all cost elements 
that pertain to the program from initial concept all the way through 
operations, support, and disposal. An LCCE encompasses all past (or 
sunk), present, and future costs for every aspect of the program, 
regardless of funding source. 

Life-cycle costing enhances decision making, especially in early 
planning and concept formulation of acquisition. Design trade-off 
studies conducted in this period can be evaluated on a total cost 
basis, as well as on a performance and technical basis. A life-cycle 
cost estimate can support budgetary decisions, key decision points, 
milestone reviews, and investment decisions. 

The LCCE usually becomes the program’s budget baseline. Using the LCCE 
to determine the budget helps to ensure that all costs are fully 
accounted for so that resources are adequate to support the program. 
DOD identifies four phases that an LCCE must address: research and 
development, procurement and investment, operations and support, and 
disposal. Civilian agencies may refer to the first two as development, 
modernization, and enhancement and may include in them acquisition 
planning and funding. Similarly, civilian agencies may refer to 
operations and support as “steady state” and include them in operations 
and maintenance activities. Although these terms mean essentially the 
same thing, they can differ from agency to agency. DOD’s four phases 
are described below. 

1. Research and development include development and design costs for 
system engineering and design, test and evaluation, and other costs for 
system design features. They include costs for development, design, 
startup, initial vehicles, software, test and evaluation, special 
tooling and test equipment, and facility changes. 

2. Procurement and investment include total production and deployment 
costs (e.g., site activation, training) of the prime system and its 
related support equipment and facilities. Also included are any related 
equipment and material furnished by the government, initial spare and 
repair parts, interim contractor support, and other efforts. 

3. Operations and support are all direct and indirect costs incurred in 
using the prime system—manpower, fuel, maintenance, and support—through 
the entire life cycle. Also included are sustaining engineering and 
other collateral activities. 

4. Disposal, or inactivation, includes the costs of disposing of the 
prime equipment after its useful life. 

Because they encompass all possible costs, LCCEs provide a wealth of 
information about how much programs are expected to cost over time. 
This information can be displayed visually to show what funding is 
needed at a particular time and when the program is expected to move 
from one phase to another. For example, figure 3 is a life-cycle cost 
profile for a hypothetical space system. 

Figure 3: Life-Cycle Cost Estimate for a Space System: 

Refer to PDF for image: line graph] 

Space system life cycle: 
Phase B ATP; 
Final design; 
O&M Support start; 
Launch 1; 
Launch 2; 
IOC; 
Launch 3; 
Launch N; 
FOC. 

RDT&E: Includes development and production of first two vehicles; 
Follow-on buys occur after final design verification; 
Procurement: Includes production of follow-on buys (typically lots of 2 
or 3 SVs); 
O&M staff in place before launch 1; 
O&M: Operators and controllers through system EOL. 

Source: DOD. 

Note: O&M = operations and maintenance; 
RDT&E = research, development, test, and evaluation; 
SV = space vehicle; 
EOL = end of life; 
IOC = initial operational capacity; 
FOC = full operational capacity. 

[End of figure] 

Figure 3 illustrates how space systems must invest heavily in research 
and development because once a system is launched into space, it cannot 
be retrieved for maintenance. Other systems such as aircraft, ships, 
and information technology systems typically incur hefty operations 
costs in relation to development and production costs. Such mission 
operations costs are very large because the systems can be retrieved 
and maintained and therefore require sophisticated logistics support 
and recurring broad-based training for large user populations. Thus, 
having full life-cycle costs is important for successfully planning 
program resource requirements and making wise decisions. 

Business Case Analysis: 
 
A business case analysis, sometimes referred to as a cost benefit 
analysis, is a comparative analysis that presents facts and supporting 
details among competing alternatives. A BCA considers not only all the 
life-cycle costs that an LCCE identifies but also quantifiable and 
nonquantifiable benefits. It should be unbiased by considering all 
possible alternatives and should not be developed solely for supporting 
a predetermined solution. Moreover, a BCA should be rigorous enough 
that independent auditors can review it and clearly understand why a 
particular alternative was chosen. 

A BCA seeks to find the best value solution by linking each alternative 
to how it satisfies a strategic objective. Each alternative should 
identify the: 
 
* relative life-cycle costs and benefits; 

* methods and rationale for quantifying the life-cycle costs and 
benefits; 

* effect and value of cost, schedule, and performance tradeoffs; 

* sensitivity to changes in assumptions; and; 

* risk factors. 

On the basis of this information, the BCA then recommends the best 
alternative. In addition to supporting an investment decision, the BCA 
should be considered a living document and should be updated often to 
reflect changes in scope, schedule, or budget. In this way, the BCA is 
a valuable tool for validating decisions to sustain or enhance the 
program. 

Auditors may encounter other estimates that fall into one of the two 
main categories of cost estimates. For example, an auditor may examine 
an independent cost estimate, independent cost assessment, independent 
government cost estimates, total ownership cost, or rough order of 
magnitude estimate—all variations of a life-cycle cost estimate. 
Similarly, instead of reviewing a business case analysis, an auditor 
may review an analysis of alternatives (AOA), a cost-effectiveness 
analysis (CEA), or an economic analysis (EA). Each of these analyses is 
a variation, in one form or another, of a BCA. Table 4 looks more 
closely at the different types of cost estimates that can be developed. 

Table 4: Life-Cycle Cost Estimates, Types of Business Case Analyses, 
and Other Types of Cost Estimates: 
 
Life-cycle cost estimate: 

Estimate type: Independent cost estimate; 
Level of effort: Usually requires a large team, may take many months to 
accomplish, 
and addresses the full LCCE; 
Description: An ICE, conducted by an organization independent of 
the acquisition chain of command, is based on the same 
detailed technical and procurement information used 
to make the baseline estimate—usually the program or 
project LCCE. ICEs are developed to support new programs 
or conversion, activation, modernization, or service life 
extensions and to support DOD milestone decisions for 
major defense acquisition programs.[A] 
 
An estimate might cover a program’s entire life cycle, 
one program phase, or one high-value, highly visible, or 
high-interest item within a phase. ICEs are used primarily 
to validate program or project LCCEs and are typically 
reconciled with them. 

Because the team performing the ICE is independent, it 
provides an unbiased test of whether the program office 
cost estimate is reasonable. It is also used to identify risks 
related to budget shortfalls or excesses 

Estimate type: Total ownership cost estimate; 
Level of effort: Requires a large team, may take many months to 
accomplish, and addresses the full LCCE; 
Description: Related to LCCE but broader in scope, a total ownership 
cost estimate consists of the elements of life-cycle cost plus some 
infrastructure and business process costs not necessarily 
attributable to a program. 

Infrastructure includes acquisition and central logistics 
activities; nonunit central training; personnel administration 
and benefits; medical care; and installation, communications, 
and information infrastructure to support military bases. It is 
normally found in DOD programs. 

Business case analysis: 

Estimate type: Analysis of alternatives and cost effectiveness 
analysis; 
Level of effort: Requires a large team, may take many months to 
accomplish, and addresses the full LCCE; 
Description: AOA compares the operational effectiveness, suitability, 
and LCCE of alternatives that appear to satisfy established capability 
needs. Its major components are a CEA and cost analysis. 

AOAs try to identify the most promising of several conceptual 
alternatives; analysis and conclusions are typically used to 
justify initiating an acquisition program. An AOA also looks at 
mission threat and dependencies on other programs. 

When an AOA cannot quantify benefits, a CEA is more 
appropriate. A CEA is conducted whenever it is unnecessary 
or impractical to consider the dollar value of benefits, as when 
various alternatives have the same annual monetary benefits. 
Both the AOA and CEA should address each alternative’s 
advantages, disadvantages, associated risks, and uncertainties 
and how they might influence the comparison. 
 
Estimate type: Economic analysis and cost benefit analysis; 
Level of effort: Requires a large team, may take many months to 
accomplish, and addresses the full LCCE; 
Description: EA is a conceptual framework for systematically 
investigating problems of choice. Posing various alternatives for 
reaching an objective, it analyzes the LCCE and benefits of each one, 
usually with a return on investment analysis. 

Present value is also an important concept: Since an LCCE 
does not consider the time value of money, it is necessary to 
determine when expenditures for alternatives will be made. 

EA expands cost analysis by examining the effects of the time 
value of money on investment decisions. After cost estimates 
have been generated, they must be time-phased to allow for 
alternative expenditure patterns. Assuming equal benefits, 
the alternative with the least present value cost is the most 
desirable: it implies a more efficient allocation of resources. 

Other: 

Estimate type: Rough order of magnitude; 
Level of effort: May be done by a small group or one person; can be 
done in hours, days, or weeks; and may cover only a portion of the 
LCCE; 
Description: Developed when a quick estimate is needed and few details 
are available. Usually based on historical ratio information, it is 
typically developed to support what-if analyses and can be developed 
for a particular phase or portion of an estimate to the entire cost 
estimate, depending on available data. It is helpful for examining 
differences in high[level alternatives to see which are the most 
feasible. Because it is developed from limited data and in a short 
time, a rough order of magnitude analysis should never be considered a 
budget-quality cost estimate. 

Estimate type: Independent cost assessment; 
Level of effort: Requires a small group; may take months to accomplish, 
depending on how much of the LCCE is being reviewed; 
Description: An ICA is an outside, nonadvocate’s evaluation of a cost 
estimate’s quality and accuracy, looking specifically at a program’s 
technical approach, risk, and acquisition strategy to ensure that the 
program’s cost estimate captures all requirements. 

Typically requested by a program manager or outside source, it may be 
used to determine whether the cost estimate reflects the program of 
record. It is not as formal as an ICE and does not have to be performed 
by an organization independent of the acquisition chain of command, 
although it usually is. 

An ICA usually does not address a program’s entire life cycle. 
 
Estimate type: Independent government cost estimate; 
Level of effort: Requires a small group, may take months to accomplish, 
and covers only the LCCE phase under contract; 
Description: An IGCE is conducted to check the reasonableness of a 
contractor’s cost proposal and to make sure that the offered prices are 
within the budget range for a particular program. 

The program manager submits it as part of a request for contract 
funding. It documents the government’s assessment of the program’s most 
probable cost and ensures that enough funds are available to execute 
it. It is also helpful in assessing the feasibility of individual tasks 
to determine if the associated costs are reasonable. 
 
Estimate type: Estimate at completion; 
Level of effort: Requires nominal effort once all EVM data are on hand 
and have been determined reliable; covers only the LCCE phase under 
contract; 
Description: An EAC is an independent assessment of the cost to 
complete authorized work based on a contractor’s historical EVM 
performance. 

It uses various EVM metrics to forecast the expected final cost: 
EAC = actual costs incurred + (budgeted cost for work remaining / EVM 
performance factor). 

The performance factor can be based on many different EVM metrics that 
capture cost and schedule status to date. 
 
Source: GAO, DOD, NIH, OMB, and SCEA. 

[A] For more detail, see app. V, ICEs, 10 U.S.C. § 2434. 

[End of table] 

The Overall Significance Of Cost Estimates: 

Not an end in itself, cost estimating is part of a total systems 
analysis. It is a critical element in any acquisition process and helps 
decision makers evaluate resource requirements at milestones and other 
important decision points. 

Cost estimates: 
 
* establish and defend budgets and, 

* drive affordability analysis. 

Cost estimates are integral to determining and communicating a 
realistic view of likely cost and schedule outcomes that can be used to 
plan the work necessary to develop, produce, install, and support a 
program. 

Cost estimating also provides valuable information to help determine 
whether a program is feasible, how it should be designed, and the 
resources needed to support it. Further, cost estimating is necessary 
for making program, technical, and schedule analyses and to support 
other processes such as: 
 
* selecting sources; 

* assessing technology changes, analyzing alternatives, and performing 
design trade-offs; and; 

* satisfying statutory and oversight requirements. 

Cost Estimates In Acquisition: 
 
An acquisition program focuses on the cost of developing and procuring 
an end item and whether enough resources and funding are available. The 
end product of the acquisition process is a program capability that 
meets its users’ needs at a reasonable price. During the acquisition 
process, decisions must be made on how best to consume labor, capital, 
equipment, and other finite resources. A realistic cost estimate allows 
better decision making, in that an adequate budget can accomplish the 
tasks that ultimately increase a program’s probability of success. 

Acquisition is an event-driven process, in that programs must typically 
pass through various milestones or investment reviews in which they are 
held accountable for their accomplishments. Cost estimates play an 
important role in these milestone or investment decisions. For example, 
in government programs, a cost estimate should be validated if a major 
program is to continue through its many acquisition reviews and other 
key decision points. 

Validation involves testing an estimate to see if it is reasonable and 
includes all necessary costs. Testing can be as simple as comparing 
results with historical data from similar programs or using another 
estimating method to see if results are similar. Industry requires 
similar scrutiny throughout development, in what is commonly referred 
to as passing through specific gates. 

Once a cost estimate has been accepted and approved, it should be 
updated periodically as the program matures and as schedules and 
requirements change. Updated estimates help give management control 
over a project’s resources when new requirements are called for under 
tight budget conditions. This is especially important early in a 
project, when less is known about requirements and the opportunity for 
change (and cost growth) is greater. As more knowledge is gained, 
programs can retire some risk and reduce the potential for unexpected 
cost and schedule growth. 

Cost estimates tend to become more certain as actual costs begin to 
replace earlier estimates. This happens when risks are either mitigated 
or realized. If risks actually occur, the resulting cost growth becomes 
absorbed by the cost estimate. 

For this reason, it is important to continually update estimates with 
actual costs, so that management has the best information available for 
making informed decisions. In addition, narrow risk ranges should be 
viewed as suspect, because more cost estimates tend to overrun than 
underrun. These processes are illustrated in what is commonly called 
the “cone of uncertainty,” which are depicted in figure 4. 

Figure 4: Cone of Uncertainty: 

[Refer to PDF for image: illustration] 

Cost estimate baseline; 

Concept refinement gate: 
Technology development gate: 
Start of program and start of system integration gate: 

Uncertainty about cost estimate is high; 
Estimate becomes more certain as program progesses; 
Estimate tends to grow over time as risks are realized; 
Uncertainty is low. 

Source: GAO. 

[End of figure] 

It is important to have a track record of the estimate so one can 
measure growth from what the estimate should have been. Therefore, 
tying growth and risk together is critical because the risk 
distribution identifies the range of anticipated growth. 

The Importance Of Cost Estimates In Establishing Budgets: 

A program’s approved cost estimate is often used to create the budget 
spending plan. This plan outlines how and at what rate the program 
funding will be spent over time. Since resources are not infinite, 
budgeting requires a delicate balancing act to ensure that the rate of 
spending closely mirrors available resources and funding. And because 
cost estimates are based on assumptions that certain tasks will happen 
at specific times, it is imperative that funding be available when 
needed so as to not disrupt the program schedule. 

Because a reasonable and supportable budget is essential to a program’s 
efficient and timely execution, a competent estimate is the key 
foundation of a good budget. For a government agency, accurate 
estimates help in assessing the reasonableness of a contractor’s 
proposals and program budgets. Credible cost estimates also help 
program offices justify budgets to the Congress, OMB, department 
secretaries, and others. Moreover, cost estimates are often used to 
help determine how budget cuts may hinder a program’s progress or 
effectiveness. 

Outside the government, contractors need accurate estimates of the 
costs required to complete a task in order to ensure maximum 
productivity and profitability. Estimates that are too low can reduce 
profits if the contract is firm fixed price, and estimates that are too 
high will diminish a contractor’s ability to compete in the 
marketplace. 

While contractors occasionally propose unrealistically low cost 
estimates for strategic purposes—for example, “buying-in”—such outcomes 
can be attributed to poor cost estimating. This sometimes happens when 
contractors are highly optimistic in estimating potential risks. As a 
program whose budget is based on such estimates is developed, it 
becomes apparent sooner or later that either the developer or the 
customer must pay for a cost overrun, as case study 14 indicates. 

Case Study 14: Realistic Estimates, from Defense Acquisitions, GAO-05-
183: 
 
In negotiating the contract for the first four Virginia class ships, 
program officials stated that they were constrained in negotiating the 
target price to the amount funded for the program, risking cost growth 
at the outset. The shipbuilders said that they accepted a challenge to 
design and construct the ships for $748 million less than their 
estimated costs, because the contract protected their financial risk. 
Despite the significant risk of cost growth, the Navy did not identify 
any funding for probable cost growth, given available guidance at the 
time. The fiscal year 2005 President’s Budget showed that budgets for 
the two Virginia class case study ships had increased by $734 million. 
However, on the basis of July 2004 data, GAO projected that additional 
cost growth on contracts for the two ships would be likely to reach 
$840 million, perhaps higher. In the fiscal year 2006 budget, the Navy 
requested funds to cover cost increases expected to reach to 
approximately $1 billion. 
 
Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, [hyperlink, 
http://www.gao.gov/products/GAO-05-183], Washington, D.C.: Feb. 28, 
2005. 

[End of case study] 

Cost Estimates And Affordability: 
 
Affordability is the degree to which an acquisition program’s funding 
requirements fit within the agency’s overall portfolio plan. Whether a 
program is affordable depends a great deal on the quality of its cost 
estimate. Therefore, agencies can follow the 12-step estimating process 
we outlined in chapter 1 to ensure that they are creating and making 
decisions based on credible cost estimates. The 12-step process 
addresses best practices, including defining the program’s purpose, 
developing the estimating plan, defining the program’s characteristics, 
determining the estimating approach, identifying ground rules and 
assumptions, obtaining data, developing the point estimate, conducting 
sensitivity analysis, performing a risk or uncertainty analysis, 
documenting the estimate, presenting it to management for approval, and 
updating it to reflect actual costs and changes. Following these steps 
ensures that realistic cost estimates are developed and presented to 
management, enabling them to make informed decisions about whether 
the program is affordable within the portfolio plan. 

Decision makers should consider affordability at each decision point in 
a program’s life cycle. It is important to know the program’s cost at 
particular intervals, in order to ensure that adequate funding is 
available to execute the program according to plan. Affordability 
analysis validates that the program’s acquisition strategy has an 
adequate budget for its planned resources (see figure 5). 

Figure 5: An Affordability Assessment: 

[Refer to PDF for image: combined line graph] 

Source: DOD. 

[End of figure] 

In figure 5, seven programs A–G are plotted against time, with the 
resources they will need to support their goals. The benefit of 
plotting the programs together gives decision makers a high-level 
analysis of their portfolio and the resources they will need in the 
future. In this example, it appears that funding needs are relatively 
stable in fiscal years 1–12, but from fiscal year 12 to fiscal year 16, 
an increasing need for additional funding is readily apparent. This is 
commonly referred to as a bow-wave, meaning there is an impending spike 
in the requirement for additional funds. Whether these funds will be 
available will determine which programs remain within the portfolio. 
Because the programs must compete against one another for limited 
funds, it is considered a best practice to perform the affordability 
assessment at the agency level, not program by program. 

While approaches may vary, an affordability assessment should address 
requirements at least through the programming period and, preferably, 
several years beyond. Thus, LCCEs give decision makers important 
information in that not all programs require the same type of funding 
profile. In fact, different commodities require various outlays of 
funding and are affected by different cost drivers. Figure 6 
illustrates this point with typical funding curves by program phase. It 
shows that while some programs may cost less to develop—for example, 
research and development in construction programs differ from fixed-
wing aircraft—they may require more or less funding for investment, 
operations, and support in the out-years. 

Figure 6: Typical Capital Asset Acquisition Funding Profiles by Phase: 

[Refer to PDF for image: 5 vertical bar graphs] 

The bar graphs depict the percent of project cost for R&D, Investment, 
and O&S/Disposal for the following program types: 
Construction; 
Space; 
Ships; 
Surface vehicles; 
Fixed-wing aircraft. 

Source: GAO and DOD. 

[End of figure] 

Line graphs or sand charts like those in figure 5, therefore, are often 
used to show how a program fits within the organizational plan, both 
overall and by individual program components. Such charts allow 
decision makers to determine how and if the program fits within the 
overall budget. It is very important for LCCEs to be both realistic and 
timely, available to decision makers as early as possible. Case studies 
15 and 16 show how this often does not happen. 

Case Study 15: Importance of Realistic LCCEs, from Combating Nuclear 
Smuggling, GAO-07-133R: 
 
The Department of Homeland Security’s (DHS) Domestic Nuclear Detection 
Office (DNDO) had underestimated life-cycle costs for plastic 
scintillators and advanced spectroscopic portal monitors. Although 
DNDO’s analysis assumed a 5-year life cycle for both, DNDO officials 
told GAO that a 10-year life cycle was more reasonable. DNDO’s analysis 
had assumed annual maintenance costs at 10 percent of their procurement 
costs: maintenance costs for the scintillators would be about $5,500 
per year per unit, based on a $55,000 purchase price, and maintenance 
costs for the monitors would be about $38,000 per year per unit, based 
on a $377,000 purchase price. DNDO’s analysis had not accounted for 
about $181 million in potential maintenance costs for the monitors 
alone. With the much higher maintenance costs, and doubling the life 
cycle, the long-term implications would be magnified. 

Source: GAO, Combating Nuclear Smuggling: DHS’s Cost-Benefit Analysis 
to Support the Purchase of New Radiation Detection Portal Monitors Was 
Not Based on Available Performance Data and Did Not Fully Evaluate All 
the Monitors’ Costs and Benefits, GAO-07-133R (Washington, D.C.: Oct. 
17, 2006). 

[End of case study] 

Case Study 16: Importance of Realistic LCCEs, from Space Acquisitions, 
GAO-07-96: 

GAO has in the past identified a number of causes behind cost growth 
and related problems in DOD’s major space acquisition programs, but 
several consistently stand out. On a broad scale, DOD starts more 
weapons programs than it can afford, creating competition for funding 
that encourages low-cost estimating and optimistic scheduling, 
overpromising, suppressing bad news, and for space programs, forsaking 
the opportunity to identify and assess potentially better alternatives. 
Programs focus on advocacy at the expense of realism and sound 
management. 

With too many programs in its portfolio, DOD is invariably forced to 
shift funds to and from programs—particularly as programs experience 
problems that require more time and money. Such shifts, in turn, have 
had costly, reverberating effects. In previous testimony and reports, 
GAO has stressed that DOD could avoid costly funding shifts. 

It could do this by developing an overall investment strategy to 
prioritize systems in its space portfolio with an eye toward balancing 
investments between legacy systems and new programs, as well as between 
science and technology programs and acquisition investments. Such 
prioritizing would also reduce incentives to produce low estimates. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Acton to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Evolutionary Acquisition And Cost Estimation: 

GAO has reported that evolutionary acquisition is in line with 
commercial best practices.[Footnote 24] In evolutionary acquisition, a 
program evolves to its ultimate capabilities on the basis of mature 
technologies and available resources. This approach allows commercial 
companies to develop and produce more sophisticated products faster and 
less expensively than their predecessors. 

Commercial companies have found that trying to capture the knowledge 
required to stabilize a product design that entails significant new 
technical content is an unmanageable task, especially if the goal is to 
reduce development cycle times and get the product to the marketplace 
as quickly as possible. Therefore, product features and capabilities 
that cannot be achieved in the initial development are planned for 
development in the product’s future generations, when the technology 
has proven mature and other resources are available. 

Figure 7 compares evolutionary to single-step acquisition, commonly 
called the big bang approach. An evolutionary environment for 
developing and delivering new products reduces risk and makes cost more 
predictable. While a customer may not initially receive an ultimate 
capability, the product is available sooner, with higher quality and 
reliability and at a lower and more predictable cost. With this 
approach, improvements can be planned for the product’s future 
generations. (See case study 17.) 

Figure 7: Evolutionary and Big Bang Acquisition Compared: 

[Refer to PDF for image: illustration] 

Evolutionary acquisition approach: 

Beginning: 

1st generation (5 years): 
* Basic stealth platform; 
Needed technologies are mature. 

2nd generation (10 years): 
* Basic stealth platform; 
* Advanced avionics; 
Needed technologies are mature. 

3rd generation (15 years): 
* Basic stealth platform; 
* Advanced avionics; 
* Advanced intelligence and communications. 

Single-step acquisition approach: 

1st generation (15 years): 
* Basic stealth platform; 
* Advanced avionics; 
* Advanced intelligence and communications. 

Source: GAO. 

[End of figure] 
 
Case Study 17: Evolutionary Acquisition and Cost Estimates, from Best 
Practices, GAO-03-645T: 

The U.S. Air Force F/A-22 tactical fighter acquisition strategy was, at 
the outset, to achieve full capability in a big bang approach. By not 
using an evolutionary approach, the F/A-22 took on significant risk and 
onerous technological challenges. While the big bang approach might 
have allowed the Air Force to compete more successfully for early 
funding, it hamstrung the program with many new, undemonstrated 
technologies, preventing the program from knowing cost and schedule 
ramifications throughout development. Cost, schedule, and performance 
problems resulted. 
 
Source: GAO, Best Practices: Better Acquisition Outcomes Are Possible 
If DOD Can Apply Lessons from FA-22 Program, GAO-03-645T, Washington, 
D.C.: Apr. 11, 2003. 

[End of case study] 

Two development processes support evolutionary acquisition: incremental 
development and spiral development. Both processes are based on 
maturing technology over time instead of trying to do it all at once, 
as in the big bang approach. Both processes allow for developing 
hardware and software in manageable pieces by inserting new technology 
and capability over time. This usually results in fielding an initial 
hardware or software increment (or block) of capability with steady 
improvements over less time than is possible with a full development 
effort. 

In incremental development, a desired capability is known at the 
beginning of the program and is met over time by developing several 
increments, each dependent on available mature technology. A core set 
of functions is identified and released in the first increment. Each 
new increment adds more functionality, and this process continues until 
all requirements are met. This assumes that the requirements are known 
up front and that lessons learned can be incorporated as the program 
matures. (See figure 8.) 

Figure 8: Incremental Development: 

[Refer to PDF for image: line graphs] 

Single step: 
Capability is plotted against time, with the following depicted: 
Technology base; 
Requirements; 
Capability; 
IOC; 
FOC. 
No capability. 

Incremental: 
Capability is plotted against time, with the following depicted: 
Technology base; 
Requirements; 
Capability. 
Initial operationally useful capability. 

Source: GAO. 

Note: 
IOC = initial operational capability; 
FOC = final operational capability. 

[End of figure] 

The advantages of incremental development are that a working product is 
available after the first increment and that each cycle results in 
greater capability. In addition, the program can be stopped when an 
increment is completed and still provide a usable product. Project 
management and testing can be easier, because the program is broken 
into smaller pieces. Its disadvantages are that the majority of the 
requirements must be known early, which is sometimes not feasible. In 
addition, cost and schedule overruns may result in an incomplete system 
if the program is terminated, because each increment only delivers a 
small part of the system at a time. Finally, operations and support for 
the program are often less efficient because of the need for additional 
learning for each increment release. (See case study 18.) 

Case Study 18: Incremental Development, from Customs Service 
Modernization, GAO/AIMD-99-41: 

The U.S. Customs Service was developing and acquiring the Automated 
Commercial Environment (ACE) program in 21 increments. At the time of 
GAO’s review, Customs defined the functionality of only the first 2 
increments, intending to define more later. Customs had nonetheless 
estimated costs and benefits for and had committed to investing in all 
21 increments. It had not estimated costs and benefits for each 
increment and did not know whether each increment would produce a 
reasonable return on investment. Furthermore, once it had deployed an 
increment at a pilot site for evaluation, Customs was not validating 
that estimated benefits had actually been achieved. It did not even 
know whether the program’s first increment, being piloted at three 
sites, was producing expected benefits or was cost-effective. Customs 
could determine only whether the first increment was performing at a 
level “equal to or better than” the legacy system. 

Source: GAO, Customs Service Modernization: Serious Management and 
Technical Weaknesses Must Be Corrected, GAO/AIMD-99-41 (Washington, 
D.C.: Feb. 26, 1999). 

[End of case study] 

Spiral Development: 

In spiral development, a desired capability is identified but the end-
state requirements are not yet known. These requirements are refined 
through demonstration and risk management, based on continuous user 
feedback. This approach allows each increment to provide the best 
possible capability. Spiral development is often used in the commercial 
market, because it significantly reduces technical risk while 
incorporating new technology. The approach can, however, lead to 
increased cost and schedule risks. Spiral development can also present 
contract challenges due to repeating phases, trading requirements, and 
redefining deliverables. 

The advantage of spiral development is that it provides better risk 
management, because user needs and requirements are better defined. Its 
disadvantage is that the process is a lot harder to manage and usually 
results in increased cost and longer schedule. 

While both incremental and spiral development have advantages and 
disadvantages, their major difference is the knowledge of the final 
product available to the program from the outset. With incremental 
development, the program office is aware of the final product to be 
delivered but develops it in stages. With spiral development, the final 
version of the product remains undetermined until the final stage has 
been completed—that is, the final product design is not known while the 
system is being built. 

Even though it is a best practice to follow evolutionary development 
rather than the big bang approach, it often makes cost estimating more 
difficult, because it requires that cost estimates be developed more 
frequently. In some cases, cost estimates made for programs are valid 
only for the initial increment or spiral, because future increments and 
spirals are not the product they were at the outset. Nevertheless, this 
approach is considered a best practice because it helps avoid 
unrealistic cost estimates, resulting in more realistic long-range 
investment funding and more effective resource allocation. Moreover, 
realistic cost estimates help management decide between competing 
options and increase the probability that the programs will succeed. 

1. Best Practices Checklist: The Estimate: 

* The cost estimate type is clearly defined and is appropriate for its 
purpose. 

* The cost estimate contains all elements suitable to its type—ICA, 
ICE, IGCE, LCCE, rough order of magnitude, total ownership cost: 
development, procurement, operating and support, disposal costs, and 
all sunk costs. 
- AOA, CEA, EA, cost-benefit analysis: consistently evaluate all 
alternatives. 
- EA, cost-benefit analysis: portray estimates as present values. 

* All program costs have been estimated, including all life-cycle 
costs. 

* The cost estimate is independent of funding source and 
appropriations. 

* An affordability analysis has been performed at the agency level to 
see how the program fits within the overall portfolio. 
- The agency has a process for developing cost estimates that includes 
the 12-step best practice process outlined in chapter 1. 
- An overall agency portfolio sand chart displays all costs for every 
program. 

* The estimate is updated as actual costs become available from the EVM 
system or requirements change. 

* Post mortems and lessons learned are continually documented. 

[End of Chapter 4] 

Chapter 5: The Cost Estimate’s Purpose, Scope, And Schedule: 

A cost estimate is much more than just a single number. It is a 
compilation of many lower-level cost element estimates that span 
several years, based on the program schedule. Credible cost estimates 
are produced by following the rigorous 12 steps outlined in chapter 1 
and are accompanied by detailed documentation. The documentation 
addresses the purpose of the estimate, the program background and 
system description, its schedule, the scope of the estimate (in terms 
of time and what is and is not included), the ground rules and 
assumptions, all data sources, estimating methodology and rationale, 
the results of the risk analysis, and a conclusion about whether the 
cost estimate is reasonable. Therefore, a good cost estimate—while 
taking the form of a single number—is supported by detailed 
documentation that describes how it was derived and how the expected 
funding will be spent in order to achieve a given objective. 

Purpose: 

The purpose of a cost estimate is determined by its intended use, and 
its intended use determines its scope and detail. Cost estimates have 
two general purposes: (1) to help managers evaluate affordability and 
performance against plans, as well as the selection of alternative 
systems and solutions, and (2) to support the budget process by 
providing estimates of the funding required to efficiently execute a 
program. 

More specific applications include providing data for trade studies, 
independent reviews, and baseline changes. Regardless of why the cost 
estimate is being developed, it is important that the program’s purpose 
link to the agency’s missions, goals, and strategic objectives. The 
purpose of the program should also address the benefits it intends to 
deliver, along with the appropriate performance measures for 
benchmarking progress. 

Scope: 

To determine an estimate’s scope, cost analysts must identify the 
customer’s needs. That is, the cost estimator must determine if the 
estimate is required by law or policy or is requested. For example, 10 
U.S.C. § 2434 requires an independent cost estimate before a major 
defense acquisition program can advance into system development and 
demonstration or production and deployment. The statute specifies 
that the full life-cycle cost—all costs of development, procurement, 
military construction, and operations and support, without regard to 
funding source or management control—must be provided to the decision 
maker for consideration. 

In other cases, a program manager might want initially to address 
development and procurement, with estimates of operations and support 
to follow. However, if an estimate is to support the comparative 
analysis of alternatives, all cost elements of each alternative should 
be estimated to make each alternative’s cost transparent in relation to 
the others. 

Where appropriate, the program manager and the cost estimating team 
should work together to determine the scope of the cost estimate. The 
scope will be determined by such issues as the time involved, what 
elements of work need to be estimated, who will develop the cost 
estimates, and how much cost estimating detail will be included. Where 
the program is in its life cycle will influence the quantity of detail 
for the cost estimate as well as the amount of data to be collected. 
For example, early in the life cycle the project may have a concept 
with no solid definition of the work involved. A cost estimate at this 
point in the life cycle will probably not require extensive detail. As 
the program becomes better defined, more detailed estimates should be 
prepared. 

Once the cost analysts know the context of the estimate or the 
customer’s needs, they can determine the estimate’s scope by its 
intended use and the availability of data. For example, if an 
independent cost analyst is typically given the time and other 
resources needed to conduct a thorough analysis, the analysis is 
expected to be more detailed than a what-if exercise. For either, 
however, more data are likely to be available for a system in 
production than for one that is in the early stages of development. 

More detail, though, does not necessarily mean greater accuracy. 
Pursuing too much detail too early may be detrimental to an estimate’s 
quality. If a detailed technical description of the system being 
analyzed is lacking, along with detailed cost data, analysts will find 
it difficult to identify and estimate all the cost elements. It may be 
better to develop the estimate at a relatively high system level to 
ensure capturing all the lower-level elements. This is the value of so-
called parametric estimating tools, which operate at a higher level of 
detail and are used when a system lacks detailed technical definition 
and cost data. These techniques also allow the analyst to link cost and 
schedule to measures of system size, functionality, or complexity in 
advance of detailed design definition. 

Analysts should develop, and tailor, an estimate plan whose scope 
coincides with data availability and the estimate’s ultimate use. For a 
program in development, which is estimated primarily with parametric 
techniques and factors, the scope might be at a higher level of the 
WBS. (WBS is discussed in chapter 8.) As the program enters production, 
a lower level of detail would be expected. 

As the analysts develop and revise the estimating plan, they should 
keep management informed of the initial approach and any changes in 
direction or method.[Footnote 25] Since the plan serves as an agreement 
between the customer and cost estimating team, it must clearly reflect 
the approved approach and should be distributed formally to all 
participants and organizations involved. 

Schedule: 

Regardless of an estimate’s ultimate use and its data availability, 
time can become an overriding constraint on its detail. When defining 
the elements to be estimated and when developing the plan, the cost 
estimating team must consider its time constraints relative to team 
staffing. Without adequate time to develop a competent estimate, the 
team may be unable to deliver a product of sufficiently high quality. 
For example, a rough-order-of-magnitude estimate could be developed in 
days, but a first-time budget-quality estimate would likely require 
many months. If, however, that budget estimate were simply an update to 
a previous estimate, it could be done faster. The more detail required, 
the more time and staff the estimate will require. It is important, 
therefore, that auditors understand the context of the cost 
estimate—why and how it was developed and whether it was an initial or 
follow-on estimate. (See case study 19.) 

Case Study 19: The Estimate’s Context, from DOD Systems 
Modernization, GAO-06-215: 

Program officials told GAO that they had not developed the 2004 cost 
estimate in accordance with all SEI’s cost estimating criteria, because 
they had only a month to complete the economic analysis. By not 
following practices associated with reliable estimates—by not making a 
reliable estimate of system life-cycle costs—the Navy had decided on a 
course of action not based on sound and prudent decision making. This 
meant that the Navy’s investment decision was not adequately justified 
and that to the extent that program budgets were based on cost 
estimates, the likelihood of funding shortfalls and inadequate funding 
reserves was increased. 

Source: GAO, DOD Systems Modernization: Planned Investment in the Naval 
Tactical Command Support System Needs to Be Reassessed, GAO-06-215, 
Washington, D.C.: Dec. 5, 2005. 

[End of case study] 

After the customer has defined the task, the cost estimating team 
should create a detailed schedule that includes realistic key decision 
points or milestones and that provides margins for unforeseen, but not 
unexpected, delays. The team must ensure that the schedule is not 
overly optimistic. If the team wants or needs to compress the schedule 
to meet a due date, compression is acceptable as long as additional 
resources are available to complete the effort that fewer analysts 
would have accomplished in the longer period of time. If additional 
resources are not available, the estimate’s scope must be reduced. 

The essential point is that the team must attempt to ensure that the 
schedule is reasonable. When this is not possible, the schedule must be 
highlighted as having curtailed the team’s depth of analysis and the 
estimate’s resulting confidence level. 

2. Best Practices Checklist: Purpose, Scope, and Schedule: 

* The estimate’s purpose is clearly defined. 

* Its scope is clearly defined. 

* The level of detail the estimate is to be conducted at is consistent 
with the level of detail available for the program. For example, an 
engineering buildup estimate should be conducted only on a well-defined 
program. 

* The team has been allotted adequate time and resources to develop the 
estimate. 

[End of Chapter 5] 

Chapter 6: 

The Cost Assessment Team: 

Cost estimates are developed with an inexact knowledge of what the 
final technical solution will be. Therefore, the cost assessment team 
must manage a great deal of risk—especially for programs that are 
highly complex or on technology’s cutting edge. Since cost estimates 
seek to define what a given solution will ultimately cost, the estimate 
must be bound by a multitude of assumptions and an interpretation of 
what the historical data represent. This tends to be a subjective 
effort, and these important decisions are often left to a cost 
analyst’s judgment. A cost analyst must possess a variety of skills to 
develop a high-quality cost estimate that satisfies the 12 steps 
identified in chapter 1, as figure 9 illustrates. 

Figure 9: Disciplines and Concepts in Cost Analysis: 

[Refer to PDF for image: illustration] 

Cost Analysis: 
 
Economics: 
* Break-even analysis; 
* Foreign exchange rates; 
* Industrial base analysis; 
* Inflation;
* Labor agreements; 
* Present value analysis.

Budgeting: 
* Budget appropriations; 
* Internal company (industry); 
* Program specific. 

Engineering: 
* Design; 
* Materials; 
* Performance parameters; 
* Production engineering; 
* Production process; 
* Program development test; 
* Scheduling; 
* System integration. 

Computer science/mathematics: 
* Analysis of commercial models; 
* Analysis of proposals; 
* Development of cost estimating relationship; 
* Model development; 
* Programming. 

Statistics: 
* Forecasting; 
* Learning curve applications; 
* Regression analysis; 
* Risk/uncertainty analysis; 
* Sensitivity analysis. 

Accounting: 
* Cost data analysis; 
* Financial analysis; 
* Overhead analysis; 
* Proposal analysis. 

Interpersonal skills: 
* Approach; 
* Estimate; 
* Knowledge. 
 
Public and government affairs: 
* Appropriations process; 
* Auditors; 
* Legislative issues; 
* Outside factors. 

Source: GAO. 

[End of figure] 

Each discipline in figure 9 applies to cost estimating in its own 
unique way. For example, having an understanding of economics and 
accounting will help the cost estimator better understand the 
importance of inflation effects and how different accounting systems 
capture costs. Budgeting knowledge is important for knowing how to 
properly allocate resources over time so that funds are available when 
needed. Because cost estimates are often needed to justify enhancing 
older systems, having an awareness of engineering, computer science, 
mathematics, and statistics will help identify cost drivers and the 
type of data needed to develop the estimate. It also helps for the cost 
estimator to have adequate technical knowledge when meeting with 
functional experts so that credibility and a common understanding of 
the technical aspects of the program can be quickly established. 
Finally, cost estimators who are able to “sell” and present their 
estimate by defending it with solid facts and reliable data stand a 
better chance of its being used as a basis for program funding. In 
addition, cost estimators need to have solid interpersonal skills, 
because working and communicating with subject matter experts is vital 
for understanding program requirements. 

Team Composition And Organization: 

Program office cost estimates are normally prepared by a 
multidisciplinary team whose members have functional skills in 
financial management, engineering, acquisition and logistics, 
scheduling, and mathematics, in addition to communications.[Footnote 
26] The team should also include participants or reviewers from the 
program’s operating command, product support center, maintenance depot, 
and other units affected in a major way by the estimate.[Footnote 27] 
Team members might also be drawn from other organizations. In the best 
case, the estimating team is composed of persons who have experience in 
estimating all cost elements of the program. Since this is seldom 
possible, the team leader should be familiar with the team members’ 
capabilities and assign tasks accordingly. If some are experienced in 
several areas, while others are relatively inexperienced in all areas, 
the team leader should assign the experienced analysts responsibility 
for major sections of the estimate while the less experienced analysts 
work under their supervision. 

An analytic approach to cost estimates typically entails a written 
study plan detailing a master schedule of specific tasks, responsible 
parties, and due dates. For complex efforts, the estimating team might 
be organized as a formal, integrated product team. For independent 
estimates, the team might be smaller and less formal. In either case, 
the analysis should be coordinated with all stakeholders, and the study 
plan should reflect each team member’s responsibilities. 

What is required of a cost estimating team depends on the type and 
purpose of the estimate and the quantity and quality of the data. More 
detailed estimates generally require larger teams, more time and 
effort, and more rigorous techniques. For example, a rough-order-of-
magnitude estimate—a quick, high-level cost estimate—generally requires 
less time and effort than a budget-quality estimate. In addition, the 
estimating team must be given adequate time to develop the estimate. 
Following the 12 steps takes time and cannot be rushed—rushing would 
significantly risk the quality of the results. 

One of the most time consuming steps in the cost estimating process is 
step 6: obtaining the data. Enough time should be scheduled to collect 
the data, including visiting contractor sites to further understand the 
strengths and limitations of the data that have been collected. If 
there is not enough time to develop the estimate, then the schedule 
constraint should be clearly identified in the ground rules and 
assumptions, so that management understands the effect on the 
estimate’s quality and confidence. 

Cost estimating requires good organizational skills, in order to pull 
together disparate data for each cost element and to package it in a 
meaningful way. It also requires engineering and mathematical skills, 
to fully understand the quality of the data available. Excellent 
communication skills are also important for clarifying the technical 
aspects of a program with technical specialists. If the program has no 
technical baseline description, or if the cost estimating team must 
develop one, it is essential that the team have access to the subject 
matter experts—program managers, system and software engineers, test 
and evaluation analysts—who are familiar with the program or a program 
like it. Moreover, team members need good communication skills to 
interact with these experts in ways that are meaningful and productive. 

Cost Estimating Team Best Practices: 

Centralizing the cost estimating team and process—cost analysts working 
in one group but supporting many programs—represents a best practice, 
according to the experts we interviewed. Centralization facilitates the 
use of standardized processes, the identification of resident experts, 
a better sharing of resources, commonality and consistency of tools and 
training, more independence, and a career path with more opportunities 
for advancement. Centralizing cost estimators and other technical and 
business experts also allows for more effective deployment of technical 
and business skills while ensuring some measure of independence. 

A good example is in the Cost Analysis Improvement Group (CAIG) in the 
Office of the Secretary of Defense. Its cost estimates are produced by 
a centralized group of civilian government personnel to ensure long-
term institutional knowledge and no bias toward results. Some in the 
cost estimating community consider a centralized cost department that 
provides cost support to multiple program offices, with a strong 
organizational structure and support from its leadership, to be a 
model. 

In contrast, decentralization often results in ad hoc processes, 
limited government resources (requiring contractor support to fill the 
gaps), and decreased independence, since program offices typically fund 
an effort and since program management personnel typically rate the 
analysts’ performance. The major advantage of a decentralized process 
is that analysts have better access to technical experts. Under a 
centralized process, analysts should thus make every effort to 
establish contacts with appropriate technical experts. 

Finally, organizations that develop their own centralized cost 
estimating function but outside the acquiring program represent the 
best practice over organizations that develop their cost estimates in a 
decentralized or ad hoc manner under the direct control of a program 
office. One of the many benefits of centralized structure is the 
ability to resist pressure to lower the cost estimate when it is higher 
than the allotted budget. Furthermore, reliance on support contractors 
raises questions from the cost estimating community about whether 
numbers and qualifications of government personnel are sufficient to 
provide oversight of and insight into contractor cost estimates. Other 
experts in cost estimating suggested that reliance on support 
contractors can be a problem if the government cannot evaluate how good 
a cost estimate is or if the ability to track it is lacking. Studies 
have also raised the concern that relying on support contractors makes 
it more difficult to retain institutional knowledge and instill 
accountability. Therefore, to mitigate any bias in the cost estimate, 
government customers of contractor-produced cost estimates must have a 
high enough level of experience to determine whether the cost estimate 
conforms to the best practices outlined in this Guide. 

Certification And Training for Cost Estimating And EVM Analysis: 

Since the experience and skills of the members of a cost estimating 
team are important, various organizations have established training 
programs and certification procedures. For example, SCEA’s 
certification program provides a professional credential to both 
members and nonmembers for education, training, and work experience and 
a written examination on basic concepts and methods for cost 
estimating. Another example is the earned value professional 
certification offered by the Association for the Advancement of Cost 
Engineering International that PMI’s College of Performance Management 
endorses; it requires candidates to have the requisite experience and 
the ability to pass a rigorous written exam. 

Under the Defense Acquisition Workforce Improvement Act, DOD 
established a variety of certification programs through the Defense 
Acquisition University (DAU).[Footnote 28] DAU provides a full range of 
basic, intermediate, and advanced certification training; assignment-
specific training; performance support; job-relevant applied research; 
and continuous learning opportunities. Although DAU’s primary mission 
is to train DOD employees, all federal employees are eligible to attend 
as space is available. One career field is in business, cost 
estimating, and financial management. Certification levels are based on 
education, experience, and training. Since this certification is 
available to all federal employees, it is considered a minimum training 
requirement for cost estimators. 

In addition to the mandatory courses in table 5, DAU encourages 
analysts to be trained in courses identified in its Core Plus 
Development Guide. These courses cover a wide range of cost estimating 
and earned value topics, such as acquisition reporting concepts and 
policy requirements, analysis of alternatives, baseline maintenance, 
basic software acquisition management, business case analysis, business 
management modernization, contract source selection, cost as an 
independent variable, economic analysis, EVM system validation and 
surveillance, integrated acquisition for decision makers, operating 
and support cost analysis, principles of schedule management, program 
management tools, and risk management. The standards for the business, 
cost estimating, and financial management levels of certification are 
shown in table 5. 

Table 5: Certification Standards in Business, Cost Estimating, and 
Financial Management in the Defense Acquisition Education, Training, 
and Career Development Program: 
 
Level: I, Desired; 
Education: Baccalaureate. 
 
Level: I, Mandatory; 
Experience: 1 year of acquisition in business, cost estimating, or 
financial management; 
Training: 
ACQ 101: Fundamentals of Systems Acquisition Management and 2 of the 
following: 
BCF 101: Fundamentals of Cost Analysis; 
BCF 102: Fundamentals of Earned Value; 
BCF 103: Fundamentals of Business Financial Management. 
 
Level: II, Desired: 
Education: Baccalaureate; 
Experience: 2 additional years in business, cost estimating, or 
financial management. 
 
Level: II, Mandatory; 
Experience: 2 years of acquisition in business, cost estimating, or 
financial management; 
Training: 
ACQ 201: (Parts A & B) Intermediate Systems Acquisition and; 
BCF 205: Contractor Business Strategies and, if not taken at Level I, 
BCF 101: Fundamentals of Cost Analysis or, 
BCF 102: Fundamentals of Earned Value Management or, 
BCF 103: Fundamentals of Business Financial Management and one of the 
following: 
BCF 203: Intermediate Earned Value Management or, 
BCF 204: Intermediate Cost Analysis or, 
BCF 211: Acquisition Business Management. 
 
Level: III, Desired; 
Education: Baccalaureate or 24 semester hours among 10 courses[A] or 
Master’s; 
Experience: 4 additional years of acquisition in business, cost 
estimating, or financial management. 
 
Level: III, Mandatory; 
Training: BCF 301: Business, Cost Estimating, and Financial Management 
Workshop. 

Source: DAU. 

[A] The 10 courses are accounting, business finance, contracts, 
economics, industrial management, law, marketing, organization and 
management, purchasing, and quantitative methods. 

[End of table] 

When reviewing an agency’s cost estimate, an auditor should question 
the cost estimators about whether they have both the requisite formal 
training and substantial on-the-job training to develop cost estimates 
and keep those estimates updated with EVM analysis. Continuous learning 
by participating in cost estimating and EVM conferences is important 
for keeping abreast of the latest techniques and maximizing lessons 
learned. Agency cost estimators and EVM analysts, as well as GAO’s 
auditors, should attend such conferences to keep their skills current. 
Maintaining skills is essential if subject matter experts are to be 
relied on to apply best practices in their roles. 

While formal training is important, so is on-the-job training and first-
hand knowledge from participating in plant and site visits. On-site 
visits to see what is being developed and how engineering and 
manufacturing are executed are invaluable to cost estimators and 
auditors. To understand the complexity of the tasks necessary to 
deliver a product, site visits should always be included in the audit 
plan. 

SEI’s Checklists and Criteria for Evaluating the Cost and Schedule 
Estimating Capabilities of Software Organizations lists six requisites 
for reliable estimating and gives examples of evidence needed to 
satisfy them. It also contains a checklist for estimating whether an 
organization provides its commitment and support to the estimators. 
SEI’s criteria are helpful for determining whether cost estimators have 
the skills and training to effectively develop credible cost estimates. 
(See appendix VIII for a link to SEI’s material.) 

While much of this Cost Guide’s focus is on cost estimating, in chapter 
18 we focus on EVM and how it follows the cost estimate through its 
various phases and determines where there are cost and schedule 
variances and why. This information is vitally important to keeping the 
estimate updated and for keeping abreast of program risks. Because of 
performance measurement requirements (including the use of EVM), OMB 
issued policy guidance in August 2005 to agency chief information 
officers on improving information technology projects. OMB stated that 
the Federal Acquisition Institute (co-located with DAU) was expanding 
EVM system training to the program management and contracting 
communities and instructed agencies to refer to DAU’s Web site for a 
community of practice that includes the following resources:[Footnote 
29] 
 
* 6 hours of narrated EVM tutorials (Training Center), 

* descriptions and links to EVM tools (Tools), 

* additional EVM-related references and guides (Community Connection), 

* DOD policy and contracting guidance (Contract Documents and DOD 
Policy and Guidance), 

* a discussion forum (Note Board), and 

* an on-line reference library (Research Library). 

Such resources are important for agencies and auditors in understanding 
what an EVM system can offer for improving program management. 

3. Best Practices Checklist: Cost Assessment Team: 

* The estimating team’s composition is commensurate with the assignment 
(see SEI’s checklists for more details). 
- The team has the proper number and mix of resources. 
- Team members are from a centralized cost estimating organization. 
- The team includes experienced and trained cost analysts. 
- The team includes, or has direct access to, analysts experienced in 
the program’s major areas. 
- Team members’ responsibilities are clearly defined. 
- Team members’ experience, qualifications, certifications, and 
training are identified. 
- The team participated in on-the-job training, including plant and 
site 
visits. 

* A master schedule with a written study plan has been developed. 

* The team has access to the necessary subject matter experts. 

[End of Chapter 6] 

Chapter 7: 

Technical Baseline Description Definition And Purpose: 

Key to developing a credible estimate is having an adequate 
understanding of the acquisition program—the acquisition strategy, 
technical definition, characteristics, system design features, and 
technologies to be included in its design. The cost estimator can use 
this information to identify the technical and program parameters that 
will bind the cost estimate. The amount of information gathered 
directly affects the overall quality and flexibility of the estimate. 
Less information means more assumptions must be made, increasing the 
risk associated with the estimate. Therefore, the importance of this 
step must be emphasized, because the final accuracy of the cost 
estimate depends on how well the program is defined. 

The objective of the technical baseline is to provide in a single 
document a common definition of the program—including a detailed 
technical, program, and schedule description of the system—from which 
all LCCEs will be derived—that is, program and independent cost 
estimates. At times, the information in the technical baseline will 
drive or facilitate the use of a particular estimating approach. 
However, the technical baseline should be flexible enough to 
accommodate a variety of estimating methodologies. It is also critical 
that the technical baseline contain no cost data, so that it can be 
used as the common baseline for independently developed estimates. 
[Footnote 30] 
 
In addition to providing a comprehensive program description, the 
technical baseline is used to benchmark life-cycle costs and identify 
specific technical and program risks. In this way, it helps the 
estimator focus on areas or issues that could have a major cost effect. 

Process: 

In general, program offices are responsible for developing and 
maintaining the technical baseline throughout the life cycle, since 
they know the most about their program. A best practice is to assign an 
integrated team of various experts—system engineers, design experts, 
schedulers, test and evaluation experts, financial managers, and cost 
estimators—to develop the technical baseline at the beginning of the 
project. The program manager and the senior executive oversight 
committee approve the technical baseline to ensure that it contains all 
information necessary to define the program’s systems and develop 
the cost estimate. 

Furthermore, the technical baseline should be updated in preparation 
for program reviews, milestone decisions, and major program changes. 
The credibility of the cost estimate will suffer if the technical 
baseline is not maintained. Without explicit documentation of the basis 
of a program’s estimates, it is difficult to update the cost estimate 
and provide a verifiable trace to a new cost baseline as key 
assumptions change during the course of the program’s life. 

It is normal and expected that early program technical baselines will 
be imprecise or incomplete and that they will evolve as more 
information becomes known. However, it is essential that the technical 
baseline provide the best available information at any point in time. 
To try to create an inclusive view of the program, assumptions should 
be made about the unknowns and should be agreed on by management. These 
assumptions and their corresponding justifications should be documented 
in the technical baseline, so their risks are known from the beginning. 

Schedule: 

The technical baseline must be available in time for all cost 
estimating activities to proceed on schedule. This often means that it 
is submitted as a draft before being made final. The necessary lead 
time will vary by organization. One example is the CAIG in the Office 
of the Secretary of Defense, which requires that the Cost Analysis 
Requirements Description be submitted in draft 180 days before the 
Defense Acquisition Board milestone and in final form 45 days before 
the milestone review. 

Contents: 
 
Since the technical baseline is intended to serve as the baseline for 
developing LCCEs, it must provide information on development, testing, 
procurement, installation and replacement, operations and support, 
planned upgrades, and disposal. In general, a separate technical 
baseline should be prepared for each alternative; as the program 
matures, the number of alternatives and, therefore, technical baselines 
decreases. Although technical baseline content varies by program (and 
possibly even by alternative), it always entails a number of sections, 
each focusing on a particular aspect of the program being assessed. 
Table 6 describes typical technical baseline elements. 

Table 6: Typical Technical Baseline Elements: 
 
Element: System purpose; 
Description: Describes the system’s mission and how it fits into the 
program; should give the estimator a concept of its complexity and 
cost. 

Element: Detailed technical system and performance characteristics; 
Description: Includes key functional requirements and performance 
characteristics; the replaced system (if applicable); who will develop, 
operate, and maintain the system; descriptions of hardware and software 
components (including interactions, technical maturity of critical 
components, and standards); system architecture and equipment 
configurations (including how the program will interface with other 
systems); key performance parameters; information assurance; 
operational concept; reliability analysis; security and safety 
requirements; test and evaluation concepts and plans. 

Element: Work breakdown structure; 
Description: Identifies the cost and technical data needed to develop 
the estimate. 

Element: Description of legacy or similar systems; 
Description: A legacy (or heritage or predecessor) system has 
characteristics similar to the system being estimated; often the new 
program is replacing it. The technical baseline includes a detailed 
description of the legacy hardware and software components; technical 
protocols or standards; key performance parameters; operational and 
maintenance logistics plan; training plan; phase-out plan; and the 
justification for replacing the system. 

Element: Acquisition plan or strategy; 
Description: Includes the competition strategy, whether multiyear 
procurement will be used, and whether the program will lease or buy 
certain items; it should identify the type of contract awarded or to be 
awarded and, if known, the contractor responsible for developing and 
implementing the system. 

Element: Development, test, and production quantities and program 
schedule; 
Description: Includes quantities required for development, test (e.g., 
test assets), and production; lays out an overall development and 
production schedule that identifies the years of its phases—the 
schedule should include a standard Gantt chart with major events such 
as milestone reviews, design reviews, and major tests—and that 
addresses, at a high level, major program activities, their duration 
and sequence, and the critical path. 

Element: System test and evaluation plan; 
Description: Includes the number of tests and test assets, criteria for 
entering into testing, exit criteria for passing the test, and where 
the test will be conducted. 

Element: Deployment details; 
Description: Includes standard platform and site configurations for all 
scenarios (peacetime, contingency, war) and a transition plan between 
legacy and new systems Safety plan Includes any special or unique 
system safety considerations that may relate to specific safety goals 
established through standards, laws, regulations, and lessons learned 
from similar systems. 
 
Element: Training plan; 
Description: Includes training for users and maintenance personnel, any 
special certifications required, who will provide the training, where 
it will be held, and how often it will be offered or required. 

Element: Disposal and environmental effect; 
Description: Includes identification of environment impact, mitigation 
plan, and disposal concept. 
 
Element: Operational concept; 
Description: Includes program management details, such as how, where, 
and when the system will be operated; the platforms on which it will be 
installed; and the installation schedule. 

Element: Personnel requirements; 
Description: Includes comparisons to the legacy system (if possible) in 
salary levels, skill-level quantity requirements, and where staff will 
be housed. 

Element: Logistics support details; 
Description: Includes maintenance and sparing plans, as well as planned 
upgrades. 

Element: Changes from the previous technical baseline; 
Description: Includes a tracking of changes, with a summary of what 
changed and why. 

Source: DOD, DOE, and SCEA. 

[End of table] 

Programs following an incremental development approach should have a 
technical baseline that clearly states system characteristics for the 
entire program. In addition, the technical baseline should define the 
characteristics to be included in each increment, so that a rigorous 
LCCE can be developed. For programs with a spiral development approach, 
the technical baseline tends to evolve as requirements become better 
defined. In earlier versions of a spiral development program, the 
technical baseline should clearly state the requirements that are 
included and those that have been excluded. This is important, since a 
lack of defined requirements can lead to cost increases and delays in 
delivering services, as case study 20 illustrates. 

Case Study 20: Defining Requirement, from United States Coast Guard, 
GAO-06-623: 

The U.S. Coast Guard contracted in September 2002 to replace its search 
and rescue communications system, installed in the 1970s, with a new 
system known as Rescue 21. The acquisition and initial implementation 
of Rescue 21, however, resulted in significant cost overruns and 
schedule delays. By 2005, its estimated total acquisition cost had 
increased to $710.5 million from 1999’s $250 million, and the schedule 
for achieving full operating capability had been delayed from 2006 to 
2011. GAO reported in May 2006 on key factors contributing to the cost 
overruns and schedule delays, including requirements management. 
Specifically, GAO found that the Coast Guard did not have a rigorous 
requirements management process. 

Although the Coast Guard had developed high-level requirements, it 
relied solely on the contractor to manage them. According to Coast 
Guard acquisition officials, they had taken this approach because of 
the performance-based contract vehicle. GAO’s experience in reviewing 
major systems acquisitions has shown that it is important for 
government organizations to exercise strong leadership in managing 
requirements, regardless of the contracting vehicle. 

Besides not effectively managing requirements, Rescue 21 testing 
revealed numerous problems linked to incomplete and poorly defined user 
requirements. For example, a Coast Guard usability and operability 
assessment of Rescue 21 stated that most of the operational 
advancements envisioned for the system had not been achieved, 
concluding that these problems could have been avoided if the contract 
had contained user requirements. 

A key requirement was to “provide a consolidated regional geographic 
display.” The contractor provided a capability based on this 
requirement but, during testing, the Coast Guard operators believed 
that the maps did not display sufficient detail. Such discrepancies led 
to an additional statement of work that defined required enhancements 
to the system interface, such as screen displays. 

GAO reported that if deploying Rescue 21 were to be further delayed, 
Coast Guard sites and services would be affected in several ways. Key 
functionality, such as improved direction finding and improved coverage 
of coastal areas, would not be available as planned. Coast Guard 
personnel at those sites would continue to use outdated legacy 
communications systems for search and rescue operations, and coverage 
of coastal regions would remain limited. In addition, delays could 
result in costly upgrades to the legacy system in order to address 
communications coverage gaps, as well as other operational concerns. 

Source: GAO, United States Coast Guard: Improvements Needed in 
Management and Oversight of Rescue System Acquisition, GAO-06-623, 
Washington, D.C.: May 31, 2006. 

[End of case study] 

Fully understanding requirements up front helps increase the accuracy 
of the cost estimate. While each program should have a technical 
baseline that addresses each element in table 6, each program’s aspects 
are unique. In the next section, we give examples of system 
characteristics and performance parameters typically found in 
government cost estimates, including military weapon systems and 
civilian construction and information systems. 

Key System Characteristics and Performance Parameters: 

Since systems differ, each one has unique physical and performance 
characteristics. Analysts need specific knowledge about them before 
they can develop a cost estimate for a weapon system, an information 
system, or a construction program. 

While the specific physical and performance characteristics for a 
system being estimated will be dictated by the system and the 
methodology used to perform the estimate, several general 
characteristics have been identified in the various guides we reviewed. 
Table 7 lists general characteristics shared within several system 
types. 

Table 7: General System Characteristics: 

System: Aircraft; 
Characteristics: 
* Breakdown of airframe unit weight by material type; 
* Combat ceiling and speed; 
* Internal fuel capacity; 
* Length; 
* Load factor; 
* Maximum altitude; 
* Maximum speed (knots at sea level); 
* Mission and profile; 
* Weight; 
- Type: Airframe unit weight, combat, empty, maximum gross, 
payload, structure; 
* Wetted area; 
* Wing; 
- Type: Wingspan, wing area, wing loading. 

System: Automated information systems; 
Characteristics: 
* Architecture; 
* Commercial off-the-shelf software used; 
* Customization of commercial off-the-shelf software; 
* Expansion factors; 
* Memory size; 
* Processor type; 
* Proficiency of programmers; 
* Programming language used; 
* Software sizing metric. 

System: Construction; 
Characteristics: 
* Changeover; 
* Environmental impact; 
* Geography; 
* Geology; 
* Liability; 
* Location: 
- Type: Land value, proximity to major roads, relocation expenses; 
* Material type: 
- Type: Composite, masonry, metal, tile, wood shake; 
* Number of stories; 
* Permits; 
* Public acceptance; 
* Square feet; 
* Systemization. 

System: Missiles; 
Characteristics: 
* Height; 
* Length; 
* Payload; 
* Propulsion type; 
* Range; 
* Sensors; 
* Weight; 
* Width. 
 
System: Ships; 
Characteristics: 
* Acoustic signature; 
* Full displacement; 
* Full load weight; 
* Length overall; 
* Lift capacity; 
* Light ship weight; 
* Margin; 
* Maximum beam; 
* Number of screws; 
* Payload; 
* Propulsion type; 
* Shaft horsepower. 

System: Space; 
Characteristics: 
* Attitude; 
* Design life and reliability; 
* Launch vehicle; 
* Mission and duration; 
* Orbit type; 
* Pointing accuracy; 
* Satellite type; 
* Thrust; 
* Weight and volume. 

System: Tanks and trucks; 
Characteristics: 
* Engine; 
* Height; 
* Horsepower; 
* Length; 
* Weight; 
* Width; 
* Payload. 
 
Source: DOD and GAO. 

[End of table] 

Once a system’s unique requirements have been defined, they must be 
managed and tracked continually throughout the program’s development. 
If requirements change, both the technical baseline and cost estimate 
should be updated so that users and management can understand the 
effects of the change. When requirements are not well managed, users 
tend to become disillusioned, and costs and schedules can spin out of 
control, as case study 21 demonstrates. 

Case Study 21: Managing Requirements, from DOD Systems Modernization, 
GAO-06-215: 
 
The Naval Tactical Command Support System (NTCSS) was started in 1995 
to help U.S. Navy personnel manage ship, submarine, and aircraft 
support activities. At the time of GAO’s review, about $1 billion had 
been spent to partially deploy NTCSS to about half its intended sites. 
In December 2005, GAO reported that the Navy had not adequately 
conducted requirements management and testing activities for the 
system. For example, requirements had not been prioritized or traced to 
related documentation to ensure that the system’s capabilities would 
meet users’ needs. As a result, failures in developmental testing had 
prevented NTCSS’s latest component from passing operational testing 
twice over the preceding 4 years. From the Navy’s data, the recent 
trend in key indicators of system maturity, such as the number and 
nature of reported system problems and change proposals, showed that 
problems with NTCSS had persisted and that they could involve costly 
rework. In addition, the Navy did not know the extent to which NTCSS’s 
optimized applications were meeting expectations—even though the 
applications had been deployed to 229 user sites since 1998—because 
metrics to demonstrate that the expectations had been met had not been 
defined and collected. 

Source: GAO, DOD Systems Modernization: Planned Investment in the Naval 
Tactical Command Support System Needs to Be Reassessed, GAO-06-215 
Washington, D.C.: Dec. 5, 2005. 

[End of case study] 

Case study 21 shows that an inability to manage requirements leads to 
additional costs and inefficient management of resources. To manage 
requirements, they must first be identified. The bottom line is that 
the technical baseline should document the underlying technical and 
program assumptions necessary to develop a cost estimate and update 
changes as they occur. Moreover, the technical baseline should also 
identify the level of risk associated with the assumptions so that the 
estimate’s credibility can be determined. As we stated previously, the 
technical baseline should mature in the same manner as the program 
evolves. Because it is evolutionary, earlier versions of the technical 
baseline will necessarily include more assumptions and, therefore, more 
risk, but these should decline as risks become either realized or 
retired. 

4. Best Practices Checklist: Technical Baseline Description: 

* There is a technical baseline: 
- The technical baseline has been developed by qualified personnel such 
as system engineers. 
- It has been updated with technical, program, and schedule changes, 
and it contains sufficient detail of the best available information at 
any given time. 
- The information in the technical baseline generally drives the cost 
estimate and the cost estimating methodology. 
- The cost estimate is based on information in the technical baseline 
and has been approved by management. 

* The technical baseline answers the following: 
- What the program is supposed to do—requirements; 
- How the program will fulfill its mission—purpose; 
- What it will look like—technical characteristics; 
- Where and how the program will be built—development plan; 
- How the program will be acquired—acquisition strategy; 
- How the program will operate—operational plan; 
- Which characteristics affect cost the most—risk. 

[End of Chapter 7] 

Chapter 8: Work breakdown Structure: 

A work breakdown structure is the cornerstone of every program because 
it defines in detail the work necessary to accomplish a program’s 
objectives. For example, a typical WBS reflects the requirements, what 
must be accomplished to develop a program, and provides a basis for 
identifying resources and tasks for developing a program cost estimate. 
A WBS is also a valuable communication tool between systems 
engineering, program management, and other functional organizations 
because it provides a clear picture of what needs to be accomplished 
and how the work will be done. Accordingly, it is an essential element 
for identifying activities in a program’s integrated master schedule. 
In addition, it provides a consistent framework for planning and 
assigning responsibility for the work. Initially set up when the 
program is established, the WBS becomes successively detailed over time 
as more information becomes known about the program. 

A WBS is a necessary program management tool because it provides a 
basic framework for a variety of related activities like estimating 
costs, developing schedules, identifying resources, determining where 
risks may occur, and providing the means for measuring program status 
using EVM. Furthermore, a well structured WBS helps promote 
accountability by identifying work products that are independent of one 
another. It also provides the framework to develop a schedule and cost 
plan that can easily track technical accomplishments—in terms of 
resources spent in relation to the plan as well as completion of 
activities and tasks—enabling quick identification of cost and schedule 
variances. 

Best Practice: Product-Oriented WBS: 
 
A WBS deconstructs a program’s end product into successive levels with 
smaller specific elements until the work is subdivided to a level 
suitable for management control. By breaking work down into smaller 
elements, management can more easily plan and schedule the program’s 
activities and assign responsibility for the work. It also facilitates 
establishing a schedule, cost, and EVM baseline. Establishing a product-
oriented WBS is a best practice because it allows a program to track 
cost and schedule by defined deliverables, such as a hardware or 
software component. This allows a program manager to more precisely 
identify which components are causing cost or schedule overruns and to 
more effectively mitigate the root cause of the overruns. 

A WBS breaks down product-oriented elements into a hierarchical 
structure that shows how elements relate to one another as well as to 
the overall end product. A 100 percent rule is followed that states 
that “the next level of decomposition of a WBS element (child level) 
must represent 100 percent of the work applicable to the next higher 
(parent) element.”[Footnote 31] This is considered a best practice by 
many experts in cost estimating, because a product-oriented WBS 
following the 100 percent rule ensures that all costs for all 
deliverables are identified. Failing to include all work for all 
deliverables can lead to schedule delays and subsequent cost increases. 
It can also result in confusion among team members. To avoid these 
problems, standardizing the WBS is a best practice in organizations 
where there is a set of program types that are standard and typical. 
This enables an organization to simplify the development of the top-
level program work breakdown structures by publishing the standard. It 
also facilitates an organization’s ability to collect and share data 
from common WBS elements among many programs. The more data that are 
available for creating the cost estimate, the higher the confidence 
level will be. 

Its hierarchical nature allows the WBS to logically sum the lower-level 
elements that support the measuring of cost, schedule, and technical 
analysis in an EVM system. A good WBS clearly defines the logical 
relationship of all program elements and provides a systematic and 
standardized way for collecting data across all programs. Therefore, a 
WBS is an essential part of developing a program’s cost estimate and 
enhancing an agency’s ability to collect data necessary to support 
future cost estimates. Moreover, when appropriately integrated with 
systems engineering, cost estimating, EVM, and risk management, a WBS 
provides the basis to allow program managers to have a better view into 
a program’s status, facilitating continual improvement. 

A WBS is developed and maintained by a systems engineering process that 
produces a product-oriented family tree of hardware, software, 
services, data, and facilities. It can be thought of as an illustration 
of what work will be accomplished to satisfy a program’s requirements. 
The WBS diagrams the effort in small discrete pieces, or elements, to 
show how each one relates to the others and to the program as a whole. 
These elements such as hardware, software, and data are further broken 
down into specific lower-level elements. The lowest level of the WBS is 
defined as the work package level. 

The number of levels for a WBS varies from program to program and 
depends on a program’s complexity and risk. Work breakdown structures 
need to be expanded to a level of detail that is sufficient for 
planning and successfully managing the full scope of work. However, 
each WBS should, at the very least, include three levels. The first 
level represents the program as a whole and therefore contains only one 
element—the program’s name. The second level contains the major program 
segments, and level three contains the lower-level components or 
subsystems for each segment. These relationships are illustrated in 
figure 10, which depicts a very simple automobile system WBS. 

Figure 10: A Product-Oriented Work Breakdown Structure: 

[Refer to PDF for image: illustration] 
 
Level 1: 
Automobile system. 

Level 2: 
Chassis;
Shell;
Interior; 
Exterior; 
Powertrain. 

Level 3: 
Subcomponent; 
Subcomponent; 
Subcomponent. 

Source: © 2005 MCR, LLC, “Developing a Work Breakdown Structure.” 

[End of figure] 

In figure 10, all level 2 elements would also have level 3 
subcomponents; chassis is the example in the figure. For some level 2 
elements, level 3 would be the lowest level of breakdown; for others, 
still lower levels would be required. The elements at each lower level 
of breakdown are called “children” of the next higher level, which are 
the “parents.” The parent–child relationship allows for logical 
connections and relationships to emerge and a better understanding of 
the technical effort involved. It also helps improve the ability to 
trace relationships within the cost estimate and EVM system. 

In the example in figure 10, the chassis would be a child of the 
automobile system but the parent of subcomponents 1–3. In constructing 
a WBS, the 100 percent rule always applies. That is, the sum of a 
parent’s children must always equal the parent. Thus, in figure 10, the 
sum of chassis, shell, interior, and so on must equal the automobile 
system. In this way, the WBS makes sure that each element is defined 
and related to only one work effort, so that all activities are 
included and accounted for. It also helps identify the specialists who 
are needed to complete the work and who will be responsible so that 
effort is not duplicated. 

It is important to note that a product-oriented WBS reflects cost, 
schedule, and technical performance on specific portions of a program, 
while a functional WBS does not provide that level of detail. For 
example, an overrun on a specific item in figure 10 (for example, 
powertrain) might cause program management to change a specification, 
shift funds, or modify the design. If the WBS were functionally based 
(for example, in manufacturing, engineering, or quality control), then 
management would not have the right information to get to the root 
cause of the problem. Therefore, since only a product-oriented WBS 
relates costs to specific hardware elements—the basis of most cost 
estimates—it represents a cost estimating best practice. Case study 22 
highlights problems that can occur by not following this best practice. 


Case Study 22: Product-Oriented Work Breakdown Structure, from Air 
Traffic Control, GAO-08-756: 

Federal Aviation Administration (FAA) required the use of EVM on its 
major information technology investments. GAO found key components not 
fully consistent with best practices. We reported that leading 
organizations establish EVM policies that require programs to use a 
product-oriented structure for defining work products. FAA’s policy and 
guidance are not consistent with best practices because it requires its 
programs to establish a standard WBS using a function-oriented 
structure. FAA work is thus delineated by functional activities, such 
as design engineering, requirements analysis, and quality control. A 
product-oriented WBS would reflect cost, schedule, and technical 
performance on specific deliverables. 

Without a product-oriented approach, program managers may not have the 
detailed information needed to make decisions on specific program 
components. For example, cost overruns associated with a specific radar 
component could be quickly identified and addressed using a product-
oriented structure. If a function-oriented structure were used, these 
costs could be spread out over design, engineering, etc. 

FAA program managers using a product-oriented WBS need to transfer 
their data to FAA’s required function-oriented WBS when reporting to 
management. EVM experts agree that such mapping efforts are time-
consuming, subject to error, and not always consistent. Until FAA 
establishes a standard product-oriented WBS, program officials may not 
be obtaining the information they need. 

Source: GAO, Air Traffic Control: FAA Uses Earned Value Techniques to 
Help Manage Information Technology Acquisitions, but Needs to Clarify 
Policy and Strengthen Oversight, GAO-08-756, Washington, D.C.: July 18, 
2008. 

[End of case study] 

Since best practice is for the WBS prime mission elements to be product-
oriented, the WBS should not be structured or organized at a second or 
third level according to any element not a product or not being in or 
itself a deliverable: 

* design engineering, requirements analysis, logistics, risk, quality 
assurance, and test engineering (all functional engineering efforts), 
aluminum stock (a material resource), and direct costs (an accounting 
classification);[Footnote 32] 
 
* program acquisition phases (for example, development and procurement) 
and types of funds used in those phases (for example, research, 
development, test, and evaluation); 
 
* rework, retesting, and refurbishing, which should be treated as 
activities of the WBS element; 
 
* nonrecurring and recurring classifications, for which reporting 
requirements should be structured to ensure that they are segregated; 
 
* cost saving efforts—such as total quality management initiatives and 
acquisition reform initiatives—included in the elements they affect, 
not captured separately;

* the organizational structure of the program office or contractor; 

* the program schedule—instead the WBS will drive the necessary 
schedule activities; 
 
* meetings, travel, and computer support, which should be included in 
the WBS elements they are associated with; 
 
* generic terms (terms for WBS elements should be as specific as 
possible); and; 

* tooling, which should be included with the equipment being produced. 

While functional activities are necessary for supporting a product’s 
development, the WBS should not be organized around them. Only products 
should drive the WBS, not common support activities. Moreover, the WBS 
dictionary should state where the functional elements fall within the 
products and how the statement of work elements come together to make 
specific products. 

Common WBS Elements: 

In addition to including product-oriented elements, every WBS includes 
program management as a level 2 element and other common elements like 
integration and assembly, government furnished equipment, and 
government testing. Table 8 lists and describes common elements that 
support the program. For instance, systems engineering, program 
management, integration, and testing are necessary support functions 
for developing, testing, producing, and fielding hardware or software 
elements. 

Table 8: Common Elements in Work Breakdown Structures: 

Common element: Integration, assembly, test, and checkout; 
Description: All effort of technical and functional activities 
associated with the design, development, and production of mating 
surfaces, structures, equipment, parts, materials, and software 
required to assemble level 3 equipment (hardware and software) elements 
into level 2 mission equipment (hardware and software) 

Common element: System engineering; 
Description: The technical and management efforts of directing and 
controlling a totally integrated engineering effort of a system or 
program. 

Common element: Program management; 
Description: The business and administrative planning, organizing, 
directing, coordinating, controlling, and approval actions designated 
to accomplish overall program objectives not associated with specific 
hardware elements and not included in systems engineering. 

Common element: Training; 
Description: Deliverable training services, devices, accessories, aids, 
equipment, and parts used to facilitate instruction in which personnel 
will learn to operate and maintain the system with maximum efficiency. 

Common element: Data; 
Description: The deliverable data that must be on a contract data 
requirements list, including technical publications, engineering data, 
support data, and management data needed to configure management, cost, 
schedule, contractual data management, and program management. 

Common element: System test and evaluation; 
Description: The use of prototype, production, or specifically 
fabricated hardware and software to obtain or validate engineering data 
on the performance of the system in developing program (in DOD, 
normally funded from research, development, test, and evaluation 
appropriations); also includes all effort associated with design and 
production of models, specimens, fixtures, and instrumentation in 
support of the system-level test program. 

Common element: Peculiar support equipment; 
Description: Equipment uniquely needed to support the program: 
vehicles, equipment, tools, and the like to fuel, service, transport, 
hoist, repair, overhaul, assemble and disassemble, test, inspect, or 
otherwise maintain mission equipment, as well as equipment or software 
required to maintain or modify the software portions of the system. 

Common element: Common support equipment;
Description: Equipment not unique to the program and available in 
inventory for use by many programs. 

Common element: Operational and site activation; 
Description: Installation of mission and support equipment in the 
operations or support facilities and complete system checkout or 
shakedown to ensure operational status; may include real estate, 
construction, conversion, utilities, and equipment to provide all 
facilities needed to house, service, and launch prime mission 
equipment. 

Common element: Facilities; 
Description: Includes construction, conversion, or expansion of 
existing industrial facilities for production, inventory, and 
contractor depot maintenance required as a result of the specific 
system. 

Common element: Initial spares and repair parts; 
Description: Includes the deliverable spare components, assemblies, and 
subassemblies used for initial replacement purposes in the materiel 
system equipment end item. 

Source: DOD. 

[End of table] 
 
Therefore, in addition to having a product-oriented WBS for the prime 
mission equipment that breaks down the physical pieces of, for example, 
an aircraft, information technology system, or satellite, the WBS 
should include these common elements to ensure that all effort is 
identified at the outset. This, in turn, will facilitate planning and 
managing the overall effort, since the WBS should be the starting point 
for developing the detailed schedule. Figure 11 shows a program WBS, 
including common elements, for an aircraft system. 

Figure 11: A Work Breakdown Structure with Common Elements: 

[Refer to PDF for image: illustration] 

Level 1: 
Aircraft system; 

Level 2: 
Air vehicle; 
System engineering/Program management; 
System test and evaluation; 
Data; 
Training. 

Level 3: 
Airframe; 
Propulsion;
Fire control.
 
Source: © 2005 MCR, LLC, “Developing a Work Breakdown Structure.” 

[End of figure] 

While the top-level WBS encompasses the whole program, the contractor 
must also develop a contract WBS that extends the lower-level 
components to reflect its responsibilities. See figure 12. 

Figure 12: A Contract Work Breakdown Structure: 

[Refer to PDF for image: illustration] 

Source: DOD. 

[End of figure] 

Figure 12 shows how a prime contractor may require its subcontractor to 
use the WBS to report work progress. In this example, the fire control 
effort (a level 3 element in the prime contractor’s WBS) is the first 
level for the subcontractor. Thus, all fire control expenditures at 
level 1 of the subcontractor’s contract WBS would map to the fire 
control element at level 3 in the program WBS. This shows how a 
subcontractor would break a level 3 item down to lower levels to 
accomplish the work, which when rolled up to the prime WBS, would show 
effort at levels 4–7. Always keep in mind that the structure provided 
by the prime contractor WBS will identify the work packages that are 
the responsibility of the subcontractor. The subcontractor will also 
need to decompose the work further in its own WBS as well. 

WBS Development: 

A WBS should be developed early to provide for a conceptual idea of 
program size and scope. As the program matures, so should the WBS. Like 
the technical baseline, the WBS should be considered a living document. 
Therefore, as the technical baseline becomes further defined with time, 
the WBS will also reflect more detail. For example, as specification 
requirements become better known and the statement of work is updated, 
the WBS will include more elements. As more elements are added to the 
WBS, the schedule is capable of greater definition, giving more insight 
into the program’s cost, schedule, and technical relationships. 

It is important that each WBS be accompanied by a dictionary of the 
various WBS elements and their hierarchical relationships. A WBS 
dictionary is simply a document that describes in brief narrative 
format what work is to be performed in each WBS element. Each element 
is presented in an outline to show how it relates to the next higher 
element and what is included to ensure clear relationships. With minor 
changes and additions the WBS dictionary can be converted into a 
statement of work. Although not the normal approach, the dictionary may 
also be expanded by the program manager to describe the resources and 
processes necessary for producing each element in cost, technical, and 
schedule terms. Also, since the WBS is product related, it is closely 
related to, and structured somewhat the same as, an indented bill of 
materials for the primary product. Like the WBS, its dictionary should 
be updated when changes occur. After the program is baselined, updating 
the WBS should be part of a formal process, as in configuration 
management. 

Standardized WBS: 

Standardizing the WBS is considered a best practice because it enables 
an organization to collect and share data among programs. Standardizing 
work breakdown structures results in more consistent cost estimates, 
allows data to be shared across organizations, and leads to more 
efficient program execution. WBS standardization also facilitates cost 
estimating relationship development and allows for common cost measures 
across multiple contractors and programs. Not standardizing WBSs causes 
extreme difficulty in comparing costs from one contractor or program to 
another, resulting in substantial expense to government estimating 
agencies when collecting and reconciling contractor cost and technical 
data in consistent format. 

The standardized WBS logic should support the engineering perspective 
on how the program is being built. The WBS should be a communication 
tool that can be used across all functions within the program. To 
foster flexibility, WBS standardization should occur at a high 
level—such as WBS level 3—so that lower levels can be customized to 
reflect how the specific program’s work will be managed. For high-risk 
or costly elements, however, management can make decisions to 
standardize the WBS to whatever level is necessary to properly gain 
insight. Thus, the WBS should be standard at a high level, with 
flexibility in the lower levels to allow detailed planning once the 
schedule is laid out. Furthermore, the same standard WBS should be used 
for developing the cost estimate and the program schedule and setting 
up the EVM performance measurement baseline. Relying on a standard WBS 
can enable program managers to better plan and manage their work and 
helps in updating the cost estimate with actual costs—the final 
critical step in our twelve steps to a high-quality cost estimate. 

A standardized product-oriented WBS can help define high-level 
milestones and cost driver relationships that can be repeated in future 
applications. In addition to helping the cost community, standard WBSs 
can result in better portfolio management. Programs reporting to a 
standard WBS enable leadership to make better decisions about where to 
apply contingency reserve and where systemic problems are occurring, 
like integration and test. Using this information, management can take 
action by adjusting investment and obtaining lessons learned. As a 
result, it is easier to manage programs if they are reporting in the 
same format. 

Besides the common elements shown in table 8, DOD has identified, for 
each defense system, a standard combination of hardware and software 
that defines the end product for that system. In its 2005 updated WBS 
handbook, DOD defined and described the WBS, provided instructions on 
how to develop one, and defined specific defense items.[Footnote 33] 
The primary purpose of the handbook is to develop the top levels of the 
WBS with uniform definitions and a consistent approach. Developed 
through the cooperation of the military services, with assistance from 
industry associations, its benefit is improved communication throughout 
the acquisition process. 

In addition to defining a standard WBS for its weapon systems, DOD has 
developed a common cost element structure that, while not a product-
oriented WBS, standardizes the vocabulary for cost elements for 
automated information systems undergoing DOD review.[Footnote 34] The 
cost element structure is also designed to standardize the systems, 
facilitating the validation process. Furthermore, DOD requires that all 
the cost elements be included in LCCEs for automated information 
systems submitted for review. Table 9 gives an example of the cost 
element structure for an automated information system. 

Table 9: Cost Element Structure for a Standard DOD Automated 
Information System: 

Element 1 and subelements: 
1.0 Investment: 
1.1 Program management;
1.1.1 Personnel;
1.1.2 Travel;
1.1.3 Other government support;
1.1.4 Other;
1.2 Concept exploration; 
1.2.1 Engineering analysis investment & specification;
1.2.2 Concept exploration hardware;
1.2.3 Concept exploration software;
1.2.4 Concept exploration data; 
1.2.5 Exploration documentation; 
1.2.6 Concept exploration testing; 
1.2.7 Facilities;
1.2.8 Other;
1.3 System development;
1.3.1 System design & specification; 
1.3.2 Prototype & test site investment; 
1.4 System procurement; 
1.4.1 Deployment hardware; 
1.4.2 System deployment software; 
1.4.3 Initial documentation; 
1.4.4 Logistics support equipment; 
1.4.5 Initial spares; 
1.4.6 Warranties; 
1.5 Outsource investment; 
1.5.1 Capital investment; 
1.5.2 Software development; 
1.5.3 System user investment; 
1.6 System implementation; 
1.6.1 Training; 
1.6.2 Integration, test, acceptance;
1.6.3 Common support equipment;
1.6.4 Site activation & facilities;
1.6.5 Initial supplies;
1.6.6 Engineering change;
1.6.7 Initial logistics support;
1.6.8 Office furniture & furnishings;
1.6.9 Data upload & transition;
1.6.10 Communication;s
1.6.11 Other;
1.7 Upgrades;
1.7.1 Upgrade development;
1.7.2 Life cycle upgrades;
1.7.3 Central mega center upgrades;
1.8 Disposal & reuse;
1.8.1 Capital recoupment;
1.8.2 Retirement;
1.8.3 Environmental & hazardous,

Element 2 and subelements: 
2.0 System operations & support; 
2.1 System management;
2.1.1 Personnel;
2.1.2 Travel;
2.1.3 Other government support;
2.1.4 Other; 
2.2 Annual operations; 
2.2.1 Maintenance investment; 
2.2.2 Replenishment spares; 
2.2.3 Replenishment supplies; 
2.3 Hardware maintenance; 
2.3.1 Hardware maintenance; 
2.3.2 Maintenance support; 
2.3.3 Other hardware maintenance; 
2.4 Software maintenance; 
2.4.1 Commercial off-the-shelf software; 
2.4.2 Application & mission software;
2.4.3 Communication software; 
2.5 Megacenter maintenance; 
2.6 Data maintenance; 
2.6.1 Mission application data; 
2.6.2 Standard administrative data; 
2.7 Site operations; 
2.7.1 System operational personnel; 
2.7.2 Utility requirement; 
2.7.3 Fuel 
2.7.4 Facilities lease & maintenance; 
2.7.5 Communications; 
2.7.6 Base operating & support; 
2.7.7 Recurring training & fielding; 
2.7.8 Miscellaneous support; 
2.8 Environmental & acceptance hazardous; 
2.9 Contract leasing.

Element 3 and subelements: 
3.0 Legacy system phase-out; 
3.1 System management; 
3.1.1 Personnel; 
3.1.2 Travel; 
3.1.3 Other government support;
3.2 Phase-out investment; 
3.2.1 Hardware; 
3.2.2 Software; 
3.2.3 Hazardous material handling; 
3.3 Phase-out operations & support; 
3.3.1 Hardware maintenance;
3.3.2 Software maintenance; 
3.3.3 Unit & subunit operations; 
3.3.4 Megacenter operations; 
3.3.5 Phase-out contracts. 

Source: DOD. 

[End of table] 

This standard WBS should be tailored to fit each program. In some 
cases, the cost element structure contains built-in redundancies that 
provide flexibility in accounting for costs. For example, logistics 
support costs could occur in either investment or operations and 
support. However, it is important that the cost element structure of 
the automated information system not double count costs that could be 
included in more than one cost element. While the structure is 
flexible, the same rules as those of a WBS apply, in that children are 
assigned to only one parent. (Appendix IX contains numerous examples 
of standard work breakdown structures for, among others, surface, sea, 
and air transportation systems; military systems; communications 
systems; and systems for construction and utilities.) 

WBS And Scheduling: 

The WBS should be used as the outline for the integrated master 
schedule, using the levels of indenture down to the work package level. 
Since the WBS defines the work in lower levels of detail, its framework 
provides the starting point for defining all activities and tasks that 
will be used to develop the program schedule. 

The lowest level of the WBS is the work package. Within the work 
packages, the activities are defined and scheduled. When developing the 
program schedule, the WBS—in outline form—should be simply cut and 
pasted into the software. From there, the lower-level work packages and 
subsequent activities and tasks are defined. 

Accordingly, the WBS provides a logical and orderly way to begin 
preparing the detailed schedule, determining the relationships between 
activities, and identifying resources required to accomplish the 
tasks. Therefore, high-level summary tasks and all the detailed tasks 
in the schedule should map directly to the WBS to ensure that the 
schedule encompasses the entire work effort. 

WBS and EVM: 

By breaking the work into smaller, more manageable work elements, a WBS 
can be used to integrate the scheduled activities and costs for 
accomplishing each work package at the lowest level of the WBS. This is 
essential for developing the resource-loaded schedule that forms the 
foundation for the EVM performance measurement baseline. Thus, a WBS is 
an essential part of EVM cost, schedule, and technical monitoring, 
because it provides a consistent framework from which to measure 
progress. This framework can be used to monitor and control costs based 
on the original baseline and to track where and why there were 
differences. In this way, the WBS serves as the common framework for 
analyzing the original cost estimate and the final cost outcome. 

When analysts use cost, schedule, and technical information organized 
by the WBS hierarchical structure, they can summarize data to provide 
management valuable information at any phase of the program. 
Furthermore, because a WBS addresses the entire program, managers at 
any level can assess their progress against the cost estimate plan. 
This helps keep program status current and visible so that risks can be 
managed or mitigated quickly. Without a WBS, it would be much more 
difficult to analyze the root cause of cost, schedule, and technical 
problems and to choose the optimum solution to fix them. 

The WBS also provides a common thread between EVM and the integrated 
master schedule (IMS)—the time-phased schedule DOD and other agencies 
use for assessing technical performance. This link to the WBS can allow 
for further understanding of program cost and schedule variances. When 
the work is broken down into small pieces, progress can be linked to 
the IMS for better assessments of cost, technical, schedule, and 
performance issues. The WBS also enhances project control by tying the 
contractual work scope to the IMS, which DOD commonly uses to develop a 
program’s technical goals and plans. 

WBS And Risk Management: 

The WBS is also valuable for identifying and monitoring risks. During 
the cost estimating phase, the WBS is used to flag elements likely to 
encounter risks, allowing for better contingency planning. During 
program execution, the WBS is used to monitor risks using the EVM 
system, which details plans to a level that is needed to accomplish all 
tasks.

In scheduling the work, the WBS can help identify activities in the 
schedule that are at risk because resources are lacking or because too 
many activities are planned in parallel to one another. In addition, 
risk items can be mapped to activities in the schedule and the results 
can be examined through a schedule risk analysis (more detail is in 
appendix X). 

WBS Benefits: 

Elements of a WBS may vary by phase, since different activities are 
required for development, production, operations, and support. 
Establishing a master WBS as soon as possible for the program’s life 
cycle that details the WBS for each phase provides many program 
benefits: 
 
* segregating work elements into their component parts; 

* clarifying relationships between the parts, the end product, and the 
tasks to be completed; 

* facilitating effective planning and assignment of management and 
technical responsibilities; 

* helping track the status of technical efforts, risks, resource 
allocations, expenditures, and the cost and schedule of technical 
performance within the appropriate phases, since the work in phases 
frequently overlaps; 

* helping ensure that contractors are not unnecessarily constrained in 
meeting item requirements; and; 

* providing a common basis and framework for the EVM system and the 
IMS, facilitating consistency in understanding program cost and 
schedule performance and assigning to the appropriate phase. Since the 
link between the requirements, WBS, the statement of work, IMS, and the 
integrated master plan provides specific insights into the relationship 
between cost, schedule, and performance, all items can be tracked to 
the same WBS elements. 

As the program or system matures, engineering efforts should focus on 
system-level performance requirements—validating critical technologies 
and processes and developing top-level specifications. As the 
specifications are further defined, the WBS will better define the 
system in terms of its specifications. After the system concept has 
been determined, major subsystems can be identified and lower-level 
functions determined, so that lower-level system elements can be 
defined, eventually completing the total system definition. The same 
WBS can be used throughout, updating and revising it as the program or 
system development proceeds and as the work in each phase progresses. 
One of the outputs of each phase is an updated WBS covering the 
succeeding phases. 

In summary, a well-developed WBS is essential to the success of all 
acquisition programs. A comprehensive WBS provides a consistent and 
visible framework that improves communication; helps in the planning 
and assignment of management and technical responsibilities; and 
facilitates tracking engineering efforts, resource allocations, cost 
estimates, expenditures, and cost and technical performance. Without 
one, a program is most likely to encounter problems, as case studies 23 
and 24 illustrate. 

Case Study 23: Developing Work Breakdown Structure, from NASA, GAO-04-
642: 
 
For more than a decade, GAO had identified NASA’s contract management 
as a high-risk area. NASA had been unable to collect, maintain, and 
report the full cost of its programs and projects. Because of 
persistent cost growth in a number of NASA programs, GAO was asked to 
assess 27 programs—10 in detail. GAO found that only 3 of the 10 had 
provided a complete breakdown of the work to be performed, despite 
agency guidance calling for projects to break down the work into 
smaller units to facilitate cost estimating and program management and 
to help ensure that relevant costs were not omitted. Underestimating 
full life-cycle costs creates the risk that a program may be 
underfunded and subject to major cost overruns. It may be reduced in 
scope, or additional funding may have to be appropriated to meet 
objectives. Overestimating life-cycle costs creates the risk that a 
program will be thought unaffordable and it could go unfunded. Without 
a complete WBS, NASA’s programs cannot ensure that its LCCEs capture 
all relevant costs, which can mean cost overruns. Inconsistent WBS 
estimates across programs can cause double counting or, worse, costs 
can be underestimated when historical program costs are used for 
projecting future costs for similar programs. Among its multiple 
recommendations, GAO recommended that NASA base its cost estimates for 
each program on a WBS that encompassed both in-house and contractor 
efforts and develop procedures that would prohibit proposed projects 
from proceeding through review and approval if they did not address the 
elements of recommended cost estimating practices. 

Source: GAO, NASA: Lack of Disciplined Cost-Estimating Processes 
Hinders Effective Program Management, GAO-04-642, Washington, D.C.: 
May 28, 2004. 

[End of case study] 

Case Study 24: Developing Work Breakdown Structure, from 
Homeland Security, GAO-06-296: 

The Department of Homeland Security (DHS) established U.S. Visitor and 
Immigrant Status Indicator Technology (US–VISIT) to collect, maintain, 
and share information, including biometric identifiers, on selected 
foreign nationals entering and exiting the United States. Having 
reported that the program had not followed effective cost estimating 
practices, GAO recommended that DHS follow effective practices for 
estimating future increments. 

GAO then reported on the cost estimates for the latest increment in 
February 2006, finding US–VISIT’s cost estimates still insufficient. 
For example, they did not include a detailed WBS and they omitted 
important cost elements such as system testing. The uncertainties 
associated with the latest system increment cost estimate were not 
identified. Uncertainty analysis provides the basis for adjusting 
estimates to reflect unknown facts and circumstances that could affect 
costs, and it identifies risk associated with the cost estimate. 

Program officials stated that they recognized the importance of 
developing reliable cost estimates and initiated actions to more 
reliably estimate the costs of future system increments. For example, 
US–VISIT chartered a cost analysis process action team to develop, 
document, and implement a cost analysis policy, process, and plan for 
the program. Program officials had also hired additional contracting 
staff with cost estimating experience. 

Source: GAO, Homeland Security: Recommendations to Improve Management 
of Key Border Security Program Need to Be Implemented, GAO-06-296, 
Washington, 
D.C.: Feb. 14, 2006). 

[End of case study] 

5. Best Practices Checklist: Work Breakdown Structure: 

* A product-oriented WBS represents best practice. 

* It reflects the program work that needs to be done. It: 
- clearly outlines the end product and major work for the program; 
- contains at least 3 levels of indenture; 
- is flexible and tailored to the program. 

* The 100 percent rule applies: the sum of the children equals the 
parent. 
- The WBS defines all work packages, which in turn include all cost 
elements and deliverables. 
- In addition to hardware and software elements, the WBS contains 
program management and other common elements to make sure all the work 
is covered. 

* Each system has one program WBS but may have several contract WBSs 
that are extended from the program WBS, depending on the number of 
subcontractors. 

* The WBS is standardized so that cost data can be collected and used 
for estimating future programs. It: 
- facilitates portfolio management, including lessons learned; 
- matches schedule, cost estimate, and EVM at a high level; 
- is updated as changes occur and the program becomes better defined; 
- includes functional activities within each element that are needed to 
support each product deliverable; 
- is the starting point for developing the program’s detailed schedule; 
- provides a framework for identifying and monitoring risks and the 
effectiveness of contingency plans; 
- provides for a common language between the government program 
management office, technical specialists, prime contractors, and 
subcontractors. 

* The WBS has a dictionary that: 
- defines each element and how it relates to others in the hierarchy; 
- clearly describes what is included in each element; 
- describes resources and functional activities needed to produce the 
element product;
- links each element to other relevant technical documents. 

[End of Chapter 8] 

Chapter 9: Ground Rules and Assumptions: 

Cost estimates are typically based on limited information and therefore 
need to be bound by the constraints that make estimating possible. 
These constraints usually take the form of assumptions that bind the 
estimate’s scope, establishing baseline conditions the estimate will be 
built from. Because of the many unknowns, cost analysts must create a 
series of statements that define the conditions the estimate is to be 
based on. These statements are usually made in the form of ground rules 
and assumptions (GR&A). By reviewing the technical baseline and 
discussing the GR&As with customers early in the cost estimating 
process, analysts can flush out any potential misunderstandings. GR&As: 

* satisfy requirements for key program decision points, 

* answer detailed and probing questions from oversight groups, 

* help make the estimate complete and professional, 

* present a convincing picture to people who might be skeptical, 

* provide useful estimating data and techniques to other cost 
estimators, 

* provide for reconstruction of the estimate when the original 
estimators are no longer available, and, 

* provide a basis for the cost estimate that documents areas of 
potential risk to be resolved. 

Ground Rules: 
 
Ground rules and assumptions, often grouped together, are distinct. 
Ground rules represent a common set of agreed on estimating standards 
that provide guidance and minimize conflicts in definitions. When 
conditions are directed, they become the ground rules by which the team 
will conduct the estimate. The technical baseline requirements 
discussed in chapter 7 represent cost estimate ground rules. Therefore, 
a comprehensive technical baseline provides the analyst with all the 
necessary ground rules for conducting the estimate. 

Assumptions: 

Without firm ground rules, the analyst is responsible for making 
assumptions that allow the estimate to proceed. In other words, 
assumptions are required only where no ground rules have been provided. 
Assumptions represent a set of judgments about past, present, or future 
conditions postulated as true in the absence of positive proof. The 
analyst must ensure that assumptions are not arbitrary, that they are 
founded on expert judgments rendered by experienced program and 
technical personnel. Many assumptions profoundly influence cost; the 
subsequent rejection of even a single assumption by management could 
invalidate many aspects of the estimate. Therefore, it is imperative 
that cost estimators brief management and document all assumptions 
well, so that management fully understands the conditions the estimate 
was structured on. Failing to do so can lead to overly optimistic 
assumptions that heavily influence the overall cost estimate, to cost 
overruns, and to inaccurate estimates and budgets. (See case study 25.) 
 
Case Study 25: The Importance of Assumptions, from Space 
Acquisitions, GAO-07-96: 

Estimated costs for DOD’s major space acquisition programs increased 
about $12.2 billion, nearly 44 percent, above initial estimates for 
fiscal years 2006 through 2011. Such growth has had a dramatic effect 
on DOD’s overall space portfolio. To cover the added costs of poorly 
performing programs, DOD shifted scarce resources from other programs, 
creating a cascade of cost and schedule inefficiencies. 

GAO’s case study analyses found that program office cost estimates—
specifically, assumptions they were based on—were unrealistic in eight 
areas, many interrelated. In some cases, such as assumptions regarding 
weight growth and the ability to gain leverage from legacy systems, 
past experiences or contrary data were ignored. In others, such as when 
contractors were given more program management responsibility or when 
growth in the commercial market was predicted, estimators assumed that 
promises of reduced cost and schedule would be borne out but did not 
have the benefit of experience to factor them into their work. 

GAO also identified flawed assumptions that reflected deeper flaws in 
acquisition strategies or development approaches. For example, five of 
six programs GAO reviewed assumed that technologies would be 
sufficiently mature when needed, even though they began without a 
complete understanding of how long it would take or how much it would 
cost to ensure that they could work as intended. In four programs, 
estimators assumed few delays, even though the programs adopted highly 
aggressive schedules while attempting to make ambitious leaps in 
capability. In four programs, estimators assumed funding would stay 
constant, even though space and weapons programs frequently experienced 
funding shifts and the Air Force was in the midst of starting a number 
of costly new space programs to replenish older ones. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Action to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Global And Element-Specific Ground Rules And Assumptions: 

GR&As are either global or element specific. Global GR&As apply to the 
entire estimate; element-specific GR&As are driven by each WBS 
element’s detailed requirements. GR&As are more pronounced for 
estimates in the development phase, where there are more unknowns; they 
become less prominent as the program moves through development into 
production. 

While each program has a unique set of GR&As, some are general enough 
that each estimate should address them. For example, each estimate 
should at a minimum define the following global GR&As: program 
schedule, cost limitations (for example, unstable funding stream or 
staff constraints), high-level time phasing, base year, labor rates, 
inflation indexes, participating agency support, and government 
furnished equipment.[Footnote 35] 

One of the most important GR&As is to define a realistic schedule. It 
may be difficult to perform an in-depth schedule assessment early to 
uncover the frequent optimism in initial program schedules. Ideally, 
members from manufacturing and the technical community should be 
involved in developing the program schedule, but often information is 
insufficient and assumptions must be made. In this case, it is 
important that this GR&A outline the confidence the team has in the 
ability to achieve the schedule so that it can be documented and 
presented to management. 

One major challenge in setting realistic schedules is that the 
completion date is often set by external factors outside the control of 
the program office before any analysis has been performed to determine 
whether it is feasible. Another predominant problem is that schedule 
risk is often ignored or not analyzed—or when it is analyzed, the 
analysis is biased. This can occur on the government (customer) or 
contractor side or both. Risk analysis conducted by a group independent 
of the project manager has a better chance of being unbiased than one 
conducted by the program manager. However, it should also be noted that 
many organizations are not mature enough to acknowledge or to apply 
program schedule or cost risk realism because of the possible 
repercussions. For example, a contractor may be less likely to identify 
schedule or cost risk if it fears negative reaction from the customer. 
Likewise, the customer may be unwilling to report cost or schedule risk 
from fear that the program could be canceled. 

Sometimes, management imposes cost limitations because of budget 
constraints. The GR&A should then clearly explain the limitation and 
how it affects the estimate. Usually, cost limitations are handled by 
delaying program content or by a funding shortfall if program content 
cannot be delayed. In many cases, such actions will both delay the 
program and increase its final delivered cost. Either way, management 
needs to be fully apprised of how this GR&A affects the estimate. 

Estimates are time phased because program costs usually span many 
years. Time phasing spreads a program’s expected costs over the years 
in which they are anticipated to aid in developing a proper budget. 
Depending on the activities in the schedule for each year, some years 
may have more costs than others. Great peaks or valleys in annual 
funding should be investigated and explained, however, since staffing 
is difficult to manage with such variations from one year to another. 
Anomalies are easily discovered when the estimate is time phased. Cost 
limitations can also affect an estimate’s time phasing, if there are 
budget constraints for a given fiscal year. Additionally, changes in 
program priority will affect funding and timing—often a program starts 
with high priority but that priority erodes as it proceeds, causing 
original plans to be modified and resulting in later delivery and 
higher cost to the government. These conditions should be addressed by 
the estimate and their effects adequately explained. 

The base year is used as a constant dollar reference point to track 
program cost growth. Expressing an estimate in base year dollars 
removes the effects of economic inflation and allows for comparing 
separate estimates “apples to apples.” Thus, a global ground rule is to 
define the base year dollars that the estimate will be presented in and 
the inflation index that will be used to convert the base year costs 
into then-year dollars that include inflation. At a minimum, the 
inflation index, source, and approval authority should be clearly 
explained in the estimate documentation. Escalation rates should be 
standardized across similar programs, since they are all conducted in 
the same economic environment, and priority choices between them should 
not hinge on different assumptions about what is essentially an 
economic scenario common to all programs. 

Some programs result from two or more agencies joining together to 
achieve common program goals. When this happens, agreements should lay 
out each agency’s area of responsibility. An agency’s failing to meet 
its responsibility could affect the program’s cost and schedule. In the 
GR&A section, these conditions should be highlighted to ensure that 
management is firmly aware that the success of the estimate depends on 
the participation of other agencies. 

Equipment that the government agrees to provide to a contractor can 
range from common supply items to complex electronic components to 
newly developed engines for aircraft. Because the estimator cannot 
predict whether deliveries of such equipment will be timely, 
assumptions are usually made that it will be available when needed. It 
is important that the estimate reflect the items that it assumes 
government will furnish, so that the risk to the estimate if items are 
delayed can be modeled and presented to management. In general, 
schedules represent delivery of material from external sources, 
including the government, with date-constrained milestones. A better 
approach is to include the supplier’s work to produce the product by a 
summary activity in the schedule, examine the possibility of delayed 
delivery, include that risk in a schedule risk analysis, and monitor 
the work of the supplier as the date approaches. 

In addition to global GR&As, estimate-specific GR&As should be tailored 
for each program, including: 

* life-cycle phases and operations concept; 

* maintenance concepts; 

* acquisition strategy, including competition, single or dual sourcing, 
and contract or incentive type; 

* industrial base viability; 

* quantities for development, production, and spare and repair parts; 

* use of existing facilities, including any modifications or new 
construction; 

* savings for new ways of doing business; 

* commonality or design inheritance assumptions; 

* technology assumptions and new technology to be developed; 

* technology refresh cycles; 

* security considerations that may affect cost; and 

* items specifically excluded from the estimate. 

The cost estimator should work with members from the technical 
community to tailor these specific GR&As to the program. Information 
from the technical baseline and WBS dictionary help determine some of 
these GR&As, like quantities and technology assumptions. The element-
specific GR&As carry the most risk and therefore should be checked for 
realism and should be well documented in order for the estimate to be 
considered credible. 

Assumptions, Sensitivity, And Risk Analysis: 

Every estimate is uncertain because of the assumptions that must be 
made about future projections. Sensitivity analysis that examines how 
changes to key assumptions and inputs affect the estimate helps 
mitigate uncertainty. Best practice cost models incorporate the ability 
to perform sensitivity analyses without altering the model so that the 
effect of varying inputs can be quickly determined (more information is 
in chapters 13 and 14). For example, suppose a decision maker 
challenges the assumption that 5 percent of the installed equipment 
will be needed for spares, asking that the factor be raised to 10 
percent. A sensitivity analysis would show the cost impact of this 
change. Because of the implications that GR&As can have when 
assumptions change, the cost estimator should always perform a 
sensitivity analysis that portrays the effects on the cost and schedule 
of an invalid assumption. Such analysis often provides management with 
an invaluable perspective on its decision making. 

In addition to sensitivity analysis, factors that will affect the 
program’s cost, schedule, or technical status should be clearly 
identified, including political, organizational, or business issues. 
Because assumptions themselves can vary, they should always be inputs 
to program risk analyses of cost and schedule. A typical approach to 
risk analysis emphasizes the breadth of factors that may be uncertain. 
In a risk identification exercise, the goal is to identify all 
potential risks stemming from a broad range of sources. A good starting 
point would be to examine the program’s risk management database to 
determine which WBS elements these risks could affect. Another option 
would be to examine risks identified during a program’s integrated 
baseline review—a risk based assessment of the program plan to see 
whether the requirements can be met within cost and schedule 
assumptions. Regardless of what method is used to identify risk, it is 
important that more than just cost, schedule, and technical risks are 
examined. For example, budget and funding risks, as well as risks 
associated with start-up activities, staffing, and organizational 
issues, should also be considered. Therefore, risks from all sources 
such as external, organizational, and even project management 
practices, in addition to the technical challenges, need to be 
addressed. 

Well-supported assumptions should include documentation of an 
assumption’s source and should discuss any weaknesses or risks. Solid 
assumptions are measurable and specific. For example, an assumption 
that states “transaction volume will average 500,000 per month and is 
expected to grow at an annual rate of 5 percent” is measurable and 
specific, while “transaction volumes will grow greatly over the next 5 
years” is not as helpful. By providing more detail, cost estimators can 
perform risk and sensitivity analysis to quantify the effects of 
changes in assumptions. 

Assumptions should be realistic and valid. This means that historical 
data should back them up to minimize uncertainty and risk. 
Understanding the level of certainty around an estimate is imperative 
to knowing whether to keep or discard an assumption. Assumptions tend 
to be less certain earlier in a program, and become more reliable as 
more information is known about them. A best practice is to place 
all assumptions in a single spreadsheet tab so that risk and 
sensitivity analysis can be performed efficiently and quickly. Explicit 
assumptions should be available, but assumptions are also sometimes 
implicit—implicit assumptions should be documented as well. 

Certain ground rules should always be tested for risk. For example, the 
effects of the program schedule’s slipping on both cost and schedule 
should always be modeled and the results presented to management. This 
is especially important if the schedule was known to be aggressive or 
was not assessed for realism. Too often, we have found that when 
schedules are compressed, for instance to satisfy a potential 
requirements gap, the optimism in the schedule does not hold and the 
result is greater costs and schedule delays. Case study 26 gives 
examples of what happens in such situations. 

Case Study 26: Testing Ground Rules for Risk, from Space Acquisitions, 
GAO-07-96: 

Advanced Extremely High Frequency Satellite Program. The first AEHF 
launch was originally scheduled for June 2006. In response to a 
potential gap in satellite coverage because of the launch failure of 
the third Milstar satellite, DOD accelerated the schedule by 18 months, 
aiming for December 2004. An unsolicited contractor proposal stated 
that it could meet this date, even though not all AEHF’s requirements 
had been fully determined. The program office thus knew that the 
proposed schedule was overly optimistic, but the decision was made at 
high levels in DOD to award the contract. DOD did not, however, commit 
the funding to support the activities and manpower needed to design and 
build the satellites more quickly. Funding issues further hampered 
development efforts, increased schedule delays, and contributed to cost 
increases. 

National Polar-orbiting Operational Environmental Satellite System. 
When the NPOESS estimate was developed, the system was expected to be 
heavier, require more power, and have more than twice as many sensors 
as heritage satellites. Yet the program office estimated that the new 
satellites would be developed, integrated, and tested in less time than 
heritage satellites. Independent cost estimators highlighted to the 
NPOESS program office that the proposed integration schedule was 
unrealistic, compared to historical satellite programs. Later, the CAIG 
cautioned the program office that the system integration assembly and 
test schedule were unrealistic and the assumptions used to develop the 
estimate were not credible. 

Space Based Infrared System High Program. The SBIRS schedule proposed 
in 1996 did not allow enough time for geosynchronous Earth orbit system 
integration. And it did not anticipate the program design and 
workmanship flaws that eventually cost the program considerable delays. 
The schedule was also optimistic with regard to ground software 
productivity and time needed to calibrate and assess satellite health. 
Delivery of highly elliptical orbit sensors was delayed by almost 3 
years, the launch of the first geosynchronous Earth orbit satellite by 
6 years. 

Wideband Gapfiller Satellites. The request for proposals specified that 
the available WGS budget was $750 million for three satellites and that 
the ground control system was to be delivered within 36 months. 
Competing contractors were asked to offer maximum capacity, coverage, 
and connectivity in a contract that would use existing commercial 
practices and technologies. However, greater design complexity and 
supplier quality issues caused the WGS schedule to stretch to 78 months 
for the first expected launch. DOD’s history had been 55–79 months to 
develop satellites similar to WGS, so that while DOD’s experience was 
within the expected range, the original 36-month schedule was 
unrealistic. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Action to 
Address Unrealistic Initial Cost Estimates of Space Systems; GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Above and beyond the program schedule, some programs can be affected by 
the viability of the industrial base. Case study 27 illustrates. 

Case Study 27: The Industrial Base, from Defense Acquisitions, GAO-05-
183: 

For the eight case study ships GAO examined, cost analysts relied on 
the actual cost of previously constructed ships, without adequately 
accounting for changes in the industrial base, ship design, or 
construction methods. Cost data available to Navy cost analysts were 
based on higher ship construction rates from the 1980s. These data were 
based on lower costs because of economies of scale, which did not 
reflect the lower procurement rates after 1989. 

According to the shipbuilder, material cost increases on the CVN 76 and 
CVN 77 in the Nimitz class of aircraft carriers could be attributed to 
a declining supplier base and commodity price increases. Both carriers’ 
material costs had been affected by more than a 15 percent increase in 
metals costs that in turn increased costs for associated components. 

Moreover, many of the materials used in the construction of aircraft 
carriers are highly specialized and unique—often produced by only one 
manufacturer. With fewer manufacturers competing in the market, the 
materials were highly susceptible to cost increases. 

After the Seawolf submarine program was cancelled and, over a period of 
6 years, submarine production had decreased from three to four 
submarines per year to one, many vendors left the nuclear submarine 
business to focus on more lucrative commercial product development. 
Prices for highly specialized material increased, since competition and 
business had diminished. 

For example, many vendors were reluctant to support the Virginia class 
submarine contract because costs associated with producing small 
quantities of highly specialized materials were not considered worth 
the investment—especially for equipment with no other military or 
commercial applications. 

Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, GAO-05-183, 
Washington, D.C.: Feb. 28, 2005. 

[End of case study] 

Another area in which assumptions tend to be optimistic is technology 
maturity. Having reviewed the experiences of DOD and commercial 
technology development, GAO has found that programs that relied on 
technologies that demonstrated a high level of maturity were in a 
better position to succeed than those that did not. Simply put, the 
more mature technology is at the start of a program, the more likely it 
is that the program will meet its objectives. 

Technologies that are not fully developed represent a significant 
challenge and add a high degree of risk to a program’s schedule and 
cost. Programs typically assume that the technology required will 
arrive on schedule and be available to support the effort. While this 
assumption allows the program to continue, the risk that it will prove 
inaccurate can greatly affect cost and schedule. Case studies 28 and 29 
provide examples. 

Case Study 28: Technology Maturity, from Defense Acquisitions, GAO-05-
183: 

The lack of design and technology maturity led to rework, increasing 
the number of labor hours for most of the case study ships. For 
example, the design of the LPD 17, in the San Antonio class of 
transports, continued to evolve even as construction proceeded. When 
construction began on the DDG 91 and DDG 92, in the Arleigh Burke class 
of destroyers—the first ships to incorporate the remote mine hunting 
system—the technology was still being developed. As a result, workers 
were required to rebuild completed ship areas to accommodate design 
changes. 

Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, GAO-05-183, 
Washington, D.C.: Feb. 28, 2005. 

[End of case study] 

Case Study 29: Technology Maturity, from Space Acquisitions, GAO-07-96: 

The Advanced Extremely High Frequency (AEHF) program of communications 
satellites faced several problems of technology maturity. They included 
developing a digital processing system that would support 10 times the 
capacity of Milstar’s medium data rate, the predecessor satellite, 
without self-interference and using phased array antennas at extremely 
high frequencies, which had never been done before. In addition, the 
change from a physical process to an electronic process for crypto 
rekeys had not been expected at the start of AEHF. Milstar had required 
approximately 2,400 crypto rekeys per month and had been done 
physically. AEHF’s proposed capability was approximately 100,000—too 
large for physical processing. Changing the rekeys to electronic 
processing was revolutionary and led to unexpected cost and schedule 
growth. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Action to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Cost estimators and auditors should not get trapped by overly 
optimistic technology forecasts. It is well known that program 
advocates tend to underestimate the technical challenge facing the 
development of a new system. Estimators and auditors alike should 
always seek to uncover the real risk by performing an uncertainty 
analysis. In doing so, it is imperative that cost estimators and 
auditors meet with engineers familiar with the program and its new 
technology to discuss the level of risk associated with the technical 
assumptions. Only then can they realistically model risk distributions 
using an uncertainty analysis and analyze how the results affect the 
overall cost estimate. 

Once the risk uncertainty and sensitivity analyses are complete, the 
cost estimator should formally convey the results of changing 
assumptions to management as early and as far up the line as possible. 
The estimator should also document all assumptions to help management 
understand the conditions the estimate was based on. When possible, the 
analyst should request an updated technical baseline in which the new 
assumptions have been incorporated as ground rules. Case study 30 
illustrates an instance of management’s not knowing the effects of 
changing assumptions. 

Case Study 30: Informing Management of Changed Assumptions, 
from Customs Service Modernization, GAO/AIMD-99-41: 

The Automated Commercial Environment (ACE) was a major U.S. Customs 
Service information technology system modernization effort. In November 
1997, it was estimated that ACE would cost $1.05 billion to develop, 
operate, and maintain between 1994 and 2008. GAO found that the agency 
lacked a reliable estimate of what ACE would cost to build, deploy, and 
maintain. 

The cost estimates were understated, benefit estimates were overstated, 
and both were unreliable. Customs’ August 1997 cost-benefit analysis 
estimated that ACE would produce cumulative savings of $1.9 billion 
over a 10-year period. The analysis identified $644 million in 
savings—33 percent of the total estimated savings—resulting from 
increased productivity. Because this estimate was driven by Customs’ 
assumption that every minute “saved” by processing transactions or 
analyzing data faster using ACE rather than its predecessor system 
would be productively used by all workers, it was viewed as a best 
case upper limit on estimated productivity improvements. 

Given the magnitude of the potential savings, even a small change in 
the assumption translated into a large reduction in benefits. For 
example, conservatively assuming that three-fourths of each minute 
saved would be used productively by three-fourths of all workers, the 
expected benefits would be reduced by about $282 million. Additionally, 
the analysis excluded costs for hardware and systems software upgrades 
at each port office. Using Customs’ estimate for acquiring the initial 
suite of port office hardware and systems software, and assuming a 
technology refreshment cycle of every 3 to 5 years, GAO estimated this 
cost at $72.9 million to $171.8 million. 

Because Customs did not have reliable information on ACE costs and 
benefits and had not analyzed viable alternatives, it did not have 
adequate assurance that ACE was the optimal approach. In fact, it had 
no assurance at all that ACE would be cost-effective. Furthermore, it 
had not justified the return on its investment in each ACE increment 
and therefore would not be able to demonstrate whether ACE would be 
cost-effective until it had spent hundreds of millions of dollars to 
acquire the entire system. 

GAO recommended that Customs rigorously analyze alternative approaches 
to building ACE and, for each increment, use disciplined processes to 
prepare a robust LCCE, prepare realistic and supportable benefit 
expectations, and validate actual costs and benefits once an increment 
had been piloted. 

Source: GAO, Customs Service Modernization: Serious Management and 
Technical Weaknesses Must Be Corrected, GAO/AMD-99-41, Washington, 
D.C.: Feb. 26, 1999. 

[End of case study] 

6. Best Practices Checklist: Ground Rules and Assumptions: 

* All ground rules and assumptions have been: 
- Developed by estimators with input from the technical community. 
- Based on information in the technical baseline and WBS dictionary. 
- Vetted and approved by upper management. 
- Documented to include the rationale behind the assumptions and 
historical data to back up any claims. 
- Accompanied by a level of risk of each assumption’s failing and its 
effect on the estimate. 

* To mitigate risk, 
- All GR&As have been placed in a single spreadsheet tab so that risk 
and sensitivity analysis can be performed quickly and efficiently. 
- All potential risks including cost, schedule, technical, and 
programmatic (e.g., risks associated with budget and funding, start up 
activities, staffing, and organizational issues) have been identified 
and traced to specific WBS elements. 
-- A schedule risk analysis has been performed to determine the 
program schedule’s realism. 
-- A cost risk analysis, incorporating the results of the schedule risk 
analysis, has been performed to determine the program’s cost estimate 
realism. 

* Budget constraints, as well as the effect of delaying program 
content, have been defined. 
- Peaks and valleys in time-phased budgets have been explained. 
- Inflation index, source, and approval authority have been identified. 
- Dependence on participating agencies, the availability of government 
furnished equipment, and the effects if these assumptions do not hold 
have been identified. 
- Items excluded from the estimate have been documented and 
explained.
- Technology was mature before it was included; if its maturity was 
assumed, the estimate addresses the effect of the assumption’s failure 
on cost and schedule. 

* Cost estimators and auditors met with technical staff to determine 
risk distributions for all assumptions; the distributions were used in 
sensitivity and uncertainty analyses of the effects of invalid 
assumptions. Management has been briefed, and the results have been 
documented. 

[End of Chapter 9] 

Chapter 10: Data: 

Data are the foundation of every cost estimate. How good the data are 
affects the estimate’s overall credibility. Depending on the data 
quality, an estimate can range anywhere from a mere guess to a highly 
defensible cost position. Credible cost estimates are rooted in 
historical data. Rather than starting from scratch, estimators usually 
develop estimates for new programs by relying on data from programs 
that already exist and adjusting for any differences. Thus, collecting 
valid and useful historical data is a key step in developing a sound 
cost estimate. The challenge in doing this is obtaining the most 
applicable historical data to ensure that the new estimate is as 
accurate as possible. One way of ensuring that the data are applicable 
is to perform checks of reasonableness to see if the results are 
similar. Different data sets converging toward one value provides a 
high degree of confidence in the data. 

Performing quality checks takes time and requires access to large 
quantities of data. This is often the most difficult, time-consuming, 
and costly activity in cost estimating. It can be exacerbated by a 
poorly defined technical baseline or WBS. However, by gathering 
sufficient data, cost estimators can analyze cost trends on a variety 
of related programs, which gives insight into cost estimating 
relationships that can be used to develop parametric models. 

Before collecting data, the estimator must fully understand what needs 
to be estimated. This understanding comes from the purpose and scope of 
the estimate, the technical baseline description, the WBS, and the 
ground rules and assumptions. Once the boundaries of the estimate are 
known, the next step is to establish an idea of what estimating 
methodology will be used. Only after these tasks have been performed 
should the estimator begin to develop an initial data collection plan. 

Data Collection: 

Data collection is a lengthy process and continues throughout the 
development of a cost estimate and through the program execution 
itself. Many types of data need to be collected—technical, schedule, 
program, and cost data. Once collected, the data need to be normalized. 
Data can be collected in a variety of ways, such as from databases of 
past projects, engineering build-up estimating analysis, interviews, 
surveys, data collection instruments, and focus groups. After the 
estimate is complete, the data need to be well documented, protected, 
and stored for future use in retrievable databases. Cost estimating 
requires a continual influx of current and relevant cost data to remain 
credible. The cost data should be managed by estimating professionals 
who understand what the historical data are based on, can determine 
whether the data have value in future projections, and can make the 
data part of the corporate history. 

Cost data should be continually supplemented with written vendor 
quotes, contract data, and actual cost data for each new program. 
Moreover, cost estimators should know the program acquisition plans, 
contracting processes, and marketplace conditions, all of which can 
affect the data. This knowledge provides the basis for credibly using, 
modifying, or rejecting the data in future cost estimates.

Knowing the factors that influence a program’s cost is essential for 
capturing the right data. Examples are equivalent source lines of code, 
number of interfaces for software development, number of square feet 
for construction, and the quantity of aircraft to be produced. To 
properly identify cost drivers, it is imperative that cost estimators 
meet with the engineers and other technical experts. In addition, by 
studying historical data, cost estimators can determine through 
statistical analysis the factors that tend to influence overall cost. 
Furthermore, seeking input from schedule analysts can provide valuable 
knowledge about how aggressive a program’s schedule may be. 

Cost estimates must be based on realistic schedule information. Some 
costs such as labor, quality, supervision, rented space and equipment, 
and other time-related overheads depend on the duration of the 
activities they support. Often the cost estimators are in synch with 
the baseline schedule with the early estimates, but they also have to 
keep in touch with changes in the schedule, since schedule changes can 
lead to cost changes. 

In addition to data for the estimate, backup data should be collected 
for performing cross-checks. This takes time and usually requires 
travel to meet with technical experts. It is important to plan ahead 
and schedule the time for these activities. Scheduling insufficient 
time can affect the estimator’s ability to collect and understand the 
data, which can then result in a less confident cost estimate. 

Common issues in data collection include inconsistent data definitions 
in historical programs compared to the new program. Understanding what 
the historical data include is vital to data reliability. For example, 
are the data skewed because they are for a program that followed an 
aggressive schedule and therefore instituted second and third shifts to 
complete the work faster? Or was a new manufacturing process 
implemented that was supposed to generate savings but resulted in more 
costs because of initial learning curve problems? Knowing the history 
behind the data will allow for its proper allocation for future 
estimates. 

Another issue is whether the data are even available. Some agencies may 
not have any cost databases. Data may be accessible at higher levels 
but information may not be sufficient to break them down to the 
lower levels needed to estimate various WBS elements. Data may be 
incomplete. For instance, they may be available for the cost to build a 
component, but the cost to integrate the component may be missing. 
Similarly, if data are in the wrong format, they may be difficult to 
use. For example, if the data are only in dollars and not hours, they 
may not be as useful if the labor and overhead rates are not available. 

Sometimes data are available, but the cost estimator cannot gain access 
to them. This can happen when the data are highly classified or 
considered competition sensitive. When this is the case, the cost 
estimator may have to change the estimating approach to fit the data 
that are available. Case study 31 gives an example. 

Case Study 31: Fitting the Estimating Approach to the Data, from 
Space Acquisitions, GAO-07-96: 

The lack of reliable technical source data hampers cost estimating. 
Officials GAO spoke with believed that cost estimation data and 
databases on which to base cost estimates were incomplete, 
insufficient, and outdated. They cited the lack of reliable historical 
and current cost, technical, and program data and expressed concern 
that available cost, schedule, technical, and risk data were not 
similar to the systems they were developing cost estimates for. In 
addition, some expressed concern that relevant classified and 
proprietary commercial data might exist but were not usually available 
to the costestimating community working on unclassified programs. Some 
believed that Air Force cost estimators needed to be able to use all 
relevant data, including those contained in National Reconnaissance 
Office cost databases, since the agency builds highly complex, 
classified satellites in comparable time and at comparable costs per 
pound. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Action to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Types Of Data: 
 
In general, the three main types of data are cost data, schedule or 
program data, and technical data. Cost data generally include labor 
dollars (with supporting labor hours and direct costs and overhead 
rates), material and its overhead dollars, facilities capital cost of 
money, and profit associated with various activities. Program cost 
estimators often do not know about specific dollars, so they tend to 
focus mostly on hours of resources needed by skill level. These 
estimates of hours are often inputs to specialized databases to convert 
them to cost estimates in dollars. 

Schedule or program data provide parameters that directly affect the 
overall cost. For example, lead-time schedules, start and duration of 
effort, delivery dates, outfitting, testing, initial operational 
capability dates, operating profiles, contract type, multiyear 
procurement, and sole source or competitive awards must all 
be considered in developing a cost estimate. 

Technical data define the requirements for the equipment being 
estimated, based on physical and performance attributes, such as 
length, width, weight, horsepower, and size. When technical data are 
collected, care must be taken to relate the types of technologies and 
development or production methodologies to be used. These change over 
time and require adjustments when estimating relationships are being 
developed. 

Cost data must often be derived from program and technical data. 
Moreover, program and technical data provide context for cost data, 
which by themselves may be meaningless. Consider the difference between 
these two examples: 
 
* Operations and maintenance utilities cost $36,500. 

* The Navy consumes 50,000 barrels of fuel per day per ship. 

In the operations and maintenance example, the technical and program 
descriptors are missing, requiring follow-up questions like: What 
specific utilities cost $36,500? Gas or electricity or telephone? What 
time does this cost represent? A month or a year? and When were these 
costs accrued? In the current year or 5 years ago? In the Navy example, 
a cost estimator would need to investigate what type of ship consumes 
50,000 barrels per day—aircraft carrier? destroyer?—and what type of 
fuel is consumed.[Footnote 36] 

It is essential that cost estimators plan for and gain access, where 
feasible, to cost and technical and program data in order to develop a 
complete understanding of what the data represent. Without this 
understanding, a cost estimator may not be able to correctly interpret 
the data, leading to greater risk that the data can be misapplied. 

Sources Of Data: 

Since all cost estimating methods are data-driven, analysts must know 
the best data sources. Table 10 lists some basic sources. Analysts 
should use primary data sources whenever possible. Primary data are 
obtained from the original source, can usually be traced to an audited 
document, are considered the best in quality, and are ultimately the 
most useful. Secondary data are derived rather than obtained directly 
from a primary source. Since they were derived, and thus changed, from 
the original data, their overall quality is lower and less useful. In 
many cases, secondary data are actual data that have been “sanitized” 
to obscure their proprietary nature. Without knowing the details, 
analysts will find such data of little use. 

Table 10: Basic Primary and Secondary Data Sources: 
Data type: 
Primary 
Secondary: 

Data type: Basic accounting records; 
Source: Primary. 

Data type: Data collection input forms; 
Source: Primary. 

Data type: Cost reports; 
Source: Primary, Secondary.

Data type: Historical databases; 
Source: Primary, Secondary.

Data type: Interviews; 
Source: Primary, Secondary.

Data type: Program briefs; 
Source: Primary, Secondary.

Data type: Subject matter experts; 
Source: Primary, Secondary.

Data type: Technical databases; 
Source: Primary, Secondary.

Data type: Other organizations; 
Source: Primary, Secondary.

Data type: Contracts or contractor estimates; 
Source: Secondary.

Data type: Cost proposals; 
Source: Secondary.

Data type: Cost studies; 
Source: Secondary.

Data type: Focus groups; 
Source: Secondary.

Data type: Research papers; 
Source: Secondary.

Data type: Surveys; 
Source: Secondary. 

Source: DOD and NASA. 

[End of table] 

Cost estimators must understand whether and how data were changed 
before deciding whether they will be useful. Furthermore, it is always 
better to use actual costs rather than estimates as data sources, since 
actual costs represent the most accurate data available. 

While secondary data should not be the first choice, they may be all 
that is available. Therefore, the cost estimator must seek to 
understand how the data were normalized, what the data represent, how 
old they are, and whether they are complete. If these questions can be 
answered, the secondary data may be useful for estimating and would 
certainly be helpful for cross-checking the estimate for reasonableness.

Sources of historical data include business plans, catalog prices, 
contract performance reports, contract funds status reports, cost and 
software data reports, forward pricing rate agreements, historical cost 
databases, market research, program budget and accounting data from 
prior programs, supplier cost information, historical or current vendor 
quotes, and weight reports. In the operating and support area, common 
data sources include DOD’s Visibility and Management of Operating and 
Support Costs management information system. Cost estimators should 
collect actual cost data from a list of similar and legacy programs. 
Since most new programs are improvements over existing ones, data 
should be available that share common characteristics with the new 
program. 

Historical data provide the cost estimator insight into actual costs on 
similar programs, including any cost growth since the original 
estimate. As a result, historical data can be used to challenge 
optimistic assumptions. For example, a review of the average labor 
rates for similar tasks on other programs could be a powerful reality 
check against assumptions of skill mixes and overall effort. In 
addition, historical data from a variety of contractors can be used to 
establish generic program costs or they can be used to establish cost 
trends of a specific contractor across a variety of programs. 

Historical data also provide contractor cost trends relative to 
proposal values, allowing the cost estimator to establish adjustment 
factors if relying on proposal data for estimating purposes. 
Additionally, insights can be obtained on cost accounting structures to 
allow an understanding of how a certain contractor charges things like 
other direct costs and overhead. 

However, historical cost data also contain information from past 
technologies, so it is essential that appropriate adjustments are made 
to account for differences between the new system and the existing 
system with respect to such things as design characteristics, 
manufacturing processes (automation versus hands-on labor), and types 
of material used. This is where statistical methods, like regression, 
that analyze cost against time and performance characteristics can 
reveal the appropriate technology-based adjustment. 

CPRs and cost and software data reports are excellent sources of 
historical cost data for DOD programs. The CPR is the primary report of 
cost and schedule progress on contracts containing EVM compliance 
requirements. It contains the time-phased budget, the actual cost, and 
earned value, which is the budgeted value of completed work. 

By reviewing CPR data, the cost analyst can gain valuable insights into 
performance issues that may be relevant to future procurements. For 
instance, CPR data can provide information about changes to the 
estimate to complete (or the total expected cost of the program) and 
the performance measurement baseline, and it explains the reason for 
any variances. Before beginning any analysis of such reports, the 
analyst should perform a cursory assessment to ensure that the 
contractor has prepared them properly. 

The several ways of analyzing cost data reports all use three basic 
elements in various combinations: 

* budgeted cost for work scheduled (BCWS), or the amount of budget 
allocated to complete a specific amount of work at a particular time;

* budgeted cost for work performed (BCWP), also known as earned value, 
which represents budgeted value of work accomplished; and; 

* actual cost of work performed (ACWP), or actual costs incurred for 
work accomplished.[Footnote 37] 

Cost data reports are often used in estimating analogous programs, from 
the assumption that it is reasonable to expect similar programs at 
similar contractors’ plants to incur similar costs. This analogy may 
not hold for the costs of hardware or software but may hold in the 
peripheral WBS areas of data, program management, or systems 
engineering. If the analyst can then establish costs for the major 
deliverables, such as hardware or software, a factor may be applied for 
each peripheral area of the WBS, based on historical data available 
from cost reports. Sometimes, the data listed in the WBS include 
elements that the analyst may not be using in the present 
estimate—spares, training, support equipment. In such cases, these 
elements should be removed before the data are analyzed. 

Rate and factor agreements contain rates and factors agreed to by the 
contractor and the appropriate government negotiator. Because the 
contractor’s business base may be fluid, with direct effect on these 
rates and factors, such agreements do not always exist. Information in 
them represents negotiated direct labor, overhead, general and 
administrative data, and facilities capital cost of money. These 
agreements could cover myriad factors, depending on each contractor’s 
accounting and cost estimating structure. Typical factors are material 
scrap, material handling, quality control, sustaining tooling, and 
miscellaneous engineering support factors. 

The scope of the estimate often dictates the need to consult with other 
organizations for raw data. Once government test facilities have been 
identified, for example, those organizations can be contacted for 
current cost data, support cost data, and the like. Other government 
agencies could also be involved with the development of similar 
programs and can be potential sources of data. Additionally, a number 
of government agencies and industry trade associations publish cost 
data that are useful in cost estimating. 

The Defense Contract Management Agency (DCMA) and the Defense Contract 
Audit Agency (DCAA) help DOD cost analysts obtain validated data. Both 
agencies have on-site representatives at most major defense contractor 
facilities. Navy contractor resident supervisors of shipbuilding, for 
example, help obtain validated data. Before a contract is awarded, DCMA 
provides advice and services to help construct effective solicitations, 
identify potential risks, select the most capable contractors, and 
write contracts that meet customers’ needs. In evaluating contract 
proposals, DCMA assists in the review of the proposal assumptions to 
identify how tightly scope was constrained to reduce risk premiums in 
the proposed cost. After a contract is awarded, DCMA monitors 
contractors’ performance and management systems to ensure that cost, 
product performance, and delivery schedules comply with the contract’s 
terms and schedule. It is common for DCMA auditors to be members of 
teams assembled to review elements of proposals, especially in areas of 
labor and overhead rates, cost, and supervision of man-hour 
percentages. 

DCMA analysts often provide independent estimates at completion for 
programs; they are another potential source of information for cost 
analysts.

DCAA performs necessary contract audits for DOD. It provides accounting 
and advisory services for contracts and subcontracts to all DOD 
components responsible for procurement and contract administration. 
Cost analysts should establish and nurture contacts with these 
activities, so that a continual flow of current cost-related 
information can be maintained. Although civil agencies have no 
comparable organizations, DCMA and DCAA occasionally provide support to 
them. 

Another area of potential cost data are contractor proposals. Analysts 
should remember that a contractor proposal as a source of data is a 
proposal—a document that represents the contractor’s best estimate of 
cost. Proposals also tend to be influenced by the amount the customer 
has to spend. When this is the case, the proposal data should be viewed 
as suspect, and care should be taken to determine if the proposal data 
are supportable. Because of this, an estimate contained in a 
contractor’s proposal should be viewed with some caution. During source 
selection in a competitive environment, for instance, lower proposed 
costs may increase the chances of receiving a contract award. This 
being so, it is very important to analyze the cost data for realism. A 
proposal can nonetheless provide much useful information and should be 
reviewed, when available, for the following: 

* structure and content of the contractor’s WBS; 

* contractor’s actual cost history on the same or other programs; 

* negotiated bills of material; 

* subcontracted items; 

* government-furnished equipment compared to contractor-furnished 
equipment lists; 

* contractor rate and factor data, based on geography and makeup of 
workforce; 

* a self-check to ensure that all pertinent cost elements are included; 

* top-level test of reasonableness; 

* technological state-of-the-art assumptions; and 

* estimates of management reserve and level of risk. 

Because of the potential for bias in proposal data, the estimator must 
test the data to see whether they deviate from other similar data 
before deciding whether they are useful for estimating. This can be 
done through a plant visit, where the cost estimator visits the 
contractor to discuss the basis for the proposal data. As with any 
potential source of data, it is critical to ensure that the data apply 
to the estimating task and are valid for use. In the next two sections, 
we address how a cost estimator should perform these important 
activities. 

Data Applicability: 

Because cost estimates are usually developed with data from past 
programs, it is important to examine whether the historical data apply 
to the program being estimated. Over time, modifications may have 
changed the historical program so that it is no longer similar to the 
new program. For example, it does not make sense to use data from an 
information system that relied on old mainframe technology when the new 
program will rely on server technology that can process data at much 
higher speeds. Having good descriptive requirements of the data is 
imperative in determining whether the data available apply to what 
is being estimated. 

To determine the applicability of data to a given estimating task, the 
analyst must scrutinize them in light of the following issues: 

* Do the data require normalization to account for differences in base 
years, inflation rates (contractor compared to government), or calendar 
year rather than fiscal year accounting systems? 

* Is the work content of the current cost element consistent with the 
historical cost element? 
 
* Have the data been analyzed for performance variation over time (such 
as technological advances)? Are there unambiguous trends between cost 
and performance over time? 
 
* Do the data reflect actual costs, proposal values, or negotiated 
prices and has the type of contract been considered? 

Proposal values are usually extremely optimistic and can lead to overly 
optimistic cost estimates and budgets. Furthermore, negotiated prices 
do not necessarily equate to less optimistic cost estimates. 
 
* Are sufficient cost data available at the appropriate level of detail 
to use in statistical measurements? 

* Are cost segregations clear, so that recurring data are separable 
from nonrecurring data and functional elements (manufacturing, 
engineering) are visible?

* Have risk and uncertainty for each data element been taken into 
account? High-risk elements usually cause optimistic cost estimates. 
 
* Have legal or regulatory changes affected cost for the same 
requirement? 

* When several historical values are available for the same concept, 
are they in close agreement or are they dispersed? 

If they are in close agreement, as long as the definitions agree they 
should provide valuable insight. If they are different, perhaps the 
issues are not settled, the approaches are still at variance, and 
historical data may not be as useful for estimating current programs’ 
costs. 

Once these questions have been answered, the next step is to assess the 
validity of the data before they can be used to confidently predict 
future costs. 

Validating And Analyzing The Data: 

The cost analyst must consider the limitations of cost data before 
using them in an estimate. Historical cost data have two predominant 
limitations: 
 
* the data represent contractor marketplace circumstances that must be 
known if they are to have future value, and 

* current cost data eventually become dated. 

The first limitation is routinely handled by recording these 
circumstances as part of the data collection task. To accommodate the 
second limitation, an experienced cost estimator can either adjust the 
data (if applicable) or decide to collect new data. In addition, the 
contract type to be used in a future procurement—for example, firm 
fixed-price, fixed-price incentive, or cost plus award fee—may differ 
from that of the historical cost data. Although this does not preclude 
using the data, the analyst must be aware of such conditions, so that 
an informed data selection decision can be made. A cost analyst must 
attempt to address data limitations by: 

* ensuring that the most recent data are collected, 

* evaluating cost and performance data together to identify 
correlation, 

* ensuring a thorough knowledge of the data’s background, and 

* holding discussions with the data provider. 

Thus, it is best practice to continuously collect new data so they can 
be used for making comparisons and determining and quantifying trends. 
This cannot be done without background knowledge of the data. This 
knowledge allows the estimator to confidently use the data directly, 
modify them to be more useful, or simply reject them. 

Once the data have been collected, the next step is to create a scatter 
plot to see what they look like. Scatter plotting provides a 
wealth of visual information about the data, allowing the analyst to 
quickly determine outliers, relationships, and trends. In scatter 
charts, cost is typically treated as the dependent variable and is 
plotted on the y axis, while various independent variables are plotted 
on the x axis. These independent variables depend on the data collected 
but are typically technical—weight, lines of code, speed—or operational 
parameters—crew size, flying hours. These statistics provide 
information about the amount of dispersion in the data set, which is 
important for determining risk. 

The cost estimator should first decide which independent variables are 
most likely to be cost drivers and then graph them separately. The 
extent to which the points are scattered will determine how likely it 
is that each independent variable is a cost driver. The less scattered 
the points are, the more likely it is that the variable is a cost 
driver. Eventually, the analyst will use statistical techniques to 
distinguish cost drivers, but using scatter charts is an excellent way 
to reduce their number. 

The cost estimator should also examine each scatter chart in unit space 
to determine if a linear relationship exists. Many relationships are 
not linear; in such cases, the estimator can often perform a 
transformation to make the data linear. If the data appear to be 
exponential when plotted in unit space, the analyst should try plotting 
the natural log of the independent variable on the y axis. If the data 
appear to represent a power function, the analyst should try plotting 
the natural log of both the cost and the independent variable. In both 
cases, the goal is to transform the data appropriately to reveal a 
linear relationship, because most cost estimating relationships are 
based on linear regression. 

After analyzing the data through a scatter plot, the estimator should 
calculate descriptive statistics to characterize and describe the data 
groups. Important statistics include sample size, mean, standard 
deviation, and coefficient of variation. Calculating the mean provides 
the estimator with the best estimate, because it is the average of the 
historical data. To determine the dispersion within the data set, the 
estimator must calculate the standard deviation. Finally, the estimator 
should calculate the coefficient of variation so that variances between 
data sets can be compared. 

The coefficient of variation is calculated by dividing the standard 
deviation by the mean.[Footnote 38] This provides a percentage that can 
be used to examine which data set has the least variation. Once the 
statistics have been derived, creating visual displays of them helps 
discern differences among groups. Bar charts, for example, are often 
useful for comparing averages. Histograms can be used to examine the 
distribution of different data sets in relation to their frequency. 
They can also be used for determining potential outliers. (Chapter 11 
has more information on statistical approaches.) 

Many times, estimates are not based on actual data but are derived by 
subjective engineering judgment. All engineering judgments should be 
validated before being used in a cost estimate. Validation involves 
cross-checking the results, in addition to analyzing the data and 
examining the documentation for the judgment. Graphs and scatter charts 
can often help validate an engineering judgment, because they can 
quickly point out any outliers. 

It is never a good idea to discard an outlier without first 
understanding why a data point is outside the normal range. An outlier 
is a data point that is typically defined as falling outside the 
expected range of three standard deviations. Statistically speaking, 
outliers are rare, occurring only 0.3 percent of the time. If a data 
point is truly an outlier, it should be removed from the data set, 
because it can skew the results. However, an outlier should not be 
removed simply because it appears too high or too low compared to the 
rest of the data set. Doing so is naïve. Instead, a cost estimator 
should provide adequate documentation as to why an outlier was removed 
and this documentation should include comparisons to historical data 
that show the outlier is in fact an anomaly. If possible, the 
documentation should describe why the outlier exists; for example, 
there might have been a strike, a program restructuring, or a natural 
disaster that skewed the data. If the historical data show the outlier 
is just an extreme case, the cost estimator should retain the data 
point; otherwise, it will appear that the estimator was trying to 
manipulate the data. This should never be done, since all available 
historical data are necessary for capturing the natural variation 
within programs. 

EVM Data Reliability: 

In chapter 3, we discussed top-level EVM data reliability tasks such 
as: 

* requesting a copy of the EVM system compliance letter showing the 
contractor’s ability to satisfy the 32 guidelines; 

* requesting a copy of the IBR documentation and final briefing to see 
what risks were identified and what weaknesses, if any, were found; 

* determining whether EVM surveillance is being done by qualified and 
independent staff; and; 

* determining the financial accounting status of the contractor’s EVM 
system to see whether any adverse opinions would call into question the 
reliability of the accounting data. 

In addition to these tasks, auditors should perform a sanity check to 
see if the data even make sense. For example, the auditor should review 
all WBS elements in the CPR to determine whether there are any data 
anomalies such as: 

* negative values for BCWS, BCWP, ACWP, estimate at completion (EAC), 
or budget at completion (BAC);

* large month-to-month performance swings (BCWP) not attributable to 
technical or schedule problems (may indicate cost collection issues); 

* BCWS and BCWP data with no corresponding ACWP; 

* BCWP with no BCWS or ACWP; 

* ACWP with no BCWS or BCWP; 

* large and continuing unexplained variances between ACWP and BCWP; 

* inconsistencies between EAC and BAC (for example, EAC with no BAC or 
BAC with no EAC); 

* ACWP greater than EAC; 

* BCWP or BCWS exceed the BAC. 

Despite the fact that these anomalies should be rare and fully 
explained in the variance analysis portion of the report, unfortunately 
we have found programs that submit CPRs with these types of errors. 
Case study 32 highlights this issue. 

Case Study 32: Data Anomalies, from Cooperative Threat Reduction, GAO-
06-692: 

The EVM system the contractor was using to record, predict, and monitor 
progress contained flawed and unreliable data. GAO found serious 
discrepancies in the data, such as improper calculations and accounting 
errors. For example, from September 2005 through January 2006 the 
contractor’s EVM reports had not captured almost $29 million in actual 
costs for the chemical weapons destruction facility project. EVM 
current period data were not accurate because of historical data 
corruption, numerous mistakes in accounting accruals, and manual budget 
adjustments. The mistakes underestimated the true cost of the project 
by ignoring cost variances that had already occurred. 

For example, the Moscow project management task had been budgeted at a 
cost of $100,000. According to the January 2006 EVM report, the work 
was complete, but the actual cost was $2.6 million—an overrun of 
approximately $2.5 million that the EVM report failed to capture. Such 
data were misleading and skewed the project’s overall performance. 
Unreliable EVM data limited DOD’s efforts to accurately measure 
progress on the Shchuch’ye project and estimate its final completion 
date and cost. 

GAO recommended that the Secretary of Defense direct the Defense Threat 
Reduction Agency, in conjunction with the U.S. Army Corps of Engineers, 
to ensure that the contractor’s EVM system contain valid, reliable data 
and that the system reflect actual cost and schedule conditions; 
withhold a portion of the contractor’s award fee until the EVM system 
produced reliable data; and require the contractor to perform an IBR 
after awarding the contract for completing Building 101. 

Source: GAO, Cooperative Threat Reduction: DOD Needs More Reliable Data 
to Better Estimate the Cost and Schedule of the Shchuch’ye Facility, 
GAO-06-692, Washington, D.C.: May 31, 2006. 

[End of case study] 

Data Normalization: 

The purpose of data normalization (or cleansing) is to make a given 
data set consistent with and comparable to other data used in the 
estimate. Since data can be gathered from a variety of sources, they 
are often in many different forms and need to be adjusted before being 
used for comparison analysis or as a basis for projecting future costs. 
Cost data are adjusted in a process called normalization, stripping 
out the effect of certain external influences. The objective of data 
normalization is to improve data consistency, so that comparisons and 
projections are more valid and other data can be used to increase the 
number of data points. Data are normalized in several ways. 

Cost Units: 

Cost units primarily adjust for inflation. Because the cost of an item 
has a time value, it is important to know the year in which funds were 
spent. For example, an item that cost $100 in 1990 is more expensive 
than an item that cost $100 in 2005 because of the effects of inflation 
over the 15 years that would make the 1990 item more expensive when 
converted to a 2005 equivalent cost. Costs may also be adjusted for 
currency conversions. 

In addition to inflation, the cost estimator needs to understand what 
the cost represents. For example, does it represent only direct labor 
or does it include overhead and the contractor’s profit? Finally, cost 
data have to be converted to equivalent units before being used in a 
data set. That is, costs expressed in thousands, millions, or billions 
of dollars must be converted to one format—for example, all costs 
expressed in millions of dollars. 

Sizing Units: 

Sizing units normalize data to common units—for example, cost per foot, 
cost per pound, dollars per software line of code. When normalizing 
data for unit size, it is very important to define exactly what the 
unit represents: What constitutes a software line of code? Does it 
include carriage returns or comments? The main point is to clearly 
define what the sizing metric is so that the data can be converted to a 
common standard before being used in the estimate. 

Key Groupings: 

Key groupings normalize data by similar missions, characteristics, or 
operating environments by cost type or work content. Products with 
similar mission applications have similar characteristics and traits, 
as do products with similar operating environments. For example, space 
systems exhibit characteristics different from those of submarines, but 
the space shuttle has characteristics distinct from those of a 
satellite even though they may share common features. Costs should also 
be grouped by type. For example, costs should be broken out between 
recurring and nonrecurring or fixed and variable costs. 

Technology Maturity: 

Technology maturity normalizes data for where a program is in its life 
cycle; it also considers learning and rate effects. The first unit of 
something would be expected to cost more than the 1,000th unit, just 
as a system procured at one unit per year would be expected to cost 
more per unit than the same system procured at 1,000 units per year. 
Technology normalization is the process of adjusting cost data for 
productivity improvements resulting from technological advancements 
that occur over time. 

In effect, technology normalization is the recognition that technology 
continually improves, so a cost estimator must make a subjective 
attempt to measure the effect of this improvement on historical program 
costs. For instance, an item developed 10 years ago may have been 
considered state of the art and the costs would be higher than normal. 
Today, that item may be available off the shelf and therefore the costs 
would be considerably less. 

Therefore, technology normalization is the ability to forecast 
technology by predicting the timing and degree of change of 
technological parameters associated with the design, production, and 
use of devices. Being able to adjust the cost data to reflect where the 
item is in its life cycle, however, is very subjective, because it 
requires identifying the relative state of technology at different 
points in time. 

Homogeneous Groups: 

Using homogeneous groups normalizes for differences between historical 
and new program WBS elements in order to achieve content consistency. 
To do this type of normalization, a cost estimator needs to gather cost 
data that can be formatted to match the desired WBS element definition. 
This may require adding and deleting certain items to get an apples-to-
apples comparison. A properly defined WBS dictionary is necessary to 
avoid inconsistencies. 

Recurring And Nonrecurring Costs: 

Embedded within cost data are recurring and nonrecurring costs. These 
are usually estimated separately to keep one-time nonrecurring costs 
from skewing the costs for recurring production units. For this 
reason, it is important to segregate cost data into nonrecurring and 
recurring categories. 

Nonrecurring Costs: 

SCEA defines nonrecurring costs as the elements of development and 
investment costs that generally occur only once in a system’s life 
cycle. They include all the effort required to develop and qualify an 
item, such as defining its requirements and its allocation, design, 
analysis, development, qualification, and verification. Costs for the 
following are generally nonrecurring: 

* manufacturing and testing development units, both breadboard and 
engineering, for hardware, as well as qualification and life-test 
units; 

* retrofitting and refurbishing development hardware for 
requalification; 

* developing and testing virtually all software before beginning 
routine system operation; nonrecurring integration and test efforts 
usually end when qualification tests are complete;
 
* providing services and some hardware, such as engineering, before and 
during critical design review; 

* developing, acquiring, producing, and checking all tooling, ground 
handling, software, and support equipment and test equipment. 

Recurring Costs: 

As defined by SCEA, recurring costs are incurred for each item produced 
or each service performed. For example, the costs associated with 
producing hardware—that is, manufacturing and testing, providing 
engineering support for production, and supporting that hardware with 
spare units or parts—are recurring costs. Recurring integration and 
testing, including the integration and acceptance testing of production 
units at all WBS levels, also represent recurring costs. In addition, 
refurbishing hardware for operational or spare units is a recurring 
cost, as is maintaining test equipment and production support software. 
In contrast, maintaining system operational software, although 
recurring in nature, is often considered part of operating and support 
costs, which might also have nonrecurring components. 

Similar to nonrecurring and recurring costs are fixed and variable 
costs. Fixed costs are static, regardless of the number of quantities 
to be produced. An example of a fixed cost is the cost to rent a 
facility. A variable cost is directly affected by the number of units 
produced and includes such things as the cost of electricity or 
overtime pay. Knowing what the data represent is important for 
understanding anomalies that can occur as the result of production unit 
cuts. 

The most important reason for differentiating recurring from 
nonrecurring costs is in their application to learning curves. Simply 
put, learning curve theory applies only to recurring costs. Cost 
improvement or learning is generally associated with repetitive actions 
or processes, such as those directly tied to producing an item again 
and again. Categorizing as recurring or variable costs that are 
affected by the quantity of units being produced adds more clarity to 
the data. An analyst who knows only the total cost of something does 
not know how much of that cost is affected by learning. 

Inflation Adjustments: 

In the development of an estimate, cost data must be expressed in like 
terms. This is usually accomplished by inflating or deflating cost data 
to express them in a base year that will serve as a point of reference 
for a fixed price level. Applying inflation is an important step in 
cost estimating. If a mistake is made or the inflation amount is not 
correct, cost overruns can result, as case study 33 illustrates. 

Case Study 33: Inflation, from Defense Acquisitions, GAO-05-183: 

Inflation rates can significantly affect ship budgets. Office of the 
Secretary of Defense (OSD) and OMB inflation indexes are based on a 
forecast of the implicit price deflator for the gross domestic product. 
Until recently, the Navy had used OSD and OMB inflation rates; 
shipbuilding industry rates were historically higher. As a result, 
contracts were signed and executed using industry-specific inflation 
rates while budgets were based on the lower inflation rates, creating a 
risk of cost growth from the outset. For the ships reviewed, this 
difference in inflation rates explained 30 percent of the $2.1 billion 
cost growth. The Navy had changed its inflation policy in February 
2004, directing program offices to budget with what the Navy believed 
were more realistic inflation indexes, anticipating that this would 
help curtail requests for prior-year completion funds. 

Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, GAO-05-183, 
Washington, D.C.: Feb. 28, 2005. 

[End of case study] 

Applying inflation correctly is necessary if the cost estimate is to be 
credible. In simple terms, inflation reflects the fact that the cost of 
an item usually continues to rise over time. Inflation rates are used 
to convert a cost from its current year into a constant base year so 
that the effects of inflation are removed. When cost estimates are 
stated in base-year dollars, the implicit assumption is that the 
purchasing power of the dollar has remained unchanged over the period 
of the program being estimated. Cost estimates are normally prepared in 
constant dollars to eliminate the distortion that would otherwise be 
caused by price level changes. This requires the transformation of 
historical or actual cost data into constant dollars. 

For budgeting purposes, however, the estimate must be expressed in then-
year dollars to reflect the program’s projected annual costs by 
appropriation. This requires applying inflation to convert from base-
year to then-year dollars. Cost estimators must make assumptions about 
what inflation indexes to use, since any future inflation index is 
uncertain. In cases in which inflation decreases over time, applying 
the wrong inflation rate will result in a higher cost estimate. Worse 
is the situation in which the inflation is higher than projected, 
resulting in costs that are not sufficient to keep pace with inflation, 
as illustrated in case study 33. Thus, it is imperative that inflation 
assumptions be well documented and that the cost estimator always 
perform uncertainty and sensitivity analysis to study the effects of 
changes on the assumed rates. 

Selecting The Proper Indexes: 

The cost estimator will not have to construct an index to apply 
inflation but will select one to apply to cost data. Often, the index 
is directed by higher authority, such as OMB. In this way, all programs 
can be compared and aggregated with the same escalation rate, since 
they are all being executed in the same economic circumstances. This 
doesn’t mean that the forward escalation rates are correct—in fact, 
escalation rates are difficult to forecast—but that program comparisons 
will at least not be confused by different assumptions about 
escalation. When the index is not directed, a few general guidelines 
can help the cost estimator select the correct index. Because all 
inflation indexes measure the average rate of inflation for a 
particular market basket of goods, the objective in making a choice is 
to select the one whose market basket most closely matches the program 
to be estimated. The key is to use common sense and objective judgment. 
For example, the consumer price index would be a poor indicator of 
inflation for a new fighter aircraft, because the market baskets 
obviously do not match. Labor escalation would be affected by different 
factors than, say, fuel or steel costs. Although the selected index 
will never exactly match the market basket of costs, the closer the 
match, the better the estimate. 

Weighted indexes are used to convert constant, base-year, dollars to 
then-year dollars and vice versa. Raw indexes are used to change the 
economic base of constant dollars from one base year to another. 
Contract prices are stated in then-year dollars, and weighted indexes 
are appropriate for converting them to base-year dollars. Published 
historical cost data are frequently, but not always, normalized to a 
common base year, and raw indexes are appropriate for changing the base 
year to match that of the program being estimated. It is important that 
the cost estimator determine what year dollars cost data are expressed 
in, so that normalization for inflation can be done properly. 

Schedule risk can affect the magnitude of escalation in a cost 
estimate. The escalation dollars are often estimated by applying a 
monthly escalation rate (computed so that compounding monthly values 
equates to the forecasted annual rate) to dollars forecasted to be 
spent in each month. If the schedule is delayed, a dollar that would 
have been escalated by, say, 30 months might now be escalated for 36 
months. Even if the cost estimate in today’s dollars is an accurate 
estimate, a schedule slip would affect the amount of escalation. 

In addition, the question of escalating the contingency reserve arises. 
Some cost estimating systems calculate the contingency on base-year 
dollars but do not escalate the contingency, perhaps because they do 
not have a way to determine when the dollars will be spent. In a cost 
risk analysis, in contrast, the contingency reserve is computed during 
the simulation using the risk in the line-item costs. If the simulated 
line-item costs are then subjected to escalation during the same 
simulation, the process effectively escalates the contingency. This is 
appropriate, since contingency money is just more money needed to be 
spent on the statement of work, and it should be affected by escalation 
as is any other money spent. 

Data Documentation: 

After the data have been collected, analyzed, and normalized, they must 
be documented and stored for future use. One way to keep a large amount 
of historical data viable is to continually supplement them with every 
new system’s actual return costs and with every written vendor quote or 
new contract. Although data have many sources, the predominant sources 
are the manufacturers who make the item or similar items. It can take 
years for a cost estimator to develop an understanding of these sources 
and to earn the trust of manufacturers regarding the use of their 
proprietary and business-sensitive data. Once trust has been 
established and maintained for some time, the cost estimator can 
normally expect a continual flow of useful data. 

All data collection activities must be documented as to source, work 
product content, time, units, and assessment of accuracy and 
reliability. Comprehensive documentation during data collection greatly 
improves quality and reduces subsequent effort in developing and 
documenting the estimate. The data collection format should serve two 
purposes. First, the format should provide for the full documentation 
and capture of information to support the analysis. Second, it should 
provide for standards that will aid in mapping other forms of cost 
data. 

Previously documented cost estimates may provide useful data for a 
current estimate. Relying on previous estimates can save the cost 
estimator valuable time by eliminating the need to research and conduct 
statistical analyses that have already been conducted. For example, a 
documented program estimate may provide the results of research on 
contractor data, identification of significant cost drivers, or actual 
costs, all of which are valuable to the cost estimator. Properly 
documented estimates describe the data used to estimate each WBS 
element, and this information can be used as a good starting point for 
the new estimate. Moreover, relying on other program estimates can be 
valuable in understanding various contractors and providing cross-
checks for reasonableness. 

Because many cost documents are secondary sources of information, the 
cost estimator should be cautious. When using information from 
documented cost estimates, the analyst should fully understand the 
data. For example, if a factor was constructed from CPRs, the cost 
estimator should ask the following questions to see if the data are 
valid for the new program: 

* What was the base used in the ratio? 

* Are the WBS elements consistent with those of the system being 
estimated—for example, is data management included in the data or the 
systems engineering and program management element? 

* Was the factor computed from the ACWP or the EAC? 

* What percentage complete is the contract? 

7. Best Practices Checklist: Data: 
 
* As the foundation of an estimate, data: 
- Have been gathered from historical actual cost, schedule and program, 
and technical sources; 
- Apply to the program being estimated; 
- Have been analyzed for cost drivers; 
- Have been collected from primary sources, if possible, and secondary 
sources as the next best option, especially for cross-checking results; 
- Have been adequately documented as to source, content, time, units, 
assessment of accuracy and reliability, and circumstances affecting the 
data; 
- Have been continually collected, protected, and stored for future 
use; 
- Were assembled as early as possible, so analysts can participate in 
site visits to understand the program and question data providers. 

* Before being used in a cost estimate, the data were: 
- Fully reviewed to understand their limitations and risks; 
- Segregated into nonrecurring and recurring costs; 
- Validated, using historical data as a benchmark for reasonableness; 
- Current and found applicable to the program being estimated; 
- Analyzed with a scatter plot to determine trends and outliers; 
- Analyzed with descriptive statistics; 
- Normalized to account for cost and sizing units, mission or 
application, technology maturity, and content so they are consistent 
for comparisons; 
- Normalized to constant base-year dollars to remove the effects of 
inflation, and the inflation index was documented and explained. 

[End of Chapter 10] 

Chapter 11: Developing A Point Estimate: 

In this chapter, we discuss step 7 in the high-quality estimating 
process. Step 7 pulls all the information together to develop the point 
estimate—the best guess at the cost estimate, given the underlying 
data. High-quality cost estimates usually fall within a range of 
possible costs, the point estimate being between the best and worst 
case extremes. (We explain in chapter 14 how to develop this range of 
costs using risk and uncertainty analysis.) The cost estimator must 
perform several activities to develop a point estimate: 

* develop the cost model by estimating each WBS element, using the best 
methodology, from the data collected; 

* include all estimating assumptions in the cost model; 

* express costs in constant-year dollars; 

* time-phase the results by spreading costs in the years they are 
expected to occur, based on the program schedule; and; 
 
* add the WBS elements to develop the overall point estimate. 

Having developed the overall point estimate, the cost estimator must 
then: 

* validate the estimate by looking for errors like double counting and 
omitted costs and ensuring that estimates are comprehensive, accurate, 
well-documented, and credible (more information on validation is in 
chapter 15); 

* compare the estimate against the independent cost estimate and 
examine where and why there are differences; 

* perform cross-checks on cost drivers to see if results are similar; 
and; 

* update the model as more data become available or as changes occur 
and compare the results against previous estimates.

We have already discussed how to develop a WBS and GR&As, collect and 
normalize the data into constant base-year dollars, and time-phase the 
results. Once all the data have been collected, analyzed, and 
validated, the cost estimator must select a method for developing the 
cost estimate. 

Cost Estimating Methods: 

The three commonly used methods for estimating costs are analogy, 
engineering build-up, and parametric. An analogy uses the cost of a 
similar program to estimate the new program and adjusts for 
differences. The engineering build-up method develops the cost estimate 
at the lowest level of the WBS, one piece at a time, and the sum of the 
pieces becomes the estimate. The parametric method relates cost to one 
or more technical, performance, cost, or program parameters, using a 
statistical relationship. 

Which method to select depends on where the program is in its life 
cycle. Early in the program, definition is limited and costs may not 
have accrued. Once a program is in production, cost and technical data 
from the development phase can be used to estimate the remainder of the 
program. Table 11 gives an overview of the strengths, weaknesses, and 
applications of the three methods. 

Table 11: Three Cost Estimating Methods Compared : 

Method: Analogy; 
Strength: 
* Requires few data; 
* Based on actual data;
* Reasonably quick;
* Good audit trail.
Weakness: 
* Subjective adjustments; 
* Accuracy depends on similarity of items; 
* Difficult to assess effect of design change; 
* Blind to cost drivers; 
Application: 
* When few data are available; 
* Rough-order-of-magnitude estimate; 
* Cross-check. 

Method: Engineering build-up; 
Strength: 
* Easily audited 
* Sensitive to labor rates
* Tracks vendor quotes
* Time honored
Weakness: 
* Requires detailed design 
* Slow and laborious
* Cumbersome
Application: 
* Production estimating 
* Software development 
* Negotiations 
 

Method: Parametric; 
Strength: 
* Reasonably quick; 
* Encourages discipline;
* Good audit trail;
* Objective, little bias;
* Cost driver visibility;
* Incorporates real-world effects (funding, technical, 
risk); 
Weakness: 
* Lacks detail; 
* Model investment;
* Cultural barriers;
* Need to understand model’s behavior;
Application: 
* Budgetary estimates; 
* Design-to-cost trade studies; 
* Cross-check; 
* Baseline estimate; 
* Cost goal allocations. 

Source: © 2003, MCR, LLC, “Cost Estimating: The Starting Point of EVM.” 

[End of table] 

Other cost estimating methods include: 

* expert opinion, which relies on subject matter experts to give their 
opinion on what an element should cost;[Footnote 39] 
 
* extrapolating, which uses actual costs and data from prototypes to 
predict the cost of future elements; and; 

* learning curves, which is a common form of extrapolating from actual 
costs. 

In the sections below, we describe these methods and their advantages 
and disadvantages. Finally, we discuss how to pull all the methods 
together to develop the point estimate. 

Analogy Cost Estimating Method: 

An analogy takes into consideration that no new program, no matter how 
state of the art it may be technologically, represents a totally new 
system. Most new programs evolve from programs already fielded that 
have had new features added on or that simply represent a new 
combination of existing components. The analogy method uses this 
concept for estimating new components, subsystems, or total programs. 
That is, an analogy uses actual costs from a similar program with 
adjustments to account for differences between the requirements of the 
existing and new systems. A cost estimator typically uses this method 
early in a program’s life cycle, when insufficient actual cost data are 
available but the technical and program definition is good enough to 
make the necessary adjustments. 

Adjustments should be made as objectively as possible, by using factors 
(sometimes scaling parameters) that represent differences in size, 
performance, technology, or complexity. The cost estimator should 
identify the important cost drivers, determine how the old item relates 
to the new item, and decide how each cost driver affects the overall 
cost. All estimates based on the analogy method, however, must pass 
the “reasonable person” test—that is, the sources of the analogy and 
any adjustments must be logical, credible, and acceptable to a 
reasonable person. In addition, since analogies are one-to-one 
comparisons, the historical and new systems should have a strong 
parallel. 

Analogy relies a great deal on expert opinion to modify the existing 
system data to approximate the new system. If possible, the adjustments 
should be quantitative rather than qualitative, avoiding subjective 
judgments as much as possible. An analogy is often used as a cross-
check for other methods. Even when an analyst is using a more detailed 
cost estimating technique, an analogy can provide a useful sanity 
check. Table 12 shows how an analogy works. 

Table 12: An Example of the Analogy Cost Estimating Method: 

Parameter: Engine; 
Existing system: F-100; 
New system: F-200; 
Cost of new system (assuming a linear relationship): [Empty]. 

Parameter: Thrust; 
Existing system: 12,000 lbs; 
New system: 16,000 lbs; 
Cost of new system (assuming a linear relationship): [Empty]. 

Parameter: Cost; 
Existing system: $5.2 million; 
New system: [Empty]; 
Cost of new system (assuming a linear relationship): (16,000/12,000) x 
$5.2 million = $6.9 million. 

Source: © 2003, Society of Cost Estimating and Analysis (SCEA), 
“Costing Techniques.” 

[End of table] 

The equation in table 12 implicitly assumes a linear relationship 
between engine cost and amount of thrust. However, there should be a 
compelling scientific or engineering reason why an engine’s cost is 
directly proportional to its thrust. Without more data (or an expert on 
engine costs), it is hard to know what parameters are the true drivers 
of cost. Therefore, when using the analogy method, it is important that 
the estimator research and discuss with program experts the 
reasonableness of technical program drivers to determine whether they 
are significant cost drivers. 

The analogy method has several advantages: 
 
* It can be used before detailed program requirements are known. 

* If the analogy is strong, the estimate will be defensible. 

* An analogy can be developed quickly and at minimum cost. 

* The tie to historical data is simple enough to be readily understood. 

Analogies also have some disadvantages: 

* An analogy relies on a single data point. 

* It is often difficult to find the detailed cost, technical, and 
program data required for analogies. 

* There is a tendency to be too subjective about the technical 
parameter adjustment factors. 

The last disadvantage can be best explained with an example. If a cost 
estimator assumes that a new component will be 20 percent more complex 
but cannot explain why, this adjustment factor is unacceptable. The 
complexity must be related to the system’s parameters, such as that the 
new system will have 20 percent more data processing capacity or will 
weigh 20 percent more. Case study 34 highlights what can happen when 
technical parameter assumptions are too optimistic. 

Case Study 34: Cost Estimating Methods, from Space Acquisitions, GAO-
07-96: 

In 2004, Advanced Extremely High Frequency (AEHF) satellite program 
decision makers relied on the program office cost estimate rather than 
the independent estimate the CAIG developed to support the production 
decision. The program office estimated that the system would cost about 
$6 billion, on the assumption that AEHF would have 10 times more 
capacity than Milstar, the predecessor satellite, at half the cost and 
weight. However, the CAIG concluded that the program could not deliver 
more data capacity at half the weight, given the state of the 
technology. In fact, the CAIG believed that to get the desired increase 
in data rate, the weight would have to increase proportionally. As a 
result, the CAIG estimated that AEHF would cost $8.7 billion and 
predicted a $2.7 billion cost overrun. 

The CAIG relied on weight data from historical satellites to estimate 
the program’s cost, because it considered weight to be the best cost 
predictor for military satellite communications. The historical data 
from the AEHF contractor showed that the weight had more than doubled 
since the program began and that the majority of the weight growth was 
in the payload. The Air Force also used weight as a cost predictor but 
attributed the weight growth to structural components rather than the 
more costly payload portion of the satellite. The CAIG stated that 
major cost growth was inevitable from the program start because 
historical data showed that it was possible to achieve a weight 
reduction or an increase in data capacity but not both at the same 
time. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Action to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Engineering Build-Up Cost Estimating Method: 

The engineering build-up cost estimating method builds the overall cost 
estimate by summing or “rolling-up” detailed estimates done at lower 
levels of the WBS. Because the lower-level estimating associated with 
the build-up method uses industrial engineering principles, it is often 
referred to as engineering build-up and is sometimes referred to as a 
grass-roots or bottom-up estimate. 

An engineering build-up estimate is done at the lowest level of detail 
and consists of labor and materials costs that have overhead and fee 
added to them. In addition to labor hours, a detailed parts list is 
required. Once in hand, the material parts are allocated to the lowest 
WBS level, based on how the work will be accomplished. In addition, 
quantity and schedule have to be considered in order to capture the 
effects of learning. Typically, cost estimators work with engineers to 
develop the detailed estimates. The cost estimator’s focus is to get 
detailed information from the engineer in a way that is reasonable, 
complete, and consistent with the program’s ground rules and 
assumptions. The cost estimator must find additional data to validate 
the engineer’s estimates. 

An engineering build-up method is normally used during the program’s 
production, because the program’s configuration has to be stabilized, 
and actual cost data are required to complete the estimate. The 
underlying assumption of this method is that historical costs are good 
predictors of future costs.The premise is that data from the 
development phase can be used to estimate the cost for production. 
As illustrated in table 13, the build-up method is used when an analyst 
has enough detailed information about building an item—such as number 
of hours and number of parts—and the manufacturing process to be used.

Table 13: An Example of the Engineering Build-Up Cost Estimating 
Method: 

 
Problem: Estimate sheet metal cost of the inlet nacelle for a new 
aircraft; 
Similar aircraft: F/A-18 inlet nacelle; 
Solution: Apply historical F/A-18 variance for touch labor effort and 
apply support labor factor to adjust estimated touch labor hours; 
Result: 2,000 hours x 1.2 = 2,400 touch labor hours and 2,400 labor 
hours x 1.48 = 3,522 labor hours (touch labor plus support labor) 
estimate for new aircraft. 

Problem: Standard hours to produce a new nacelle are estimated at 2,000 
for touch labor; adjust to reflect experience of similar aircraft and 
support labor effort; 
Similar aircraft: F/A-18 inlet nacelle experienced a 20% variance in 
touch labor effort above the industrial engineering standard. In 
addition, F/A-18 support labor was equal to 48% of the touch labor 
hours; 
Solution: [Empty]; 
Result: Average labor rates would then be used to convert these total 
labor hours into costs. 

Source: © 2003, Society of Cost Estimating and Analysis (SCEA), 
“Costing Techniques.” 

[End of table] 


Because of the high level of detail, each step of the work flow should 
be identified, measured, and tracked, and the results for each outcome 
should be summed to make the point estimate. 

* The several advantages to the build-up technique include: 

* the estimator’s ability to determine exactly what the estimate 
includes and whether anything was overlooked, 

* its unique application to the specific program and manufacturer, 

* that it gives good insight into major cost contributors, and 

* easy transfer of results to other programs. 

Some disadvantages of the engineering build-up method are that 

* it can be expensive to implement and it is time consuming, 

* it is not flexible enough to answer what-if questions, 

* new estimates must be built for each alternative, 

* the product specification must be well known and stable, 

* all product and process changes must be reflected in the estimate, 

* small errors can grow into larger errors during the summation, and 

* some elements can be omitted by accident. 

Parametric Cost Estimating Method: 

In the parametric method, a statistical relationship is developed 
between historical costs and program, physical, and performance 
characteristics. The method is sometimes referred to as a top-down 
approach. Types of physical characteristics used for parametric 
estimating are weight, power, and lines of code. Other program and 
performance characteristics include site deployment plans for 
information technology installations, maintenance plans, test and 
evaluation schedules, technical performance measures, and crew size. 
These are just some examples of what could be a cost driver for a 
particular program. Sources for these cost drivers are often found in 
the technical baseline, cost analysis requirements document or cost 
analysis data requirement. The important thing is that the attributes 
used in a parametric estimate should be cost drivers of the program. 
The assumption driving the parametric approach is that the same factors 
that affected cost in the past will continue to affect future costs. 
This method is often used when little is known about a program except 
for a few key characteristics like weight or volume. 

Using a parametric method requires access to historical data, which may 
be difficult to obtain. If the data are available, they can be used to 
determine the cost drivers and to provide statistical results and can 
be adjusted to meet the requirements of the new program. Unlike an 
analogy, parametric estimating relies on data from many programs and 
covers a broader range. Confidence in a parametric estimate’s results 
depends on how valid the relationships are between cost and the 
physical attributes or performance characteristics. Using this method, 
the cost estimator must always present the related statistics, 
assumptions, and sources for the data. 

The goal of parametric estimating is to create a statistically valid 
cost estimating relationship using historical data. The parametric CER 
can then be used to estimate the cost of the new program by entering 
its specific characteristics into the parametric model. CERs 
established early in a program’s life cycle should be continually 
revisited to make sure they are current and the input range still 
applies to the new program. In addition, parametric CERs should be well 
documented, because serious estimating errors could occur if the CER is 
improperly used. 

Parametric techniques can be used in a wide variety of situations, 
ranging from early planning estimates to detailed contract 
negotiations. It is always essential to have an adequate number of 
relevant data points, and care must be taken to normalize the dataset 
so that it is consistent and complete. In software, the development 
environment—that is, the extent to which the requirements are 
understood and the strength of the programmers’ skill and experience—is 
usually the major cost driver. Because parametric relationships are 
often used early in a program, when the design is not well defined, 
they can easily be reflected in the estimate as the design changes 
simply by adjusting the values of the input parameters. 

It is important to make sure that the program attributes being 
estimated fall within (or, at least, not far outside) the CER dataset. 
For example, if a new software program was expected to contain 1 
million software lines of code and the data points for a software CER 
were based on programs with lines of code ranging from 10,000 to 
250,000, it would be inappropriate to use the CER to estimate the new 
program. 

To develop a parametric CER, cost estimators must determine the cost 
drivers that most influence cost. After studying the technical baseline 
and analyzing the data through scatter charts and other methods, the 
cost estimator should verify the selected cost drivers by discussing 
them with engineers. The CER can then be developed with a mathematical 
expression, which can range from a simple rule of thumb (for example, 
dollars per pound) to a complex regression equation. 

The more simplified CERs include rates, factors, and ratios. A rate 
uses a parameter to predict cost, using a multiplicative relationship. 
Since rate is defined to be cost as a function of a parameter, the 
units for rate are always dollars per something. The rate most commonly 
used in cost estimating is the labor rate, expressed in dollars per 
hour. 

A factor uses the cost of another element to estimate a new cost using 
a multiplier. Since a factor is defined to be cost as a function of 
another cost, it is often expressed as a percentage. For example, 
travel costs may be estimated as 5 percent of program management costs. 

A ratio is a function of another parameter and is often used to 
estimate effort. For example, the cost to build a component could be 
based on the industry standard of 20 hours per subcomponent. 

Rates, factors, and ratios are often the result of simple calculations 
(like averages) and many times do not include statistics. Table 14 
contains a parametric cost estimating example. 

Table 14: An Example of the Parametric Cost Estimating Method: 
 
Program attribute: A cost estimating relationship (CER) for site 
activation (SA) is a function of the number of workstations (NW); 
Calculation: SA = $82,800 + ($26,500 x NW). 

Program attribute: Data range for the CER; 
Calculation: 7 – 47 workstations based on 11 data points. 

Program attribute: Cost to site activate a program with 40 
workstations; 
Calculation: $82,800 + ($26,500 x 40) = $1,142,800. 

Source: © 2003, Society of Cost Estimating and Analysis (SCEA), 
“Costing Techniques.” 

[End of table] 

In table 14, the number of workstations is the cost driver. The 
equation is linear but has both a fixed component (that is, $82,800) 
and a variable component (that is, $26,500 x NW). 

In addition, the range of the data is from 7 to 47 workstations, so it 
would be inappropriate to use this CER for estimating the activation 
cost of a site with as few as 2 or as many as 200 workstations. 

In fact, at one extreme, the CER estimates a cost of $82,800 for no 
workstation installations, which is not logical. Although we do not 
show any CER statistics for this example, the CERs should always be 
presented with their statistics. The reason for this is to enable the 
cost estimator to understand the level of variation within the data and 
model its effect with uncertainty analysis. 

CERs should be developed using regression techniques, so that 
statistical inferences may be drawn. To perform a regression analysis, 
the first step is to determine what relationship exists between cost 
(dependent variable) and its various drivers (independent variables). 
This relationship is determined by developing a scatter chart of the 
data. If the data are linear, they can be fit by a linear regression. 
If they are not linear and transformation of the data does not produce 
a linear fit, nonlinear regression can be used. The independent 
variables should have a high correlation with cost and should be 
logical. 

For example, software complexity can be considered a valid driver of 
the cost of developing software. The ultimate goal is to create a fit 
with the least variation between the data and the regression line. This 
process helps minimize the statistical error or uncertainty brought on 
by the regression equation. 

The purpose of the regression is to predict with known accuracy the 
next real-world occurrence of the dependent variable (or the cost), 
based on knowledge of the independent variable (or some physical, 
operational, or program variable). Once the regression is developed, 
the statistics associated with the relationship must be examined to see 
if the CER is a strong enough predictor to be used in the estimate. 
Most statistics can be easily generated with the regression analysis 
function of spreadsheet software. Among important regression statistics 
are: 

* R-squared, 

* statistical significance, 

* the F statistic, and, 

* the t statistic. 

R-squared: 

The R-squared (R2) value measures the strength of the association 
between the independent and dependent (or cost) variables. The R2 value 
ranges between 0 and 1, where 0 indicates that there is no relationship 
between cost and its independent variable, and 1 means that there is a 
perfect relationship between them. Thus, the higher R2 is the better. 
An R2 of 91 percent in the example in table 14, for example, would mean 
that the number of workstations (NW) would explain 91 percent of the 
variation in site activation costs, indicating that it is a very good 
cost driver. 

Statistical Significance: 

Statistical significance is the most important factor for deciding 
whether a statistical relationship is valid. An independent variable 
can be considered statistically significant if there is small 
probability that its corresponding coefficient is equal to zero, 
because a coefficient of zero would indicate that the independent 
variable has no relationship to cost. Thus, it is desirable that the 
probability that the coefficient is equal to zero be as small as 
possible. How small is denoted by a predetermined value called the 
significance level. For example, a significance level of .05 would mean 
there was a 5 percent probability that a variable was not statistically 
significant. Statistical significance is determined by both the 
regression as a whole and each regression variable. 

F Statistic: 

The F statistic is used to judge whether the CER as a whole is 
statistically significant by testing to see whether any of the 
variables’ coefficients are equal to zero. The F statistic is defined 
as the ratio of the equation’s mean squares of the regression to its 
mean squared error, also called the residual. The higher the F 
statistic is, the better the regression, but it is the level of 
significance that is important. 

t Statistic: 

The t statistic is used to judge whether individual coefficients in the 
equation are statistically significant. It is defined as the ratio of 
the coefficient’s estimated value to its standard deviation. As with 
the F statistic, the higher the t statistic is, the better, but it is 
the level of significance that is important. 

The Parametric Method: Further Considerations: 

The four statistics described above are just some of the statistical 
analyses that can be used to validate a CER. (For more information on 
statistics or hardware cost estimating, a good reference is the 
Parametric Estimating Handbook.[Footnote 40]) Once the statistics have 
been evaluated, the cost estimator picks the best CER—that is, the one 
with the least variation and the highest correlation to cost. 

The final step in developing the CER is to validate the results, using 
a data set different from the one used to generate the equation, to see 
if the results are similar. Again, it is important to use a CER 
developed from programs whose variables are within the same data range 
as those used to develop the CER. Deviating from the CER variable input 
range could invalidate the relationship and skew the results. We 
note several other pitfalls associated with CERs. 

Always question the source of the data underlying the CER. Some CERs 
may be based on data that are biased by unusual events like a strike, 
hurricane, or major technical problems that required a lot of rework. 
To mitigate this risk, it is essential to understand the data the CER 
is based on and, if possible, to use other historical data to check the 
validity of the results. 

All equations should be checked for common sense to see if the 
relationship described by the CER is reasonable. This helps avoid the 
mistake that the relationship adequately describes one system but does 
not apply to the one being estimated. 

Normalizing the data to make them consistent is imperative to good 
results. All cost data should be converted to constant base years. In 
addition, labor and material costs should be broken out separately, 
since they may require different inflation factors to convert them to 
constant dollars. Moreover, independent variables should be converted 
into like units for various physical characteristics such as weight, 
speed, and length. 

Historical cost data may have to be adjusted to reflect similar 
accounting categories, which might be expressed differently from one 
company to another. 

It is important to fully understand all CER modeling assumptions and to 
examine the reliability of the dataset, including its sources, to see 
if they are reasonable. 

Among the several advantages to parametric cost estimating are its:
 
* Versatility: If the data are available, parametric relationships can 
be derived at any level, whether system or subsystem component. And as 
the design changes, CERs can be quickly modified and used to answer 
what-if questions about design alternatives. 

* Sensitivity: Simply varying input parameters and recording the 
resulting changes in cost can produce a sensitivity analysis. 

* Statistical output: Parametric relationships derived from statistical 
analysis generally have both objective measures of validity 
(statistical significance of each estimated coefficient and of the 
model as a whole) and a calculated standard error that can be used in 
risk analysis. This information can be used to provide a confidence 
level for the estimate, based on the CER’s predictive capability. 

* Objectivity: CERs rely on historical data that provide objective 
results. This increases the estimate’s defensibility. 

Disadvantages to parametric estimating include: 

* Database requirements: The underlying database must be consistent and 
reliable. It may be time-consuming to normalize the data or to ensure 
that the data were normalized correctly, especially if someone outside 
the estimator’s team developed the CER. Without understanding how the 
data were normalized, the analyst has to accept the database on faith— 
sometimes called the black-box syndrome, in which the analyst simply 
plugs in numbers and unquestioningly accepts the results. Using a CER 
in this manner can increase the estimate’s risk. 

* Currency: CERs must represent the state of the art; that is, they 
must be updated to capture the most current cost, technical, and 
program data. 
 
* Relevance: Using data outside the CER range may cause errors, because 
the CER loses its predictive ability for data outside the development 
range.

* Complexity: Complicated CERs (such as nonlinear CERs) may make it 
difficult for others to readily understand the relationship between 
cost and its independent variables. 

Parametric Cost Models: 

Many cost estimating models are based on parametric methods. They may 
estimate hardware or software costs. Depending on the model, the 
database may contain cost, technical, and programmatic data at the 
system, component, and subcomponent level. Parametric models typically 
consist of several interrelated CERs and are often computerized. They 
may involve extensive use of cost-to-noncost CERs, multiple independent 
variables related to a single cost effect, or independent variables 
defined in terms of weapon system performance or design characteristics 
rather than more discrete material requirements or production 
processes. Information technology databases and computer modeling may 
be used in these types of parametric cost estimating systems. 

When using parametric models, many times the underlying data are 
proprietary, so access to the raw data may not be available. When the 
inputs to the parametric models are qualitative, as often happens, they 
should be objectively assessed. In addition, many parameters should be 
selected to tailor the model to the specific hardware or software 
product that is being estimated. Therefore, it is also important to 
calibrate the parametric model to best reflect the particular situation 
or environment in which the product will be developed. Finally, the 
model should be validated using historical data to determine how well 
it predicts costs. 

Parametric models are always useful for cross-checking the 
reasonableness of a cost estimate that is derived by other means. As a 
primary estimating method, parametric models are most appropriate 
during the engineering concept phase when requirements are still 
somewhat unclear and no bill of materials exists. When this is the 
situation, it is imperative that the parametric model is based on 
historical cost data and that the model is calibrated to those data. To 
ensure that the model is a good predictor of costs, it should 
demonstrate that it actually reflects or replicates known data to a 
reasonable degree of accuracy. In addition, the model should 
demonstrate that the cost-to-nocost estimating relationships are 
logical and that the data used for the parametric model can be verified 
and traced back to source documentation. 

Using parametric cost models has several advantages: 

* They can be adjusted to best fit the hardware or software being 
estimated. 

* Cost estimates are based on a database of historical data. 

They can be calibrated to match a specific development environment. 

Their disadvantages are that: 

* their results depend on the quality of the underlying database, 
 
* they require many inputs that may be subjective, and 

* accurate calibration is required for valid results. 

Expert Opinion: 

Expert opinion is generally considered too subjective but can be useful 
in the absence of data. It is possible to alleviate this concern by 
probing further into the experts’ opinions to determine if real data 
back them up. If so, the analyst should attempt to obtain the data and 
document the source. 

The cost estimator’s interviewing skills are also important for 
capturing the experts’ knowledge so that the information can be used 
properly. However, cost estimators should never ask experts to estimate 
the costs for anything outside the bounds of their expertise, and they 
should always validate experts’ credentials before relying on their 
opinions. 

The advantages of using an expert’s opinion are that: 

* it can be used when no historical data are available; 

* it takes minimal time and is easy to implement, once experts are 
assembled; 

* an expert may give a different perspective or identify facets not 
previously considered, leading to a better understanding of the 
program; 
 
* it can help in cross-checking for CERs that require data 
significantly beyond the data range; 

* it can be blended with other estimation techniques within the same 
WBS element; and; 

* it can be applied in all acquisition phases. 

Disadvantages associated with using an expert’s opinion include 

* its lack of objectivity, 

* the risk that one expert will try to dominate a discussion to sway 
the group or that the group will succumb to the urge to agree, and; 

* it is not very accurate or valid as a primary estimating method. 

The bottom line is that because of its subjectivity and lack of 
supporting documentation, expert opinion should be used sparingly and 
only as a sanity check. Case study 35 shows how relying on expert 
opinion as a main source for a cost estimate is unwise. 

Case Study 35: Expert Opinion, from Customs Service Modernization, 
GAO/AIMD-99-41: 

The U.S. Customs Service Automated Commercial Environment (ACE), a 
major information technology systems modernization effort, was 
estimated in November 1997 to cost $1.05 billion to develop, operate, 
and maintain between 1994 and 2008. GAO’s 1999 review found that the 
agency lacked a reliable estimate of what ACE would cost to build, 
deploy, and maintain. Instead of using a cost model, Customs had used 
an unsophisticated spreadsheet to extrapolate the cost of each ACE 
software increment. 

Further, Customs’ approach to determining software size and reuse was 
not well supported or convincing and had not been documented. For 
example, Customs had estimated the size of each ACE software 
increment—most increments had still been undefined—by extrapolating 
from the estimated size of the first increment, based on individuals’ 
undocumented best judgments about functionality and complexity. 

Last, Customs did not have any historical project cost data when it 
developed the $1.05 billion estimate, and it had not accounted for 
relevant, measured, and normalized differences in the increments. For 
instance, it had not accounted for the change in ACE’s architecture 
from a mainframe system that had been written in COBOL and C++ to a 
combined mainframe and Internet-based system that was to be written in 
C++ and Java. Such a fundamental change would clearly have a dramatic 
effect on system costs and should have been explicitly addressed in 
Customs’ cost estimates. 

Source: GAO, Customs Service Modernization: Serious Management and 
Technical Weaknesses Must Be Corrected, GAO/AMD-99-41, Washington, 
D.C.: Feb. 26, 1999. 

[End of case study] 

Other Estimating Methods: Extrapolation from Actual Costs: 
 
Extrapolation uses the actual past or current costs of an item to 
estimate its future costs. The several variants of extrapolation 
include: 

* averages, the most basic variant, a method that uses simple or moving 
averages to determine the average actual costs of units that have been 
produced to predict the cost of future units; 

* learning curves, which account for cost improvement and are the most 
common variant; and; 

* estimates at completion, which use actual cost and schedule data to 
develop estimates of costs at completion with EVM techniques; EACs can 
be calculated with various EVM forecast techniques to take into account 
factors such as current performance. 

Extrapolation is best suited for estimating follow-on units of the same 
item when there are actual data from current or past production lots. 
This method is valid when the product design or manufacturing process 
has changed little. If major changes have occurred, careful adjustments 
will have to be made or another method will have to be used. When using 
extrapolation techniques, it is essential to have accurate data at the 
appropriate level of detail, and the cost estimator must ensure that 
the data have been validated and properly normalized. When such data 
exist, they form the best basis for cost estimates. Advantages 
associated with extrapolating from actual costs include their 

* reliance on historical costs to predict future costs, 

* great credibility and reliability for estimating costs, and 

* ability to be applied at whatever level of data—labor hours, material 
dollars, total costs. 

The disadvantages associated with extrapolating from actual costs are 
that: 

* changes in the accounting of actual costs can affect the results, 

* obtaining access to actual costs can be difficult, 

* results will be invalid if the production process or configuration is 
not stable, and; 
 
* it should not be used for items outside the actual cost data range. 

Other Estimating Methods: Learning Curves: 

Using the cost estimating methods discussed in this chapter can 
generate the cost of a single item. However, a cost estimator needs to 
determine whether that cost is for the first unit, the average unit, or 
every unit. And given the cost for one unit, how should a cost 
estimator determine the appropriate costs for other units? The answer 
is in the use of learning curves. Sometimes called progress or 
improvement curves, learning curve theory is based on the premise that 
people and organizations learn to do things better and more efficiently 
when they perform repetitive tasks. A continuous reduction in labor 
hours from repetitive performance in producing an item often results 
from more efficient use of resources, employee learning, new equipment 
and facilities, or improved flow of materials. This improvement can be 
modeled with a mathematical CER that assumes that as the quantity of 
units to be produced doubles, the amount of effort declines by a 
constant percentage. 

Workers gain efficiencies in a number of areas as items are repeatedly 
produced. The most commonly recognized area of improvement is worker 
learning. Improvement occurs because as a process is repeated, workers 
tend to become physically and mentally more adept at it. Supervisors, 
in addition to realizing these gains, become more efficient in using 
their people, as they learn their strengths and weaknesses. 
Improvements in the work environment also translate into worker and 
supervisory improvement: Studies show that changes in climate, 
lighting, and general working conditions motivate people to improve. 

Cost improvement also results from changes to the production process 
that optimize placement of tools and material and simplify tasks. In 
the same vein, organizational changes can lead to lower recurring 
costs, such as instituting a just-in-time inventory or centralizing 
tasks (heat and chemical treatment processes, tool bins, and the like). 
Another example of organizational change is a manufacturer’s agreeing 
to give a vendor preferred status if it is able to limit defective 
parts to some percentage. The reduction in defective parts can 
translate into savings in scrap rates, quality control hours, and 
recurring manufacturing labor, all of which can result in valuable time 
savings. In general, it appears that more complex manufacturing tasks 
tend to improve faster than simpler tasks. The more steps in a process, 
the more opportunity there is to learn how to do them better and 
faster. 

Another reason for contractor improvement is that in competitive 
business environments, market forces require suppliers to improve 
efficiency to survive. As a result, some suppliers may competitively 
price their initial product release at a loss, with the expectation 
that future cost improvements will make up the difference. This 
strategy can also discourage competitors from entering new markets. For 
the strategy to work, however, the assumed improvements must 
materialize or the supplier may cease to exist because of high losses. 

In observing production data (for example, manufacturing labor hours), 
early analysts noted that labor hours per unit decreased over time. 
This observation led to the formulation of the learning curve equation 
Y = AXb and the concept of a constant learning curve slope (b) that 
captures the change in Y given a change in X.[Footnote 41] The unit 
formulation states that “as the number of units doubles, the cost 
decreases by a 
constant percent.” In other words, every time the total quantity 
doubles, the cost decreases by some fixed percentage. Figure 13 
illustrates how a learning curve works. 

Figure 13: A Learning Curve: 

[Refer to PDF for image: line graph]

Cumulative average hours per unit (as a percent of first unit) plotted 
against Cumulative number of units. 

Two lines: 
90% curve ratio; 
80% curve ratio. 

Source: © 1994, R. Max Wideman, FCSCE, “A Pragmatic Approach to Using 
Resource Loading, Production and Learning Curves on Construction 
Projects.” 

[End of figure] 

Figure 13 shows how an item’s cost gets cheaper as its quantities 
increase. For example, if the learning curve slope is 90 percent and it 
takes 1,000 hours to produce the first unit, then it will take 900 
hours to produce the second unit. Every time the quantity doubles—for 
example, from 2 to 4, 4 to 8, 8 to 16—the resource requirements will 
reduce according to the learning curve slope. 

Determining the learning curve slope is an important effort and 
requires analyzing historical data. If several production lots of an 
item have been produced, the slope can be derived from the trend in the 
data. Another way to determine the slope would be to look at company 
history for similar efforts and calculate it from those efforts. Or the 
slope could be derived from an analogous program. The analyst could 
look at slopes for a particular industry—aircraft, electronics, 
shipbuilding—sometimes reported in organizational studies, research 
reports, or estimating handbooks. Slopes can be specific to functional 
areas such as manufacturing, tooling, and engineering, or they may be 
composite slopes calculated at the system level, such as aircraft, 
radar, tank, or missiles. 

The first unit cost might be arrived at by analogy, engineering build-
up, a cost estimating relationship, fitting the actual data, or another 
method. In some cases, the first unit cost is not available. Sometimes 
work measurement standards might provide the hours for the 5th unit, or 
a cost estimating relationship might predict the 100th unit cost. This 
is not a problem as long as the cost estimator understands the point on 
the learning curve that the unit cost is from and what learning curve 
slope applies. With this information, the cost estimator can easily 
solve for the 1st unit cost using the standard learning curve formula Y 
= AXb. 

Because learning can reduce the cost of an item over time, cost 
estimators should be aware that if multiple units are to be bought from 
one contractor as part of the program’s acquisition strategy, reduced 
costs can be anticipated. Thus, knowledge of the acquisition plan is 
paramount in deciding if learning curve theory can be applied. If so, 
careful consideration must be given to determining the appropriate 
learning curve slope for both labor hours and material costs. In 
addition, learning curves are based on recurring costs, so cost 
estimators need to separate recurring from nonrecurring costs if the 
results are not to be skewed. Finally, these circumstances should be 
satisfied before deciding to use learning curves:[Footnote 42]  
 
* much manual labor is required to produce the item; 

* the production of items is continuous and, if not, then adjustments 
are made; 

* the items to be produced require complex processes; 

* technological change is minimal between production lots; 

* the contractor’s business process is being continually improved; and; 

* the government program office culture (or environment) is 
sufficiently known. 

Particular care should be taken for early contracts, in which the cost 
estimator may not yet be familiar enough with program office habits to 
address the risk accurately (for example, high staff turnover, 
propensity for scope creep, or excessive schedule delays). 

Production Rate Effects On Learning: 
 
It is reasonable to expect that unit costs decrease not only as more 
units are produced but also as the production rate increases. This 
theory accounts for cost reductions that are achieved through economies 
of scale. Some examples are quantity discounts and reduced ordering, 
processing, shipping, receiving, and inspection costs. Conversely, if 
the number of quantities to be produced decreases, then unit costs can 
be expected to increase, because certain fixed costs have to be spread 
over fewer items. At times, an increase in production rate does not 
result in reduced costs, as when a manufacturer’s nominal capacity is 
exceeded. In such cases, unit costs increase because of such factors as 
overtime, capital purchases, hiring actions, and training costs. 

Another aspect of improvement is the continuity of the production line. 
Production breaks may occur because of program delays (budgetary or 
technical), time lapses between initial and follow-on orders, or labor 
disputes. They may occur as a result of design changes that may require 
a production line to shut down so it can be modified with new tools and 
equipment or a new configuration. Production lines can also shut down 
for unexpected recalls that require repairs for previously produced 
items. How much learning is lost depends on how long the production 
line is shut down. 

To determine the effect of a production break on the unit cost two 
questions need answering: 

1. How much learning has been lost (or forgotten) because of the break 
in production? 

2. How will this loss of learning affect the costs of future production 
items? 

The cost estimator should always consider the effect of a production 
break on the cost estimate. (See case study 36.) 

Case Study 36: Production Rate, from Defense Acquisitions, GAO-05-183: 

Costs on the CVN 76 and CVN 77 Nimitz aircraft carriers grew because of 
additional labor hours required to construct the ships. At delivery, 
CVN 76 had required 8 million additional labor hours to construct; CVN, 
77, 4 million. As the number of hours increased, total labor costs grew 
because the shipbuilder was paying for additional wages and overhead 
costs. Increases in labor hours stemmed in part from underestimating 
the labor hours. The shipbuilder had negotiated CVN 76 for 
approximately 39 million labor hours—only 2.7 million more labor hours 
than the previous ship—CVN 75. However, CVN 75 had been constructed 
more efficiently, because it was the fourth ship of two concurrent ship 
procurements. CVN 76 and CVN 77, in contrast, were procured as single 
ships. 

Single ship procurements have historically been less efficient than two-
ship procurements. The last time the Navy procured a carrier as a 
single-ship procurement, 7.9 million more hours were required—almost 3 
times the number estimated for CVN 76 (2.7 million more hours). In 
addition, a 4-month strike in 1999, during the construction of CVN 76, 
had led to employee shortages in key trades and learning losses, 
because many employees were not returning to the shipyard. According to 
Navy officials, the shipbuilder was given $51 million to offset the 
strike’s effect. 
 
Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, GAO-05-183, 
Washington, D.C.: Feb. 28, 2005. 

[End of case study] 

Pulling The Point Estimate Together: 

After each WBS element has been estimated with one of the methods 
discussed in this chapter, the elements should be added together to 
arrive at the total point estimate. The cost estimator should validate 
the estimate by looking for errors like double-counting and omitted 
costs. The cost estimator should also perform, as a best practice, 
cross-checks on various cost drivers to see if similar results can be 
produced. This helps validate the estimate. The cost estimator should 
also compare the estimate to an independent cost estimate. The estimate 
and the independent cost estimate should also be reconciled at this 
time. (Chapter 15 discusses validating the estimate.) 

DOD’s major defense acquisition programs are required to develop 
independent cost estimates for major program milestones; other agencies 
may not require this practice. An independent cost estimate gives an 
objective measure of whether the point estimate is reasonable. 
Differences between them should be examined and discussed to achieve 
understanding of overall program risk and to adjust risk around the 
point estimate. 

Finally, as the program matures through its life cycle, as more data 
become available, or as changes occur, the cost estimator should update 
the point estimate. The updated point estimate should be compared 
against previous estimates, and lessons learned should be documented. 
(More detail is in chapter 20.) 

8. Best Practices Checklist: Developing a Point Estimate: 

* The cost estimator considered various cost estimating methods: 
- Analogy, early in the life cycle, when little was known about the 
system being developed: 
-- Adjustments were based on program information, physical and 
performance characteristics, contract type. 
- Expert opinion, very early in the life cycle, if an estimate could be 
derived no other way. 
- The build-up method later, in acquisition, when the scope of work was 
well defined and a complete WBS could be determined. 
- Parametrics, if a database of sufficient size, quality, and 
homogeneity was available for developing valid CERs and the data were 
normalized correctly. 
-- Parametric models were calibrated and validated using historical 
data. 
- Extrapolating from actual cost data, at the start of production. 

* Cost estimating relationships were considered: 
 Statistical techniques were used to develop CERs: 
-- Higher R-squared; 
- Statistical significance, for determining the validity of statistical 
relationships; 
-- Significance levels of F and t statistics. 
- Before using a CER, the cost estimator 
-- Examined the underlying data set to understand anomalies; 
-- Checked equations to ensure logical relationships; 
-- Normalized the data; 
-- Ensured that CER inputs were within the valid dataset range; 
-- Checked modeling assumptions to ensure they applied to the 
program. 
- Learning curve theory was applied if: 
-- Much manual labor was required for production; 
-- Production was continuous or adjustments had to be made; 
-- Items to be produced required complex processes; 
-- Technological change was minimal between production lots; 
-- The contractor’s business process was being continually improved. 

* Production rate and breaks in production were considered. 

* The point estimate was developed by aggregating the WBS element cost 
estimates by one of the cost estimating methods. 
- Results were checked for accuracy, double-counting, and omissions and 
were validated with cross-checks and independent cost estimates. 

[End of Chapter 11] 

Chapter 12: Estimating Software Costs: 

Software is a key component in almost all major systems the federal 
government acquires. Estimating software development, however, can be 
difficult and complex. To illustrate, consider some statistics: a 
Standish Group International 2000 report showed that 31 percent of 
software programs were canceled, more than 50 percent overran original 
cost estimates by almost 90 percent, and schedule delays averaged 
almost 240 percent.[Footnote 43] Moreover, the Standish Group reported 
that the number of software development projects that are completed 
successfully on time and on budget, with all features and functions as 
originally specified, rose only from 16 percent in 1994 to 28 percent 
in 2000.[Footnote 44] 

Most often, creating an estimate based on an unachievable schedule 
causes software cost estimates to be far off target. Playing into this 
problem is an overwhelming optimism about how quickly software can 
be developed. This optimism stems from a lack of understanding of how 
staffing, schedule, software complexity, and technology all 
interrelate. Furthermore, optimism about how much savings new 
technology can offer and the amount of reuse that can be leveraged from 
existing programs also cause software estimates to be underestimated. 
Case study 37 gives an example. 

Case Study 37: Underestimating Software, from Space Acquisitions, 
GAO-07-96: 

The original estimate for the Space Based Infrared System for 
nonrecurring engineering, based on actual experience in legacy sensor 
development and assumed software reuse, was significantly 
underestimated. Nonrecurring costs should have been two to three times 
higher, according to historical data and independent cost estimators. 
Program officials also planned on savings from simply rehosting 
existing legacy software, but those savings were not realized because 
all the software was eventually rewritten. It took 2 years longer than 
planned to complete the first increment of software. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Action to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96 
(Washington, D.C.: Nov. 17, 2006). 

[End of case study] 

Our work has also shown that the ability of government program offices 
to estimate software costs and develop critical software is often 
immature. Therefore, we highlight software estimation as a special case 
of cost estimation because of its significance and complexity in 
acquiring major systems. This chapter supplements the steps in cost 
estimating with what is unique in the software development environment, 
so that auditors can better understand the factors that can lead to 
software cost overruns and failure to deliver required functionality on 
time. Auditors should remember that all the steps of cost estimating 
have to be performed for software just as they have to be performed for 
hardware.

The 12 steps of cost estimating described in chapter 1 and summarized 
in table 15 also apply to software. That is, the purpose of the 
estimate and the estimating plan should be defined in steps 1 and 2, 
software requirements should be defined in step 3, the effort to 
develop the software should be defined in step 4, GR&As should be 
established in step 5, relevant technical and cost data should be 
collected in step 6, and a method for estimating the cost for software 
development and maintenance should be part of the point estimate in 
step 7. Moreover, sensitivity analysis in step 8, risk and uncertainty 
analysis in step 9, documenting the estimate in step 10, presenting 
results to management in step 11, and updating estimates with actual 
costs in step 12 are all relevant for software cost estimates. 

Table 15: The Twelve Steps of High-Quality Cost Estimating Summarized: 
 
Step: 1; 
Summary: Define the estimate’s purpose. 
Step: 2; 
Summary: Develop the estimating plan. 

Step: 3; 
Summary: Define the program characteristics, the technical baseline. 

Step: 4; 
Summary: Determine the estimating structure, the WBS. 

Step: 5; 
Summary: Identify ground rules and assumptions. 

Step: 6; 
Summary: Obtain the data. 

Step: 7; 
Summary: Develop the point estimate and compare it to an independent 
cost estimate. 

Step: 8; 
Summary: Conduct sensitivity analysis. 

Step: 9; 
Summary: Conduct a risk and uncertainty analysis. 

Step: 10; 
Summary: Document the estimate. 

Step: 11; 
Summary: Present the estimate to management for approval. 

Step: 12; 
Summary: Update the estimate to reflect actual costs and changes. 

Source: GAO. 

[End of table] 

In this chapter, we discuss some of the best practices for developing 
reliable and credible software cost estimates and fully understanding 
typical cost drivers and risk elements associated with software 
development. 

Unique Components Of Software Estimation: 
 
Since software is not tangible like hardware, it can be more ambiguous 
and difficult to comprehend. In addition, software is built only once, 
whereas hardware is often mass produced, once design and testing 
are complete. Unlike hardware, for which the industry changes more 
slowly, software changes constantly, making it difficult to collect 
good data for cost estimating. Despite these differences, software 
estimating is otherwise similar to hardware estimating in that it 
follows the same basic development process.[Footnote 45] For 
instance, both use the same types of estimating methods—analogy, 
engineering build-up, parametric. 

Size and complexity are cost drivers for both. Finally, how quickly 
hardware and software can be produced depends on the developer’s 
capability, available resources, and familiarity with the environment.

Software is mainly labor intensive, and all the tasks associated with 
developing it are nonrecurring—there is no production phase. That is, 
once the software is developed, it is simple to produce a copy of it. 
How much effort is required to develop software depends on its size and 
complexity. Thus, estimating software costs has two basic elements—the 
software to be developed and the development effort to accomplish it. 

Estimating Software Size: 

Cost estimators begin a software estimate by predicting the sizes of 
the deliverables that must be constructed. Software sizing is the 
process of determining how big the application being developed will be. 
The size depends on many factors. For example, software programs that 
are more complex, perform many functions, have safety-of-life 
requirements, and require high reliability are typically bigger than 
simpler programs. 

Estimating software size is not easy and depends on having a detailed 
knowledge about a program’s functions in terms of scope, complexity, 
and interactions. Not only is it hard to generate a size estimate for 
an application that has not yet been developed, but the software 
process also often experiences requirements growth and scope creep that 
can significantly affect size and the resulting cost and schedule 
estimates. 

Programs that do not track and control these trends typically overrun 
their costs and experience schedule delays. Methods for measuring size 
data include COSMIC (Common Software Measurement International 
Consortium) Functional Sizing Method, function point analysis, object 
point analysis, source lines of code, and use case (described in table 
16). 

Table 16: Sizing Metrics and Commonly Associated Issues: 
 
COSMIC functional sizing: 

Metric: Measures the size of software based on functional user 
requirements; sizes software independently of the technology to be used 
to implement it, focusing on practices and procedures the software must 
follow to meet user needs. COSMIC points are based on four different 
data movements: entry, exit, read, and write. Each one constitutes a 
COSMIC function point. 
The method can be used to determine the software size of various 
applications including business, real-time (telecommunications, process 
control), embedded software (cellular phones, electronics), and 
infrastructure software (operating system software). 
Advantages: Sizing is easily understood and simplified because all data 
movements have the same value; sizing does not depend on data 
attributes; It applies to real-time and embedded systems and allows for 
end-user and developer viewpoints; standards exist for counting.
Disadvantages: Recently developed, so benchmarking data are limited; 
not accurate for counting highly algorithmic software; detailed 
information about data movements takes time to collect; automated 
counting does not exist. 

Metric: Function point analysis; Considers how many functions a program
does rather than how many instructions it contains; functions typically 
include user inputs (add, change, delete), outputs (reports), data 
files to be updated by the application, interfaces with other 
applications, and inquiries (searches or retrievals). Each function is 
weighted for complexity and total count is adjusted for the effect of 
14 characteristics such as data communications, transaction rate, 
installation ease, and whether there are multiple sites. Accurate 
counting requires in-depth knowledge of standards, experience, and, 
preferably, function point certification. Function point analysis is
linked directly to system requirements and functionality, so size 
analysis is measured in terms users can understand. The size estimates 
(and resulting cost and schedule estimates) can be based on quantifiable
analysis through the project life cycle as requirements change. 
Function points are particularly useful in many development 
environments that might use unified modeling language, commercial off-
the-shelf components, or object-oriented approaches to software 
development and implementation. 
Advantages: Many types of data sources can be used throughout 
development: user or estimator interviews, requirements and design 
documents, data dictionaries and models, end user guides, screen 
captures; not dependent on language or technology; count is unaffected 
by language or tools used to develop the software; counts are available
early in development from requirements and design specifications; 
nontechnical users can understand what function points are measuring;
function points can be used to determine requirements (or scope creep); 
counts are fully documented and auditable; standards are established
and reviewed often by the International Function Point Users Group; 
counting can be quick and efficient. 
Disadvantages: Counting involves subjectivity; difficult to derive 
requirements from top-level specifications; does not capture technical 
and design constraints; untrained or inexperienced people can develop
inconsistent function point counts; definitions can be confusing; 
automated function point analysis counting does not exist; database is 
not as big as for source line of code counts; counts tend to 
underestimate algorithmic intensive systems. 

Metric: Object point analysis: Uses integrated computer-aided software 
engineering tools (CASE) to count number of screens, reports, and third-
generation modules for basic sizing; CASE tools take over the job of 
manually writing software code by using graphical user interface 
generators, libraries of reusable components, and other design tools. 
Object points focus on actors involved in the solution and any actions 
they must take. One benefit of using objects (i.e., actors) is that 
similar behaviors can be grouped into classes, allowing for behaviors 
from upper classes (parent) to be inherited by lower classes 
(children). Inheritance results in reduced coding effort; each count is 
weighted for complexity, summed to a total count, and adjusted for 
reuse. 
Advantages: Relies on a graphical user interface; automates manual 
activities; objective measures; easier calculations; accounts for reuse 
through inheritance. 
Disadvantages: Counts occur at the end of design; no standards for 
counting; and not widely used and therefore validated productivity 
metrics are not available. 

Metric: Reports, interfaces, conversions, extensions, and 
forms/workflows (RICEF/W); Commonly used to size the effort associated 
with implementing Enterprise Resource Planning (ERP) systems; 
identifies changes that need to be made to configure the ERP system so 
that it satisfies user needs and fits within the target operating 
environment. Can be used to add functionality through custom 
development. RICEF/W needs to be adjusted for complexity. 
Advantages: Represents ERP modifications and enhancements that do not 
require custom development; 
Disadvantages: Specific to ERP systems; no standards for counting; does 
not capture costs for integrating bolt-on functionality. 

Metric: Source lines of code (SLOC): Considers the volume of code 
required to develop the software; includes executable instructions and 
data declarations and normally excludes comments and blanks. Estimation 
is by analogy, engineering expertise, or automated code counters. SLOC 
sizing is particularly appropriate for projects preceded by similar ones
(e.g., same language, developers, type of application); helps ensure 
that experience is aligned to future development. When developing lines 
of code counts, it is critical to define what is and is not included.
When developing databases or relying on software cost models, 
consistency in defining what the lines of code include is key. 
Advantages: Widely used for many years; can be used to estimate real 
time systems easily counted, manually or by automated code counter; 
objective; large databases of historical program sizes are available;
can obtain precise counts of existing software using the USC Code 
Counter. 
Disadvantages: No standard definition of what should be counted as lines
of code (e.g., physical line vs. logical statement); different lines of 
code count for the same function, depending on language and programmer’s
style; hard to capture lines of code for commercial off-the-shelf 
systems; hard to translate lines of code counts between other 
programming languages such as object oriented code; variations in 
definition make it hard to compare studies using SLOC; hard to estimate 
program SLOC early; emphasizes coding effort, which is small compared to
overall software development effort. 

Metric: Use cases and use case points: Defines interactions between 
external users and the system to achieve a goal (e.g., capture 
fingerprint or facial biometric to enroll applicants). A use case model 
describes a system’s functional requirements, consists of all users and 
use cases (tasks performed by the end user of a system that has a 
useful outcome), and identifies reuse by use case inclusions and 
extensions. Sizing count is arrived at by categorizing use cases as 
small, medium, or large and applying an average “use case points per 
category.” Adding a complexity factor to the sizing count based on 
number and types of users and transactions improves the count accuracy. 
Advantages: Applies to interactive end-user applications and devices 
users interact with; intuitive to stakeholders and development team;
identifies opportunities for software reuse; traceable to development 
team’s plans and output; increasingly applied to real-time systems;
can be mapped to test cases and business scenarios, which helps in 
staggered deployment. 
Disadvantages: Often yields an inaccurate final estimate if the system
engineering process is immature and historical data are lacking; no 
standards for counting; developer must be using object oriented design 
techniques so required documentation is available; estimate cannot be 
done until design document with the defined use case is available; 
requires a design team with a great deal of experience with object 
oriented design. 

Source: DOD, NASA, SCEA, and industry. 

[End of table] 

While software sizing can be approached in many ways, none are accurate 
because the “size” of software is an abstract concept. Moreover, with 
the exception of COSMIC and function points, none of the methods table 
16 describes has a controlling body for internationally standardizing 
the counting rules. In the absence of a universal counting convention, 
different places may take one of the source definitions for the basic 
approach and then “standardize” the rules internally. This can result 
in different counts. Therefore, it is critical that the sizing method 
used is consistent. The test of a good sizing method is that two 
separate individuals can apply the same rules to the same problem and 
yield almost the same result. Before choosing a sizing approach, one 
must consider the following questions of maturity and applicability: 
 
* Are the rules for the sizing technique rigorously defined in a widely 
accepted format? 

* Are they under the control of a recognized, independent controlling 
body? 

* Are they updated from time to time by the recognized, independent 
controlling body? 

* Does the controlling body certify the competency (and, hence, 
consistency) of counters who use their rules? 

* Are statistical data available to support claims for the consistency 
of counting by certified counters? 

* How long have the rules been stable? 

Auditors should know a few things about software sizing. The first is 
that reused and autogenerated software source lines of code should be 
differentiated from the total count. Reused software (code used 
verbatim with no modifications), adapted software (code that needs to 
be redesigned, may need to be converted, and may need some code added), 
and autogenerated software provide the developer with code that can be 
used in a new program, but none of these comes for free, and additional 
effort is usually associated with incorporating them into a new 
program. For instance, the effort associated with reused code depends 
on whether significant integration, reverse engineering, and additional 
design, validation, and testing are required. But if the effort to 
incorporate reused software is too great, it may be cheaper to write 
the code from scratch. As a result, the size of the software should 
reflect the amount of effort expected with incorporating code from 
another source. This can be accomplished by calculating the equivalent 
source lines of code, which adjusts the software size count to reflect 
the fact that some effort is required. 

Software porting is a special case of software reuse that is getting 
increasing visibility in cost estimation with respect to specific 
technologies, such as communications systems (waveforms). Porting 
represents hidden pitfalls, depending on the amount of capability to be 
transferred from special purpose processors (such as field-programmable 
gate arrays). Also, the quality of software commenting and 
documentation and the modularity of the initial code’s design and 
implementation greatly affect the porting of standard code in general 
purpose processors. Therefore, assumptions regarding savings (for 
example, assume less effort is required and no testing is necessary) 
from reused, adapted, and autogenerated software code should be looked 
at skeptically because of the additional work to research the code and 
provide necessary quality checks. As a minimum, regression testing will 
be required before integrating the software with the hardware for this 
type of code. 

Second, while function points generate counts for real-time software, 
like missile systems, they are not optimal in capturing the complexity 
associated with high levels of algorithmic software. Therefore, for 
programs that require high levels of complex processing like operating 
systems, telephone switching systems, navigation systems, and process 
control systems, estimators should base the count on COSMIC points or 
SLOC rather than function points to adequately capture the additional 
effort associated with developing algorithmic software.

Finally, choosing a sizing metric depends on the software application 
(purpose of the software and level of reliability needed) and the 
information that is available. Since no one way is best, cost 
estimators should work with software engineers to determine which 
metric is most appropriate. Since SLOCs have been used widely for years 
as a software sizing metric, many organizations have databases of 
historical SLOC counts for various completed programs. Thus, source 
lines of code tend to be the most predominant method for sizing 
software. If the decision is made to use historical source lines of 
code for estimating software size, however, the cost estimator needs to 
make sure that the program being estimated is similar in size, 
language, and application to the historical data. For programs for 
which no analogous data are available but detailed requirements and 
specifications have been developed, function point counting is 
appropriate, as long as the software does not contain many algorithms; 
if it does, then COSMIC points or SLOC should be used. And, if computer-
assisted software engineering tools are being used to develop the 
software, then object point analysis is appropriate. No matter which 
metric is chosen, however, the actual results can vary widely from the 
estimate, so that any point estimate should be accompanied by an 
estimated range of probability. (We discuss software and other cost 
estimating risk and uncertainty analyses in chapter 14.) 

When completing a software size estimate, it is preferable to use two 
different methodologies, if available, rather than relying on a single 
approach. Software estimates based on several different approaches that 
are compared and merge toward a consensus is the best practice. In 
addition, it is extremely important to include the expected growth in 
software size from requirements growth or underestimation (that is, 
optimism). Adjusting the software size to reflect expected growth from 
requirements being refined, changed, or added or initial size estimates 
being too optimistic and less reuse than expected is a best practice. 
This growth adjustment should be made before performing an uncertainty 
analysis (discussed in chapter 14). Understanding that software will 
usually grow, and accounting for it by using historical data, will 
result in more accurate software sizing estimates. Moreover, no matter 
what sizing convention is used, it is a best practice to continually 
update the size estimate as data become available so that growth can be 
monitored and accounted for. 

Estimating Software Development Effort: 

Once the initial software sizing is complete, it can be converted into 
software development effort—that is, an estimate of the human resources 
needed for the software’s development. It is important to note whether 
the effort accounts only for the WBS elements associated with the 
actual development of the software or also includes all the other 
nondevelopment activities. 

Table 53 in appendix IX, for example, shows a typical WBS for ground 
software development. The table shows that many other activities 
outside the actual coding of software are part of a typical software 
acquisition. These activities should also be estimated as part of the 
development effort. In particular, software management and control, 
software systems engineering, test-bed development, system integration 
and testing, quality assurance, and training are all activities that 
should be performed in a customized software solution acquisition. 

The level of effort required for each activity depends on the type of 
system being developed. For example, military and systems software 
programs require more effort than Web programs of the same size. Since 
variations in activities can affect overall costs, schedules, and 
productivity rates by significant amounts, it is critical to 
appropriately match activities to the type of software project being 
estimated. For example, safety critical software applications composed 
of complex mathematical algorithms require higher levels of effort 
because stringent quality and certification testing must be satisfied. 
Moreover, operating systems that must reflect real time updates and 
great reliability will need more careful design, development, and 
testing than software systems that rely on simple calculations. 

To convert software size into software development effort, the size is 
usually divided by a productivity factor like number of source lines of 
code, or function points, developed per labor work month. The 
productivity factor depends on several aspects, like the language used; 
whether the code is new, reused, or autogenerated; the developer’s 
capability; and the development tools used. It is best to use 
historical data from a similar program to develop the productivity 
factor, so that it best represents the development environment. If 
historical productivity factors are not available, an estimator can use 
a factor based on industry averages, but this will add more uncertainty 
to the estimate. It is important to note, however, that a productivity 
factor—based on the coding phase only—cannot be used to estimate the 
entire software development effort. When a productivity factor is used, 
all parameters associated with its computation need to be considered. 
Once the productivity factor has been selected, the corresponding labor 
hours can be generated. 

Some considerations in converting labor hours to cost are, first, that 
a cost estimator needs to determine how many productive hours are being 
assumed in a typical developer’s work day. This is important because 
assuming 8 hours of productive coding is unrealistic: staff meetings 
and training classes cut into valuable programming time, so that the 
number of effective work hours per day is typically 6 hours rather than 
8. Further, the number of work days per year is not the same from 
company to company because of differences in vacation and sick leave 
offered and the country the developers live in. In the United States, 
fewer vacation days tend to be provided than in countries in Europe, 
but in other countries like Japan less time is provided. All these 
issues need to be considered and calibrated to the program being 
estimated. In fact, multiple studies on the impact of overtime have 
shown that except for a short increase in effort over the first 1 or 2 
months, overtime does not have a significant impact on the life of the 
program. 

The sizing value usually represents only the actual software 
development effort, so the cost estimator needs to use other methods to 
estimate all the other activities related to developing the software. 
Sometimes factors (such as percentage of development effort) are 
available for estimating these additional costs. Software cost 
estimating models often provide estimates for these activities. If a 
model is not used or not available, then the cost estimator must 
account for the cost of the other labor as well as nonlabor costs, such 
as hardware and licenses. Accurately estimating all these tasks is 
challenging, because they are affected by a number of risks. (Some are 
identified in table 17; appendix XV contains a more comprehensive 
list of risks.) 

Table 17: Common Software Risks That Affect Cost and Schedule: 

Risk: Sizing and technology; 
Typical cost and schedule element: 
* Overly optimistic software engineers tending to underestimate the 
amount of code needed; 
* Poor assumptions on the use of reused code (requiring no 
modification) or adapted code (requiring some redesign, recoding, and 
retesting); 

* Vague or incomplete requirements, leading to uncertain size counts; 

* Not planning for additional effort associated with commercial off-the-
shelf software (e.g., systems engineering, performance testing, 
developing glue code). 

Risk: Complexity; 
Typical cost and schedule element: 
* Programming language: the amount of design, coding, and testing 
(e.g., object-oriented languages require more up-front design but 
result in less coding and testing); 
* Applications: software purpose and reliability (e.g., criticality of 
failure, loss of life); 
* Hardware limitations with respect to the need for more efficient 
code; 
* Number of modules affecting integration effort; 
* Amount of new code to be developed; 
* Higher quality requiring more development and testing but resulting 
in less and easier-to-perform maintenance; 
* Safety critical software requires more design, coding, and testing. 
 
Risk: Capability; 
Typical cost and schedule element: 
* Developers with better skill can deliver more effective software with 
fewer defects, allowing for faster software delivery; 
* Optimistic assumption that a new development tool will increase 
productivity; 
* Optimistic assumption about developer’s productivity, leading to cost 
growth, even if sizing is accurate; 
* Geographically dispersed development locations, making communication 
and coordination more difficult. 
 
Risk: Management and executive oversight; 
Typical cost and schedule element: 
* Management’s dictating an unrealistic schedule; 
* A decision to concurrently develop hardware and software, increasing 
risk; 
* Incorporating a new method, language, tool, or process for the first 
time; 
* Incomplete or inaccurate definition of system requirements; 
* Not handling creeping requirements proactively; 
* Inadequate quality control, causing delays in fixing unexpected 
defects; 
* Unanticipated risks associated with commercial off-the-shelf software 
upgrades and lack of support. 

Source: SCEA and industry. 

[End of table] 

Scheduling Software Development: 

The schedule for getting the work accomplished should also be 
estimated. Too often, software development programs tend to run late 
because of requirements creep or poor quality control. Other times, the 
schedule is driven by some arbitrary date dictated by management or the 
customer. Optimism may be based on management’s thinking that if more 
people are added to the development team, the product can be developed 
faster. Unfortunately, the opposite usually happens: the larger the 
development team, the less its members are able to communicate with one 
another or work effectively. In addition, the more complex the software 
development effort is, the harder it will be to find the right staff 
for the job. Scheduling is complicated and is affected by many factors. 
A cost estimator should understand the intricate interdependencies that 
affect the schedule: 
 
* staff availability; 

* an activity’s dependence on prior tasks; 

* the concurrence of scheduled activities; 
 
* the activities that make up the critical path; 

* the number of shifts working and effective work hours per shift; 

* available budget; 

* whether overtime can be authorized; 

* downtime from meetings, travel, sickness; 

* geographic location of workers, including time zones. 

Significantly large software development efforts frequently experience 
cost and schedule growth. This is because of the complexities inherent 
in managing configuration, communications, and design assumptions that 
typically hinder software development productivity. In addition, 
increased software schedule has a ripple effect on other collateral 
support efforts such as program management and systems engineering. 
Hardware programs experience the same problems. 

Management pressure on software developers to keep to an unrealistic 
schedule presents other problems. For example, to meet schedule 
constraints, the developer may minimize the time for requirements 
analysis, which can affect the quality of the software developed. In 
addition, developers may skip documentation, which could result in 
higher software maintenance costs. Moreover, developers may decide to 
build more components in parallel, defer functionality, postpone 
rework, or minimize functional testing, all to reduce schedule time. 
While these actions may save some time up front, they result in 
additional time, effort, and risk for the program. 

Rework should be included in every software development schedule 
because it is unwise to assume that software can be delivered without 
any defects. Therefore, if rework is not accounted for in the schedule, 
it will have to be accounted for when it occurs, which will cause 
problems in the sequencing of remaining tasks. It should be noted that 
if a software schedule does not include effort for rework, then the 
schedule will be unexecutable, and the maturity of the developing 
organization is questionable for assuming that all requirements will 
pass testing the very first time. Rework effort should include the time 
and resources associated with diagnosing the problem, designing and 
coding the fix, and then retesting until the problem is resolved. To 
adequately account for rework, the schedule should anticipate a certain 
number of defects based on historical experience, and time and effort 
should be allocated for fixing them. We discuss scheduling more 
thoroughly in chapter 18, including how to account for these risks so 
that schedule is realistic. 

Software Maintenance: 

Once the software has been developed, tested, and installed in its 
intended location, it must be maintained, just like hardware. Often 
called the operational phase for software, its costs must be accounted 
for in the LCCE. During this phase, software is maintained by fixing 
any defects not discovered in testing (known as corrective 
maintenance), modifying the software to work with any changes to its 
physical environment (adaptive maintenance), and adding new 
functionality (perfective maintenance). When adding capability, the 
effort is similar to a minidevelopment effort and the cost drivers are 
the same as in development. Software maintenance may also be driven by 
technology upgrades (adaptive maintenance) and users requesting 
enhancements (perfective maintenance). In addition to providing help 
desk support to users of the software, perfective maintenance often 
makes up the bulk of the software maintenance effort. 

The level of maintenance required depends on several factors. How 
complex the software is will determine how much maintenance is needed. 
In addition, if requirements from development were deferred until the 
software was in maintenance mode, or the requirements were too vague 
and not well understood, then additional perfective maintenance will be 
necessary. The quality of the developed software will also affect 
maintenance. If the software was rigorously tested, then less 
corrective maintenance will be needed. In addition, software that is 
well documented will be easier to de-bug and will provide maintainers a 
better understanding of how the software was designed, making 
modifications more streamlined. 

In addition to the need to maintain the software code, costs are 
associated with help desk support that need to be included in the 
software’s operation and support phase. Effort will be spent on trouble 
calls and generating defect tickets for software maintenance and should 
be included as part of the software cost estimate. 

Parametric Software Estimation: 

Software development cost estimating tools—or parametric tools—can be 
used to estimate the cost to develop and maintain software. Parametric 
tools are based on historical data collected from hundreds of actual 
projects that can generate cost, schedule, effort, and risk estimates 
based on inputs provided by the tool user. Among other things, these 
inputs generally include the size of the software, personnel 
capabilities, experience, development environment, amount of code 
reuse, programming language, and labor rates. Once the data have been 
input, the tool relies on cost estimating relationships and analogies 
to past projects to calculate the software cost and schedule estimates. 
When these data are not available to the cost estimator, most tools 
have default values that can be used instead. 

Parametric tools should be used throughout the development life cycle 
of the software. They are especially beneficial in the early stages of 
the software life cycle, when requirement specifications and design are 
still vague. For example, these tools provide flexibility by accepting 
multiple sizing metrics, so that estimators can apply different sizing 
methods and examine the results. Additionally, parametric-based 
estimates can be used to understand tradeoffs by analyzing the relative 
effects of different development scenarios, determine risk areas that 
can be managed, and provide the information necessary for monitoring 
and control of the program. 

The tools allow estimators to manipulate various inputs to gauge the 
overall sensitivity to parameter assumptions and then assess the 
overall risk, based on the certainty of those inputs. Developers who 
use tools in development can discover potential problems early enough 
to mitigate their impact. 

As the project matures and actual data become available, the precision 
of the cost estimates produced by a parametric tool are likely to 
improve. For this to happen, the tool must be calibrated with actual 
data from completed programs so it can be adjusted to reflect the 
actual development environment. Since most models are built on industry 
averages, simply using default values in the tool may lead to skewed 
results. Calibration avoids this by using known inputs and outcomes to 
adjust the relationships in the model. Therefore, calibration is 
necessary for ensuring more accurate estimates. 

When a parametric tool is used, it is essential to ensure that the 
estimators are trained and experienced in applying it and interpreting 
the results. Simply using a tool does not enhance the estimate’s 
validity. Using a tool correctly by calibrating it to the specific 
program is necessary for developing a reliable estimate. In addition, 
the following issues should be well understood before unquestioningly 
accepting the results of a parametric tool: 

* Ensure that autogenerated code is properly captured by the model, in 
terms of increased productivity and the effort required to design, 
develop, document, and produce the code.

* Output from the tool may include different cost and effort estimates 
or activities and phases that would have to be mapped or deleted to 
conform to the specific program. Not understanding what is in the 
output could lead to overestimating or underestimating the program. 

* Some models limit the size of the development program for which they 
can forecast the effort. Sizes outside of the tool range may not fit 
the program being estimated. 

* Data are often proprietary so the models are only as accurate as 
their underlying data allow them to be. Therefore, results from the 
model should be cross-checked. 

* Each model has different sensitivities to certain parameters and 
“opinions” on desirable staff levels. Therefore, various models offer 
different schedule duration results. For particularly small or large 
software programs, a schedule predicted by a commercial parametric 
model 
needs to be crosschecked. 

* Where a detailed build structure or spiral development is to be 
modeled, the commercial model implementation and results should be 
closely monitored. The same is true for significant integration of 
commercial off-the-shelf software (COTS) or government off-the-shelf 
software (GOTS) with development software (or hardware). 

In addition to these issues, it is important to note that many models 
do not address the costs associated with database development. If 
databases will be required as part of the software solution, and the 
model used to estimate the software does not account for the cost of 
database development, then this cost must be estimated separately. The 
cost for database development will depend on the size and complexity of 
the source data. Cost drivers for database development include the 
number of feeder systems, data elements, and users as well as the 
software to be used to develop the new database. 

Commercial Off-the-Shelf Software: 

Using commercial off-the-shelf software has advantages and 
disadvantages, and auditors need to understand the risks that come with 
relying on it. One advantage is that development time can be faster. 
The software can provide more user functionality than custom software 
and may be flexible enough to accommodate multiple hardware and 
operating environments. Also, help desk support can be purchased with 
the commercial license, which can help reduce software maintenance 
costs. 

Among the drawbacks to off-the-shelf software is the learning curve 
associated with its use, as well as integrating it into the new 
program’s environment. In addition, most commercial software is 
developed for a broad spectrum of users, so it tends to address only 
general functions. More specific functions must be customized and 
added, and glue-code may be required to enable the software to interact 
with other applications. And, because the source code is usually not 
provided to customers of commercial off-the-shelf software, it can be 
hard to support the software in-house. When upgrades occur, the 
software may have to be reintegrated with existing custom code. Thus, 
it can be wrong to think that commercial software will necessarily be 
an inexpensive solution. 

Estimators tend to underestimate the effort that comes before and after 
implementing off-the-shelf software. For example, requirements 
definition, design, and testing of the overall system must still be 
conducted. Poorly defined requirements can result in less than optimal 
software selection, necessitating the development of new code to 
satisfy all requirements. This unexpected effort will raise costs and 
cause program delays. In addition, adequate training and access to 
detailed documentation are important for effectively using the 
software. 

Furthermore, since commercial software is subject to intense market 
forces, upgrades can be released with minimal testing, causing 
unpredictable problems, such as defects and systems incompatibilities. 
When this happens, additional time is needed to analyze the cause of 
failures and fix them. Finally, interfaces between the software and 
other applications may need to be rewritten every time the software is 
upgraded. While software developers can address all these issues, they 
take some time to accomplish. Therefore, adequate planning should be 
identified and estimated by the cost estimator to ensure that enough 
time and resources are available to perform them. 

Enterprise Resource Planning Software: 

Enterprise resource planning (ERP) refers to the implementation of an 
administrative software system based on commercial off-the-shelf 
software throughout an organization. ERP’s objective is to integrate 
information and business processes—including human resources, finance, 
manufacturing, and sales—to allow information entered once into the 
system to be shared throughout an organization. ERP systems force 
business process reengineering, allowing for improved operations that 
can lead to savings down the road. To achieve savings requires an 
extensive knowledge of business processes so that users will optimize 
automation, programming skills, and change management in the new work 
processes. Although an ERP system is configured commercial software and 
should be treated as such, we highlight this type of effort because of 
the unique difficulty of estimating its implementation costs and 
duration. 

Organizations implementing ERP systems risk cost overruns and missed 
deadlines. According to a Gartner report, “For 40 percent of 
enterprises deploying ERP systems through 2009, the actual time and 
money spent on these implementations will exceed original estimates by 
at least 50 percent (0.7 probability).”[Footnote 46] 

At the heart of an ERP system are thousands of packages—built from 
database tables—that need to be configured to match end business 
processes. Each table has a decision switch that opens a specific 
decision path. By confining themselves to only one way to do a task, 
stove-piped units become integrated under one system. Deciding which 
switches in the tables to choose requires a deep understanding of the 
existing business operating processes. Thus, as table switches are 
picked, these business processes become reengineered to conform to the 
ERP’s way of doing business. As a result, change management and buy-in 
from the end users are crucial to the ERP system’s ultimate success. 

Cost estimators and auditors need to be aware of the additional risks 
associated with ERP implementation. 

Table 18 describes some of these risks and best practices for avoiding 
them. 

Table 18: Best Practices Associated with Risks in Implementing ERP: 

Risk: Training; 
Best practice: Staff are trained in the new ERP system’s software and 
the new processes; agencies teach workers how the ERP system will 
affect their business processes, developing their own training programs 
if necessary; providing mentoring and support for the first year of 
implementation eases the transition to the new system; obtaining user 
buy-in can be accomplished by communicating and marketing the benefits 
and new capabilities the ERP system will offer. 

Risk: Integrating and testing; 
Best practice: Agencies build and test links from their established 
software to the new ERP software links system or buy add-ons that are 
already integrated with the new system; they estimate and budget costs 
carefully, planning either way to test ERP integration from a process-
oriented perspective. 
 
Risk: Interfacing with legacy systems; 
Best practice: Since interfacing the ERP’s system software with legacy 
systems can be very expensive, carefully determining early on how both 
systems will pass data is paramount; preparing a business case to 
evaluate whether to maintain the legacy system is worth the added 
costs. 
 
Risk: Customizing; 
Best practice: Customizing core ERP software can be costly, especially 
since the ERP system’s elements are linked; perhaps use commercial add-
ons if the software cannot handle at least one business process. 
 
Risk: Converting and analyzing data; 
Best practice: Cost estimators look at the agency’s data conversion and 
analysis needs to see whether, for example, the cost of converting data 
to a new client server setup is accounted for, data from the ERP system 
and external systems have to be combined for analysis, the ERP budget 
should include data warehouse costs, or programming has to be 
customized. 
 
Risk: Following up installation; 
Best practice: Agencies plan for follow-up activities after 
installation, building them into their budget, keeping the team who 
implemented the ERP system onboard to keep the agency informed of its 
progress, and providing management with knowledge of the ERP project’s 
benefits. 

Source: GAO, DOD, and Derek Slater, “The Hidden Costs of Enterprise 
Software,” CIO Enterprise Magazine, Jan. 15, 1998. 

[End of table] 

Other costs associated with ERP system implementations include costs 
for adding “bolt-ons,” which are separate supplemental software 
packages that deliver capability not offered by the ERP system. Bolt-
ons connect to the ERP system using standard application programming 
interfaces or extensible markup language schema, which allow for data 
to pass between both systems. Costs for interfacing the bolt-on with 
the ERP system need to be identified and estimated. In addition, the 
number of bolt-ons that need to be integrated, as well as the type and 
size of the bolt-on functionality, will drive the cost of the 
interface. 

Experts agree that the ERP postimplementation stabilization period 
tends to be underestimated, because people tend to be too optimistic 
about how long training and the transition period will last. As a 
result, there is a risk for cost growth if management does not do a 
good job of selling the benefits of ERP. To successfully implement an 
ERP system, management has to be committed to freeing up resources to 
get the job done. This means that seasoned staff will need to be pulled 
away from their day jobs to focus on the effort to be fully effective. 
In addition, training tends to be underestimated in terms of both 
length and timing. To better plan for this effort, management needs to 
create a sense of urgency for change and provide early communication 
and adequate training in order to ensure successful implementation. 

Software Costs Must Also Account For Information Technology
Infrastructure And Services: 

Studies have shown that information technology (IT) services outside 
software development and maintenance (for example, hardware cost, help 
desk, upgrade installation, training) can make up a majority of total 
ownership costs. In fact, OMB reports that 77 percent of the overall IT 
budget for fiscal year 2009 will support steady state IT operations 
while only 23 percent will be used for development, modernization, and 
enhancement. 

Even systems such as ships, aircraft, and mission control centers have 
major IT infrastructure and services components to them. In fact, some 
IT systems encounter over 90 percent of their costs in the 
infrastructure and services required to support and run them. Yet when 
we read of costs, successes, failures, and challenges in IT systems, 
the vast majority of the systems typically refer to the software 
portions only, ignoring the IT services and infrastructure components. 
Making matters more difficult for those estimating IT systems are the 
numerous definitions of IT infrastructure. One useful definition is 
that it consists of the equipment, systems, software, and services used 
in common across an organization, regardless of mission, program, or 
project. IT infrastructure also serves as the foundation on which 
mission, program, or project-specific systems and capabilities are 
built. 

While we have already discussed software development and maintenance, 
we discuss in this section estimating the information technology 
services, hardware systems, and facilities required to support software 
and systems. 

Unique Components Of It Estimation: 

Unlike software, IT estimation is in some ways simpler than software 
development estimation, since IT infrastructure and services are more 
tangible. However, IT estimation is fraught with issues such as:
 
* What is the cost of the system engineering to define the IT system? 

* How much computing power is needed to support a system? 

* How many help desk personnel are needed to support X users? 

* How can costs be contained while still achieving innovation? 

* How can the value of the IT investment be quantified against its 
costs? 

* How do buy and lease decisions affect expenses and profitability? 

* How can we make tradeoffs between technology and costs? 

* What kind of application initiatives are needed to support the 
business? 

* How many vendors and how much vendor interface is required to run the 
IT operation? 

* How many sites does the IT infrastructure support? 

* How many and how clearly defined or stable are the requirements for 
the IT to align itself with the business goals? 

Simply getting a quote from a vendor for an IT system is rarely 
sufficient for IT cost estimation. While quotes often do not include 
many important cost elements, the cost estimator will still need to 
consider these elements. They include: 

* help desk support services supplied internally for applications and 
equipment; 

* facilities costs; 

* costs of on-going installation, maintenance, repair, and trouble 
shooting; 

* employee training, both formal training and self-training. 

To further complicate the effort, many vendors offer IT infrastructure 
either as a “software as a service” platform or as just “cloud 
computing.”[Footnote 47] Vendor-operated IT infrastructure hardware can 
be viable if issues such as loss of control, security, and potential 
resource sharing are acceptable. However, such vendor-operated 
infrastructure does not usually eliminate the costs of ongoing IT 
services to provide users help desk support, local computing, setup 
training, and other infrastructure services. The cost estimator must be 
aware that these costs should be considered, whether the infrastructure 
is to be owned by the government, leased, or owned and operated by 
vendors under contract with the government. 

Major Cost Drivers Associated with IT Estimation: 
 
Many factors that affect IT costs need to be considered when developing 
an IT cost estimate. Various examples of cost drivers, organized by 
physical attributes of the IT infrastructure, are listed next, along 
with performance and complexity requirements and economic 
considerations. 

1. Physical attributes that drive IT costs: 

* Application software, system software, and database storage size; 

* End user hardware list (e.g., laptops, CPU, printers); 

* Facility requirements (power, cooling); 

* Infrastructure hardware list (UNIX Servers, Windows servers, WAN/LAN 
equipment); 

* Number of application software, system software, and database items; 

* Number of application software, system software, and database users 
(concurrent, causal); 

* Number of inbound and outbound application software and database 
interfaces; 

* Number of unique platforms supported; 

* Operating locations; 

* Physical and organizational entities. 

2. Performance and complexity attributes: 

* Business requirements; 

* Complexity of infrastructure environment (e.g., disparate platforms, 
loose vs. tight coupling); 

* User type (professional, concurrent, casual); 

* Criticality and reliability of systems; 

* Expected service level (system administration, database 
administration, help desk Tier I, Help Desk Tier II, Help Desk Tier 
III); 

* Experience with systems; 

* Infrastructure hardware complexity (small, medium, large); 

* IT project type (ERP, SOA, Web application, data mart); 

* Number of transactions per second; 

* Number of vendors; 

* Process experience and rigor; 

* Security requirements; 

* System complexity (hardware or software); 

* Usage patterns (transaction rates). 

3. Economic factors and considerations: 

* Acquisition strategy; 

* Hardware leasing and purchasing agreements; 

* Labor rates; 

* Sourcing strategy; 

* Replacement and upgrade policies; 

* Software leasing and purchasing agreements (enterprise, user based); 

* Test plan; 

* Training strategy; 

* Years of operating. 

Common Risks for IT Infrastructure: 

Many of the risks that affect software cost estimating apply to IT 
infrastructure. For example, in estimating the costs of any effort, a 
consideration should be made whether the risks of the investment 
justify the inclusion of an independent verification and validation 
contractor. In situations where the risks are very high, such as 
potential loss of life, the overall schedule may need to be extended to 
accommodate the additional reviews and testing required. For IT 
infrastructure, the set of risks in table 19 should be considered. 

Table 19: Common IT Infrastructure Risks: 

Risk: Financial; 
Technical, management, and logistic requirements that increase costs: 
* Cost overruns; 
* Funding cuts and delays. 

Risk: Logistics and equipment; 
Technical, management, and logistic requirements that increase costs: 
* Contingency equipment availability; 
* Physical storage of equipment on arrival and security; 
* Supply availability. 
 
Risk: Schedule; 
Technical, management, and logistic requirements that increase costs: 
* Unscheduled changes and delays; 
* Nonconformance, not starting, and failures; 
* Reliance on external subcontractors and organizations. 

Risk: Personnel; 
Technical, management, and logistic requirements that increase costs: 
* Changes of personnel among customer or vendor; 
* Lack of skills or knowledge; 
* Not aware of policy or procedures or inadequate personnel to support 
help desk and deployment; 
* Time lost for end user training, trouble shooting, and down time. 
 
Risk: Project management; 
Technical, management, and logistic requirements that increase costs: 
* No quality control or management process built into plan; 
* Absence of issue, change request, or configuration management logs; 
* Inconsistent project documentation or lack of IT process model; 
* Information security; 
* Lack of detailed site information; 
* Lack of issue identification or trends; 
* Lack of reporting; 
* Poor planning; 
* Requirements not well defined; 
* Role confusion; 
* Unaware of customer site requirements. 
 
Risk: Technical; 
Technical, management, and logistic requirements that increase costs: 
* Adequate capacity; 
* Additional hardware or software requirements to fully support system 
* Compatibility or whether data in the relevant process flow from end 
to end; 
* Disasters; 
* Hardware or software failure; 
* Incorrect images or version loaded; 
* Integration with existing systems; 
* New design not working; 
* Unplanned or unapproved changes; 
* Version control problems. 

Risk: User; 
Technical, management, and logistic requirements that increase costs: 
* Confusion about customer and vendor responsibilities; 
* Inability to perform core or noncore business activities; 
* Loss of data; 
* Not aware of vendor schedule or activities; 
* User expectations. 

Source: GAO. 

[End of table] 

Estimating Labor and Material Costs Associated with IT Infrastructure: 

Labor and material nonrecurring and recurring efforts are associated 
with IT infrastructure. For estimating the nonrecurring effort, staff 
loading of the IT infrastructure is similar to software development 
during early architecture and design. Once the design is complete, the 
recurring effort associated with actual implementation and deployment 
can be accomplished, based on a distribution of organizational demand 
for IT. 

IT recurring operations costs include costs similar to the maintenance 
of general fixed facilities. For example, facilities costs such as 
power, security, and general facilities support apply to IT 
infrastructure recurring operations. Furthermore, costs for purchased 
software licenses, training, technical refreshment, and various service 
level agreements also need to be considered. Finally, since the cost of 
hardware changes daily as does the requirement for computing power in 
items like servers, designing with a 50 percent reserve in capacity is 
prudent since systems tend to grow. Many labor services categories need 
to be considered when developing an IT infrastructure labor cost 
estimate. Table 20 describes typical labor categories.[Footnote 48] 

Table 20: Common Labor Categories Described: 
 
Category: Project stakeholder; 
Description: A person invested in the project’s success while not 
participating in its execution or implementation; includes end users, 
managers, and external clients whose success is somehow tied to the 
project’s success. Stakeholders work with the product management team 
to ensure that the solution developed meets the project’s original 
needs. Stakeholder participation and availability are vital to the 
success of any project; 
Common titles: [Empty]. 
 
Category: Management; 
Description: Performs project planning, staffing, and tracking; is 
involved with daily operational activities, ensuring that resources are 
used effectively and services are delivered; 
Common titles: Configuration manager, database manager, IT manager, 
project manager. 
 
Category: Analyst; 
Description: Generally involved in planning and defining needs and
requirements for IT projects and related support systems and in ongoing 
systems support, often bridging the user or customer and the technical 
team. Generally has domain or specialty knowledge of a certain type of 
system, technology, or discipline used to apply technology to address 
business and user requirements; 
Common titles: Business process, requirements, or system analyst; 
network or telecommunications analyst; support analyst; operations
analyst; database analyst; UI analyst; security analyst. 

Category: Architect; 
Description: Develops high-level system design plans to meet the
organization’s needs and comply with its policies; can help formulate 
policies and plans that support the organization, particularly as they 
pertain to technologies used to carry out policies and procedures; 
Common titles: Systems architect or engineer; IT or data architect; 
network architect; storage architect. 

Category: Technician; 
Description: Involved primarily in the physical setup, support, and 
maintenance of systems according to well defined plans and procedures, 
including system setup, installation, upgrades, and troubleshooting; 
Common titles: Desktop or PC technician; network engineer or 
technician; hardware technician; telecommunications technician. 
 
Category: Test/QA; 
Description: Primarily verifies the integrity and performance of 
systems being deployed and operated; develops test plans and 
procedures, collecting and tracking defect data and problem reports and 
serves an auditing function to ensure compliance with policies and 
procedures; 
Common titles: IT auditor, QA analyst, application tester, call center 
agent. 
 
Category: Documentation; 
Description: Prepares or maintains documentation pertaining to 
programming, systems operation, and user documentation, including user 
manuals and online help screens; 
Common titles: Technical or report writer; online help publisher; 
content developer; documentation specialist. 
 
Category: Training; 
Description: Prepares and updates courseware and training materials and 
conducts training classes or events; 
Common titles: Instructor, training developer, instructional designer, 
end user. 

Category: Administrator; 
Description: Generally involved with the ongoing administration, 
maintenance, and support of specific systems to ensure they operate 
properly and effectively; associated with a specific system or type of 
system such as a platform, database, network, or enterprise 
application; 
Common titles: Network, system, or enterprise application 
administrator; system administrator; Web or telecommunications 
administrator; database administrator; security administrator; storage 
administrator; help desk specialist (tier I, tier II, tier III). 

Category: Computer operator; 
Description: Computer operators not included in support of IT 
infrastructure and IT services; 
Common titles: [Empty]. 

Category: Indirect support; 
Description: Secretarial, reception, and other labor in support of IT 
services and infrastructure personnel and systems; 
Common titles: [Empty]. 

Category: Contract labor; 
Description: Vendors that provide services under contract to support IT 
infrastructure; 
Common titles: [Empty]. 
 
Source: GAO. 
[End of table] 

9. Best Practices Checklist: Estimating Software Costs: 
 
* The software cost estimate followed the 12-step estimating process: 
- Software was sized with detailed knowledge of program scope, 
complexity, and interactions, and the cost estimators worked with 
software engineers to determine the appropriate sizing metric. 
- It was sized with source lines of code, function, object, feature 
point, or other counts. 

* The software sizing method was appropriate: 
- Source lines of code were used if requirements were well defined and 
if 
there was a historical database of code counts for similar programs and 
a standard definition for a line of code. 
- Function points were used if detailed requirements and specifications 
were available, software did not contain many algorithmic functions, 
and an experienced and certified function point counter was available. 
- COSMIC points were used if functional user requirements are known and 
the application is for business, real-time, embedded, or infrastructure 
software. 
- Object points were used if computer-aided software engineering tools 
were used to develop the software. 
- Reports, interfaces, conversions, extensions and forms/workflow were 
used for ERP programs. 
- Use cases and use case points were used if system and user 
interactions were defined. 
- Autogenerated and reused source lines of code were identified 
separately from new and modified code to account for pre- and 
postimplementation efforts.
- Several methods were used to size the software to increase the 
accuracy of the sizing estimate. 
- The final software size was adjusted for growth based on historical 
data, and growth is continually monitored over time. 

*Software cost estimates included: 
- Development labor costs for coding and testing, other labor 
supporting 
software development, and nonlabor costs like purchasing hardware 
and licenses. 
- Productivity factors for converting software size into labor effort, 
based on historical data and calibrated to match program size and 
development environment. 
- Industry average productivity factors and risk ranges (no historical 
data were available). 
- Assumptions about productive labor hours in a day and work days in a 
year. 
- Development schedules accounting for staff availability, prior task 
dependencies, concurrent and critical path activities, number and 
length of shifts, overtime allowance, down time, and worker locations. 
- Costs for help desk support, database development, and corrective, 
adaptive, and preventive maintenance as part of the software’s life 
cycle cost. 
- Time and effort associated with rework to fix defects. 
- Training cost estimators to calibrate parametric tools to match 
the program and cross-checking model results for accuracy. 
- Estimators' accounting for integrating commercial off-the-shelf 
software into the system, including developing custom software and glue-
code. 
- Impact of risks facing ERP system implementations as outlined in 
table 18. 
- Costs associated with interfacing bolt-on applications for ERP 
systems. 

* IT infrastructure and services components of the software cost 
estimate included: 
- Costs associated with the physical attributes of the IT 
infrastructure, the performance and complexity requirements, and 
economic considerations. 
- Impact of risks affecting IT infrastructure, as outlined in table 19. 
- Costs associated with labor and material nonrecurring and recurring 
efforts. 

[End of Chapter 12] 

Chapter 13: Sensitivity Analysis: 

As a best practice, sensitivity analysis should be included in all cost 
estimates because it examines the effects of changing assumptions and 
ground rules. Since uncertainty cannot be avoided, it is necessary to 
identify cost elements that represent the most risk and, if possible, 
cost estimators should quantify the risk. This can be done through both 
a sensitivity analysis and an uncertainty analysis (discussed in the 
next chapter). 

Sensitivity analysis helps decision makers choose the alternative. For 
example, it could allow a program manager to determine how sensitive a 
program is to changes in gasoline prices and at what gasoline price a 
program alternative is no longer attractive. By using information from 
a sensitivity analysis, a program manager can take certain risk 
mitigation steps, such as assigning someone to monitor gasoline price 
changes, deploying more vehicles with smaller payloads, or decreasing 
the number of patrols. 

For a sensitivity analysis to be useful in making informed decisions, 
however, carefully assessing the underlying risks and supporting data 
is necessary. In addition, the sources of the variation should be well 
documented and traceable. Simply varying the cost drivers by applying a 
subjective plus or minus percentage is not useful and does not 
constitute a valid sensitivity analysis. This is the case when the 
subjective percentage does not have a valid basis or is not based on 
historical data. 

In order for sensitivity analysis to reveal how the cost estimate is 
affected by a change in a single assumption, the cost estimator must 
examine the effect of changing one assumption or cost driver at a time 
while holding all other variables constant. By doing so, it is easier 
to understand which variable most affects the cost estimate. In some 
cases, a sensitivity analysis can be conducted to examine the effect of 
multiple assumptions changing in relation to a specific scenario. 

Regardless of whether the analysis is performed on only one cost driver 
or several within a single scenario, the difference between sensitivity 
analysis and risk or uncertainty analysis is that sensitivity analysis 
tries to isolate the effects of changing one variable at a time, while 
risk or uncertainty analysis examines the effects of many variables 
changing all at once. 

Typically performed on high-cost elements, sensitivity analysis 
examines how the cost estimate is affected by a change in a cost 
driver’s value. For example, it might evaluate how the number of 
maintenance staff varies with different assumptions about system 
reliability values or how system manufacturing labor and material costs 
vary in response to additional system weight growth. 

Sensitivity analysis involves recalculating the cost estimate with 
different quantitative values for selected input values, or parameters, 
in order to compare the results with the original estimate. If a small 
change in the value of a cost element’s parameter or assumption yields 
a large change in the overall cost estimate, the results are considered 
sensitive to that parameter or assumption. Therefore, a sensitivity 
analysis can provide helpful information for the system designer 
because it highlights elements that are cost sensitive. In this way, 
sensitivity analysis can be useful for identifying areas where more 
design research could result in less production cost or where increased 
performance could be implemented without substantially increasing cost. 
This type of analysis is typically called a what-if analysis and is 
often used for optimizing cost estimate parameters. 

Sensitivity Factors: 

Uncertainty about the values of some, if not most, of the technical 
parameters is common early in a program’s design and development. Many 
assumptions made at the start of a program turn out to be inaccurate. 
Therefore, once the point estimate has been developed, it is important 
to determine how sensitive the total cost estimate is to changes in the 
cost drivers. Some factors that are often varied in a sensitivity 
analysis are: 

* a shorter or longer economic life; 

* the volume, mix, or pattern of workload; 

* potential requirements changes; 

* configuration changes in hardware, software, or facilities; 

* alternative assumptions about program operations, fielding strategy, 
inflation rate, technology heritage savings, and development time; 

* higher or lower learning curves; 

* changes in performance characteristics; 

* testing requirements; 

* acquisition strategy, whether multiyear procurement, dual sourcing, 
or the like; 

* labor rates; 

* growth in software size or amount of software reuse; and 

* down-scoping the program. 

These are just some examples of potential cost drivers. Many factors 
that should be tested are determined by the assumptions and performance 
characteristics outlined in the technical baseline description and 
GR&As. Therefore, auditors should look for a link between the technical 
baseline parameters and the GR&As to see if the cost estimator examined 
those that had the greatest effect on the overall sensitivity of the 
cost estimate. 

In addition, the cost estimator should always include in a sensitivity 
analysis the assumptions that are most likely to change, such as an 
assumption that was made for lack of knowledge or one that is outside 
the control of the program office. Case study 38 shows some assumptions 
that can affect the cost of building a ship. 

Case Study 38: Sensitivity Analysis, from Defense Acquisitions, GAO-05-
183: 
 
Given the uncertainties inherent in ship acquisitions, such as 
introducing new technologies and volatile overhead rates over time, 
cost analysts face a significant challenge in developing credible 
initial cost estimates. The Navy must develop cost estimates as long as 
10 years before ship construction begins, before many program details 
are known. Cost analysts therefore have to make a number of assumptions 
about ship parameters like weight, performance, and software and about 
market conditions, such as inflation rates, workforce attrition, and 
supplier base. 

In the eight case study ships we examined, other unknowns led to 
uncertain estimates. Labor hour and material costs were based on data 
from previous ships and on unproven efficiencies in ship construction. 
GAO found that analysts often factored in savings based on expected 
efficiencies that never materialized. For example, they anticipated 
savings from implementing computer-assisted design and computer-
assisted manufacturing for the San Antonio class transport LPD 17, but 
the contractor had not made the requisite research investments to 
achieve the proposed savings. Similar unproven or unsupported 
efficiencies were estimated for the Arleigh Burke class destroyer DDG 
92 and Nimitz class aircraft carrier CVN 76. Changes in the 
shipbuilders’ supplier base also created uncertainties in their 
overhead costs. 

Despite these uncertainties, the Navy did not test the validity of the 
cost analysts’ assumptions in estimating construction costs for the 
eight case study ships and did not identify a confidence level for 
estimates. 

Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, GAO-05-183, 
Washington, D.C.: Feb. 28, 2005. 

[End of case study] 

Steps In Performing A Sensitivity Analysis: 

A sensitivity analysis addresses some of the estimating uncertainty by 
testing discrete cases of assumptions and other factors that could 
change. By examining each assumption or factor independently, while 
holding all others constant, the cost estimator can evaluate the 
results to discover which assumptions or factors most influence the 
estimate. A sensitivity analysis also requires estimating the high and 
low uncertainty ranges for significant cost driver input factors. To 
determine what the key cost drivers are, a cost estimator needs to 
determine the percentage of total cost that each cost element 
represents. The major contributing variables within the highest 
percentage cost elements are the key cost drivers that should be 
varied in a sensitivity analysis. A credible sensitivity analysis 
typically has five steps: 

1. identify key cost drivers, ground rules, and assumptions for 
sensitivity testing; 

2. reestimate the total cost by choosing one of these cost drivers to 
vary between two set amounts—for example, maximum and minimum or 
performance thresholds;[Footnote 49] 

3. document the results; 

4. repeat 2 and 3 until all factors identified in step 1 have been 
tested independently; 

5. evaluate the results to determine which drivers affect the cost 
estimate most. 

Sensitivity analysis also provides important information for economic 
analyses that can end in the choice of a different alternative from the 
original recommendation. This can happen because, like a cost estimate, 
an economic analysis is based on assumptions and constraints that may 
change. Thus, before choosing an alternative, it is essential to test 
how sensitive the ranking of alternatives is to changes in assumptions. 
In an economic analysis, sensitivity is determined by how much an 
assumption must change to result in an alternative that differs from 
the one recommended. For example, an assumption is considered sensitive 
if a 10–50 percent change yields a different alternative, very 
sensitive if the change is less than 10 percent. 

Assumptions and cost drivers that have the most effect on the cost 
estimate warrant further study to ensure that the best possible value 
is used for that parameter. If the cost estimate is found to be 
sensitive to several parameters, all the GR&As should be reviewed, to 
assure decision makers that sensitive parameters have been carefully 
investigated and the best possible values have been used in the final 
point estimate. 

Sensitivity Analysis Benefits and Limitations: 

A sensitivity analysis provides a range of costs that span a best and 
worst case spread. In general, it is better for decision makers to know 
the range of potential costs that surround a point estimate and the 
reasons behind what drives that range than to just have a point 
estimate to make a decision from. Sensitivity analysis can provide a 
clear picture of both the high and low costs that can be expected, with 
discrete reasons for what drives them. Figure 14 shows how sensitivity 
analysis can give decision makers insight. 

Figure 14: A Sensitivity Analysis That Creates a Range around a Point 
Estimate: 

[Refer to PDF for image: Illustration] 

Point Estimate: $10 billion. 

Increase in life-cycle estimate: 

Description: Increase the number of cost penalties in airframe 
development CER: +$40.0 million ((0.4%): $10.040 billion. 

Description: Double the development testing: +$50.5 million (0.5%): 
$10.090 billion. 

Description: Increase airframe weight: +$1.009 million (10%): $11,099 
billion. 

Description: Eliminate concurrent production quantities: +$22.0 million 
(0.2%): $11.121 billion. 

Description: Increase quantity of materials in aircraft: +$1,668 
million (15%): $12,789 billion. 

Decrease in life-cycle estimate: 

Description: Use 88% learning curve instead of 91%: -$60.0 million 
(0.6%): $9.940 billion; 

Description: Eliminate integration and assembly cost add-on: -$50.0 
million (0.5%): $9.89 billion. 

Description: Reduce airframe weight: -$100.0 million (1.0%): $9.79 
billion. 

Description: Improve aircraft maintainability: -$40.0 million (0.4%): 
$9.75 billion. 

Description: Reduce peacetime flying hours: -#390.0 million (4.0%): 
$9.36 billion. 

Source: GAO. 

[End of figure] 

In figure 14, it is very apparent how certain assumptions affect the 
estimate. For example, increasing the quality of materials in the 
aircraft has the biggest effect on the highest cost estimate—adding 
$1,668 million to the point estimate—while reducing the number of 
flying hours is the biggest driver for reducing the cost 
estimate—reducing the flying hours saves $390 million. Using visuals 
like this can quickly display what-if analyses that can help management 
make informed decisions. 

A sensitivity analysis also reveals critical assumptions and program 
cost drivers that most affect the results and can sometimes yield 
surprises. Therefore, the value of sensitivity analysis to decision 
makers lies in the additional information and understanding it brings 
to the final decision. Sensitivity analysis can also make for a more 
traceable estimate by providing ranges around the point estimate, 
accompanied by specific reasons for why the estimate could vary. This 
insight allows the cost estimator and program manager to further 
examine potential sources of risk and develop ways to mitigate them 
early. Sensitivity analysis permits decisions that influence the 
design, production, and operation of a system to focus on the elements 
that have the greatest effect on cost. 

Sensitivity analysis is limited in that it examines only the effect of 
changing one assumption or factor at a time. But the risk of several 
assumptions or factors varying simultaneously, and its effect on the 
overall point estimate, should be understood.[Footnote 50] In the next 
chapter, we discuss risks and uncertainty analyses. 

10. Best Practices Checklist: Sensitivity Analysis: 

* The cost estimate was accompanied by a sensitivity analysis that 
identified the effects of changing key cost driver assumption and 
factors. 
- Well-documented sources supported the assumption or factor ranges. 
- The sensitivity analysis was part of a quantitative risk assessment 
and not based on arbitrary plus or minus percentages. 
- Cost-sensitive assumptions and factors were further examined to see 
whether design changes should be implemented to mitigate risk. 
- Sensitivity analysis was used to create a range of best and worst 
case costs. 
- Assumptions and performance characteristics listed in the technical 
baseline description and GR&As were tested for sensitivity, especially 
those least understood or at risk of changing. 
- Results were well documented and presented to management for 
decisions. 

* The following steps were taken during the sensitivity analysis: 
- Key cost drivers were identified. 
- Cost elements representing the highest percentage of cost were 
determined and their parameters and assumptions were examined. 
- The total cost was reestimated by varying each parameter between its 
minimum and maximum range. 
- Results were documented and the reestimate was repeated for each 
parameter that was a key cost driver. 
- Outcomes were evaluated for parameters most sensitive to change. 
* The sensitivity analysis provided a range of possible costs, a point 
estimate, and a method for performing what-if analysis. 

[End of Chapter 13] 

Chapter 14: Cost Risk And Uncertainty: 

In chapter 13, we discussed sensitivity analysis and how it is useful 
for performing what-if analysis, determining how sensitive the point 
estimate is to changes in the cost drivers, and developing ranges of 
potential costs. A drawback of sensitivity analysis is that it looks 
only at the effects of changing one parameter at a time. In reality, 
many parameters can change at the same time. Therefore, in addition to 
a sensitivity analysis, an uncertainty analysis should be performed to 
capture the cumulative effect of additional risks. 

Because cost estimates predict future program costs, uncertainty is 
always associated with them. For example, data from the past may not 
always be relevant in the future, because new manufacturing processes 
may change a learning curve slope or new composite materials may change 
the relationship between weight and cost. Moreover, a cost estimate is 
usually composed of many lower-level WBS elements, each of which comes 
with its own source of error. Once these elements are added together, 
the resulting cost estimate can contain a great deal of uncertainty. 

The Difference Between Risk And Uncertainty: 

Risk and uncertainty refer to the fact that because a cost estimate is 
a forecast, there is always a chance that the actual cost will differ 
from the estimate. Moreover, lack of knowledge about the future is only 
one possible reason for the difference. Another equally important 
reason is the error resulting from historical data inconsistencies, 
assumptions, cost estimating equations, and factors typically used to 
develop an estimate. 

In addition, biases are often found in estimating program costs and 
developing program schedules. The biases may be cognitive—often based 
on estimators’ inexperience—or motivational, where management 
intentionally reduces the estimate or shortens the schedule to make the 
project look good to stakeholders. Recognizing the potential for error 
and deciding how best to quantify it is the purpose of risk and 
uncertainty analysis.[Footnote 51] 

It is inaccurate to add up the most likely WBS elements to derive a 
program cost estimate, since their sum is not usually the most likely 
estimate for the total program, even if they are estimated without 
bias.[Footnote 52] Yet summing costs estimated at the detailed level to 
derive a point estimate is the most common approach to estimating a 
total program. Simulation of program risks is a better way to estimate 
total program cost, as we discuss below.

Quantifying risk and uncertainty is a cost estimating best practice 
addressed in many guides and references. DOD specifically directs that 
uncertainty be identified and quantified. The Clinger-Cohen Act 
requires agencies to assess and manage the risks of major information 
systems, including the application of the risk-adjusted return on 
investment criterion in deciding whether to undertake particular 
investments.[Footnote 53] 

While risk and uncertainty are often used interchangeably, in 
statistics their definitions are distinct: 

Risk is the chance of loss or injury. In a situation that includes 
favorable and unfavorable events, risk is the probability that an 
unfavorable event will occur. 

Uncertainty is the indefiniteness about the outcome of a situation. It 
is assessed in cost estimate models to estimate the risk (or 
probability) that a specific funding level will be exceeded.[Footnote 
54] 

Therefore, while both risk and uncertainty can affect a program’s cost 
estimate, enough data will never be available in most situations to 
develop a known frequency distribution. Cost estimating is analyzed 
more often for uncertainty than risk, although many textbooks use both 
terms to describe the effort. 

Point Estimates Alone Are Insufficient For Good Decisions: 

Since cost estimates are uncertain, making good predictions about how 
much funding a program needs to be successful is difficult. In a 
program’s early phases, knowledge about how well technology will 
perform, whether the estimates are unbiased, and how external events 
may affect the program is imperfect. For management to make good 
decisions, the program estimate must reflect the degree of uncertainty, 
so that a level of confidence can be given about the estimate. 

Quantitative risk and uncertainty analysis provide a way to assess the 
variability in the point estimate. Using this type of analysis, a cost 
estimator can model such effects as schedules slipping, missions 
changing, and proposed solutions not meeting user needs, allowing for a 
known range of potential costs. Having a range of costs around a point 
estimate is more useful to decision makers, because it conveys the 
level of confidence in achieving the most likely cost and also informs 
them on cost, schedule, and technical risks. 

Point estimates are more uncertain at the beginning of a program, 
because less is known about its detailed requirements and opportunity 
for change is greater. In addition, early in a program’s life cycle, 
only general statements can be made. As a program matures, general 
statements translate into clearer and more refined requirements that 
reduce the unknowns. However, more refined requirements often translate 
into additional costs, causing the distribution of potential costs to 
move further to the right, as illustrated in figure 15. 

Figure 15: Changes in Cost Uncertainty across the Acquisition Life 
Cycle: 

[Refer to PDF for image: Illustration] 

This figure plots cost against time and illustrates the following data 
points: 

Concept formulation: 
Cost estimate: $125 million. 

Development: 
Cost estimate: $175 million. 

Implementation: 
Cost Cost estimate: $230 million. 

Source: GAO. 

While the point estimate increases in figure 15, the uncertainty range 
around it decreases. More is learned as the project matures. First, a 
better understanding of the risks is achieved, and either some risk is  
retired or some form of risk handling lessens the potential cost or 
effect on schedule. Second, the program is understood better and, most 
probably, more requirements are added or overlooked as elements are 
added, which has a tendency to increase costs along with reducing the 
variance. Thus, a point estimate, by itself, provides no information 
about the underlying uncertainty other than that it is the value chosen 
as most likely. 

A confidence interval, in contrast, provides a range of possible costs, 
based on a specified probability level. For example, a program with a 
point estimate of $10 million could range in cost from $5 million to 
$15 million at the 95 percent confidence level. In addition, the 
probability distribution, usually in the form of a cumulative 
distribution or S curve (described below) can provide the decision 
maker with an estimate of the probability that the program’s cost will 
actually be at some value or lower. Conversely, 1.0 minus this 
probability is the probability that the project will overrun that 
value. 

Using an uncertainty analysis, a cost estimator can easily inform 
decision makers about a program’s potential range of costs. Management, 
in turn, can use these data to decide whether the program fits within 
the overall risk range of the agency’s portfolio. 

Budgeting To A Realistic Point Estimate: 
 
Over the years, GAO has reported that many programs overrun their 
budgets because original point estimates are unrealistic. Case studies 
39 and 40 are examples. 

Case Study 39: Point Estimates, from Space Acquisitions, GAO-07-96: 
 
Estimated costs for DOD’s major space acquisitions increased about 
$12.2 billion, or nearly 44 percent, above initial estimates for fiscal 
year 2006 through fiscal year 2011. GAO identified a variety of reasons 
for this. The most notable are that weapons programs have incentives to 
produce and use optimistic cost and schedule estimates to compete 
successfully for funding and that DOD starts its space programs before 
it has assurance that the capabilities it is pursuing can be achieved 
within its resource and time constraints. 

At the same time, the cost growth resulted partly from DOD’s using low 
cost estimates to establish program budgets, finding it necessary later 
to make funding shifts with costly, reverberating effects. In 2003, a 
DOD study found that the space acquisitions system was strongly biased 
to produce unrealistically low cost estimates throughout the process. 
The study found that most programs at contract initiation had a 
predictable cost growth of 50 percent to 100 percent. It found that the 
unrealistically low projections of program cost and the lack of 
provisions for management reserve seriously distorted management 
decisions and program content, increased risks to mission success, and 
virtually guaranteed program delays. GAO found most of these conditions 
in many DOD programs. 

Source: GAO, Space Acquisitions: DOD Needs to Take More Acton to 
Address Unrealistic Initial Cost Estimates of Space Systems, GAO-07-96, 
Washington, D.C.: Nov. 17, 2006. 

[End of case study] 

Case Study 40: Point Estimates, from Defense Acquisitions, GAO-05-183: 

For several case study ships, the costs of materials increased 
dramatically above the shipbuilder’s initial plan. Materials cost was 
the most significant component of cost growth for the CVN 76 in the 
Nimitz class of aircraft carriers, the LPD 17 in the San Antonio class 
of transports, and the SSN 775 in the Virginia class of submarines. The 
growth in materials costs resulted, in part, from Navy and shipbuilders 
under budgeting for these costs. 

For example, the materials budget for the first four Virginia class 
submarines was $132 million less than quotes received from vendors and 
subcontractors. The shipbuilder agreed to take on the challenge of 
achieving lower costs in exchange for providing in the contract that 
the shipbuilder would be reimbursed for cost growth in high-value, 
specialized materials. 

In addition, the materials budget for the CVN 76 and CVN 77 was based 
on an incomplete list of materials needed to construct the ships, 
leading to especially sharp increases in estimated materials costs. In 
this case, the Defense Contract Audit Agency criticized the 
shipbuilder’s estimating system, particularly the system for materials 
and subcontract costs, stating that the resulting estimates “do not 
provide an acceptable basis for negotiation of a fair and reasonable 
price.” Underbudgeting of materials contributed to cost growth, 
recognized in the fiscal year 2006 budget. 
 
Source: GAO, Defense Acquisitions: Improved Management Practices Could 
Minimize Cost Growth in Navy Shipbuilding Programs, GAO-05-183, 
Washington, D.C.: Feb. 28, 2005. 

[End of case study] 

We have found that budgeting programs to a risk-adjusted estimate that 
reflects a program’s risks is critical to its successfully achieving 
its objectives. However, programs have developed optimistic estimates 
for many reasons. Cost estimators may have ignored program risk, 
underestimated data outliers, relied on historical data that may be 
misleading for a new technology, or assumed better productivity than 
the historical data supported, causing narrow uncertainty ranges. 
Decision makers may add their own bias for political or budgetary 
reasons. For example, they may make optimistic assumptions by assuming 
that a new program will perform much better than its predecessor in 
order to justify a preconceived notion, to fit the program within 
unrealistic budgetary parameters, or just to sell the program. 

One way to determine whether a program is realistically budgeted is to 
perform an uncertainty analysis, so that the probability associated 
with achieving its point estimate can be determined. A cumulative 
probability distribution, more commonly known as an S curve—usually 
derived from a simulation such as Monte Carlo—can be particularly 
useful in portraying the uncertainty implications of various cost 
estimates. Figure 16 shows an example of a cumulative probability 
distribution with various cost estimates mapped to a certain 
probability level. 

Figure 16: A Cumulative Probability Distribution, or S Curve: 

[Refer to PDF for image: S-curve graph] 

Probability of occurrence plotted against dollars in thousands: 
 
Probability of occurrence: The risk-adjusted primary estimate = $825, 
or 40% probable; 

Probability of occurrence: 50% probability = $907.9. 

Probability of occurrence: 70% probability = $1,096. 

Source: GAO and NASA. 

[End of figure] 

In figure 16, one can readily see that given what is known about 
program risks and uncertainties, the least this hypothetical program 
could cost is about $500,000, at about 5 percent probability; the most, 
$1,700,000 or less, at about 95 percent probability. Using an S curve, 
decision makers can easily understand what the likelihood of different 
funding alternatives will imply.[Footnote 55] 

For example, according to the S curve in figure 16, the point estimate 
has up to a 40 percent chance of being met, meaning there is a 60 
percent chance that costs will be greater than $825,000. On the basis 
of this information, management could decide to add $82,900 to the 
point estimate to increase the probability to 50 percent or $271,000 to 
increase the confidence level to 70 percent. The important thing to 
note, however, is the large cost increase between the 70 percent and 95 
percent confidence levels—about $600,000—indicating that a substantial 
investment would be necessary to reach a higher level of certainty. 

Management can use the data in an S curve to choose a defensible level 
of contingency reserves. While no specific confidence level is 
considered a best practice, experts agree that program cost estimates 
should be budgeted to at least the 50 percent confidence level, but 
budgeting to a higher level (for example, 70 percent to 80 percent, or 
the mean) is now common practice. Moreover, they stress that 
contingency reserves are necessary to cover increased costs resulting 
from unexpected design complexity, incomplete requirements, technology 
uncertainty, and industrial base concerns, to name a few uncertainties 
that can affect programs. 

How much contingency reserve should be allocated to a program beyond 
the 50 percent confidence level depends on the program cost growth an 
agency is willing to risk. Some organizations adopt other levels 
like the 70th or 80th percentile (refer to the S curve above) to: 
 
1. reduce their anxiety about success within budget, 

2. make some provision for risks unknown at the time but likely to 
appear as the project progresses, and, 

3. reduce the probability that they will have to explain overruns or 
rebaseline because they ran out of reserve budget. 

The amount of contingency reserve should be based on the level of 
confidence with which management chooses to fund a program, based on 
the probabilities reported in the S curve. In figure 16, management 
might choose to award a contract for $907,900 but fund the program at 
$1,096,000. This alternative would provide an additional $188,000 in 
contingency reserve at the 70 percent confidence level. The result 
would be only a 30 percent chance that the program would need 
additional funding, given the identification and quantification of the 
risks at the time of the analysis. 

Another benefit of using an S curve is that management can proactively 
monitor a program’s costs, because it knows the probability for 
incurring overruns. By understanding which input variables have a 
significant effect on a program’s final costs, management can devote 
resources to acquire better knowledge about them so that risks can be 
minimized. Finally, knowing early what the potential risks are enables 
management to prepare contingencies to monitor and mitigate them using 
an EVM system once the program is under contract. 

The bottom line is that management needs a risk-adjusted point estimate 
based on an estimate of the range of confidence to make wise decisions. 
Using information from an S curve with a realistic probability 
distribution, management can quantify the level of confidence in 
achieving a program within a certain funding level. It can also 
determine a defensible amount of contingency reserve to quickly 
mitigate risk. 

Developing a Credible S Curve Of Potential Program Costs: 

Since an S curve is vital to knowing how much confidence management can 
have in a given point estimate, it is important to know the activities 
in developing one. Seven steps are associated with developing a 
justifiable S curve: 

1. determine the program cost drivers and associated risks; 

2. develop probability distributions to model various types of 
uncertainty (for example, program, technical, external, organizational, 
program management including cost estimating and scheduling); 

3. account for correlation between cost elements to properly capture 
risk; 

4. perform the uncertainty analysis using a Monte Carlo simulation 
model; 

5. identify the probability level associated with the point estimate; 

6. recommend sufficient contingency reserves to achieve levels of 
confidence acceptable to the organization; and; 

7. allocate, phase, and convert a risk-adjusted cost estimate to then-
year dollars and identify high-risk elements to help in risk mitigation 
efforts. 

To take these steps, the cost estimator must work with the program 
office and technical experts to collect the proper information. Short-
changing or merely guessing at the first two steps does not lead to a 
credible S curve and can give management a false sense of confidence in 
the information. 

Step 1: Determine Program Cost Drivers and Associated Risks: 
 
In chapter 13, we noted that one of the benefits of a sensitivity 
analysis is a list of the program cost drivers. Since numerous risks 
can influence the estimate, they should be examined for their sources 
of uncertainty and potential effect, and they should be modeled to 
determine how they can affect the uncertainty of the cost estimate. For 
example, undefined or unknown technical information, uncertain economic 
conditions, unexpected schedule problems, requirements growth, security 
level changes, and political issues are often encountered during a 
program’s acquisition. Each of these risks can negatively or positively 
affect a program’s cost. This means that uncertainty can cause the 
actual cost or schedule to differ from any current plan either in a 
positive or beneficial direction or in a negative or harmful direction. 
In addition, new technologies may be proposed that can fail outright, 
causing rework and unexpected cost growth. 

Risks are also associated with the estimating process itself. For 
instance, historical data from which to make a credible estimate can be 
lacking. When this happens, a cost estimator has no choice but to 
extrapolate with existing methods or develop a new estimating approach. 
No matter the method, some error will be introduced into the estimate. 

Accounting for all possible risks is necessary to adequately capture 
the uncertainty associated with a program’s point estimate. Far from 
exhaustive, table 21 describes some of the many sources of risk. It is 
only a starting point, since each program is unique. 

Table 21: Potential Sources of Program Cost Estimate Uncertainty: 

Uncertainty: Business or economic; 
Definition: Variations from change in business or economic assumptions; 
Example: Changes in labor rate assumptions—e.g., wages, overhead, 
general and administrative cost—supplier viability, inflation indexes, 
multiyear savings assumptions, market conditions, and competitive 
environment for future procurements. 

Uncertainty: Cost estimating; 
Definition: Variations in the cost estimate despite a fixed 
configuration baseline; 
Example: Errors in historical data and cost estimating relationships, 
variation associated with input parameters, errors with analogies and 
data limitations, data extrapolation errors, optimistic learning and 
rate curve assumptions, using the wrong estimating technique, omission 
or lack of data, misinterpretation of data, incorrect escalation 
factors, overoptimism in contractor capabilities, optimistic savings 
associated with new ways of doing business, inadequate time to develop 
a cost estimate. 
 
Uncertainty: Program; 
Definition: Risks outside the program office control; 
Example: Program decisions made at higher levels of authority, indirect 
events that adversely affect a program, directed funding cuts, multiple 
contractor teams, conflicting schedules and workload, lack of 
resources, organizational interface issues, lack of user input when 
developing requirements, personnel management issues, organization’s 
ability to accept change, other program dependencies. 

Uncertainty: Requirements; 
Definition: Variations in the cost estimate caused by change in the 
configuration baseline from unforeseen design shifts; 
Example: Changes in system architecture (especially for system of 
systems programs), specifications, hardware and software requirements, 
deployment strategy, critical assumptions, program threat levels, 
procurement quantities, network security, data confidentiality. 

Uncertainty: Schedule; 
Definition: Any event that changes the schedule: stretching it out may 
increase funding requirements, delay delivery, and reduce mission 
benefits; 
Example: Amount of concurrent development, changes in configuration, 
delayed milestone approval, testing failures requiring rework, 
infeasible schedule with no margin, overly optimistic task durations, 
unnecessary activities, omission of critical reviews. 
 
Uncertainty: Software; 
Definition: Cost growth from overly optimistic assumptions about 
software development; 
Example: Underestimated software sizing, overly optimistic software 
productivity, optimistic savings associated with using commercial off-
the-shelf software, underestimated integration effort, lack of 
commercial software documentation, underestimating the amount of glue 
code needed, configuration changes required to support commercial 
software upgrades, changes in licensing fees, lack of support for older 
software versions, lack of interface specification, lack of software 
metrics, low staff capability with development language and platform, 
underestimating software defects. 

Uncertainty: Technology; 
Definition: Variations from problems associated with technology 
maturity or availability; 
Example: Uncertainty associated with unproven technology, obsolete 
parts, optimistic hardware or software heritage assumptions, 
feasibility of producing large technology leaps, relying on lower 
reliability components, design errors or omissions. 
 
Source: DHS, DOD, DOE, NASA, OMB, SCEA, and industry. 

[End of table] 

Collecting high-quality risk data is key to a successful analysis of 
risk. Often there are no historical data from which to derive the 
information needed as inputs to a risk analysis of cost or schedule. 
Usually most risk data are derived from in-depth interviews or in risk 
workshops. In other words, the data used in program risk analyses are 
often based on individuals’ expert judgment, which depends on the 
experience of the interviewees and may be biased. The success of data 
collection depends also on the risk maturity of the organization’s 
culture. It is difficult to collect useful risk analysis data when the 
organization is indifferent or even hostile to expressing risk in the 
program. Obtaining risk information from staff outside the acquisition 
program office can help balance potential optimism. 

After identifying all possible risks, a cost estimator needs to define 
each one in a way that facilitates determining the probability of each 
risk occurring, along with the cost effect. To do this, the estimator 
needs to identify a range of values and their respective probabilities— 
either based on specific statistics or expressed as best case, worst 
case, and most likely—and the rationale for choosing the variability 
discussed. While the best practice is to rely on historical data, if 
these data are not available, how qualitative judgment was applied 
should be explained (e.g., not planning for first time success in 
testing). Because the quality and availability of the data affect the 
cost estimate’s uncertainty, these should be well documented and 
understood. For example, a cost estimate based on detailed actual data 
in significant quantities will yield a more confident estimate than one 
based on an analogy using only a few data points. 

Since collecting all this information can be formidable, it should be 
done when the data are collected to develop the estimate. Interviews 
with experts familiar with the program are good sources of how varied 
the risks are for a particular cost element. However, experts do not 
always think in extremes. They tend instead to estimate probability 
ranges that represent only 60 percent to 85 percent of the possible 
outcomes, so adjustments may have to be made to consider a wider 
universe of risks. In addition, the technical baseline description 
should address the minimum and maximum range, as well as the most 
likely value for critical program parameters. 

Several approaches, ranging from subjective judgment to complex 
statistical techniques, are available for dealing with uncertainty. 
Here we describe different ways of determining the uncertainty of a 
cost estimate. 

Cost Growth Factor: 

Using the cost growth factor, the cost estimator reflects on 
assumptions and judgments from the development of the cost estimate and 
then makes a final adjustment to the estimate. This is usually a 
percentage increase, based on historical data from similar programs, or 
an adjustment solicited from expert opinion and based on experience. 
This yields a revised cost estimate that explicitly recognizes the 
existence of uncertainty. It can be applied at the total program level 
or for one or more WBS elements. The advantages of this approach are 
that it is easy to implement, takes little time to perform, and 
requires minimal detail. Its several problems are that it requires 
access to a credible historical database, the selection of comparative 
projects and adjustment factors that can be subjective, and new 
technologies or lessons learned that may cause historical data to be 
less relevant. 

Expert Opinion: 

An independent panel of experts can be gathered to review, understand, 
and discuss the system and its costs, in order to quantify the 
estimate’s uncertainty and adjust the point estimate. This approach is 
often used in conjunction with the Delphi technique, in which several 
experts provide opinions independently and anonymously. The results are 
summarized and returned to the experts, who are then given the 
opportunity to change or modify their opinions, based on the opinions 
of the group as a whole. If successful, after several such iterations, 
the expert opinions converge. 

The strengths of this approach are directly related to the diversity 
and experience of the panel members. The major weaknesses are that it 
can be time consuming and experts can present biased opinions. For 
example, some of the largest risks facing a program may stem from a new 
technology for which there is little previous experience. If the risk 
distributions rest on the beliefs of the same experts who may be 
stakeholders, it could be difficult to truly capture the program risks. 
A typical rule of thumb is that lower and upper bounds estimated by 
experts should be interpreted as representing the 15 percent and 85 
percent levels, respectively, of all possible outcomes. Therefore, the 
cost estimator will need to adjust the distribution bounds to account 
for skew (see step 2 for more on this issue). Cost estimators can also 
mitigate bias by avoiding leading questions and by questioning all 
assumptions to see if they are backed by historical data. 

The analytic hierarchy process, like the Delphi technique, is another 
approach to making the best of expert opinion. It can be applied to the 
opinion of either an individual or a panel of experts and mitigates the 
problems of bias that result from group think or dominating 
personalities. The analytic hierarchy process provides a structured way 
to address complicated decisions: it relies on a framework for 
quantifying decision elements and evaluating various alternatives. This 
process allows for effective decision making because it captures both 
subjective and objective evaluation parameters, which can lead to less 
bias and help determine the best overall decision. The approach relies 
on mathematics to organize pair-wise comparisons of decision components 
and prioritizes the results to arrive at a stable outcome. 

Mathematical Approaches: 

Mathematical approaches rely on statistics to describe the variance 
associated with an analogy or a cost estimating relationship. The most 
common approach is to collect data on the optimistic, most likely, and 
pessimistic range (the “3-point estimate”) for the risk or the cost 
element schedule activity duration. Statistics like the standard error 
of the estimate and confidence intervals are more difficult to collect 
from program participants and are not commonly used. Some distributions 
use more exotic inputs such as “shape parameters” that are often 
difficult to collect, even in the most in-depth interviews. Therefore, 
the 3-point estimate and an idea about the distribution shape can be 
used to define the probability distribution to be used in a simulation. 
Probability distributions are used either to characterize risks that 
are assigned to cost elements or activity durations or as estimates of 
uncertainty in costs or durations that may be affected by several 
risks. With either of these approaches, in the simulation the lower-
level WBS element cost probabilities are combined to develop an overall 
program cost estimate probability distribution. 

A benefit of this approach is that it complements the decomposition 
approach to cost estimating. In addition, the emergence of commercial 
software models means that Monte Carlo simulation can be implemented 
quickly and easily, once all the data have been collected. Some 
drawbacks to the approach include input distributions that can be 
various, correlation between cost elements needs to be included, and 
decision makers may not always accept the output. In addition, high-
quality risk data are sometimes difficult and may be expensive to 
collect. 

Technology Readiness Levels: 

NASA and the Air Force Space Command, among other organizations, 
address uncertainty by applying readiness levels, which capture the 
risk associated with developing state-of-the-art technology. They have 
historically developed technology readiness levels to indicate how 
close a given technology is to being available. Technology readiness 
levels are rated on a scale from 1 to 9, with 1 representing paper 
studies of a technology’s feasibility and 9 representing technology 
completely integrated into a finished product. In appendix XII, we list 
and describe nine technology readiness levels. 

Knowing a technology’s readiness level allows a cost estimator to judge 
the risk inherent in assuming it will be available for a given program. 
For example, GAO has determined that level 7—demonstration of a 
prototype in an operational environment—is the level of technological 
maturity that constitutes low risk for starting a product development 
program. One needs to be cautious, since programs can inflate the 
level. There should be specific evidence that a program has achieved 
the claimed technology readiness level, such as that physical and 
functional interfaces are clearly defined, raw materials are available, 
and manufacturing procedures are set up and undergoing testing for 
proof of concept before accepting a claim as true. 

Software Engineering Institute Maturity Models: 

SEI has developed a variety of models that provide a logical framework 
for assessing whether an organization has the necessary process 
discipline to repeat earlier successes on similar projects. 
Organizations that do not satisfy the requirements for the “repeatable” 
level are by default judged to be at the initial level of 
maturity—meaning that their processes are ad hoc, sometimes even 
chaotic, with few of the processes defined and success dependent mainly 
on the heroic efforts of individuals. The lower the maturity, the 
higher the risk that a program will incur cost overruns. 

In addition to evaluating software risks, SEI’s risk evaluation method 
can be tailored to address hardware and organizational risks with a 
program. This method includes identifying and quantifying risk using a 
repeatable process for eliciting risks from experts. Furthermore, using 
SEI’s taxonomy, the risk evaluation method provides a consistent 
framework for employing risk management methods and mitigation 
techniques. 

Schedule Risk Analysis: 

Schedule risk analysis captures the risk that schedule durations may 
increase from technical challenges, lack of qualified personnel, and 
too few staff to do the work. Schedule risk analysis examines the 
effect of activities and events slipping on a program’s critical path 
or the longest path through the network schedule. A program schedule 
delay will have cost effects for all aspects of a program, including 
systems engineering and program management. It also analyzes how 
various activities affect one another because of precedence 
relationships—activity C cannot begin until activities A and B are 
finished—and how a slip in one activity affects the duration of other 
activities when concurrence is high among tasks. By applying 
probabilistic distributions to capture the uncertainty with traditional 
early start–late start and early finish–late finish schedule durations, 
using optimistic, pessimistic, and most likely values, a cost estimator 
can draw a better picture of the true critical path and any cost 
effects to the program. In addition, this analysis addresses the 
feasibility of the program plan as well as the effect of not meeting 
the anticipated finish date. 

Risk Cube (Probability Impact Matrix) Method: 

The risk cube method prioritizes uncertainties that could jeopardize 
program cost, schedule, performance, and quality objectives in terms of 
probability of occurrence and cost effect. Subject matter experts, 
typically engineers and others familiar with the program, define the 
risk factors, probabilities, and cost effect for each identified risk. 
Using these data, the cost estimator develops the expected cost overrun 
by multiplying the cost impact by each risk factor’s probability of 
occurrence. A common technique for engaging those knowledgeable about 
the program is creating a two-dimensional matrix like the one in figure 
17. 

Figure 17: A Risk Cube Two-Dimensional Matrix: 

[Refer to PDF for image: illustration] 

The illustration shows an steadily increasing amount of risk, from low 
to medium, to high when plotted as follows: 

Probability: The likelihood that an objective will not plan be met if 
the current plan is used; 

plotted against: 

Consequence: The program penalty incurred if the objective is not 
obtained. 

Source: GAO. 

[End of figure] 

In the risk cube (P-I matrix) method, risks are mapped onto the matrix, 
based on the severity of the consequence—ranging from low risk = 1 to 
high risk = 5—and the likelihood of their occurring—ranging from low 
likelihood = 1 to high = 5. Risks that fall in the upper right quadrant 
are the most likely to occur and have the greatest consequences for the 
program, compared to risks that fall into the lower left quadrant. 

When risks are plotted together, management can quickly determine which 
ones have top priority. For a risk cube (P-I matrix) analysis to be 
accurate, complete lists of all risks are needed, as well as accurate 
probabilities of occurrence and cost impacts. Determining the cost 
impact will vary by program and WBS element, but a cost impact could, 
for example, be categorized as “60 percent more funding is required to 
resolve a technical shortfall that has no acceptable workarounds.” Once 
the cost impacts are identified, they are mapped to the appropriate WBS 
elements to help identify risk mitigation steps that would be most 
beneficial. 

The advantages of using this approach are that those knowledgeable 
about the program can readily understand and relate to risks presented 
in this manner and that decision makers can understand the link between 
specific risks and consequences. A disadvantage is that engineers may 
not always know the cost impacts and may not account for the full 
spectrum of possible outcomes. Moreover, this method can underestimate 
total risk by omitting the correlation between technical risk and level 
of effort in activities like program management. 

Risk Scoring: 
 
Risk scoring quantifies and translates risks into cost impacts. Risk 
scoring is used to determine the amount of risk, preferably using an 
objective method in which the intervals between a score have meaning—a 
score of 1 = low risk, a score of 5 = medium risk, and a score of 10 = 
high risk. This method is used most often to determine technical risk 
associated with hardware and software. The following categories are 
used for hardware: technology advancement (level of maturity), 
engineering development (current stage of development), reliability 
(operating time without failure), producibility (ease to manufacture), 
alternative item (availability of back-up item), and schedule (amount 
of aggressiveness). Table 22 is an example of the hardware risk scoring 
matrix.[Footnote 56] 
 
Table 22: A Hardware Risk Scoring Matrix: 

Risk score: 0 = low, 5 = medium, 10 = high: 

Risk category: 1. Technology advancement; 
0: Completed, state of the art; 
1–2: Minimum advancement required; 
3–5: Modest advancement required; 
6–8: Significant advancement required; 
9–10: New technology. 

Risk category: 2. Engineering development; 
0: Completed, fully tested; 
1–2: Prototype; 
3–5: Hardware and software development; 
6–8: Detailed design; 
9–10: Concept defined. 
 
Risk category: 3. Reliability; 
0: Historically high for same system;
1–2: Historically high on similar systems;
3–5: Modest problems known;
6–8: Serious problems known;
9–10: Unknown. 
 
Risk category: 4. Producibility; 
0: Production and yield shown on same system;
1–2: Production and yield shown on similar system;
3–5: Production and yield feasible;
6–8: Production feasible and yield problems; 
9–10: No known production experience. 
 
Risk category: 5. Alternative item; 
0: Exists or availability on other items not important;
1–2: Exists or availability on other items somewhat important;
3–5: Potential alternative in development;
6–8: Potential alternative in design;
9–10: Alternative does not exist and is required. 
 
Risk category: 6. Schedule; 
0: Easily achieved; 
1–2: Achievable; 
3–5: Somewhat challenging;
6–8: Challenging; 
9–10: Very challenging.

Source: © 2003, Society of Cost Estimating and Analysis (SCEA), “Cost 
Risk Analysis.” 

[End of table] 

In addition to hardware, categories for software include technology 
approach (level of innovation), design engineering (current stage of 
development), coding (code maturity), integrated software (based on the 
source lines of code count), testing (amount completed), alternatives 
(availability of back-up code), and schedule (amount of 
aggressiveness). A software risk scoring matrix is shown in table 23. 

Table 23: A Software Risk Scoring Matrix: 

Risk score: 0 = low, 5 = medium, 10 = high: 

Risk category: 1. Technology advancement; 
0: Proven conventional analytic approach, standard methods; 
1–2: Undemonstrated conventional approach, standard methods; 
3–5: Emerging approaches, new applications; 
6–8: Unconventional approach, concept in development; 
9–10: Unconventional approach, concept unproven. 

Risk category: 2. Design engineering; 
0: Design complete and validated; 
1–2: Specifications defined and validated; 
3–5: Specifications defined; 
6–8: Requirements defined; 
9–10: Requirements partly defined. 

Risk category: 3. Coding; 
0: Fully integrated code available and validated; 
1–2: Fully integrated code available; 
3–5: Modules integrated; 
6–8: Modules exist but not integrated; 
9–10: Wholly new design, no modules exist. 

Risk category: 4. Integrated software; 
0: Thousands of instructions;
1–2: Tens of thousands of instructions;
3–5: Hundreds of thousands of instructions;
6–8: Millions of instructions; 
9–10: Tens of millions of instructions. 

Risk category: 5. Testing; 
0: Tested with system; 
1–2: Tested by simulation; 
3–5: Structured walk-throughs conducted; 
6–8: Modules tested but not as a system; 
9–10: Untested modules. 

Risk category: 6. Alternatives; 
0: Alternatives exist, alternative design not important; 
1–2: Alternatives exist, design somewhat important; 
3–5: Potential for alternatives in development; 
6–8: Potential alternatives being considered; 
9–10: Alternative does not exist but is required. 

Risk category: 7. Schedule and management; 
0: Relaxed schedule, serial activities, high review cycle frequency, 
early first review; 
1–2: Modest schedule, few concurrent activities, review cycle 
reasonable; 
3–5: Modest schedule, many concurrent activities, occasional reviews, 
late first review; 
6–8: Fast track on schedule, many concurrent activities; 
9–10: Fast track, missed milestones, review at demonstrations only, no 
periodic reviews. 

Source: U.S. Air Force. 

[End of table] 

Technical engineers score program elements between 0 and 10 for each 
category and then rank the categories according to the program’s 
effect. Next, each element’s risk score is translated into a cost 
impact by (1) multiplying a factor by an element’s estimated cost (for 
example, a score of 2 increases the cost of an element by 10 percent) 
or (2) multiplying a factor by predetermined costs (a score of 2 has a 
cost impact of $50,000) or (3) developing a weighted average risk 
assessment score that is mapped to a historical cost growth 
distribution. 

After using one or several of these methods to determine the cost risk, 
the estimator’s next step is to choose probability distributions to 
model the risk for each WBS cost element that has uncertainty. 

Step 2: Develop Probability Distributions to Model Uncertainty: 
 
Uncertainty is best modeled with a probability distribution that 
accounts for all possible outcomes according to the probability that 
they will occur. Figure 18 gives an example of a known distribution 
that models all outcomes associated with rolling a pair of dice. 

Figure 18: The Distribution of Sums from Rolling Two Dice: 

[Refer to PDF for image: illustration] 

Probability plotted versus Value; 

0 probability: that outcome is less than 2; 
50% probability: that the outcome is above or below 7 (this is the 
median); 
100% probability: that outcome will not exceed 12. 

Source: GAO. 

[End of figure] 

In figure 18, the horizontal axis shows the potential value of dice 
rolls, while the vertical axis shows the probability associated with 
each roll. The value at the midpoint of all rolls is the median. In the 
example, the median is also the most likely value (that is, average = a 
roll of 7), because the outcomes associated with rolling a pair of dice 
are symmetric. 

Besides descriptive statistics, probability distributions provide other 
useful information, such as the boundaries of an outcome. For example, 
the lower bound in figure 18 is 2 and the upper bound is 12. By 
examining the distribution, it is easy to see that both the upper and 
lower bounds have the lowest probability of occurring, while the 
chances of rolling a 6, 7, or 8 are much greater. 

It is difficult to pick an appropriate probability distribution for the 
point cost estimate as a whole, because it is composed of several 
subsidiary estimates based on the WBS. These WBS elements are often 
estimated with a variety of techniques, each with its own uncertainty 
distributions that may be asymmetrical. Therefore, just simply adding 
the most likely WBS element costs does not result in the most likely 
cost estimate because the risk distributions associated with the 
subelements differ. 

One way to resolve this issue is to create statistical probability 
distributions for each WBS element or risk by specifying the risk shape 
and bounds that reflect the relative spread and skewness of the 
distribution. The probability distribution represents the risk shape, 
and the tails of the distribution reflect the best and worst case 
outcomes. Even though the bounds are extremes and unlikely to occur, 
the distribution acknowledges the possibility and probability that they 
could happen. Probability distributions are typically determined using 
the 3-point estimates of optimistic, most likely, and pessimistic 
values to identify the amount of spread and skewness of the data. 
However, if risks are used directly, they will be assigned to specific 
cost elements or activities in a schedule and will perform 
appropriately in a simulation.[Footnote 57] 

Using a simulation tool such as Monte Carlo, a cost estimator can 
develop a statistical summation of all probable costs, allowing for a 
better understanding of how likely it is that the point estimate can be 
met. A Monte Carlo simulation also does a better job of capturing risk, 
because it takes into consideration that some risks will occur while 
others may not. Furthermore, the simulation can adjust the risks beyond 
the upper and lower bounds to account for the fact that experts do not 
typically think in extremes. Figure 19 shows why different WBS element 
distributions need to be statistically summed in order to develop the 
overall point estimate probability distribution. 

Figure 19: A Point Estimate Probability Distribution Driven by WBS: 

[Refer to PDF for image: illustrations] 

Inputs: 
Probability distributions for each cost element in a system’s work 
breakdown structure; 

Outputs: 
A cumulative probability distribution of the system’s total cost. 

Source: NASA. 

Note: RPE = reference point estimate. 

[End of figure] 

In figure 19, the sum of the reference point estimates has a low level 
of probability on the S curve. In other words, there is only a 20 
percent chance or less of meeting the point estimate cost. Therefore, 
in order to increase the confidence in the program cost estimate, it 
will be necessary to add more funding to reach a higher level of 
confidence. 

Next to knowing the bounds or 3-point estimates for the uncertainty of 
the WBS element or risk, choosing the right probability distribution 
for each WBS element is important for capturing the uncertainty 
correctly. For any WBS element, selecting the probability distribution 
should be based on how effectively it models actual outcomes. Since 
different distributions model different types of risk, knowing the 
shape of the distribution helps in visualizing how the risk will affect 
the overall cost estimate uncertainty. A variety of probability 
distribution shapes are available for modeling cost risk. Table 24 
lists eight of the most common probability distributions used in cost 
estimating uncertainty analysis. 

Table 24: Eight Common Probability Distributions: 

Distribution: Bernoulli; 
Description: Assigns probabilities of “p” for success and “1 – p” for 
failure; mean = “p”; variance = “1 – p”; 
Shape: [Refer to PDF for image]; 
Typical application: With likelihood and consequence risk cube models; 
good for representing the probability of a risk occurring but not for 
the impact on the program. 

Distribution: Beta; 
Description: Similar to normal distribution but does not allow for 
negative cost or duration, this continuous distribution can be 
symmetric or skewed; 
Shape: [Refer to PDF for image]; 
Typical application: To capture outcomes biased toward the tail ends of 
a range; often used with engineering data or analogy estimates; the 
shape parameters usually cannot be collected from interviewees. 

Distribution: Lognormal; 
Description: A continuous distribution positively skewed with a 
limitless upper bound and known lower bound; skewed to the right to 
reflect the tendency toward higher cost; 
Shape: [Refer to PDF for image]; 
Typical application: To characterize uncertainty in nonlinear cost 
estimating relationships; it is important to know how to scale the 
standard deviation, which is needed for this distribution. 

Distribution: Normal; 
Description: Used for outcomes likely to occur on either side of the
average value; symmetric and continuous, allowing for negative costs 
and durations. In a normal distribution, about 68% of the values fall 
within one standard deviation of the mean; 
Shape: [Refer to PDF for image]; 
Typical application: To assess uncertainty with cost estimating 
methods; standard deviation or standard error of the estimate is used 
to determine dispersion. Since data must be symmetrical, it is not as 
useful for defining risk, which is usually asymmetrical, but can be 
useful for scaling estimating error. 

Distribution: Poisson; 
Description: Peaks early and has a long tail compared to other 
distributions; 
Shape: [Refer to PDF for image]; 
Typical application: To predict all kinds of outcomes, like the number 
of software defects or test failures. 

Distribution: Triangular; 
Description: Characterized by three points (most likely, pessimistic, 
and optimistic values), can be skewed or symmetric and is easy to 
understand because it is intuitive; one drawback is the absoluteness of 
the end points, although this is not a limitation in practice since it 
is used in a simulation; 
Shape: [Refer to PDF for image]; 
Typical application: To express technical uncertainty, because it works 
for any system architecture or design; also used to determine schedule 
uncertainty. 

Distribution: Uniform; 
Description: Has no peaks because all values, including highest and 
lowest possible values, are equally likely; 
Shape: [Refer to PDF for image]; 
Typical application: With engineering data or analogy estimates. 

Distribution: Weibull; 
Description: Versatile, can take on the characteristics of other 
distributions, based on the value of the shape parameter “b”— e.g., 
Rayleigh and exponential distributions can be derived from it[A]; 
Shape: [Refer to PDF for image]; 
Typical application: In life data and reliability analysis because it 
can mimic other distributions and its objective relationship to 
reliability modeling. 

Source: DOD, NASA, SCEA, and Industry. 

[A] The Rayleigh and exponential distributions are a class of 
continuous probability distribution. 

[End of table] 

The triangular, lognormal, beta, uniform, and normal distributions in 
table 24 are the most common distributions that cost estimators may use 
to perform an uncertainty analysis. They are generally sufficient, 
given the quality of the information derived from interviews and the 
granularity of the results. However, many other types of distributions 
are discussed in myriad literature sources and are available through a
variety of statistical tools. 

The point to remember is that the shape of the distribution is 
determined by the characteristics of the risks they represent. If they 
are applied to WBS elements, they may combine the impact of several 
risks, so it may take some thought to determine the most appropriate 
distribution to use. For a CER, it is a best practice to use prediction 
interval statistical analysis to determine the bounds of the 
probability distribution because it is an objective method for 
determining variability. The prediction interval captures the error 
around a regression estimate and results in a wider variance for the 
CER. 

When there is no objective way to pick the distribution bounds, a cost 
estimator will resort to interviewing several people—especially 
experienced personnel both within and outside the program—about what 
the distribution bounds should be. Promising anonymity to the 
interviewees may help secure their unbiased thoughts. Separating the 
risk analysis function organizationally from the program and program 
manager often provides the needed independence to withstand political 
and other pressures for biased results. 

One way to avoid the potential for experts to be success oriented when 
choosing the upper and lower extremes of the distribution is to look 
for historical data that back up the distribution range. If historical 
data are not available, it may be necessary to adjust the tails to 
account for the fact that being overly optimistic usually results in 
programs costing more and taking longer than planned. Thus, it is 
necessary to skew the tails to account for this possibility in order to 
properly represent the overall risk. Organizations should, as a best 
practice, examine and publish default distribution bounds that cost 
estimators can use when the data cannot be obtained objectively. 

Once all cost element risks have been identified in step 1 and 
distributions have been chosen to model them in step 2, correlation 
between the cost elements must be examined in order to fully capture 
risk, especially risk related to level-of-effort cost elements. 

Step 3: Account for Correlation between Cost Elements: 

Because different WBS elements’ costs may be affected by the same 
external factors, some degree of correlation exists between them. 
Correlation identifies the relationship between WBS elements such that 
when one WBS element’s cost is high within its own probability 
distribution, the other WBS element will also show a high cost in its 
own probability distribution. Thus, correlated cost elements should 
rise and fall together. Without correlating the two elements, 
inconsistent scenarios where one is high and the other is low could 
occur during the simulation, causing erroneous results. Therefore, a 
change in one WBS element’s cost will usually be found with a change in 
the same direction (if positive correlation) in another element’s cost. 
If this is so for many elements, the cumulative effect tends to 
increase the range of possible costs. Consider the following examples: 

* If a supplier delivers an item late, other scheduled deliveries could 
be missed, resulting in additional cost. 

* If technical performance problems occur, unexpected design changes 
and unplanned testing may result, affecting the final schedule and 
cost. 

* If concurrence is great between activities, a slip in one activity 
could have a cascading effect on others, resulting in a high degree of 
schedule and cost uncertainty. 

* If the number of software lines of code depends heavily on the 
software language and the definition of what constitutes a line of 
code, a change in the counting definition or software language will 
change the number of lines of code affecting both schedule and cost. 

As these examples show, many parts of a cost estimate may move 
together, and when they do, summing their costs results in 
reinforcement in both negative and positive directions. Therefore, 
mitigating a risk that affects two or more WBS cost elements can reduce 
uncertainty on several cost elements. A case in point is the standing 
army effect, which occurs when a schedule slip in one element results 
in delays in many other cost elements as staff wait to complete their 
work. As such, interelement correlation must be addressed so that the 
total cost probability distribution properly reflects the risks. 

To properly capture functional correlation, the cost model should be 
structured with all dependencies intact. For instance, if the cost of 
training is modeled as a factor of hardware cost, then any uncertainty 
in the hardware cost will be positively correlated to the risk in 
training cost. Thus, when the simulation is run, risks fluctuating 
within main cost element probability distributions will accurately flow 
down to dependent WBS elements. 

One of the advantages of a cost estimating relationship based cost 
model is the manner in which the statistical analysis used to derive 
the CERs can also be drawn on to identify and, in some cases, quantify 
the correlations between various cost risk elements. It is also 
important to ensure that uncertain cost method inputs (weight, labor 
rates) are correlated appropriately. 

In some cases, however, it may be necessary to inject correlation to 
“below the line” dependent elements to account for correlated risk. 
These elements are typically level-of-effort support activities, like 
systems engineering and program management. In addition, correlation 
may have to be injected into the cost model to account for effects that 
the model may not capture. For example, a program risk may be that the 
length of an aircraft wing increases. If that happens, a larger engine 
than was originally estimated would then be required. Because this risk 
effect is not correlated in the cost model, it must be injected into 
the risk scenario. 

Estimators should examine the correlation coefficients from the 
simulation model to determine the amount of correlation that already 
exists in the cost model. As a rule of thumb, it is better to insert an 
overall nominal correlation of 0.25 than to show no correlation at all. 
This will prevent the simulation from drawing a low value for one 
element and a high value for another element, causing a cancellation of 
risk when both elements are positively correlated. 

Regardless of which approach is taken, it is important to note that 
correlation should never be ignored. Doing so can significantly affect 
the cost risk analysis, resulting in a dramatically understated 
probability distribution that can create a false sense of confidence in 
the resulting estimate. Therefore, highly risky programs should show a 
wide range of possible costs. (More information on correlation and how 
to account for schedule risk affecting the cost estimate is in appendix 
X.) 

Step 4: Perform Uncertainty Analysis with a Monte Carlo Simulation: 

The most common technique for combining the individual elements and 
their distributions is Monte Carlo simulation.[Footnote 58] In one 
approach, the distributions for each cost element are treated as 
individual populations from which random samples are taken. In another 
approach, each risk is modeled and assigned to the WBS elements it 
affects; in this approach, a risk may affect more than one WBS 
element’s cost, and a WBS element’s cost may be affected by more than 
one risk. In either case, during the simulation a cost model is 
recalculated thousands of times by repeatedly drawing random values 
from each WBS distribution or distribution of risk factors, so that 
many, thousands of, or nearly all possible outcomes are taken into 
account. The simulation’s output illustrates (1) the likelihood of 
achieving the program’s cost objectives, given the current plan and 
risks as they are known and quantified; (2) the likelihood of other 
possible outcomes, which can be a way to determine the cost value that 
has an acceptable probability of being exceeded; and (3) by 
sensitivity, the high-priority risks or WBS elements as a guide to 
effective risk mitigation. 

Not a new concept, Monte Carlo simulation has been a respected method 
of analyzing risk in engineering and science for more than 60 years. 
Mathematicians working on the Manhattan project used it during 
World War II and this technique was used to determine the value of pi 
(p) to within 6 decimal points. Developed by a mathematician who 
pondered the probabilities associated with winning a card game of 
solitaire, Monte Carlo simulation is used to approximate the 
probability outcomes of multiple trials by generating random numbers. 
In determining the uncertainty associated with a program’s point 
estimate, a Monte Carlo simulation randomly generates values for 
uncertain variables over and over to simulate a model. 

Without the aid of simulation, the analyst generally produces a single 
outcome for the total program cost, usually by adding up the 
individual WBS element cost estimates. This value is not necessarily 
the most likely or average scenario. In fact, without a risk analysis, 
it is not known how adequate this single-point estimate is likely to be 
for handling the program risks. But after hundreds or thousands of 
trials, one can view the frequency distribution of the results and 
determine the certainty of any outcome. Performing an uncertainty 
analysis using Monte Carlo simulation quantifies the amount of cost 
risk within a program. Only by quantifying the cost risk can management 
make informed decisions about risk mitigation strategies and provide a 
benchmark against which to measure progress. 

To perform an uncertainty analysis, each WBS element’s risk or risk 
factor is assigned a specific probability distribution of feasible 
values. In setting up the simulation, any identified causality may be 
modeled. Also, correlations are specified, including identified 
correlated elements and estimated strength of the correlation. These 
are automatically taken into account by the software during the 
simulation, where a random draw from each distribution is taken and the 
results are added up. This random drawing among distributions is 
repeated thousands of times with statistical software in order to 
determine the frequency distribution. Since the simulation’s inputs are 
probability distributions, the outputs are also distributions. The 
result is a distribution of random total program costs based on the 
overall mean and standard deviation. Rather than being normal, the 
total cost distribution is usually lognormal. This happens because the 
overall cost distribution is derived from the lower-level WBS elements, 
each of which has unique distributions. Since many of these underlying 
distributions tend to be skewed to the right, the overall distribution 
is typically lognormal. This makes sense since most cost estimates tend 
to overrun rather than underrun. This distribution can also be 
converted to an S curve like the S curves shown in figures 16 and 19. 

An advantage of using a Monte Carlo simulation is that both good and 
bad effects can be modeled, as well as any offsets that occur when both 
happen at the same time. In addition, Monte Carlo simulation not only 
recognizes the uncertainty inherent in the point estimate but also 
captures the uncertainty with all other possible estimates, allowing 
for a better analysis of alternatives. Using this technique, management 
can base decisions on cost estimate probabilities rather than relying 
on a single point estimate with no level of confidence attached. 

Step 5: Identify the Probability Associated with the Point Estimate: 

After the simulation has been run and causality and correlation have 
been accounted for, the next step is to determine the probability 
associated with the point estimate. The cumulative probability 
distribution resulting from the Monte Carlo simulation provides the 
cost estimator and management with risk-adjusted estimates and 
corresponding probabilities. The output of the simulation is useful for 
determining the level of probability in achieving the point estimate, 
along with a range of possible outcomes bounded by minimum and maximum 
costs. This probability can then be weighed against available funding 
to understand the confidence one can place in the program’s meeting its 
objectives. 

Uncertainty analysis using a Monte Carlo simulation communicates to 
stakeholders how likely a program is to finish at the estimated cost 
and schedule, how much cost contingency reserve is needed to provide 
the desired degree of certainty that the estimate will be adequate, and 
the likely risks so that proactive responses can be developed.[Footnote 
59] It also determines how different two competing alternatives are in 
terms of cost. In addition, estimating future costs with probabilities 
is better than just relying on a point estimate, because informed 
decisions can be made regarding all possible outcomes. 

Because we can never know all the risks until the program is finally 
complete, the risk analysis and cost risk simulation exercise should be 
conducted periodically through the life of the program. Organizations 
often require such an analysis before major milestone decision points. 

Step 6: Recommend Sufficient Contingency Reserves: 

The main purpose of risk and uncertainty analysis is to ensure that a 
program’s cost, schedule, and performance goals can be met. The 
analysis also communicates to decision makers the specific risks that 
contribute to a program’s cost estimate uncertainty. Without this 
knowledge, a program’s estimated cost could be understated and subject 
to underfunding and cost overruns, putting it at risk of being reduced 
in scope or requiring additional funding to meet its objectives. 
Moreover, probability data from an uncertainty analysis can result in 
more equitable distribution of budget in an EVM system, ensuring that 
the most risky cost elements receive adequate budget up front. 

Using information from the S curve, management can determine the 
contingency reserves needed to reach a specified level of confidence. 
The difference in cost between the point estimate and the desired 
confidence level determines the required contingency reserve. Because 
cost distributions tend to be right skewed (that is, the tendency is 
for costs to overrun rather than underrun), the mean of the 
distribution tends to fall somewhere between the 55 percent and 65 
percent confidence levels. Therefore, if it is decided to fund a 
program at the 50 percent confidence level, there is still a chance 
that the program will need additional funding because the expected 
value is at a higher confidence level. Moreover, extremely risky 
programs will require funding at a level closer to the 65 percent 
confidence level or higher. Since each program is unique and so are its 
risks, there are no set rules as to what level of contingency is 
sufficient. Decision makers have to decide the level of confidence at 
which to set the budget. Having adequate funding is paramount for 
optimal program execution, since it can take many months to obtain 
necessary funding to address an emergent program issue. Without 
available risk funding, cost growth is likely. 

We caution that the validity of the results depends on the knowledge, 
experience, and data regarding a program’s risks. When the uncertainty 
analysis has been poorly executed, management may have a false sense of 
security that all risks have been accounted for and that the analysis 
was based on sound data. When this happens, program decisions will be 
based on bad information. Thus, it is imperative that the cost 
estimators properly correlate cost elements and consider a broad range 
of potential program risks rather than simply focusing on the risks 
that most concern the program office or contractor. Furthermore, to 
ensure that best practices have been followed and to prevent errors 
such as not properly accounting for correlation between cost elements, 
it is a best practice to vet the uncertainty analysis through a core 
group of experts to ensure that results are robust and valid. 

In addition, to ensure that accurate information is available for 
performing uncertainty analysis, the estimate should be continually 
updated with actual costs and any variances recorded. This will enable 
organizations to identify areas where estimating was difficult or 
sources of risk were not considered. Doing so will guard against 
presenting misleading results to management and will result in 
continuous improvements in the uncertainty analysis process. 

A program’s early phases entail a lot of uncertainty, and the amount of 
contingency funding required may exceed acceptable levels. Management 
may gain insight from the uncertainty analysis by acting to reduce risk 
to keep the program affordable. It may also examine different levels of 
contingency reserve funds to understand what level of confidence the 
program can afford. Most importantly, management needs to understand 
that any uncertainty analysis or risk mitigation is only as good as the 
comprehensiveness of risks and uncertainties identified. Unknown risks 
could still cause problems, and these are difficult, if not impossible, 
to quantify. 

Step 7: Allocate, Phase, and Convert a Risk-Adjusted Cost Estimate to 
Then-Year Dollars and Identify High-Risk Elements: 

Uncertainty is calculated on the total cost estimate results, not year 
by year. Therefore, since a budget is requested in then-year dollars, 
it is necessary to convert the cost estimate into then-year dollars by 
phasing the WBS element costs over time. Because WBS element results at 
a specific confidence level will not sum to the parent levels, it will 
be necessary to pick the level in the WBS from which risk dollars are 
to be managed. The difference between the point estimate and the risk 
result at the selected confidence level is the amount of contingency 
reserve to be set aside for mitigating risks in lower WBS level 
elements. 

Once the amount of contingency reserve has been identified, reserves 
need to be identified and set aside for the WBS elements that harbor 
the most risks so that funding will be available to mitigate risks 
quickly. To identify which WBS elements may need contingency reserve, 
results from the uncertainty analysis are used to prioritize risks, 
based on probability and impact as they affected the cost estimate 
during the simulation. Knowing which risks are important will guide the 
allocation of contingency reserve. 

Risk Management: 

Risk and uncertainty analysis is just the beginning of the overall risk 
management process. Risk management is a structured and efficient 
process for identifying risks, assessing their effect, and developing 
ways to reduce or eliminate risk. It is a continuous process that 
constantly monitors a program’s health. In this process, program 
management develops risk handling plans and continually tracks them to 
assess the status of program risk mitigation activities and abatement 
plans. In addition, risk management anticipates what can go wrong 
before it becomes necessary to react to a problem that has already 
occurred. Identifying and measuring risk by evaluating the likelihood 
and consequences of an undesirable event are key steps in risk 
management. The risk management process should address five steps: 

1. identify risks, 

2. analyze risks (that is, assess their severity and prioritize them), 

3. plan for risk mitigation, 

4. implement a risk mitigation plan, and, 

5. track risks. 

Steps 1 and 2 should have already been taken during the risk and 
uncertainty analysis. Steps 3–5 should begin before the actual work 
starts and continue throughout the life of the program. Over time, some 
risks will be realized, others will be retired, and some will be 
discovered: Risk management never ends. Establishing a baseline of risk 
expectations early provides a reference from which actual cost risk can 
be measured. The baseline helps program managers track and defend the 
need to apply risk reserves to resolve problems. 

Integrating risk management with a program’s systems engineering and 
program management process permits enhanced root cause analysis and 
consequence management, and it ensures that risks are handled at the 
appropriate management level. Furthermore, successful risk mitigation 
requires communication and coordination between government and the 
contractor to identify and address risks. A common database of risks 
available to both is a valuable t