Full text of "On the feasibility of creating a comparable database for nonrecurring cost analysis under dual source competition"

See other formats

NPS-54-87-004

NAVAL POSTGRADUATE SCHOOL

Monterey, Calilornia

ON THE FEASIBILITY OF CREATING A COMPARABLE

DATABASE FOR NONRECURRING COST ANALYSIS

UNDER DUAL SOURCE COMPETITION

Dan C . Boger
//

and

Shu S. Liao
May 1987

Approved for public release; distribution unlimited.

Prepared for: Naval Air Systems Command

Cost Analysis Division
Washington, D. C. 20361

FedDocs
D 208.IM/2
NPS-54-87-004

NAVAL POSTGRADUATE SCHOOL
Monterey, California

RADM. R. C. Austin
Superintendent

David A. Schrady
Provost

The research summarized herein was sponsored by the Naval Air
Systems Command, Cost Analysis Division under work order
N0001986WR6347W.

Reproduction of all or part of this report is authorized.

This report was prepared by:

1 I AS')'^ A 'lON •^Aijt _ _

■“■■"" DUD tCY KNOX LIBKAIr^^^"

REPORT DOCUMENTATION PAG EnaVAL POSTGRADUATE SCHOOL

Id REPORT SECURITY CLASSimCAT.ON
Unclassified

ID RESTRICTIVE markifMPNTEREV LA ^^3943-6101

2a SECURITY CLASSIFICATION AUTHORITY

3 DlSTRlBuriON/ availability OF REPORT

Approved for public release; distribution
uni imited.

2b DECLASSIFICATION /DOWNGRADING SCHEDULE

4 PERFORMING ORGANIZATION REPORT NUMBER(S)

NPS-54-87-004

S MONITORING ORoANiZATlON REPORT NUMBER(S)

oa NAME OF PERFORMING ORGANIZATION
Naval Postgraduate School

60 OFFICE SYMBOL
(If applicable)

7a NAME OF MONlTORiNG ORGANIZATION

Naval Air Systems Command
Cost Analvsis Division... . ..

bc ADDRESS {City. State, and ZIP Code)
Monterey, CA 93943

7b ADDRESS (C/fy, Stare, and ZIPCode)
Washington, D.C. 20361

8a. NAME OF FUNDING , SPONSORING

ORGANIZATION Naval Air Systems
Command

Cost Analysis Division

So OFFICE SYMBOL
(If applicable)

9 PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER

N0001986WR6347W

8c. ADDRESS (C/fy, Stare, and ZIPCode)
Washington, D.C. 20631

10 SOURCE OF FUNDING NUMBERS

PROGRAM
ELEMENT NO

PROJECT

TASK

NO.

WORK UNIT
ACCESSION NO

11 riTit {tnciuae Security OassitKation) (UNCLASSIFIED)

On the Feasibility of Creating a Comparable Database for Nonrecurring Cost Analysis
Under Dual Source Competition

12 personal AUThOR(S)

Dan C . Boqer and Shu

S. Liao

13a TYPE OF REPORT

13b TIME COVERED

14 DATE OF REPORT {Year. Month. Day)

15 PAGE COUNT

Final report

FROM TO

January 1986 to Sept. 1986

3 3 DP .

16 SUPPLEMENTARY NOTATION

COSATI CODES

18 SUBJECT TERMS {Continue on reverse if necessary and identify by block number)

Field

GROUP

SUB-GROUP

Dual source commetition, nonrecurring costs, cost

drivers, cost estimation relationships

19 abstract {Continue on reverse if neceisary and identify by block number)

This study examines the nature of nonrecurring cost items and develops systematic
procedures for the government and contractors to follow in order to generate a
database of nonrecurring costs which is both consistent and comparable. The objective
was to develop a method for generating an analyzable set of data for parametric
estimation of nonrecurring costs. Cost elements were analyzed and their cost
drives were identified.

ZO^STRlBuTlON/ AVAILABILITY OP ABSTRACT
(SjUNCLASSIFIEO/UNLIMITED □ SAME AS RPT

□ DTIC USERS

21 ABSTRACT SECURITY CLASSIFICATION

22a NAME OF RESPONSIBLE INDIVIDUAL

22b T ELEPHONE (/nc/uc/e Area Code)

22c. OFFICE SYMBOL

DD FORM 1473, 84 mar 83 APK edition may D€ used until exhausted SECURITY CLASSIFICATION OF THIS PAGE

All other editions are obsolete

ON THE FEASIBILITY OF CREATING A COMPARABLE DATABASE
FOR NONRECURRING COST ANALYSIS UNDER
DUAL SOURCE COMPETITION

Dan C . Boger
and

Shu S. Liao

Department of Administrative Sciences
Naval Postgraduate School
Monterey, California 93943

May 1987

EXECUTIVE SUMMARY

A major task in cost estimation is to determine the cost
behavior of a cost element so that the amount of the cost element
may be estimated when the factors driving the cost change. This
report examines the cost behavior of major nonrecurring cost
elements and explores the feasibility of creating a comparable data
base for parametric analysis of nonrecurring costs when a second
supplier is to be established for competitive procurement.

To estimate nonrecurring costs associated with establishing a
second source, both analogy and parametric CER methods have been
used. However, our examination of contractors' cost estimation
methods indicates that government cost analysts face several
difficulties under current cost data reporting systems. These are
discussed in detail in chapter 2.

Since nonrecurring costs consist of several categories of cost
items, a parametric cost estimating model with a small number of
available observations and a large number of potential explanatory
variables would be unreliable, even if possible. A feasible
solution is to disaggregate the nonrecurring costs into relatively

homogeneous groups of cost items for data accumulation purposes.
With a consistent data base and relatively homogeneous cost items,
a parametric model for each group may be constructed with a
relatively small number of observations, a typical constraint in
major weapon system cost estimation.

Our research effort focused on identifying the components of
nonrecurring second source start-up costs, their cost behavior, and
cost drivers. Our findings can be summarized below.

First, it is necessary to separate nonrecurring material costs
from labor and overhead costs for estimating purpose. The main
reason is that the cost drivers for each are not always the same.
Lumping material and labor costs together would require adding more
explanatory variables to the CER model.

Second, parametric CER models for labor cost must be based on
labor hours, not on labor dollars as some cost analysts do now.

Third, the most important explanatory variable that needs to
be operationalized is the complexity of the weapon system. Most of
the manpower needed is variable and is driven, according to
contractors, by the amount of labor hours. In fact, some contractors
have used certain surrogate measures for the level of complexity.
Standardizing the measure of system complexity is necessary if a
parametric CER is to be developed.

Our recommendation is that material and labor costs should be
separately reported for each category defined in the main body of
the report and that defining the complexity level of a system holds
the key to developing viable parametric CER models for nonrecurring
costs .

CHAPTER 1

INTRODUCTION

Competition has been hailed by virtually every corner of the public
and private sectors as a vehicle for reducing the cost of acquiring
weapon systems. However, in order to introduce competition in
major weapon system procurement, a second source of supply must be
created. The cost to establish a competitive second supply source
can be high and difficult to estimate. This report examines the
feasibility of creating a comparable data base for parametric
analysis of nonrecurring costs when a second supplier is to be
established for competitive procurement.

BACKGROUND

In dual sourcing decisions, it is necessary that the nature and
extent of nonrecurring costs be fully analyzed. Nonrecurring costs
are important since they represent those investment costs which
must be incurred before the second source produces any output. It
should be noted that the only way dual sourcing will produce
overall cost savings for the Government is for the present value of
the eventual recurring cost savings to offset the nonrecurring
investment costs. Therefore, good measurement and models of
nonrecurring investment costs are required in order to determine
whether dual sourcing can save costs.

A major task in cost estimation is to determine the cost
behavior of a cost element so that the amount of the cost element

may be estimated when the factors driving the cost change. Our
earlier research has identified several difficulties with the
current state of the estimation of nonrecurring costs of dual
sourcing. 1 First among these difficulties is that there is an
inconsistent treatment of nonrecurring cost elements with their being
predefined as either fixed or variable. The second difficulty is
that different methods were used to aggregate costs into the cost
elements used for analysis. These difficulties result in the data
of nonrecurring costs compiled by specific sources being noncom-
parable with data compiled by other sources, thus rendering the
data unusable for analytical purposes. The lack of acceptable
nonrecurring cost models can be partially attributed to these
difficulties .

To address this problem, it is necessary to develop systematic
procedures for the government and contractors to follow in order to
generate a database of nonrecurring costs which is both consistent
and comparable. These procedures should provide specific and
particular guidelines for all parties to a contract to follow in
order to produce a consistent and comparable database of nonrecurr-
ing costs for parametric analysis.

OBJECTIVE OF THE PROJECT

The objective of the project is to examine the nature of

1 Dan C. Boger and Shu S. Liao, "An Analysis of Quantity-
Split and Nonrecurring Costs Under a Competitive Procurement
Environment (Vol. I)," Technical Report NPS54-85-08, Naval
Postgraduate School, Monterey, CA, September 1985.

nonrecurring cost items and develop systematic procedures for the
government and contractors to follow in order to generate a
database of nonrecurring costs which is both consistent and
comparable. Potential changes to RFP ' s and RFQ ' s will be analyzed
with recommendations made as to how such changes could result in
the provision of an analyzable set of data from each government
contractor and how these sets may be aggregated to provide weapons-
system peculiar databases for future analysis. The key element in
such a system will be the provision of consistent and comparable
databases .

RESEARCH METHODOLOGY

Due to the paucity of prior research on nonrecurring costs under
dual source environments, the study relies heavily on first-hand
data gathering. Five missile system contractors were visited by
the research team jointly or separately. One of the five defense
contractors is a prime contractor for a sole-source missile
program. The intention was to see the differences and similarities
between a sole sourced program and a dual sourced program. The
remaining four are either the original source or the second source
of dual-sourced missile programs. Individuals interviewed included
cost-estimating specialists, financial managers, program managers,
contract managers, and engineer ing/technical division managers.

The interview focused on the following three questions;

1. the criteria for classifying nonrecurring costs in
setting up production facility.

2. the classification of nonrecurring cost items as variable
or fixed costs.

3. the method of tracking and accounting start-up nonrecurr-
ing costs to segregate from follow-on production efforts.

SCOPE AND LIMITATIONS OF THE STUDY

As mentioned earlier, this study relied heavily on first-hand data
gathering from defense contractors' files due to the paucity of
prior research. Some contractors, however, were reluctant to
discuss even their procedures, because of their perceptions of the
"competition-sensitive" nature of this information. In addition to
this problem of cooperation, the conduct of this study was further
hampered by two built-in difficulties related to the maturity of
the programs studied. The second supplier was either established
several years ago or had only recently been announced. For the
former, some of the techniques used to estimate the startup costs
were often lost to corporate memory. For the latter, either the
methods to be used for cost estimation had not been decided on or
the information was considered to be competition sensitive. As a
result, information disclosed tends to be of a general nature.
However, the researchers were able to partially reconstruct the
picture from supplemental documents accompanying proposals submitted
to the Government procurement office. Therefore, the results of
the research provide the reader with a comprehensive but incomplete
overview of cost elements and costing methods used by the defense
aerospace industry in the treatment of nonrecurring investment
costs in a reprocurement environment.

CHAPTER 2

DATABASE AND PARAMETRIC COST ESTIMATION

Cost estimation is a projection into the future from a database
compiled from past experiences. Accuracy of the projection depends
on many things, not the least of which is the validity of the
database. The need for a comparable and consistent database has
long been recognized in the cost estimation community. This
chapter discusses various cost estimation techniques, their
database requirements, and related current Government regulations.

COST ESTIMATING METHODS

In subsequent chapters reference will be made to the different
methods of cost estimation used by defense aerospace contractors.
This section discusses the three basic methods most widely used.

Analogy Method

The method of analogy is based on direct comparisons with historical
information on like components of existing systems. 2 There is
intuitive and logical appeal to the popularity of this method. For
example, the use of new structural material for aircraft often
requires the development of special cutting and forming techniques
with manufacturing labor requirements that differ significantly
from those based on a sample of primarily aluminum airframes.

2 C. A. Batchelder, et al . , An Introduction to Equipment Cost
Estimating , Memorandum RM-6103-SA, The Rand Corporation, Santa
Monica, CA, December 1969, pp. 2, 6-8.

Faced with this problem when titanium was first considered for usel
in airframe manufacture, airframe companies developed standard-hourB
values for titanium fabrication on the basis of shop experience ini
fabricating test parts and sections. Although the experience ofB
fabricating test parts is not identical to the effort necessary toll
build the entire airframe, an estimate can be made by consideringll
the similarities and extrapolating . 3 I

Most often it works as follows. The item(s) to be produced is
first compared with a recent project to determine if there are any
common elements. If it is found that a significant portion of the
new system is similar to the former one, the hours of effort and
material expended to produced the previous item are used as a base
estimate. Differences in the design and performance of the two
systems are then considered. Estimates are made for these and
added to the base.

The strength of the analogy method is that it can provide
sufficient accuracy for the least cost in the shortest time. 4 xhe :
key requirement is to ascertain the similarity of items to be
compared. Therefore, the database requirement is mainly concerned
with establishing an objective basis to characterize an item for
analogy. This requirement is best exemplified by recent Air Force

3 Batchelder, pp. 7-8; Joseph W. Lemire, Jr., "Cost
Estimating Methods Utilized by the Defense Aerospace Industry in
the Production of Technical Data, Masters' Thesis, Naval
Postgraduate School, Monterey, CA, June 1985, p. 21.

^ Lemire, pp. 35-36.

attempts to establish a cost estimating model for tools . 5
Engineering Method

The industrial engineering approach involves breaking down the
system into separate segments of work. These segments are then
examined in detail and estimates are made for each segment. The
detailed estimates are then consolidated into a total estimate for
the overall system. This method is normally used when a thorough,
detailed analysis is required for all the processes involved.

The system must be relatively well-defined before this method
can be used. It is often applied using a work breakdown structure
(WBS) in which the system is organized in levels and each of the
levels is comprised of a number of elements. A cost is estimated
for each component element and totaled at each level. An overall
estimate consists of the total of all the levels.

The engineering method is more expensive than analogy,
requiring more time and personnel and is used most often when the
data cannot be directly fitted to an analogy. 6

Parametric Method

The parametric method is a statistical approach in which cost
estimating relationships (CERs) with parametric explanatory

® See James L. Storrs, "Taking the Art Out of Tool
Estimating; Business Research Report, Proceedings of 1985 Federal
Acquisition Research Symposium, Defense Systems Management College,
pp. 241-45.

® Lemire, p. 36.

variables, such as weight, speed, power, frequency, and thrust are
used to predict costs. For example, in the area of airframe
manufacturing, known CERs exist in terms of dollars per pound of
weight, per pound of thrust, and so forth. These relationships are
used with the variables of a new airframe to develop its estimated
cost. This method is applied at a higher level of aggregation than
the industrial engineering approach.

Worthwhile use of this method requires that sufficient data
exist and that historical costs are fairly consistent for statis-
tical analysis. In an area where there are constant changes in
technology, the CERs are invalidated faster than new ones can be
developed. Hence, the parametric method is difficult to apply in a
case like this.

CURRENT GOVERNMENT REGULATIONS

The government published Military Standard 881A in order to

standardize the process of defense material acquisition. It is a

summary of the upper three levels of a work breakdown structure. A

work breakdown structure is defined by Military Standard 881A as;

...a product-oriented family tree composed of hardware,
services and data which result from project engineering
efforts during the development and production of a defense
material item, and which completely defines the project/-
program. A WBS displays and defines the product (s) to be
developed or produced and relates the elements of work to be
accomplished to each other and to the end product. ^

Military Standard 881A is to be used by both contractors and

^ Department of Defense, "Work Breakdown Structure for Defense
Material Items," Military Standard 881A, 25 April 1985, p. 2.

the government during the acquisition process. The work breakdown
structure has become a foundation of DoD cost estimation data
requirements .

In 1973, the Secretary of Defense introduced the Contractor
Cost Data Reporting (CCDR) system which was intended "to maximize
effective utilization of data resources... to provide the the
primary database for use in most cost estimating efforts. "8 The
CCDR is, by law, the fundamental cost data reporting guide for all
contractors doing business with the government.

Although the Military Standard 881A and CCDR guidelines are
clear, there is no systematic procedure for applying them. An
examination of contractors' WBS and cost estimating relationships
based upon them shows that, at the lowest level of the WBS, both
analogy and parametric approaches can be found in costing the work
element .

AN OVERVIEW OF CURRENT PRACTICE

To estimate nonrecurring costs associated with establishing a
second source, both analogy and parametric CER methods have been
used. However, government cost analysts face several difficulties
under the current cost data reporting system. This section
discusses these difficulties.

Difficulty of Using The Analogy Method At the Aggregate Level

® Department of Defense Instruction 7000.11, "Contractor Cost
Data Reporting (CCDR)," September 5, 1973. This instruction was
reissued on March 27, 1984.

The analogy method as applied here bases the second source's
nonrecurring costs, especially tooling and test equipment costs, on
the original producer's cost. A study by the Institute for Defense
Analysis (IDA) cited the opinion of a cost analyst that the cost of
special tooling and test equipment is about 80^ of the amount
incurred by the original source. 9 But available data examined by
the IDA authors do not allow generalization of this estimate. Our
earlier work also found other attempts to use the analogy method for
nonrecurring cost estimation.

Results from our interviews with contractors shed some light
on the difficulty of using the analogy method at the aggregate
level. First, the sole source or the original source supplier of a
system is also the developer of the system. Some of the production
tooling costs are actually incurred during the development phase
(sometimes called system planning phase in contractor's WBS ) and
never specifically identified. Therefore, the data for nonrecurring
costs are not entirely consistent with what might be incurred by a
second source. Second, a study by Garrick shows that the original
source's production capacity is sometimes larger than necessary. 10
A second source, if determined to be desirable, is most likely to
be sized to some production rate smaller than the original source.

3 See G. G. Daly, H. P. Gates, and J. A. Schuttinga, The
Effect of Price Competition on Weapon System Acquisition Costs , the
Institute for Defense Analysis, P-1435, Arlington, VA, 1979.

P. M. Garrick, "Estimating the Savings from Competitive
Acquisitions: A Review of Previous Investigations," Chapter III of
a draft Institute for Defense Analysis report, "Competition as an
Acquisition Strategy: An Assessment of Selected Army Weapon System
Procurements . "

However, initial production start-up costs are not separated from
production rate related start-up costs under the current cost data
reporting system. These initial costs are driven by design
considerations, while the rate-related costs are driven by the
number of units. Therefore, use of the analogy method faces the
built-in difficulty of noncomparable data.

Difficulty of Using Parametric CER models at the Aggregate Level

The second method commonly used by government cost analysts calls
for using parametric CERs . However, parametric models are based on
statistical analysis of historical data. As mentioned earlier,
worthwhile use of this method requires that a sufficient data base
exists and that historical costs are fairly consistent for statist-
ical analysis. Due to the limited number of second-sourced major
programs and the absence of systematic accumulation of nonrecurring
cost data it should be obvious that any attempt to construct a
parametric CER for nonrecurring costs is premature at this stage in
time .

REQUIREMENTS FOR A PARAMETRIC MODEL

Since nonrecurring costs consist of several catagories of cost
items, a parametric cost estimating model with a small number of
available observations and a large number of potential explanatory
variables would be unreliable, even if possible. A feasible
solution is to disaggregate the nonrecurring costs into relatively
homogeneous groups of cost items for data accumulation purposes.

With a consistent data base and relatively homogeneous cost items,
a parametric model for each group may be constructed with a
relatively small number of observations, a typical constraint in
major weapon system cost estimation.

Our research effort focused on identifying the components of
nonrecurring second source start-up costs, their general patterns
of variation, and cost drivers. It is important to identify the
cost drivers of most nonrecurring cost items, as the development of
a parametric cost estimating model requires a comparable database
of costs as well as cost drivers. The results of our field study
are discussed in the next chapter.

CHAPTER 3

ANALYSIS OF COST BEHAVIOR OF NONRECURRING COSTS

To become a competitive second supplier the new source incurs
nonrecurring costs in six areas: the preparation of bid and
proposal, research and development, data management, production
start-up, test and evaluation, and project management. This
chapter will present research findings regarding functions typically
included in each cost element and the cost estimating methods,
which includes cost behavior and cost drivers, used by the contrac-
tors studied. A summary of findings and our recommendations can be
found in Chapter 4 .

BID AND PROPOSAL

This group of cost items include preparation of bid packages,
contract bidding, and document preparation for meeting reporting
requirements. Since these costs are considered to be part of
future business expenses, they are included in overhead expenses
and distributed across all government contracts. We found that for
all contractors, the amount budgeted for a project proposal is
considered a fixed cost and is determined by the analogy method.

RESEARCH AND DEVELOPMENT

Cost Components

Included in this group of cost items are engineering efforts
ranging from initial design analysis and interpretation/translat ion

through design integration. It is important to note that the level
three drawings that become government property in the technical'
data package are not the same as the level-three drawings maintained
in-house by the first source. The original contractor is not
required to reveal the processes used to translate the drawings
into hardware. The level-three drawings available to the Government
and the second source shows dimensions, surface finishes, and
materials, but no processes.

A second source may be provided a technical data package
containing level-three drawings and a system engineer to help
interpret the drawings but may still have to redo everything the
original source has done, except for the original development of
the system. In the reinterpretation of the drawings to redevelop
the drawing base in accordance with the way that second contractor
does business, it can be as though it were redesign.

It should be pointed out that the research and development
cost discussed in this section refers to initial design analysis,
interpretat ion/translat ion , or reverse engineering. It is not the
independent research and development (IR&D) covered by Cost
Accounting Standard 420. The cost discussed here is typically
accounted for by the second source as a part of "Design, Testing &
Evaluation Engineering Labor".

Cost Estimation Method

The R & D tasks involved in becoming a competitive second source
depend largely on the acquisition strategy employed by the Govern-

merit. At one extreme, the second source may be required to produce
a model identical in every aspect to the sample provided by the
original source, or the so-called "Chinese copy". To do this, the
second source and its subcontractors must do reverse engineering.
In this case the R & D costs are roughly the same as the R & D
costs incurred by the first producer.

At the other end of the spectrum, the contractor can be
relatively free of constraints as long as the final product is
compatible in form-fit-function. This strategy allows the second
source to explore some cost savings alternatives, but the savings
may be offset by the additional R & D costs incurred.

The amount of material cost varies with the competition
strategy and system complexity. Direct labor and overhead costs
are driven by the amounts of hours for various engineering special-
ties. For cost estimation purposes, it is logical to estimate the
number of engineering hours needed and apply the applicable labor
and overhead rates to derive the total direct labor and overhead
costs. Contractors' cost estimating personnel indicated that
engineering hours are variable and are affected by both the
complexity of the system and whether the contractor has prior
experience with a similar system.

TECHNICAL DATA

Nonrecurring cost items related to technical data include technical
manuals and drawings, engineering data, and data management. It
should be noted that, although some separate contracts may exist

which purchase technical data, these costs are incremental,
nonrecurring costs from the standpoint of second sourcing decision.
Some technical data packages include clauses which prohibit the
second source from altering technical manuals and drawings. In
such cases, engineering changes must be proposed to and made by the
first source. This process involves Engineering Change Proposals
( ECP ) and Engineering Change Orders (ECO). The typical cost
components are discussed below.

Technical Manuals and Drawings

The nonrecurring cost items in this category include the preparation
of technical manuals and maintenance support for preliminary
design. Design personnel are supported by draftpersons in the
preparation of drawings, parts lists, system correlation drawings,
and design instructions.

These costs are considered variable; the causal factor is the
complexity of the task, which determines the level of effort
necessary to deliver a finished product. For cost estimation
purposes, it is logical to estimate the number of drafting hours
needed and apply the applicable labor and overhead rates to derive
the total direct labor and overhead costs. Contractors' cost
estimating methods vary, with one using the total number of
drawings and another using the number of drawings to be modified.
These are then multiplied by an average hours per drawing to obtain
total hours.

If there is an ECO involved, cost estimation relies on a

complexity score and a schedule impact index. As an example,
complexity may be measured by the range or weight of a missle. The
value for the complexity score is influenced by the degree of
uncertainty concerning design requirements, the level of risk due
to technical advancement, whether the program is joint or single
service, and whether it will be a single or multi-purpose system.

Engineering Data

Cost items in this category include the preparation of a variety of
coordinating and correlating documents (e.g., drawings, plans,
procedures, specifications, computer programs, and flow diagrams),
corrective design, and documentation update.

As in technical manuals and drawings, these costs are considered
variable; the causal factor is the complexity of the task, which
determines the number of labor hours necessary to deliver a finished
product. It is interesting to note that, although different
classes of technical personnel are used and separately identified,
no such distinction was made in computing the cost data in all
supporting documents we examined accompanying contract proposals.
Only the total number of direct labor hours are used in cost
computation. This rudimentary method seems to be an accepted
practice. Therefore, it seems to be logical to group cost items
related to technical manuals and engineering data into a single
element. This grouping is advantageous for parametric model
building and estimation purposes. However, for budgetary purposes,
these costs may need to be separately reported.

Data Management

Tasks such as maintaining master documents at the latest approved
revision level, assuring that all data items are prepared and
submitted in accordance with contract requirements, and monitoring
subcontract data are included in this cost element. For one group
of contractors this cost element was treated as a variable cost of
total engineering labor hours. Under this circumstance, this cost
element can be grouped with the preceding two cost elements. For
another group of contractors, data management cost was estimated on
a per Contractor Data Requirement List (CDRL) item basis. It
should be noted, however, that the number of CDRL items can be an
alternative measure of the complexity of a system, since more
tracking of business/financial items may be required for a more
complex system. Therefore, it is reasonable to conclude that,

although different measures have been used as cost drivers for
technical data handling, these different measures all reflect the
degree of complexity of the system.

PRODUCTION START-UP COSTS

Nonrecurring costs incurred by the second source in production
start-up generally fall into one of three categories: tooling,

production engineering, and buildings or facilities. Test equipment
will be treated in the following section.

Tooling

Tooling equipment needed to produce the contract item is included

in this cost element. It should be noted that some contractors
distinguish initial tooling from rate tooling, but some do not. As
we noted above, initial tooling costs tend to be driven by design
and complexity considerations while rate tooling costs tend to be
driven by the number of production units. However, separation into
these two categories is not necessary in concept if both cost
drivers are available for model building. All contractors stressed
that production phase special tooling equipment is not directly
charged to the government at the outset. However, the cost to the
government remains the same, as this cost item is charged to the
government through amortization. Alternative policies merely
change the form of recoupment from the government for these
expenditures .

Labor and labor related costs account for mast of the cost
items in this cost element. Material costs include purchased tools
or tool manufacturing material if the tool is made in house. This
presents a problem for forecasting purpose, as making the tool in-
house increases labor costs but decreases the cost of tool purchas-
es. Tool manufacturing material was typically estimated according
to tool manufacturing and tool proofing labor hours based on prior
history .

The major factors affecting labor estimates are the complexity
of the system and the production rate. It is apparent that the
former is more related to the initial tooling and the latter is
related to rate tooling. In most cases, the distinction between
initial tooling and rate tooling is not made. Where the distinction

is made, a rather crude method is used by the contractor to project
nonrecurring tooling costs. The nonrecurring ramp (to increase
monthly rate) is treated as variable dependent on quantity and
rate. A ratio of dollar increase to rate increase is calculated by
using prior experience, and this delta factor is applied to prior
costs ,

Production Engineering

The original industrial engineering efforts to translate design
into an efficient manufacturing technique (for example, redesigning
plant layout and operations planning) are included in this cost
element .

This cost element is estimated by determining the engineering
hours on the basis of the level of effort required. Therefore, the
real driver is the complexity of the system involved. In this
case, it is clear that the complexity of the weapon system is
serving as a proxy for the complexity of production of the system.
In cases where this is not true, production complexity will have to
be treated directly.

Buildings or Facilities

Unless the program is going to be a long term effort, no new
facilities will usually be built. Contractors track existing
programs to see when they will terminate and what facilities will
become available for the new program. Therefore, the cost items
involved are primarily related to conversion of existing facilities

and construction of special facilities.

The cost is estimated by the number of square feet of floor
space adjusted for the level of effort required to modify it. The
the level of effort requires a surrogate measure, such as the
degree of similarity between the existing program and the proposed
new program. The number of square feet of floor space required is
variable, depending on the production rate desired.

TEST AND EVALUATION

Nonrecurring costs incurred by a second source in test and evaluation
include first article test quantities, test equipment, and related
engineering costs. Cost items included in this element exhibit two
kinds of cost behavior: variable cost and fixed/variable cost.

Variable costs include assembly and test labor and materials.
These cost items clearly are related to the quantity of test items.
For comparison across programs, however, one must add another
explanatory variable, complexity of the system.

Fixed/var iable costs, which are both time dependent and
quantity dependent, are also affected by the complexity of the test
and test equipment. Cost items included in this category consist
of hardware as well as engineering support costs.

In the case of quality assurance, engineering inspection is
estimated as a percentage of the number of hours required to build
test units. Receiving inspections are factored against total
engineering material.

PROGRAM MANAGEMENT

Included in this cost element are program planning and technical
direction, system engineering, administrative support, procurement
planning, liaison with the government, preliminary design reviews,
systems requirements review, management support, subcontract
selection and coordination, scheduling, and cost analysis.

Cost estimation depends upon the structure of the program
management effort. In the pure or functional program management
structure all resources, personnel and material, are dedicated to a
specific program. Contractors view this cost element as mostly
fixed and use past experience with similar programs.

In a matrix program management structure, personnel in the
organization may work on a variety of projects and material
resources are drawn from a shared pool. In this case, labor
estimates are variable and estimated on a labor hours basis
according to prior experience with similar programs. In essence,
the degree of complexity determines the labor hours.

FIRST SOURCE COSTS

In developing a competitive environment, technical transfer to and
coordination with the second source are the two areas in which
nonrecurring costs accrue to the original source of supply.

Technical Transfer

There are two types of nonrecurring costs in technical transfer.
These are the costs of the engineering task of preparing a technical

data/reprocurement package and the costs of production of hardware,
for example, missiles. Missile hardware in various stages of
completion and special test equipment are transferred to the second
source to assist in the learning process. Note that the hardware
involved is the training version which differs from the tactical
unit in that the tactical unit warhead, rocket motor, and hydraulic
actuation system are replaced with a ballast, a signal processor,
and a tape recorder unit with cartridge loading capability. In
some cases, however, for cost estimation purposes, no distinction
is made between the two versions since the cost of the deleted
items is not significantly different from the cost of the added
items .

The data/reprocurement package cost estimation is done by
experienced engineering staff based on manhours, much as the
preparation of technical manual and drawings discussed earlier in
this chapter. The missile hardware costs are variable. The total
cost of labor and material are driven by the quantity supplied as
well as the complexity of the system.

Coordination

Coordination with the second source is primarily the task of
training and supporting the second source's engineers. The costs
involved are primarily travel expenses and labor costs.

Cost estimation is based on the time required which is related
to the quality/experience of the engineers as well as the adequacy
of the data package.

CHAPTER 4

SUMMARY OF FINDINGS AND RECOMMENDATIONS

This study was designed to examine the nature of nonrecurring cost
items and develop systematic procedures for the government and
contractors to follow in order to generate a database of nonrecurr-
ing costs which is both consistent and comparable. The objective
was to develop a method for generating an analyzable set of data
for parametric estimation of nonrecurring costs.

This study relies heavily on first-hand data gathering. Five
missile system contractors were visited. Individuals interviewed
included cost-estimating specialists, financial managers, program
managers, contract managers, and engineering/technical division
managers. Cost estimation documents supporting contractors'
proposals were examined to determine the cost estimation practice
and identify relevant cost drivers.

SUMMARY OF FINDINGS

In Chapter 2, we have shown that, given current practices of cost
reporting, it is impractical to develop a viable model for
estimating nonrecurring costs when second sourcing is contemplated.
The analogy method at the aggregate level has been shown to be
unreliable. The development of a parametric CER model, however,
requires a sufficient data base which is not currently available.
To develop such a data base, knowledge of major nonrecurring cost
categories and their cost drivers is necessary. Exhibit 1 summarizes

the results of our endeavour for this purpose. It contains the
cost elements discussed in the previous chapter, measures of the
cost elements, whether that particular element is fixed or variable,
and the cost drivers for that element. Relative sizes for the
major groupings of cost elements are discussed in our previous
report . 1 1

A few notable observations should be mentioned. First, it is
necessary to separate nonrecurring material costs from labor and
overhead costs for estimating purpose. The main reason is that the
cost drivers for each are not always the same. Lumping material
and labor costs together requires adding more explanatory variables
to a CER model, which would quickly exhaust the degrees of freedom
because of the limited database for major weapon systems.

Second, parametric CER models for nonrecurring labor costs must
be based on labor hours, not on labor dollars as some cost analysts
do now. There are several reasons for this. For one, all contrac-
tors interviewed indicate that they estimate anticipated manpower
levels (in labor hours) required to support the contract before
they cost it. For another, although different types of engineering
elements are distinguished in the Work-Breakdown-Structure, all
engineering hours for a given work element are grouped together for
costing purpose. Furthermore, different contractors have different
labor and overhead rates, which would cloud the development of CER
models if labor dollars are used as the dependent variable.

Third, the most important explanatory variable that needs to

H Boger and Liao, September 1985,

be operationalized is the complexity of the weapon system. Most of
the manpower needed is variable and is driven, according to
contractors, by both labor hours and complexity. When prodded to
define how this is predicted, contractors' cost estimating personnel
typically indicated that they use the analogy method if the firm
has prior experience or a commercially available model which uses
several relatively subjective inputs (such as complexity) if a task
is new. One can infer that "prior experience" was most likely the
result of using such a model when the firm performed a specific
task for the first time. Note that the use of such a model
requires calibration of the model according to the complexity level
of the item. Therefore, defining the complexity level of a system
in objectively measurable terms is of paramount importance if a
reliable parametric CER model for nonrecurring cost is to be
developed. In fact, most contractors have used certain surrogate ‘
measures for complexity level, such as the number of Contract Data
Requirement List items, the number of drawings, etc.

COMPLEXITY AND ITS MEASUREMENT

Throughout Chapter 3 the complexity of a weapon system is repeatedly
mentioned as the main cost driver. Picking those particular
variables which can precisely measure the complexity of a missile
system is a difficult task, since little, if any, work has been
accomplished in this area. Therefore, we propose that several
variables be examined for their adequacy in generating good cost
estimating relationships. Subsequent to this exploratory examina-

tion, requirements may be more closely focused on those variables
which yield the best results.

The candidate variables which we propose are analogous to
surrogate measures of aircraft complexity, i.e., range of the
missile, speed of the missile, weight of the missile, and weight of
the guidance and control section of the missile. That each of
these variables might serve as a proxy for complexity under certain
conditions is quite clear. Which one or two of these is best under
all, or almost all, conditions is an empirical question which
cannot be settled until data are available.

RECOMMENDATIONS

Apart from responses to our interview questions, much of our
information about the behavior of nonrecurring cost items comes
from supporting documents accompanying contractors' proposals.
Therefore, implementing a modified cost reporting system to
facilitate the development of a parametric CER model for nonrecurring
cost would not be a major task.

Two major pieces of information are missing from current
contract proposal supporting documents that are needed to develop a
consistent database for statistical analysis. The first is the
breakdown of costs and labor hours according to the major cost
elements outlined in Chapter 3. The second is data on explanatory
variables, the cost drivers.

The cost breakdown available in the supporting documents
accompanying contractors' proposals typically follows the Work

Breakdown Structure which does not distinguish between recurring and
nonrecurring costs, because the Work Breakdown Structure is based
upon work elements and not costs. If the breakdown of costs and
labor hours according to the major cost elements outlined in
Chapter 3 is to be accomplished, it requires only regrouping of
costs and labor hours already available to the contractor.

Our recommendation is that material and labor costs should be
separately reported. The main reason is that separate CER models
must be developed for each, one with material cost in dollars as
the dependent variable and another with labor hours as the dependent
variable. The labor hours can then be used as the basis for
estimating labor and overhead cost according to each contractor's
labor and overhead rates structure. The secondary reasons for
separate CER models are that cost drivers for each are not always
the same and that separate models would require fewer explanatory
variables, a major consideration when the database is limited.

Defining the complexity level of a system holds the key to
developing viable parametric CER models for nonrecurring costs.
The RCA/PRICE model and the Air Force's tool cost estimating model
are the prime examples of using complexity level as the explanatory
variable. The cost drivers for tooling and test equipment costs
developed by Naval Weapon Center in China Lake include per unit
hardware cost as well as production rate. 12 xhe conventional method
for estimating hardware costs typically takes into consideration

1 2 Advanced Intercept Air-to-Air Missile (AIAAM) Life Cycle
Cost Estimates (U), Naval Weapons Center, Code 081, NWC TM 4899,
September 1982. (C)

the capability of the system via inclusion of physical or performance
characteristics of the system. For example, aircraft and missiles
typically use some measures of speed, range, etc. It may be
logical to examine the feasibility of using capability as a measure
of the complexity of a system. The production rate variable was
confirmed in our study as a major cost driver. However, we believe
that the China Lake CER model would have been more reliable had
they estimated materials cost and labor hours separately.

Exhibit I

flajor Nonrecurrinq Cost Eleients and Cost Drivers

Cost Eleients

Measure

Cost Behavior

Cost Drivers

Bid 4 Proposal:

(Included in contractor’s overhead expenses)

1 B: R 4 D:

Material

Variable

Coaplexity, lethods of second source developient

Labor and overhead

hrs.

Variable

Coiplexity, prior experience

C: Technical Data:

Technical lanuals/draiinqs

hrs.

Variable

Coiplexity, 1 of drawings, 1 of drawings to be aodified

Engineering data

hrs.

Variable

Coiplexity

Data Manageient

hrs.

Variable

Coiplexity, 1 of CDRL iteis

D: Production Start-Up:

Tooling (initial)

hrs.

Variable

Coiplexity

Tooling (rate)

hrs.

Variable

Production rate, predeteriined raip ratio

Production engineering

hrs.

Variable

Coiplexity

Facility conversion

s.f.

Variable

Level of efforts depends on siiilarity to previous prograi

Special construction

s.f.

Variable

Production rate

E: Test and Evaluation:

Hardware- Labor

hrs.

Variable

Quantity of test iteis, coiplexity

Material

Variable

Quantity of test iteas, coiplexity

Quality assurance

hrs.

Variable

Fabrication hours, engineering laterial

F: Prograi Manageient:

(functional structure)

hrs.

Fixed

Past experience

(latrix structure)

hrs.

Variable

Coiplexity

G: First Source Cost:

Technical Transfer- Data

hrs.

Variable

Coiplexity

Harewa

e $

Variable

Complexity, quantity supplied

Coordination

L . .

Variable

Experience of engineer k adequacy of data

DISTRIBUTION LIST

Agency No

Defense Technical Information Center
Cameron Station
Alexandria, VA 22314

Mr. James Weathersby
Code 524

Naval Air System Command
Washington, D. C. 20361

Mr. Tom Gilbride
Code 524

Naval Air System Command
Washington, D. C. 20361

Dr . Dan Nussbaum

Naval Center for Cost Analysis

Washington, D. C. 20350-1100

COL Robert Skipp

Air Force Business Research Management Center
Building 125, Rm. 2063
Wright-Patterson AFB , OH 45433-5000

Director

Air Force Cost Center (AFCCE)

c/o HQ USAF/ACC

Washington, D. C. 20330-5018

Dr. Robert F. Williams
Director

Army Procurement Research Office
Ft. Lee, VA 23801-6045

Dudley Knox Library, Code 0142
Naval Postgraduate School
Monterey, CA 93943

Office of Research Administration
Code 012

Naval Postgraduate School
Monterey, CA 93943

Professor Dan C. Boger

Department of Administrative Sciences
Naval Postgraduate School
Monterey, CA 93943

Professor Shu S. Liao

Department of Administrative Sciences
Naval Postgraduate School
Monterey, CA 93943

of copies