T ECHNICAL A RTICLE

Management of Project
Contingency and Allowance
Jan Terje Karlsen and Jon Lereim
ABSTRACT: Cost overruns in engineering projects occur frequently because a certain margin of
risk is inherent in all projects. As a result, risk management is continuously gaining the attention of the engineering industries. Reserves or contingencies represent the additional funding
required to account for the cost of risk. However, many corporations have different practices for
estimating and managing such reserves. In addition to discussing and clarifying the risk reserves
terminology, this article presents several techniques and methods for estimating such reserves.
Authority to use the respective parts of the risk reserves is also discussed.
KEY WORDS: Contingency, cost overruns, estimating, and risk management

n engineering and construction

projects, plans and cost estimates are
usually drawn to ensure that the work
is carried out to the desired quality, in the
allowed time, and within budget.
However, we often hear about projects with
delays and where the costs tend to be
higher than depicted by budgets [2].
With all the experience and research
in cost estimation, what are the reasons for
these results? One answer is that the
management faces an information /
influence dilemma in the early phase of
the project. Figure 1 shows that the
possibility to influence and change the
project is highest in the early stage of the
project because conceptual decisions have
not yet been made or taken effect.
However, at the same time access to
information is limited. It is hard to obtain
knowledge about what the final cost and
schedule will be because relatively little
detailed and precise information is
available about the major cost and
schedule drivers [1].
The lack of information in the early
stage of the project creates risk. Because of
the level of technology and complexity
involved in many industrial projects, the
risk might be high. However, risk has been
recognized as an inevitable part of
industrial projects. The traditional view of
risk is negative, representing loss, hazard
and adverse consequences. Risks are a
major cause of problems, disputes,
disruptions, delays and cost overruns [1].
Management should accept a certain
level of risk because of unforeseen cost,
which they incur during the project period.
The risk manifests itself in unforeseen

expenditure, which was not envisioned at

the planning and cost estimation stage. To
cover the unforeseen conditions, a risk
reserve or buffer should be estimated. This
risk reserve can be divided into three parts;
contingency, allowance, and budget
reserve.
In table 1, a literature survey shows
that there are different uses and practices
regarding this terminology. The list of
authors is not complete. Some authors
differentiate between contingency and

Figure 1 The Information/Influence Gap

Table 1 Literature Review

24

Cost Engineering Vol. 47/No. 9 SEPTEMBER 2005

allowance, while others only use the term

contingency. The literature survey also
reveals that few authors discuss the
estimation of budget reserve.
Results from the literature survey also
indicate that procedures or rules that clarify
the roles and states who has access to the
respective parts of the risk reserves have not
been a subject for theoretical discussion.
There are different perceptions and
practices regarding estimation of costs and
the use of risk reserves. If different
practices are present within the same
corporation, it is unlikely that there will be
a consistent estimation and project
budgeting practice. The ultimate
consequences of such an approach are that
the major decision-making processes for
selection of prospects and initiatives with
the highest potential may suffer.
Experience also indicates that in those
projects where risk reserves are estimated,
they are often based on the manager's
experience and intuition.
It is therefore of utmost importance to
establish a standard and practice for
deriving a consistent set of cost estimations,
including uncertainty and the use of risk
reserves figures as the basis for the project
budget and master control estimate.

The purpose of this article is to discuss

and clarify the risk reserves terminology. In
addition, several techniques and methods
for estimating such reserves will be
presented. This article also explains and
clarifies the roles, and states who has
authority to use the reserves during the
project period.
This research is important because
there are few studies about estimation and
management of risk reserves. Knowledge
about risk reserve and cost estimation can
provide a distinct advantage in a rapidly
changing business environment. The
purpose of risk reserves estimation is to
ensure that the budget set aside for project
execution is realistic and sufficient to
contain the risk of unforeseen cost
increases.
The rest of the article is structured as
follows: risk analysis is discussed with a
focus on identification and assessment of
risk items. Various risk reserves estimation
techniques are presented. Build-up of cost
and budget figures are discussed, as well as
who should control the risk reserves. This is
illustrated with a mini-case. Conclusions
will be drawn as well as the highlighting of
recommendations and implications.
Cost Consequences of Risk Elements
In most situations, cost estimates are
point estimates. These can also be defined
as single value estimates based on the most
likely values of the cost elements.
According to David H. Picken and Stephen
Mak, these single estimates may or may not
accurately indicate the possible value of
cost estimate, and they do not indicate the
possible range of values an estimate may
assume [4]. The danger of this approach is
that the technique does not include risk in
the estimation of costs.
An important part of cost estimation is
the process of risk analysis. For many
engineering and construction companies,
risk analysis represents something new.
Although project risk management has
been practiced for almost 20 years, most of
the methods, techniques and its
application are more recent phenomena.
All projects are exposed to riskfrom
the first initiation to delivery and
implementation. The sources of risk can
be divided into internal and external
factors. Internal factors may include
project complexity, internal management
policy,
routines
and
procedures,
competence and management skills, in-

Figure 2 Inaccuracy of Estimates

house culture and a degree of bureaucracy.

External sources may include inflation,
currency exchange fluctuation, weather
conditions, government policy, technology
development, culture and language.
As a consequence of all these risk
sources, considerable attention should
been given to ensuring comprehensive
identification and objective assessment of
project risks to provide a clear
understanding of the extent of risk
exposure faced by a project. Many
techniques and tools that work well when
used properly have been developed to
support these stages in the risk process.
The starting point of the risk analysis is
the risk identification, whose purpose it is
to identify all the risk items that could
affect the project in terms of performance,
cost, or schedule.
This step is of
paramount importance in the frame of risk
management. It does not matter how good
the risk assessment is, if you have not
identified the key risks that might affect the
project.
The second stage of the risk analysis is
to assess and quantify the effect of each risk
item. The purpose of this step is to
determine the magnitude of the individual
risks and to rank them with respect to cost.
To achieve this aim it is necessary to
determine the probability of occurrence
and the consequence severity of the events
identified in the previous step as risk items.
For each risk item, one should assess the
variety in the consequence/impact.
Risk analysis goes hand in hand with
cost estimation.
By performing risk
analysis, project management will get an
early warning of potential cost overruns.

Cost Engineering Vol. 47/No. 9 SEPTEMBER 2005

The impact of the risks can be assessed and

then evaluated to determine risk reserves.
Risk Reserves Estimation
Many techniques and approaches have
been proposed in literature for the
estimation of risk reserves. According to
Osama Moselhi, most of the methods are
traditional algorithmic methods [3]. Some
of the methods are deterministic and others
are stochastic (probabilistic) in nature.
The most traditional method of
allowing for risk is to add a percentage
figure to the most likely estimate of final
cost of the known works. The amount
added is usually called a contingency. It
ranges normally between five percent and
10 percent. The purpose of this approach
is to ensure that the project budget is
sufficient to contain any cost incurred by
risks.
However, this is a subjective method
based on gut feeling and intuition, and
therefore has several weaknesses [4,5].
One problem is that the method is in
danger of being overly simplistic and
heavily dependent on an estimator's faith
in his/her own experience.
Another problem may be that the
percentage figure is, most likely, arbitrarily
arrived at and not appropriate for the
specific project.
A third weakness is that there may be a
tendency to double count risk, because
some estimators are inclined to include
contingencies in their estimate.
An additional weakness is that a
percentage addition still results in a single
figure prediction of estimated costs,
implying a degree of certainty that is
simply not justified.

25

The result of this problem is illustrated

in figure 2. Depending on the variance of
the cost estimates, a percentage addition
can create very different results. As we can
see from figure 2, the lower the variance of
the cost estimates, the higher the
probability of no cost overruns. The
method may also turn into a convenient
routine that requires very little
investigation and does not encourage
creativity in cost estimating practice.
As a result of these weaknesses, this
method should be used with caution in the
project.
Another technique that is similar to
the method discussed above is where a
contingency percentage is added to each
cost item in the work breakdown structure
(WBS). Even though this method is more
detailed than the first method, it still has
many of the same weaknesses. For
example, the results heavily depend on the
estimator's knowledge and experience.
A different approach to incorporating
risks in the cost estimates is to apply a range
estimation method. This method follows
the PERT estimating procedure based on a
set of three-point estimates [6]. These
estimates are a lowest value (L), the most
likely value (M) and the highest value (H)
of the cost elements in WBS.
These three estimates for each cost
item can be based on knowledge and
experience or on data collected from
previous projects. The lowest and highest
values should be estimated in such a way
that there is only a 1 in 20 (five percent)
chance that cost values are exceeded,
either downward or upward.
Based on the set of three-point
estimates, one can define a subjective
probabilistic distribution. Since it is
generally observed that the estimator's
assessment of range estimates tend to be
conservatively biased for the upper-bound
value assigned, the probability distribution
can be assumed by a beta-distribution
skewed to the right [6]. In figure 3, such an
asymmetric probability distribution curve is
illustrated.
A weakness with this PERT estimating
procedure is that it assumes that all cost
items are independent random variables.
In practice this is wrong, cost items are
often correlated with one another. The
problem with this assumption is that it
tends to underestimate the variance
associated with the total project cost [3].

26

Figure 3 Beta Distribution of a Cost Item

Figure 4 Monte Carlo Simulation Model

As an alternative to the PERT

procedure, a straightforward method of
obtaining
information
about
the
distribution of total project cost is through
the use of Monte Carlo simulation. This is
an approach where a simulation model is
created as illustrated in figure 4.
The model is basically a cost
breakdown structure (CBS) where each
cost item in the structure is a single point
estimate. To include the risk in the model,
all identified and quantified risk items are
placed in the model where the risk will
have influence on the costs.
Some risks may only have an effect on
a work package while others may influence
the total project. Each risk item is range
defined by the item's lowest, most likely
and highest effect.
A triangular
distribution is normally used. The risk
effect on a cost item can either be
measured in percentage added or in cost
added.
There are several computer programs
that can be used for simulation purposes.
One example is the program called

Cost Engineering Vol. 47/No. 9 SEPTEMBER 2005

definitive scenario.
For a realistic
probability distribution of the total project
cost, 1000 to 10 000 cost sets or simulation
might be used. Any correlation between
the cost items or the risk items should be
specified in the program.
If such
correlations are ignored, the simulation
results tend to underestimate the variance
of the cost.
The result of the simulation is a
normal distribution curve of total cost as
shown in figure 5. From this curve the
cumulative probability distribution of total
project cost can be derived. This curve
gives information about the probability of
exceeding or not exceeding a specific cost
sum.
Build-up of Cost and Budget Figures
Most estimators tend to be
conservatively biased in their range
estimates [6]. As a result, the base estimate
tends to be less than the even chance
estimate (P50/50). The base estimate or
most likely value is an estimate where
theoretically no risk items are included.

But such a project is non-existing, and

therefore risk reserves have to be estimated
and included.
The first type of risk reserve that can be
added to the base estimate is called
contingency. By adding contingency to the
most likely estimate, one achieves the
expected value, i.e., the value that has the
same probability (50 percent) of overrun as
underrun.
When the management wants to be
even surer that there will not be any cost
overrun, for example only a 15 percent
chance of overrun, a risk reserve allowance
should be added.
An allowance is generally described as
a reserve for covering risks that can result in
unforeseen expenditures. Usually there is
no fixed rule on how to determine the right
amount of allowance. However, in Norway
all government projects exceeding 70
million US dollars must add an allowance
so that there is only a 15 percent chance of
cost overrun. When the allowance has
been added to the most likely estimate, one
gets a control estimate.
For simplicity reasons, many
corporations use contingency as a generic
term for both contingency and allowance
reserves.
The risk reserves - contingency and
allowance - estimated as shown, only cover
the risk items included in the simulation
model. These risk reserves do not cover
changes in the project's assumptions or
other major unforeseen impacts.
For example, contractor bankruptcy or
major client induced scope changes are to
be considered as extraordinary risks and
should be covered by a budget risk reserve.
The size of the budget reserve is to be
decided by the project owner and depends
on the desire to avoid an overrun situation.
When adding the budget reserve to the
control estimate, one gets a total budget.
The build-up of cost and budget
figures may be performed as shown in
figure 6.
Why should contingency preferably be
the figure lifting the estimate from most
likely to the expected value? In a project
there is normally a large number of
activities.
Successive
estimation
techniques derive figures for the single
activities
and
the
corresponding
accumulated figures for the entire project.
If the single activities are independent of
each other, the expected value of the total

costs is equal to the sum of expected values

for the single activities.
Therefore, it is appropriate to use the
expected value level as the specified level
for contingency. The analogy is a portfolio
in which the total value converges toward
the expected value, although the individual
elements may vary positively and
negatively. However, in many projects
there are dependencies between single
activities and group of activities. Thus a
deviation from the estimate on one activity

may have a knock-on effect on the total

estimate. One may then question whether
the expected value of the total is the
appropriate level for the contingency.
Practice Concerning Bid Estimation
The construction industry phases
fierce competition and the industry is
characterized by having very small profit
margins on their delivery projects.
The estimation practice is thereby
pressing the bid estimate to such low

Figure 5 Cumulative Distribution of Total Cost

Figure 6 Build-Up of Cost Figures

Cost Engineering Vol. 47/No. 9 SEPTEMBER 2005

27

figures that essentially no margins are

remaining for the contractor. The reality is
that the so-called base estimate, which
represents the most likely estimate (the
most frequent figure), at best will be used
as official bid/offer to the client. Often the
actual bid is even below the base estimate.
In such situations there is a larger
probability of exceeding the bid than
running below the bid.
Such practice most likely leads to an
overly optimistic estimate and means that
the tiny margins might be eaten up, and
one ends up losing money on the contract.
The contractors actually are running a high
economic risk because of this estimation
practice.
Situations may arise in which
management is firm and confident about a
bid below the base estimate. It can be
argued that it is wise from a strategic point
of view with respect to keeping market
share and/or competence development
within the company. The lack of profit
must then be considered as a strategic
investment where management is
responsible. However not all the projects
can follow such a practice.
Practice Concerning Level of Contingency
Frequently, there is a significant gap
between the actual as-built costs and the
cost estimations performed prior to project
sanctions, start-up and execution. There
are numerous reasons for that, including
scope creep, change in functionality,
change in boundary conditions, etc. In
addition, it may be partly due to nonsatisfactory estimation of cost uncertainty
and a non-consistent practice of derivation

of contingency. The latter is further

influenced by the estimation of the total
uncertainty range.
The plain definition of contingency
shall be the figure between most likely and
the predefined probability level, normally
up to the 50/50 estimate or expected value.
If no significant variation between extreme
figures is accounted for, the expected value
or 50/50 estimate is not far from the most
likely value. The contingency addition
may therefore be under estimated
compared to the real uncertainty picture at
the time of first estimation.
If the cost is derived from the base
estimate, plus the underestimated
contingency, one will expect a gap between
the as-built cost and the first estimate.
Because of such experiences, many
define "contingency" in the 70/30
percentile and even higher, in order to
compensate for their experience on
previous projects. This is not consistent
with the basic principles and methodology
for estimation.
Authority to Use the Risk Reserves
It is of primary importance to clarify
the roles and state who has authority over
the respective parts of the budget figures.
The project manager should have authority
over the base cost, plus contingency and
allowance. This is the accumulated figure,
called the control estimate, based on the
projects original scope.
The base estimate should represent
the most likely plain estimate for the
activity and project. The project manager
should be in charge of the estimate,
including contingency and allowance in

Table 2 For Projects Well-Known and Mature

Table 3 For Projects With Significant Uncertainty

28

Cost Engineering Vol. 47/No. 9 SEPTEMBER 2005

order to perform effective project

management.
Concerning budget reserve, this
additional risk reserve portion requires a
close dialogue between project manager
and project owner. It is the project owner
or the steering group that has the formal
authority over the budget risk reserve,
frequently also called management
reserves. This is illustrated in figure 6.
An Example From the Industry
Presented and analyzed in this section
is a shipbuilding case, to illustrate the
previous discussion about authority to use
the risk reserves.
Multiple sources were used in the data
collection phase to enhance both
construction and content validity. The
primary source was interviews with the
project manager and project owner. For
this purpose, a structured interview
protocol was used in all the interviews. In
addition, project reports and archives were
studied to provide reliability while
conducting the interviews.
In the mid-1990s, a Norwegian
company decided to build six new ships.
The company appointed a project manager
and a task force of engineers to work with
design requirements. In addition to a
custom-made design, the requirements
were that the ships be equipped with the
newest technology. As stated by the project
manager, a top-down (analogous)
estimating technique was used for
preparation of cost estimates.
Several shipbuilding yards around the
world were invited to tender. Evaluation
and selection of the winning shipbuilding
yard was based on an evaluation with focus
on the following factors of price, quality,
risk, technical solution, and financial
strength.
According to the project manager, two
types of risk analysis have been performed
on the project. One of these was daily or
operational risk management, i.e., identify,
analyze, and initiate proper response to
reduce the risk in the project. The other
form of risk analysis performed in the
project had a more strategic character and
involved Monte Carlo simulation.
The result from the simulation was
that a risk reserve was added to the base
estimate up to the 90/10 percentile.
Control of the risk reserve was divided
between the steering group, the line
organization and the project owner. The

steering group had the authority to use 1/6

of the risk reserve, the line organization 2/6
and the project owner 3/6.
The project manager had no control
over the risk reserves, but a rule was made
that he had the authority to use up to 1
million US dollars in each case of the
reserve controlled by the steering group. If
more money was needed, he had to make a
request to the steering group. When the
reserve controlled by the steering group
was spent, money should be transferred
first from the line organization and after
that from the project owner to the steering
group.
In this project it seems they have done
a thorough job on estimating the risk
reserve. However, according to the project
manger they did not differentiate between
contingency and allowance. Further, it
looks like the risk reserve does not include
a budget reserve, which covers changes in
the project's assumption and other major
unforeseen impacts.
It is the authors view that the project
manager should have had more control
over the risk reserve, as shown in figure 6.
The project manager's limited authority to
use the reserve could result in constrained
management flexibility.
Unexpected
development in the project can happen
suddenly and to handle such a situation
quickly, the project manager should
control a larger portion of the risk reserve.
On the other side, the distribution of the
risk reserve between several actors hinders
misuse of it.
iven that there is a well
established practice in deriving
uncertainty in cost estimation by
using probabilistic methods and tools, the
risk reserves could be deduced as follows:

The distribution density function for

the
activity/object
should
be
developed.
If it is not specified what kind of
distribution, use PERT as the default
distribution type. Ask for estimation of
optimistic, pessimistic and most likely,
respectively.
Calculate expected value, i.e., 50/50
estimate.
Contingency is the difference between
expected value and most likely figure
in the estimate.
If the management wants, for example,
an 80/20 estimate (only 20 percent

chance of cost overrun), an additional

allowance should be included.
A budget reserve should be estimated 4.
to cover other major unforeseen
impacts.

If a satisfactory set of statistical data is

not present for the single activities or group
of activities, the distribution can be derived
when the base estimate is established and is
defined as the most likely figure. A
solution then is to introduce the
contingency figure that lifts the estimation
from most likely to the expected figure.
Tables of possible contingency figures are
given in tables 2 and 3, for mature and
poorly defined projects, respectively.
When performing deterministic
estimates only, the derived plain base line
figures should as far as possible represent
the most likely or equivalent as-built
reference figures. One may then just add
to these figures a contingency representing
the proposed percentage addition as given
in tables 2 and 3 for mature and immature
projects, respectively.
The total approved budget should,
however, as a general rule, be of such a
nature that it is significantly less likely to
experience cost overruns than cost
underruns. A budget reserve should thus
be added to the control estimate.
The budget reserve for the specific
project of concern should reflect the
following aspects:

5.

6.

International Transactions, (1997):

A.06.1-A.06.6.
Picken, David H. and Stephen Mak.
Risk Analysis in Cost Planning and its
Effect on Efficiency in Capital Cost
Budgeting. Logistics Information
Management, 14, 5/6 (2001): 318327.
Thompson, A. and J.G. Perry.
Engineering Construction Risks: a
Guide to Project Risk Analysis and
Assessment: Implications for Project
Clients and Project Managers.
London: Thomas Telford, 1992.
Yeo, K.T. Risks, Classification of
Estimates,
and
Contingency
Management.
Journal
of
Management in Engineering, 6, 4
(1990): 458-470.

RECOMMENDED READING
1. Burger,
Riaan.
Contingency,
Quantifying the Uncertainty. Cost
Engineering, 45, 8 (2003): 12-17.
2. Nassar, Khaled. Cost Contingency
Analysis for Construction Projects
Using
Spreadsheets.
Cost
Engineering, 44, 9 (2002): 26-29.
3. Rolstadas,
Asbjorn.
Praktisk
Prosjektstyring. Trondheim, Norway:
Tapir Forlag, 2001.

ABOUT THE AUTHORS

Jan Terje Karlsen is an associate
The overall business risk attitude of the professor with the Norwegian School of
company, are you a risk taker, neutral, Management in Oslo, Norway. He can be
contacted
by
e-mail
at:
or risk adverse
The size of the project in the total [email protected]
portfolio, and its contribution to the
Jon Lereim is a professor with the
uncertainty of the total project
Norwegian School of Management in
portfolio.
Oslo, Norway. He can be contacted by eREFERENCES
mail at: [email protected]
1. Chen, Dong and Francis T. Hartman.
A Neural Network Approach to Risk
Technical Articles - Each month, Cost
Assessment
and
Contingency Engineering journal publishes one or more peerAllocation. AACE International reviewed technical articles. These articles go
through a blind peer review evaluation prior to
Transactions, (2000): 07.01-07.06.
2. Dillon, Robin L., John Richard and publication. Experts in the subject area judge the
technical accuracy of the articles. They advise the
Detlof von Winterfeldt. Assessment of authors on the strengths and weaknesses of their
Cost
Uncertainties
for
Large submissions and what changes can be made to
Technology Projects: A Methodology improve the article.
and an Application. Interfaces, 32, 4
(2002): 52-66.
3. Moselhi, Osama. Risk Assessment and
Contingency Estimating. AACE

Cost Engineering Vol. 47/No. 9 SEPTEMBER 2005

29