Professional Documents
Culture Documents
Allocation and Management of Cost Contingency in Projects
Allocation and Management of Cost Contingency in Projects
in Projects
Mohammed Wajdi Hammad 1; Alireza Abbasi, Ph.D. 2; and Michael J. Ryan, Ph.D. 3
Abstract: Contingency is crucial for successfully managing projects since it provides a buffer against risk development. This research
introduces a new methodology to estimate and allocate cost contingency during the planning phase, as well as managing cost contingency
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 04/29/16. Copyright ASCE. For personal use only; all rights reserved.
during the execution phase of the project. The cost contingency allocation technique is based on the activity’s contribution to the overall cost
variance of the project, which includes the cost, uncertainty, and whether the activity is on the critical path or not. New measures have been
introduced in the proposed cost contingency management method to control the expenditure and balance of activities’ cost contingency and
estimate future need. The new cost contingency management method is applied to a real-life bridge maintenance project. The result shows that
the proposed cost contingency methodology is adaptive and robust enough to accommodate typical disruptions in the project environment.
DOI: 10.1061/(ASCE)ME.1943-5479.0000447. © 2016 American Society of Civil Engineers.
Author keywords: Project management; Cost contingency; Contingency estimation; Contingency allocation; Contingency management;
Project control; Monte Carlo simulation.
mance is assumed among the project activities and a probability National Aeronautics and Space Administration (NASA) in an ef-
distribution is assigned to each activity, all possible project costs will fort to understand the difference between the estimated costs and
be normally distributed with a mean cost similar in value to the one actual costs of software projects. He found that the main factor that
that is calculated by the addition of the activities’ mean costs, and contributed to the increase in the number of successful projects is
the variance of the possible project cost results to be equal to the the improvements in the cost estimation techniques.
accumulation of all activities’ cost variances. The proposed meth- Lhee et al. (2014) proposed a two-step artificial neural network
odology provided the performance status classification, which can (ANN)-based method for estimating the optimal contingency for
be used as a practical guide for the management decision making. construction projects. The model was developed by manipulating
Idrus et al. (2011) proposed a method to estimate cost contin- an intermediate form of contingency (i.e., contingency rate) as
gency using a flexible and rational approach that could accommo- the output variable of the neural network to predict the owner’s
date contractors’ subjective judgment based on risk analysis and the contingency on construction projects. A comparison of the perfor-
fuzzy expert system concept. The proposed method was applied to mances was performed with the one-step ANN-based model. Based
building and infrastructure projects in Malaysia. They applied risk on the net statistics such as average error, mean squared error, cor-
analysis as the main concept to identify and assess the level of risk relation values, and statistical analysis about error rates, the predic-
of each risk factor, while the fuzzy expert system was the method tive performance of the two-step model was shown to be better than
used to assess the level of risk in the risk assessment step. Develop- that of the one-step model. They also claimed that the two-step ap-
ment of the proposed model can be described in the following seven proach has the potential to improve an owner’s budgetary decisions,
stages: development of a conceptual model for cost contingency, reducing the risk of either underutilizing or overcommitting funds.
determination of risk factors for the model, development of the Eldosouky et al. (2014) have done research to trial and put
fuzzy expert system, model testing, tuning, and validation. Project an end to the incorrect opinion that “adding contingency funds
cost contingency was modeled as the function of the level of risk to the tender price of a project may lead to loss of the tender.” They
factors identified as risk magnitude. They found that the predictions proposed an approach for determining and monitoring cost contin-
given by the system were within 20% accuracy compared to actual gency reserve (CCR) for a project. Control of CCR is interfaced
cost contingencies. with earned value management. An application was carried out to
Turskis et al. (2012) reviewed the evolution of concepts, the use real-life projects. The proposed approach for determination of CCR
of reserves, robust itineraries, and contingency of time and cost in for a project with established context for the purpose of tendering
the construction industry. They presented the main trends in differ- includes the following six phases: acquiring project cost estimate
ent approaches to risk management in construction and construc- and high-quality schedule; describing project schedule and cost
tion project processes, such as reliability of production, discount uncertainty; preparing risk register; performing integrated cost/
methods, methods of reporting, and integrated management to- schedule Monte Carlo simulation; determining performance meas-
gether with trends towards standardization of the investment pro- urement baseline (PMB), CCR, management reserve, and nonspe-
cess management. They concluded that the risks can be analyzed cific risk provision; and finally, monitoring and control of project
depending on a context. Risk should be treated as a state, in which PMB and risk budgets. The authors determined that if a contractor
there is a possibility of a loss. One of the most relevant concepts of prepares a detailed cost estimate for the purpose of tendering for a
risk is the dispersion of actual and expected results. This concept certain project accompanied by a high-quality critical path method
leaves room for contingency. They also stated that processes in schedule, then the contractor will be anxious for performing project
construction industry, as well as implementation of construction risk analysis and determining CCR as proposed in this study. On
projects, are governed by specific laws. The need to standardize the other hand, if the contractor does not prepare such information,
legislation results from international cooperation in the execution they will not accept to pay attention to project risk management and
of contracts and requirements of banks. This promotes standardi- the determination of CCR according to the proposed approach or
zation, especially at the stage of preparation for and implementation any equivalent approach.
of investments. In summary, most of the papers found in the literature were
Gharaibeh (2014) investigated the issue of project cost control about estimating cost contingency, with very few about managing
in major power transmission projects, to understand the reasons cost contingency in a project and even fewer about combined the
behind cost overrun. The Delphi method was used to identify prob- cost contingency estimation and management. Touran (2003), Idrus
lems of controlling the project cost, and identifying lessons learned. et al. (2011), Uzzafer (2013), and Lhee et al. (2014) all show differ-
The Delphi study is normally a qualitative research method using ent methods to estimate cost contingency. Some require the project
questionnaires as a research instrument where a panel of experts is manager to have prior knowledge, such as fuzzy logic and artificial
asked to provide feedback and answers on selected questions nor- neural networks, while others require to make assumptions about
mally involving several iterations or rounds of questions (Linstone variables or collecting data. Eldosouky et al. (2014) presented a
and Turloff 1975). A three-round semistructured questionnaire method to manage contingency on a project. This method requires
correctly calculating contingencies. Idrus et al. (2011) require the The method allows the project manager to carefully plan activities
project manager to have knowledge regarding fuzzy logic, while based on the past and current spending pattern. Thirdly, the pro-
Lhee et al. (2014) require prior knowledge regarding artificial neu- posed method is adaptable and able to accommodate disruptions
ral networks (ANNs). in the project environment by allowing the project manager to se-
The management method provided by Barraza and Bueno lect the required confidence interval.
(2007) can be used in any project, but does not propose how to
allocate contingency to each activity. The management method
provided by Eldosouky et al. (2014) requires identifying all risks Proposed Cost Contingency Management
involved in the project, which can be difficult and very time con- Framework
suming to obtain the required data.
Table 1 shows a summary of all models found regarding con- Very few studies have been conducted on the subject of cost con-
tingencies in project management. In summary, the following is tingency management during project execution; such research is
what differentiates the proposed method from the previous nearly nonexistent. To fill this gap, this paper proposes a new
based on the fraction of PVA to PVP , regardless of the uncer- allocated to it for the given reporting period.
tainty, cost, or whether the activity is on the critical path or not. 9. Calculate remaining planned value (RPVA )
On the other hand, if the project manager chooses α to be RPVA ¼ ð1 − % completeÞ × PVA ð6Þ
less than 100%, for instance, a 90% confidence interval, the
first part of the equation will be considered as there is uncer- The RPV calculates how much planned value is remaining
tainty involved. This is where an activity’s contribution to the for the activity to complete. It is calculated by multiplying the
total cost variance is considered. The contribution to variance percentage remaining to complete the activity by the PVA .
answers questions such as “What percentage of the variance or 10. Calculate actual remaining contingency (ARCA )
uncertainty in the target forecast is caused by activity X?” It is
calculated by squaring the rank correlation coefficients and ARCA ¼ CVA þ CCA ð7Þ
normalizing them to 100%. Contribution to variance indicates ARCA is different that RCA because RCA calculated the
the relative importance by showing the percentage of the remaining contingency with respect to a given reporting
forecast variance contributed by each assumption (Oracle period. On the other hand, ARCA calculates how much of
Corporation 2012). The correlation coefficient shows the de- the allocated total cost contingency to each activity is re-
gree to which assumptions and forecasts change together. maining.
Since the project is made up of individual activities that are 11. Determine total remaining cost (TRCA )
connected to each other, the activities that have a significant
impact on the forecast will have a high correlation coefficient. TRCA ¼ ARCA þ RPVA ð8Þ
The higher the absolute value of the correlation coefficient,
the higher the contribution to variance and the stronger the This calculates the total cost (including contingency) re-
relationship. This means that any activity with a high cost maining at the end of each reporting period.
(affecting the budget of the project), or an activity that sits 12. Calculate planned updated remaining contingency for each
on the critical path (affecting the duration of the project), activity (PURCA )
or has a high uncertainty the higher the correlation, then
the stronger the relationship to the target forecast. This will PURCA ¼ CCA − ECSA ð9Þ
allow the project manager to find out which assumptions Calculates the minimum cost contingency for each activity
are influencing the forecasts the least and can pay less atten- that should remain at the end of the reporting period. For ex-
tion or discard them altogether, and focus more on the activ- ample, if an activity is 60% complete, then it is expected that at
ities that will have the most impact on the forecast. least 40% of the contingency is remaining.
4. Determine percentage completion for each activity: each activ- 13. Calculate actual updated remaining contingency for each ac-
ity is evaluated at the end of the reporting period and the per- tivity (AURCA )
centage completion is estimated by the project manager.
5. Record actual cost for each activity (ACA ): this is the cost IF % complete ¼ 100; then AURCA ¼ 0;
spent on an activity during the project execution and is re-
corded at the end of every reporting period. otherwise; AURCA ¼ ðRPVA =RPVP Þ × RCP þ PURCA ð10Þ
6. Calculate the expected contingency spent for each activity
(ECSA ) This means that if the % complete of an activity is 100%, then
the AURC is set to be zero. This is because when an activity is
ECSA ¼ % complete × CCA ð3Þ complete then any remaining contingency (or contingency overrun)
The ECS calculates the maximum amount of cost contin- will be distributed to other activities, either to be used as extra cush-
gency, for each activity, that is expected to be spent at the end ion against future uncertainties (in the case of excess contingency
of the reporting period. For example, if an activity is 60% com- remaining) or to distribute the overrun among other activities (in
plete, then the activity is expected to spend 60% of its allo- the case of spending more contingency than expected). At the end
cated contingency. of the reporting period, the remaining contingency of the project
7. Calculate cost variance for each activity (CVA ) (RCP ) will be distributed to other activities based on the share
of the RPVA of the activity in the RPVP of the project. The project
CVA ¼ ð% complete × PVA Þ − ACA ð4Þ manager should then decide if any further action is required to be
A positive CVA indicates that the activity is under budget, taken to keep the project within the original budget. If PURCP is
and no contingency was required. If the activity has completed greater than AURCP , the project has more cost contingency re-
and the CVA is still positive, then any remaining PVA will be maining than expected. If the PURCA is less than AURCA , the ac-
added to the remaining CCA and used for future remaining tivity has less remaining cost contingency than expected. If PURCP
pier). The project also included repairing two major gantries. The Each activity has an influence on the total cost variance of the
bridge project lasted for 21 months with a budget of $3 million, but project; this is used to allocate contingencies to each activity using
for the sake of this research, only the fabrication and installation of Eq. (1). The results for the cost contingency allocation are shown in
three new ladders are studied, which had a planned duration of just Table 2. The cost contingencies are allocated based on activity’s
over 2 months with a budget of $135,652. Since the project dura- contribution to the overall cost variance. This means that an activity
tion is only 39 days, the reporting period is chosen to be 10 days. with a high uncertainty, high cost, and that lies on the critical path,
The following are assumed when calculating cost contingencies will have a high contribution to the cost variance of the project and
for this paper: will therefore have a higher share of the cost contingency, and the
1. All activities are normally distributed to 10% of the mean cost. opposite is true.
To allow for this assumption, according to Barraza and Bueno The following is a detailed demonstration of how cost contin-
(2007), independent performance is to be assumed and probabil- gency is allocated to the activity 1.1, which requires the use of
ity distribution to be assigned to all activities. “Therefore all pro- Eq. (2)
ject costs will be normally distributed with a mean cost similar
in value to the one that is calculated by the addition of the ac- CCA ¼ ð1 − αÞ × ð% contribution to variance × CCP Þ
tivities’ mean costs, and the variance of the possible project cost
results to be equal to the accumulation of all activities’ cost þ α × ðPVA =PVP Þ × CCP
variance (as a result from the central limit theorem)” (Barraza
and Bueno 2007). As mentioned earlier, the project manager has chosen an 80%
2. Any remaining contingency, for say, activity A, should not be confidence interval, α ¼ 80%. The % contribution to variance
transferred to activity B until activity A is 100% complete. This for activity 1.1 was found to be 0.39% (refer to Table 2). The cost
assumption prevents spending the contingency on a first-come, contingency for the project (CCP ) was $5,372 (as discussed ear-
first-served basis, but instead focusing on keeping all activities lier), which is calculated as the difference between the total cost
under the target cost. It also allows identifying which activities of the project (TCP ) and the planned value of the project (PVP ),
are experiencing a cost overrun and requiring to be monitored refer to Eq. (1) and task 1 in Table 2. The planned value for activity
more closely (and in some instances take immediate actions). 1.1 (PVP ) is 4,500 (refer to Table 2), while the planned value for
The PVP for the project was estimated to be $126,500 (without the project (PVP ) is $126,500 (refer to Table 2). Therefore, the CCA
contingency). All work on this project was done in-house using is $155
Fig. 2. Results of total project cost obtained from Monte Carlo simulation
1.3.1.4 0.18 $3,000 $228 $3,228 the first reporting period is $230. This means that there is $230
1.3.1.5 0.08 $2,000 $227 $2,227 remaining (including planned value and contingency), which is cal-
1.3.2 1.06 $14,000 $1,142 $15,142 culated using Eq. (8) ($230 + $0).
1.3.2.1 0.14 $3,000 $228 $3,228 For the activity 1.2.1, the total PV assigned is $60,000 (as seen
1.3.2.2 0.32 $3,000 $230 $3,230
in Table 2), but since the activity is only 20% complete, the amount
1.3.2.3 0.36 $3,500 $230 $3,730
1.3.2.4 0.19 $3,000 $228 $3,228 of PV shown in Table 3 is $12,000 ($60,000 × 20%) while the ACA
1.3.2.5 0.04 $1,500 $227 $1,727 is $15,000. The ECSA is $246, which is calculated using Eq. (3)
1.3.3 2.33 $20,000 $1,156 $21,156 (20% × $1,228). This means that activity 1.2.1 is expected to spend
1.3.3.1 0.92 $6,000 $236 $6,236 a maximum of $246 of the $1,228 (from Table 2) assigned to it
1.3.3.2 0.49 $4,000 $231 $4,231 by the end of the first reporting period. The CVA is –$3,000, which
1.3.3.3 0.53 $5,000 $232 $5,232 is calculated using Eq. (4) (20% × $60,000–$15,000). Since CVA
1.3.3.4 0.36 $3,000 $230 $3,230
1.3.3.5 0.03 $2,000 $227 $2,227
is a negative value, this means that activity 1.2.1 exceed the planned
value and therefore the contingency will be used. The remaining
contingency (RCA ) is −$2,754, which is calculated using Eq. (5)
($246–$3,000). This means that activity 1.2.1 used $2,754 more
$4,500 contingency than what it was expected to use by the end of the
0.2 × ð0.39% × $5,301Þ þ 0.8 × × $5,301 ¼ $155
$126,500 first reporting period ($246 was expected). The ARCA is thus
The TCA for activity 1.1 is calculated [using Eq. (1)] as −$1,772, which is calculated using Eq. (7) (–$3,000 þ $1,228).
$4,500 þ $155 ¼ $4,655. This means that the cost overrun for activity 1.2.1 was larger than
For the first reporting period (after 10 days), activity 1.1 is 100% the total contingency assigned to it; $1,228 was originally assigned
complete and only 20% of the activity 1.2.1 is completed (refer to to activity 1.2.1 (refer to Table 2). The RPV is $48,000, calculated
the project consumed more contingency than expected. of the contingency for activities under 1.3.1, but some of the PVA
For the second reporting period (after 20 days), shown in was also saved. Work under 1.3.2 was 67% complete with ACA of
Table 4, activity 1.2.1 is 87% complete and the ACA was $55,000. $6,200, which is just over the PVA of $6,000; this is shown with a
The amount of contingency expected to be spent (ECSA ) was CVA of –$200. The ECSA is $458, which means that all activities
$1,068, i.e., 87% of the total allocated contingency to the activity under 1.3.2 are expected to spend a maximum of $458 of the total
(refer to Table 2 for total allocated contingency). The CVA for ac- $1,142 (refer to Table 2) allocated to it. The RCA and ARCA for the
tivity 1.2.1 was –$2,800. Since the value is negative, this means that second pier showed a positive value ($258), meaning that the cost
activity 1.2.1 is spending more money than its allocated PV. The overrun (CVA ¼ −$200) had been covered by the contingency and
allocated PVA for activity 1.2.1 in the second reporting period was there is still $258 remaining.
$52,200 (Table 4), while the ACA exceeded this amount and there- Overall, the project in the third reporting period had an AURCP
fore contingency will have to be used to cover the cost overrun. of $3,372 while the PURCP is $1,831. This shows that the project
Looking at the RCA , the results show a negative value (–$1,732) was financially doing much better (1.8 times) than expected.
but with a significant improvement when compared to the RCA The project was supposed to have $1,831 (out of $5,372 originally
value for activity 1.2.1 in the first reporting period (Table 4). The allocated to the project) remaining contingency at the end of
ARCA was –$1,572 (–$2,800 þ $1,228). Since the ARCA is a neg- the third reporting period but actually managed to have $3,372
ative value, this means that the cost overrun could not be covered remaining. This means that for the third reporting period, the
by the total contingency assigned to activity 1.2.1. The RPVA is project is doing better than expected and managed to recover
$7,800 and the total remaining cost (TRCA ) was $6,228. Overall, from the overrun achieved in the first and second reporting
the project had an AURCP at the end of the second reporting period periods.
of $2,572, which was much lower than the PURCP ($4,074). For the fourth and last reporting period (after 39 days), shown in
Again, this shows that the project is still consuming contingency Table 6, all work on the project is now completed. The ACP was
more than expected. $127,500 and the PVP was $126,500, which means that the project
had spent $1,000 more than was planned to be spent. The total re- Discussion
maining cost (TRCP ) at the end of the project was $4,372. This
shows that the cost contingency was enough to cover all cost over- It has been shown that contingency can have a tremendous impact
runs in the project and there was still $4,372 remaining at the end of on project outcome for project participants (Dey et al. 1994;
the project, whereas the expected contingency spent (ECSA ) at the Baccarini 2004) as it provides a buffer against risk development.
last reporting period (end of the project) is $5,372 (as shown in A contingency that is too large can result in poor cost management,
Table 6). uneconomic completion of a project, and a lack of available funds