International Journal of Project Management 21 (2003) 261–269

www.elsevier.com/locate/ijproman

Modelling global risk factors affecting construction

cost performance
Daniel Baloia,*, Andrew D.F. Priceb
a
Universidade Eduardo Mondlane, Campus Universitario—GIU, PO Box 257, Maputo, Mozambique
b
Department of Civil and Building Engineering, Loughborough University, Leicestershire LE11 3TU, UK

Received 19 September 2001; received in revised form 28 November 2001; accepted 12 February 2002

Abstract
This paper discusses the core issues of global risk factors modelling, assessment and management. The research reported upon
forms part of a larger study that aims to develop a fuzzy decision framework for contractors to handle global risk factors affecting
construction cost performance at a project level. Major global risk factors affecting cost performance were identified through an
extensive literature review and preliminary discussions with construction contractors. The main decision perspectives namely nor-
mative and behavioural were explored. Different decision-making technologies, both classical and emergent, such as classical
management science techniques and DSSs, KBSs were explored and evaluated. Preliminary indications show that Fuzzy Set Theory
is a viable technology for modelling, assessing and managing global risk factors affecting construction cost performance and thus a fuzzy
decision framework for risk management can be successfully developed.
# 2003 Elsevier Science Ltd and IPMA. All rights reserved.
Keywords: Cost performance; Decision Support System; Fuzzy Set Theory; Risk management; Uncertainty

1. Introduction construction ranging from the simplest to more complex

projects such as nuclear plants, environmental restoration,
This paper discusses the core issues of risk modelling, transport systems and oil and gas platforms have
assessment and management with emphasis on global increasingly faced cost overruns. Raftery [1] pointed out
risk factors that affect cost performance. The research that construction projects tend to have poor reputation
reported upon forms part of a larger study that aims to for excessive time and cost overruns. Morris and Hough
develop a fuzzy decision framework for contractors to [2], during a study of records for different types of pro-
handle global risk factors affecting construction cost jects funded by the World Bank between 1974 and 1988,
performance. Major risk factors affecting cost perfor- found that 63% out of 1778 projects had experienced
mance were identified through an extensive literature significant cost overruns. Kaming et al. [3] studied fac-
review complemented by a workshop with construction tors influencing time and cost performance on high-rise
contractors in Mozambique. A questionnaire survey projects in Indonesia and concluded that cost and time
will be conducted at a later date to confirm the factors overruns were very frequent. Therefore, poor cost per-
identified earlier and to determine the most critical. The formance of construction projects seems to be the norm
findings of the questionnaire will form the basis for struc- rather than the exception particularly in most developing
tured interviews using the Repertory Grid technique to countries where the problem is more acute.
elicit relevant knowledge so as to develop a knowledge Various factors significantly influence construction
based decision support system. costs from the estimating stage to project completion.
Poor cost performance of construction projects has Some factors are intrinsically related to construction
been a major concern for both contractors and clients. organisations which are solely responsible for managing
Despite the large number of reported cases, it seems that them, whereas others are closely related to the socio-cul-
tural, economic, technological and political environments
* Corresponding author. Tel.: +258-1-490088; fax: +258-1-491025. within which such organisations operate. The latter are
E-mail addresses: baloi@zebra.uem.mz (D. Baloi), a.d.f.price@ usually called global risk factors. As a principle,
lboro.ac.uk (A.D.F. Price). contractors are not normally responsible for risk factors
0263-7863/03/$30.00# 2003 Elsevier Science Ltd and IPMA. All rights reserved.
doi:10.1016/S0263-7863(02)00017-0
262 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269

outside their control and traditional forms of contracts 3. Definition of risk

should provide a fair and sensible allocation of risks
between the various parties. However, contractors in Risk has different meanings to different people; that is,
developing countries often have to bear most of the the concept of risk varies according to viewpoint, attitudes
construction risks including those for which they have and experience. Engineers, designers and contractors view
little control. Unfortunately, many contractors are risk from the technological perspective; lenders and
unfamiliar with these risk factors and do not have the developers tend to view it from the economic and
experience and knowledge to manage them effectively. financial side; health professionals, environmentalists,
As a consequence, conflicts, poor quality, late comple- chemical engineers take a safety and environmental
tion, poor cost performance and business failures are perspective. Risk is therefore generally seen as an
commonplace in the construction industry. abstract concept whose measurement is very difficult [1].
Risk management is nowadays a critical factor to The Oxford Advanced Learner’s Dictionary—1995 ed.
successful project management, as projects tend to be defines risk as the: ‘‘chance of failure or the possibility
more complex and competition increasingly tougher. of meeting danger or of suffering harm or loss’’. In
There is a direct relationship between effective risk construction projects, risk may be defined as the like-
management and project success since risks are assessed lihood of a detrimental event occurring to the project.
by their potential effect on the objectives of the project. Since the objectives of construction projects are usually
Contractors have traditionally used high mark-ups to stated as targets established for function, cost, time and
cover risk but as their margins have become smaller this quality, the most important risks in construction are the
approach is no longer effective. In addition, the construc- failure to meet these targets. However, risks are not
tion industry has witnessed significant changes particu- always associated with negative outcomes. Risks may
larly in procurement methods with clients allocating represent opportunities as well, but the fact that most
greater risks to contractors. risks usually have negative outcomes has led individuals
Evidence shows that there is a gap between the existing to consider only their negative side.
risk management techniques and their practical appli-
cation by construction contractors [4–6]. Many reasons
have been put forward to explain why this is the case 4. Risk management
but it seems that assessment and analysis are the most
controversial issues. There is thus a need to explore new Risk management is a process comprising the following
directions in risk management so as to respond to the main steps: risk management planning, risk identifica-
expectations of construction industry towards an effective tion, risk assessment, risk analysis, risk response, risk
and efficient modelling, assessment and management of monitoring and risk communication [1,6].
risk.

5. Taxonomy of risks
2. Objectives
Many different classifications of risk have been
The overall aim of the research discussed in this paper developed over the years, however, most of these have
is to develop a fuzzy decision framework for a systema- considered the source criteria as the most important.
tic modelling, analysis and management of global risk Following this criteria, a broad classification of
factors affecting construction cost performance from construction project risks could be: technical, construc-
contractor’s perspective and at a project level. The study tion, legal, natural, logistic, social, economic, financial,
is concerned with financial risk rather than hazard. In commercial and political [5,6]. However, apart from the
order to achieve the overall objective the research will: source criteria, there have been other forms of classify-
ing risks, which take different perspectives. A classifica-
 review literature on risk modelling, analysis and tion taking into account the location of the impact of
management; risks in the elements of the project was suggested by Tah
 review management science theories and techni- [7]. It is also usual to categorise risks into dynamic/static,
ques; corporate/individual, internal/external, positive/negative,
 determine the most critical risk factors affecting acceptable/unacceptable and insurable/non-insurable.
construction cost performance;
 elicit relevant knowledge related to the most cri-
tical factors; and 6. Developing countries
 formalise such knowledge, develop, test-evaluate
a fuzzy decision framework for handling global As stated by Ofori [8], the ‘‘structural problems of
risk factors. construction industry in developing countries are more
 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269 263

fundamental, more serious, more complex, and, over- tion, which focused on the costs and revenue variables.
all, much more pressing than those confronting their Cost-related variables included labour, material and
counterparts elsewhere’’. Common problems affecting overheads whereas revenue-related variables included
construction industry in developing countries include taxation, repatriation restrictions and foreign exchange
lack of management skills, shortage of skilled labour, rates. Kaming [3] studied factors influencing construction
low productivity, shortage of supplies, bad quality of time and cost overruns on high-rise projects in Indone-
supplies and lack of equipment. Apart from technical sia and concluded that inflation, inaccurate estimates,
issues, management-related problems are one of the project complexity, weather conditions, project location
most important aspects facing construction contractors and local regulation were the main contributors.
since they have to deal with substantial constraints such as The factors affecting cost performance of construction
incomplete information, unpredictable client behaviour, projects seem to be well appreciated in the literature,
and uncertain project circumstances. however, there are some gaps. Firstly, most of the current
techniques and tools are founded on statistical decision
theory models. Contractors rarely use these techniques
7. Research into risk factors affecting cost performance and tools in practice. In this regard Mulholland and
Christian [16] asserted that there was a lack of an
Research into the poor cost performance of construction accepted method of risk assessment and management in
projects has pointed out several variables as risk drivers. the construction industry. More often than not con-
The cost performance of a single project is often in terms struction contractors and other practitioners rely on
of cost growth, i.e. the percentage difference between the assumptions, rules of thumb, experience and intuitive
final contract amount and the contract award amount [9]. judgement which can not be fully described by pre-
The final contract amount includes all additions, altera- scriptive or normative models. Individual knowledge
tions and deductions resulting from project changes. and experience, however, need to be accumulated and
More often than not, variations and claims are inherent structured to facilitate the analysis and retrieval by
in construction projects because uncertainties lead, others. Secondly, considerable research into risk man-
invariably, to the need for adjustments. However, varia- agement has mainly focused on the easily quantifiable
tions and claims have been the main reasons for disputes variables and has neglected other important factors.
due to both conflicting interests of the parties and the There is thus a need to combine both prescriptive and
complexity of contractual provisions when dealing with behavioural models in the modelling, assessment and
the valuation of variations and settlement of claims. management of risks since construction contractors do
Conflicting interests that lead to adversarial relation- not only rely on prescriptive models to make decisions
ships between clients and contractors have their economic but they also make an extensive use of experience, rules
roots namely within clients’ costs and contractors’ profits. of thumb and intuitive judgement. It is also suggested
Many studies on poor cost performance of construction that global risk factors pose more challenges to con-
projects have focused on change-order rate, which can tractors than others categories of risks particularly in
be defined as the ratio between the amount of change- developing countries. The recommendations and models
orders and the contract award amount. Factors found produced so far need to be complemented in order to
to influence change-order rate include level of competi- account for the complexity and dynamism surrounding
tion, project size and type. Okpala and Aniekwu [10] construction projects. Although some sources of uncer-
argued that delay and cost overruns could be reduced tainties are very difficulty to assess, analyse and to
through the increase of human efficiency. In addition, communicate, reliance on arbitrary assumptions may be
they identified price fluctuations, fraudulent practices misleading and controversial. Furthermore, overlooking
and ‘kickbacks’ as the major factors of poor cost per- or ignoring risks does not make them go away.
formance in construction projects in Nigeria. Further,
research by Jahren and Ashe, Elinwa and Buba, [11, 12]
found similar variables as the most influencing factors 8. Global risk factors
of project cost overruns. Risk factors associated with
political instability, fluctuations in currency, corruption, Construction organisations operate within an environ-
interest rates and material availability were considered ment and not a vacuum. They are inevitably influenced by
the main causes of additional costs in privatised infra- and constantly interacting with their environment. As
structure projects in developing countries [13]. Kangari such, construction organisations and, consequently con-
and Lucas [14] mentioned that all government-funded struction projects, are open systems rather than closed
projects in developing countries were political in nature. systems [17]. Closed systems are rigid and do not adapt
Political risk, in turn, invariably leads to bribery and to changing environments. An example of a closed sys-
corruption. Ashley and Bonner [15] developed a model tem is computer that is designed to perform specific and
for political risk assessment in international construc- repetitive tasks within specific conditions. Construction
264 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269

organisations’ efficiency and effectiveness largely depend Global risk, on the other hand, refer to risks factors
upon how managers scan the external project environ- that are not directly present in cost estimates yet they
ment, identify the critical factors and accordingly adapt may lead to significant financial disasters. Global risks
their organisations. The project environment can be are called so because they transcend the boundaries of
subdivided as follows: the organisation yet they have large impact on it. Con-
tractors have less control over these risk factors and
 inner layer or internal environment; contracts should provide fair and sensible allocation of
 operational environment; and these risks between the parties. However, in practice,
 outer layer or general environment. contractors in developing countries have to bear most of
construction risks.
Both general and operational environments are external Acts of God represent risks with extremely low prob-
environments. The general environment is broad in ability of occurrence, but can have huge negative
scope and comprises five basic domains, namely tech- impacts on projects if they occur. This category of risk
nological, social, physical, economic and political [17]. come under force majeure under the terms of contract
Risk factors affecting cost performance are classified in and includes events such as heavy floods, landslide,
this study as Organisation-specific, Global and Acts of earthquake and hurricanes. These risks are excusable
God. These risk categories can be broken down to and generally insurable but non-compensatable risks.
achieve a more detailed and comprehensive picture. The unique and main groups of global risks factors,
Organisation-specific risks are internal risks related to identified from the extensive literature search and
an organisation’s resources and management. Organi- preliminary discussions with construction contractors,
sation-specific risks include factors that are considered include economic, political, design, level of competition,
to be under contractor’s control, for example risks related fraudulent practices and construction-specific risk. Several
to labour skills and availability, materials delivery and authors, such as Jahren and Ashe, Ashley and Bonner,
quality, equipment reliability and availability, and and Akinci and Fisher [11,15,19] have highlighted the
management efficiency [18]. In principle, contractors are importance of these risk factors, which are shown in
solely responsible for managing these risks. Fig. 1.

Fig. 1. Main groups of global risk factors.

 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269 265

9. Current practices in risk management data, explanation capacity, difficulty at development

and appropriate domain.
Risk management is beneficial if implemented in a sys-
tematic manner from the planning stage through project
completion so as to make better and more informed deci- 11. Handling uncertainty
sions. The unsystematic and arbitrary management of
risks can endanger the success of the project since most The main objective of risk management is to reduce
risks are very dynamic throughout the project lifetime. uncertainty and thus improve decision-making. Uncer-
Indeed, risks may vary from appraisal, design, tender- tainty has many different sources and different types, [21].
ing, construction and commissioning. To meet these The main types of uncertainty include error, impreci-
demands, human and organisational dimensions play a sion, variability, vagueness, ambiguity and ignorance.
key role in the whole process of risk management. In The diversity in terms for different types of uncertainty
fact, risk management is both an art and a science. makes the modelling process a very difficult task
However sophisticated the analytical tools are, they are because the information concerning each specific uncer-
only a complement to the process. tainty is scarce. Several formal techniques for managing
Despite the large body of knowledge and continuous the different types of uncertainty have been developed
developments of the risk management discipline, it but there has not been any consensus on the appro-
seems that practitioners have not fully appreciated its priateness of such techniques so far. It seems that there
importance. Apart from the high-risk sectors such as oil is no best theory of uncertainty and therefore the choice
exploration and petrochemical, there is a significant gap of the most appropriate technique depends upon the
between the existing theory and practice of risk specific problem [22]. Four main approaches have been
management in the construction industry [4–6]. used for handling uncertainty in KB-DSS, namely
Probability Theory, Certainty Factor Theory, Demp-
ster–Shafer Theory and Fuzzy Set Theory.
10. Risk decision-making and uncertainty

Decision-making problems are broadly categorised 12. Probability Theory

into deterministic, stochastic/risk and uncertain. Deter-
ministic problems are those in which data are known The classical approach to address uncertainty is the
with certainty; whereas stochastic problems are those in Bayesian theory of probability. Probability Theory has
which data are not known with certainty but can be been used to model precisely described, repetitive experi-
represented by a probability distribution; and uncertain ments with observable but uncertain outcomes. The basic
refer to those problems in which data are not known. assumption in the classical theory of probability is that all
Most risk management problems fall into the last two uncertainties are measures of randomness or subjective
categories because they are poorly structured problems measures of confidence. The main concern of random
for which few algorithms or mechanical methods exist. processes is the statistics. The heavy reliance on the
Indeed, risk management relies heavily on experience, Probability Theory as the only effective and reliable
subjectivity and human judgement. Poorly structured methodology to deal with uncertainty has historical
problems can not be formulated at the desired level of roots. Probability Theory has well-established and
precision due to the surrounding uncertainty. However, sound scientific foundations and has been widely used
recent developments in information technologies have for centuries. Probabilitics argue that random methods
made it possible to improve the formulation of poorly are the only effective methods for dealing with uncer-
structured problems so as to aid decision making. In tainty. Deterministic approaches are commonplace in
this regard much research effort has put on the devel- ‘hard’ sciences and the pursuit for precision in such
opment of decision support systems (DSS) and, parti- areas is more favoured than qualitative models.
cularly, knowledge based systems (KBS). According to
Sprague and Watson [20] a DSS has the following main
features: computer-based system helps decision-makers; 13. Certainty Theory
confronts poorly structured problems; and has direct
interaction. There are several DSS development techniques Certainty Theory was formulated by Buchanan and
for applications in the construction industry and they Shortliffe [23] to handle uncertainty in the medical
include: rule-based expert systems; case-based expert expert system (an expert system which diagnoses micro-
systems; model-based expert systems; neural network bial infections) MYCIN. It was developed in attempt to
based systems; and genetic algorithm based systems. overcome some of the weaknesses of Probability The-
The choice of an appropriate technique depends mainly ory. Certainty Theory fundamentals are the concepts of
upon the difficulty at knowledge acquisition, required ‘‘certainty measures’’ which are associated with ‘‘factual
266 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269

statements’’. Certainty measures comprise numbers where the accuracy and precision are considered impor-
ranging from 1 to +1 where 1 represents complete tant. Certainty Factor Theory provides a simple method
certainty that a proposition is false and +1 represents for handling uncertainty but it is said as lacking an
complete certainty that a proposition is true. The fac- adequate theoretical underpinning [26]. The Dempster–
tual statements are rules comprising antecedents and Shafer Theory is richer in terms of semantics since it
consequents. allows an expression of partial knowledge. Its main
shortcoming is the elicitation and interpretation of
belief functions, [21]. Fuzzy Set Theory has proved to be
14. Dempster–Shafer Theory of Evidence a powerful technology in modelling unstructured pro-
blems particularly in the electronic industry. In recent
Dempster–Shafer Theory of Evidence [24] is usually years there has been a significant increase of its appli-
called epistemic probability because it provides an cation in the construction industry, particularly for
alternative model for the assessment of numerical modelling uncertainty in KBSs.
degrees of belief. Dempster–Shafer attempted to distin-
guish between uncertainty and ignorance and instead of
probabilities, belief functions are used. The main dis- 17. Uncertainty and risk management in construction
tinctions between Bayesian models of numerical degrees
and Dempster–Shafer model are the following: It has already been mentioned that construction risk
management is mainly based upon experience, assump-
 belief functions of Dempster–Shafer are set tions and human judgement. Consequently, risk man-
functions rather than point values; agement is mainly cognitive in nature. Kahneman and
 rejection of the law of additivity for belief in Tversky [27] uncovered important aspects of cognitive
disjoint propositions; and and non-cognitive types of uncertainty. Klir and Foger
 Dempster–Shafer Theory has an operation for [28] discussed two types of uncertainty namely ambiguity
the pooling of evidence from various sources. and vagueness.
Ambiguity is generally due to a non-cognitive root
causes. Ambiguity is a state in which an expression or
15. Fuzzy Set Theory word may have a number of distinct meanings and only
the context may help clarifying the real meaning. Evi-
Fuzzy Set Theory is a branch of modern mathematics dence theory has proved to be the most suitable mathe-
that was formulated by Zadeh [25] to model vagueness matical framework to handle ambiguity type of
intrinsic to human cognitive processes. Since then, it has uncertainty.
been used to tackle ill-defined and complex problems Vagueness arises when the outcome of an experiment
due to incomplete and imprecise information that char- can not be properly observed. A vague expression is ill
acterise the real-world systems. It is, therefore, suitable defined and lacks preciseness or sharpness. For exam-
for uncertain or approximate reasoning that involve ple, the global risk factor ‘‘Strong competition’’ does
human intuitive thinking. Fuzzy Set Theory uses linguistic not have an exact meaning, because the qualifier
variables and membership functions with varying grades ‘‘strong’’ may assume several degrees of intensity.
to model uncertainty inherent in natural language. ‘‘Strong competition’’ may involve a wide spectrum of
human perceptions, as there are no rigorous definition
of what ‘‘Strong competition’’ is.
16. Appropriateness of the approaches A particular type of vagueness is fuzziness. Fuzziness
is a kind of imprecision where the transition from a
Probability Theory considers all uncertainty random, membership state of an element to a set is gradual. It is,
however, not all types of uncertainty are random. A nevertheless, most frequent in situations where human
great deal of management issues in construction does judgement is an essential feature such as reasoning,
not comply with randomness properties. They are learning and decision-making process.
mainly cognitive and thus do not lend themselves to
precise measurement. Several authors have highlighted
some serious shortcomings related to the Bayesian statis- 18. Fuzzy Set Theory
tics. Zadeh [25] suggested that the assumptions associated
with Bayesian function namely, mutual exclusivity of Fuzzy Set Theory has been used to tackle ill-defined
events, conditional independence and exhaustivity of and complex problems due to incomplete and imprecise
events do not always hold. In addition, there is no dis- information that characterise the real-world systems.
tinction between randomness and ignorance. Probabil- Zadeh stated that ‘‘as the complexity of a system
istic approaches are best suited to mechanistic systems increases, human ability to make precise yet significant
 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269 267

statements about its behaviour diminishes until a purpose of providing a means of approximate char-
threshold is reached beyond which precision and sig- acterisation of phenomena that are too complex or too
nificance become mutually exclusive’’—the Principle of ill defined to be amenable to description in conventional
Incompatibility. Then, it follows that modelling complex quantitative terms. Any risk factor presented in Fig. 1
or ill-defined systems can not be made precisely. Fuzzy can be characterised using linguistic variables through
Set Theory is not intended to replace Probability Theory its likelihood of occurrence and its severity or impact.
but rather to provide solutions to problems that lack Examples of linguistic variables are expressions such as,
mathematical rigour inherent to Probability Theory. ‘‘need for job’’, ‘‘number of bidders’’, and ‘‘market
Essentially, Fuzzy Set Theory is an extension of the conditions’’. These linguistic variables may assume dif-
classical Boolean or binary logic. In Classical Set the- ferent values such as ‘‘very high’’, ‘‘high’’, ‘‘moderate’’,
ory, a set is a collection of objects having a general ‘‘low’’ and ‘‘very low’’, which are fuzzy sets (member-
property, for example, a set of clients. In Classical ship functions) and represent the perception of the
Logic, an element is, therefore, either or not a member decision-maker on the magnitude of any risk factor.
of a set. The boundaries of concepts are very rigid or
‘crisp’ and there is no room for grey or ‘in between’
states. There are no intermediate grades of membership 20. Membership functions
between full and non-membership. This deterministic
‘yes-or-no’ or ‘dichotomous’ approach is nowadays A fuzzy set is a set whose elements have varying
widespread practice in systems modelling, reasoning degrees of membership. The degrees of membership of
process and computing. The main problem with binary an element are expressed by a membership function.
approach is that it fails to convey information effectively, Membership functions in Fuzzy Set Theory play a similar
that is, the states between full and non-membership are role to that of probability distribution functions in
ignored yet they are very important. Meanwhile, most Probability Theory, that is, membership functions are
real-world systems are very complex and ill defined to used to represent uncertainty. A membership function is
be well understood and modelled precisely. a function that maps a universe of objects, X, onto the
The essence of fuzziness, in contrast to binary or dual unit interval [0, 1]. The universe of objects represents the
logic, is that the transition from a membership to non- elements of the set and the interval corresponds to the
membership state of an element of a set is gradual set of grades. The grades of membership in fuzzy sets
rather than abrupt. Thus, Fuzzy Set Theory allows a may fall anywhere in the interval [0, 1]. A degree of 0
generalisation of the classical set concept to model (zero) means that an element is not a member of the set
complex or ill-defined systems. The main concepts at all. A degree of 1 (one) represents full membership. In
associated with Fuzzy Set Theory, as applied to decision contrast with ‘crisp’ sets that have only one membership
systems are membership functions, linguistic variable, function, fuzzy sets have a large number of membership
natural language computation, linguistic approxima- functions.
tion, fuzzy set arithmetic operations, set operations and
fuzzy weighted average. Details concerning these topics
can be found in [25,29]. 21. Modelling global risk factors using Fuzzy Set Theory

Global risk factors that affect construction cost

19. Linguistic variable performance can be modelled using Fuzzy Set Theory
and a fuzzy decision support system can be developed.
Research in cognitive psychology suggests that indi- For that purpose it is necessary: to determine the most
viduals base their thinking on conceptual patterns and significant risk factors; define the different linguistic
mental images rather than on any quantity or numbers. variables, which correspond to the constructs used by
Although natural language is imprecise it conveys valu- construction professionals to describe risk factors;
able information and despite the vagueness inherent in define the membership functions for each linguistic
natural language humans can understand each other variable and build a knowledge base. Significant risk
perfectly. The concept of linguistic variables lies at the factors, linguistic variables, membership functions and
core of Fuzzy Set Theory, since the basics of Fuzzy Set knowledge base can be determined through literature
Theory is the manipulation of linguistic expressions search, questionnaires and interviews with construction
instead of numbers. The values assumed by linguistic industry practitioners. The knowledge base is a reposi-
variables are words. A linguistic variable differs from a tory of experts’ knowledge and experience. The knowl-
numerical variable in that its values are not numbers edge in the knowledge base can be represented using
but words or sentences in a natural or artificial lan- ‘‘production rules’’, which is a knowledge representa-
guage. Since words in general are less precise than tion system that seems to capture well risk knowledge,
numbers, the concept of linguistic variables serves the because a great deal of knowledge and heuristics used
268 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269

by construction contractors in risk decision-making is in devise reliable membership functions, and build the
the form of rules (for example, IF subcontractor A is very knowledge base, as important ingredients of a fuzzy
experienced, THEN performance is good; IF inflation is decision support system. Both the inputs and outputs of
medium and exchange rate is medium THEN economic the system are qualitative and in natural language,
risk is medium). The main advantage of production which facilitates its use and understanding. The decision
rules are that they are easy to understand, provide framework would be an important tool for contractors
modularity in design, which enables addition, deletion to increase their awareness, identify global risk factors
and change of rules independently. In addition to the affecting cost performance, assess their impact and like-
knowledge base, databases and analytical models can be lihood and take appropriate measures in order to reduce
incorporated in the decision support system in order to their impact on cost performance.
enhance its performance.
The inputs to the decision support system are the
assessments of the different global risk factors specific to References
a project in linguistic terms (high, medium, low). The
system checks the knowledge base and databases and [1] Raftery J. Risk analysis in project management. London: E &
performs natural language computations and produces FN Spon, 1994.
the risk impact for each group of risk factors as well as [2] Morris P, Hough G. The anatomy of major projects. New-York:
the overall risk (combination of partial risk impacts) John Wiley, 1987.
[3] Kaming PF, Olomolaiye PO, Holt GD, Harris FC. Factors
and the corresponding likelihood in linguistic terms. influencing construction time and cost overruns on high-rise
For example the overall risk impact can be ‘‘low’’ with projects in Indonesia. Construction Management and Economics
‘‘high’’ likelihood. The system can also provide recom- 1997;15:83–94.
mendations on the most appropriate risk response strate- [4] Perry JG, Hayes RW. Risk management in construction projects.
gies in the light of risk analysis results. The decision-maker Proc. Inst. Civ. Engrs Part I 1985;78:499–521.
[5] Thompson P, Perry J. Engineering Construction Risks. A guide
can then make her/his judgement and take appropriate to project risk analysis and risk management. London: Thomas
measures to mitigate project risks and thus improve the Telford, 1992.
likelihood of project success. [6] Flanagan R, Norman G. Risk management and construction.
London: Blackwell, 1993.
[7] Tah JHM, Thorpe A, McCaffer R. Contractor project risks con-
tingency allocation using linguistic approximation. Engineering,
22. Conclusions Construction and Architectural Management 1996;3(4):251–69.
[8] Ofori G. Research on construction industry development at the
This paper discussed the core issues of global risk crossroads. Construction Management and Economics 1993;11:
factors modelling, assessment and management. Evi- 175–85.
dence shows that poor cost performance of construction [9] McKim R, Hegazy T, Attalla M. Project performance control in
reconstruction projects. Journal of Construction Engineering and
projects seems to be the norm rather than the exception, Management 2000;126(2):137–41.
and both clients and contractors suffer significant [10] Okpala D, Aniekwu A. Causes of high costs of construction in
financial losses due to cost overruns. It is suggested that Nigeria. Journal of Construction Engineering and Management
global risk factors pose more challenges to construction 1988;114(2):233–45.
contractors, which are less familiar with them. In addi- [11] Jahren C, Ashe A. Predictors of cost-overrun rates. Journal of
Construction Engineering and Management 1990;116(3):548–51.
tion, they lack effective techniques and tools to handle [12] Elinwa A, Buba S. Construction costs factors in Nigeria. Journal
these risks. of Construction Engineering and Management 1993;119(4):698–
Risk management has traditionally been largely based 713.
on experience and subjective judgement; that is, it features [13] Rosenbaum DB. Risk Questions hamper deals. Engineering
News Record 1997;24:11.
humanistic systems that are not characterised by preci-
[14] Kangari R, Lucas CL. Managing international operations: a
sion. Unlike mechanistic systems, humanistic systems guide for engineers. Architects and Construction Managers,
have proved impervious to mathematical analysis and USA, ASCE: 1997.
computer simulation. A humanistic system is a system [15] Ashley D, Bonner J. Political risks in international construction.
that is heavily dependent on human judgement and per- Journal of Construction Engineering and Management, 113 (3)
ceptions. As such, there is a need to explore new directions 1987:447–67.
[16] Mulholland B, Christian J. Risk assessment in construction
towards risk modelling, assessment and management. schedules. Journal of Construction Engineering and Management
Different techniques for handling uncertainty were 1999;125(1):8–15.
evaluated and preliminary results indicate that global [17] Walker A. Project management in construction. UK: Blackwell,
risk factors affecting construction cost performance can 1996.
be successfully modelled, assessed and managed using [18] Smith N, et al., editors. Managing risk in construction projects.
Oxford: Blackwell Science, 1999.
Fuzzy Set Theory and Decision Support System tech- [19] Akinci B, Fisher M. Factors affecting contractors’ risk of cost
nologies. For that purpose it is necessary to determine overburden. Journal of Management in Engineering 1998;14(1):
significant global risk factors, define linguistic variables, 67–75.
 D. Baloi, A.D.F. Price / International Journal of Project Management 21 (2003) 261–269 269

[20] Sprague RH, Watson H. Decision support systems. Putting the- [25] Zadeh LA. Fuzzy Sets. Information and Control 1965;8:338–53.
ory into practice. USA: Prentice-Hall, 1986. [26] Blockely DI, Baldwin JF. Uncertain inference in knowledge-
[21] Krause P, Clark D. Representing uncertain knowledge: an artifi- based systems. Journal of Engineering Mechanics 1987;III(4):
cial intelligence approach. Oxford: Intellect Press, 1993. 467–81.
[22] Saffioti A, Umkehrer E. Comparing uncertainty management [27] Kahneman D Slovic P, Tversky A. Judgement under uncertainty:
techniques. Microcomputers in Civil Engineering 1994;9:367–80. heuristics and biases. Cambridge, 1982; 185, September 1124–1131.
[23] Buchanan BG, Shortliffe EH. Rule-based expert systems—the [28] Klir GJ Folger TA. Fuzzy Sets, uncertainty and information
MYCIN experiments of the standard heuristic programming Englewood-Cliffs, 1988.
project. Reading: Addison-Wesley. Reading, MA. [29] Schmucker KJ. Fuzzy Sets, natural language computation, and
[24] Shafer G. Mathematical theory of evidence. Princeton, NJ: Uni- risk analysis. Rockville, Maryland: Computer Science Press,
versity Press, 1976. 1984.