You are on page 1of 16

09613218 # 1998 E & FN Spon

Linking procurement systems to project priorities

Mohan M. Kumaraswamy and Sunil M. Dissanayaka

Department of Civil & Structural Engineering, The University of Hong Kong , Pokfulam Road , Hong
Kong SAR
E - mail : mohan@ hkucc. hku . hk ; h9595041 @ hkusub. hku . hk

The main objective of this work is to assess the impacts of various procurement variables on project
performance , in comparison to the impacts of non-procurement related variables, such as project conditions
and team characteristics . Since the impacts of specic procurement variables , such as contract type, have
been investigated before, it was decided to rst develop a comprehensive framework of procurement
options. This framework is based on a holistic overview of procurement systems that included, for
example, sub-systems of work packaging, and type of contract. A model was next developed to link the
procurement system variables to project outcomes. Additional variables such as the characteristics of the
project and the project participants were incorporated into this model as intervening variables. A survey
in Hong Kong tested the relative strengths of such procurement sub-systems and intervening variables, in
terms of their inuences on project outcomes. The interim results are described in this paper. This
includes observations that cost and time overruns were not signicantly inuenced by the chosen
intervening variables. Such observations led to the identication of particular needs to further probe the
inuences of team performance levels, as well as of procurement sub-systems. This evaluation of the other
relationships in this proposed model will assist clients and their advisers to design more appropriate
procurement systems, that should be geared to their particular project priorities .

L objectif principal de ce travail est d e valuer les incidences de diverses variables d approvisionnement sur
l exe cution d un projet, par rapport aux incidences de variables qui ne sont pas lie es aux approvisionnements,
comme les conditions du projet et les caracte ristiques de l e quipe. Les incidences des variables propres a un
approvisionnement , comme le type de contrat, ayant de ja e te examine es, on a de cide de constituer tout
d abord un cadre complet des options d approvisionnement . Ce cadre repose sur une vue ge ne rale holistique
des syste mes d approvisionnement qui comprenaient, par exemple , des sous-syste mes de lots de travail et la
type de contrat. On a ensuite re alise un mode le pour lier les variables du syste me d approvisionnement aux
re sultats du projet. D autres variables comme les caracte ristiques du projet et les participants ont e te inte gre es
dans le mode le comme variables d intervention . Dans le cadre d une e tude re alise e a Hong Kong , on a teste
les re sistances relatives de tels sous-syste mes d approvisionnement et variables d intervention , en termes de
leurs inuences sur les re sultats du projet. Cet article donne les re sultats provisoires. On a notamment
observe que les de passements de dates et de cou ts ne subissaient pas de fac: on importante l inuence des
variables d intervention choisies. Ces observations ont conduit au recensement des besoins particuliers pour
mieux sonder l inuence des niveaux de performance des e quipes ainsi que des sous-syste mes
d approvisionnement . Cette e valuation des autres relations dans le mode le propose devrait aider les clients
et leurs conseillers a concevoir des syste mes d approvisionnement plus approprie s et qui devraient e tre axe s
sur les priorite s particulie res de chaque projet.

Keywords: construction , project, procurement , performance , Hong Kong

The signicance of procurement system while necessary for project success , is not
systems sufcient to ensure it. For example, Rowlinson
(1988) found management, organization and con-
It has become increasingly evident over recent textual variables to be more strongly associated
years, that an appropriate choice of procurement with project performance than procurement form.

Bu i l d in g Re s e a r ch & In f o r m at io n (1998) 26(4), 223238 223


KUMARASWAMY AND DISSANAYAKA

He also found that different procurement forms necessary to dene the latter in terms of a client s
could be associated with different levels of per- particular priorities ; and identify (and allow for)
formance. While construction projects have grown the other factors and variables that may also affect
in complexity and magnitude, crossing national performance . For example, the importance of the
and natural barriers , innovative modalities of client s and designer s characteristics , in determin-
procurement have also been developed. These ing project outcomes in a given scenario, have
present prospective clients with many possible previously been highlighted (Naoum and Musta-
procurement paths within the complex network of pha, 1995; Walker, 1995); while the inuence of
potential arrangements for procuring design, human factors have not been overlooked
construction , management and nancial services. (Rwelamila and Hall, 1994). Proper communica-
For example, Build Operate Transfer arrange- tions and teamwork based on the commitment
ments in themselves can incorporate many possi- and proactive attitudes of all project participants ,
ble variations . have been conrmed as valuable in boosting
performance levels (Latham, 1994).
Inappropriate procurement strategies may lead to
cost and time overruns , claims and disputes on Innovative approaches, such as formalized but
projects as previously highlighted (Masterman, non-contractual (non-binding) partnering ar-
1992; Abdel-Meguid and Davidson, 1996). Con- rangements between the client=client s represen-
versely , appropriate procurement strategies are tatives and a contractor, may be superposed on
needed to help achieve optimal solutions in procurement systems to generate the aforemen-
terms of cost, time and quality. They can also tioned communications and positive team spirits ,
contribute positively to other aspects of perform- even within traditionally adversarial contractual
ance, such as meeting agreed targets, as outlined arrangements (Baden Hellard, 1993). However,
by Jagger (1995) and detailed (in the context of evidence from the industry and the literature
the UK construction industry) in the Latham indicates an initial need to design a suitable
Report (1994). system with appropriate frameworks that will
facilitate such teamwork and enable success.
At the macro level, broader (e.g. national)
procurement initiatives could also lead to longer- It is thus important to identify the relationships
term benets through the development and between the various parts of procurement sys-
upgrading of domestic contractors , other con- tems and different aspects of success. Particular
struction organizations and national industries attention may then be focused on certain pro-
themselves. This has been demonstrated by case curement sub-systems, and on choosing options
studies, for example in Malaysia (Abdul Aziz therein, when addressing specic client priorities
and Ofori, 1996), Singapore (Ofori, 1996) and Sri for a given project. For example, it is generally
Lanka (Kumaraswamy and Dissanayaka, 1997). suggested that a client who is looking for an
However, this paper will focus on identifying economical design in a reduced (fast-tracked)
linkages between procurement systems and per- time frame and=or single-point responsibility
formance outcomes at the project level, as is the may need to consider Design and Construct
principal concern of clients. Furthermore , since options.
industry conditions do affect project performance
levels, while also differing from country to
country, it should be noted that this paper draws In search of a suitable model
primarily on project data from Hong Kong.
However, the general conclusions will also be Skitmore and Marsden (1988) suggested a multi-
of use to other contractual regimes, given the attribute technique based on the UK National
experienced -based perceptions that have been Economic Development Ofce procurement path
captured in this research. selection decision chart (NEDO, 1985). Such
approaches allowed for choosing between differ-
ent options related to the design, management
Missing links and construction functional arrangements. How-
ever, they generally fell short of incorporating all
When looking for linkages between project pro- aspects=sub-systems of a procurement system
curement systems, and good performance it is such as work packaging and participant selection

224
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

methods. Neither did they incorporate provision the CIB W92 Working Commission on Procure-
for other variables that may interact with the sub- ment Systems as the framework within which
systems, such as project-specic: conditions (e.g. construction is brought about, acquired or ob-
difcult location ); team characteristics (e.g. lack tained (Sharif and Morledge, 1994).
of cohesion); and external factors (e.g. construc-
tion boom accompanied by materials and labour The general framework illustrated in Fig. 1 was
shortages). developed for this research exercise , in keeping
with the latter denition . The breakdown of a
It was therefore decided at the outset of this typical procurement system into ve principal
research to incorporate such additional variables sub-systems (and further sub-sub-systems there-
in the model. Linkages between all procurement from) is followed by an indication of the many
sub-systems and project performance were mod- choices to be made in assembling each sub-
elled to include the intervening interactions with system. For example, this broad overview of a
other factors such as project parameters, project procurement system thus incorporates the front-
team characteristics and external conditions. It end decisions particularly on a complex
was expected that the development of a compre- and=or high value project about the nature
hensive model in this manner, would facilitate and magnitude of different contracts or work
more informed and intelligent choices of pro- packages. Appropriate work packaging can be
curement systems. crucial to the achievable economies and efcient
management of many projects (Kumaraswamy ,
This paper presents the ndings from the rst set 1997). The principal sub-systems were thus
of data collected in Hong Kong to identify these conceptualized as follows:
links and model such relationships between
procurement sub-systems and project outcomes,
(1) Work packages
may be designed to be large enough to
while also incorporating the contributions of
attract international interest, if needed for
other factors such as project, client and other
purposes of greater price competition , or
participant characteristics , together with external
for deploying advanced technologies eco-
conditions.
nomically (for example, if special expertise
was needed to design and=or build com-
The holistic view of procurement systems
plex structures such as double-curvature
adopted in this research, coupled with broader
arch dams or tunnels in poor ground
overviews of performance and of the project
conditions ). Alternatively , the large and=or
process itself , led to more comprehensive results
complex work packages may be sliced , to
than were obtainable through many previous
keep them within the capabilities of local
studies. This is because many such studies had
construction organizations . For example,
mostly focused on identifying relationships be-
vertical and=or horizontal slicing is pos-
tween certain aspects of performance (such as
sible on a roadworks project, by dividing it
time performance) and specic procurement sub-
into different projects along the length of
systems (e.g. contract type). Focused studies of
the road and=or into separate parcels for
that nature often precluded the interactive effects
earthworks, surfacing and services, respec-
of other sub-systems and other dimensions of
tively (Kumaraswamy and Dissanayaka,
performance .
1997).
(2) Functional Groupings
Developing a general procurement of the design, construction and manage-
framework ment functions may be based on either: a
separated approach of independent design
Procurement has been dened as the action or and construction as in the traditional
process of acquiring or obtaining material, prop- method, whether within a sequential or
erty or services at the operational level (McGraw - fast-track structure; an integrated ap-
Hill, 1984). Building Procurement has been iden- proach such as in Design and Construct
tied as the amalgam of activities undertaken by or Build Operate Transfer (BOT); or a
a client to obtain a building (Franks, 1984). Management-led approach such as in
Construction Procurement has been dened by Management Contracting.

225
KUMARASWAMY AND DISSANAYAKA

CONSTRUCTION PROJECT PROCUREMENT SYSTEMS SYSTEM

TYPE OF CONTRACT

WORK FUNCTIONAL PAYMENT FORM OF SELECTION


PACKAGING GROUPING MODALITIES CONTRACT METHODOLOGIES

SUB-SYSTEMS

Contractors Joint
differentiation Consultants Project Venture
magnitude Currency Managers partners
(contract value) Timing
Valuation Method Negotiate Register
large
Prequalify Negotiate
geographical Fixed Advance 2 Open
medium
divisions Price (Yes/No) envelope Tender
small (of Cost- 2
contracts) Milestone Negotiate envelope
plus Lump
degree Sum
of over- functional Fee Monthly
lap/ divisions % BoQ
sequencing Target
fast-tracking Schedule
disciplinary of rates
divisions Mixed
General Special Other
Conditions (from Conditions Contract
Standard forms) Document
Separated Management
Integrated led FIDIC Dispute
resolution/
minimization
Design & Construction
Institute of Architects Special Arbitration
sequential Management
Build (eg: HK or USA) Risks
accelerated Mediation
Turnkey ICE (Institution
(fast Management Technology
tracked) of Civil Engineers, Transfer/ Adjudication
Contracting
UK) exchange
th
6 Ed. Dispute
BOT NEC Resolution
Government Adviser
Civil Engineering
Building Partnering
Design & Build
ACP (Airport Core
Programme) type

Notes: contains the sub-sub-systems within each sub-system


(i.e. indicates a break-down of the sub-system)

Each sub-system has series of options as indicated above


All possible sub-sub-systems and options are not indicated here

Fig. 1. Hierarchy of procurement options.

226
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

(3) Payment Modalities On the other hand, the procurement framework


could vary from Cost Reimbursement to developed in this study, as shown in Fig. 1, also
Lump Sum Fixed Price. The timing and includes a suitable selection sub-system. The
currency of payments could also vary . latter would ideally lead to the selection of the
best available design, management and con-
struction teams, which in turn should enhance
Items (2) and (3) above may be combined under performance . However, the compromises that are
Basic Contract Type, e.g. as for a Design and often reached in practice, such as in selecting the
Build Cost Reimbursement contract. less expensive teams (as often required on
public sector projects, to satisfy rigid account-
(4) Standard sets of Contract Conditions ability criteria ) could impact negatively on
(such as the FIDIC, ICE, Hong Kong potential performance levels . This could arise
Government Conditions for Civil Engineer-
from the poor performance of one or more of the
ing Works, or for Design and Build ), may
teams, from their incompatibilities (with each
be chosen for convenience or because they
other), from communication problems or even
have been tried and tested and are now
from personality clashes (Kumaraswamy , 1995).
trusted by the industry. Alternatively ,
Proposals to address the potentially poorer per-
special sets of conditions may be designed
formance of less competent contractors for
for a given programme as in the New
example, have previously been assessed by Holt
Airport Core Programme in Hong Kong, or
et al. (1993) and Hatush and Skitmore (1997).
for use by a particular large client or
Consultant.
Parts of a chosen procurement system may also
(5) Selection Methodologies turn out to be incompatible with particular
of consultants and constructors may be project goals or priorities . For example, a
open, or based on registered or pre-quali- remeasurement payment modality may not
ed short-lists. Comparisons of such prac- provide cost certainty, whereas a Design and
tices and possible alternatives are described Construct functional grouping may not be the
in (Holt et al., 1993) and (Kumaraswamy , best to evoke high quality levels .
1996). For example, different weighting
techniques can be used in balancing (as- Furthermore , a systems approach suggests the
signing weightings to) the nancial capa- criticality of interactions between the sub-systems
cities , technical capabilities and pricing (and also sub-sub-systems) that will affect the
differentials in the proposals of different output of the parent system. Such critical inter-
bidders, depending on the project require- faces suggest that the selected options from
ments. within each sub-system (or sub-sub-system) must
at least be compatible, while at best they could
be synergistic, when assembled into the total
procurement system. Since the latter is itself an
Selecting appropriate options
open system interacting with many other project
The selection of suitable options from each sub- and environmental systems, it is reasonable to
system (or sub-sub-system) and the development conclude that the most appropriate procurement
of an appropriate procurement system would be system must necessarily depend on the project
valuable to construction clients in increasing their scenario or project prole that can be derived
chances of project success in a particular project from: (1) the contextual conditions, including the
scenario. However, other factors such as the client s characteristics , project parameters and
competencies and performance of the various external factors relating to the project; and (2)
interacting project teams, as well as external the client s needs and priorities , including the
factors will also affect project outcomes. For desired risk allocation among participants .
example, adverse weather or market conditions,
or client-initiated design changes may impact on
Contextual conditions
such targets. If the latter are not adjusted, such an
adverse scenario could lead to poor performance, Conditions and constraints that bear on the
even if there had been an optimal choice of project and its environment could individually
procurement system at the outset. or collectively inuence the choice of procurement

227
KUMARASWAMY AND DISSANAYAKA

system. For example, relatively low value and less Desired risk allocation
complex building projects for a reliable client
The choices of types of contract ( functional
(who from past experiences or reputation may be
groupings and payment modalities ) and con-
relied upon to full their obligations ) may not
tract conditions have been observed (through
warrant sophisticated contractual arrangements to
interviews of practitioners, as well as by the
get a reasonable job done. But special project,
authors own experiences) to be largely based on
participant or environmental conditions and cri-
the desired allocation of risks between the differ-
tical priorities may warrant special consideration
ent groups=teams (e.g. clients, consultants , con-
and specic measures. Examples of conditions to
tractors , nominated sub-contractors , domestic sub-
be considered when proceeding through the steps
contractors , suppliers , insurers). For any project, it
of procurement path selection include:
is essential that each group=team be fully aware
of the risks undertaken and=or expected of them,
Type of project (e.g. housing estate, road, so that they can provide for these in terms of
dam, ofce building refurbishment). proper personnel, pricing and precautions.
Size of project (e.g. value ; number of stories;
Risk management has been said to involve the
oor area; km of road).
steps of: (1) risk identication ; (2) risk analysis ;
Complexity of project (e.g. special ground and (3) risk response. Risk identication ideally
conditions or technology requirements). models all the uncertainty and risk elements of a
particular project; while risk analysis investi-
Type of client (e.g. public=private=mixed;
gates the effects of such risk elements on project
experienced=one-off; project staff calibre and
viability ; for example by assessing the probabil-
their strengths, weaknesses and management
ities, the quantiable impacts (in terms of cost,
style).
time or both) and the sensitivities of the project
Availability of information at project incep- outcomes to such impacts. The risk response
tion and points at which any remaining then depends on the client s capacities and
information will be required=be available. priorities . The responses have been classied as:
Nature and status of local construction risk avoidance or reduction, e.g. by redesign;
industry, including available capacities of
risk transfer, e.g. by contracting out that
potential project participants, scarcity of
responsibility to some other project partici-
work in particular elds, competitiveness ,
pant, such as the designer, the contractor, the
etc.
supplier, the sub-contractor or the insurer; or
Availability of materials and equipment that
risk retention; where the residual risk which
are required for the works.
is perhaps unquantiable or too expensive to
Performance of available contractors and economically transfer is retained by the
consultants on previous (similar) projects in client.
the area in terms of meeting cost, quality
In an ideal world and for overall economies or net
and time targets; safety records and client
savings, each risk element should be transferred
satisfaction levels: as compared with the
to those project participants best equipped to
procurement modalities used.
handle them. But a client (or any other party)
Local familiarity and condence in=disillu - may exercise their preference to transfer out some
sionment with, particular types of procure- element of risk and pay a premium in exchange
ment with reasons (as perhaps veriable from for a greater certainty in protection against that
local professional institutions and techno- aspect. The risk allocation preferences may thus
commercial networks ). be considered along with the other client s needs
and priorities .
Effects of relevant political , legal and eco-
nomic systems, including market conditions.
Specic location , special weather and envir- Clients needs and priorities
onmental concerns.
Table 1(a) was formulated by assembling sets of
Any other special conditions . possible priorities of different types of construc-

228
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

Table 1(a). Rating prospective procurement systems against client priorities


Possible client priorities Priority weighting System ratings
(w) (0 to 10)
A B C

Ra wRa Rb wRb Rc wRc

1. Level of design competition


2. Level of price competition
3. Economy
4. Value for money
5. Life cycle costs
6. Cost certainty
7. Speed
8. Time certainty
9. Urgency to complete project
10. Urgency to commence construction
11. Importance of intermediate milestones
12. Aesthetic value
13. Durability
14. Innovations
15. Quality assurance
16. Construction risks allocation
17. Design risks allocation
18. Financial risks allocation
19. Other risks allocation
20. Need for mid-project design changes= exibility
21. Need to be kept informed
22. Need to be involved
23. Need to assign single-point responsibility
24. Need to delegate decision-making
25. Desire for good communications
26. Health and Safety concerns during construction
27. Health and Safety concerns during operations
28. Importance of planning
29. Importance of controls
30. Technology transfer=exchange
31. Technology innovations
32. Operational guarantees
33. Design life certainty
34. Maintainability
35. Constructability
36. Minimal environmental impacts
37. Disputes (and claims) minimization
38. Others
Total of weighted ratings for each system
Key: A, B, C, . . . etc.: shortlisted procurement systems that would suit contextual conditions w weighting; R Rating of
each prospective system (e.g. Ra for system A), say from 0 to 10 (so wRa is the weighted rating for system A against the
corresponding priority).

tion clients in the rst column. A provision was systems) as in Fig. 1. It is therefore proposed that
made (in the second column) to add weightings the following steps may be taken in selecting a
for identied relative importance levels of each of procurement system to suit a given project
such priorities on any given project. Different scenario:
procurement frameworks (A, B, C, . . .) may then
be assembled from appropriate options as chosen (1) Eliminate options within the general pro-
from the different sub-systems (and sub-sub- curement framework (as in Fig. 1) that are

229
KUMARASWAMY AND DISSANAYAKA

clearly incompatible with the foregoing external and internal compatibilities [al-
contextual conditions . ready checked once in steps (1) and (3)].
(2) Identify options which appear desirable at
rst sight, in achieving the desired risk While simpler versions of similar scoring systems
allocations and project priorities . using such matrices have been suggested pre-
viously (Franks, 1984; NEDO, 1985; Skitmore and
(3) Prepare a shortlist of up to about six
Marsden, 1988; Love and Skitmore, 1996), the
potentially suitable (prospective ) procure-
presently proposed approach is expected to be
ment systems by assembling a few separate
more comprehensive, in for example, incorporat-
sets of mutually compatible options each
ing an initial consideration of contextual condi-
set of which contains one appropriate
tions , together with wider ranges of priorities and
option from each sub-system or sub-sub-
options.
system as in Fig. 1.
(4) Assign ratings (R) to each such potentially Table 1(b) was next developed to demonstrate a
suitable (prospective ) system against the specic (although hypothetical) example of the
chosen priorities in Table 1(a), i.e. based on application of the general scoring system indi-
how well they are expected to perform in cated in Table 1(a). A set of ten important client
meeting that priority , say on a scale from 0 priorities for the hypothetical project was chosen,
to 10 (or even extend to a negative scale as indicated in the rst column of Table 1(b). The
where they are clearly counter-productive other possible client priorities were taken to be
in respect of that priority ). negligible (of 0 value ) in this example. Priority
weightings were assigned as shown. In practice
(5) Assign weightings (w) to the priorities
they would be assigned after an iterative process
themselves also on a scale, say, from 0 to
involving the main stakeholders, such as client s
10 (those with 0 weighting being totally
key personnel and main end users.
irrelevant in this project scenario).
(6) Compute the total sum of the weighted In this example, three potentially suitable pro-
ratings for each prospective procurement curement systems A , B , and C were as-
system. Propose the system with the high- sembled as follows, by combining compatible
est total and if necessary re-check for options to also suit project conditions. (Note:

Table 1(b). Hypothetical demonstration of a comparison of three shortlisted (potentially suitable) procurement
systems for a particular project
Possible client priorities Priority weighting System ratings
(w) (0 to 10)
A B C

Ra wRa Rb wRb Rc wRc

1. Level of design competition 6 6 36 9 54 7 42


2. Level of price competition 6 8 48 9 54 9 54
3. Economy 7 4 28 8 56 7 49
4. Life cycle costs 7 8 56 3 21 7 49
5. Cost certainty 8 4 32 9 72 8 64
6. Speed 8 4 32 7 56 8 64
7. Aesthetic value 8 7 56 5 40 8 64
8. Need for mid-project design changes= exibility 7 5 35 2 14 9 63
9. Need for Single-point responsibility 6 2 12 8 48 6 36
10. Disputes (and claims) minimization 5 2 10 7 35 7 35
Total of weighted ratings for each system: 345 450 520
The above ratings and weightings are hypothetical. They would also vary with the project and external condition, as well with
client priorities. Descriptions of the shortlisted systems A , B and C are provided in the text.
Key: w weighting; R Rating of each prospective system (e.g. Ra for system A ), say from 0 to 10 (so wRa is the weighted
rating for system A against the corresponding priority).

230
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

examples of possible options for each sub-system and ratings of the procurement systems consid-
are indicated in Fig. 1). ered are also hypothetical.

A : Traditional separated (design and con-


struction functions), sequential; re-meas-
urement with monthly payments; tenders Modelling linkages between
obtained from prequalied tenderers for procurement systems and project
construction , and consultants selected on a performance
2 envelope ( technical and fee proposals
sealed in separate envelopes ) system; ICE
Structuring the linkages
(Institution of Civil Engineers) Conditions
of Contract. The foregoing recommended strategy for selecting
an appropriate project-specic procurement sys-
B : Design & Build; guaranteed maximum
tem, presupposes an appreciation of the compat-
price (xed price lump sum) with pay-
ibilities (or otherwise) of the various options
ments at milestones; tenderers selected
with the contextual conditions of the project and
from a prequalied shortlist; Hong Kong
with each other. For example, certain options may
Government Conditions of Contract for
be more (or less) desirable depending on
Design & Build projects.
particular client characteristics including those of
C : Construction Project Manager, overseeing the principal decision-makers=advisers, the speci-
both design and construction teams for a c project parameters and=or external conditions .
fee; work divided into a series of pack- Project parameters would include project size ,
ages=contracts (to aid fast-tracking ): specic location , construction type and complex-
mostly xed price remeasure, but some ity; while external conditions would include the
lump sum contracts; selection of project market, legal, political , weather and other such
manager and designers on 2 envelope conditions impacting on the given project. The
systems; selection of contractors by project model (Fig. 2) incorporates the inputs from such
managers in a two-stage system (prequa- contributors to the selection of the most appro-
lifying and receiving tenders from) specia- priate procurement system for a given scenario.
list contractors and sub-contractors for
each work package; NEC (New Engineer- Even though the chosen procurement system
ing Contract) Conditions. includes a sub-system that should assist in the
selection of appropriate design, management and
The sub-systems and systems would be devel- construction teams, the performance of such teams
oped in practice so that they would make the best can hardly ever be predicted precisely. Project
use of opportunities provided by (and counteract outcomes or performance levels will thus be
the threats posed by) the contextual conditions affected by the performance of the project partici-
surrounding the project, as discussed in a pant teams, as also represented in the model (Fig.
previous sub-section. 2). This model itself has been structured on the
basis of linkages as identied from the literature ,
Ratings can then be assigned for each shortlisted the experiences of the authors and of four semi-
system in turn, against each chosen priority, by a structured pilot interviews in Hong Kong (of three
set of experts who are experienced in the senior industry practitioners and one researcher
management of such procurement systems. Some with particular knowledge of this eld).
possible ratings are indicated in Table 1(b). The
ratings are weighted by the corresponding
Evaluating the linkages
priority weightings. The totals of the weighted
ratings are then aggregated as indicated, upon Next, it was necessary to evaluate the relative
which proposed system C appears to be the strengths of the various linkages, so as to increase
most appropriate under the assumed conditions, the condence levels when assessing: (1) the
weightings and ratings in this hypothetical relative impacts of the different contributors to
example. It should be noted that Table 1(b) has project outcomes as in Fig. 2; and (2) the ratings
been assembled for demonstration purposes only that may be assigned to prospective systems as in
and the indicated weightings of client priorities Table 1; when choosing a procurement system for

231
KUMARASWAMY AND DISSANAYAKA

INITIAL CHANGED

(BOTH INITIAL AND CHANGED)

DESIGN TEAM
PROJECT * PERFORMANCE
PARAMETERS

PROCUREMENT E
E
SYSTEM X
X PROJECT
(INCLUDING WORK T
T * OUTCOMES
PACKAGING, E
E MANAGEMENT TEAM
FUNCTIONAL GROUPING R
R PERFORMANCE
PAYMENT MODALITIES N
N
SELECTION METHODOLOGY A
A
AND CONTRACT L
L CLIENT CONDITIONS)
CHARACTERISTICS,
OBJECTIVES AND
PRIORITIES.
TYPE OF DECISION *
CONTRACTOR
MAKERS/ADVISERS PERFORMANCE

(BOTH INITIAL AND CHANGED)

CONDITIONS CONDITIONS
NOTES:-

MANAGEMENT TEAM 5 EITHER INDEPENDENT PROJECT MANAGER OR


CONSULTANT WHO IS MANAGING PROJECT FOR CLIENT * OTHER INPUTS/RELATIONSHIPS

DECISION MAKERS/ 5 CLIENTS DECISION MAKERS OR PROJECT MANAGERS/


ADVISORS CONSULTANTS WHO ADVISE CLIENT

Fig. 2. Model of basic linkages between procurement system and project outcomes.

a given project. Such evaluations were expected to It was decided to focus on building projects for
assist in modelling the relationships so as to this questionnaire survey, so that the specic
provide a better knowledge-base for clients and project data sets obtained would be more
their advisers who need to decide on appropriate comparable . Thirty questionnaires were directed
procurement systems. to selected senior industry personnel, who were
personally contacted by the authors. In addition,
The second phase of this research project thus 225 questionnaires were sent to organizations
commenced with the development of a compre- selected randomly from available databases of
hensive questionnaire. This contained two parts clients (e.g. developers, government organiza-
with questions on general accumulated experi- tions ), consultants and contractors. Table 2
ences over many projects, and specic informa- indicates a breakdown of the groups to which
tion from a recent project. The former queried questionnaires were sent in mid-1996 and the
general perceptions as to, for example, the pattern of responses therefrom. The considerably
strengths and weaknesses of different procure- lower response rate from the random sample was
ment sub-systems and options therein, with not entirely unexpected given the 10 page length
particular reference to their effects in inuencing and detailed nature of the questionnaire; hence
certain project outcomes. The latter sought spe- the decision to approach a selected sample in
cic data on a particular building project in terms parallel. Follow -up interviews were conducted
of, for example, project and participant character- with some of the respondents to clarify some
istics, performance outcomes and perceptions as responses and perform cross -checks. These inter-
to the inuence of particular procurement sub- views were intended to check on any apparent
systems on such outcomes. While the information misunderstandings or biases; and also to probe
sought in the rst part was necessarily qualitative , any perceived trends, for example.
elements of quantication were introduced by
Likert type ratings requested in response to some
Observations from the survey
questions. The information sought in the project-
specic second part of the questionnaire was both Although only 37 completed questionnaires were
quantitative and qualitative. received in total, each of these contained 10 pages

232
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

Table 2. Composition of the surveyed samples


Target group Selected sample Random sample

Dispatched Received Interviewed Dispatched Received

Clients=Project Managers 17 11 (65%) 8 45 2 (4%)


Consulting Engineers 7 5 (71%) 3 40 3 (8%)
Architects 0 0 0 35 0 (0%)
Contractors 5 4 (80%) 3 85 12 (14%)
Quantity Surveyors 1 0 (0%) 20 0 (0%)
TOTAL 30 20 (67%) 14 225 17 (8%)

of useful data and some of the respondents were marginally more important than others, perhaps
interviewed as indicated above. A summary of depending again on the conditions and priorities
the relevant results from the primary analysis of of a given project.
this data is presented in this paper. Detailed
results and derived relationships are expected to The relative importance (or Relative Importance
be presented in a subsequent paper, following Index as described by Kometa et al., 1994) and
further data collection and analysis that will focus sumarized below; RI p of the pth sub-system
on the key areas identied in this primary expressed as:
analysis .
N
X
ri
For example, one of the questions required i1
respondents to assign a rating (ranging on a RI p (1)
N3 H
scale from 0 [not signicant] to 4 [extremely
signicant]) according to their perception of where ri rating given by the ith respondents
relative importance of each of the different (ranging from 0 to 4) 0 is not signicant and 4
procurement sub-systems. The relative impor- is extremely signicant; H highest rating (4 in
tance of each sub-system was then compiled on this case); N total number of respondents.
a scale from 0 to 1 as in Equation 1. The relative
signicance percentage of each sub-system was The above Relative Importance was then con-
next computed as in Equation 2, so that the verted to percentage terms as a Relative Sig-
summations totalled 100 in each case, for nicance Percentage. The Relative Signicance
convenient comparison as relative percentages. Percentage ( RS p ) of the pth sub-system is ex-
The averages of responses within each group (i.e. pressed as:
separately for the 21 clients=their representative
project managers or consultants and for the 16 ( RI ) p
RS p 3 100 (2)
contractors ), were computed to yield a prole as K
X
in Table 3. This suggests that none of the sub- ( RI ) p
systems can be neglected, while some may be p 1

Table 3. Perceived relative signi cance of procurement sub-systems


Sub-system Relative signi cance percentages (and ranks)

Clients Contractors

Work packaging 21.0 (2) 22.5 (1)


Functional grouping 19.4 (3) 21.2 (3)
Payment modality 16.8 (5) 15.8 (5)
Selection methodology 19.4 (3) 18.5 (4)
Contract conditions 23.4 (1) 22.0 (2)
Relative percentages, i.e. totalling 100.
Clients include their representatives, e.g. consultants=project managers.

233
KUMARASWAMY AND DISSANAYAKA

N
where K number of sub-systems (5 in this case) X
and the other variables are as in Equation 1. (j Ril Ri2 j)
i 1
or PD
N
3 100
Little disagreement was noted between the two X
(j Ri1 R j2 j)
groups (i.e. clients and contractors) based on the i1
foregoing Relative Signicance values. The rela-
tive rankings were also compared using a Rank The Percentage Agreement (PA) is then obtained
Agreement Factor (RAF) as computed according from:
to the procedure described by Okpala and PA 100 PD
Aniekwu (1988) and summarized below. The
RAF was found to be 0.60 and the Percentage Table 4 indicates the quantied collective re-
Agreement related to the above rankings was sponses of the client and contractor groups to
then quantied as 75.0 (using the formula another set of general (non-project-specic) ques-
below). This indicated a reasonable, degree of tions related to the perceived inuences of differ-
agreement between the groups as to the relative ent functional groupings (such as design &
importance of the sub-systems. construct or traditional=separated groupings of
the design, construction and management func-
In general, for any two groups, if the ranks of tions ) on specic performance outcomes, namely
the ith item in Groups 1 and 2 are Ri1 and Ri2 , in meeting time, cost and quality targets. These
respectively ; and if N is the total number of responses were obtained on a scale ranging from
items; then the Rank Agreement Factor (RAF) is relatively low (1) and average (3) to relatively
expressed as: high (5), in terms of their positive impact in
achieving either time, cost or quality targets. Each
N
! response was weighted by the experience of the
X
j Ri1 Ri2 j respondent in working with such a specic
RAF
i1 functional grouping on a scale that assigned 0.2
N for no personal experience in that grouping, 0.6
for moderate experience, and 1.0 for substantial
experience in that particular grouping.
The maximum possible RAF ( RAFmax ) could also
be computed as follows , assuming that when the
The individual ratings were thus weighted by
rank of the ith item in Group 1 is i, then the rank
either 0.2, 0.6 or 1.0 and combined by adding the
of the corresponding item in Group 2 is j, where
weighted ratings and dividing by the maximum
j N i 1 (i.e. ranked at the other extreme)
possible score in each case. Each gure was
! multiplied by 100 to convert it to a percentage.
N
X The resulting weighted averages (which could
j Ri1 R j2 j
range from 0 to 100) are compared in Table 4.
i1
RAFmax
N
The Table 4 summary suggests that there are no
very strong perceptions as to the superiority of
The Percentage Disagreement (PD) is then ex- any one functional grouping in meeting specic
pressed as 100 3 RAF=RAFmax : targets, although the response patterns are inter-

Table 4. Perceived positive impact of different functional groupings on principal performance outcomes
Functional grouping Perceived impact on time (T), cost (C) and quality (Q) outcomes

Clients (average perception) Contractors (average perception)

T C Q T C Q

Traditional separated 62.2 67.4 70.0 61.0 57.4 65.1


Design & Construct 74.7 66.5 44.5 86.4 60.9 63.4
Management Contracting 61.9 63.2 65.4 77.4 70.4 79.1
Construction Management 65.6 65.6 66.7 73.1 65.5 73.8

234
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

esting. For example, they reinforce the perception groups (computed as described previously
of achieving better quality through traditional= in this section in relation to Table 3);
separated or management type groupings;
(3) the numbers of specied variables initially
better construction time performance through
postulated by the authors, so as to collec-
design and construct; and also interesting
tively represent each such attribute ade-
(although not always considerable) differences quately; for example variables such as (a)
between the average perceptions of clients and
client involvement in planning; and (b)
contractors . A general conclusion that contributes
client s speed of decision making under
to the subsequent study is that the functional
the client characteristics attribute;
grouping sub-system alone is unlikely to sig-
nicantly determine project success against any (4) the numbers of the foregoing variables that
of the above criteria, thus reinforcing the need to were found to be suitable for factor analy-
investigate the contributions of other sub-systems sis on the basis of initial statistical tests
in the procurement system, as well as the including the Kaiser Meyer Olkin s mea-
contributions of the intervening variables , such sure of sampling adequacy (. 0.5) and the
as project and team characteristics . Bartlett test of sphericity (at a 0.05 sig-
nicance level) as described in the SPSS
The project-specic questions in the questionnaire software manual (Norusis, 1990).
covered a wide range of items as deemed useful
(5) the numbers of factors which were
for modelling the relationships in Fig. 2. For
derived to represent the foregoing corre-
example, Table 5 summarizes the initial results of
lated variables [in item=column (4)] more
the analysis of the group of attributes (as listed
concisely for purposes of multiple re-
in the rst column of the table), in respect of the
gression analysis .
sub-sample of traditional separated type con-
tracts. Specically, while the sixth column in The foregoing factors [in column (5)] were
Table 5 contains the variables (including factors derived in a data-reduction process , in order to
described below) as nally derived to represent reduce the large numbers of variables that were
these attributes, the rst ve columns indicate: initially postulated, to a more manageable num-
ber. [It would be noted from column (1) that 65
(1) the attributes postulated as inuencing
variables were initially postulated.] The factor
project performance outcomes, in addition
analysis technique was employed for this data
to procurement system and team perfor-
reduction using the SPSS statistical software
mances;
(Norusis, 1990). Factor analysis itself, has been
(2) the percentage agreement on the impor- found to be a useful statistical technique in
tance rankings of such postulated variables representing a large number of variables in terms
between the Client and Contractor of a smaller number of hypothetical variables or

Table 5. Sets of variables and representative factors (derived from the traditional separated functional grouping
type projects)
Attribute Percentage Number of Number of correlated No. of representative Number of factors=
(1) agreement on variables variables used for factors extracted by variables used for
relative postulated factor analysis factor analysis multiple regression
importance (3) (4) (5) (6)
(2)

Client characteristics 71 29 16 6 19
Contractor characteristics 69 20 14 6 12
Project parameters 100 5 5 3 3
Design team 67 5 3 2 3
characteristics
External factors 89 6 naa naa 1a
a
na not applicable, as an analysis of the responses as to the relative importance of the six variables chosen to represent
external factors revealed very little internal differences. In view of this apparently close correlation, just one factor (as an
average of the six variables) was selected to represent them conveniently.

235
KUMARASWAMY AND DISSANAYAKA

factors. The main advantage of factor analysis is in the multiple regression 0.44 in the time
that a smaller set of uncorrelated factors is much overrun relationship and 0.23 in the cost overrun
easier to understand and use in further analysis , relationship . The low predictive capabilities of
than a larger set of correlated variables . In this the foregoing attributes, in explaining a major
case, the factors which accounted for 80% of the component of the time and cost overruns have
total variance (in the attributes) were extracted conrmed the need to seek alternative explana-
using the strategy and procedures described by tions for the major part of these performance
Lewis-Beck (1994). shortfalls . For example, based on the foregoing
results the authors identied the need to focus
The derived factors [as shown in column (5)], the next stage of research on assessing the
were used as independent variables in the impacts of (1) the different procurement sub-
subsequent regression analysis, together with systems compared to (2) the performance levels
the residual variables [obtained from column attained by different teams. The latter can
(3) column (4)], i.e. those that were not found arguably also be related back to improving
to be suitably correlated (and therefore not prediction capabilities (relating to potential per-
chosen for the factor analysis ). formance) during selection of the different teams,
since this in itself relates to one of the ve
For example, in the case of the rst attribute of identied sub-systems in this study. More data
client characteristics : 16 of the 29 initially from a larger project sample would also be
postulated variables were found to be correlated . useful.
Six representative factors were derived by factor
analysis from the 16 correlated variables. These
were combined with the 13 (i.e. 29 16) un- Conclusions arising from this survey
correlated variables, to yield 19 representative
independent variables (including the derived The hierarchy of procurement options , as devel-
factors ) for the regression analysis as shown oped in this study, is a useful representation of
in column (6). the choices facing contemporary construction
clients and their advisers ; while the proposed
As described in the previous paragraph, the model of basic linkages between procurement
representative factors [in column (5)] together systems and project outcomes is a viable and
with the residual independent variables [column valuable framework for deriving the signicant
(3) column (4)], were nally taken [as in relationships . The proposed matrix of compari-
column (5)] to represent each of the correspond- sons between a range of potential project prio-
ing attributes. Multiple regression exercises were rities and a set of shortlisted prospective
next carried out with the objective of discerning procurement systems is another useful tool devel-
any signicant underlying relationships between oped to help in gearing the procurement system
these attributes and each performance outcome to suit the project-specic priorities . These models
in turn. If signicant relationships were dis- which have been developed on the basis of the
cerned in these cases, it would be possible to literature review , the experiences of the authors
both (a) predict time (or cost ) overruns on and the initial interviews in this research exercise ,
projects where the signicant variables were provided the basis for the initial survey in Hong
known in advance and (b) control these variables Kong, that was described herein.
to reduce the potential for such time (or cost)
overruns . It may also be noted that multiple The results from this Hong Kong-based survey
regression was chosen as a suitable technique in indicate: (1) the perceived general importance of
view of the many variables that were still needed all ve sub-systems of procurement (as depicted
to represent each attribute. in Fig. 1); (2) the absence of any strong
perception of the superiority of a particular
For example, the initial exercise, sought to functional grouping in meeting principal project
discern relationships between the foregoing ve targets in general; (3) the general agreement
sets of attributes and time overruns and cost between client and contractor groups on the
overruns respectively , in turn. However, the foregoing two aspects; (4) the collective percep-
results from these two initial exercises indicated tions of the lack of a predominant impact on cost
low R-square (coefcient of correlation) values and time overruns, of client, contractor and

236
LINKING PROCUREMENT SYSTEMS TO PROJECT PRIORITIES

design team characteristics , project parameters Holt, G.D., Olomolaiye, P.O. and Harris, F. (1993)
and external conditions; and (5) the consequential A conceptual alternative to current tendering
need to analyse the impacts on project outcomes practice, Building Research and Information, 21(3),
167 72.
of the performance levels of each team, the Jagger, D. (1995) Editor s Introduction, Journal of
inuence of the corresponding team selection Construction Procurement, 2(1), 83 6.
sub-systems and of the other procurement sub- Kometa, S.T., Olomolaiye, P.O. and Harris, F.C. (1994)
systems themselves, in more depth. Attributes of U.K. construction clients inuencing
project consultants performance, Construction Man-
agement and Economics, 12(5), 433 43.
It was conrmed that the detailed quantitative Kumaraswamy, M.M. (1995) Synergistic strategies for
analysis of the data obtained from the survey, mega-projects, NICMAR Journal of Construction
should be supplemented by: further interviews Management, India, X(V1), 163 75.
with experts and experienced practitioners , to Kumaraswamy, M.M. (1996) Contractor evaluation and
probe specic aspects emerging as signicant in selection a Hong Kong perspective, Building and
Environment Journal, 31(3), 273 82.
different projects; and specic case studies in the Kumaraswamy, M.M. (1997) Repackaging construction
next stage of the research. The problems encoun- megaprojects and redening technology transfer,
tered in making sense of the multitude of Information Technology Support for Construction
interactive contributors to performance outcomes Process Reengineering, CIB W78 Conference, Cairns,
in each project, were in fact anticipated and the Australia, July 1997, Proceedings, pp. 215 24.
Kumaraswamy, M.M. and Dissanayaka, S.M. (1997)
consequential blend of quantitative and qualita- Synergising construction research with industry
tive approaches that have been designed to development, First International Conference on Con-
discern the underlying patterns have yielded struction Industry Development, Singapore, Decem-
interesting interim results that encourage the ber, Vol. 1, pp. 182 189.
researchers to probe deeper. Latham, M. (1994) Constructing the Team, HMSO,
London.
Lewis-Beck, M.S. (ed.) (1994) Factor Analysis and Related
Comparative studies in other contractual regimes Techniques, Sage, Singapore.
will also help to validate the general relation - Love, P.E.D. and Skitmore, M. (1996) Construction
ships identied in the model although the project delivery systems: an analysis of selection
relative strengths may vary with differing condi- criteria weighting, CIB W92 North meets South
Procurement Systems Symposium Proceedings , Dur-
tions. The validated models in each contractual ban, South Africa, R.G. Taylor (ed.), pp. 329 42.
regime would then assist clients and their Masterman, J.W.E. (1992) An Introduction to Building
advisers in selecting procurement systems that Procurement Systems, E & FN Spon, London.
will be geared to their particular project priorities McGraw-Hill (1984) Dictionary of Scientic and Technical
after incorporating allowances for the contextual Terms, 3rd edition.
Naoum, S.G. and Mustapha, F.H. (1995) Relationship
conditions of a particular scenario. This should in between the building team, procurement methods
turn lead to higher performance levels in and project performance, Journal of Construction
achieving desired project outcomes. Procurement, 1(1), 50 63.
NEDO (1985) Thinking about Building, National Econom-
ic Development Ofce, Building EDC, HMSO.
Norusis, M.J. (1990) SPSS Base System Users Guide, SPSS
References Inc., Chicago.
Ofori, G. (1996) Linking project procurement to con-
Abdel-Meguid, T.A. and Davidson, C.H. (1996) Man- struction industry development: the case of Singa-
aged claims procurement strategy (MCPS): a pore, CIB W92 North meets South Procurement
preventive approach, CIB W92 North meets South Systems Symposium Proceedings, Durban, South
Procurement Systems Symposium Proceedings, Dur- Africa, R.G. Taylor (ed.), pp. 473 82.
ban, South Africa, R.G. Taylor (ed.), pp. 11 20. Okpala, D.C. and Aniekwu, A.N. (1988) Causes of high
Abdul Aziz, A. and Ofori, G. (1996) Developing world- costs of construction in Nigeria, Journal of Construc-
beating contractors through procurement policies: tion Engineering and Management, 114(2), 233 44.
the case of malaysia, CIB W92 North meets South Rowlinson, S.M. (1988) An Analysis of Factors affecting
Procurement Systems Symposium Proceedings, Dur- Project Performance, PhD thesis Brunel University,
ban, South Africa, R.G. Taylor (ed.), pp. 1 10. UK.
Baden Hellard, R. (1993) Project Partnering, Thomas Rwelamila, P.D. and Hall, K.A. (1994) An inadequate
Telford, London. traditional procurement system? Where do we go
Franks, J. (1984) Building Procurement Systems, The from here?, CIB W92 East meets West Procurement
Chartered Institute of Building, UK. Systems Symposium Proceedings, Hong Kong, S.
Hatush, Z. and Skitmore, M. (1997) Criteria for Rowlinson (ed.), CIB Publication No. 175, pp.
contractor selection, Construction Management & 107 114.
Economics, 15(1), 19 38. Sharif, A. and Morledge, R. (1994) A functional

237
KUMARASWAMY AND DISSANAYAKA

approach to modelling procurement systems inter- procurement system?, Construction Management &
nationally and the identication of necessary Economics, 6(1), 71 89.
support frameworks, East Meets West CIB W92 Walker, D. (1995) The inuence of client and project
Conference, Hong Kong, CIB Publication 175, pp. team relationships upon construction time per-
295 305. formance, Journal of Construction Procurement , 1(1),
Skitmore, R.M. and Marsden, D.E. (1988) Which 4 20.

238

You might also like