Page 1
A cost of quality analysis of a building project: towards a complete methodology
by
Mark Hall* and Cyril Tomkins
University of Bath School of Management Working Paper Series
2000.01
*Communications to be sent to Dr Mark Hall, Agile Construction Initiative,
School of Management, University of Bath,
BATH, BA2 7AY, U.K. e-mail: [email protected]
Page 2
University of Bath School of Management Working Paper Series
University of Bath School of Management Claverton Down
Bath BA2 7AY
United Kingdom Tel: +44 1225 826742 Fax: +44 1225 826473
http://www.bath.ac.uk/management
2000.01 Mark Hall and Cyril Tomkins
A cost of quality analysis of a building project: towards a complete methodology
2000.02 Bruce A. Rayton The Residual Claim of Rank and File Employees
2000.03 Bruce A. Rayton Firm Performance and Compensation Structure: Performance Elasticities of Average Employee Compensation
2000.04 Stephen Brammer Endogenous Fixed Costs, Integer Effects and Corporate Performance
2000.05 Felicia Fai and Nicholas von Tunzelmann
Scale and Scope in Technology: Large Firms 1930/1990
2000.06 Areti Krepapa Market Orientation and Customer Satisfaction in the Service Dyad
2000.07 Areti Krepapa Interpreting to Learn: Theory and Propositions
2000.08 Felicia Fai and Nicholas von Tunzelmann
Industry-specific Competencies and Converging Technological Systems: Evidence from Patents
2000.09 Phillip J. McKnight and Cyril Tomkins
How much do CEOs gain from stock options when share prices change?
Page 3
A cost of quality analysis of a building project: towards a
complete methodology
A number of studies have been published that claim to carry out cost of quality studies on construction projects.
These studies, however, have largely ignored the contribution of prevention and appraisal costs to COQ and
have limited their analysis to the effect for the main contractor. This paper presents a methodology for assessing
the ‘complete’ COQ for construction projects and reports on the findings of a building project in the UK on
which the methodology was piloted. The company that applied this approach has now extended it to other
projects.
Key words: quality management; cost of quality; cost of non-conformance; building
Introduction
The cost of quality (COQ) approach to the measurement, management and control of quality defects in the
production process is well established in manufacturing and service industries generally, having been in place in
its current form for five decades (Kumar and Brittain, 1995). It forms part of a collection of management
methods, which have been introduced to industries around the world, and is related to, and forms a subset of,
total quality management (TQM) (Tomkins et al, 1997).
Although the COQ technique has had a mixed response and is far from being universally applied, notable
successes in its application can be found across several of industries. These range over an early classic
application in Texas Instruments (Ittner, 1988), applications in manufacturing (see Dale and Plunkett, 1991 for a
few examples ), aspects of medicine (inter alia Robert et al, 2000 and Fernandes et al, 1997), the food industry
(inter alia Chase, 1998 and Stein and Miscikowski, 1999) and ‘high-hazard’ industries (Carrol, 1998).
Moreover., with an increasing interest in ISO 9000 certification, Kumar and Brittain (1995) considered that
COQ and related activities would gain more prominence across industrial activity generally.
One industry in which the potential of COQ is beginning to be recognised is construction. A number of studies
have been conducted in the USA (Burati et al, 1992; Davis et al, 1989), in Autralia (Love and Li, 1999; Love et
Page 4
1
al, 1999; Love and Li, 2000a; Love et al, 2000; Love and Li, 2000b), in Scandinavia (Nylen, 1999) and in the
UK (Abdul-Rahman et al, 1996; Barber et al, 2000). However, all previous studies of the implementation of
COQ in construction have been only partial. They have focused on rework in the construction process (and in
isolated instances, also on design) but ignored the far broader scope entailed in a complete COQ methodology.
The study reported in this paper sought to address this deficiency by interpreting the COQ methodology in far
broader terms and attempting to apply the complete methodology to a construction project. As far as we can
ascertain, this is the first time this has been attempted. Further, the study sought to adopt an overt supply-chain
perspective, aimed at achieving an optimal, integrated framework based on COQ (Yasin et al, 1999). This
perspective has also been largely ignored or poorly addressed in previous studies in construction.
Conceptual basis of COQ
COQ derives from the philosophy, espoused by Philip Crosby and others; that “quality is free” because it is the
lack of quality that increases costs (Crosby, 1979). Joseph Juran, defined the cost of poor quality as “those costs
that would disappear if our products and processes were perfect” (Juran, 1988). Estimates of the cost of quality
(or, more accurately, the cost of poor quality or non-conformance with specification) vary across industries and
between companies. In general, unless focused efforts are taken to minimise them, they are estimated to fall
between 10% and 30%, with most analyses putting them at around 20% (Atkinson et el., 1991; Nylen, 1999).
The COQ methodology is laid out broadly in BS 6143 – Parts 1 & 2 (British Standards Institute (BSI), 1990;
1992). These documents introduce the process cost model and prevention, appraisal and failures (PAF) model.
They stress the link between cost and quality – that it is of little use to achieve the required quality at a cost that
is prohibitively high and uncompetitive. Equally, achieving a competitive cost by degrading quality is also
inappropriate. The Lundvall-Juran curve (Figure 1) illustrates the traditional theory underlying COQ and shows
how cost and quality operate as a trade-off within the COQ methodology. Together, costs that arise through the
need for prevention and appraisal activities and costs due to failures represent the unnecessary additional cost
incurred in the product if all processes could operate correctly the first time. Specifically, the separate elements
of the PAF model can be defined as:
• Prevention – those activities undertaken to ensure that failures do not occur, e.g. education, training, studies
Page 5
2
• Appraisal – those activities undertaken to ensure that the finished product meets the desired quality, e.g.
quality checking systems
• Failures – a measure of failures to achieve the required quality of finished product and subsequent activities
undertaken to rectify those failures. They can be is subdivided into internal and external failures:
→ Internal – e.g. scrap, rework, delays
→ External – e.g. repairs, complaints, compensation
<< bring in Figure 1 about here >>
The optimum defect level shown in Figure 1 will vary according to the nature of the project – the more severe
the consequences of failure, the higher the requirement for quality performance. This view of the relationship
between prevention, appraisal and failure costs has been challenged in the literature. The modern view now is
more in accord with Morse and Poston (1989), Atkinson et al. (1991) and Foster (1996). The quality axis should
be defined more precisely as percentage conformance to specification and then one should recognise that, while
the relationship in Figure 1 is broadly appropriate for expressing static relationships, one must, over a longer
time-frame, recognise the dynamics of changing technology and knowledge and how even the acceptable degree
of compliance with specification can change. Recognising these dynamics, the objective is not just to estimate
the cost curves for prevention and appraisal and failures in order to find the optimal level of quality, but to link
identified failures and their causes with technological or work process improvements such that failure costs are
driven, over time, to be as near zero as possible (perhaps also with a tighter agreement on specifications). If
materials and processes can be relied upon to radically reduce the incidence of failures, this would enable a
reduction of prevention and appraisal activities such that almost all the PAF costs could be removed (although
there will always be some prevention costs associated with, for example, health and safety regulations). It was
with an awareness of this more complete picture of COQ and the possibilities for a more on-going and dynamic
linkage of COQ estimates with continuous improvement, that the authors approached a company in the
construction industry with a proposal to launch the pilot study. The results of this case study are reported in this
paper.
Page 6
3
The case study
A case study construction project was selected in order to investigate and analyse the outcomes of the
implementation of a complete COQ exercise as outlined in the BSI documents. The project was a £2.5 million
building contract undertaken on a design and build basis. The project was an office development of low
technical complexity. From the beginning of the project a great deal of emphasis was placed upon the
programme, with the client requiring a completion date during the Christmas period – a total construction period
of 38 weeks. Given our interest in developing continuous improvement through the use of the COQ
methodology, the focus for the site staff was to record and cost any internal failures as they were observed and
pursue, at that time, such additional information as was needed to make cost estimates and identify causes for
failures. Hence, this was not to be a COQ estimation procedure at the end of a period of operation or the end of a
project as many previous studies have been. Data were provided through the direct experience and observation
by individuals while supplementary information was found in the site diaries, from invoices and orders, the
project programme, allocation sheets and the bill of quantities. The quality failures data were supplemented by
data concerning prevention and appraisal costs for the project, acquired by a researcher. The researcher also
spent a considerable amount of time providing advice on implementation, ensured validity of data collection,
ensured confidentiality and mediated between individuals and organizations in order to bring academic rigour
and integrity to the collection and analysis of the data. Workshops were held at the outset to explain the nature of
quality and the importance of performance measurement and to stress the purpose of the study as being one of
learning and understanding rather than monitoring and surveillance. Efforts were also made to involve all parties
to the contract. The study was introduced to key subcontractors and suppliers in initial start-up meetings while
the consultants were also introduced to the exercise at an early stage.
Methodology
Studies elsewhere have relied primarily on after-the-event interviews with ‘key participants’ to provide data. For
example, Abdul-Rahman et al (1996) relied on sporadic interviews by a researcher to complete a quality matrix.
Elsewhere, data were collected through post-project interviews (Burati et al, 1992; Love et al, 1999; Love and
Li, 2000; Love, 2000). The approaches adopted in these studies had the drawback of relying on the memory of
the interviewees. Furthermore, they focused on quality failures for the main contractor only. This point is
Page 7
4
important because, in most developed construction markets such as the UK (and North America and Australia,
where the other cited studies were conducted) main contractors carry out very little physical work themselves.
Rather, they tend to manage and co-ordinate the input of a wide range and number of subcontractors and
suppliers (Harvey and Ashworth, 1993). Therefore, in terms of measuring cost of quality failures during a
construction project, the focus should be on the entire supply chain and the data collected during the construction
project while all relevant parties are still available and focusing on the project in hand. Furthermore, the previous
studies cited tended to concentrate on rework. However, this is merely one aspect of quality failures. The
definition in the literature is much broader and, in fact, when viewed in its broadest context of ‘cost of non-
conformance’ (Foster, 1996) can be seen as a means of covering the entire construction management process
rather than merely the physical construction of the product. What should be measured is any disruption to the
construction of the finished product, however that may be manifest. In this sense, what is being measured is the
flow of value to the client – where the process is interrupted or goes awry in some way, this is construed as non-
value adding. This view suggests a great virtue of the PAF model. When the costs of quality failures are
combined with costs incurred for prevention and appraisal, the sum of the costs can be seen as non-value adding
cost to the client. When deducted from the construction costs, the remainder can be classified as value adding.
Therefore, the PAF model identifies value adding activity by defining and measuring activity that is not value
adding.
The study reported in this paper developed a methodology gained from insights gained in a previous study
(Barber et al, 2000) which found that staff sensitivities to error exposure led to an unwillingness to participate in
the research and, consequently, led to the danger of under-reporting. Aware of this danger, the COQ exercise
was established in close co-operation with the site staff. They were involved at every stage of the introduction of
COQ, with their questions and concerns being addressed as fully as possible. In that there was a need for full
involvement from the project staff, it was important to ensure that they had ‘ownership’ of the research – that
they were not only involved in provision of data but also in the implementation and continuing development of
the methodology. The method applied in this study drew on the approach outlined in BS6143 but was modified
in line with the methodology described in Barber et al (2000) and Barber and Tomkins (1997). On this occasion,
the objective of the research was to conduct a ‘complete’ COQ study (prevention, appraisal and failures analysis)
on the pilot construction project. This represented a substantial change from previous studies purporting to
examine the COQ or cost of non-conformance (CONC) in construction projects. This is because those previous
Page 8
5
studies failed to consider prevention and appraisal costs in any meaningful way. Where they had included some
consideration of costs of quality beyond those of failure in the physical product of the construction process
(Love and Li, 1999) this consideration has been cursory at best. The study reported here represented the first
attempt to conduct a detailed COQ analysis within the construction industry in accordance with the literature. To
achieve this, the site staff were required to monitor quality failures on site through self-monitoring and
observation. This meant that the project site staff were personally responsible for recording incidents and
suggesting possible causes for the manifested effects. Consequently, the process relied on an open, ‘no-blame
culture’, where the tendency to hide or conceal quality incidents was mitigated in the knowledge that the exercise
would be used in a constructive, problem-solving and learning spirit. Each quality incident (a less pejorative
euphemism for ‘internal quality failure’) was recorded and costed by the site-based staff as it arose. In the spirit
of inclusiveness, all staff were encouraged to get involved in recording any incidents that seemed to be relevant
to the exercise, relating either to their own activities or those of other parties in the supply chain (e.g. suppliers
and subcontractors, designers and the client). Thus, the project manager, site agent, quantity surveyor and junior
staff were all involved in recording and collating data. The incidents were then categorized by activity within
PAF process model (see Table 1). Not only did this approach enable a more complete picture of quality failures
to be established, it also developed data collection and performance measurement competencies among the
participating staff. Initially, the participating staff had problems in determining exactly what constituted a quality
failure. It was decided that, in this instance, a failure would be defined as any incident that impeded the process
of construction of the building. Thus, the methodology adopted went beyond interpreting quality failures as
failure to meet specification, although it could be interpreted as complying with specification first time and
without unplanned delay. Instead, a quality failure was interpreted as an activity that failed to proceed as
‘planned’ or, in other words, an activity that was inefficient. In this sense, it reflected the view of total
performance for production being a combination of availability, efficiency and quality (Ollila and Malmipuro,
1999; Yasin et al, 1999). The researcher was in frequent contact with the staff gathering the data, to ensure that
they were thorough and appropriate and costed correctly.
An important aspect in implementing the COQ exercise on site, and in ensuring acceptance by site staff was that
the system used was simple to understand and operate. An onerous or complex system would have alienated the
staff and reduced their co-operation with the exercise and, consequently, its potential efficacy. The system
centred on a ‘logsheet’ (see Appendix). This instrument had the virtue of being immediately recognisable to the
Page 9
6
site staff. The logsheet was used as a costing mechanism, with resources (labour, materials and plant) that had
been used over and above those planned, being recorded and priced using standard pricing data. In general, the
resources were deployed by subcontractors. Where subcontractors and suppliers had incurred costs that were
difficult for the staff to ascertain, estimates were made and appropriate sums included on the logsheets. One
difficulty in costing the failures lay in estimating the cost of delays to the construction process. It was established
that such delays were only relevant where they effected the construction programme’s critical path. Two
approaches were considered:
1. Devise a proxy cost for each day of delay, related to the contract liquidated and ascertained damages
(LADs). This approach had been adopted in a previous study (Barber et al, 2000).
2. Estimate the costs related to accelerating aspects of the work to ensure the programme remained unaffected.
It was agreed that the second option was more appropriate as it reflected real activities taken on site.
The logsheet was supplemented by use of a ‘whiteboard’, located in the site office. This was useful in making the
exercise visible, thereby reducing any mistrust at the site level. The whiteboard was open for use by any
personnel involved in the construction project, included those employed by subcontractors and suppliers. It acted
as a first point of consideration, where possible failures were simply listed by people who had observed, or were
aware of, their occurrence. The facts of each incident were then considered and a decision made on whether it
should be recorded on a logsheet. The quality failures data were supplemented by data from the bill of quantities,
which were used to compile figures for prevention and appraisal costs.
While it cannot be said that the approach described above would capture every single quality failure or capture
all prevention and appraisal costs, compared with previous studies, it can probably be regarded as being the most
thorough and rigorous available.
The overall aim of the COQ exercise was to achieve a number of anticipated outcomes. These were:
1. A carefully constructed PAF model (British Standards Institute, 1990). Combined with a view of production
as a process as outlined above and in the British Standard (1992), the goal was to produce a tool that could
be used to support senior management decision-making as an on-going improvement process.
2. An investigation into and understanding of the causes of quality problems through causal analysis.
Page 10
7
3. An absolute figure for COQ failures, that might have a strong impact, were it to be disseminated and against
which future projects could be compared allowing measurement of the effects of learning, policy changes
and continuous improvement initiatives.
4. Perhaps, most importantly, the effect of encouraging a cultural change among those who participated in the
exercise, focusing their attention on quality issues generally and the importance of minimising quality
failures, both at a site level and along the supply chain.
Findings and analysis
The quality failures data were analysed on a frequent, cumulative basis to provide a developing picture of the
cost of internal quality failures for the construction project. These data were supplemented by figures for
prevention and appraisal activities, taken primarily from the bill of quantities. The findings are divided into three
sections, the first analysing the cost data, the second providing an analysis of the cost of delay as a subset of the
quality failures and the third section illustrating the causal data.
Cost analysis
After checking the logsheets for validity through regular reviews, the data were divided between pre-determined
processes in order to populate the PAF model (see Table 1). The processes were identified as elements of the
building being studied. They were chosen for both their generic nature and their usefulness to the participating
company, who wished to compare results across a number of projects. An inherent flexibility of the PAF model
is that ‘process’ can be defined in a number of ways. For example, the ‘processes’ could be specific trades or
subcontractors, as shown by Love (2000).
<< bring in Table 1 about here >>
Table 1 shows the breakdown of the COQ costs for different processes for the construction project. The total
COQ (the sum of the non-value adding costs) recorded for the project was £428,441, while the cost of quality
failures, which formed a part of that figure, was £135,140. The BSI documentation provides only a vague
description of the type of incidents that might be included in the external failures category, and includes
reference to loss of customer goodwill, etc. However, the original COQ categorisation (Feigenbaum, 1956;
Page 11
8
Oakland, 1993) restricts external quality failures to costs that occur after the product or services have been
transferred to the customer (Tsai, 1998). In this definition, snagging-type activities (included in this analysis)
would be external failures and this would mean that these had also been included within the study. Lost customer
goodwill, damaged company reputation, etc. could be included in the category of “lost opportunities” (Field, and
Swift, 1996) and would remain difficult or impossible to measure.
The total cost to the client, shown in Table 1 excludes design fees, contingencies and profit to the contractor, as
the exercise was limited to the construction process only. The appraisal and prevention costs were identified
through a review of the bill of quantities (BQ). It was established that a substantial proportion of the
preliminaries section of the BQ were inherently either appraisal or prevention costs. While they may have been
necessary, they did not, in fact, contribute to the final constructed product and, thus, were non-value adding. For
example, substantial items were:
• Hoarding – preventing ingress and escape from the construction site.
• Non-productive labour – employed in moving materials and equipment around the site, etc.
• Site staff – to ensure that the works progress to plan while preventing accidents, resolving problems as they
arose and ensuring that quality met the prescribed specification, etc.
The costed logsheets included time for site staff where this was relevant. While the time site staff were normally
involved in prevention and appraisal activities was included within the model as prevention and appraisal costs,
when they were ‘diverted’ from their usual activities by having to spend extra time to resolve quality failures the
additional costs associated with this were included in the calculations. Were there to be no quality failures, the
consequent ‘freeing up’ of staff time could mean savings in overheads as staff could be deployed elsewhere. For
each BQ item, a view was taken on whether it was prevention, appraisal or both, and a decision was taken on
where to allocate the cost within the PAF model. For example, the staff costs (the largest single non-value adding
cost) were classified as both prevention and appraisal and were distributed between the construction processes.
There was inevitably a degree of subjectivity in this operation. In some instances, it was unclear whether an item
was prevention or appraisal (indeed, the difficulty in distinguishing between prevention and appraisal costs is
noted in the British Standard and, in the instance of this study, included items such as non-productive labour and
temporary accommodation), which led to the inclusion of the ‘Other’ category.
Page 12
9
In total, 180 incidents were recorded, of which 166 were deemed to merit inclusion within the PAF model (an
average of 4.3 incidents per week). These incidents were resourced and costed by site staff using standard data.
The distribution of the incidents is shown in Table 2. The table shows that the vast majority of incidents were
defined as small or very small in terms of cost, but that together, they accounted for only 27% of the total cost of
quality failure. By far the largest contribution to the cost of quality failures (52%) were from the four incidents
categorized as either large or very large. In seeking a Pareto-relationship (Saket et al, 1986), it was found that the
19 incidents costed over £1000 (representing 12% of the total number of incidents) accounted for 73% of the
total cost of all quality failures (a relationship that was, in fact, closer to Deming’s (1986) 85/15 rule). This
finding would appear to support previous studies (Barber et al, 2000).
<< bring in Table 2 about here >>
When the data within the PAF model were analysed more closely (Table 3) it was found that quality failures
accounted for 5.84% of the contract sum while prevention, appraisal and other activities accounted for 12.68%
of the contract sum. This meant that the ‘value adding’ proportion of the contract sum was 81.48%. Focusing
specifically on the individual processes, it was found that, as a proportion of the cost of each process, failures
were greatest in the groundworks and substructure (C) and finishes (J) and least in demolition (B). While there
were a large number of failures relating to groundworks and substructure, there were relatively few relating to
finishes. The highest number of quality failures related to the external envelope (E), although the average cost of
these incidents was lower than other processes and the cost of the failures as a percentage of the cost of the
process was also relatively low. It should be noted that the high average cost of the failures associated with
finishes was partly due to the fact that, in some instances, a number of incidents relating to specific trades were
aggregated onto single logsheets. Had these been separately listed, the number of failures would have been far
greater. Table 3 is particularly interesting as it shows the relationship between money spent on prevention,
appraisal and other non-value adding activities in relation to the quality failures costs. It can be seen that the
highest of these costs, as a proportion of the associated process, were in preliminaries (A) and groundworks and
substructure.
<< bring in Table 3 about here >>
Page 13
10
Finally, the cost of quality failures were analysed in terms of their development over the period of the
construction project (Figure 2). It was found that the cost of quality failures steadily increased as the project
progressed until the final two months, where a sharp rise in the cost of quality failures was evident. There are a
number of explanations for this phenomenon. Firstly, the finishes were undertaken during this period which
represented a concentration of quality failures in a short space of time. Secondly, supervision (prevention and
appraisal costs) were scaled down during this period as staff moved to different projects, thus providing greater
scope for errors and mistakes to occur. However, perhaps the most important factor was that two of the largest
incidents occurred at this time. However, these were, in fact, ‘latent’ failures in that they had occurred earlier in
the project (during groundworks) but had not become apparent until the end of the project. Thus, the distribution
of quality failures over the duration of the project is probably distorted.
<< bring in Figure 2 about here >>
Analysis of delay
One aspect of the measurement of quality incidents that has caused controversy in previous studies was an
accurate measurement of the impact of programme delays (Nylen, 1999; Barber et al, 2000). Where programme
delays affected the critical path, they were found to have a significant effect on the cumulative cost of quality as
they impacted on later activities. Previous measurement of this phenomenon has been unsatisfactory – for each
day of delay, the costs were deemed to be equal to the contractual LADs for the project. The LADs were used as
a proxy measure for costs incurred through acceleration and increased resourcing, whether LADs were actually
levied or not. For this study, programme delays were measured more precisely. Where an event impacted the
critical path, an estimate was made of the resources that would be required to accelerate the work in order to get
the project back on programme. The resources were noted separately and incorporated within the overall cost of
the incident. As a result of this closer examination, Table 4 was produced. It can be seen that value adding costs
formed 81% of the total project cost, while prevention, appraisal and other non-value adding costs totalled 13%
of the project cost. Internal quality failures accounted for 5% of the project cost while the cost of programme
delays (which formed a subset of quality failures) formed only 1% of the project cost.
<< bring in Table 4 about here >>
Page 14
11
Causal analysis
The quality failures were analysed to determine cause and divided into the following categories:
• Communications e.g. poor information control, misunderstandings.
• Plant and equipment e.g. breakdowns, punctures.
• Personnel e.g. carelessness, lack of training, poor workmanship, sickness.
• Design e.g. mistakes that ‘get on to’ the construction site.
• Management e.g. lack of planning, errors, poor organisation.
• Suppliers (including subcontractors) e.g. poor selection, errors and mistakes.
• Force majeure e.g. third parties, weather, ground conditions.
These categories were identified in previous studies (Barber et al, 2000) and supplemented by additional
categories that became apparent during the analysis itself. The findings of the causal analyses were aggregated
by absolute numbers and cost failures, shown in Table 5. In terms of absolute numbers of incidents, force
majeure type incidents were very infrequent (2%). The majority of incidents were attributable to errors and
mistakes by specific individuals (22%) or to supplier errors (27%). Management and communication problems,
both attributable to the main contractor, totalled 25% of the incidents. It was surprising, bearing in mind the
proportion of work that was subcontracted, that suppliers did not account for a larger proportion of the incidents.
However, it should be remembered that many of the incidents attributed to personnel were the fault of
subcontractors’ employees, while the majority of the plant was supplied by third parties. Furthermore, when the
quality failures were analysed in terms of relative cost, it was found that the vast majority of incidents (55%)
were attributable to suppliers.
<< bring in Table 5 about here >>
This analysis of the causes of the quality failures arising during the project should be viewed with caution. The
categories simplify a more complex picture. For example, mistakes by specific individuals might be attributable
to the main contractor’s or suppliers’ employees and their ‘root’ causes were diverse, including lack of training
and inexperience. Similarly, the ‘root’ cause of suppliers’ errors may in fact have been poor selection of specific
suppliers in the first instance, or poor co-ordination of different trades. This is an important point, as the
Page 15
12
tendency in viewing the figures is to attribute blame. However, their purpose was to provide an overview of the
issues and indicate the direction for corrective measures and change management.
An unexpected finding of the causal analysis activity was a general impression of closer integration throughout
the supply chain. Representatives of the client, main contractor, principle consultants, key subcontractors and
major suppliers, were brought together in a joint forum, in a spirit of openness and participation. In the ensuing
dialogue, selected quality failure incidents were discussed from the multiple perspectives of the different
participants, which led to a shared understanding of the complex series of events that gave rise to many of the
incidents up for consideration. Thus, participation in the causal analysis process itself prompted a change in
attitude among the individuals working on the project. There was a realisation that problems were rarely the
consequence of any specific individual but the result of a wide range of causes, that organizations and their
employees worked within constraints often beyond their control and that sharing problems and working together
to find solutions would not only benefit the client in delivery of a final product (the building) of greater integrity
but also benefit themselves in improved margins, reduced stress and aggravation and enhanced reputations.
More specifically, the causal analysis sessions were recorded and analysis of those recordings revealed a number
of common issues (‘root’ causes) that indicated directions for possible improvement in the future and learning
themes that, together, could lead to reduced quality failure costs on future projects. These included:
• More careful selection of suppliers and subcontractors – selection on a basis of best out-turn value rather
than lowest initial cost.
• In design and build contracts, a closer and earlier involvement of the main contractor in the design process
with more consideration of buildability issues.
• Consideration of ways in which information from the planning stage could be transferred to the design and
construction stages more effectively.
• More involvement of key suppliers and subcontractors in the design stage of the project which would mean
an earlier commitment to those suppliers and subcontractors by the main contractor. This would suggest
that, for certain trades and services, strategic partnering arrangements should be established.
• Identification of common and recurring mistakes and errors that could be considered at the beginning of
future, similar projects and where effort by site staff can be directed.
Page 16
13
• Better consideration of the training needs of suppliers’ and subcontractors’ employees and a co-ordinated,
joint approach to setting training targets and seeing that these are achieved.
• A long-term strategic approach to tackling a culture of complacency that was identified to exist among
suppliers of, for example, plant and certain manufactured products that were incorporated into the finished
building.
Discussion
While the analysis of COQ on the case study project was exemplary (within the construction sector) in terms of
the richness of the data collected and the detail and depth of analysis, the findings were by no means completely
valid (hence the paper title suggests moving towards a complete methodology). There were a number of areas
where data were likely to be under-reported or were incomplete. For example, although the findings suggest that
the cost of quality failures was quite low (approximately 6%), there is a possibility that the findings were an
underestimate of the ‘true’ cost of quality failures. Apart from the likelihood of general under-reporting, the
costed incidents failed to capture the ‘full’ cost implications along the supply chain. One reason is because it was
impossible for the site staff to make an entirely accurate assessment of the implications of quality failures for
suppliers and subcontractors beyond a ‘reasonable’ estimate. Participation from subcontractors and suppliers was
admirable but reporting and/or confirmation of costs incurred was patchy and there was doubt as to whether the
suppliers recognised the full implications of incidents for themselves and their suppliers in any case.
A further factor, which should be considered in a ‘complete’ analysis, is the extent to which overheads (to which
the project made a contribution) should be included for prevention and appraisal activities. Indeed, there is a
reasonable argument that many overheads are non-value adding (e.g. training, quality assurance systems,
checking invoices, etc.) and the project’s entire contribution to overheads should be included in the PAF model.
Additionally, bearing in mind that the contract was design and build, the analysis also failed to consider
incidences of quality failures within the design process itself and, therefore, the cost of the design element was
removed from the contract cost in conducting the analysis. While some quality failure logsheets were recorded
during the design phase where perceived quality failures had occurred, it was found difficult to integrate these
into the model as there seemed to be considerable overlap with construction phase events – the logsheets from
the design phase became manifest in the construction phase. If this were the case, design fees should remain in
Page 17
14
the analysis as the implications of design quality failures could be regarded as having being been captured after
all. This remains an unresolved issue. Incidentally, the presence of implicit prevention and appraisal costs within
contract sum applies to design fees as much as it applies to other supplier activity.
An additional concern was that the prevention and appraisal costs were also underestimated. This was because of
the high degree of subcontracted work. While it was possible to make a fairly close analysis of the prevention
and appraisal costs incurred by the main contractor, subcontract and supplier contributions to the project cost
were included as being entirely value adding. However, their costs would also include an element for prevention
and appraisal, which was not considered in the analysis. Were each subcontractor and supplier to have costs of a
similar order to that of the main contractor (prevention and appraisal costs accounted for 13% of the project
cost) the total prevention and appraisal costs for the project would be far higher. While it would be difficult to
assess the prevention and appraisal costs for all subcontractors and suppliers, it might be possible to estimate the
total through the supply chain based on the assessment of a representative sample. Another approach would be to
conduct a Pareto analysis of suppliers and subcontractors, identifying those that contributed the most substantial
value to the project and undertake a detailed analysis of their prevention and appraisal costs only. Naturally, such
an approach would have confidentiality implications.
Finally, while our objective to introduce a dynamic procedure for continuing improvement based upon COQ was
achieved to a considerable extent in so far as data gathering and interpretation was on-going and used for
correction, where possible, throughout the project, there was no significant move to reduce prevention and
appraisal costs as a result of reducing failure costs. It would seem that this requires a perception that failure
reduction has been achieved consistently over several projects before confidence is gained to cut prevention and
appraisal activities. Even so, a very senior manager in the company, but not associated with the initial
application of this pilot study stated informally:
I was sceptical at first, but I have been won over and now we will be applying cost of quality analysis in
this way over other projects. Ongoing cost of quality enables root cause detection and fast rectification.
To this we would add that to fast rectification can be added the prospect of prevention and appraisal cost
reduction.
Page 18
15
Conclusions
When conducting the study, it was found that, rather than merely capturing the cost of quality failures in terms of
rework, what was really being captured was the cost of failures in the management and communication process.
The categories arising during the causal analysis revealed that rework and scrap were only one element of failure
cost and that a study limited to them does not reveal the range of causes leading to quality failures. Further, it
was found that when quality incidents occurred, their effect was to further disrupt the management and
communication process, by diverting managerial effort. When seen from this perspective, COQ has the potential
to become much more than a tool of ‘final’ analysis. Rather, it could be seen as a potential means of measuring
and assessing the waste resident within the entire procurement process, with causal analysis being a means of
identifying ways of reducing said waste.
An important finding of the research exercise was that it appeared to refute the contention that cost of quality
failures represent the largest category of the PAF relationship (Israel and Fisher, 1991) representing “70-85% of
the COQ in most organizations” (Johnson, 1995). This assertion would explain why the previous studies in
construction have focused almost entirely on the quality failures element to the exclusion of prevention and
appraisal costs. Indeed, it has been argued elsewhere that analysis of COQ should ignore prevention and
appraisal costs as these are vital in achieving the required quality (Nylen, 1999), although, as argued earlier, this
does not take fully into account the dynamics of possible continued improvement. The findings from this study
suggest that quality failures are actually quite small and that prevention and appraisal costs are much higher,
providing the greatest scope for reducing non-value adding costs. This would suggest that the aim should be to
find ways of reducing the cost of quality failures while ensuring that the related prevention and appraisal costs do
not escalate beyond the return in failure reduction, as illustrated in Figure 1. Over time, as systems and processes
are developed to eradicate quality failures, and these become embodied within the organization’s policies and
procedures (for example, strategic partnering) prevention and appraisal costs might also be further reduced.
Thus, a step-change would be observed in the prevention and appraisal curve. A thorough COQ analysis such as
the one presented in this paper provides the necessary information to investigate ways of reducing all aspects that
fail to add value to the final constructed product. It would also be interesting to determine whether the
relationship between prevention, appraisal and failure costs observed in this study was unique to the construction
Page 19
16
industry or whether inaccurate studies elsewhere have led to a mistaken belief that quality failures are so large
relative to prevention and appraisal costs.
The case study investigation satisfactorily fulfilled the first three aims of the exercise, namely to develop:
1. A carefully constructed PAF model to support senior management decision-making.
2. An understanding of the causes of quality problems through causal analysis.
3. An absolute figure for COQ failures against which future projects can be compared allowing measurement
of the effects of learning, policy changes and continuous improvement initiatives.
However, the fourth item (a change in culture and attitude to quality) is difficult to measure. One observed
impact of the process, which may qualify as a change in attitude and culture, was a gradual improving of
relationships along the entire supply chain. Although this may have occurred during the natural course of the
project, in terms of quality, it manifested itself in a better understanding and appreciation of the constraints,
imposed both by third parties and other parties along the supply chain, on organizations achieving appropriate
quality standards. This led to the realization that, particularly when the project is under design and build, the
quality and integrity of the constructed product is reliant not only on organizations focusing on their own
activities but also on how those activities impinge upon others and how, through communication and
thoughtfulness, they can help others to deliver quality. Supply chain relationships and interaction are the key to
delivering quality and attention should be focused on this aspect in improvement attempts.
As the study progressed, it emerged that there were a number of areas that might receive closer examination in
future studies. First, although a review of the available literature suggests that this is the most thorough and
rigorous COQ study to have been conducted in the construction industry, there remains scope for improving the
data validity and reliability. For example, future studies might use more complete costing, including better and
more thorough attention to costs manifest down the supply chain. Similarly, future studies could involve closer
analysis of programme delays and the critical path. The fact that an activity was delayed, but not on the initial
critical path, should not necessarily exclude consideration of the delay within the costing process. The delay on
initially non-critical activities may alter the subsequent critical path, making previously non-critical activities
critical. For example, it might be worth considering the potential to link COQ analysis into the software designed
to map construction programmes.
Page 20
17
Finally, the study produced a large amount of information of both a strategic and operational nature. The
participating contractor now has a substantial amount of data at its disposal. Indeed, so pleased was the company
with the results of the case study that they have extended the exercise to a number of new projects and the
quantity of data, and their understanding of how they add value to the construction process, is set to increase
dramatically. If they are able to disseminate the knowledge resident in the data across the company, could result
in important benefits in reduced transaction and quality failure costs and improved supplier and subcontractor
relationships. This could enable the company to differentiate itself from its competitors (notoriously difficult to
do in the construction industry) through delivery of better and more reliable quality at the same or better margins
for the contractor and its supply chain while maintaining a competitive price for its clients. The key difficulty for
the company lies in finding a means of transferring the lessons learnt throughout its organization and engaging
the supply chain in that learning process.
References
Abdul-Rahman, H., Thompson, P. and Whyte, I. (1996) Capturing the cost of non-conformance on construction
sites: an application of the quality cost matrix, International Journal of Quality and Reliability Management, 13
(1), 48-60.
Atkinson, J.H., Hohner, G., Mundt, B., Troxel, R.B. and Winchell, W. (1991) Current trends in cost of quality:
linking the cost of quality and continuous improvement, Montvale, New Jersey: National Association of
Accountants.
Baiman, S., Fischer, P. E. and Rajan, M. V. (2000) Information, contracting and quality costs, Management
Science, 46 (6), 776-789.
Barber, P., Graves, A., Hall, M., Sheath, D. and Tomkins, C. (2000) Quality failure costs in civil engineering
projects, International Journal of Quality and Reliability Management, 17 (4/5), 479-492.
Barber, P. and Tomkins, C. (1997) The application of cost of quality analysis to civil engineering: an analysis of
drivers, Agile Working Paper, University of Bath.
British Standards Institute (1990), BSI Handbook 22. BS 6143 Guide to the economics of quality : Part 2
Prevention, appraisal and failure model (Revised), London: British Standards Institute.
British Standards Institute (1992), BSI Handbook 22. BS 6143 Guide to the economics of quality : Part 1
Process cost model (Revised), London: British Standards Institute.
Page 21
18
Burati Jr., J. L., Farrington, J. J. and Ledbetter, W. B. (1992) Causes of quality deviations in design and
construction, Journal of Construction Engineering and Management, 118 (1), 34-49.
Carrol, J. S. (1998) Organizational learning activities in high-hazard industries: the logic underlying self-
analysis, Journal of Management Studies, 35 (6), 699-717.
Chase, N. (1998) Accounting for quality: counting costs, reaping returns, Quality Online,
[http://www.qualitymag.com/articles/1998/oct98/1098f1.html].
Crosby, P. B. (1979) Quality is free: The art of making quality certain, London: McGraw-Hill.
Dale, B.G. and Plunkett, J.J. (1991) Quality costing, London: Chapman and Hall.
Davis, K., Ledbetter, W. R. and Burati Jr., J. L. (1989) Measuring design and construction quality costs, Journal
of Construction Engineering and Management, 115 (3), 385-400.
Deming, W. E. (1986) Out of crisis: quality, productivity and competitive position, Cambridge: Cambridge
University Press.
Feigenbaum, A. V. (1956) Total quality control, Harvard Business Review, 34 (6), 93-101.
Fernandes, C. M. B., Walker, R. Price, A. Marsden, J. and Haley, L. (1997) Root cause analysis of laboratory
delays to an emergency department, The Journal of Emergency Medicine, 15 (5), 735-739.
Field, S. W. and Swift, K. G. (1996) Effecting a quality change: an engineering approach, London: Arnold.
Foster Jr., S. T. (1996) An examination of the relationship between conformance and quality-related costs,
International Journal of Quality and Reliability Management, 13 (4), 50-63.
Harvey, R. C. and Ashworth, A. (1993) The construction industry of Great Britain, Oxford: Butterworth-
Heinemann.
Israel, A. and Fisher, B. (1991) Cutting quality costs, Quality Progress, 24.
Ittner, C. (1988) Texas Instruments: Materials and Controls Group, Harvard Business School Case 189-029.
Johnson, M. A. (1995) The development of measures of the cost of quality for an engineering unit, International
Journal of Quality and Reliability Management, 12 (2), 86-100.
Juran, J. M. (1988) Juran’s quality control handbook, London: McGraw-Hill.
Kumar, K. and Brittain, J. C. (1995) Cost of quality: evaluating the health of British manufacturing industry, The
TQM Magazine, 7 (5), 50-57.
Love, P. E. D., Heng, L. and Mandal, P. (1999) Rework: a symptom of a dysfunctional supply-chain, European
Journal of Purchasing and Supply Management, 5, 1-11.
Page 22
19
Love, P. E. D. and Li, H. (1999) Knocking down walls and building bridges – overcoming the problems
associated with quality certification, Construction Management and Economics (forthcoming).
Love, P.E.D., and Li, H. (2000) Quantifying the causes and costs of rework in
construction, Construction Management and Economics, 18 (4), 479-490.
Love, P.E.D. (2000) The development and empirical testing of a method for quantifying rework costs in
construction projects, Engineering Construction and Architectural Management (forthcoming).
Love, P.E.D., Li, H., and Holt, G.D. (2000) Some observations on the indirect costs of construction rework,
Building and Environment (forthcoming).
Love, P.E.D., and Li, H. (2000). Overcoming the Problems Associated with
Quality Certification, Construction Management and Economics, 18 (2), 139-149.
Morse, W.J. and Poston, K.M. (1989) Accounting for quality costs – a critical component of CIM” in
Campanella, J. (ed), Quality costs: ideas and applications, Volume II, Milwaukee: ASQC Quality Press.
Nylen, K-O. (1999) Civil works – Unique projects or a repeatable process?, unpublished PhD thesis,
Stockholm: Royal Institute of Technology.
Oakland, J. S. (1993) Total quality management, 2nd Ed., Oxford: Butterworth-Heinemann.
Ollila, A. and Malmipuro, M. (1999) Maintenance has a role in quality, The TQM Magazine, 11 (1), 17-21.
Robert, G., Brown, J. and Garvican, L. (2000) Cost of quality management and information provision for
screening: colorectal cancer screening, Journal of Medical Screening, 7 (1), 31-34.
Saket, M. M., McKay, K. J. and Horner, R. M. W. (1986) Some applications of the principle of cost-
significance, International Journal of Construction Management and Technology, 1 (3), 5-22.
Stein, E. W. and Miscikowski, D. K. (1999) FAILSAFE: supporting product quality with knowledge-based
systems, Expert Systems with Applications, 16, 365-377.
Tomkins, C., Madigan, D., May, B. and Steckelmacher, P. (1997) Modern developments in cost management:
implications for management accounting in the civil engineering industry, Agile Working Paper, University of
Bath.
Tsai, W-H. (1998) Quality cost measurement under activity-based costing, International Journal of Quality and
Reliability Management, 15 (7), 719-752.
Yasin, M. M., Czuchry, A. J., Dorsch, J. J. and Small, M. (1999) In search of an optimal cost of quality: an
integrated framework of operational efficiency and strategic effectiveness, Journal of Engineering and
Technology Management, 16, 171-189.
Page 23
20
Appendix
Cause EffectRef. Resource Quantity Unit Rate Total CostDate Engineers hours 0Time QS hours 0
Previous Sheets Contract Surveyors hours 0Process For PAF model Foremen hours 0
Description of Event:- Gangerman hours 0Labourers hours 0Scaffolders hours 0Steel Fixers hours 0Joiners hours 0Welders hours 0Painters hours 0Drivers inc lorry hours 0
Programme Effects:- Purchasing hours 0Admin. hours 0Inspectors hours 0Carpenters hours 0Contracts man hours 0
Activity Being Undertaken:- Bricklayers hours 0Site manager hours 0360 excav inc dri hours 0
Learning Outcomes:- Other Costs:- Cost (C) Subtotal 0
b/f A 0b/f B 0
Additional Plant Required:- Cost (A) b/f C 0
0Additional Materials Required:- Cost (B)
0 0 Total 0
17-Nov-99
in these cells.For example, subcontract
example
if thought necessaryas appropriate
Immediate ’apparent’ cause should be put here as prompt for later analysis
lump sum quotations couldgo here.
The sum of this item is here
Resources Required
The person completingthe sheet can placeanything not included above
This is a description of a Cost of Quality Incident. As you can see, the text wraps automatically in this box - one simply has to click on the box and type as usual. The boxes below are formatted to ’wrap’ in the same way.
This box has been found to be paticularly important in previous Agile studies as a non-productive activity (quality failure) that effects the critical path in the programme can have profound cost implictions
This box also acts as a ’prompt’ for later analysis.
This box is intended to provide the opportunity to record any obvious ways things could be done differently in future.
COST OF QUALITY LOGSHEET
For recording items of plant involved in incident
For recording materials involved in incident
See ’Cause’ box.
Optimum
Cost of preventionand appraisal
Cost of quality failures
Improving quality
Increasingcost
Total costs of prevention, apprasal
and failures
Figure 1. Optimisation of quality and cost (Foster, 1996)
Page 24
21
Table 1. PAF model illustrating costs for case study project
Prevention Appraisal Others Failures
Preliminaries 332,513 56,930 9,052 7,491 9,182 249,858
Demolition 65,121 2,450 1,754 1,247 403 59,267
Groundworks and substructure 282,764 32,117 9,894 5,734 73,764 161,255
Frame 215,825 8,597 6,155 4,376 5,902 190,795
External envelope 365,072 19,821 11,211 7,545 16,280 310,215
Internal construction activities 153,902 12,388 4,473 3,181 3,601 130,259
Mechanical and electrical installation 207,884 7,821 5,600 3,982 4,314 186,167
Roof 134,857 7,995 6,579 3,434 3,024 113,825
Finishes 129,094 8,891 3,752 2,668 11,884 101,899
External works 426,542 17,152 12,280 8,731 6,786 381,593
Total 2,313,574 174,162 70,750 48,389 135,140 1,885,133
Total cost to client £)
ProcessNon-value adding costs (£) Value adding
costs (£)
Table 2. Distribution of quality failures
Totals 166 135,140 814
10001+ (very large) 2 53,518 26,759
5001-10000 (large) 2 16,141 8,071
1001-5000 (medium) 15 29,194 1,946
501-1000 (small) 18 12,374 687
1-500 (very small) 129 23,913 185
Category (£)Number of Incidents
Total Cost of Incidents (£)
Average Cost/ Incident (£)
Page 25
22
Table 3. PAF model expressed in proportions
P, A and O* Failures
Preliminaries 332,513 14 22.10 2.76 75.14
Demolition 65,121 2 8.37 0.62 91.01
Groundworks and substructure 282,764 30 16.89 26.09 57.03
Frame 215,825 10 8.86 2.73 88.40
External envelope 365,072 41 10.57 4.46 84.97
Internal construction activities 153,902 16 13.02 2.34 84.64
Mechanical and electrical installation 207,884 15 8.37 2.08 89.55
Roof 134,857 7 13.35 2.24 84.40
Finishes 129,094 17 11.86 9.21 78.93
External works 426,542 14 8.95 1.59 89.46
Total 2,313,574 166 12.68 5.84 81.48* P = Prevention, A = Appraisal, O = Others
Non-value adding costs (%)No. quality failures
ProcessTotal cost to
client (£)Value adding
costs (%)
Page 26
23
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
April May June July August September October November December January
Month
CO
Q/m
onth
(£)
0
20000
40000
60000
80000
100000
120000
140000
160000
Cum
ulat
ive
CO
Q (
£)
Figure 2. Distribution of cost of quality failures over the period of the case study project
Table 4. Delay costs as a proportion of other project costs
Prevention, appraisal and others 293,300 12.68
Quality failures 109,454 4.73
Delays to programme 25,685 1.11
Value adding costs 1,885,137 81.48
Total project costs 2,313,576
Cost (£) Percentage
Table 5. Causal analysis of quality failures
Page 27
24
Communications 15 9.04 4746 3.51
Plant and equipment 21 12.65 4439 3.28
Personnel 37 22.29 24155 17.87
Design 18 10.84 7859 5.82
Management 26 15.66 17405 12.88
Suppliers 46 27.71 74368 55.03
Force majuere 3 1.81 2168 1.60
Total 166 135139
PercentageCauseNo. of failures
PercentageCost of
incidents (£)