This article was downloaded by:[HEAL-Link Consortium]

On: 16 July 2008

Access Details: [subscription number 786636650]
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Production

Research
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713696255
Quality costs: a critique of some 'economic cost of
quality' models
J. J. Plunkett a; B. G. Dale a
a
Department of Management Sciences, UMIST, Manchester, U.K

Online Publication Date: 01 November 1988

To cite this Article: Plunkett, J. J. and Dale, B. G. (1988) 'Quality costs: a critique of
some 'economic cost of quality' models', International Journal of Production
Research, 26:11, 1713 — 1726
To link to this article: DOI: 10.1080/00207548808947986
URL: http://dx.doi.org/10.1080/00207548808947986

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

INT. J. PROD. RES., 1988, VOL. 26, NO. I I, 1713-1726

Quality costs: a critique of some 'economic cost of quality' models

J. J. PLUNKETTt and B. G. DALEt

While carrying out a study of the collection and use of quality-relatedcosts in

manufacturingindustry the authors found in the literature many notional models
purporting to indicate the relationships between the major categories of quality
costs, and a few sets of real data. Despite being based on common principles,
there are wide differences between some of the models, and between the models
and real data. The paper categorizes and discusses the models in the light of the
research experience. It is concluded that many of the models are inaccurate and
misleading, and serious doubts are cast on the concept of an optimum quality
level corresponding to a minimum point on the total quality-cost curve.

l. Introduction
In the course of a two-year study of the collection and use of quality-related
costs in manufacturing industry (Plunkett 1986) it became apparent that the costs
and economics of many quality-related activities, including investment in prevention
and appraisal activities, are not known (Plunkett and Dale 1986). It is perhaps
surprising therefore to find published in the literature a number of diagrammatic
representations of quality-costs models which purport to show relationships
between the major quality-cost categories. Many discussions of quality costs in the
literature feature such a model, but many quality-management practitioners are
doubtful about the representations and feel that they can be misleading.
The general suppositions underlying all these notional models are that invest-
ment in prevention and appraisal activities will bring handsome rewards from
reduced failure costs, and that further investment in prevention activities will show
profits from reduced appraisal costs. However, despite these common principles
there are wide differences between some of the models. This paper discusses the
various notional models and their implications and, in the light of our research,
compares them with data from the literature.

2. A classification of quality-cost models

For the purpose of this discussion, models of quality-cost relationships published
in the literature can be conveniently arranged into the following five groups.
Group A. Notional diagrams used in standard textbooks to indicate the prin-
ciples of reducing failure and appraisal costs by increasing expenditure on preven-
tion activities. Typical diagrams taken from Kirkpatrick (1970), Robertson (1971),
Caplen (1972), and Besterfield (1979) are shown in Fig. 1.
Group B. Diagrams based on actual data published by Krzikowski (1963), Kohl
(1976), Campanella and Corcoran (1983), Huckett (1985), and Hagan (1986). These
are shown in Fig. 2.

Received June 1987.

t Department of Management Sciences. UMIST, Manchester, U.K.
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1714 J. J. Plunkett and B. G. Dale

BESTERFIELD 1979 CAPlEN 1972

< TOTAL DIRECT OPERATNO

QUALITY COSTS nD"".'" j

PREvENTI()lIj "NO
APPRAISAL COSTS

'oo':-v-"'-----------==--v:
PRODUCT DEFECT LEVEL- '"

DECREASING

Reproduced by permission of Prentice-Hall Inc. Reproduced by permission of Century Hutchison Ltd.

KIRKPATRICK 1970
ROBERTSON 1971 INCREASING
COST

r ~,~au""",co,,
--
QUALlTV CONTROl COSTa
(PREVENTION COSTS)

'.R """'c;';;oo;;;o,.......---,=~
HIGH
MINIMUM
TOTAL COST

LL.-==:2::~~~
: PREVENTION
FAILURE

INCREASINGQUALITY
OF COf\FORMANCE
QUALITY lEVEL

Reproduced with permission of Pitman Publishing Ltd.

Figure I. Group A.

Group C. Models from Lockyer (1985) and from the trammg literature of a
.Ieading automotive manufacturer, company A (Anon. 1985), both of which appear
to be based on experience, are grouped with a model constructed from data taken
from an illustrative case developed by Veen (1974). They are shown in Fig. 3.
Group D. Notional diagrams all showing similar characteristics and taken from
publications devoted specifically to quality costs, i.e. BS 6143 (1981), Urwick Group
(1981), and Harrington (1976). These are shown in Fig. 4.
Group E. Notional models which are similar in principle to the others but which
differ in an important detail (i.e. the position of the optimum cost in the quality
spectrum). They are taken from Juran (1974), Veen (1974), Thoday (1976), and Cam-
panella and Corcoran (1983). They are shown in Fig. 5.
Even a cursory examination of the groups reveals some notable differences, the
most striking of which is the dissimilarity between the notional models (Figs I, 4
and 5) and those based on actual data (Fig. 2). The remaining group, group C
(Fig. 3), bears closer resemblance to the data plotted in Fig. 2 than to any other
group. Another striking feature is the way in which some of the notional models
conveniently, and in some cases inexplicably, combine major quality-cost categories.
Lastly there is the diversity of measures of quality used as abscissae in indicating the
relationships between the major quality-cost categories.
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

Quality costs 1715

2.1. Group A
The models shown in Fig. I are a selection of notional models taken from text-
books, where they were clearly intended only to convey the idea that there can be a
trade-off between prevention and appraisal costs and failure costs (though as Cox
1979 rightly points out, such trade-offs are not always allowable). It is not intended
here to analyse these models in detail, but it is worth noting Caplen's (1972) com-
bination of failure and appraisal costs vs Besterfield's (1979) preference for combin-
ing prevention and appraisal costs in indicating the principle.
It is also interesting to note that in three of the four models the minimum cost
coincides with an approximately equal division of costs between failure and
prevention/appraisal costs, which is in broad agreement with Caplen's (1972) cri-
terion for the determination of inspection levels. However, in the remaining model
Caplen (1972) shows prevention costs alone to be equal to combined failure and
appraisal costs. The latter situation is far removed from any real situation found in
our research or elsewhere in the literature.
The combination of cost categories in these ways can seem logical, but it is
dependent upon one's objective. If, for example, the objective is to reduce failure
costs, there may be little point in distinguishing between prevention and appraisal
costs. If, on the other hand, the objective is to reduce failure and/or appraisal costs,
then maybe it is not important to separate them. The research experience was that
quality managers are constantly looking for preventive measures which will reduce
appraisal and/or prevention costs, but this does not justify combining them. It is our
view that each cost category should be plotted separately. It is worth noting also
that these notional models do not distinguish between internal and external failure
costs.

2.2. Group B
The data plotted in Fig. 2 were the only published data found which could be
used to check the notional models. They are also useful, because cost-conscious
companies are always keen to learn by how much their quality costs might be
reduced by improved management of quality.
It will be noticed immediately that all the data are plotted against a linear time
base rather than against some measure of quality performance. While it is reason-
able to suppose that the cost reductions were accompanied by improvements in
quality, the time/quality relationships, or their effects on the costs diagrams, are not
known. Indeed, it is unlikely that the cost or quality profiles between inception of a
cost-reduction programme and post-warranty pay-off will be linear with time.
Ideally, costs should be plotted against measures of quality performance.
Krzikowski's (1963) data show the relative proportions of the constituent cost
categories. In the original paper the total quality costs are also expressed as percent-
ages (6,4-4,4%) of some undefined cost. From the text it appears that they are
percentages of manufacturing cost. A feature of the data is the relatively high pro-
portion of appraisal costs (60-75%) and low failure costs (12-20%) by comparison
with typical figures of 30% and 65% respectively from more recent literature. The
slight upturn in total costs in the last year is explained in the text of the paper as
being entirely owing to a one-off additional cost for training inspectors.
Kohl's (1976) data are closer to what is regarded as typical by many quality-cost
commentators, and suggests that fairly modest increases in prevention and appraisal
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1716 J. J. Plunkett and B. G. Dale

KRZIKOWSKI 1963

11158 '951 111118 1I1S9 11180 11161 Hie:! 1967 HI68 19611 1970 Hll1 1I~72 '1173 11174

S" CAMPANELLA AND CORCORAN 1983

~ ,
~
~
.
, ~OT" .. aUAlITYCOST

~
r
.! 4

, ~~NOUAL'TY COST
"PPR"ISALC~

PREVENTION COST

JFMAIolJJASOJIIDJ FM"MJJ"SOND 11180 11181 '98:2 11183 I~

1(177 1978

© American Society for Quality Control.

Reprinted by permission.

HAGAN 1986

r------------:=I'OO

"IJ

Reproduced by Courtesy of Marcel Dekker Inc.

Figure 2. Group B.

expenditure can produce dramatic reductions in failure costs in the beginning, but
that a point is reached such that, if the prevention and appraisal expenditures are
not increased further, there is no further significant reduction in failure costs. The
BS 6143 (1981) arid Thoday (1976) models (Figs. 4 and 5 respectively) appear to
support this view by indicating a continuing reduction in failure costs whilst increas-
ing prevention and appraisal costs. Mortiboy's (1985) explanation of these relation-
ships is that in the early stages of a quality-cost reduction programme quality
managers tackle problems themselves, improving quality and reducing failure costs
at little or no extra cost to the company. The quality manager is then fully occupied
trying to maintain the higher level of quality attained, with the same level of
resources, leaving him no time to tackle further problems. Hence there is no further
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

Quality costs 1717

improvement without taking on extra resources, which subsequently increases the

joint prevention and appraisal expenditure. A similar situation arises when quality-
improvement groups are set up to tackle quality problems. Improvement continues
only while the group activity is sustained and, in the words of Juran, 'management
hold the gains'.
Huckett's (1985) published data are limited to total quality costs expressed as a
percentage of manufacturing cost. However, it is evident from the text (and from the
fact that the company, Rank Xerox at Micheldean, were joint winners of the 1984
British Quality Award for quality assurance throughout the production process)
that the company achieved very high standards of quality conformance but their
total quality costs are continuing to fall, suggesting that if indeed there is an
optimum cost it occurs close to perfection. (It is perhaps pertinent to note here
Sekiya's (1985) assertion that, whilst it is possible to set optimal quality levels, those
companies that set as their goal the attainment of the very finest quality will win in
the end.) Interesting cost items included in Huckett's quality costs are 'exceeding
requirements-work performed for which there is no need' and 'lost opportunities-
lost business due to failure to meet customer requirements'. If these refer directly to
their standard products it is reasonable to include them, but otherwise 'work for
which there is no need' may be a matter of subjective judgement, debatable, and
perhaps more a matter of efficiency than quality.
Campanella and Corcoran's (1983) data corroborate the impressions gained
from other real data, and from research experience, that, in particular, making sig-
nificant reductions in total quality costs has been a slow process brought about
mainly by modest increases in prevention costs. An interesting feature of the data is
that reductions in appraisal costs contributed nearly as much to the reduction in
total quality costs as did the reduction in failure ('unquality') costs. The potential for
large savings to be made by reducing appraisal costs is a popular theme among the
quality fraternity, and indeed Richardson (1983) and Kohl (1976) have claimed that
substantial savings were achieved by this means.
Hagan's (1986) data, gathered over 11 years, broadly corroborate Kohl's (1976)
data in that appraisal and prevention costs are maintained at a sensibly constant
proportion of sales income whilst failure costs are continually reduced. It indicates
also a transfer of expenditure from appraisal to prevention in the early years, even
in the face of falling sales revenue. It then appears that the company featured by
Hagan felt they had the right balance between these two quality-cost categories and
maintained them at a constant proportion of their increasing sales revenue for 4 or
5 years. Their judgement in this appears to be vindicated by the continuing decline
in failure cost and total costs of quality. The value of sales increased dramatically
over the last few years and there is a suspicion that, as drawn, the fall in the rate of
quality improvement (as indicated by the change in the rate of fall of Failure costs) is
accompanied by a slight relative increase in appraisal expenditure compared with
prevention expenditure, but this is not certain. It would be interesting to know
whether, in this case, the dramatic increase in sales income in the tenth year was
owing to increased output, or to monetary inflation and changes in exchange rates.

2.3. Group C
The models from Lockyer (1983), company A (Anon. 1985), and one based on
illustrative data taken from Veen (1974), have been grouped together in Fig. 3 and
shown in their original forms because they have much in common. Firstly, they bear
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1718 J. J. Plunkett and B. G. Dale

\~" POSSiBlE
SAVINGS FROM
A"BASIC'
QUALITY COST
SYSTEM

PROCESS CAPABiliTY INCREASING

UNBALANCED BASIC B"SlC COMPREHENSIVE
Reproduced by permission of Pitman Publishing Ltd. SITUATION INSPECTION CONTROL OU"LITY CONTRO\.

DEVELOPMENT STAGES

'0
eo ~n COMPANY A 1985-
.~ REPLOTTED
c.. .j .

"'.1>

.
~c
§ s
a

I II IV
UNCERTA~TY A\\oo'~ENI'lG ENlIGHTENMmT WISDOM DEVELOPMENT STAGES
'"
DEVelOPMENT STAGES

Figure 3. Group C.

a greater resemblance in shape and proportion to the actual data curves shown in
Fig. 2 than to the notional models shown in Figs. I, 4 and 5, although none of the
three is claimed to be based on actual data. Secondly, the style of presentation of all
three leads to some ambiguity because it is not entirely clear whether or not the
curves are of cumulative costs. Thirdly, the diagram derived from Veen's illustrative
data (1974) and the model advanced by company A (Anon. 1985) are strikingly
similar, despite differences in labelling and magnitude of costs.
Examination of the models in this group raises a number of interesting points.
For example, Lockyer's diagram straightaway draws attention to the problem of
definitions of quality and what expenditures to include among quality costs. Most
contributors to the literature on the topic do not bother to define their terms.
Lockyer, however, defines failure costs in his model as 'correction costs', thereby
perhaps omitting scrap, warranty and other failure costs. He also uses 'Total quality
assurance costs for quality acceptable to customer' and suggests that the relation-
ships he indicates are only true for one customer-perceived quality level. Whilst
Lockyer's concepts may be more difficult to grasp he at least defines his terms, and
the use of process capability has advantages over undefined measures of quality.
However, the way the model is labelled in the original reference suggests that the
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

Quality costs 1719

costs are plotted cumulatively, but it is evident from the text that they are in the
usual non-cumulative form. If one keeps in mind the narrow view taken of failure
costs in this case (i.e. correction costs only), the uncertainty of the relationship
between degree of quality and process capability (which is usually taken to mean the
ability to meet a given specification), and its limitation to 'one customer-perceived
quality level', it would be unwise to claim that valid comparisons can be made with
other models. Nevertheless, the failure, appraisal and prevention curves are of
similar form and relationship as in the models based on real data though the total, if
drawn, would pass through a minimum, and correspond with Lockyer's ideas
expressed in the accompanying text:
Where the manufacturing capability is low for a quality acceptable to the cus-
tomer, the total quality assurance costs are high, the failure costs predominating.
As the manufacturing ability improves (usually by increasing appraisal and-
especially-prevention costs), the failure costs drop very steeply. When the manu-
facturing capability is "matched" to the customer's quality requirements, the
total quality assurance costs will be at a minimum.
Veen's illustrative data first came to the writers' notice, plotted in the form
shown in Fig. 3, in a private communication. In his original paper, Veen publishes
the model shown in Fig. 5 and uses other data in an ingenious presentation about
balancing various kinds of quality costs. Veen does not plot these other data
graphically. They have been picked up and presented (with acknowledgement), in
the form shown in Fig. 3, by a contributor to a BSI technical committee. The pres-
entation is included here to illustrate how readily seemingly substantive models can
be created and given credibility by taking data from reputable sources and using
them for purposes for which they were not intended. Worse still, the presentation of
the contrived model is ambiguous and misleading. The presentation is ambiguous
because the labelling of the curves is inconsistent with the label indicating 14% of
turnover as possible savings. If it is assumed that the curves are absolute cost
curves, and the costs of each of the major cost categories are summed at each devel-
opment stage, quality costs are 32, 43, 39 and 26% respectively. These figures
cannot be manipulated to give rise to a 14% saving as indicated on the diagram.
(Moreover, it is misleading to suggest that, in the general case, quality costs may be
such a high proportion of turnover.) Hence it is clear that the top curve must be a
total-costs curve and that the magnitude of the cost categories at each stage is
represented by the distance between adjacent curves.
Company A's diagram (taken from their supplier training course notes) plots
quality costs as a percentage of sales vs Crosby's (1979) four stages of quality man-
agement. The similarity to the diagram based on Veen's illustrative data has already
been pointed out. It should be noted also that the total quality costs indicated at
each management stage approximate closely to those indicated by Crosby (1979).
It is clear from the shading and labelling of this model that the costs are plotted
cumulatively and therefore it is not directly comparable with most other models,
hence it has been re-plotted in the more usual non-cumulative form (Fig. 3). The
replotted model has a number of interesting features. The first is that in progressing
from 'Uncertainty' to 'Awakening' a modest increase in appraisal expenditure pro-
duces a substantial reduction in total quality costs, which is attractive to the sup-
plier, and a dramatic transfer of the remaining failure costs from the external-failure
category to the internal-failure category, which suits the purchaser's objectives very
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1720 J. J. Plunkett and B. G. Dale

well. Further development of a company's quality management system sees a trans-

fer of expenditure (and emphasis) from appraisal to prevention activities, accompa-
nied by substantial reductions in internal failure, external failure and total costs. It is
interesting also to note that, as drawn in the diagram, the reduction in external
failure costs is 25% greater than the reduction in total costs, whilst the internal
failure costs are approximately doubled. The message company A is giving to its
suppliers is therefore very clear-quality costs should be in-house.
However, the most interesting feature of the model is .the shape of the total cost
curve. It is interesting in two respects. Firstly, it does not pass through a minimum.
This causes one to conjecture whether this is because, in company A's experience,
quality costs do not (or should not) pass through a minimum, or whether they do
not wish to give their suppliers the idea that there is an optimum level of quality
(which may be some way away from perfection, and which would conflict with
company A's operating policy of never-ending improvement). Clearly it is in the
manufacturer's interests to get their suppliers to pursue perfection. Secondly, the
shape of the total-costs curve is similar to Huckett's (Fig. 2) total quality-costs
curve, which is based on actual data. Also, the form of the curves in the converted
model is not dissimilar to those produced from Kohl's data (Fig. 2).
The use of Crosby's (1979) stages of development of quality management as
abscissa complicates analysis and comparisons of the model because the relation-
ship between quality achieved and the state of development of a company's quality
management is not known. Whilst it is recognized that the model is only notional, it
does appear that the company's ideas of a quality-cost model are closer to the truth
than many others.
It is worth noting here that a realistic diagram based on Crosby's stages of
development can only be constructed after the stage of 'wisdom' had been reached,
i.e. with hindsight. This is because measurement of quality costs is among the
quality-management techniques being developed. Crosby suggests that attempts to
collect quality costs in the early stages of development will identify only about one-
sixth of them. Thus the diagram purports to represent actual costs, as opposed to
measurable costs, at each stage of development.

2.4. Group D
The models shown in Fig. 4 are similar in many ways to the models in Fig. I.
However, they have been segregated for discussion because they appear in pub-
lications devoted specifically to quality costs, and hence might be expected to reflect
accurately the relationships between the major cost categories and the cost savings
to be expected from investment in prevention or appraisal activities. In addition,
two of the models-BS 6143 (1981) and Urwick Group (1981)-are widely publicized,
and carry the authority and prestige that attaches to publications of the BSI and to
a leading firm of management consultants.
The quality-costs model published in BS 6143 confuses and misleads. It does so
by failing to separate prevention and appraisal costs, thereby suggesting that
reduction of failure costs by increasing the combined expenditure on prevention and
appraisal is the prime strategy for reducing quality costs. In fact, reduction of
appraisal costs by increasing preventive measures may be a reasonable and worth-
while alternative to reducing failure costs. Indeed, in some cases it is the easiest and
quickest way to achieve substantial reductions in quality costs. However, expendi-
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

Quality costs 1721

BS 6143 1981

.,
~,
~I
HARRINGTON 1976
~I
f:S:ISI
.'
01

0'
~I
~,
rz.z.a
C"~l~O"A81.E OVA",,,

TOTAL' .... ""E CO"S

cos's

TOTAL COSTS I
,,
I

I
.~
I
,
I s

MEASURE OF QUALIfY PRODUCT DEFECT lEVEL

(QUALITY LEVE;L)

Reproduced by permission of the Reprinted by permission from QUALlTY,

British Standards Institute a Hitchcock Publication

URWICK GROUP 1981 URWICK GROUP 1981

FALlflE COSTS
, B with data
85% 10'l1.

.§

W O f ' l S E _ QVALITY _BE;T1E~ W O R S E _ QuALITY _ B E l T E R

Reproduced by permission of the Urwick

Management Centre/Price Walerhous~
Figure 4, Group D.

ture on appraisal, above a certain threshold level, tends to move costs from external
to 'internal failure categories, and savings derive mainly from detecting faults sooner.
(It seems curious that in view of the strength of inspection activities in industry, and
recently, the attention paid to sampling plans and techniques, the overall economics
of appraisal activities are not discussed in depth in the quality-cost literature.)
The model misleads by suggesting:
(i) that huge reductions can be achieved at little cost;
(ii) that the economic balance occurs at a point where 70% of the total quality
costs are failure costs (corresponding closely to the quality-cost distribution
labelled 'typical' in the text of the Standard); and
(iii) that the optimum quality measure lies approximately two-thirds of the way
across the quality-level range. (This point may be countered by arguing that
the quality range represented is only the high end of the range, but this
would not be consistent with the way prevention and appraisal costs are
represented in the diagram.)
The Urwick Group's (1981) model in their booklet Quality Costs-A Key to
Survival is even more confusing and misleading. In their (praiseworthy) attempts to
promote expenditure on prevention, they separate out prevention costs, but
inexplicably combine appraisal and failure costs in their model (Fig. 4). (Note: there
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1722 J. J. Plunkett and B. G. Dale

is some ambiguity about the latter because a point on the 'Appraisal/Failure Costs'
line is labelled 'Halving failure costs'.) Like the model from BS 6143, it suggests very
substantial returns on investment in prevention, and that the optimum may lie some
two-thirds of the way across the quality spectrum. However, it differs in the respect
that optimum quality in this model coincides with a point where prevention costs
are about equal to the combined appraisal/failure costs. (There may be instances of
such quality-cost distributions, but none has been met with in our research.) It is
interesting to note, however, that their 'balanced costs' position does not coincide
with the minimum total cost, but has a 70: 30 split between appraisal/failure costs
and prevention costs, corresponding exactly to the BS 6143 split between failure and
prevention/appraisal costs.
In the same booklet the Urwick Group publishes quality-cost data from a
typical sample of clients as in Table I. These data have been superimposed by
Plunkett (1986) on to a version of their quality-costs model by converting absolute
costs to percentages, with interesting results (Fig. 4). It will be seen that all of their
typical sample of clients are well to the left of the optimum, suggesting rather poor
quality performance. It is difficult to accept that seven typical companies from such
a diverse range of industries all have mediocre standards of quality performance.
(Standards of quality performance in this context means performance with respect
to costs. It does not necessarily mean that the companies market mediocre-quality
products. So far as their customers are concerned they may manufacture excellent
products. What is implicit in quality-cost models is that improved quality and
quality-cost performances accrue from increased expenditure on prevention activ-
ities. It does not follow that when prevention expenditure is relatively low, outgoing
product quality is poor.) Hence one is led to suspect the model and/or the data may
be inaccurate. Closer examination of the data shows the failure costs to be 66-94%
(average 85%) of total costs, which seems high by comparison with other published
data. It may be that the Urwick Group's methods of calculating failure costs tend to
maximize them (e.g. by valuing scrap at sales value). Consultants often maximize
costs they are trying to reduce, thereby maximizing claimed savings. If these failure
costs are deflated so that they average 70% (i.e. close to the generally accepted
average) the effect is only to move the average from A to B (Fig. 4), i.e. still well to
the left of the 'balanced costs' and optimum costs positions, thereby suggesting that
the greater inaccuracy lies in the model.

Annual quality costs

No. of
Industry employees Failure Appraisal Prevention Total
Footwear 750 1,100,000 175,000 25,000 1,300,000
Board manufacturing 500 800,000 40,000 10,000 850,000
Building systems 750 120,000 50,000 10,000 180,000
Foundry A 200 350,000 50,000 15,000 415,000
Foundry B 300 500,000 50,000 30,000 580,000
Mechanical handling 300 500,000 30,000 5,000 535,000
equipment
Glass manufacturing 150 200,000 20,000 10,000 230,000
Overall percentage 87·5 10 2·5 100
of total

Table 1.
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

Quality costs 1723

CAMPANELLA & CORCORAN 1983

JURAN 1980 '0 '0

\
O.~
rOTAL
00
T CXl
00
LOSS OUE TO
omen",
(FALlJlE COST)

~
§• _~~~~_~~_~~T

COST OF APPRAISAL QPnl>Uo4POINT

PLUS PREVENTION

100'110
DEFECTIVE QUALITY OF CONFORMANCE

Reproduced by permission of the © American Society/or Quality Control.

McGraw-Hill Book Company Reprinted by permission
THODAY 1976 VEEN 1974

HCREASl'la QUALITY _ Pert.cllo"

DEGREE OF' PERFECTION ...
Reproduced by permission 0/ the Reproduced by permission of
Institute of Quality Assuraru:e the EOQC
Figure 5. Group E.

2.5. Group E
Although the models in Fig. 5 have broadly similar characteristics to the models
in Figs 1 and 4 in that they indicate the possibility of balancing the different types of
quality costs at some optimum level, they differ in two important details: (i) the rate
of change of cost with quality is much greater; and (iii the optima are positioned
much closer to perfection.
Campanella and Corcoran's (1983) diagram is so similar to Juran's version that
it must surely have been taken from the earlier publications of either Juran (1974) or
Juran and Gryna (1980). Although it is clear that the models are only notional,
effects are clearly exaggerated, as indicated for example by the idea of costs increas-
ing to infinity, which of course they cannot. It also seems curious that a model
suggesting such a pronounced and precise optimum should appear alongside real
data (Campanella and Corcoran 1983-Fig. 2) which exhibit no such characteristic.
It is clear from the text of Veen's (1974) paper that his model is the result of
experience and some careful thought. It is interesting to note his view that the rates
of reduction in both appraisal and failure costs can be maintained constant almost
to the point of product perfection-but only by a rapid increase in prevention
expenditure. The position on the scale of quality at which this occurs (98% of
perfection) endorses the views implied (but not quantified) in the other models in
this group. It is interesting also to note Veen's implied view that the optimum may
be attained with a constant and relatively low level of expenditure on prevention
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1724 J. J. Plunkett and B. G. Dale

activities. A useful point paraphrased from the text of Veen's paper is that, in the
case of a market which inclines towards consistently higher quality demands, it may
be advantageous to operate to the right of the minimum cost. This means paying
relatively greater attention to prevention. In this wayan organization is always one
step ahead of developments and can think about tomorrow's problems instead of
having to devote all its energy to today's problems.
Thoday's (1976) model (which is also reproduced by Chisholm 1982), incorpo-
rates the view that total quality cost has two main components, the cost of achiev-
ing quality and the cost of failure to achieve quality. He calls these internal and
external quality costs respectively. By Thoday's definitions internal quality costs are
made up of prevention and appraisal costs, whilst external quality costs comprise
external failure costs. Thus his cost groupings correspond with those of the BS 6143
model. (Thoday's external failure costs contain elements which some practitioners
may not readily accept as external failure costs or even as quality costs, e.g. preven-
tive maintenance on products; loss of goodwill, including loss of customers and
additional sales promotion necessary if quality level is too low; and national work-
shops.)
However, Thoday's model is also interesting in two other respects. Firstly, it
postulates the notion of minimum levels of costs, which increase rapidly as one
approaches or departs from perfection. Hence the optimum cost occurs near the
high end of the quality spectrum, not two-thirds of the way across as in the BS 6143
and Urwick Group models. The second feature is that Thoday claims that the
rapidly rising costs follow exponential curves. (Chisholm 1982 states: 'It can be
shown that the cost of achieving quality increases as an exponential curve as quality
approaches perfection. The cost of failure similarly increases exponentially as
quality deviates from perfection' and cites Thoday 1976 as his source. In the writers'
view the curves are not 'shown' to be exponential; they are only claimed to be so.]
When the 'internal' and 'external' costs are added together the resulting total cost
goes through a minimum value. It is but a small step from the concept of a
minimum cost to the notion of optimal quality as presented in BS 6143, and the
Urwick Group notional quality models. The notional models of Veen, Lockyer, and
the automotive manufacturing company (Fig. 3), and the data of Campanella and
Corcoran, Kohl, and Huckett (Fig. 2), do not appear to support the idea that the
curves are exponential, or that there is an optimum level of quality corresponding to
minimum quality costs. Krzikowski's data alone show a slight upturn in total costs
at a high level of quality for reasons which have been explained.

3. Concluding remarks
Although quality-cost models have been taken from a number of sources and
grouped and compared with each other to bring into question the validity of the
popular notional models, the legitimacy of doing so may itself be questioned. For
example, although all the models (with the exception of Lockyer-Fig. 3) use absol-
ute cost or a simple variation of it as the ordinate, abscissae range from measures of
quality dimensioned only at their extremes to undimensioned process capability,
arbitrary quality-management development stages, or time. It is clear also from the
contributions of Huckett (1985) and Thoday (1976) that different authors may
include among their quality costs items which other practitioners would not readily
accept as being quality costs. In addition, combining different categories of failure
costs, and prevention with appraisal costs, as in most notional models, or combin-
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

Quality costs 1725

ing appraisal and failure costs, as in the Urwick Group model (Fig. 4), obscures the
relationships between categories of activity and between their associated costs. It
may also be argued that the diagrammatic representations are only notional and are
intended only to illustrate principles. Nevertheless, it is felt that the comparisons
and criticisms made are valid and that even notional models should reflect, so far as
is possible, the existing state of knowledge. It is important to get it right, not just to
provide a predictive tool but also because it can have a strong bearing on the
quality philosophy of a company. Indeed, the authors believe that the notion of
optimum quality levels has seriously inhibited the development of total quality man-
agement in companies.
It may be, of course, that hard evidence is available to support the notion that
quality-cost curves are exponential and give rise to cost minima, but the authors
have not found it. It would be unfortunate if the notion of an optimum quality cost,
given such prominence and wide publicity, was based on speculation.
Some of the models imply nonsensically extravagant returns on investment in
prevention and/or appraisal. Indeed, it is noticeable that whenever investment in
prevention expenditure is discussed there is an implicit expectation of a large return
on the investment. While recognizing that companies are in business to make profit,
and that there may be competition for limited resources, the paybacks expected (and
sometimes proffered) from investment in quality seem unusually high, and since high
returns are usually associated with high risk, one wonders whether, as a general
rule, the benefits from investment in prevention are seen in business circles as being
problematical.
The view that many of the widely publicized quality-economics models are inac-
curate, and may even be of the wrong form, is widely supported by quality man-
agers. Everyone would like to have a valid model which they could use to assess
their present situation and predict the effects of changes. So far as is known there
are insufficient data available to construct such a model, though there should be
enough collective experience available to make a reasonable' hypothesis as to the
shape and the relative proportions of the constituent costs in the diagram. For
example, there are reasonably good grounds for believing that optimum quality, if it
exists, is near to the highest attainable standard of quality. The model should take
account of the fact thai investments in prevention through capital or resources often
involve relatively large stepwise increases. It should also reflect the often consider-
able time lag to be expected between investment in prevention and/or appraisal and
reduction in failure costs, and especially warranty costs.

Acknowledgments
The financial support of the Science and Engineering Research Council, through
their research contract GR/C33475 which led to the production of this paper, is
gratefully acknowledged.
The authors also wish to thank the publishers of the models shown for per-
mission to reproduce them in this paper.

References
ANON., 1985, Statistical Process Control Course Notes (Company A).
BESTERFIELD, D. H., 1979, Quality Control (Englewood Cliffs, New Jersey: Prentice-HalI).
BS 6143,1981, Guide to the Determination and Use of Quality Related Costs (London: British
Standards Institution).
CAMPANELLA, J., and CORCORAN, F. J., 1983, Principles of quality costs. Quality Progress, 16,
No.4, 16-21.
Downloaded By: [HEAL-Link Consortium] At: 10:32 16 July 2008

1726 Quality costs

CAPLEN, R. H., 1972, A Practical Approach to Quality Control (London: Business Books).
CHISHOLM, C. U., 1982, Quality assurance: a review of production progress. Quality Assur-
ance, 8, No.2, 55-60.
Cox, B., 1979, Interface of quality costing and terotechnology. The Accountant, 21 June,
800-801.
CROSBY, P. B., 1979, Quality is Free (New York: McGraw-Hili).
HAGAN, 1. T., 1986, Quality Management Handbook (eds. Walsh, Wurster & Kimber), (New
York: Marcel Dekker).
HARRINGTON, J. H., 1976, Quality costs-the whole and its parts-part 1. Quality (Illinois), 15,
No.5, pp. 34-35. .
HUCKETT, J. D., 1985, An outline of the quality improvement process. International Journal of
Quality and Reliability, 2, No.2, 5-14.
JURAN, J. M. (ed.), 1974, Quality Control Handbook (New York: McGraw-Hili).
JURAN,1. M., and GRYNA, F. M., 1980, Quality Planning and Analysis (New York: McGraw-
Hill).
KIRKPATRICK, E. G., 1970, Quality Control for Managers and Engineers (Chichester: Wiley).
KOHL, W. F., 1976, Hitting quality costs where they live. Quality Assurance, 2, No.2, 59-64.
KRZIKOWSKJ, 1963, Quality control and quality costs within the mechanical industry. Pro-
ceedings of the 7th EOQC Conference, Copenhagen, pp. 129-145.
LOCKYER, K. G., 1983, Production Management (London: Pitman).
MORTIBOYS, R. J., 1985 (Consultant to the National Quality Campaign}--private communica-
tion with Plunkett.
PLUNKETT, J. J., 1986, A study of the collection and use of quality related costs in manufac-
turing industry. Ph.D. Thesis, UMIST, UK.
PLUNKETT, J. J., and DALE, B. G., 1986, Quality costs: the economics of some engineering
practices. Chartered Mechanical Engineer, 33, No. 11,33-35.
RICHARDSON, D. W., 1983, Cost benefits of quality control-a practical example from
industry. BSI News, October 14-15.
ROBERTSON, A. G., 1971, Quality Control and Reliability (London: Pitman).
SEKIYA, S., 1985, Quality control at Toyota Motor Corporation. Productivity Digest, 4, No.2,
68-87.
THODAY, W. R., 1976, The equation of quality and profit. Quality Assurance 2, No.2, 48-52.
URWICK GROUP, 1981, Quality Costs-a Key to Survival (Urwick Group pamphlet).
VEEN, B., 1974, Quality costs. Quality (EOQC publication), No.2, pp. 55-59.
Tandis qu'ils effectuaient une etude de la collection et de I'utilisation des
rapports cout-qualite dans les industries de fabrication, lea auteurs ont decouvert
dans la documentation de nombreux modeles notionnels pretendant indiquer les
relations entre les principales categories de couts de qualite et quelques ensem-
bles de donnees reelles, Malgre Ie fait qu'ils soient fondes sur des principles
comrnuns, iI existe d'importantes differences entre certains des modeles et entre
les modeles et les donnees reelles, L'article categorise et examine les modeles a la
lumiere de l'experience de la recherche. II en conclut que beaucoup de modeles
sont incorrects et trompeurs et mettent serieusement en doute Ie concept d'un
niveau de qualite optimale correspondant a un point minimum sur la courbe
totale du rapport cout-qualite.

Bei einer Untersuchung der Erfassung und Verwendung von qualitatsab-

hangigen Kosten fanden die Verfasser in der Literatur zwar viele abstrakte
Modelle, die Anspruch darauf erheben, die Beziehungen zwischen den hauptsach-
lichen Qualitatskostenkategorien aufzuzeigen, aber nur wenige echte Daten.
Obwohl die Modelle auf den g1eichen Grundsatzen basieren, bestehen merkliche
Unterschiede sowohl zwischen einigen der Modelle als auch zwischen den Mod-
ellen und den wirklichen Daten. Diese Abhandlung kategorisiert und bespricht
die Modelle aus der Sicht der bei der Untersuchung gemachten Erfahrungen.
Wir kommen zu dem Ergebnis, daf viele der Modelle ungenau und irrefiihrend
sind, und ziehen die Vorstellung einer optimalen Qualitatsstufe, der ein
Minimum auf der Giitegesamtkostenkurve entspricht, ernstlich in Zweifel.