Section 8b Economics of Quality Notes
Section 8b Economics of Quality Notes
Section 8b Economics of Quality Notes
Section
The Economics
of Quality
QUALITY MANAGEMENT
8
Section
Introduction
T raditionally, it has been common for the measures of quality within industry
to be in the form of indices such as: percentage scrap, rework hours,
number of warranty claims etc. These measures have not proved
themselves to be an effective report to management. It is difficult to get worked
up about fractions of percentage points! The tendency to roll figures together has
also been instrumental in muddying the water. If we create an agglomeration of
percentage measures of scrap, rework and warranty claims, who is to say if a figure
of 0.65 is good or bad? At what point do we get excited? At what point do we get
mad?
The result of the confusion caused by the use of somewhat arbitrary indices for the
measurement of quality has been the low priority given to quality issues compared
to those issues for which hard cash figures have been available. It is, after all, not
unreasonable for management to concentrate on areas where they can identify the
financial consequences of their actions.
The challenge, then, was to represent quality issues in a more meaningful way for
management action (this means in terms of money!). Businesses are all about
money and this reporting format ensures that quality is on the same footing as
other major issues. Armand V. Feigenbaum took up this challenge in his book
“Total Quality Control” published in 1951. The American Society of Quality
Control (ASQC) added a publication called “Quality Costs - What and How” and
latterly the BSI have published BS 6143 “Guide to the Determination & Use of
Quality Related Costs”. These all give advice, although it is sometimes conflicting,
on setting up and running quality costing systems.
1
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
bottom line of the organisation since they all represent money spent doing things
wrong or doing the wrong thing.
Secondly, it can be used to pinpoint areas where action needs to be taken, prioritise
the areas for action and give an indication of the resources that should be
committed to improvement in that area.
A survey by UMIST in the late 1980’s showed that less than 35% of companies are
aware of their Quality Costs and less than that actually employed a Quality Costs
system as prescribed in the literature. A few companies had no identification of
basic costs such as scrap and rework. It is unlikely that the situation has improved
significantly in the intervening few years.
Prevention costs will include things like: training, quality planning and
administration, any costs incurred in running quality circles and supplier
development activities. Appraisal costs will involve: inspectors time; purchase,
running and calibration costs of inspection equipment; report approval etc. Failure
costs will include: scrap, rework, design changes to solve problems created by the
primary design (all internal costs) and warranty (external).
2
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
These categories appear simple enough at first glance. However, there are, in fact,
large grey areas and much debate is to be had about activities such as first off
inspection (prevention or appraisal?), dealer PDI checks (failure or appraisal?)etc.
Whilst there may or may not be a ‘correct’ answer it is very important that these
issues are addressed at an early stage and a consistent approach adopted.
Quality Costs within an organisation will tend to be broken down as in the figure
shown below. The exact percentages will vary from company to company but the
general form will remain the same. Similarly, the general point that a correctly
targeted increase in prevention costs will produce a more than compensating
reduction in failure and appraisal costs is true although the ratio will vary from case
to case.
Overall Reduction
(both appraisal &
failure reduced)
Failure Cost
Appraisal Cost
Prevention Cost
The illustration above shows the second weakness of the traditional Quality Cost
system, that is the complexity of determining the effect of an increase in prevention
spending on the other categories.
Phil Crosby believed that the splitting of quality costs into three categories was too
complex and unnecessary. He proposed two simple categories based upon the idea
that the definition of Quality was “conformance to requirements”, where the
requirements were set by the customer (internal or external, as appropriate). He
3
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
argued that, if you were not undertaking normal business activities, you were taking
action either to ensure that something was right first time or to deal with the fact
that it had not been done right first time. The former he called “The Price of
Conformance” (POC) and the latter, which includes inspection since this is done
when we are not sure we have done it right first time, “Price of Non-
Conformance” (PONC).
In this simplified scenario the categorisation process is easier to achieve and the
“bad costs” (PONCs) can be easily separated from the “good costs” (POCs) and
the interaction more clearly seen.
We should note at this point that Quality Costing documentation does not form
part of the normal accounting structure. It can be seen as an alternative view of the
disposal of finance and fulfils the role of a management-reporting document to
target waste in the business; a function not adequately covered by conventional
accounting structures.
The level of detail to which Quality Costs are taken is very much a matter for the
individual organisation. It is frequently argued that the Quality Cost report should
only take the data that currently exists and collate it into a different structure rather
than establish an entirely new system. This may be possible in a system that has a
high level of detail in its present form but may be somewhat problematical if there
is insufficient depth. A Quality Cost system requires a high level of definition so
that costs can be attacked appropriately, too little detail may obscure the actions
required for maximum benefit. To be weighed against this is the additional effort
and complexity engendered by operating a system of Quality Costing that is all-
embracing. The principal rule to apply is to keep the system as simple as possible
whilst fully satisfying the organisation’s requirements.
4
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
show up in different departments to the one where the problem exists (and thus
the one capable of resolving it), this can lead to departments taking the “blame-
thrower” approach if a negative management culture, as discussed in earlier
sections, is present. Much care may, therefore, be necessary to allocate costs to
source rather than where they become visible. The correct environment for using
Quality Costs is one where the emphasis is on continuous improvement.
A pilot study of Quality Costs in a part of the organisation may be a useful starting
point before going on to develop a company-wide system incorporating any
lessons learned in the process.
Very careful examination of what is to count as a Quality Cost and what is not
should be undertaken at the planning and implementation phases. Do we include
overheads for example? The conventional answer is ‘no’ since most of the costs
we are dealing with are themselves overheads. Do we include the costs of lost
customers? Again, the conventional answer is ‘no’ since this would largely be
guesswork. Do we include the redundant material caused by a design change?
Here, the answer is usually ‘yes’ since it was the original error that caused redundant
material to be made. Consistency is the watchword; apply the same logic to each
situation as it arises.
The last paragraph illustrates the problem with Quality Costing; it is somewhat
subjective and does not fully include all potential consequential costs (the sort of
costs such as lost customers which Deming describes as “Unknown and
Unknowable”). It is, however, better than the ignorance in which many
organisations currently live and, if you understand the shortfalls can still provide a
very powerful tool for driving and monitoring continuous improvement.
Review all activities in the prescribed area and determine into which of the three
categories below they fall:
They are done because something has not been done right the first
time (PONC).
They are done to ensure that subsequent activities are done right
the first time (POC).
They are part of the way in which we would normally run the
business if everything was done right the first time.
5
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
value to the organisation! This was clearly a nonsensical statement that damaged
the initiative to evaluate Quality Costs almost irreparably.
The first two categories should be measured in some sensible way; hours, heads etc.
in addition to material losses. This information should be passed to the accounts
department for financial evaluation and then collated as the Total Cost of Quality.
The subsequent breakdown of the figures should be calculated to assist the process
of improvement in the organisation. This may vary from organisation to
organisation but will generally be devolved to departmental level with the focus on
major PONC’s.
If neither of the above reasons accounts for the increase you will need to revisit the
system design to see where the process is failing. The most likely place is the
translation of the data into improvement actions.
6
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
G enichi Taguchi is a Japanese engineer who has become one of the most
recent additions to the Quality “Hall of Fame”. He has attained pre-
eminence due to two significant contributions to the field of Quality: the
first is the so-called ‘loss function’ which will be discussed below and the second is
off-line QC involving a novel approach to the design of experiments which is not
dealt with in this module.
Lower Upper
Tolerance Tolerance
Cost
Measured Characteristic
Taguchi Loss Function
Cost
Measured Characteristic
Traditional thinking is that any product that falls inside the tolerance limits is
“good”. The unspoken assumption here is that they are equally good and that no
cost is incurred. Following the logic through we can see that any product falling
outside the limits is bad and a cost equivalent to the full cost of producing that
product is incurred (often referred to as the scrap cost). In this simple scenario we
7
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
have assumed that reworking the product is either not possible or is uneconomic.
Again, the hidden implication is that all products outside the limits are equally bad.
The usual derivation of tolerances further throws this attitude into doubt. They are
usually based upon what was done last time or the draughtsman’s ‘best guess’.
There also exists an element of barter in the generally adversarial relationship
between design and production with designers wanting to tie production down to
extremely tight tolerances and manufacturing wanting to be able to drive a bus
through them. Seen in this light tolerances can be viewed as, at best, somewhat
arbitrary.
Taguchi states that to regard the transition from good to bad as a step change is not
logical. He contends that, provided the nominal has been specified correctly, any
deviation from this target value will have a detrimental effect on the performance
of the product and will therefore cause an overall “loss to society”. This concept is
probably one of the more esoteric of Taguchi’s ideas. A good example may be to
consider the thickness of a polythene sheet used by farmers to protect crops; if the
sheet thickness is low (but within tolerance) it may tear more easily and allow the
weather to damage the crops. The costs generated by this failure will be outside the
company but very real. Firstly, farmers will incur additional replacement costs;
secondly, the reduced crop yield will increase the price in the marketplace, a loss
borne by all society. Since the loss is usually greater than the company’s gain from
selling a poorer product the Japanese have coined the phrase “worse than a thief”
which refers to the fact that when a thief steals money only his victim is worse off
and not society as a whole.
Sony, Japan
Sony, USA
C B A A B C
8
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
The above diagram is an illustration of the loss function as a long-term loss to the
company, and appeared in a Japanese newspaper called “The Asashi” in 1979. The
article discussed the preference of American consumers for television sets built by
Sony in Japan over those built at an identical plant in the USA. The key
performance characteristic is colour density. The ‘A’ band represents excellent
colour density; the ‘B’ band good colour density and the ‘C’ band acceptable colour
density. Outside of the limits of the ‘C’ bands colour density is deemed
unacceptable and the TV is considered a reject.
Clearly from the figure, Sony Japan was producing defects whilst the American
plant was not. However, the key fact is that the chances of having an ‘A’ or ‘B’
grade TV from the Japanese plant was much greater than the American one where
the odds of getting any grade were roughly similar. The “in tolerance is OK”
attitude was costing a lot of sales for the American plant. The Taguchi belief that
variation from the nominal is expensive seems much closer to the truth in this case.
Taguchi states that the loss function takes the quadratic form shown above for all
“nominal the best” type characteristics and the appropriate half of that shape for
“bigger the better” and “smaller the better” features. Whether this is in fact strictly
the case is debatable. However the principle that deviation from the target is
expensive regardless of tolerances and that the rate of deterioration of the situation
increases with distance from the target is sensible.
The principal implications of the loss function for the way we do business is that it
gives a financial credibility to the argument for continuous improvement of
processes beyond the current specification limits and it firmly indicates the
reduction of variation in our processes as the way to increased profitability and
success. In addition to this, if we are able to quantify the important constant in the
loss function, we can see in financial terms the benefits of reducing the variation in
a process and set this against the cost of doing so. This gives a very good basis for
evaluating improvement projects either in absolute terms or against other projects
competing for the same funding.
It is, however, difficult to accurately define the equation for the loss function for
any given process the general form taken is shown below but the constants are
difficult to determine and are frequently underestimates due to the unpredictability
of consequential losses.
L(y) = k(y-m)2
Where:
k ........... is a constant for the given characteristic. It depends upon the importance
of y.
9
©Warwick Manufacturing Group
Revision date: October 06
T H E E C O N O M I C S O F Q U A L I T Y
µµµ
10
©Warwick Manufacturing Group
Revision date: October 06