|
Manufacturing A Winner: Salute Uncertainty
Why are bookmakers usually richer than punters? Why for that
matter, are casinos, football pool companies, insurance companies,
fruit machine companies, seldom pushed for cash? It is easy
to dismiss their success as 'working the odds', but it is
more instructive to learn from their success.
A bookmaker manipulates the odds he offers and the extent
to which he lays-off bets, not particularly on his appreciation
of horse-flesh, but to ensure that he will profit in the long
term whatever the outcome of individual races. The punter
generally bases his strategy on one particular outcome. When
a UK or British company gambles, it more often than not adopts
the role of punter, rather than bookie.
Salute Uncertainty
Uncertainty abounds. Whether it relates to technology, fashion,
weather, politics, geology, ecology, domestic and foreign
economies, strikes, exchange rates or numerous other factors,
all organisations face uncertainty. Uncertainty means risk.
The risk is not necessarily borne directly. Nationalised industries
transfer the risk to the state; monopolies to the consumer;
easily definable risks can be underwritten by insurance companies.
For the company operating in a real market however, the most
important risks will not be transferable. They will relate
to the company's products, processes or services, and the
market to which they are offered. If this entailed no risk,
then it would provide no profit, since anybody could do it.
It is fashionable to advocate the minimisation of uncertainty.
Up to a point, that must be right - but it is deciding when
that point has been reached that is difficult. Reducing uncertainty
costs money and takes time. Spend too much of the former and
the profit has gone, spend too much of the latter and the
opportunity has gone.
In the long term, a company's competitive advantage is based
on the knowledge, skills and tools that it has accumulated.
These assets can reduce uncertainty and lessen risk, but in
a dynamic world they cannot eliminate it. History records
many companies, countries, dynasties and empires which were
overwhelmed by events they would presumably have chosen to
avoid, but which they failed to foresee.
Trying to avoid all risk can only cause stagnation. Risk
must be recognised explicitly and treated appropriately. That
is what this article is about.
Perhaps the common and mundane risk concerns the decision
to start making products now for which there cannot yet be
expected to be any firm orders. It is essentially the same
problem as buying plant and machinery, or hiring staff to
meet demand, way beyond that covered by 'orders-on-hand'.
The punter approach is to make a guess at what the future
is likely to offer, and proceed on that basis. The guess is
usually referred to as a forecast and can be produced by quite
sophisticated methods.
Unfortunately, this very sophistication can help to foster
an illusion of accuracy. A computer-produced forecast of 273
may indeed be a better forecast than a sales manager's hunch
of about 250 - but only in the sense that the probability
of the former is 2.1% whereas the probability of the latter
is 1.9%.
In horse-racing terms, this is nothing more than the observation
that, in races with large fields, the favourite seldom wins.
To make any headway, it is necessary to drop the perception
of forecasts as 'future facts' and adopt the language of the
bookmaker.
To say that the best forecast is 273 is not very helpful,
if the probability of an outcome within plus or minus 100
of it is only 60%. But, if the probability of being more than
5 in error is not more than 5%, then the implications are
very different.
This extra dimension is a quantification of uncertainty,
and it has profound implications. If a forecast is inherently
subject to a large element of uncertainty, then a number of
stratagems may be cost-effective:
- shorter manufacturing and supply lead times (even at higher
costs)
- smaller raw material purchases (losing bulk discounts)
- more versatile employees and machinery
- smaller production runs
- more frequent re-scheduling
- more capacity, accepting lower average utilisation
- expansion into more diverse markets, thus broadening the
risk base
- hire rather than buy decisions
- use of sub-contract labour to match the margin of uncertainty
- developing fast MIS systems, so that periods of ignorance
are minimised
- manufacture and assembly programmes which delay minor
product differentiation until the latest possible moment.
How was it for you?
One of the first steps in recognising the existence of uncertainty
is measuring the success of one's own forecasts by comparing
what actually happened with what was forecast. It is surprising
how few companies - even quite major ones - do this. The exercise
can be very enlightening:
- it may be possible to improve forecasts by removing systematic
errors
- the average error may be large enough to affect management
plans in one of the ways suggested above
- elaborate forecasting methods (whether manual or computer-based)
may be found to be no better (and sometimes worse) than
very simple methods such as using last month's sales as
a forecast for next month's. Abandoning inaccurate and over-elaborate
methods not only saves time and money, but it removes the
unwarranted illusion of precision.
If the errors are too small, then their cause should be investigated.
Using our horse racing analogy, it may be that the field
has been rigged. The obvious route to low errors to adjust
effort so as to meet the forecast.
Since it is much easier to reduce effort than to increase
it, a modest forecast allows scope for selective under-performance.
A company which always meets its sales targets is therefore
probably failing to meet market demand. In due course, rival
companies will mop up the unsatisfied demand and imperil both
forecasting accuracy and company profits.
Which one to run with?
In a dynamic world, no success lasts forever. Sooner or later
a cheaper or better product will emerge, fashion will swing
adversely, or a quite different means of satisfying the market
requirement will emerge. Innovation is therefore an essential
ingredient of success. It provides no guarantee however.
In fact, the economies of innovation are more closely related
to Darwinian self-selection than they are to a typical accountant's
project appraisal.
The more outrageous the innovation, the less likely it is
to succeed, but the greater its potential if it does.
Once again, the punter's vision and that of the bookmaker
differ:
- the punter:
- sees one project with a probability of success which he
subjectively assesses at 25%. The potential rewards seem
very attractive, but the market, product or process are
unfamiliar. He would prefer to back cost reduction programmes
which promise less, but are more likely to deliver.
- the bookmaker:
- combines the punter's one improbable project with nine
similar, but quite independent, ones. He calculates that
the likelihood of at least one of them succeeding is just
less than 95%, and that there is an almost evens chance
that three or more will come off. He backs all ten projects
initially, but is quite ruthless in withdrawing funds if
the odds get longer or more promising projects emerge.
The bookmaker's motives are not mere greed, but a recognition
that around the world other bookmakers (with perhaps less
to lose and more to gain) are also gambling on projects whose
success would be his loss.
Since there is no chance that all those competing bookmakers
will always be unsuccessful our bookmaker recognises that
a 'do nothing' strategy is inevitably fatal in the long term.
In practice, the bookmaker viewpoint of product development
has to be interpreted according to the institution adopting
it:
- for the small (embryo) company it provides the motivation
to keep trying - large companies are vulnerable, particularly
if they have become complacent
- for the successful medium-sized company, a little modesty
is advised. Luck invariably has a role in all successes,
even if the element of hard work seems overwhelming. But
luck needs to be courted continually.
- for the large company, it suggests that internal diversity
should be welcomed and exploited. Competitive projects should
be encouraged and the failures accepted as inevitable, rather
than the result of some fundamental weakness. Using the
horse-racing analogy again: unlikely horses should be ruthlessly
withdrawn from the race, but their jockeys should not be
condemned.
- for politicians, this is perhaps another reason why centrally
planning isn't very effective: however good it is, one plan
seldom prove as good as the best of several.
And there's more
Acceptance of uncertainty has other implications. Deming,
the American Statistician, best known for his post-war contribution
to the Japanese quality revolution, does not favour payment-by-result
schemes.
He argues that to the extent that success is a matter of
luck, there is little point in rewarding it, since the recipients
cannot consistently repeat it and the unlucky under-performers
are unnecessarily demoralised.
In these circumstances our bookmaker would only consider
rewarding sequences of better-than-average performance which
he considered unlikely on the basis of chance alone.
To conclude
Assuming that it is possible to predict the future with accuracy
provides an comfortable illusion of security. Single point
predictions also make the arithmetic of many financial and
commercial calculations very much easier. In fact it is so
much more convenient and comfortable to take the 'punter view'
that whole organisations collude with each other to brush
aside or minimise the 'What .. If...' questions.
Unfortunately the world is uncertain and the
rewards are greater, and the disappointments less, for those
that recognise this fact.
_______
|
