6.3 Use of probabilities
Clearly, these approaches to risk reduction are not without cost. Market research
costs money. A supplier who is required to be committed to a fixed price contract, or
option, would need that fixed price, or the price of the option, to take account of the
risk that its own costs might increase. Insurance businesses do not cover risks for noth-
ing. We may feel, however, that these are prices worth paying to reduce the risk to a
point where the project would become acceptable.Problems of using sensitivity analysis
Despite its benefits to the decision maker, sensitivity analysis has its weaknesses.l Sensitivity analysis does not provide us with any rule as to how to interpret the
results. We could carry out more research and possibly reinforce our confidence in
the original estimates, but often the doubt can never be completely removed.
l The sensitivities are not directly comparable from one factor to the next. In Example
6.1, it might seem that the life of the project (14.8 per cent sensitive) is less of a prob-
lem than the amount of the initial investment (13.7 per cent sensitive). However,
this is not true. The initial investment is to be made immediately and, for that rea-
son alone, is likely to be known with more certainty than the life reaching well into
the future. Also, if it is found that the cost of making the initial investment creeps
to above £56,865, the project could be cancelled. Only when the project has been
undertaken and finance committed would we discover whether or not the estimate
of the project’s life was overoptimistic.6.3 Use of probabilities
Perhaps more useful than looking at the amount by which a particular factor may vary
from its estimate, before it renders a project unprofitable, is to look at how likely it is
to do so. If the whole range of possible outcomes for each factor could be estimated
together with each one’s statistical probability of occurrence, a much fuller picture
would be obtained. These could be combined to discover the range of possibilities and
probabilities for the NPV of the project as a whole. Before we go on to look at how this
could be done, it should be pointed out that reliably identifying possible outcomes
and their probabilities is very difficult to achieve in practice.
In ascribing probabilities to various aspects, decision makers might use either or
both of the following:l Objective probabilities. These are based on past experience of the outcomes and
their likelihood of occurrence. For example, if we know what the annual demand
for a particular product has been over the years, we could assume that this would
define the possible outcomes and their probabilities for next year. Suppose that dur-
ing each of six of the past ten years the demand for the product had been 2 million
units and it had been 3 million for each of the other four years, we could conclude
that next year’s demand will either be 2 million or 3 million with a 0.6 and 0.4 prob-
ability, respectively. If we have reason to believe that the past is not a good guide
to the future because, for example, the product has just been superseded by a more
technologically advanced version, this approach could not be justified.‘