CHAPTER NINE

Decision making under

Uncertainty
Introduction
Decision making, particularly long –term decisions, has to be taken under conditions of risk and uncertainty.

WHAT IS RISK AND UNCERTAINTY?

Uncertainty
Uncertainty simply reflects that there is more than one possible outcome for a given event but there is little previous
statistical evidence to enable the possible outcome to be predicted.

Risk
Risk is where that uncertainty can be quantified in some way.

It is normal to quantify the risk in terms of a probability distribution, generally derived from statistical data in the past.

Methods
There are several methods of dealing with risk in decision-making. The methods which are dealt with in this syllabus
are:

 Expected values
 Value of perfect information
 Decision trees
 Maximax
 Maximin
 Minimax regret

You are also required to be aware of the practical ways of dealing with uncertainty.

Expected values
The concept
The expected value represents the average outcome which would be achieved if a decision were to be repeated many
times.

Expected value (EV) = weighted arithmetic mean of possible outcomes

=  (x, p(x))

This formula represents the sum () of each possible outcome (x) multiplied by its probability of occurring
(p(x)).

Question 1

Decision Making Under Uncertainty Page 1

A decision maker must select one of three mutually exclusive projects. The outcome from each project depends on the
state of the market, which can be diminishing, static or expending. The profit for each project under each of the three
outcomes is shown in the following payoff (profit) table:
Project Project Project
1 2 3
State of the marketing Probability
Diminishing 0.4 100 0 180
Static 0.3 200 500 190
Expanding 0.3 1,000 600 200.

Required:
Determine the expected value (EV) for each project and select which project to pursue.

Advantages of EV
 It reduces the information to one number for each choice.
 The idea of an average is easily understood.

Limitations of EV
The probabilities of the different possible outcomes may be difficult to estimate.
The average may not correspond to any of the possible outcomes.
Unless the same decision has to be made many times, the average will not be achieved; it is therefore unsuitable
for decision-making in ‘one-off’ decisions.
The average gives no indication of the spread of possible results (i.e. it ignores risk).
The accuracy of the results depends on the accuracy of the probability distribution used.

Profit Tables
A profit table (payoff matrix) shows all possible ‘payoffs’ (NPVs, contribution, profits, etc) which may result from a
decision-maker’s chosen strategy.

Question 2
A baker pays 10c per cake and sells each cake for 30c. At the end of a day any cakes not sold must be thrown away.
On any particular day the level of demand follows the following probability distribution:

Number of cakes sold 20 40 60

Probability 0.3 0.5 0.2

Required:
(a) Construct a profit table to show the possible outcomes.
(b) Calculate the daily order the baker should place in order to maximise the expected value of daily profits.

Value of Perfect Information

The Concept

Imagine that, in a situation of uncertainty, it is possible to buy an accurate forecast which predicts with certainty what
the uncertain variable is going to be each time a decision has to be made.

The value of perfect information is the maximum amount a decision-maker would be willing to pay for advance
information to know which outcome will occur.

Decision Making Under Uncertainty Page 2

Even if the baker were to order a daily forecast which would provide him with accurate data about daily sales, he
would still face risk. The risk arises because although the daily forecasts would be accurate, the baker would not know
what the forecasts would say for each day.

Expected value with perfect information is the expected daily profit the baker would earn if he ordered the forecast
and acted on it. Some days the forecast will say that demand will be 20 cakes, other days 40 cakes and so on. It is
assumed that the probability distribution of the forecast is the same as the probability distribution of the underlying
variable (i.e. demand for cakes).

Value of perfect information = EV with perfect information – EV without perfect information

Question 3
A baker pays 10c per cake and sells each cake for 30c. At the end of a day any cakes not sold must be thrown away.
On any particular day the level of demand follows the following probability distribution:

Number of cakes sold 20 40 60

Probability 0.3 0.5 0.2

The baker opts to buy a daily forecast which tells him in advance of placing the days order what demand for that day
will be with certainty.

Required:
Calculate the value of perfect information.

Decision Trees
The Concept
Decision-making often involves multi-stage decisions. At each stage of the decision-making process, the decision
maker has to choose between two or more decisions. The possible outcomes of each decision will be specified, along
with the associated probability. Having made the first decision, a second decision or possibly even more decisions
may be required.

A ‘decision’ tree helps to visualise and evaluate outcomes in the decision-making process. It is a pictorial
representation of the decisions which need to be made at each stage, along with other potential outcomes and
associated probabilities.

Conventions and Process

The following conventions are used in drawing decision trees:

 Decision fork (point) – this is a point at which a decision-maker has to decide between two or more decisions

Action a

Action b

Action c

 Chance fork (outcome point) – this occurs where there are several possible outcomes. Normally, for each
decision taken, there will be two or more possible outcomes.

Probability Outcome B

Probability Outcome A
Decision Making Under Uncertainty Page 3
Having drawn the decision tree, it is necessary to calculate the expected outcome at each decision fork. To do this
process, start at the right-hand side of the decision tree and work back to each decision fork to identify which is the
best decision at each fork.

Ultimately, the decision tree enables the decision-maker to determine the decision to make at the first stage.

Question 4
The following information relates to Seven Tree Ltd, a company which is considering whether to develop and market
a product.
Probability
Development
Being successful 0.75
Being unsuccessful 0.25

Estimated development costs would be $180,000.

If successful, the product will be marketed with the following probabilities:

Probability Profit / (Loss)
Being very successful 0.4 $540,000
Being moderately successful 0.3 $100,000
Being failure 0.3 ($400,000)

The above profits / losses figures include the effect of the development costs.
Required:
Draw a decision tree to illustrate the above problem, and recommend the best course of action.

Question 5
Slim Foods is considering launching a non-sugar snack bar into a new market. Although the company has not yet
undertaken any market research, the marketing director estimates that the product has a 60% chance of success and a
40% chance of failure in the market.

A market research company has offered to do research in the new market prior to any decision being made whether to
launch the product. Management believes that there is a 60% chance the market research will recommend the launch
and a 40% chance it will advise Slim Foods not to launch the product. The cost of the market research will be
$30,000.

The market research company has admitted that their research findings do not always turn out to be expected once the
product has been launched. They advise that if they recommend the launch, there would be an 80% chance the product
would succeed and a 20% chance it would fail. If they do not recommend the launch, there would be a 30% chance the
product would succeed if management were to launch it and a 70% chance it would fail.

In all cases, if the product succeeded, the present value of future profits from the product would be $10 million
(excluding the market research costs) and, if the product failed, the net present value of the loss would be $4.5 million
(excluding the market research costs).

Decision Making Under Uncertainty Page 4

The directors of the company are trying to decide whether to accept the offer of the market research company or to
make a decision based on the gut feel of the marketing director.

Required:
(a) Draw a decision tree to illustrate the possible decisions and their associated potential outcomes.
(b) Advise management how they should proceed.
(c) Calculate the value of the imperfect information provided by the market research company.

Question 6
Mesho Co is a manufacturer of baby equipment and is planning to launch a revolutionary new style of sporty
pushchair. The company has commissioned market research to establish possible demand for the pushchair and the
following information has been obtained.

If the price is set at $425 demand is expected to be 1,000 pushchairs, at $500 it will be 730 pushchairs and at $600 it
will be 420 pushchairs. Variable costs are estimated at either $170, $210 or $260.

A decision needs to be made on what price to charge.

Required:
(a) Produce a table showing the expected contribution for each of the nine possible outcomes.
(4 marks)
(b) Explain what is meant by maximax, maximin and minimax regret decision rules, using the information in
the information in the scenario to illustrate your explanations. (10 marks)

(c) Explain the use of expected values and sensitivity analysis and suggest how Mesho Co could make use of
such techniques. (6 marks)
(Total = 20 marks)

Question 7
A software company has just won a contract worth $80,000 if it delivers a successful product on time, but only
$40,000 if this is late. It faces the problem now of whether to produce the work in house or to subcontract it. To
subcontract the work would cost $50,000, but the local sub-contractor is so fast and reliable as to make it certain that
successful software is produced on time.

If the work is produced in house the cost would be only $20,000 but, based on past experience, would have only a
90% chance of being successful. In the event of the software not being successful, there would be insufficient time to
re-write the whole package internally, but there would still be the options of either a ‘late rejection’ of the contract (at
a further cost of $10,000) or of ‘late subcontracting’ of the work on the same terms as before. With this late start, the
local sub-contractor is estimated to have only a 50:50 chance of producing the work on time or of producing it late. In
this case the subcontractor still has to be paid $50,000, regardless of whether he meets the deadline or not.

Required:
(a) Draw a decision tree for the software company, using squares for decision points and circles for outcome
(chance) points, including all relevant data on the diagram.
(b) Calculate the expected values as appropriate and recommend a course of action to the software company
with reasons.

Risk Attitude and Decision Rules

Decision Making Under Uncertainty Page 5

Risk preference describes the attitude of a decision-maker towards risk – as there is a relationship between risk and
reward.

 Risk averse – a risk averse decision maker considers risk in making a decision, and will not select a course of
action that is more risky unless the expected return is higher and so justifies the extra risk.
 Risk seeker – a risk seeker decision maker also considers risk in making a decision. A risk seeker, unlike a risk
averse decision-maker, will take extra risks in the hope of earning a higher return.
 Risk neutral – a risk neutral decision maker ignores risk in making a decision. A risk neutral decision maker
will select the course of action with the highest expected return, regardless of the risk.

Decision Rules
Choosing between mutually exclusive courses of action on the basis of worst, most likely or best possible outcome
can be stated as decision rules.

The choice may be based on a maximax, maximin, or a minimax regret decision rule, and expected value.

Maximax

The decision maker will select the course of action with the highest possible pay-off (the best of the best).
The maximax decision rule is the decision rule for the risk seeker.

Maximin decision rule

The decision maker will select the course of action with the highest expected return under the worst possible
conditions. This decision rule might be associated with a risk averse decision maker.

Minimax regret decision rule

The decision maker selects the course of action with the lowest possible regret. It aims at minimising the regret from
making the wrong decision.

Regret is the opportunity cost of having made the wrong decision, given the actual conditions that apply in the future.

Expected value (EV) –

Select the option that gives the highest EV. Those who use EVs may be described as risk neutral (i.e. they are not
concerned with the amount of risk associated with each option only the amount of the expected return).

Question 8
A decision maker must select one of three mutually exclusive projects. The outcome from each project depends on the
state of the market, which can be diminishing, static or expending. The profit for each project under each of the three
outcomes is shown in the following payoff (profit) table:
Project Project Project
1 2 3
State of the marketing Probability
Diminishing 0.4 100 0 180
Static 0.3 200 500 190
Expanding 0.3 1,000 600 200.
Required:
Determine which project should be chosen, using each of the following decision rules:
(a) expected values
(b) maximax
(c) maximin
(d) minimax regret

Question 9

Decision Making Under Uncertainty Page 6

Mr Stall run runs a market stall selling vegetables and fruit. He buys a product for £20 per case. He can sell the
product for £40 per case in his stall. The product is perishable and it is not possible to store it, instead any cases unsold
at the end of the day can be sold off as scrap for £2 per case.

Purchase orders must be made before the number of orders is known. He has kept records of demand over the last 150
days.
Demand/day Number of days
10 45
20 75
30 30
Required:
(a) Prepare a summary of possible net daily margins using a pay off table.
(b) Advise Mr Stall:
(i) How many cases to purchase if he uses expected values.
(ii) How many cases to purchase if he uses maximin / maximax.
(iii) How many cases to purchase if he uses minimax regret.

Sensitivity Analysis and Simulation

Sensitivity analysis and simulation provide alternatives to the methods used above to deal with risk and uncertainty in
decision making.

Sensitivity Analysis

 Sensitivity analysis calculates how responsive a decision is to changes in any of the variables used to calculate it.
 It looks at one variable at a time and measures how much the variable can change by (in percentage terms) before
the decision changes.

Question 10
The baker is considering launching a new type of small cake, the Esterhazy. The baker will not launch the Esterhazy if
it will lose money in the first year. if it will breakeven or make a profit, the Esterhazy will be launched.

An accountant has prepared a forecast profitability analysis for the Esterhazy, which shows that it will be profitable.
The accountant’s analysis is as follows:
$
Selling price 3
Variable costs 1.5
Contribution per unit 1.5
Budgeted sales per day 50
Daily contribution 75
Additional daily fixed costs 50
Additional daily profits 25

Because the cake is forecast to make a profit, the baker has decided to launch it. The baker is worried, however, about
how reliable the accountant’s estimates are and wishes to know how sensitive his decision is to changes in the
underlying estimates.

Required:
(a) Calculate how sensitive the decision to launch the Esterhazy to the changes in:
(i) Selling price
(ii) Volume of daily sales
(iii) Additional fixed costs
(b) State to which of these variables the decision is most sensitive.

Advantages
Decision Making Under Uncertainty Page 7
 It is not a complicated theory to understand.
 It forces managers to identify the underlying variables, indicate where additional information would be most
useful, and helps to expose confused and inappropriate forecasts.
 An indication is provided of those variables to which profitability or value is most sensitive. And the extent to
which those variables may change before the investment breakeven.
 It provides an indication of why a project might fail. Once these critical variables have been identified,
management would review them to assess whether or not there is a strong possibility of events occurring which
will lead to a negative NPV.
 It serves as an aid in the preparation of contingency plans, should key parameters show unfavourable variations
ex-post.

Disadvantages
The method requires that changes in each of the key variables are isolated. But management is more interested in
the combination of the effects of changes in two or more variables. Looking at factors in isolation is unrealistic
since they are often inter-dependent.
It does not examine the probability that any particular variation in cost or revenue might occur.
It is not in itself a decision rule. Management must weigh the information provided by the analysis in deciding
whether the investment is worthwhile.

SIMULATION
A mathematical model constructed to represent the operation of a real life process or situation.
Simulation is a technique which allows more than one uncertain variable. Models can be generated which ‘simulate’
real-world environment within which the decision must be made.

One example of a mathematical model used in simulation is the ‘Monte Carlo’ method.

Stages
1. Specify the major variables (excessive detail will over-complicate)
2. Specify the relationship between the variables.
3. Attach probability distributions to each variable and assign random numbers to reflect the distribution.
4. Simulate the environment by generating random numbers
5. Record the outcome of each simulation
6. Repeat each simulation many times to obtain a probability distribution of the likely outcomes.

Application of simulation.
 Medical diagnosis
 Gambling
 Air force training
 Traffic scheduling

Advantages
 It overcomes the limitations of limitations of sensitivity analysis by examining the effects of all possible
combinations of variables and their realisations.
 It therefore provides more information about the possible outcomes and their relative probabilities
 It is useful for problems which cannot be solved analytically by other means.

Limitations
It is not a technique for making a decision, only for getting more information about the possible outcomes.
It can be very time consuming without a computer
It could prove expensive in designing and running the simulation on a computer.
It relies on reliable estimates of the probability distributions of the underlying variables.

Decision Making Under Uncertainty Page 8

Reducing Uncertainty
Focus groups
Much of the uncertainty which companies face in the real world relates to new products and whether they will be
successful. To reduce this uncertainty, focus groups may be used prior to the launch of the product.

 A group of people are asked to give their opinion about a new product or service. The discussion takes place in an
interactive environment in which participants are free to give their opinions and discuss them with other members
of the group.
 Members of the group are chosen at random. Often they are approached by employees of the marketing
organisation in the street and asked to participate.
 Prior to the meeting, the members of the group may be screened to ensure they belong to the target market to
which the product is aimed.
 A moderator may be present to ease the discussion.
 During the meeting, the participants may be observed, usually without their knowledge, by marketing
professionals who examine their body language, facial expressions and group behaviour.

Market Research
Market research is a process of systematically and objectively gathering, recording and analysing information. This
information may relate to:
 Customers
 General trends in the market
 Competitors
 Government regulations
 Economic trends
 Technological advancements, and
 Any other factors that constitute the business environment.

Market research can be used to help companies make decisions about the development and marketing of new
products. The earlier the market research is conducted in the development of a product, the better, from a risk point of
view.

Market research can be based on primary or secondary data.

 Primary data means the company collects it own original data, for example, by conducting interviews.
 Secondary data means that already published data is used, such as published statistics.

Question 11

Decision Making Under Uncertainty Page 9

Cement Co is a company specialising in the manufacture of cement, a product used in the building industry. The
company has found that when weather conditions are good, the demand for cement increases since more building
work is able to take place. Last year the weather was so good, and the demand for cement was so great, that Cement
Co was unable to meet demand. Cement Co is now trying to work out the level of cement production for the coming
year in order to maximise profits. The company doesn’t want to miss out on the opportunity to earn large profits by
running out of cement again. However, it doesn’t want to be left with large quantities of the product unsold at the end
of the year, since it deteriorates quickly and then has to be disposed of. The company has received the following
estimates about the probable weather conditions and corresponding demand levels for the coming year:

Weather Probability Demand

Good 25% 350,000 bags
Average 45% 280,000 bags
Poor 30% 200,000 bags

Each bag of cement sells for $9 and costs $4 to make. If cement is unsold at the end of the year, it has to be disposed
of at a cost of $0.50 per bag.

Cement Co has decided to produce at one of the three levels of production to match forecast demand. It now has to
decide which level of cement production to select.

Required:
(a) Construct a pay off table to show all the possible profit outcomes. (8 marks)

(b) Decide the level of cement production the company should choose, based on the following decision rules:

(i) Maximin (1 mark)

(ii) Maximax (1 mark)
(iii) Expected value (4 marks)
You must justify your decision under each rule, showing all necessary calculations.
(c) Describe the ‘maximin’ and ‘expected value’ decision rules, explaining when they might be used and the
attitudes of the decision makers who might use them. (6 marks)
(Total = 20 marks)

Question 12

SHC specialises in the provision of sports/exercise and medical/dietary advice to clients. The service is provided on a
residential basis and clients stay for whatever number of days suits their needs.

The budgeted estimates for the year ending 30 April 2014 are as follows.

Decision Making Under Uncertainty Page 10

(a) The maximum capacity of the company is 50 clients per day for 350 days in a year.
(b) Clients will be invoiced at a fee per day. The budgeted occupancy level will vary with the client fee level per day
and is estimated at different percentages of maximum capacity as follows.
Client fee Occupancy as a percentage
Per day Occupancy level of maximum capacity

K180 High 90%

K200 Most likely 75%
K220 Low 60%

(c) Variable costs are also estimated at one of three levels per client per day. The high, most likely and low levels per
client day are K95, K85 and K70 respectively.

The range of cost levels reflects only the possible effect of the purchase prices of the goods and services.

Required:

(a) Prepare a summary which shows the budgeted contribution earned by SHC for the year ended 30 April
2014 for each of the nine possible outcomes. (6 marks)

(b) State the client’s fee strategy for the year to 30 April 2014 which will result from the use of each of the
following decision rules.

(i) Maximax
(ii) Maximin
(iii) Minimax regret

Your answer should explain the basis of operation of each rule. Use the information from your
answer to (a) as relevant and show any additional working calculations as necessary.
(10 marks)

(c) The probabilities of variable costs levels occurring at the high, most likely and low levels provided in the
question are estimated at 0.1, 0.6 and 0.3 respectively.

Using the information available, determine the client’s fee strategy which will be chosen where
maximisation of expected value of contribution is used as the decision basis. (4 marks)

(Total: 20 marks)

Decision Making Under Uncertainty Page 11