FOUNDATION LEVEL 2

QUANTITATIVE ANALYSIS
DECISION THEORY
Introduction
Decision making is concerned with choosing a course of action from among available alternatives.
Terminology and definitions.
1. Alternative decisions - These are choices available for selection. They are sometimes referred
to as acts. They are strategies that may be chosen by the decision maker (and therefore
controllable) Decision alternatives are assumed to be mutually exclusive i.e. the choice of one
precludes the choice of the another.

2. States are nature- These are circumstances that affect the outcomes of a decision but are
beyond the control of the decision maker. They are also referred to as external actions. The
chances or probabilities of these occurring may or may not be available.

3. Outcomes/ conditional payoffs, Act- event combination.

This results from using a given alternative when a certain state of nature occurs. In other words,
for each combination of an alternative decision and an event, there will exists an outcome. The
outcome. The outcome will result in a pay off that may be expressed in monetary and non-
monetary terms.

4. Opportunity loss- This is the amount that a decision maker would lose by not taking up the
best course of action. sometimes called the amount of regret. For any state of nature, this is the
 difference between the consequences of any course of action and the best possible alternative.
Sometimes it is called the amount of regret.

Steps in decision making

1. Define the problem at hand. The problem often presents itself as an objective. The
quantification of an objective is often called an objective function. It is used to evaluate the
alternative courses of action and to provide the basis of choosing the best alternative.

2. List the possible alternatives (that will enable the objective to be achieved)

3. Identify possible events (states of nature). Events are often uncontrollable factors outside the
decision maker’s control.

4. List the possible payoffs for each combination of alternative and outcome. These are known
as conditional payoffs.

5. Select an appropriate mathematical tool from amongst decision theory models.

6. Apply the model and hence recommend the decision.

7. If possible, perform sensitivity analysis (also known as what if analysis or post- optimality
analysis, before making the final recommendation.

Decision making environments

Under these conditions events are known with certainty. All the information about which
state of nature will occur is known for sure. Conditions of certainty do not occur much in
reality. Certainty is when there is no doubt of the outcome of an event. E.g., The
transportation model assumes that demand at every destination is known, supply
from every source known, unit transportation cost is known and is therefore a
deterministic model i.e., contrast probabilistic model.

2. Conditions of risk-
Under this environment It is not known exactly which state of nature will occur, however
there is sufficient past data to enable the computation of probabilities for decision making.
Although we do not know for sure what outcome (state of nature) will occur, we can
estimate the chance (probability) of occurrence of each state. Chance of occurrence can be
estimated subjectively or from past experience.
3. Conditions of uncertainty- under these conditions no prior data exists for computing
probabilities (unlike conditions of risk). The states of nature can be enumerated but the
probabilities cannot be attached to them. This may be because of insufficient data or
because the trend may not be repeated in the future.

4. Conditions of competition- under these conditions the decision maker is not alone in the
environment. There is an active opponent competing with the decision maker. The body of
knowledge for this environment is known as game theory.

Decision making under conditions of uncertainty.

1. Maximax Criteria (extreme optimism)

This criterion is based on extreme optimism and the decision-maker will select that
alternative that corresponds to the MAXImum of the MAXimum payoffs.
Users: risk takers, large firms which are able to absorb huge losses in case state of nature
turns out unfavorable.

2. Maximin criteria (extreme optimism)

This criterion is based upon the conservative approach that the worst possible is going to
happen. The decision maker will consider each strategy and locate or identify the minimum
payoff for each alternative.

The decision maker will then choose the alternative which corresponds to the MAXImum
of the MINimum payoffs.

Users: Risk avoider, small firms and which cannot absorb large losses in case things turn
out unfavorable.

3. Hurwicz criteria (criterion of realism)

To overcome the disadvantages of extreme pessimism of the maximin and the extreme
optimism of the Maximax Hurwicz introduced a criterion of coefficient of optimism, ά.

This concept allows the decision maker to take into account both the maximum and the
minimum payoff for each decision alternative and assign them weights according to his
degree of optimism.

If ά is the degree of optimism then 1- ά will be the degree of pessimism

h=ά* Maximum payoff + 1-ά* minimum payoff

4. Laplace criteria (equally likely)

This criterion uses all the information by assigning equal probabilities to the possible
payoffs for each of the action and then selects that alternative that corresponds to the
maximum expected payoff.

Assign equal probabilities (1/n) to each payoff of a strategy having (n) possible payoffs.

Determine the expected payoff value for each alternative.

Select that alternative that corresponds to a maximum of the expected payoffs.

Also know as the criterion pf rationality. In effect converts problem from one of under
uncertainty to one under environment of risk.

5. Minimax regret criterion (savage criterion)

The decision maker would therefore select the alternative which will MINimize his
MAXimum regret.

The first step is to determine the amount of regret for the payoff of each alternative.

The regret amount of the ith alternative when event j has occurred is given as

ith regret= maximum pay-off (for the alternative) – ith payoff for the jth event

Determine the maximum regret for each alternative.

Choose that alternative that corresponds to the minimum of the maximum regret.