Decision theory and decision-making

Decision-making under risk:

As mentioned earlier, decision-making under
risk has multiple outcomes under each course
of action. The probability of occurrence of
each outcome is known, in other words there
is information. Each outcome under each
course of action has a stated value as
described by the problem situation.
 Decision-making under risk
. . . Expected Monetary Value: . . .

The matrix format is used for decision-making

under risk problem formulation and solution.
This section will present several solution
procedures. The first decision criteria uses the
Expected Monetary Value (EMV).
 Decision-making under risk
Expected Monetary Value: . . .
The expected monetary value of the k th
alternative is given by the formula:
1
EMV = ∑ = pi Vik
i=0
Where pi is the probability of the ith state of
nature occurring.
 Decision-making under risk
. . . Expected Monetary Value: . . .
The Law of Long-run Success:
The validity of the EMV principle relies upon
the mathematical law of long run success.
Assume that the decision-maker will make N
similar but independent decisions in the future.
If the decision-maker should use the EMV
principle, then the total value received from all
N decision would be SN.
 Decision-making under risk
. . . Expected Monetary Value: . . . The Law of Long-run Success:

If the decision-maker should use any other

principle of choice, then the total value
received would be RN. The law of long run
success says that, as the number of decisions
becomes very large (approaches infinity), then
the probability of SN exceeding RN
approaches 1. This can be expressed
symbolically as follows:
 Decision-making under risk
. . . Expected Monetary Value: . . . The Law of Long-run Success:

Lim Prob (SN > RN) 1;

as N ∞.
Note that there must be a large number of
similar and independent decisions. Also,
certain individual decisions may have bad
outcomes, but in the long run the decision-
maker will be better off having used the
EMV principle.
 Decision-making under risk

… Expected Monetary Value: … An example

A decision-maker is confronted with two
mutually exclusive investment opportunities.
The decision-maker can choose one
investment, or refuse both. In addition, three
possible states of nature may occur in the
future and the decision-maker has determined
the probabilities of these states.
 Decision-making under risk
… Expected Monetary Value: … An example
States of Nature
S1 S2 S3
P(S1) = .5 P(S2) = .1 P(S3) = .4
e A1
v Invest in
ETB ETB ETB
i
t
A 200,000 -100,000 -100,000
A2
a
n Invest in
ETB ETB ETB
r
e
B 50,000 50,000 -20,000
t A3
l Invest in
A neither 0 0 0
 Decision-making under risk
… Expected Monetary Value: … An example

The next step is to calculate the expected

monetary value (EMV) for each of the
alternatives:

.
 Decision-making under risk
. . . Expected Monetary Value: . . . Am example
Decision . . .
EMV (A) = .5 (200,000) + .1 (-100,000)
+ .4 (-100,000) = ETB 50,000

EMV (B) = .5 (50,000) + .1 (50,000) +. 4

(-20,000) = ETB 22,000

EMV (Neither) = .5 (0) + .1 (0) + .4 (0)

= ETB 0
 Decision-making under risk:
… Expected Monetary Value: … An example

Thus, according to the expected monetary value principle, the

decision-maker should choose investment A.
 Decision-making under risk
… Expected Monetary Value: … An example

Now, suppose the decision-maker is asked to

rank the alternatives according to that person's
personal preferences (considering both the
values and the probabilities). The decision-
maker expresses the following paired
comparisons:
 Decision-making under risk
… Expected Monetary Value: … An example

Investment B > Investment A

Investment B > Investment in
Neither
Investment in Neither > Investment A
 Decision-making under risk
… Expected Monetary Value: … An example

Rank Alternatives
1 Investment B
2 Investment in Neither
3 Investment A
 Decision-making under risk
… Expected Monetary Value: … An example

According to the decision-maker’s personal

attitude toward the values and their associated
risk, the decision-maker chose Investment B.
What has gone wrong? Why is the answer
inconsistent with that derived using the
expected monetary value principle?
 Decision-making under risk
The St. Petersburg Paradox : . . .

Let’s take another example. Suppose that the

decision-maker is asked to bid for the right to
make the following gamble. A fair coin is
tossed until the first head appears. If n is the
number of tosses required to obtain the first
head, then you would be paid Birr 2n.
 Decision-making under risk
. . . The St. Petersburg Paradox : . . .

How much would the decision-maker be willing

to pay for the right to participate in this gamble?
Probably not very much - may be Birr 4 or Birr
5.
 Decision-making under risk

. . . The St. Petersburg Paradox : . . .

The ironic part of this particular gamble

(known in probability theory as the St.
Petersburg Paradox) is its expected
monetary value. It is calculated as follows:
 Decision-making under risk
. . . The St. Petersburg Paradox : . . .

EMV = (½) 2 + (¼) 4 + (1 / 8 ) 8 + …+(½ n) 2n

= 1+1+1+… = Infinity
 Decision-making under risk
. . . The St. Petersburg Paradox : . . .

The expected monetary value of this gamble is infinity,

contrary to our willingness to pay only a few Birr for
the right to gamble.
 Decision-making under risk
. . . The St. Petersburg Paradox : . . .

The probability of winning much money in this

gamble is quite small. EMV is not a valid guide
in this situation simply because it ignores the
decision-maker’s attitude toward accepting risk,
that is, his attitude toward the likelihood of
incurring significant losses or gains.
 Decision-making under risk
. . .The St. Petersburg Paradox : . . .
When is expected monetary value a valid guide for
decision-making? Generally, if the values involved
in the decision are relatively small, then EMV will
be an accurate measure of the decision-maker’s
attitude toward the alternatives. Otherwise, more
sophisticated techniques of evaluation will be
necessary.
 Decision-making under risk
The Expected Utility Value: . . .

EUV is suggested as a substitute to EMV

when the decision-maker states values, other
than monetary ones, to express personal
preferences to risk. These values are called
utiles and are calculated as follows:
 Decision-making under risk
. . . The Expected Utility Value: . . .

One method of quantifying utility is to assume

hypothetical “reference” contracts whose
maximum gain and loss are greater than any in
the decision-maker’s payoff matrix.
 Decision-making under risk
. . . The Expected Utility Value: . . .

The decision-maker then is given the choice

between the reference contracts, with a
probability P of winning, and a specific value
in the payoff matrix, which he could receive
with certainty.
 Decision-making under risk
. . . The Expected Utility Value: . . .

The decision-maker is then asked to state the

value of P such that the decision-maker is
indifferent between the two alternatives offered
to that person. The value of P will be the utile
associated with that specific value in the payoff
matrix.
 Decision-making under risk
. . . The Expected Utility Value: . . .
States of Nature
S1 S2 S3
P(S1) = .5 P(S2) = .1 P(S3) = .4
e
Invest in
v
i A
200,000 - 100,000 - 100,000
t
a
Invest in
n
r
B
50,000 50,000 - 20,000
e
t
Invest in
0 0 0
l
A
neither
 Decision-making under risk

. . . The Expected Utility Value: . . .

Our reference contract must have
consequences of greater than Birr 200,000 and
less than Birr -100,000. Accordingly, let us
choose our reference lottery so that the
decision-maker has a probability P of winning
Birr 250,000, and a probability 1 - P of loosing
Birr 150,000.
 Decision-making under risk
. . . The Expected Utility Value: . . .
We ask the decision-maker the following
question: “Suppose you have a choice between
the reference lottery and a definite loss of Birr
100,000 (from payoff matrix), what would be
the value of P such that you are indifferent
between the two alternatives?
 Decision-making under risk
. . . The Expected Utility Value: . . .
The decision-maker might say: “If P is 2 then
I would be exactly indifferent between the
reference lottery and a definite loss of Birr
100,000.” Thus, associated with Birr -100,000
will be a P equal to .2
 Decision-making under risk
. . . The Expected Utility Value: . . .
Next, we give the decision-maker the choice
between the reference lottery, and a definite
loss of Birr 20,000. What would be the value
of P such that the decision-maker would be
indifferent between these two alternatives?
The decision-maker might indicate that a P
of .6 would make that person indifferent.
 Decision-making under risk
. . . The Expected Utility Value: . . .

We continue to ask the decision-maker the

same sort of questions fro 0 value and profits
of Birr 50,000 and Birr 100,000. Finally, we
arrive at the following values:
 Decision-making under risk
. . . The Expected Utility Value: . . .
Monetary value P (utiles)
-150,000 0
-100,000 .2
-20,000 .6
0 .7
+50,000 .8
+200,000 .95
+250,000 1
 Decision-making under risk
. . . The Expected Utility Value: . . .

States of Nature
S1 S2 S3
P(S1) = .5 P(S2) = .1 P(S3) = .4
e
Invest
v
i in A
.95 .2 .2
t
a
Invest
n
r
in B
.8 .8 .6
e
t
l
Invest
A
in
.7 .7 .7
neither
 Decision-making under risk

. . . The Expected Utility Value: . . .

Ranking the alternatives according to expected
utiles, we get the following:

Rank Alternative
1 Investment B
2 Neither
3 Investment A
 Decision-making under risk
. . . The Expected Utility Value: . . .

Expected utile (A) = .5 (.95) + .1 (.2) + .4 (.2)

= .575
Expected utile (B) = .5 (.8) + .1 (.8) + .4 (.6)

= .72
Expected utile (neither) = .5 (.7) + .1 (.7) + .4 (.7)
= .70
 Decision-making under risk
Utility functions:

Several authors have indicated that it is

possible under certain conditions to quantify
the decision-maker’s subjective preferences
toward monetary values.
 Decision-making under risk
, , , Utility functions: . . .
These quantified utilities describe the
decision-maker’s utility function, which
assumes a specific form when plotted. The
three most common types of utility functions
are shown below:
- The linear utility function
- Decreasing marginal utility
function
- Increasing marginal utility
 Decision-making under risk
Linear utility function: . . .

If the decision-maker’s utility function is

linear, as in the next graph, then maximizing
expected utility is equivalent to maximizing
expected monetary value.
 Decision-making under risk
. . . Linear utility function: . . .

Linear utility functions generally exist when

the difference between the maximum and
minimum values in the decision-maker’s
payoff matrix is relatively small. “Relatively
small’ is a term strictly subjective to the
decision-maker.
 Decision-making under risk
. . . Linear utility function:

Monetary Value
 Decision-making under risk
Decreasing marginal utility: . . .

In the next graph, the decision-maker receives

a diminishing marginal utility for additional
monetary values, (i.e. each additional Birr
provides utility at a decreasing rate). A
person with this shaped utility function is said
to have an aversion to risk.
 Decision-making under risk
. . . Decreasing marginal utility: . . .

For example, suppose the decision-maker can choose

a gamble where there is a .5 probability of winning
Birr100 and a .5 probability of losing Birr 50; or he
can choose to accept Birr 25 with 100 % certainty.
 Decision-making under risk

. . . Decreasing marginal utility: . . .

In this case, both alternatives have the same

expected monetary values (Birr 25). For the
decision-maker who has an aversion to risk,
Birr 25 with certainty will have a higher
expected utility than the first alternative.
 Decision-making under risk
. . . Decreasing marginal utility:

Monetary Value
 Decision-making under risk
Increasing marginal utility: . . .

As shown in the next graph, the decision-

maker receives an increasing marginal utility
for additional monetary values (i.e. each
additional Birr provides utility at an increasing
rate).
 Decision-making under risk
. . . Increasing marginal utility: . . .

Such individuals are said to have a preference

for risk. They would be willing to accept a
disproportionately large increase in risk for an
increase in expected monetary value.
 Decision-making under risk
. . . Increasing marginal utility:

Monetary Value
 Decision-making under risk
Typical utility function:

Empirical studies indicate that the decision-

maker experiences the three types of utility
functions depending on the amount of
monetary value.
 Decision-making under risk
. . . Typical utility function:

As in the graph below, for monetary values in

the range of a – b, the decision maker
expresses “indifference;” for the range b – c,
preference to risk, and for c – d, aversion to
risk.
 Decision-making under risk
. . . Typical utility function:

a b c d
Monetary Value
 Decisin-making under risk
Problems with the utility function:

It should now be obvious that utility is the

most appropriate measure of effectiveness for
decision-making. But how practical is it?
Several serious questions have been raised
about utility; and below just a few of them are
mentioned:
 Decision-making under risk
. . . Problems with the utility function: . . .
1. Whose utility function is appropriate for
the overall organization? Should it be the
manager’s, the board of director’s, or the
general manager’s?
2. Can an individual’s utility function be
said to accurately represent that of the whole
organization?
 Decision-making under risk
. . . Problems with the utility function:

3. Will decision-makers be willing to take

the time and effort necessary to construct a
utility function for every decision situation?
(This is more challenging because utility is
constantly in flux.)
 Decision-making under risk
Other decision-making procedures:
Four more decision-making procedures that
consider risk will be presented:
1. Most probable future principle;
2. Aspiration level principle;
3. Expectation-variance principle; and
4. Chebyshev’s Inequality.
 Decision-making under risk
Most Probable Future Principle:
Under certain conditions the decision--maker
tends to suppress the risk involved in a
decision situation. The decision-maker simply
locates the state of nature, which is most likely
(i.e. has the highest probability associated with
it) and acts as if this state of nature were
certain.
 Decision-making under risk
. . . Most Probable Future Principle: . . .

Thus, the decision-maker chooses the

alternative, which has the highest value,
associated with the most likely state of nature.
 Decision-making under risk
. . . Most Probable Future Principle: . . .

When is the most probable future principle

appropriate? It may be appropriate when one
state of nature is far more likely than any
other, and when the possible values to the
decision-maker are of approximately the same
magnitude.
 Decision-making under risk
. . . Most Probable Future Principle: . . .

The decision-maker need not take the time and

trouble to determine the probabilities
associated with all states of nature. If all the
probabilities had been determined, then the
EMV principle would be a more appropriate
approach.
 Decision-making under risk
. . . Most Probable Future Principle: . . .

Some examples of where the most probable future

principle is used are as follows:

1. A person who does not insure own

automobile is acting as if that person will not have
an accident.
 Decision-making under risk
. . . Most Probable Future Principle: . . .

2. A person buys government debentures

ignoring the likelihood of default.

3. A person puts own money in a bank

ignoring the likelihood of the bank closing.
 Decision-making under risk
. . . Most Probable Future Principle: . . .
An example:
Assume that a decision-maker is considering
investing in one of three companies. All three
companies are considered to be stable with
only about a .15 probability of bankruptcy.
 Decision-making under risk
. . . Most Probable Future Principle: . . .
An example -
This means the companies have a .85 probability
of being profitable. Investments in company
A, B, and C are expected to earn rates of
returns of 8%, 9%, and 10%, respectively, if
the companies remain profitable. In which
company should the decision-maker invest?
 Decision-making under risk
. . . Most Probable Future Principle: . . . An example -

States of Nature
S1 S2 S3 \
P(S1) = .85 P(S2) = .02 P(S3) = .01 ... ... ...

e
Invest
v
i in A
8% 5% 4% ... ... ...
t
a Invest \
n
r in B
9% 4% 4% ... ... ...
e
Invest \
t
l
A in C
10% 4% 3% ... ... ...
 Decision-making under risk
. . . Most Probable Future Principle: An example –

The solution:
1. Identify the state of nature with the highest
probability: P(S1) = .85.

2. Identify and select the alternative with the highest

value if S1 occurs:
Alternative C: value of 10%
 Decision-making under risk
. . . Aspiration Level Principle: . . .

There is strong evidence indicating that

decision-makers do not act with a view
towards optimizing. One reason for this is the
limited capacity of human beings to obtain and
process all relevant information.
 Decision-making under risk
. . . Aspiration Level Principle: . . .

For example, there may in fact be several

hundred alternatives available to the decision-
maker; however, the decision-maker may have
the time and money necessary to evaluate only
ten or so of these alternatives.
 Decision-making under risk
. . . Aspiration Level Principle: . . .

In this situation, the decision-maker can no

longer long for an optimal result; instead the
decision-maker hopes for some “satisfactory
result." This satisfactory result is referred to
as the decision-maker's aspiration level - a
value that the decision-maker aspires to obtain
from the decision.
 Decision-making under risk
. . . Aspiration Level Principle: . . .

The aspiration level principle involves

screening alternatives until one is found which
has a reasonable probability of achieving the
aspiration level.
 Decision-making under risk
. . . Aspiration Level Principle: . . .
If the decision-maker must choose from a set
of alternatives that are already evaluated, then
the decision-maker would choose that
alternative with the highest probability of
providing a satisfactory result (i.e. the person
chooses the alternative which maximizes the
probability of achieving the stated aspiration
level).
 Decision-making under risk
. . . Aspiration Level Principle: . . .
The aspiration level principle often is applied
in situations where it is difficult or costly to
evaluate all alternatives or states of nature. In
some cases where alternatives are transient in
nature, the decision-maker is forced to either
accept or reject an alternative on the spot
without being able to wait for other
alternatives to be discovered
 Decision-making under risk
. . . Aspiration Level Principle: . . .

Merchants are constantly faced with this type

of situation with customers' offers for certain
commodities.
 Decision-making under risk
. . . Aspiration Level Principle: . . .
Also, the aspiration level principle is commonly
used when it is difficult or impossible to
numerically evaluate outcomes. By classifying
outcomes as either “acceptable” or
“unacceptable,” the decision-maker would
choose the alternative with the highest
probability of achieving an acceptable outcome.
•
 Decision-making under risk
Expectation-Variance Principle: . . .
For major decisions, the principles of choice
previously mentioned are considered
inadequate. By ignoring risk or using only one
parameter as a guide, the decision-maker is
over-simplifying a very complex reality. The
result of using such principles generally will
be commensurate with their simplicity.
 Decision-making under risk
. . . Expectation-Variance Principle: . . .
Thus, it is often recommended that for major
decisions involving large possible gains and
losses that the decision-maker develops, a
more complex model for decision-making.
The expectation-variance principle is one
such model.
 Decision-making under risk
. . . Expectation-Variance Principle: . . .
This principle states that any one parameter by
itself will be an insufficient guide for decision-
making. Expected monetary value should be
considered; however, the risk involved in
attaining the expectation is also relevant
 Decision-making under risk

. . . Expectation-Variance Principle: . . .

A good indicator of risk is felt to be the

variance of the monetary values. Thus,
both expectation and variance should be
considered by the decision-maker.
 Decision-making under risk
. . . Expectation-Variance
Principle: . . .

A good indicator of risk is felt to be the

variance of the monetary values. Thus, both
expectation and variance should be considered
by the decision-maker.
 Decision-making under risk
. . . Expectation-Variance Principle: . . .
If we assume that our decision-maker has an
aversion to risk in the range of values
organized in the payoff matrix, then the
following will be true:
1. If two alternatives have the same
expected monetary values, then the alternative
with the smallest variance will be preferable.
 Decision-making under risk
. . . Expectation-Variance Principle: . . .
2. If two alternatives have the same
variance, then the one with the highest
expected monetary value will be preferable.
3. Even if neither expectations nor
variances are equal, it is still suggested that
both of these parameters be considered for
major decisions.
 Decision-making under risk
. . . Expectation-Variance Principle: . . .

With these values, the decision-maker will be

in a better position for choosing an alternative,
which is consistent with that person's
subjective preferences for monetary values and
that same person's willingness to accept risk in
order to obtain the monetary values.
 Decision-making under risk
. . . Expectation-Variance Principle: . . .

Alternatives EMV () Variance (2)

1 25,000 100,000,000

2 12,000 4,000,000

3 22,000 64,000,000
 Decision-making under risk
. . . Expectation-Variance Principle: . . .

In decision-making, however, point estimators

are limited in their usefulness. Interval
estimators are superior to point estimators for
decision-making.
 Decision-making under risk
Chebyshev’s Inequality:

Specification of expected value and variance

generally will not mean much to a person not
adept in the use of these parameters. For
example, consider the three alternatives
available with the following expectations and
variances:
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .

Alternatives EMV () Variance (2)

1 25,000 100,000,000

2 12,000 4,000,000

3 22,000 64,000,000
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .

These would determine points if we had a

Cartesian coordinate system with 2 on the Y-
axis and u on the X-axis, and are therefore
called point estimates. Given these point
estimates, which alternative should the
decision-maker choose? It would be difficult
to decide:
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .

5 10 15 20 25
EMV = µ in 000
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .
Point estimates are often felt to provide
insufficient information, particularly, when
major decisions are involved. It is therefore
suggested that we convert our point estimates
to interval estimates, thus enabling us to
interpret the point estimates by means of
probabilities.
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .

An acceptable (yet somewhat weak) approach

for accomplishing this is through the use of
Chebychev’s Inequality.
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .

If V is a random variable with mean µ and

standard deviation δ, then for any constant k,
the probability that V assumes a value less
than µ - kδ or greater than µ + kδ is less than
(1/k)2 .
 Decision-making under risk
. . . Chebyshev’s Inequality: . . .
Symbolically:
Probability ( |V - µ| > kδ) < (1/k)2.
We can re-write this in the following manner:
Prob. (µ - kδ) ≤ V ≤ (µ + kδ) ≥ 1- (1/k)2.
 Decision-making under risk
. . . Chebyshev’s Inequality: - Example
What is the probability that V (the actual
value) comes within 2 standard deviations (δ)
of the mean (µ)?

Prob. (µ - 2δ) ≤ V ≤ (µ + 2δ) ≥ 1- (1/2)2.

≥ .75
Decision-making under risk

Final Panel