Ch a Ptter 1

Decision Theory and Decision Trees

"The one word that makes a good manager decisiveness."
-

- lacocca, Lee

PREVIEW
In decision theory a set of techniques are used for
uncertainty and risk.
making decisions in the decision-environment of

Decision tree graphically displays the progression of decision and random events, in those cases
where a problem involves a sequence of decisions (including a decision on whether to obtain additional
information).
Utilities theory helps to incorporate decision-makers' attitude towards risk into the analysis of those
problems where there is possibility of large losses.

LEARNING OBJECTIVES

After studying this chapter, you should be able to

understand the steps of decision-making process.
make decision under various decision-making environments.
determine the expected value of perfect information, expect opportunity loss and expected monetary
value associated with any decision.
revise probability estimates using Bayesian analysis.
construct decision trees for making decision.
understand the importance of utility theory in decision-making.

CHAPTER OUTLINE
11.1 Introduction Self Practice Problems B
Hints and Answers
1.2 Steps of Decision-Making Process
11.7 Decision Tree Analysis
1.3 Types of Decision-Making Environments
11.8 Decision-Making with Utilities
14 Decision-making Under Uncertainty
Self Practice ProblemsC
Conceptual Questions A
Hints and Answers
Self Practice Problems A
Hints and Answers Chapter Summary
11.5 Chapter Concepts Quiz
Decision-Making Under Risk
Posterior Probabilities and Bayesian Analysis Case Study
Conceptual Questions B
 340 Operations Research: Theory and
Applications
11.1 INTRODUCTION
The success or failure that an individual or organization experiences, depends to a large extent. on the abilin
ot making acceptable decisions on time. To arrive at such a decision, a decision-maker needs to enumerat. bility
easible and viable courses of action (alternatives or strategies), the projection of consequences associaled
erate
wth cach course of action, and a measure of effectiveness (or an objective) to 1dentify the best cour
iated
of action. ourse
Decision theory is both descriptive and prescriptive business modeling approach to classify the degree
of
I d g e and compare expected outcomes due to several courses of action. The degree of knowledee
avided into four categories: complete knowledge (i.e. certainty), ignorance, FISK ana uncertainty as
shown in Fig.
11.1
Fig. 11.1
Zones of
gnoranceUncertainty Risk Certainty
Decision-making - Increasing knowledge

Decision aralysis is Irespective of the type of decision

following essential components are common to all:
model,
an analytical
approach of Decision alternatives There is a finite
at each
number of decision alternatives available to the decision-maker
comparing decision point in time when a decision is made. The number and
alternatives in terms the previous decisions type of such alternatives may depend on
made and their outcomes. Decision alternatives
of expected
as
stocking 100 units of a particular item, or may be described numerically, such
Outcomes. the likely demand of an item. non-numerically, such as conducting a market survey to know
States of nature A state of nature is an event or
For scenario that is not under the control of decision makers.
instance, it may be the state of
economy (e.g. inflation), a weather condition, a
etc. political development,
The states of nature
may be identified through Scenario Analysis where a section of
interviewed -

stakeholders, long-time people are

managers, etc., to understand states of nature
impact on a decision. that may have serious
The states of nature are
mutually exclusive and collectively exhaustive with
problem. The states of nature may be described numerically respect to any decision
such as, demand of 100
non-numerically such as, employees strike, etc. units of an item or

Payoff It is a numerical value (outcome) obtained due to the

States of nature of decision alternatives and states application of each possible combination
are outcomes duue of nature. The payoff values are
to random factors unknown states of nature. always conditional values because of
that effect the The payoff values are measured within a
specified
payoff of from decision horizon. The payoffs in most decisions areperiod (e.g. within one year, month, etc.) called the
decision alternatives. monetary. Payoffs resulting from each
combination of decision alternatives and states of natures
matrix) form as shown in Tahle 11.1.
are
displayed in a matrix (also calledpossible payof

Couses of Adie
States of Nanure Probability CAlteraaitves
P1 P1
Table 11.1 N2 P2 P12 P1n
P21 P22
General Form of P2n
Payoff Matrix
Pml
Pm2 Pmn
11.2 STEPS OF DECISION-MAKING PROCESS
The decision-making process involves the following steps:
1. Identify and define the problem.
all nossible future events (not under ne control
of
decision-maker) that arce likely to occur.
 all the courses 0f Decision Theory and Decision Trees 341
hdentuly ction available to the
the
Eypress the
payofts Pi;) resulting
(
irom decision-maker.
Apply n appro oropriate decision theory model toeach combination of course of
action and state of
select the best nature
of a eriterion
(mieasure of
effectiveness) to get optimal course of action
from the given list on tne
Example141.1 A fim manutactures three types (desired) payoff.
of
products. The fixed and variable costs
Fixed Cost (Rs) are
given below
lariable Cost
Product A per Unit (Rs)
Product B
25.000
12
35,000
Product C 9
53.000
The
ely demand (units) of the products is given below:
Poor demand
3,000
Moderate demand 7.000
High demand 11,000 Payoff is the
s he sale price of each type of
product is Rs 25, then prepare the quantitative measure
Solution Let D,. D, and D, be the poor, moderate and
payoff matrix. of the outcome from
high demand, respectively. The payoff is determined as: each pair of
Payoff Sales revenue - Cost decision alternative
and a state of
The calculations for payoff (in "000 Rs) for each
npes of product (state of nature) are shown below:
pair of alternative demand (course of action) and the
D, A =3 x 25 25 -3 x 12 = 14 D,A = 7 x 25 25 - 7 x 12 = 66
D,B 3 x 25 3 5 - 3 x 9
13 D, B = 7 x 25 35-7 x 9 =77
D,C 3 x 25 5 3 3 x 7 D, C = 7 x 25 5 3 - 7 x 7 = 73
D,A = 1 x 25 25 - 11 x 12 118

D,B 11 x 25 - 35 - 11x 9 = 141

D,C = 11 x 25 53 - 11 x 7= 145

The payoff values are shown in Table 11.2.

Product Type Aternative Demand (in 000 Rs)

14 b6 118
B 13 7 141
13 145
Table 11.2
C

113 TYPES OF DECISION-MAKING ENVIRONMENTS

diTive at an optimal decision it is essential to have an exhaustive list of decision-alternatives, knowledge
approach for decision-making. In this section
deCision environment, and use of appropriate quantitative and risk, have been discussed. The
Decision making
C ypes of decision-making environments: certainty, uncertainty, under certainty is
WICdge of these environments helps in choosing the quantitative approach for decision-making. an environment in
which future
outcomes or states
ype 1 Decision-Making under Certainty of nature are
(perfect information) of known.
th decision-maker has complete knowledge
decis1on-making environment, he would select a decision
of action). In such a case
out due to each decision-alternative (course
al v e under known state of nature. For example, the decision
altemati that yields the maximum return (payofl) Provident Fund, etc., is where complete
Vikas Patra, Public
In National Saving Certificate, Indira
in the principal at maturity
IS know.

about the future return due and

n

Vpe 2 Decision-Making under Risk Knowledge about possible outcome of

In this s1on-environment, decision-maker does
not have

than one
perlect
states or nature. In a such a case he makes
every ccision alternative. It may be duc to more
stale of nalure
an ass occurrence
of particular
pOn of the probability for
342 Operations Research: Theory and
Applications
Type 3 Decision-Making under Uncertainty Oceutence of particula
10T

is unable to specify the probabity

decision-mnaker ignOrance, because the possihla
ot CIsion environment, of decision-making
under

O r nature. However. this is not the case

are taken even
witn
Iess intormation than
uncertainty
nature are known. Thus, decisions under be the prime
minister ot the
country I5
O that Mr X will
celsions under risk. For exanmple, the probability
years from now is not known.

11.4 DECISION-MAKING UNDER UNCERTAINTY

decision-maker must arrive at a decision only
the
c of any outcome can not be quantified,the criterion ofeffectiveness (policy). The following
probability
in view
CCtual conditional payoff values, keeping discussed in this
have been
section.
T
decision-making under uncertainty
) Optimism (Maximax or Minimin) criterion
(i) Pessimism (Maximin or Minimax) criterion
(ii) Equal probabilities (Laplace) criterion
(iv) Coefficient of optimism (Hurwiez) criterion
()Regret (salvage) criterion

11.4.1 Optimism (Maximax or Minimin) Criterion

n this criterion the decision-maker ensures that he should not miss the opportunity to achieve the largest

profit (maximax) the lowest possible cost (minimin). Thus, he selects the decision alternative
pOssible or
that represents the maximum of the maxima (or minimum of the minima) payoffs (consequences or outcomes).

The working method is summarized as follows:

Decision making (a) Locate the maximum (or minimum) payoff values corresponding to each decision allernative.
under risk is an (b) Select a decision alternative with best payoff value (maximum for profit and minimum for cost).
environment in
which the probability Since in this criterion the decision-maker selects an decision-alternative with largest (or lowest) possible
of outcomes or payoff value, it is also called an optimistic decision criterion.
states of nature can
be quantified.
11.4.2 Pessimism (Maximin or Minimax) Criterion
In this criterion the decision-maker ensures that he would earn no less (or pay no more) than some specified
amount. Thus, he selects the decision alternative that represents the maximum of the minima (or minimum
of the minima in case of loss) payoff in case of profits. The working method is summarized as follows:
(a) Locate the minimum (or maximum in case of profit) payoff value in case of loss (or cost) data
corresponding to each decision alternative.
(b) Select a decision alternative with the best payoff value (maximum for profit and mimimum for loss
or cost).

Since in this criterion the decision-maker is conservative about the future and always anticipates the worst
possible outcome (minimum for profit and maximum for cost or loss), it is called a pessimistic decision
criterion. This criterion is also known as Wald's criterion.

Decision making
under uncertainty 11.4.3 Equal Probabilities (Laplace) Criterion
is an environment in
Since the probabilities of states of nature are not known, it is assumed that all states of nature will ocCur
which the probability
of outcomes or ith equal probability, i.e. each state of nature is assigned an
equal probability. As states of nature are
states of nature can mutually exclusive and collectively exhaustive, so the probability of each of these must be: 1/(number or
not be quantified. states of nature). The working method is summarized as follows:

(a) Assign equal probability valuc to Cach state of nature

by using the formula:
+ (number of states of nature).
(hi Compute the expected (or averag) payo lOr each alternative (course of action) by adding all tnc
nayoffs and dividing by the number ot possible states of nature, or
by applying the formula:
(Probability of state of nalure ) * (Pavoff value for the
and state of nature combination of alternative
i

1 the best expected payofl value j.)

(maximum for
profit and minimum for cost).
 Decision
This. riterion is also known as the
Theory and Decision Trees
343
criterion of
Some intormation the likelihood of insufficient
occurrence of states reason. This is
because except in a few case>,
of nature is available.
Coefficier of
11.4.4
Optimism (Hurwicz) Criterion
erion
This c r i t e r i
suggests thai a
decision-maker should be neither
therefore, must
display amixture of both.
Hurwicz, completely optimistic nor pessimistic
f optim
coeficient of
optimism (denoted by a) to measure the who suggested this eriterion, intr
ntroduced the idea
and,
of a
lies ween 00 aand 1. where O
between
repreents a complete decision-maker's degree of This optimism. coeiticicnn
mistic attitude about the future.
Thus, if a is thepessimistic attitude about the future and I a compiete
the coefticient
of pessimism. coefficient of optimism, then (1 o,) will
represent
The Hurwicz appr0ach suggests that the
decision-maker must select an alternative that maximizes
H (Criterion of realism) =
a
(Maximum in column) + (1 a)
(Minimum column)
-

The working method 1s summarized as follows: in

(al Decide the coefticient of

optimism o (alpha) and then coefficient of
(b) For each decision alternative select the pessimism (1 a).
largest and lowest payoff value and multiply these with a
-

and (1 a) values, respectively. Then

calculate the weighted
(c) Select an alternative with best average, H by using above formula.
weighted average payoff value.
11.4.5 Regret (Savage) Criterion
This criterion is also known as
opportunity loss decision criterion or minimax regret decision criterion
because decision-maker regrets for Regret criterion
choosing wrong decision alternative resulting in an opportunity loss attempts to minimize
of payoff. Thus, he always intends to minimize this
regret. The working method is summarized as follows: ne maxnu
(a) From the given payoff matrix, develop an opportunity-loss (or regret) matrix as follows: opportunity loss.

)
Find the best payoff corresponding to each state of nature
Subtract all other payoff values in that row from this value.
()
(b) For each decision alternative identify the worst (or maximum regret) payoff value. Record this value
in the new row.
(C) Select a decision alternative resulting in a smallest anticipated opportunity-loss value.

Example 11.2 A food products' company is contemplating the introduction of a revolutionary new
product with new packaging or replacing the existing product at much higher price (S,). It may even make
a moderate change in the composition of the existing product, with a new packaging at a small increase
in price (S,). or may mall a small change in the composition ofthe existing product, backing it with the word
NEW and a negligible increase in price (S,). The three possible states of nature or events are: (i) high
nCTease in sales (N,), (ii) no change in sales (N,) and (i1) decrease in sales (N). The marketing department
n e company worked out the payoffs in terms of yearly net profits for each of the strategies of three
in the following table:
cEnts (expected sales). This is represented

States of Naure

N
Strategies N N
3,00,000 I,50,000
7,00.000
4,50,000
5,00,000
S2 3,00,000 3,00,000
3,00,000
S3

concerned executive choose on the basis of

Srategy should the
(b) Maximax criterion
(a) Maximin criterion (d) Laplace criterion?
C) Minimax regret criterion
344 Operations Research: Theory and Applications

Solution The payoff matrix is rewritten as follows:

(a) Maximin Criterion

Strutegie
States o
S
Nature S
3,00.000
N 7,00,000 5,00,000

4,50.000 3,00,000
3,00,000
3,00,000
N I,50,000
3,00,000- Maximin Pavoff
Column (minimum) I,50,000

The maximum of column minima is 3,00,000. Hence, the company should adopt strategy S

(b) Maximax Criterion

Strategies
States of
Nature Ss
N 7,00,000 5,00,000 3,00,000

Na 3,00,000 4,50,000 3,00,000

3,00,000
N3 1,50,000

Column (maximum) 7,00,000 5,00,000 3,00,000

L Maximar Payoff

The maximum of column maxima is 7,00,000. Hence, the company should adopt strategy S.
(c) Mininmax Regret Criterion Opportunity loss table is shown below:

States of
Strategies
Nature S, S3
N 7,00,000 7,00,000 ,00,000 5,00,000 7,00,000 3,00,000
= 2,00,000 = 4,00,000

N 4,50,000 - 3,00,000
4,50,000 4,50,000 4,50,000 - 3,00,000

1,50,000 = 0
1.50.000
N3 3,00,000 i,50,000 3,00,000- 0 3,00,000 3,00,000
= 1,50,000
3,00,000 = 0

Column (maximum) 1,50,000 3,00,000 4,00,000

LMinimax Regret
Hence the company should adopt minimum opportunity loss strategy, S.

(d) Laplace Criterion Assuming that eaclh state ot nature has a probability 1/3 of occurrence. Thus.

Expected Return (Rs)

Strategy
(7,00,000 3,00,000+ 1,50,000)/3 3,83,333.33 Largest Pavoff
(5,00,000 4,50,000 0)/3
3,16.660.o6
(3,00,000 3,00,0000 3,00,000)/3 3,00,000

Since the largest expected return is from stralegy . he executive must select strategy S.

Example 11.3 A manulacturer nianutae u e t , ol which the principal ingredient is a chemical

X. At the moment, the manulacturer spends Rs T,000 per ycar on supply of X, but there is a possibilty
that the price may soon crease to four ties Spresent figure because of a worldwide shortage of tne
 chemical. T h e ree is another chemical Y, which the
Decision Theory and Decision Trees
345
the same eitecl as
ive th
chemical X.manufacturer could use in
n
order io give
Chemicals Y and Z would conjunction
n

Rs 3,000 per vear. but their prices are with a third mi

hit and minimax unlikely to rise. together cost the
che
the ma
04. then find
ptimism is 0.4,
crileria for
the course of
What action should
decision-making
and give two sets
the manufacturer take? manutacturc
action that of solutions. If the
coetfficient
Appiy
Solution The data of the minimizes the cost. or

represents profit).
problem is
summarized in the
following table (negative figu
1igures in the table
States of Nature
Courses of Action
S,(use Y and Z)
N, (Price of X increases) S,fuse X)
N, (Price 3,000 4,000
of X does not
increase) 3,000 - 1,000

) Maximin Criterion

States of Nature
Courses of Action
S

N - 3,000
4,000
N2 - 3,000
1,000
Column (minimum) -3,000 4,000
T Maximin Payoff

Maximum of column minima 3,000. Hence, the manufacturer should adopt action
= -

(i) Minimax (or opportunity loss) Criterion S.

States of Nature Courses of Action
S
S2
N 3,000 (-3,000) = 0 - 3,000 --4,000) = 1,000

- 1,000 (-3,000) = 2,000

N2 - 1,000 (-1,000) = 0

Maximum opportunity 2,000 1,000 Minimax Payoff

Hence. manufacturer should adopt minimum opportunity loss course of action S.
(1) Hurwicz Criterion Given the coefficient of optimism equal to 0.4, the coefficient of pessimism will be
1- 0.4 0.6. Then according to Hurwicz, select course of action that optimizes (maximum for profit
and minimum for cost) the payoff value

H =
a Best payoff) + (1 a) (Worst payoff) -

= a (Maximum in column) +(1 - a) (Minimum in column)

Worst Payoff H
Course of Action Best Payoff
3,000 3,000
3,000 2 ,800
4,000
1,000
2
0,6(4,000) =
Rs 2,800, the
S cost (maximum protil)= 0.401,000)
t
Course of action S, has the least
nanufacturer should adopt this.
and percentage rates of retun
Example 11.4
allemalives
investment
An investor is given the following
(Market Conditions)
States of N a t u r e
Adiun TIiyh
Low
5
10
To 25
Regular shares 12
Risky shares 10% 30
8
-12%
Property
346 Operations Research: Theory and Applications
OVer the past 300 days, 150 days have been medium market conditions and 60 days havc nad high market

ncreases. On the basis of these data, state the optimum investment strategy for the invCSent
Nagpur Univ., MB4, 19991

Solution According to the given information. the probabilities of low, medium and high market conditions
would be 90/300 or 0.30, 1 50/300 or 0.50 and 60/300 or 0.20. respectively. The expected pay-olts Tor each of the

altenatives are shown below:

Strategy

Market Conditions Regular Shares Risky Shares Property

Probability
0.10 x 0.30 0.15 x 0.30
Low 0.30 0.07 x 0.30
0.25 x 0.50
0.12 x 0.50
Medium 0.50 -0.10 x 0.50

- 0.12 x 0.20 0.18 x 0.20 0.30x 0.20

High 0.20
126 0.230
Expected Return 0.136
Since the expected return of 23 per cent is the highest for property, the investor should invest in this alternative.

cONCEPTUAL QUESTIONS A

What is a scientific decision-making process?

Discuss the
4.
1. Discuss the difference between decision-making under role of the statistical method in such a process.
certainty. under uncertainty and under risk.
an example of a good decision that you made,
which
5. Give
2 What techniques are used to solve decision-making problems resulted in a bad outcome. Also give an example of a good
under uncertainty? Which technique results in an optimistic decision that you made and that had a good outcome. Why
decision? Which technique results in a pessimistic decision? was each decision good or bad?
3 Explain the various quantitative methods that are useful for
decision-making under uncertainty.

SELF PRACTICE PROBLEMS A

The following matrix gives the payoff (in Rs) of different strategies Strategy State of Nature
(alternatives) S,. S and Ss against conditions (events) N. N
N and N N N2
State of Nature 40 60
Ng 10 20
Strategy N Na 40
P3 150
- 100 6,000 18,000
S 4,000
400 0 Select strategy using each of the following decision criteria: (a)
a
20,000 5,000
20,000 15,000 2,000 1,000 Maximax, (6) Minimax regret, (c) Maximin, (d) Minimum risk,
assuming equiprobable states.
Indicate the decision taken under the following approaches: () 4. Mr Sethi has Rs 10,000 to invest in one of three options: A, B
Pessimistic, (i) Optimistic, (ii) Equal probability, (iv) Regret, (v) or C. The retürn on his
investment depends on whether
Hurwicz criterion, the degree of optimism being 0.7. ne
economy experiences inflation, recession, or no change at all.
2. In a toy manufacturing company, suppose the product acceptance The possible returns under each economic
condition are given
data is known: below:
probabilities are not known but the following

Anticipated First Year Profit ('O00 Rs)

State of Nature
Product Line Strategy Inflation Recession No Change
Product 2,000
Partial Minimal 1,200 1,500
Acceptance Full B 3,000 800 |,000
50 C 2,500 1,000
Good 8 70 1,800
Fair 45 40
50 What should he decide,
- 25 10 using the pessimistic criterion, optimisu
Poor criterion, equally likely criterion and
regret criterion?
following9 5. A manufacturer's
optimal decision under each
of the
Determine the representative has been offered a new produ
decision criteria and show how you arrived at
it line. It he accepts the new
line he can handle it in one
Minimax of
a) Maximax, (b) Maximin, (c) Equal likelihood and (d) two ways. The best
way according to the manufacturer
woud
be to set a
regret? separate sales force to exclusively handle the
line. This would involve an ne
table for three stralegies initial investment of Rs
3 The following
is a payoff (in rupees) the office, office 1,00,000
and two states of nature: equipment and the hiring and training o
 On the
On other hand, it the
othe
salesmen.
new line is Decision Theory and Decision Trees 347
es
existing sales force, using the handled by
investment would only bee Rs existing facilities, the
30,000, the initial Territory and their experience with other products, the
present salesmen.
principally for
training his epresentative estimates the following probabilities for annual
he new product sells for Rs sales of the
250.
new
product:
normally receives 20 per cent of the sales The representative
which
sold, of which 10 per cent is paid as price on each unit Sales (in units)1,000 2,000 3,000 4,000 5,000
new product. The manufac
acturer offers commission
to pay 60
to handle
the Probability 0.10 0.15 0.40 0.30 0.05
price of each unit sold per cent of the
to the (a) Set up a regret table.
sacentative sets up a representative,
the normal 20 per Cent wllseparate sales if the (D) Find the expected regret of each course of action.
De paid. organization. Otherwise Which course of action would have been best under the
ots a 10 per cent commission.
In either case the
salesman (C
Based on the size of maximin criterion?
the

HINTS AND ANSWERS

I. ) S.(11) S, or S3 (ii)
S (iv) S. (v) S
.(a) Full. (b)
Minimal. (c) Full or
partial, (d) Partial Therefore, payoff function corresponding to S, and S, would
be
(a) S: Rs 150 (b) S3: Rs 80 (c) S; Rs 40 (d) S; Rs 55 S, =-1,00,000 + 250x {(30 10)/100 x a = -1,00,000 + 500
4. Choose A: Rs 120. Choose B: Rs 300, Choose C: Rs
Choose C: Rs 50 176.6, S =30,000+ 250 x {(20 10)/100} x a = - 30,000 + 25a
Equating the two, we get - 1,00,000 + 500a = - 30,000 +25a
5. Let S = install new sales facilities
or a = 2,800.
S = continue with existing sales facilities.

11.5 DECISION-MAKING UNDER RISK

In this decision-making environment, decision-maker has sufficient information to assign probability to the
likely occurrence of each outcome (state of nature). Knowing the probability distribution of outcomes
states of nature), the decision-maker needs to select a course of action resulting a largest expected
(average) payoff value. The expected payoff is the sum of all possible weighted payoffs resulting from
choosing a decision alternative.
The widely used criterion for evaluating decision alternatives (courses of action) under risk is the
Expected Monetary Value (EMV) or Expected Utility.

11.5.1 Expected Monetary Value (EMV)

ne expected monetary value (EMV) for a given course of action is obtained by adding payoff values multiplied Expected monetary
by the probabilities associated with each state of nature. Mathematically, EMV is stated as follows: value is obtained by
adding payoffs for
EMV (Course of action, S)= , Pi Pi each course of
i=l action, multiplied by
the probabilities
where =
number of possible states of nature
m associated with
probability of occurrence of state of nature, N, each state of
P associated with state of nature N, and course of action, S, nature.
Pii payoff

The Procedure
all possible courses of
action and states of nature. Enter the
onstruct a payoff matrix listing combination of course of action and state of
associated with each possible
COnditionalpayoff values each state of nature.
of the occurrence
of
dure along with the probabilities of action by multiplying the
conditional payotfs by the associated

alculate the EMV for each course of action.

values for each c o u r s e

these weighted
PrODabilities and adding the optinmal EMV.
tlect the course of action that yields
B. He c a n use the airport bus which costs
town A to town
Example
Rs 25
5 Mr X flies quite
is
often from
a 0.08
chance that he
W i l l m i s s the tlight.
The stay in a hotel costs
there use a taxi which will
but if he takes it, on time
for the 1light.
For Rs 350 he can

/ w i t h a 0.96 chance of being Mr A calches the plane on time, he will

on time
for the flight. I1
chance of being KS T0,000, otherwise he will lose it.
p e r cent a prolil ol
transaction that will produce EMV criterion.
de a business on the basis
of the
u should Mr X
use? Answer
"node
oftransport
348 Operations Research: Theory and Applications
of action is Shown in lable 113
courses
with various
Omputation of EMV associated
Courses of Action

States of Stayin Hotel Taxi

Bus Cost Prob. Expected
Nature Prab. Expected
Cost Prob. Expected Cost
Value Value
Talue
9,340.80 10,000 350 0.99 9,553.50
270 0.96
Catches 10.000 25 0.92 9,177 10,000 9.650
the flight = 9,975 9,730

0.04 10.80
- 350 0.01 3.50
Miss the flight 25 0.08 2.0 270

9,330 9,550
Table 11.3
Expected monetary 9,175
value (EMV)

Since EMV associated with course of action Taxi' is largest (= Rs 9,550), it is the logical alternative.

Example 11.6 The manager of a flower shop promises its customers delivery Within four hours on all
Tlower orders. All flowers are purchased on the previous day and delivered to Parker by 8.00 a!m the next
morming. The daily demand for roses is as follows.
Dozens of roses 7 0 80 90 100
Probability 0.1 0.2 0.4 0.3
The manager purchases roses for Rs 10 per dozen and sells them for Rs 30. All unsold roses are donated
to a local hospital. How many dozens of roses should Parker order each evening to maximize its profits?
What is the optimum expected profit? [Delhi Univ, MBA, Dec. 20041

Solution The quantity of roses to be purchased per day is considered as 'course of action' and the
daily demand of the roses is considered as a 'state of nature' because demand is uncertain with
kno vn probability. From the data, it is clear that the flower shop must not purchase less than 7 or
more than 10 dozen roses, per day. Also each dozen roses sold within a day yields a profit of Rs (30
-

10) Rs 20 and otherwise it is a loss of Rs 10. Thus

=

Marginal profit (MP) = Selling price - Cost 30 - 10 Rs 20

Marginal loss (ML) = Loss on unsold roses = Rs 10

Using the information given in the problem, the various conditional profit (payoff) values for each
combination of decision altermatives and state of nature are given by
Conditional profit = MP x Roses sold
ML Roses not sold
20D, if D2S
20D-10(S- D) = 30D-10S, if D<S
where D number of roses sold within a day and S =
number of roses stocked.
The resulting conditional protit values and
corresponding expected pavoffs
Table 11.4. are computed
States of Probability Conditional Profit (Rs) due to
Nature Couwses of Action Expected Payoff (Rs) due to
(Demand (Purchase per Dav) Courses of Action
per Day) 70 80
(Purchase per Day)
90 100 70
(1) (2 (3) (4) (5) ()()
80 90 I00

()(3) (14)
70 0. 140 30 120 T10
13 12 11
80 0.2 140 160 150 140
28 32 28
90 04 140 160 0
T80 170
56 64 68
Table 11.4 100 03 140 160 180
200
72
Conditional Profit 42 48 54 60
Value (Payoffs) Expected monetary value (EMV)
140 157 168 167
 Decision Theory and Decision Trees 349
Since the highest EMV O KS 168
rchase nine dozen roses
everyday. corresponds to the course of
action 90, the flower shop
ample 11.7
Exan A retailer should
that
purchases cherries
every morning at Rs 50 a
case. Any case
remains unsold the end of the
at case and sells them for 80 a
alue of Rs 20 per case (thereafter day can be
dispos
posed of the next day at a Ks ou
salvage
The following is the record of
they have no value). Past
sales for the sales Ihave ranged from 15 to 18 cases
per
past 120 days.
Cases sold
15
16 7
Number of days 12
18
24 48
Find out how many cases should the
retailer purchase perday in order to maximize his profit.
Delhi Univ., MCom, 2000; Ajmer Univ, MBA, 2003]1
Solution Let N, (i =1, 2. 3, 4) be the
possible states of nature (daily likely demand) and
4) be all possible courses of action S, (j F1, 2,
(number of cases of cherries to be
Marginal profit (MP) Selling price -
purchased).
Cost =
Rs (80 50) Rs 30
Marginal loss (ML)
- =

=
Loss on unsold cases Rs
=

(50 20) =
Rs 30
The conditional profit (payoft) values for each
combination of decision alternatives and state of nature
are given by

Conditional profit =
MP x
Cases sold -

ML *
Cases unsold
=
(80 S0) (Cases sold) -

(50 -20) (Cases unsold)

30S ifS2 N
(80 50) S - 30(N - S) = 60S - 30N i f S< N

The resulting conditional profit values and corresponding expected payoffs are computed in Table 11.5.

States of Probabilit Conditional Profit (Rs) due to Expected Payof R due

Nature Courses of Action Courses e Aeian
(Demand (Purchase per Day) Purchase per a
per Week)
4 5
15 0.1 450 420 390 360 45 42 39 6

16 0.2 450 480 450 420 90 96 90 84

450 480 S10 480 180 192 204 192

17 0.4
540 135 144 153 Table 11.5
0.3 450 480 510 162
Conditional Profit
450 474 486 474 Value (Payoffs)
Expected monetary value (EMV)

to the course of action 17, the retailer must purchase

Since the highest EMV of Rs 486 corresponds
cases of cherries every morning
demand for hiring cars on any day in à given city is as follows:
Example 11.8 The probability of
2 3 4
No. of cars demanded
0.2 0.3 0.2 0.2
0.1
Probability hire charges (variable costs of running)
to keep the daily
Cars have a fixed cost of Rs 90 each day
its daily expectation? If the company is about to go
cars, what is
S 200. If the car-hire company owns 4
car, how many
cars should t buy?
into Dusiness and currently has no

variable cost. The payoff values with 4 cars

fixed cost and Rs 200
1s
OOlution Given that Rs 90 is the
calculated as under:
atthe disposal of decision-maker are
No. of cars 3 4
demanded 200 90 x 4
400 90 x 4 600 90 x 4 800 9 0 x 4

Payoff 0 90 4
160 40 240 440
(with 4 cars) 360
350 Operations Research: Theory and
Applications
nus, the daily expectation is obtained hy multiplving the payoff values with the given cOTespondine

probabilities of demand:
(440)X0.2) Rs
any Expectation =(-360)%0.1) (- 160)%0.2) +(40)%0.3) (240(0.2)
+
+ =
S80
+
he conditional payoffs and expected pavoffs for each course of action are shown in lables I1.6 and 11.7

Demand of ro Decision to Purchase Cars

Probability Conditional Pavoff fRs) due
Cars (Course of Action)

0.1 0 90 180 270 360

1 0.2 110 20 70 160
Table 11.6
0.3 110 220 130 40
Conditional Payoff 0.2 110 220 330 240
Values 330 440
0.2 110 220

of Probability Conditional Payof (Rs) dueto Decision to Purchase Cars

(Course ef Action)

0.1 18 27 36
0.2 22 4 4 32
0.3 0 3 66 39 12
Table 11.7 0.2 0 22 44 48
Expected Payoffs 0.2 0 22 14 66 88
and EMV EMV 0 90 140 130 80
Since the EMV of Rs 140 for the course of action 2 is the highest, the company should buy 2 cars.

11.5.2 Expected Opportunity Loss (EOL)

Expected opportunity loss (EOL), also called expected value of regret, is an alternative decision criterion
for decision making under risk. The EOL is defined as the difference between the highest
and the actual profit due to choosing a particular course of action in a particular state of nature. payoff)
profit (or Hence,
FOLis the amount of payoff that is lost by not choosing a course of action
resulting to the minimum payoff
in a particular state of nature. A course of action resulting to the minimum EOL is
EOL is stated as follows.
preferred. Mathematically.

EOL (State of nature, N)= 2, Pi

where
=
opportunity loss due to state of nature,
N, and course
of action, S.
p. = probability of occurrence of state of nature, N,

The Procedure
renare conditional payoff values matrix Tor each combination of
a
course of action and state of nature
along with the associated probabilities.
Eor euch state of nature caleulate the conditional opportunity loss
the maximum payoff.
(COL) values by subtracting each
payoff from
aCalculate the EOL for each course o Con by multiplying the
probability of each state of nature
with the COL value and then adding the values.
which the EOL is minimunm
4 Select a course of action 1or
Example 11.9 A company manulactures
goods for a market in
which the technology of the
changing rapidly. The research and development department has . produet
have potential for commercial exploitation.
A further Rs ced a new product th:
60,000 is required
100 customers and cach customer mie for development testing
The company has the
product. Market research suggests that a selling price of Rs 6,000rchase,
for each unit,
at thewith total variable
costs
of manufacturing and selling estmate as Ks 2,000 for each unit
 Decision Theory and Decision Trees 351
From pTeVIOus Cxperience, it has been
customers who will buy the possible to derive a
probability distribution relating to tne
propor
Proportion of customers
product as follows:
0.04 0.0
Probability 0.12 0.16 0.20
0.10 0.10 0.20 0.40 0.20
Determine the expected opportunity losses,
given no other information than that stated above,
afe whether or not the
company should develop the and
product.
Solution Ifp is the
proportion of customers who
rafit is: (6.000 2,000) 100 p -60,000 Rsx purchase the new product, the company's conditional
(4,00,000p
=

Let N, (i 1,2,. .. 3) be the possible states

= 60,000). -

of nature, i.e. proportion of the customers

who will buy
the new product and S, (develop the product) and
of action.
S, (do not develop the product) be the two courses
The conditional profit values (payoffs) for each
pair of Ns and Ss are shown in Table 11.8.

Proportion of Conditional Profit Rs (4,00.00026

Customers Course of Action
(State of Nature)
(Develop Do nol Develop)
0.04 ,000
0.08 -28,000
0.12 12,000 Table 11.8
0.16 4,000 0 Conditional Profit
0.20 20,000 0. Values (Payoffs)

Opportunity loss values are shown in Table 11.9.

Proportion of Probabilit Conditional Profit(Rs)

Customers S
Si
(State of Nature)
0.04 0.1 ,000 44,000 0
0.08 0.1 28,000 28,000
0.12 .2 12,000 12,000 0 Table 11.9

0.4 4,000 0 4,000 Opportunity Loss

0.16
0
20,000 Values
0.20 0.2 20,000
associated with each state of nature, the expected
Using the given estimates of probabilities below:
course of action is given
opportunity loss (EOL) for each
+0.2 (12,000) +0.4 (0) + 0.2 (0) =Rs 9,600
EOL (S,) =0.1 (44,000) +0.1 (28,000)
+0.2 (0) +0.4 (4,000) +0.2 (20,000) =Rs 5,600
EOL (S,) 0.1 (0) +0.1 (0)
the expected opportunity loss, the company
should select course Expected value of
Since the company seeks to minimize perfect information
with minimum EOL.
O1 action
S, (do not develop the product) is an average (or
expected) value of
Perfect Information (EVPI) an additional
11.5.3 Expected Value of information if it were
about the occurrence of various
and accurale) information of any worth.
I decision-makers can get perfect (complete the desired payoff in the presence of any
that yields
course of action
Statesof nature, then choosing a

state of nature is easy. that optimizes

to select a particular course of action
the decision-maker
The EMV EOL criterion helps
or value of perfect information (EVPI)
additional informalion. Expected
without any additional information about the
ne
expected payoff, to pay tor getting
money required
Cpresents the maximum anmount of decision. Mathematically, it is stated as:
arriving t0 a
eurrence of various states
of nature before
information)
with perfect
EVPI (Expected profit
without pertect information)
=

(Expected profit

P, max(P,)
EMV*

i = l
 352 Operations Research: Theory and Applications
where Pij best action, S, is taken
payoff when in the presence of state of nature. N

P probability of state of nature. N,

EMV*= maximum expected monetary value
Exdmple 11.10 A company needs to increase its production beyond its existing capacity lt h
narrowed down on two alternatives in order to increase the production capacity: (a) expansion. at a Ca
st
oT Rs 8 million, or (b) modemization at a cost of Rs 5 million. Both approaches would require the samo
uount of time for implementation. Management believes that over ihe required payback period, demand
will either be
high or moderate. Since high demand is considered to be somewhat less likely than moderate
demand, the probability of high demand has been set at 0.35. If the demand is
an
high, expansion would gross
estimated additional Rs 12 million but modernization would only
grOSs an additional Rs 6 million, due
to lower
maximum production capability. On the other hand, if the demand is moderate. the
Tigures would be Rs 7 million for expansion and Rs 5 million for modermization. comparable
EVPI provides an (a) Calculate the conditional
instant way to profit in relation to various action-and-outcome combinations and states of
nature.
check whether
(6) If the company wishes to maximize its
getting any
(c) Calculate the EVPI.
expected monetary value (EMV), should it modernize or expand?
additional information
might be worthwhile. (d) Construct the conditional
opportunity loss table and also calculate EOL. [Delhi Univ, MBA, 2004
Solution (a) States of nature:
High demand and Moderate demand, and Courses of action: Expand
Modernize. and
Since probability of high demand is estimated at
(1 0.35, the probability of moderate demand must be
-

0.35) =
0.65. The calculations for conditional
profit values are shown in Table 11.10.

State of Nature Conditional Profit (million Rs) due to

(Demand Course of Action
Table 11.10 Expand (S,)
Conditional Profit
Modernize (S)
High demand (N) 12 8 =
Table 4
6 5 1
Moderate demand (N,) 7- 8 = -
5 5 0
(b) The payoff table (Table 11.10) can be rewritten as follows along with the
of nature. given probabilities of states
State of Nature Probability
(Demand Conditional Profit (million Rs)
Due to Course of
Action
Expand Modernize
Table 11.11 High demand 0.35
Payoff Table Moderate demand 0.65

The calculation of EMV for each course

of action
S, million
EMVS)-0.35(4) +0.65( 1)) =Rs 0.75
and S, is given below

Since EMV
EMV(S,)-0.35(1) +0.65(0) =
Rs 0.35 million
(S,) =0.75 million is maximum, the
(c) To calculate company must choose course of action
EVPI, first calculate EPPI by choosing S,(expand).
Multiply conditional profit associated with each course of action
optimal course of action for each
state of nature.
profit, and then add these weights as by the given probability
shown in Table 11.12. to get weighted
State of Nature Probability
(Demand)
Optimal Course Profir from Optimal
ofAction Course of Action
(Rs million)
Condirional Profit Weighted Profit
High demand 0.35
Moderate demand 0.65 4 4 x 0.35 1.40
Table 11.12 0x 0.65 =0
EPPI 1.40
 Decision Theory and Decision Trees
If optimal EMV* =F
Rs 0.75
million 353
EVPI EPPI =
corresponding S. then
to the course of action
EMV(S,)
-

1.40 0.75
=

In other words, to
get perfect information on Rs 0.65 million
paying up to Rs 0.6S million. demand pattern (high or
moderate), company should consider
d The opportunity loss values are
shown in Table 11.13.
Stote of Nature
(Demund) Probability Conditional Profit Conditional Opportunity
(Rs million) Due to Loss (Rs million)
Course of Action Due to
Course of Action

High denmand (N,) 0.35

Moderate demand (N) 0.65
0 1 Table 11.13
Since probabilities associated with each state of nature,
opportunity losses for the two courses of
P(N,) =
0.35, and P(N,) =0.65, the expected
action are:
EOL(S)=0.35(0) + 0.65(1) = Re 0.65 million
EOL(S,) =0.35(3) +0.65(0) =Rs 1.05 million
Since the expected
course of
opportunity loss, EOL (S) Re 0.65 million minimum, decision-maker must select
=

action S, so as to have smallest

expected opportunity loss.
Example 11.11 A certain piece of
location. This equipment contains an
equipment has to be purchased for construction
project at a remote a

can be
expensive part that is subject to random failure. Spares of this part
purchased at the same time the equipnment is purchased. Their unit cost is Rs
1,500 and they have
no
scrap value. If the part fails on the job and no spare is available, the
part will have to be manufactured
on a
special order basis. If this is required, the total cost including down time of the
at Rs 9,000 for each such occurrence. Based on equipment, is estimated
previous experience with similar parts, the following
probability estimates of the number of failures expected over the duration of the project are provided below:
Failure 2

Probability 0.80 0.15 0.05

a) Determine optimal EMV* and optimal number of spares to purchase initially
o) Based on opportunity losses, determine the optimal course of action and optimal value of EOL
e) Determine the expected profit with perfect information and expected vaiue of perfect information.

Solution (a) Let N, (no failure), N, (one failure) and N, (two failures) be the possible states of nature
(1.e. number of parts failures or number of spares required). Similarly, let S, (no spare purchased), S, (one
spare purchased) and S, (two spares purchased) be the possible courses of action.
The conditional costs for each pair of course of action and state of nature is shown in Table 11.14.

State of Nature Course of 4ction Purchase Emergeny Total Condirional

(Spare Required) (Number of Spare Purchased) Cost (Rs) Cost (Rs) Cost tRs)

N I,500 I,500
3,000 3,000
9,000 9,000
,500 I,500
3,000 3,000
I8,000 I8,000
I,500 9,000 10,500
3,000 3,000 Table 11.14

ol stales ol ialure, the expected onetary value can

OSing the conditional costs and the probabilitues
be ias shown in Table 11.15.
l a t e d for each of three states of nature
354 Operations Research: Theory and Applications
Conditional Cost Drue to
Weighted Cost De to
State of Nature Probability
Course of Action
Course of Action
(Space Required)

3,000 0.80 (0) 200 2.400

N 0.80 0 1.500
= 0

3,000 0.15 (9.000) 225 450

N2 0.15 9,000 1.500
= 1,350

0.05 (18.000)
Table 11.15 N3 0.05 18,000 10,500 3.000
Expected 900 525 I50
Monetary Value EMV = 2,250 1,950 3,000

Since weighted cost = Rs 1,950 is lowest due to course of action, S,, it should be chosen. If the EMV
Is expressed in terms of profit, then EMV* = EMV(S,) = - Rs 1,950. Hence, the optimal number of spares
to be
purchased initially should be one.
(6 Thecalculations for conditional opportunity loss (COL) to determine EOL are shown in Table 11.16
State of Nature Conditional Cost Due to Conditional Opportunity Loss Due to
(Space Required) Course of Action Course af Action

Table 11.16 S
Conditional N 1,500 3,000 1,500 3,000
Opportunity Loss
(COL) No 9,000 1,500 3,000 7,500 ,500
Na 18,000 10,500 3,000 15,000 7,500
Since we are
dealing with conditional costs rather than conditional
state of nature shall be considered for calculating
profits, the lower value for each
opportunity loss are shown in Table 11.17. opportunity
losses. The calculations for expected
State of Nature Probabiliry Conditional Opportunity
(Space Reguired) Loss (Cost) Due to Weighted Opportunity
Loss (Cost) Due to
Course of Acticn
Course of Action

N 0.80 1,500 3,000 0.80(0) 1,200 2.400

0
N2 0.15 7.500 0 1,500 0.15 (7,500) 0 225
Table 11.17 = 1,125

Expected N3 0.05 15,000 7,500 0 0.05 (15,000) 375 0

Opportunity Loss = 750

(EOL)
EMV 1,875 1,575 2,625
Since minimum, EOL* =EOL(S,): Rs 1,575,
therefore adopt course of action and
(c) The expected profit with perfect infornmation (EPPI) can be
S, purchase one spare.
determined by selecting the
of action for cach state of nalure, muliplying its conditional values optimal course
by the corresponding
then adding these products. The FPPl calculations are shown
in Table 11.18,. probability and

States of Probability Optimal Course Cost of Optimal Course of Action (ks)

Nature of Action
Conditional
(Space Cost Weighted
Required) (Minimunn Value) Opportunity Loss
N 0.80
0 0.80(0) =
N 0.15 500 0.15 (1,500) = 225
N 0,05 S 3,000 0.05 (3,000) = 150
Table 11.18
Total = 375
 Decision Theory and Decision Trees
355
Thus expected profit with perfect
information is, EPPI -Rs
is EVPl EPPl -

EMV = -

375 -
375.
(-1,950) Rs 1,575. It may beExpected
=
value of perfect information then
observed that, EVPI EOL* Rs 1,5/5 = =

Example 11.12 XYZ

Company manufactures parts for passenger cars and sells
parts each. The company has a policy of them in lots of 10,000
Five inspection categories, established forinspecting each lot before it is actually
shipped to the retailer.
quality control, represent the percentage of defective items
contained in each lot. These are given in the
following table. The daily inspection chart for past 100
inspections shows the following rating or breakdown inspection: Due to this the
two possible courses of action: management is considering
G) S: Shut down the entire plant operations and thoroughly inspect each machine.

Rating Proportion of Frequency

Defective ltems
Excellent (A) 0.02 25
Good (B) 0.05 30
Acceptable (C) 0.10 20
Fair (D) 0.15 20
Poor (E) 0.20 5
Total 100

(11) S:Continue production as it now exists but offer the customer a refund for defective items that
are discovered and subsequently returned.
The first alternative will cost Rs 600 while the second alternative will cost the company Re 1 for each
defective item that is returned. What is the optimum decision for the company? Find the EVPI.

Solution Calculations of inspection and refund cost are shown in Table 11.19.

Rating Defective Probability Opportuniy Loss

Rate inspect Rgund nspect
0.25 600 200 400
0.02
0.30 600 500 100 0
B 0.05
600 1,000 400
0.10 0.20
600 1,500 0 900
D 0.15 0.20 Table 11.19
600 2,000 1,400
E 0.20 0.05 Inspection and
670 EOL 170* 240 Refund Cost
1.00 600*

follows:
of refund is calculated
as
The cost
0.02 1.00 =
Rs 200
For lot A : 10,000
refund for other lots
is calculated.
Smilarly, the cost of
refund is:
Expected cost
of
200 0.25 + 500 0.30 +... + 2,000 x0.05 =Rs 670
Expected cost of inspection is: Rs 600
t 600 0.05
0.30 .
600 0.25 609
. .

Ihe plant should be shut down

for
more than
the Cost ol inspection,
refund is
Since the cost of Rs 170.
spection. Also, EVPI
EOL of inspection
=

doll with three

of nmanulacturng a daneng
is considerng a projcct
manufacturer mOvement design using
Ks T0. The tirst
EXample 11.13 A toy will be sold at
an iverage o
The doll 1,00,000 and Rs 5 per unit of
uferent m o v e m e n t designs. 1 Rs
setl up cost
lowest tooling and and variable cost
the cost of Rs I,60,000
Sears and levels will provide actjon will
have a lixcd
have a lixed cost of Rs
3,00.000
design with spring and pulleys will
dable cost. A second
va
with weights the doll and the probability
of their
of another design can o c c u r lor
4 per unit. Yet events h a l
S demand
and variable cost Rs 3 per unit. The
OCcurrence is given below:
356 Operations Research: Theory and Applications
Demand (units)
Probahilin

25,000 0.10
Light demand
1,00,000
0.70
Moderate demand
1.50,000 0.20
Heavy demand

(a) Construct a payoff table for the above project.

(b) Which is the optimum design? the demand
(C) How much can the decision-maker afford to pay in order to obtain perfect information about
Solution The calculations for EMV are shown in Table 11.20.
+Demand x Variable cost)
Payoff (Demand Selling price) (Fixed cost
Revenue - Total variable cost - Fixed cost

Expected Payof (Rs) Due to

States of Probabiliny Conditional Payof (Rs) Due 1o
Courses of Action
Nature Courses of Action
(Demand (Choice of Movements)
Gears and Spring Weights Gears and Spring Weights
Levele Action and Pulleys
Leveis Action and Pulleys
- 1,25,000 2,500 1,000 500
Light 0.10 25,000 - 10,000

Moderate 0.70 4,00,000 4,40.000 4,00,000 2.80,000 3,08.000 2.80,000

Table 11.20 1,48,000 I.50,000
Heavy 0.20 6,50,000 7,40,000 7,50,000 1,30,000
EMV and Payoff
Values EMV 4.12.500 4,55.000 4.17,500

Since EMV is largest for spring action, it is the one that must be selected.

States of Probability Caurses of Action

Nature Gears and Spring Weights Maximum Marimum Payoff
(Demand Levels Action and Pulleye Payoff xProbability
Light 0.10 25,000 - 10,000 - 1,25,000
25.000 500
Table 11.21 Moderate 0.70 4,00,000 4,40,000 4,00,000 4,40,000 3,08,000
Expected Payoff 0.20 650,000 7,40,000 7.50,000 7,50,000
with Perfec Heavy 1,50,000

Information Total 4,60,500

The maximum amount of money that the decision-maker would be willing to pay in order to obtain
perfect information regarding demand for the doll will be
EVPI Expected payoff with perfect information -
Expected payoff under uncertainty (EMV)
= 4,60,500-4,55,000 = Rs 5,500

Example 11.14 A TV dealer finds that the cost of holding a TV in stock for aweek is Rs. 50. Customers who
cannot obtain new TV sets immediately tend to go to other dealers and he estimates that for every customer
who cannot get immediate delivery he loses an average of Rs. 200. For one particular model of TV the probabilities
ofdemand of0, 1,2,3,4and 5 TV sets in a weekare 0.05, 0.10,0.20,0.30, 0.20 and 0.15. respectively
(a) How many televisions per week should the dealer order? Assume that there is no time lag between ordering
and delivery.
(b) Compute EVPI.
(c) The dealer is thinking ofspending on a small market survey to obtain additional information regarding the
demand levels. How much should he be willing to spend on such a survey. Delhi Univ, MBA. 2000]
Solution IfD denotes the demand and S the number of televisions stored (ordered). then the conditional
cost values are conmputed and are shown in Table 15.22.
50.S+200 ()- S), when D2S
COst function
50/+50(S-D). when D<S
 Decision Theory and Decision Trees
357
State ofNature
Probability
Demand) Conditional Cost(Rs.)
Course ofAction (Stock)
2
0.05
50 100 50 200 250
0.10
200 50 00 150 200 250
0.20 400 250 00 150 200 250
0.30 600 450 300 150 200 250
0.20 800 650 500 350 200 250
0.15 1000 850 700 550 400 250 Table 11.22
Expected Cost = 590 450 330
Expected Cost
250 230 250

Stale of Natare
Probahility Minimum Cost Expected Costfor
(Demandh
for Perfect Information Perfect Information
(2) (3)-u)x 12
0.05
0.10 50 5
0.20 100 20
0.30 150 45
0.20 200 40 Table 11.23
0.15 250 37.5
Expected Payoff
with perfect
ECPI = 147.5 Information
EVPI = Conditional Cost - ECPI = 230 147.5 =Rs. 82.5.

The pay-offor EMV is shown in Table 15.24

State of Natue Probability Conditioinal Payoff

Demand Course ofAction (Stock

0 0.05 i00 150 200 250

0.10 150 100 50 50
G.20 150 300 250 200 150

0.30 150 300 450 400 350

0.20 150 300 450 600 550
300 450 Table 11.24
0.15 I50 600 750
Computation of
EMV 0 140 260 340 360 340 EMV

State of Naiure Probahility Payoff for Epected Payoff

Perfect Information for Perfect nformation
(Demand)
(2) (3)(/)* (2)
0.05
50 15
0.10
0.20 00 60
4 50 135
036
600 120
0.20 Table 11.25
750 112.5
0.15 Computation of
442.5 EPPI
358 Operations Research: Theory and Applications
The expected value of perfect information is given by
EVPI EPPI EMV* = 442.5-360 Rs. 82.5.
(C)On the basis of the given data, the dealer should not be willing to spend more than Rs. 82.5 for the marker

survey.
Example 11.15 XYZ company is considering issuing 1,00,000 shares to raise capital neededfor expansion. It
is estimated that ifthe issues were made now, it would be fully taken up at a price of Rs. 30 per share. However, the
company is facing two crucial situations, both of which may influence the share price in the near future. namely
()Awage dispute with tool room operators which could lead to a strike and have an adverse effect on the
share price.
(1) The peOssibility ofa substantial contract with a large company overseas which would increase the share
price.
The four possible outcomes and their expected effect on the company's share prices are:

E: No strike and contract obtained-share price rises to Rs. 34.

E: Strike and
contract obtained-share price stays at Rs. 30.
E3: No strike and contract lost-share price raises to Rs. 32.
E4 Strike and contract lost-share price falls to Rs. 16.
Management has identified three possible strategies that the company could adopt, namely
A: Issue 1,00,000 shares now
A Issue 1,00,000 shares only after the outcomes of (i) and (ii) are known
A: Issue 5,00,000 shares now and 50,000 shares after the outcomes of (i) and (ii) are known.
(a) Determine the maximax solution. What alternative criterion
might be used?
(b) It has been estimated that the probability of a strike is 55
per cent and that there is a 65 per cent chance
ofgetting the contract, these probabilities being independent. Determine the optimum policy for the
company using the criterion of maximizing expected pay-off.
(c) Determine the expected value of perfect information for the
company.
Solution The payoff values are shown in Table 11.26

States of Probability Conditional Pavoff (Rs. lakh) Conditional Loss (Rs. lakh)
Nature AlternativeStrateg Alternative Strategy

E (1-0.55) x 0.65 0.2925 30

A A
34 0.5 x 30+0.5 x 34 3 1
4
E 0.55 0.65 0.3575 30 30 0.5 x
30 +0.5 x
30 =
30
E3 (1-0.55)(1 -0.65) 0.1575 30 32 0.5 x 30 +0.5 x 32
30.5
0.55 (1-0.65) 0.1925 1.5
Ea 30 16 0.5 x
30+0.5 x
16 23 0 4
Table 11.26 Expected Values 30 34
31

A strategy with highest minimum (maximin) payoII (1.e. s0) 1s A, and a

strategy with highest (maximax) pay-
off i.e. 34) is A,. Since highest pay-oll ot Rs. 50 lakh is obtamed
should adopt
stralegy A
corTesponding to
strategy A,. the company
(c) Calculations for expected value ol ihe pertect intormation are shown in Table
15,27.
State of Nature Maximunm Pay-off
Probability Expected Pay-offwith
34
Perfectlnformation
.2925
30 94
.3575
32 10.72
0.1575
Table 11.27 6
1925
5.04
EVPI 5.778
31.48
Hence, expected valuc ol perlect inlOrialion S: 31.48 30 Rs. 1.48.00
0.
 Decision Theory and Decision Trees 359
Example 11.1b A
Vegetable seller buys tomatoes for Rs.
45 a box and sells them for Rs. 80
box is not soid on the irst
selling day, it is worth Rs. 15 as per box. It the
normally distriubuted, with a mean of 30 boxes daily salvage. The past records indicate that demand is
should he stock?
and a standard deviation of 9 boxes. How many boxes

Solution The probability of selling at least one additional unit (box)

is given by to justify the stocking of additional unit
IL
(45- 15)
P
IP +IL
(80-45)+ (45 - 15)
30=0.462
65
where Incremental loss (IL) =Cost price
Incremental protit (IP) Selling priceSalvage price
-

-Cost price
This implies that the vegetable seller must be 46.2 per cent sure
of at least one additional unit
selling 30
before he should pay to
O = 9
stock anadditional unit. The vegetable seller should stock
additional boxes until point A is reached. If more units are stocked, 0.538
then the probability will fall below 0.462. 0.462
Fig. 11.1
The point A is at 0.1 standard deviation to the right of the mean - 30 A
3 0 . Since the standard deviation of the distribution of past
demand is 9 boxes, point A can be located as follows:
Point A Mean + Standard deviation = 30 +0.1 x 9=30.1 = 31 boxes

Hence. the fruit seiler should stock 31 boxes.

Example 11.17 A stall at a certain railway station sells for Rs. 1.50 paise a copy of daily newspaper for
which it pays Rs. 1.20. Unsold papers are returned for a refund ofRe. 0.95 a copy. Daily sales and corresponding
probabilities are as follows
Daily sales 500 600 700
0.5 0.3 02
Probability
should it order each day to get maximum expected profit?
(a) How many copies
are useless, what should be the optimal order each day? Use
(b) If unsold copies cannot be returned and
increment analysis.
0.30
Solution Given that, Incrementalprofit (IP) =Rs. (1.50- 1.20)=Re.
Incremental loss (IL)
Rs. (1.20-0.95)=Re. 0.25
=

The probability (p) of selling at

least one additional copy ofthe newspaperto justify keeping that additional
copy of new spaper 1S
IL 0.25
= 0.45.
PIP +IL 0.30 +0.25
must be at least 0.45 cumulative probability of selling
additional copy, there
nus to the ordering of an
justify below:
are compuled
at
copy. Cumulative probabilities 500 7000
Daily sales
0.50 0.30 0.20
Probability
Comulative
1.0X0 .50 0.2
probability
600 copies.
ience, the optimal order size is then
non-refundable,
O unsold copies
are

I 1.20
200 80
1.50
IPL 0.30 +1.20
500 copies.
Tence, optimal order size is. in a bakery is as follows:
cakes made
of the
The demand
pattern 3 4
d m p l e 11.18 2
0.30 0.20 0.10
demanded
0.10 0.25
No. of cakes 0.05 cakes should
Probability price is Ks 40 per unit, how many
unit and selling
preparation
is Rs 30 per
cost
the t h e bake
baker for maximizing his pro
360 Operations Research: Theory and Applications
Oution Given that incremental cost (IC) to prepare a cake is Rs 50 per unit and ihcrCiment. pi

cake is Rs 40 per unit. The cumulative probability (p) of selling

at least an additional unit
t
Sell a

0 Justify the stocking of that additional unit of cake is given by

IC 30 = 0.428
PIC+ IP 30+40

ne cumulative probabilities of greater than type are computed as shown in lable 11.28.

Cumulative Probability
Denand Probability
PDemand k) PDemand 2 k)
(No. of Cakes)
0.05 1.00
10 0.95
0.25 0.85
0.30 0.60
4 0.20 0.30
Table 11.28
5 0.10 0.10

Since P(demand 2 k) that exceeds the critical ratio.p = 0.428 is k=3 units of cake, the optimal decision
is to prepare
only 3 cakes.

11.6 POSTERIOR PROBABILITIES AND BAYESIAN ANALYSIS

An initial
probability statement to evaluate expected payoff is called a prior probability distribution, but
if the
probability statement has been revised due to additional information, then such probability a
Posterior statement is called
probabilities are
posterior probability distribution.
a

In this section we will discuss the method of

the revised computing posterior probabilities, given prior probabilities
probabilities of thee using Bayes' theorem. The analysis of problems using
states of nature and additional information, is called posterior probabilities with new expected payoffs
obtained after prior-posterior analysis.
conducting a test to Baye's Theorem Statement
improve the prior
probabilities of Let A, A2 A,, be mutually exclusive and collectively exhaustive
.,
Their probabilities P(A,).
respective nature. P(A),.., P(A,) are known. There is an experimental outcome B foroutcomes. which the conditional probabilities
P(B A), P(B | A,), P(B | A,) are also known. Given the
..,
information that outeome B has occurred, the
.

revised conditional probabilities

of outcomes 4, i.e. P(A; | B), i 1, 2,. are
=

determined by using the

following relationship: .., n

P(4; and
P(4,|B)= B)P(4, OB) P(4,) P(B|4,)
P(B) P(B) P(B)
where PB) PA)) P(B | 4)
i=l
Since each joint probability can be
expressed the
probability, i.e., P(A, B ) = P(4,)x P(B |A,)
as
product of a known marginal (prior) and conditional

Fxample 11.19 A company

1s considering to introduce a new
has
the
defined two levels of sales as
"high and low on which it wantsproduct
changes that each market level will oceur,
to its existing
to base its
product rangc.
Baye's decision logether with their costs decision and has estimatcu
rule uses the prior This information is summarized below: and consequent protits or
probabilities to
lossed
determine the Staes of Nalure Probabiliy
expected payoff for COUrses of Acticn
each decision Market the Produc
alternatives and Do not Market the
(Rs 000)
then choOses the Product (Rs O00)
one with the largest High sales 0.3
I50
expected payof. Low sales O.7
40

"conpany's
marketnE 1anag SugEess a t a
The
ne
provide further inlormation on which the company should market research survey
su may be undertaken o
base its decision.
Based on the company P
 Decision Theory and Decision Trees
nce with a certain market 361
research
information in the
light of
subsequent
organization. the marketing manager assesses its abilly ve
actual sales
achievements. This infornmation is
Market Research given Deo
(Sturver outcome) Actuai Saies
Market high Market lo"
High' sales forecast
0.5
Indecisive survey report 0.1
0.3
Low sales forecast 0.4
0.2 0.1
The market research survey costs
Rs 20,000, state
market research organization. whether or not there is case for
a
employing the

Solution The Delhi Univ, MBA, 2000]

expected monetary value (EMV)
for each course of action is shown in Table l1.29.
States of Prior
Courses of Action
Nature Probabiliy Expected Profit (000 Rs)
Market Do notMarket
Market Do not Market
High sales (N,) 0.3 150 0
Low sales (N) 45
0.7 40
28 0
EMV = 17 =
0 Table 11.29
With no additional information. the
company should choose course of action 'market the product.
However. if the company had the perfect information about
the 'low sales', then company would not go
ahead with the decision because
expected value would be (-) Rs 28,000. Thus, the value of perfect
information is the expected value of low sales.
Let outcomes of the research
survey be: high sales (S), indecisive report and low sales and (S,)
states of nature be: high market (N,) and low market (N,). The calculations for prior probabilities of (S,)
forecast
are shown in Table 11.30.

Outcome Sales Prediction

High Market (N) Low Market (N)
High sales (S,) P(S, |N) =
0.5
PS,IN)=0.1
Indecisive report (S,) P(S, IN) =
0.3 P(S1N) 04
Low sales (S3)
PS3IN) = 0.2
P(S 1N)= 0.5 Table 11.30

With additional information, the company can now revise the prior probabilities of outeomes to get posterior
probabilities. The calculations of the revised probabilities, given the sales forecast are shown in Table 11.31.

States of Nature Prior Conditional Joint Probabilit

Probability Probabilit P(S, nN) PN) P(S, I N)
P(N) PS N)
0.15
High sales (N) 0.3 PMS, | N)= 0.5
0.09
P(S, | N)-0.3
P(S,IN ) - 0.2 0.06

0.07
Low sales (N) 0.7 P(S, | N,)0.
0.28
PS, | N,)=0.4
0.35
Table 11.31
0.5
P(S, | Na)
=

Revised
0.22 0.37 0.41 Probabilities
Marginal Probability
362 Operations Research: Theory and
Applications
are.
given the sales forecast,
Posterior probabilities of actual sales,

States of Nature
Posterior Prohahili
Outcome Probabili
PN S)=P(N, S) P5)
(S) P(S) (N)
0.15 0.22 0.681
S 0.22 0.07/0.22 = 0.318

0.09/0.37 = 0.243
S 0.37 N 0.28 0.37 0.756
N
0.06/0.41l = 0.146
S 0.41 N
0.35/0.41 = 0.853
N,
Given the additional information, the revised probabilities to calculate net expected value with respect
to each outcome are shown in Table 11.32.

Sales Forecast
States of Nature Revised High Indecisive Low
Conditiona
Profit (Rs) Prob. EV (Rs) Prob. EV (Rs) Prob. EV (Rs)

High sales 130 0.681 88.53 0.243 31.59 0.146 18.98

Low sales 60 0.756-45.36 0.853 -51.18
0.318 -19.08

Expected value of sales forecast 69.45 - 13.77

32.20
Probability of occurrence 0.22 0.37 0.41
Net expected value
Table 11.32 (Expected value x Probability)
15.279 -5.095 13.202

CONCEPTUAL QUESTIONS B

1. Given the complete set of outcomes in a certain situation, how 4. Briefly explain 'expected value of perfect information' with
is the EMV determined for a
specific course of action? Explain examples.
in your own words.
5. Describe business situation where a decision-maker
2. Explain the difference between expected opportunity loss and faces a
decision under uncertainty and where a
expected value of perfect information. decision based on
maximizing the expected monetary value cannot be made. How
Indicate the difference between decision-making under risk, and do you think the decision-maker
should make the required
uncertainty, in statistical decision theory. decision?

SELF PRACTICE PROBLEMS B

1. You are given the following payoffs of three acts A, Ag and Ag

and the events E. E2 Eg Product Nature of Demand
Good Moderate Po or

States of Three Acts 0.70 0.20 .10

Nature A A 0.50 0.30 20
40 0.50
25 0 125 0.10
E
400 440 400 The estimated profit or loss in rupees under the
may be taken as
three states
650 740 750

Product Good
The probabilities of the states of nature are 0.1, 0.7 and 0.2,
ioderate Poor
respectively. Calculate and tabulate the EMV and conclude 30,000 20,000 10,000
which would prove to be the best course of action. 60,000 30,000 20,000
the problem of
2. The management of a company is faced with Z 40.000 10,000
choosing one of three products that it wants to
manutacture. -15,000
The potential demand for each product may turn out to be good,
Prepare the expected value table, and advise the
moderate or poor. The probabilities for each
of the states of about the choice of
product. management
estimated as follows.
nature were
 nalketing staff Decision Theory and Decision Trees
ed
ot a
certain 363
following payott tabieindustrial organization has
the
oes), concerning a cetain (giving profits in million The overhead expenses attributable to
pastry production are Rs
of technology advance. proposal depending 1.25 per dozen. Fresh
pastries are sold in special boxes which
upon the rate cost 50 paise each and the stale
pastries
are sold wrapped in
ordinary paper. The probability distribution of demand per week
Technological Decision IS as under
Advance
Accept Demand (in dozen) 1 2 3 4 5
Much 2
Reject Probability C.01 0.14 0.2 0.5 0.1 0.05
Little 3
Find the optimal
production level of pastries per week.
9. The local football club wants
None
(-1 your advice on the number of
4 programmes that should be printed for each game. The cost of
printing and production of programmes for each game, as
The probabilities are 0.2, 0.5 and
0.3 for Much, Little and None quoted by the local printer, is Rs 1,000 plus 4 paise per copy.
technological advance respectively. What decision Advertising revenue which has been agreed for the season is
taken? should be
Rs 800 for each
4, A physician purchases particular vaccine on
a game.
Monday each Programmes are sold for 15 paise each. A review of sales
week. The vaccine must be used within
the during the previous seasons indicates that the following pattern
otherwise it becomes worthless. The vaccine following week, is expected to be
dose and the physician costs Rs 2 per repeated during the coming season of 50
charges Rs 4 per dose. In the games:
weeks. the physician has administered past 50
the vaccine in the
following quantities: Number of Programmes Sold Number of Games
Doses per week 20 25 40 60
Number of weeks 5 10,000 5
15 25 5
Determine how many doses the
physician should
20,000 20
week. buy every 30,000 15
A grocery with a 40,000
bakery department is faced with the problem 10
of deciding how many cakes it
should buy in order to meet the Programmes not sold at the game are sold as waste paper to
day's demand. The grocer prefers not to sell day-old
goods in a paper manufacturer at one paise per copy.
competition with fresh products; leftover cakes are, therefore, a Assuming that the four options listed are the only possibilities
complete loss. On the other hand, if a customer desires a cake 0) prepare a payoff table
and all of them have been sold, the
disappointed customer will (i) determine the number of programmes that would provide
buy from elsewhere and the sales will be lost. he grocer has,
largest profit, if a constant number of programmes were
therefore, collected information on the past sales on a selected to be printed for each
game
100-day period as shown in table below: (i) calculate the profit that would arise from a perfect forecast
of the number of programmes which would be
sold at each
Sales per Day No. of Days game.
Probability 10. The probability distribution of
monthly sales of an item is aas
25 10 0.10 follows:
26 30 0.30
Monthly sales (units) Probabilities
27 50 0.50
28 10 0.1010 0.01
0.06
Construct the payoff table and the opportunity loss table. What 0.25
is the optimal number of cakes that should be bought each day? 0.30
Also find and interpret EVPI (Expected Value of Perfect
0.22
Information). One cake costs Re 0.80 and sells for Re 1
6. A producer of boats has estimated the following distribution of 0.10
demand for a particular kind of boat: 0.06
The cost of carrying inventory
No. demanded 1 2 3 4 6 (unsold during the month) is Rs
30, per unit, per month, and the cost of unit
Probability 0.14 0.27 0.27 0.18 0.09 0.04 0.01 shortage is Rs 70.
Determine the optimum stock to minimize expected cost.
Each boat costs him Rs 7,000 and he sells them for Rs 10,000
11. A modern home appliances dealer inds that the cost of
Boats left unsold at the end of the season
must be holding
each. in
a mini cooking range in stock for a month is Rs
200 (insurance,
each. How many boats should be
isposed of for Rs 6,000 minor deterioration, interest on borrowed capital, etc.). Customers
so as to maximize his expected profit?
STOck
A Small industry from the past data has found that the cost of
who cannot obtain a cooking range immediately tend to
go to
other dealers and he estimates that for every customer who
is Rs 30 if it
naking an item is Rs 25, the item's selling priceof at Hs 20 per
cannot get immediate delivery, he loses an
average of Rs 500.
a week. It can also be disposed The probabilities of a demand of 0, 1, 2, 3, 4, 5 mini
sSold within
cooking
tem at the end of the
week 8
ranges in a month are 0.05, 0.10, 0.20, 0.30, 0.20, and 0.15,

Weekly sales 3 4 5 7 respectively. Determine the optimal stock level of

cooking
10 20 40 30 0 ranges. Also find the EVPI. [Delhi Univ., MBA, 2000
No. of weeks 0
the industry should 12. A company manufacturing large electrical
ind the optimum number of items per week equipment is
anticipating the possibility of a total or a partial copper strike in
produce. the near future. It is attempting to decide whether to
which it sells at Rs
8 per dozen in special
a large amount of copper at an additional
stockpile
m a k e s pastries each. The direct
cost of pastries cost of Rs 50,000;
containing dozen
one stockpile a small amount costing an additional Rs 20,000; or to
s dozen. At the end of the week
the
stockpile no additional copper at all. The stockpiling costs,
Or the firm is Rs 4.50 per lower price of Rs 3.50 per dozen.
p a s t r i e s are sold off for a
364 Operations Research:. Theory and Applications
f

consisting of excess will charge Rs 350

storage, holding and handling costs and so The store's advetising agents
forth. the advertisement and eac
over and above the
are for
actual material costs. artwork and blockmaking
If there is a partial Rs 50 for the insertion of the
strike, the company estimates that an newspaper will charge
additional cost of Rs 50,000 for advertisement. You are required
to:
if there is no delayed orders will be incurred to the iearest 10 should te
stockpile at al. If a total strike occurs, the cost ot (a) Calcuiate how many garments
delayed orders is estimated at Rs 1,00,000 if there is purchased by the store in
order to maximize the expected
small stockpile, and Rs only a in: (i) one newspaper, (i) two
2,00,000 if there is no stockpile. The gross profit, after advertising
company estimates the probability of a total strike
as 0.1 and newspapers.
that of a partial strike as 0.3. net profit after
(b) Calculate the amount, if any, of the expected
(a) Develop newspaper, (i) two newspapers.
a conditional cost
table showing the cost of all advertising: (i) in one
in each car
outcomes and course of action 14. A car manufacturer uses a special control device
combinations.
(b) Determine the preferred course of action and its that he produces. Two alternative methods can be used to
cost. What
is EPPI? detect ard avoid a faulty device. Under the first method, each

(c) Develop a conditional device is tested before it is installed. The cost of this method
opportunity loss table.
(d) Without calculating the EOL, find tho EVPI. is Rs 2 per test. Alternatively, the control device can be installed
13. A well-known without being tested, and a faulty device can be detected and
departmental store advertises female fashion rendered after the car has been assembled, at a cost of Rs 20
garments from time to time in the Sunday press. Only one
per faulty device.
garment is advertised on each occasion. The
experience in management's Regardless of which method is used, faulty devices cannot
garments in the price
range Rs 30-40 leads the be repaired and must be discarded.
management to assess the statistical probability of demand at
various levels. after each A manufacturer purchases the control devices in batches of
advertisement, as follows: 10,000. Based on past experience, he estimates the proportion
of defective components and the associated probability to be:
Demand After Advertising After Advertising
in One
(No. of gaments)
in Two Proportion of Probability
Newspaper Newspapers Faulty Devices

30 .10 0.00 0.08 0.20

40 0.25 0.15 0.12 0.70
50 0.40 0.35
0.16 0..10
60 0.25 0.40
70 0.00 0.10 (a) Which inspection method should the manufacturer adopt?
(b) What is the expected value of perfect information (EVPI)?
The next garment to be advertised at Rs 35 will cost Rs 15 to
make and can be disposed of to the trade, if unsold, for Rs 10.

HINTS AND ANSWERS

1. EMVA) = 412.5, EMVA) = 455, EMVA,) = 417.5

where D is the number of boats sold and S is the number of
EMVX) = 26. EMV(Y) = 43, EMV(Z) = 19.5;
boats produced
Company should manufacture product Y. EMV = 4,080, stock 3 boats.
3. EMV (accept) = 2.6, EMV(reject) = 2.8; reject.
7. Conditional profit value = MP item sold - ML * item unsold
4. Conditional profit value = MP x units sold - ML x units unsold
=
(30 25) item sold - (25 20) item unsold
(4-2)D=2D ;D2S
= (30-25) D D2S
(4-2)D-2(S-D)=4D-2S D<S
(30-25) D-(25-20) (S- D) =10D-5S D<S
number
where D is the number of units demanded and S is the
where, D is the number of item sold: S is the number of item
of units stocked
produced.
EMV* EMV (Purchase 40 dozen) 54
Rs
EMV =
Rs 26,
5. Conditional profit value =
MP cake sold - MIL cake not sold produce 6 items.
8. Conditional profit value
-(1 -0.80) cake sold 0.80 cake not sold
MP boxes sold - ML x boxes unsold
Des
0.20D (8 4.50 1.25 0.50) boxes sold
0.20D-0.80(S-D); D<5S (4.50 + 1.125 5.50) boxes unsold
number of units demanded and S is the number
where D is the
of units stocked
L.75 D :D2S
EMV= Rs 5 OL = 0.22 (stock 26 units of cake) 75 D0.25 (S- D)=2 D-0.25 S D<S
6. Conditional profit value MV Rs 4.28.
produce 4 dozen
pastries.
F
MP boats sold ML boats unsold 9 (a) Conditional profitSales revenue Cost
(10,000-7,000) boats sold (7,000 6,000) boats unsold 0.15/) + 800 + (P
Des
D) 0.101 (1,000
x
+ 0.04P)
300 D 0.14 )-0.03 P-20 D< P
4,000/)- 1,000S
3,000 D- 1,000 (S- D)
=
0.11D-200 :D2P
 Decision Theory and Decision Trees
where P is the number of 365
demand for programmes. programnmes printed and D is the 13. Marginal profit (MP) Rs (35 =
15) R s 20
(b) largest expected protit =
Rs 2,260 Marginal loss (ML) Rs (15 - 10) Rs 5
accrues for P
30.000
copies per game Conditional payoff MP garments sold ML
(c)In case of perfect forecast, the
900x 0.1 +2.000 0.4 + 3,100
expected profit is: garments unsold
(a) EMV
EVPI
0.3+4.200
x
0.2 x
Rs 2.660. (S,)= Rs 600;
EMV (S,) =
Rs 775;
Rs
2,660 Rs 2.260 Rs =

400 EMV (S) = Rs 887.5;

10. Cost function 70(D-S) D2S EMV (S) =

Rs 900; EMV (S) =
Rs 850
200(S- D): D<S Max. EMV is
corresponding to course of action
S4. 1.e.
where D =

monthly demand purchase 60 garments.

(or sales); S =number of units
purchased (b) EMV (S,) Rs 800; EMV 962.5,
=

(S,) =
Rs
Since the expected cost of Rs 46 is EMV (S) = Rs 1,037.5;
minimum for course of
action 4. the optimum stock to minimize the EMV (S5 ) = Rs 1,012.5.
cost is 4 units
month. per
Max. EMV is corresponding to course of action
ie.
11. Cost function 500(D- S); D2S purchase 60 garments.
200 (S- D) ; D< S 14. (a) First alternative, Rs 20,000
Since the expected of Rs 315 is minimum for course
cost (b) EVPI = Expected cost under uncertainty

optimum stock to minimize the cost is 4 cooking of action

the 4,
ranges. Expected cost with perfect information
12 (b) small stockpile Rs
30,000 Rs 11,000 (d) Rs 19,000 20,000-(16,000 x0.2 +20,000x0.7+20,000x 0.1)
=
20,000 -

19,200 =
Rs 800 per batch.

11.7 DECISION TREE ANALYSIS

Decision-making problems discussed earlier were limited to arrive at a decision over a fixed period of time.
That is. payoffs, states of nature, courses of action and
states of nature
probabilities associated with the occurrence of
were not subject to change.
However, situations may arise when a decision-maker needs to revise his previous decisions due to
availability of additional information. Thus he intends to make a sequence of interrelated decisions over
several future periods. Such a situation is called a sequential or
multiperiod decision
process. For
example,
in the process of marketing a new product, a company usually first go for Test Marketing' and other
altemative courses of action might be either 'Intensive Testing' or Gradual Testing'. Given the various
possible consequences-good, fair, or poor, the company may be required to decide between redesigning
the product, an aggressive advertising campaign or complete withdrawal of product, etc. Based on this
decision there might be an outcome that leads to another decision and so on.
A decision tree analysis involves the construction of a diagram that shows, at a glance, when decisions Decision tree is
the graphical
are expected to be made in what sequence, their possible outcomes, and the corresponding payoffs.
display of the
A decision tree consists of nodes, branches, probability estimates, and payoffs. There are two types progression of
of nodes: decision and
random events.
Decision (or act) node: A decision node is represented by a square and represents a point of time
where a decision-maker must select one alternative course of action among the available. The
courses of action are shown as branches or arcs emerging out of decision node.
Chance (or event) node: Each course of action may result in a chance node. The chance node
is represented by a circle and indicates a point of time where the decision-maker will discover the
response to his decision.
Branches emerge from and connect various nodes and represent either decisions or states of nature. There

are two types of branches:

Decision branch: It is the branch leading away from a decision node and represents a ecourse of

action that can be chosen at a decision point.

Chance branch: It is the branch leading away from a chance node and represents the state of nature
of a set of chance events. The assumed probaililies of the states of nature are written alongside
their respective chance branch.
366 Operations Research: Theory and Applications
followed by cither
decision tree (not
branch that makes the end of the
erminal Ar.y
branch: can represent
either course
branch. A terminal branch
a

decision or chance node). is called a terminal s o that

tree are supposed to

be nmutually exclusive points
ot action. The terminal points of a decision

Cxactly action will be chosen.

one course of
or cost) and it can be
or negative (i.e. expenditure
p a y O / can be positive (i.e. revenue or sales)
associated either with decision or chanee branches.
tree to be
An illustration of a decision tree is shown in Fig. 11.2.
It is possible for a decision

deterministie or probabilistic. It can also further be divided in terms

of stages into single stage (a decision
-

under condition of certainty) and multistage (a sequence of decisions).

R

Chance branch
R

P (State of nature)
Chance node

Q (State of nature)
A3
Decision branch

A (Course of
action) R

Decision node

A3
A2 (Course of
action)

Q A3

AA R

Fig. 11.2
Decision Tree
Stage. Stuge Stage Stage 4

The optimal sequence of decisions n a tree 1s toumd by starting at the right-hand side and rolling
hackwards, At each node, an expected return is Calculaled (called position value). If the node is a chance
node, then the position value is calculated as the sum of the products of the probabilities or the branches
emanating from the chance node and neir respecUve positin values. It the node is a decision node, then

he evnected return calculated

is for each ol ns driänches and the
highest return is selected. This procedure
continues until the initial node is reached. The posilion valucs for this node corresponds to the maximum
the decision sequence.
expected return obtainable Irom
 Decision Theory and Decision Trees
Remark Decsto1 e e s VeIsus 367
trees. However, there probability trees Decision
are several basic differences:
trees are basically an extension of
probability
) The decision tree
utilizes the
at the
right-hand terminus withconcept
the
of 'rollback" to
solve a problem. This means that it starts
or
beginning decision point in orderhighest expected value of the tree and works back to the
current
multiplicity of decision points determine
It is the to
the decision or decisions that
should be made.
(i) The probability tree is that make the rollback
primarily concerned process necessary.
with
factors as a means of calculating
tree utilizes the probabilities, whereas the
probability decision
(ii) The most important feature of arriving
the decision tree, is that it
at a final
answer
into account. At takes time differences of future
any stage of the earnings
immediate cost or revenue decision tree, it may be to weigh differences in
against differences in value at the necessary
next stage.
Example 11.20 You are given the
following estimates concerning Research and Development
a
Decision Probability of Decisio programme
D Outcome
D Given Research R Nunmber Probabilioy of Pavof Vale
PDI R) Outcome Given D of Outcome
Develop PeD (Rs 000)
0.5
0.6 600
0.3 100
0.1 0
Do not develop 0.5
0.0 600
0.0 100
1.0 0
Construct and evaluate the decision tree diagram for the above data. Show
your workings for evaluation.
Solution The decision tree
of the given problem along with necessary calculations is shown in Fig. 11.3.

Probability Payoff Expected

(in 000 Rs) Payoff
(in 000 Rs)
Px| D) = 0.6
{3) 0.5 x 0.6 0.3 600 0.3 x 600 = 180

0.3
D. Develop (0.5) Px2| D)
=

0.5x 0.3 =0.15 -100 0.15 x - 100=-15

0.5 x 0.1 =0.05

P(x3| D1) = 0.1
0

165
P(x| D2) = 0

(0)0.5x0 600

0
Px2| D2)
=

0.5 x0 -100 0
D. Do not develop
(05)
Px3 D2) = I.0
0.5x 1.0 Fig. 11.3
Decision Tree

Example 11.21 A glass factory that specializes in crystal is developing a substantial backlog and for
this the firm's management is considering three courses of action: To arrange for subcontracting (S,), to
begin overtime production (S,), and to construct new facilities (S,). The correct choice depends largely
upon the future demand, which may be low, nediun, or high. By consensus, managementranks the
respective probabilities as 0.10, 0.50 and 0.40. A cost analysis reveals the efect upon the profits. This is
shown in the table below:
368 Operations Research: Theory and Applications
Course of Action
Denand Probability
S
(Begin Overtime) (Construct Facilities
(Subcontracting)
20 150
Low (L) 0.10 10
60 20
Medium (M) 0.50 50
00 200
High (H) 0.40 50

Show this decision situation in the form of a decision tree and indicate the most preferTed decision and

its corresponding expected value.

Solution A decision tree that represents possible courses of action and states of nature is shown in Fig.
.4. In order to analyze the tree, we start working backwards from the end branches.
The most preferred decision at the decision node 0 is found by calculating the expected value of each
decision branch and sclecting the path (course of action) that has the highest value.

Expected Pavoff
0.10x 10 =01 (in '000 Rs)
0.10)

L, ( p

M, (p =0.50)
0.50x 50 = 25

1Subcontracting EMV 46
H.p=0.40) 0.40 x 50
20
46

L.(p = 0.10)
0.10x-20 = - 02

S2: Begin overtime M, (p = 0.50)

0.50 x 60 30
EMV 68
H.(p =0.40)

S3: Construct facilities ) 0.40 x 100= 40

L.(p - 0.10)
0.10 x -150 = -15

M. (p=0.50)
0.50 x 20= 10

Fig. 11.4
Decision Tree
EMV 75

H.p-0.40) ) 0.40 x 200 80

Since node 3 has the highest EMV, theretore, the decision at node O will be to choose
action S i.e. construct new facilities. the course of

Examole 11.22 A businessman has Iwo

independent investment portfolios A and B, available him.
hut he lacks the capital undertake both of them
to
simullaneously. He can either choose A first and then
to
stop. or if A is not successful, then lake, B or vice
versa. The
Rit is 0.4. Both investment schemes require an initial probability of success of A is 0.6. while for
capital outlay of Rs 10,000 and
if the venture proves to be unsuccesstu. Successtul both return nothing
complelion of A will return Rs
csfil completion of B will return Rs 24,000 (over cost). Draw decision 20,000 (over cost) and
a
tree in order to
best strategy. determine the
Delhi Univ. MBA, 2000,
AMIE, 2006)
 Decision Theory and Decision Trees 369
Solution The decision tree based on the given information is
shown in Fig. 11.5. The evaluation of each
chance node and decision is given in Table 11.33.

Decision Point Outcome Probabilit Conditional Vaiue Expected Value

(Rs)
D3 () Accept A Success 0.6 20,000 12,000
Failure 0.4 -10.000 -4,000
8,000
(i1) Stop
0
D () Accept B Success 0.4 24,000 9,600
Failure 0.6 -10,000 6,000

3,600
(ii) Stop
D ) Accept 4 Success 0.6 20.000+ 3,600 23,600 14,1160
Failure 0.4 10,000 -4,000

10,160
(i1) Accept B Success 0.4 24,000 8,000 32,000 12,800 Table 11.33
Failure 0.6 -10,000 -6,000 Evaluation of
6,800 Decision and
(ii) Do nothing 0 Chance Nodes

Stop
Rs03 Expected Payoff
Rs 3,600

SuccesS D2
(0.6)
Rs20,000 Success (0.4)
EMV 3,600 0.4 x 24,000
EMV = 10.160
Accept B Rs 24,000 = 9,600

AcceptA
Failure (0.4) Rs 10,000
- Rs 10,000
Failure (0.6)
0.6x -10,000

6,00
Do nothing
(s) 0.4 x 10,000 4000
3,600
D1 Rs
Failure (0.6)
0.6x 10,000= - 6000
Rs 10,0000

Accept B Success (0.6)

Rs 20,0000
0.6 x 20,000
- 12,000

A c c e p tA
EMV 6.800
Rs 24,000 Rs 10,000

(0Suc.4) ces
EMV=8,000 0.4x 10,000
Failure (04)
4,000
D3 8,000

Rs 8,000
Re

Stop
() Fig. 11.5
Decision Tree

Since the EMV = Rs 10,160 at node D, is highest, therefore the best strategy is to accept course of
action A first and if A is successful, then accept B.
370 Operations Research: Theory and Applications
a m p l e 11.23 The Oil India Corporation (OIC) is wondering whether to go for an ofishore oil drilln

would 65 ner
be Ks bUU million with
High. If OIC bid, value
a
that is to be awarded in Bombay
Onuract drilling operation the alreadu
or move
a new
Cnance of gaining the contract. The OIC may set up
new site. The probability of success and
operation, which has already proved successful for
a

expected returns are as follows:

Outcome New Drilling Operatio Existirg Operation

Probability Expected Revenue Probabilin. ExpectedRevemue
(Rs million) (Rs million)
Success 0.75 800 0.85 700
Failure 0.25 200 0.15 350

It the Coporation do not bid or lose the contract, they can use Rs 600 million to modernize their

operation This would result in a return of either 5 per cent or 8 per cent on the sum invested with
probabilities 0,45 and 0.55. (Assume that all costs and revenue have been discounted to present value.)
(a) Construct a decision tree for the problem showing clearly the courses of action.
( By applying an appropriate decision criterion recommend whether or not the Oil India Corporation
should bid the contract. [Delhi Univ MBA, AMIE, 2001, 2005]
Solution The decision tree based the given information is shown in
on
Fig. 11.6. The evaluation of
each
chance node and decision node is
given in Table 11.34.

Expected Pavoff
(in Rs million)
5% return (0.45)
6 600 x 0.05 x 0.45 = 13.5

Modermize
Mode
EMV 39.9
8% return (0.55)
Success (0.75)
(7) 600x 0.08 x
0.55 =26.4

800 x 0.75 = 600

New drilling

operationn

EMV 650 200 x 0.25 5 0

Success (0.65) Failure (0.25)

Bid
Rs 650 Success (0.85)
10 700 x 0.85 = 595

Existing
operation
EMV=436.46
EMV 647.5 350 x 0.15 = 52.5
Failure (0.15)
Failure (0.35)

5% return (0.45)
Rs 39.9
12 600 x 0.45 x 0.05 = 13.5
Fig. 11.6
Decision Tree
Modernize
EMV = 39,9
13 600 * 0.55 x 0.08 26.4
8% return (0.55)
 Decision Theory and Decision Trees371
Decision Point
Outcome Prohahiliny Condiional Value Expected Value
(Rs)
D (i) Modermize 5% retun 0.45 600x 0.05 =30 30x 0.45 = 13.5
8% retun 0.55 600x 0.08 =488 48 0.55 =26.4

39.9
D (1) Undertake new Success 0.75 800 600
drilling operation Failure 0.25 200 50
650
ii) Move existing Success 0.85 700 595
operation Failure 0.15 350 52.5

647.5

D (i) Modernize 5% return 0.45 600 x 0.05 3 0 30x0.45 13.5

8% return 0.55 600x 0.08 48 48x 0.55 26.4
39.9
(i) Bid Success 0.65 650 422.50 Table 11.34
Failure 0.35 39.9 13.96 Evaluation of
Decision and
436.46 Chance Nodes

Since EMV, Rs 436.46 event node 2 is highest, therefore the best decision at decision node D, is
to decide for bid and if successful establish a new drilling operation.

Example 11.24 A large steel manufacturing company has three options with regard to production: (o produce
commercially (i) build pilot plant (ii) stop producing steel. The management has estimated that their pilot plant,
if built, has 0.8 chance of high yield and 0.2 chance of low yield. If the pilot plant does show a hight yield,
management assigns a probability of 0.75 that the commercial plant will also have a high yield. If the pilot plant
shows a low yield. there is only a 0.1 chance that the commercial plant will show a high yield. Finally, management's
best assessment of the yield on a commercial-size plant without building a pilot plant first has a 0.6 chance of high
yield. A pilot plant will cost Rs. 3,00,000. The profits earned under high and low yield conditions are Rs. 1,20,00,000
and Rs. 12.00,000 respectively. Find the optimum decision for the company. Punjab Tech Univ, BE, 2006
Solution A decision tree representing possible courses ofaction and states of nature are shown in Fig. 11.7.
n order to analyse the tree, we proceed backward from the end branches

High
0.6 Rs. 1,20,00,000

Low 0.4 12,00,000

ProduceCommercially Rs.
Produce

commercially
Hig! (0.75)
P 1,20,00,000
EMV
67,20,000 Low (0.25)
High
yield D2
Stop Rs. 12,00,000
EMV Rs. 0
Build Pilot Plant
High (0.1)
Rs. 3,00.000 Low yicld 87,00,000 Produce

commerCally
Rs. 1,20,00,000
EMV
.2 Low (0.9)
Son69,36,000 D3 Stop Rs. 12,00,000
Fig. 11.7
Rs. 0 Rs. 0
EMV 1,20,000
372 Operations Research: Theory and
Applications
EMV EMV (Node D,) = Rs. 87,00.000
(Node 3) =0.75 1.20,00,000-0.25x 12.00.000
x

= Rs. 87,00,000

EMV(Node 4) 0.1 x 1.20.00.000-0.9x 12,00,000 EMV (Node D,) = 1.20.000.

12,00,000-10,80,000 Rs. 1.20,000.

EMV (Node 1) =
0.8 *
87,00,000-0.2 x 1,20,000 EMV (Node D;) =69,36,000-3,00.00
= Rs. 69,36,000 = Rs. 66,36,000.

EMV (Node2) =0.6x 1,20,00,000-0.4x 12,00,000

= 72.00,000-4,800,000=Rs. 67.20,000

Since at decision node

D, the production cost of Rs 67,20,000, associated with course of action- Build
pilot plant is least, the company should build pilot
plant,.
Example 11.25 A businessman has an option of selling a product in domestic market or in
The available relevant data are export market.
given below.
ltems Export Market Domestic Market
Probability of selling 0.6 1.0
Probability of keeping delivery schedule
0.8 0.9
Penalty of not meeting delivery schedule (Rs.)
50,000 10,000
Selling price (Rs.) 9,00,000
Cost of third party inspection (Rs.) 8,00,000
30,000 Nil
Probability of collection of sale amount
0.9 0.9
If the product is not sold in
export market, it can always be sold in domestic market. There
implications like interest and time. are no other
(a) Draw the decision tree using the data given above.
(b) Should the businessman go for selling the
product in the export market? Justify your answer.
Solution i) The decision tree representing possible curses of action and states of nature is shown in
Fig.11.8.
The revenue generated from the stage ot product in export market is equal to the
inspection cost (i.e. Rs. 30,000). The tree is analyzed by selling price minus
moving backward from the end branches.
EMV (Node 9) Rs. 0.9x 8,00,000
EMV (Node 10) =Rs.7.20,000.
0.9x8,00,000 Rs. 7,20,000.
EMV(Node 7) =Rs. 0.9 8,70,000 x
0.1 x
30,000 EMV
Rs. 7,80,000.
(Node 8) Rs. 7,80,000.
=

EMV (Node 4) 0,9 7,20,000 +0.I x 7,10,000

EMV (Node 3) 0.8 7,80,000 + 0.2
=
x
Rs.7,19,000. 7,30,000 x

Rs. 7,70,000.
EMV(Node 5) 098,00,000 Rs. 72,20,000 EMV
EMV(Node 1) 0.6 7.70,000 1 0.4*7,19,000 (Node 6) 7,20,00.
EMV (Node 2) -0.9x 7.20,000+0.1x
Rs.7,49,600 7,10,000
EMV (Node Rs. 7.19,000.
D) Max. {7,49,600,
7,19,000 Rs. 7,49,600.
Hence, the businessman should go lor seling the product in the
export market.
 Decision Theory and Decision Trees
373

Aount Collected
0.9
Rs. 8.70,000
Rs.0
No
D on o t

Export
Collectedd Rs. 30,000
0.1

Export o (8Amount Collected Rs.8,70,000

0.9
EMV
7.49.600 Rs. 50,000 Not
EMV Collected Rs. 30,000
7,20,000 0.1
Export
Amount Collected
0.9
Rs. 8,00,000
Donot Rs. 0
Export
0.9
EMV= Not Collected - Rs. 0
7,20,000 0.1

DEMV=7,49.600 EMV =

7,19,000 Export nount Collected

0.9
Rs. 8,00,000
Rs. 10,000O
Do not EMV =

7,20,000
Collected Rs. 0
Export .1
Amount Collected
0.9
Rs. 8,00,000
Donot Rs.0 SKNor
Export EMV=
0.9 7,20,000 Collectea
0.1
-Rs. 0
EMV
7,19,000 0. Expo rt Amoun Collected
0.9
Rs. 8,00,000
Fig. 11.8
- Rs. 10,000 .
EMV Not Collectedd Rs. 0
7,20,000 0.1

11.8 DECISION-MAKING WITH UTILITIES

In previous sections, we discussed solution to problems based on the criterion of expected profit (or loss)
expressed in monetary terms. However, in many silualions, such criterion that involves expected monetary
payoff may not be appropriate. This is because ol the lact that different individuals attach ditferent utility
to money, under different conditions. The term utility iS the measure of preference for various alternatives
in terms of relative value for money. The utlity ol a given alternative is unique to the individual decision-
makers and unlike a simple moneary amount, can incorporate intangible factors or subjective standards

from their own value systes.

llustration Mr Ram has won Rs 1,000 in a quiz programme. In the last round he is asked either to
altemalives: (a) quil and take his winmings, (b) take a last chance in which
compete or quit. Now he has two
he has 50-50 chance of winning Rs 4,000 or nothing. The question now is: What would he do' On EMV

basis he has
EMV (b) 0.50 (4,000) + 0.50 (0) Rs 2,000
374 Operations Research: Theory and Applications
This amount is twice what he has already won. But would he really give up Rs 1,000 for a 50-50 chance
Utility provides a
way of incorporating
or Rs 4,000 or nothing? Many individuals would not because they would think of all the alternatives they
the decision maker's could do with Rs 1,000 and how they would regret it if they end up with nothing. Hence a new payoff
attitude towards risk
in arriving at a measure utility the decision-maker's attitude and preference has to be introduced.
reflecting
decision and hence The basic axioms of utility may be stated as follows:
measures the true ) I f outcome A is preferred to outcome B, then the utility UA) of outcome A is greater than the utility
value of an UB) of outcome B, and vice versa. If both are equally preferred, then U(A) = U(B)
outcome.
() If the decision-maker does not see difference between the two alternatives and outcome A occured
with probability p. and outcome C with probability (1 - p), then U(B) = pUA) + (1 - P) U(C)

Under utility criterion, a rational decision-maker will choose an alternative that optimizes the expected
utility rather than expected monetary value. Once individual's utility function is known, along with the
probability assigned to outcomes, then the total expected utility for each course of action can be obtained
by multiplying the utility values with their probabilities. The strategy that corresponds to the optimum utility
function is called the equal strategy.

11.8.1 Utility Functions

criterion such as
Utiity function describes the relative preference value that individuals have for a given
a given amount, for
money, goods, etc. Preferences are often determined by choosing between receiving
a certain activity versus a 50-50 chance of gaining a larger amount or nothing. The gamble amount is
adjusted upward or downward until the individual is indifferent to whether he receives the certain amount

or the gamble. This indifference point establishes individual's utility.

A utility function can be used to convert a decision criteria value into utils so that a decision can be
made on the basis of maximizing the expected utility value (EUV) rather than, say the EMV.

11.8.2 Utility Curve

A utility curve that relates utility values to rupee values is obtained by placing the decision-maker in various
Utility function is hypothetical decision situations and plotting the decision-maker's pattern of choices in terms of risk and
a graphically display utilities. Fig. 11.9 shows several utility curves and the risk preference associated with each.
of utility values
amount of money
Suppose, the relationship between monctary gains, losses and utilities for gains is established. As shown
g0ods, etc. in Fig. 11.9, if the utility curve is bent down non-linearly,then a large negative utility is assigned for losses.
received It is important not to make the curve bend down too steeply or to start the bending too quickly because this
may lead individuals into a situation where they attach such a heavy weightage to the possibility of loss that
they never take any risk and thus, never make any gains.

max
RiskRiAversi
skIndiof(navojder)
erence(neutrality)
Utility |
(1X0oS)
y
Aiui!
Fig. 11.9
xs

Utility Curve and

Associated Risk
Preferences
Money Rs max

On the positive side of the curve, il eventually bend away from the straight line. This indicates that
increasing units of money are resulting in smaller additional gains in utility.
 Decision Theory and Decision Trees 375

11.8.3 Construction of Utility Curvess

Suppose. there is a project that would payoff with Rs 1000 or (-) Rs 1000. Then what probability of success

would make the deciSion-maker just

about indifferent as to whether to undertake the or
Suppose he says that he would require an 80 per cent probability of success to make him indifferent about project not
undertaking the project. Suppose utility of Rs 1000 is 10 for an individual. Then
U (Re 0) = (0.80) U (Rs 1000) + (0.20) U(-Rs 1000)

0 0.80 (10) + (0.20) U(Rs 10) or U- Rs 10) =

- 40
The utility curve will have three
points:
U+1000) =10: U (0) =0 and U(-1000)=-40
Other points on the curve can be obtained by interacting with decision-maker to know his/her utility values
to a payoff.

Example 11.27 A manager must choose between two investments A and B that are calculated to yield
net profits of Rs 1,200 and Rs 1.600. respectively with probabilities subjectively estimated at 0.75 and 0.60.
Assume. the manager's utility function reveals that utilities for Rs 1,200 and Rs 1,600 are 45 and 50 utils,
respectively. What is the best choice on the basis of the expected utility value (EUV)?
Solution The expected utility value is expressed as:

EUV= 2 Uip;

where U, = utility value of state of nature i

P;probability value of state of nature i

Then EUV (A) =P,U, = 0.75 x 45 33.75 utils and EUV (B) =PpUp =0.60 x 50 30 utils

Since EUV(A) > EUV(B), therefore, the best choice is investment A.

Example 11.28 A businessman is thinking whether to invest in bonds with an assured return or in a new
venture that is likely to fetch him Rs. 20,000 or nothing with equal probabilities. The businessman says he
would prefer an assured return in it is Rs. 10,000 or more, would be indifferentto the two alternatives if the
assured return is Rs. 8,000 and would opt for the risky alternative if this amount is less than Rs. 6,000.
The businessman had purchased a minicomputer last year which is lying with him unused at present. The
minicomputer can fetch him a profit of Rs. 8,000 by way of consultancy. He says that he would not like to sell
the computer but would be indifferent to an offer ofRs. 3,000.
worth a profit of Rs. 8,000. Bidding expenses are
Recently a Govt agency invited bid offers for a contract
reimbursable if the contract materializes and not otherwise. There is an equal chance of winning or losing the
bid. After thinking over the possibilities, the businessman says that he would be indifferent to submitting the
bid at a bidding expense of Rs. 2,000. Construct the businessman's utility curve by using the CME technique.

Solution Let arbitrary utility values of 100 and 0 be assigned to the most favourable and least favourable

outcomes. 1.e.
100 utiles, U=0 utile.
U20.000
0.5 Uu 0t0.5 U0.5 100 + 0.5 x ) 25 utiles.
Given that
The businessman is indifferent between his utiliues 1or Rs. 3,000 and ofa chance of Rs. 8,000 or nothing. Thus,
0.5 U,.000t0.5 U, 0.5 > 50 t0.5 x 0 =25 utiles.

him either a gain of Rs. 8,000 or lose

The decision to enter the bid at an expense ol Rs. 2,000 may get
Since he is indiflerent to submitting the bid,
(bidding expenses of) Rs. 2.000 with equal probabilily.

U0.5UgoO.5 20
0-0.5 50 i 0.5) , 008) O 2000 50 utiles.
376 Operaions Research: Theory and Applications
Following in the summary of points for the
Utiles
businessman's utility curve:

Amount (Rs.) Uility (utiles)

100
,000 100
8f000 50
50
3,000
0 0
,000 Rs.
Fig. 11.10 5,000 5,000 10.000 15.000 20.000
Utility Curve The utility curve is plotted in Fig. 11.10. This curve
may reflect a reasonable estimate of the businessman's 50
behaviour when faced with risky ventures
Example 11.29 Mr XX has an after-tax annual of Rs. 90,000 and is considering to buy accident insurane
Tor his car. The probability of accident during the year is 0.1 (assume that at most oneaccident will oceur). in
which case the damage to the car willbe Rs. 11,600. With a utility function of Ut) = vx , what is the insurance
premium he will be willing to pay?

Solution Let Arepresent the venture when Mr XX does not by the accident insurance for his car. Then. in
case ofaccident he would spend Rs. 11,600 on damages and will be left with Rs. 78,400. In case of no accident
he retains Rs. 90,000. Then we have
U=U78.400 * 0.1 +Ug0,0000.9.
Since UCx) U, = Vx, 78,400 = 280 utiles, and U90.000
U7s400 90,000 300 utiles. Thus,
=

U-280x 0.1 +300x 0.9=298 utiles.

So the amount Rs. which will give the utility of the venture: A
x same =
(298) =
Rs. 88,804. [Because
U)= Vx or X=[U (w)]
Thus Mr XX will be indifferent to an amount of Rs. 88,804 with certainty and the ventureA. The amount he will
be willing
to pay car premiumwould
as be 90,000-88,804 Rs. i,196.

SELF PRACTICE PROBLEMS C

1. Raman Industries Ltd. has a new product that they expect has experience, test marketing results have been favourable about
great potential. At the moment they have two courses of action 70 per cent of the time.
Furthermore, products favourably tested
open to them: To test market (S,) and to drop product (S) have been successful 80
per cent of the time. However, when
If they test it, it will cost them Rs 50,000. The response could the test marketing result has been
be either positive or negative, with probabilities 0.70 and 0.30, has only been successful 30 unfavourable, the product
per cent of the time. What course
respectively. If it is positive, they could either market it with full of action should the
effort or drop the product. If they market with full scale, then the
company pursue?
3. XYZ
result might be low, medium, or high demand and the respective
Company manufactures guaranteed tennis balls. At
approximately 10 per cent of the present,
net payoffs would be- Rs 1,00,000, Rs 1,00,000 or Rs tennis balls are defective. A
defective ball leaving the
5.00.000. These outcomes have probabilities of O.25, 0.55 and
honour its guarantee. Assume
factory costs the company Re 0.50 to
0.20, respectively. If the result of the test marketing is negative that all defective balls
At a cost of Re 0.10 are returned.
per ball, the
they would decide to drop the product. I1, at any point, they drop that always company can conduct a test
the product there is a net gain of Rs 25,000 from the sale of correctly identifies both good and bad
(a) Draw a decision tree and tennis balls.
Scrap. Al financial values have been discounted to the present.
action and its
determine the optimal course of
Draw a decision tree for the problem and indicate the most expected cost.
(b) At what test cost should the
Mining Company is companyto indifferent to testing'?
preterred decision. 4. The Ore be

2. A manufacturing company has just developed a new product. a certain

piece of land should be attempting decide whether or not
On the basis of past experience, a product such as this
would 3,00,000. If there are commercial ore purchased. The land costs RS
either be SUCcessful, with an expected gross return
of Rs estimated value of property is Rs deposits on the land, the
1,00,000,or unsuCcessful, with an expected gross
return of Hs exist, however, the 5,00.000. If no ore deposits
have a Before property value is estimated at Rs
20.000. Similar products manutactured by the Company
50 cent of the time. The
purchasing
of Rs 20,000. The
the land, the 2,00,000
property can be cored at a cost
record of being successtul about per
coring will indicate if
production and marketing cOsts of the new product are expected or unfavourable tor
ore conditions are favourable
the mining. If the coring
to be Rs 50 000 probability of recoverable
ore
report is favourable
Iheconpariy is Considering
whether to markel this new However, it the coring
report is
deposits on the land is 0.8.
product or to drop it. Before making its decision, a test marketing only 0.2. Pnor to
obtaining any coring untavourable then probability
effort can be conducted at a cost of Rs 10,000.
Based on pasi estimates that the information, the
s
Odds of ore management
being present on the land is Su
 Decision Theory and Decision Trees
377
50. Management has also received
coring reports on
land similar to the ore in
question and found that 60 pieces of company decides to drill on the land it will cost Rs 200 lakh. If
the coring reports were favourable. per cent of the company does drill it may hit an oil wel.
gas well or a dr
Construct a decision tree and determine hole. Drilling experience indicates that the
probability of striking
should purchase the land, decline to whether the company an oil well is 0.4 on
type A and 0.1 on type B formation.
coring test betore purchase it, or take a Probability of hitting gas is 0.2 on type A and 0.3 on type B
making its decision. Specify the
course of action and EMV. optimal formation. The estimated discounted cash value from an oil well
is Rs 1.000 lakh and from a gas well is Rs 500 lakh.
5. XYZ company, dealing with a This
newly invented telephone device, includes everything except cost of mineral rights and cost of
is faced with the problem of
available cOurses of action:
selecting out of the
following drilling. Use the decision-tree approach and recommend whether
the company should purchase the mineral
() manufacture the device itself; or rights?
9. The Sensual Cosmetic Co. has developed a new perfume which
(i) be paid on a royalty basis by another
(ii) sell the rights for its invention for a lumpmanutfacturer: or management feels has a tremendous potential. It not only
sum. interacts with the wearer's body chemistry to create
The profit (in '000s Rs) that can be
unique
fragrance but is also especially long lasting. A total of Rs 10 lakh
expected in each case and
the probabilities
associated with the level of sales are shown in has already been spent on its development. Two marketing
the following table: plans have been devised:
(a) The first plan follows the company's usual policy of giving
Outcome small samples of the new products when other items in the
Probability Manufacture Royalties Sell All
Itself
company's product lines are purchased and placing
Rights advertisements in women's magazines. The plan would
High sales 0.1 75 35
cost Rs 5 lakh and it is believed that it might result in a high,
15
Medium sales 0.3 25 20
moderate or low market response with probability of
0.2
15
Low sales 0.6 10 0.5 and 0.3, respectively. The net profit excluding
10 15
development and promotion costs in these cases would be
Represent the company's problem in the form of a decision tree. Rs 20 lakh, Rs 10 lakh and Rs 1 lakh, respectively. Ifit
Further redraw the decision tree by later appears that the market response is going to be low
introducing the following it would be possible to launch a TV advt. campaign. This
additional information:
Would cost another Rs 7.5 lakh. It would change the market
(a) lf it manufactures itself, and the sales are medium or
high, response to high or moderate as previously described but
then the company has the opportunity of developing a new
with probability of 0.5 each.
version of its telephone;
(b) From past experience the company estimates that there is
() The second marketing plan is much more aggressive than
a 50 per cent chance of successful
the first. The emphasis would be heavily upon TV
development, advertising. The total cost of this plan would be Rs 15 lakh,
(C) The cost of development is Rs 15 and returns after
but the market response woulcd be either excellent or good,
deduction of development cost are Rs 30 and Rs 10 for
with probabilities of 0.4 and 0.6, respectively. The profit
high and medium sales, respectively.
excluding development and promotion costs would be Rs
6. An automobile owner faces the decision as to which deductible 30 lakh and Rs 25 lakh for the two outcomes.
amount of comprehensive insurance coverage to select. Advise on the sequence of strategy to be followed by the
Comprehensive coverage includes losses due to fire, vehicle company.
theft, vandalism and natural calamites. The possible choices are 10. The investment staff of TNC Bank is considering four investment
zero deductible coverage for Rs 60 per year or Rs 50 deductible proposals for a client: shares, bonds, reai estate and saving
coverage for Rs 45 per year. (The owner pays the first Rs 50 cerificates. These investments will be held for one year. The
of any loss of at least Rs 500.) Considering accidents covered past data regarding the four proposals are given below
by the comprehensive portion of the policy some of the owner's
Shares: There is a 25 per cent chance that shares will decline
utility values are given below by 10 per cent, a 30 per cent chance that they will remain stable
- 45
95 60 50
Amount (Rs) and a 45 per cent chance that they will increase in value by 15
0.2 0.4 0.45 0.47 0.5 per cent. Also the shares under consideration do not pay any
Utility
risk dividends.
a) Sketch the owner's utility curve. Is the owner a

avoider, an EMV'er or a risk taker? Bonds: These bonds stand a 40 per cent chance of increase
( ) O n the basis of the utility curve drawn in part (a), should in value by 5 per cent and 60 per cent chance of remaining
the owner take the zero deductible or the Rs 50 deductible stable and yield 12 per cent.
comprehensive coverage? Real Estate: This proposal has a 20 per cent chance of increasing
has several independent investment 30 per cent in value, a 25 per cent chance of increasing 20 per
Suppose a company
opportunities each of which has an equal chance of gaining Rs cent in value, a 40 per cent chance of increasing 10 per cent in
the
100,000 or losing Rs 60,000. What is the probability that value, a 10 per cent chance of remaining stable and a 5 per cent
on two such investments? On three chance of losing 5 per cent of its value.
Company will lose money
SUch investments? On four such investments? Savings Certificates: These certificates yield 8.5 per cent with
of independent investment
1a company has a nunmber certanty.
risk Is relaiively small
in each of which the financial Use a decision tree to structure the alternatives to the investment
opportunities,
Compared to its overall asset position, why should the company
staff, and using the expected value criterion, choose the
ry to maximize EMV, rather than expected utility? alternative with the highest expected value.
. An oil drilling company is consIdering ihe purchase of mineral
11. A company has developed a new product in its R&D laboratory.
lakh. The price includes tests to
Tights propety of Rs 100
on a

indicate whether the property has type

A geological formation or
The company has the option of setting up production facility to
market this product straightaway. It the product is successful.
formation. The company will be unable to teli
type B geological then over the three years expected product life, the returns will
until the purchäse is ma(le. It is
the type of geological formáalion be Rs 120 lakh with a probabilitly of 0.70. If the market does not
known however that 40 per cent of the land in this area has
respond favourably, then the returns will be only Rs 15 lakh, with
per cent type B tormation. If the
type A formation and
b0
a probabilitly of 0 30
378 Operations Research: Theory and Applications
As before, the returns from a successtur maket would be
F
The company is considering whether it should test-market this unsuccesstul market only Rs 15 lakh Th
product by building a small pilot plant. The chance that the test 120 lakh and from an
is Hs 40 lakh and the cost of tect
installation cost to production
market will yield favourable response is 0.80. If the test market is RS 5 lakh. Using decision-tree
gives a favourable response, then the chance of successful marketing the pilot plant
a decision tree diagram. carry
out necessary
analysis, draw
total market improves to 0.85.
to determine the optimal decisions.
If the test market give a poor response, then the chance analysis
o [Delhi Univ.. MBA, 2002
Success in the total market is only 0.30.

HINTS AND ANSWERS

5. Royalties; IMV = Rs 15.5

. The company should test market rather than drop the product.
If test market result is positive, the company should market the If p is the probability of high sale, then
product otherwise drop the product. >0.24
(i) manufacture itself when p
0.07 <p <0.24
2. Test market should be followed; EMV Rs 13,800 (1i) paid on royalty basis when
<0.071
3. (a) Do not test. Re 0.05: (b) Re 0.05 ii) sell all rights for lump sum when p
Test: If favourable purchase: If not - do not purchase; Rs 6. (b) Rs 50 deductible
64.000 7. 0.25; 0.50; 0.3125

.
Conditional Value Expected Value
Decision Point Outcome Probability
(Rs)
D3 i) Drill Oil well 0.1 1,000 100
Dry hole 0.6 0
Gas 0.3 500 150
250
Less: Cost 200
50
(i) Do not drill 0
Total 50

Oil well 0.4 1,000 400

D i) Drill
Dry hole 0.4 0 0

Gas 0.2 500 100

500
Less: Cost 200
300

0
(i) Do not drill
Total 300
*

A formation 0.4 300 120

D (i) Type
(ii) Type B formation 0.6 50 30
150
Less: Cost 100
Total 50

The company should purchase the mineral rights.

9. Aggressive Plan.

CHAPTER SUMMARY

of uncertainty and risk. In certain situations,

a
Decision theory provides a 1rame
work 1or ralional decisiOn making in an environment

Single sequence of decisions (with additional information that be available

decision-maker needs to make either
a decIsion or a may
hetween decisions). A numberofcourses of action (strategies or decision alternatives) are available for cach decision. The uncontrollabie
factors (also called states of nature) allect the payoll likely to be oblained from a course of action.
 Decision Theory and Decision Trees
379
The slate of nature likely to occur is learned only after making the decision.
of the rTespective state ot nature. However, it is possible to estimate prior probabilities
Baye's decision rule uses these prior
action and choose the one with the probabilities to determine the expected payoft for each course
largest expected payoff.
Sometimes while
making a sequence of decisions additional
The information is obtained about the probabilities of various states
nature. expected vaue ot
periect information provides, at a of
glance a view to understand whether taking a particular decision might
be of any worth.

A decision tree used to

display graphically the progression of sequential decisions and random events. The
is

helps incorporate decision maker's attitude towards risks. Problems

to concept of utilities
involving the possibility large
of losses the Baye's decision rule
isapplied
to express payoft
in terms of utilities rather than
monetary values.

CHAPTER CONCEPTS QUIZ

True or False
1. A course of action that may be chosen
by a decision maker is (c) analysis of information that is available
called an alternative.
(d) all of the above
2. In decision theory probabilities are not associated with states of 23. Which of the following criterion is not used for decision-making
nature.
under uncertainty?
3. The minimum expected opportunity loss criterion will (b) maximax
in the same decision as minimum regret.
always result (a) maximin
(c) minimax (d) minimize expected loss
4 In Hunwicz describes making the coefficient of realism describes 24. Which of the following criterion is not applicable to decision-
the degree of optimism. making under risk?
5. Hurwicz criterion of decision making always result in the highest (a) maximize expected retun
long-fun average payoff. (b) maximize return
(c) minimize expect regret
6.
The expected monetary value criterion is used for decision mak
ing under risk. (d) knowledge of likelihood occurrence of each state of nature.
25. The minimum expected opportunity loss (EOL) is
Minimax regret decision making criterion uses an opportunity loss (a) equal to EVPI (b) minimum regret
table.
(c) equal to EMV (d) both (a) and (b)
8. Maximum EMV is paid for geting perfect information.
26. The expected value of perfect information (EVPI) is
9 The difference between the highest the lowest EMV is said to be (a) equal to expected regret of the optimal decision under risk
EVPI. (b) the utility of additional information
10. The minimum expected opportunity loss is equal to the minimum (c) maximum expected opportunity loss
regret. (d) none of the above
27. The value of the coefficient of optimism (a) is needed while using
Fill in the Blanks the criterion of
11. in decision theory,.
called an alternative.
(a) equaly likely (b) maximin
12. The expected monetary value criterion is used for decision mak- (c) realism (d) minimax
ing under 28. The decision-maker's knowledge and experience may influence
13. The expected monetary value decision making criterion results in the decision-making process when using the criterion of
the
(a) maximax (b) minimax regret
average payof.
14. The coefficient of realism describes the degree of . of the (c) realism (d) maximin
29. The difference between the expected profit under conditions of
decision maker.
is risk and the expected profit with perfect information is called
15. The payoff due to equally likely criterion of decision making
(a) expected value of perfect information
same as minimum (b) expected marginal loss
of nature is available, than
.
probability of an outcome and state (c) expected opportunity loss
the decision-making environment is called.
(d) none of the abOve
The expected value of perfect information is equal tothe. 30. The concept of utility is used to
expected opportunity loss. (a) measure the utility of money
of the payoffs using the probabili (b) take into account aversion of risk
18. Expected payoff is the .
ties of the states of nature as the weighis. (c) both (a) and (b)
of the (d) none of the above
9. Posterior probabilities are the revised probabilities
to improve the prior probabilities. 31. The expected value of perfect information is equal to
ater conducting a survey
in calculating the of the (a) EPPI - Min (EMV) (b) EPPI+Max (EMV)
20. Probability tree diagram helps (c) Max. (EOL) (d) None of the above
states of nature.
32. Essential characteristics of a decision model are
(b) decision alternatives
ultiple Choice (a) states of nature
environment is (c) payoff (d) all of the above
21. A type of decision-making
(b) uncertainty 33. While using Hurwiez criterion, the coefficient of realism (a)
(a) certainty (d) all of the above (a) represents the degree of optimism
(c) risk
concerned With (b) represents the degree of pessimisms
22. Decision theory is
at an optimal decision (c) is the probability of a state of nature
(a) methods of arriving
decision in sequential manner (d) none of the above
(b) selecting optimal
380 Operations Research: Theory and Applications
34. The probability of selling at least an additional unit of an item to 35. The decision-making criterion that should be used to achieve max
justify the stocking of that unit is given by mum long-term payoff is
(a) p< IC(IC+ IP) (a) EOL (b) EMV
(b) P> ICI(IC+ IP
(c) p lC(IC + IP) (d) p IPI(IC + IP) (c) Hurwicz (d) Maximax

Answers to Quiz
1 T 2 F 3. F T 5. F 6. T 7. T 8. F 9. F 10. F
11. a course of action 15. EOL Criterion
or strategy 12. risk 13. highest 14. optimism
16. risk 17. minimum 18. weighted average 19. states of nature 20. posterior probabilities
21. (d) 22. (d) 23. (d) 29. (a)
24.(b) 25. (d) 26. (a) 27. (c) 28. (c) 30. (c)
31. (a) 3 2 . (d)
33 (a) 34 (c) 5. (b)

CASE STUDY

Case 11.1: Oil India Corporation

he corporation has recently got leasehold drilling rights on a large area in the western part of the country. No seismic
cOverage is available and to conduct a detailed survey Rs 3 million is required. If oil is struck, a large reserve may
result in a net profít of Rs 30 million, whereas a smaller marginal reserve may result in a net profit of Rs 18 million.
The cost of drilling a wildcast well is Rs 7 million.
Seismic 1s thought to be quite reliable in this area. Uncertainty pertains to whether or not a structure exists. The
company assesses a probability that the test producing good, fair or bad result is 0.40, 0.30 and 0.30, respectively.
On the basis of past drilling records and experiences indicating the probabilities of striking oil in large reserve, smaller
marginal reserve or dry hole, even in the presence of good, fair and bad reading of seismic study are as under:

Seismic Stucy Probability of Yeld

Large Reserve Margina Reserve Dry Hole
Good 0.50 0.25 0.25
Fair 0.30 0.30 0.40
Bad 0.10 0.20 0..70

Exploratory group has suggested two pOssible exploration strategies:

once on the basis of present geologic interpolation and extrapolation
Drill at
(a)
(b) Conduct a seismic study and defer drilling till seismic data is reviewed
As a member of strategie group of the company evaluate the two strategies suggested by the exploratory group.

Case 11.2: Moola Farms

Mr. Moola, the owner of Moola Farms is altempting to deeide which of three crops he should plant on his 100 acre

farm. The profit from cach crop is strongly dependent on the raintall during the growing season. He has categorized
the amount of the rainfall as substantial, moderale or light. He estimates his profit for each erop as shown in the table

below
Rainfall Estimaled Profit (Rs)
Crop A Crop B Crop C
Substantial 7,000 2,500 4,000
Moderale 3,500 3,500 4,000
Light ,000 4,000 3,000
s e d on the weather in previous seasons ad the cuTent projection lor the coming season, he estimates the
nrahahility of substantial rainfall as ).2, that or hoderale ralnlal as 0.3, and that of light rainfall as 0.5.
 Decision Theory and Decision Trees 381
Furthermore. services of forecasters could be employed to provide a detailed survey of current rainíall
prospects
as shown in the table.

Rainfall Estimated Profit (Rs)

Crop A Crop B Crop
Substantial 0.70 0.25 0.05
Moderaie 0.30 0.60 0.10

Light 0.10 0.20 .70

You in the capacity

of Block Advisor to farmers are expected to suggest
(a) Which erop should Mr. Moola plant based on existing data.
(6) Whether it would be economical for the farmer to hire the services of a forecaster. If yes, then how will the result
aftects the decision process.

Case 11.3: Matrix India

Matrix India, a company in FMCG sector is planning to launch a new product that can be initially introduced in
Westerm India or in the whole country. If the product is introduced only in Western India, the investment outlay will
be Rs 12 million. After two years, cornpany can evaluate the project to determine whether it should cover the whole
country. For such an expansion it will have to incur an additional investment of Rs l10 million. To introduce the product
in the whole country right in the beginning would involve an outlay of Rs 20 million. The product, in any case, will
have a life of 5 years, after which the plant will have zero net value.
If the product is introduced only in Western India, demand would be high or low with the probabilities of 0.8
and 0.2, respectively and annual cash inflow of Rs 4 million and Rs 2.5 million, respectively.
If the product is introduced in the whole country, right at the beginning the demand would be high or low with
the probabilities of 0.6 and 0.4, respectively and annual cash inflows of Rs 8 million and Rs 5 million, respectively.
Based on the observed demand in Western India, if the product is introduced in the entire country the following
would exist for high and low demand all India basis.
probabilities on an

s t e mndia Whole Conun

High Demnand Low Bean
High demand 0.90 0.10
Low demand 0.40 0.60

The hurdle rate applicable project is 12 per cent.

to this Advise Matrix India on the investment policy it should follow.

Support your advice

with appropriate reasoning.