D e c i s i o n T r e e

rees

eisa graphic
graphicrepresentation
treeis a
r ofthe decision process indicating
n
4decisior
atives,
states of
nature, probabilitiesattached to the states of
onalermati

ditional
benefits and
losses. It consists of a network of nodes and
types
nodes are used:
of decisionnode represented by a square
Two
wre(chance
nature (chan or event) node represented by a circle. Alternative
S W O o

(strategies) originate from the decision nodes as main branches

sofaction(

ranches). At
the eend of each decision branch, there is a state of nature
hanches).
which
Sson
emanate chance events in the form of sub-branches (chance
which emanate
Th respective payoffs
and the probabilities associated with
The
RCies)
and ithe chance events are shown alongside these branches.
ive
ative courses
courses

theterminal of
erminal the chance branches are shown the expected values of the
WCOme

The general approach used in decision tree analysis is to work backward

chance
Lauhthe tree from right to leit, computing the expected value of each leads
eWe then choose the particular branch leaving a decision node which
he chance node with the highest expected value. This is known as roll back
old back process.

Lample 14.16
Aclient asks an estate agent to sell three properties A, B and C for him
Nd agrees to pay him 5% commission in each sale. He specifies certain
0ONAditons. The estate agent must sell property Afirst, and this he must do within
ays. f and when A is sold the agent receives his 5% commission on that
Etie can then either back out at this stage or nominate and try to sell one
remaining two properties within 60 days. Ifhe does not succeed in selling
Omunatedproperty in that period, he is not given the opportunity to sell the
property on the same conditions. The prices, selling costs (incurred by the
ent whenever a sale is made) and the estate agent's estimated
Pobability ofmaking a sale are given below.
Probability of sale
Property Price of property Selling costs
0.7.
A Rs. 12,000 Rs. 400
B 225 0.6
25,000
C 450 05
50,000
) Draw
uw up an appropriate decision tree for
the estate agent.
up
What is tthe estate agent's best strategy under EMV approach?
hat is
650
Operalions Rese
Solution
The decision tree diagram for the problem is showniin
sear
Fig. 14A

ASSEMBLY O.K.
ASPER
TEST
0:7 Rs.
A C T U A L L Y G O O D

BAD AS
0.7

PER TEST
0.RIESA
0.3 Rs.50
TEST

R s .
1 0

ASSEMBLYFAULTY
ACTUALLY 0.K.AS
PER
0-2
TEST
RS.14
NO TEST BAO AS
RS.50 0-8PER TEST
Fig. 14.4
RS.S0
EMV ofnode B =Rs. [0.7x0+0.3 x 50] = Rs. 15.
EMVof node C = Rs. [0.2 x 140+0.8 x 50] = Rs. 68.
EMV of node A Rs. [0.7 x 15 +0.3 x 68] Rs. 30.90.
= =

EMV of node 1 via the branch "Test' Rs. [10 x

30.90J
=
=
Rs. 4090
EMV of node 1 via the branch 'No Test' Rs. 50. =

.Optimal EMV of node 1 is Rs. 40.90

corresponding to the opima
decision "Test the sub-assemblics.

Example 14.19
A large steel
manufacturing company has three options with regard o
production (i) produce commercially (ii) build pilot plant (ii) stop producin
steel. The management has
estimated that
chance of high yield and 0.2 chance low their pilot plant, if built, has
high yield, management assigns a of yield. If the pilot plant does snow
will also have a probability of 0.75 that the commerciul pa
high yield. If pilot plant shows a low yield, there is
the
0.l chance that the
commercial oniy
management's best assessment of theplant
will show a high yield. rih
yield on a commercial-size plan d
building a pilot plant first has a 0.6 chance
Rs. 3,00,000. The of high yield. A pilot plan D
profits earned under high and low are
1,20,00,000 and Rs. 12,00,000 yield conditionssionjor
the company. respectively. Find the optimum dec f
Solution
Fig. 14.5 represcnts the decision tree
diagram for the probi
 coecaLYC-
HIGH
0-6 RS.1,20, 00,000

HIGH
Rs. 1,20,00, 000
04 LOW
0-75

-RS.12,00,000
COMMERCIALLY 0-25
3 n70 47 /DUaN OS
HIGHYIELO_
PRODUCE

STOP
Rs.0
LOW
-RS. 12,00,000

HIGH
RS. 1,20,00,000
BUILD

1/PILOT PLANI 4
OslERS.J,000
LOW YIEL0 COMMERCIALLY
PRODUCE

0-2 0.9
OP LOW
STOP Rs.
-

12,00, 000
RS.
RS.
Fig. 14.5

EMV of chance node C = Rs. [0.75x 1,20,00,000 0.25 x

0)
=
Rs. [90,00,000 3,00,000]
Rs.87,00,000.
EMV of chance node D Rs. [0.1 x 1,20,00,000 0.9 x

= Rs. [12,00,000 10,80,000

= Rs. 1,20,000.
EMV of decision node 2 Rs. 87,00,000.
EMV of decision node 3 = Rs. 1,20,000.
EMV of chance node AA = Rs. [0.8 x87,00,000-0.2 x 1,20,000]
= Rs. [69,60,000- 24,000]
Rs. 69,36,000.
EMV of decision node 1 if pilot plant is built
= Rs. 69,36,000 - Rs. 3,00,000]

=. Rs. 66,36,000.
EMV of chance nodeB = Rs. [0.6 x 1,20,000-0.4 x 12,00,000]
Rs. [72,00,000-4,80,000]

= Rs. 67,20,000.
EMV ofdecision
decisi node 1 for alternative 'producecommercially
= Rs. 67,20,000.
but ld produce
should produce
The company the pilot plant
Anex.cokal y
Cally,auny
should not build

"myle 14.4.202N which,,

wh after paying for
Yler payine
process,
ny is currently working with D12.000.
wli a
The co
12,000. The company has he
s, labour, Currently working
lab a profit of Rs.
brings
ernai.
 ems InIn (
OOLems
Oper
(i) The company can conduct research R, which is ev aluons Rese
10,000 is expecedi
andhaving 90%probability ofsuccess. If
income will be Rs. 26,000.
(ii) The company can conduct research R2, the gyoa successful,
expected to
and having a probability of 60% success.
If Cost R
Rs.6
cesful, the
.
income will be Rs. 24,000.
(üi) The company can pay Rs. 5000 as p
royalty
royalty
will bring a gross income of Rs. 20.000. of a
a new
new
process whi
(iv) The company may continue the current process,.
Because of limited resources, only one of the two
carried out at a time. Draw the decision tree types ofres
and ind the ch canb
the company. opiimal strakegy j
Solution
The decision tree diagram is shown
Fig. 14.6. Point i is the dwi in
box. The four alternatives available to the
it. Upon failure of a particular rescarch, say
company are shown emanating te
R, there are again threc altematiw
conduct rescarch R,pay
royalty or gontinue the existing process. IfR,failsg
failure of Ri, the company is lett with only tWO Choices, t.e., either topay
or continue the existing process. royalu
R SUCCEEDS RS.
0.9 26,000
cURRENT PROCESSRs. 12,000
LICENCE
- RS. 5,000
-Rs. 20,000

CURRENT
- RS. 6,000
PROCES5RS 12,00
LICENCE0RS.20,000
-RS. 5,000

R2 5UCCEEDSRS. 24,000
0-6
CURRENT PROCE SS RS. 12,000

LICENCE
- RS. 5,000 -RS. 20,000

RI SUCCEE
0.9 DSRS. 26,000
CURRENT
PROCESS RS. 12,000

UCENCERs 20,0
- RS 5,000

cURRENT

-RS. 6,00O
PROCESS-RS. 12, 000
LICENCE
R S . 20,000
RS. 5,000
0-6
RS.
R2SUCCEE0S 24,00o Fig. 14.6
EMV of dccision node 4 (20,000-5,000)J=
Rs
Max. Rs. [12,000, =

EMV of decision nede 5 = Max. Rs. [12,000, 0,000-5,000)]=RS. R s2040

EMV of decision node C Rs. [0.4 x 15,000+0.6 =

EMV decision node D

of
X24,
x 15,000]=R
Rs. x x =
[0.9 26,000+0.1 15
 ames
J k o yAndGa

653
fheistOn 2
n o d e2 Max. Rs. [12,000, (20,000 5,000).
(20,400-6,000)]
= Rs. 15,000.
Max. Rs. [12,000,
dcson node3
(20,000-5,000),
(24,900- 10,000)]
= Rs. 15,000.
ylchance nodeA Rs. [0.9x 26,000 +0.1 x
15,000]
=Rs. 24,900.
chance node B Rs. [0.6 x24,000+0.4 x
15,000]
=Rs.
decisionnode 1 = Max. Rs. [24,900- 10,000), 12,000, 20,400.
(20,000- 5,000), (20,400 6,000)
= Rs. 15,000.

ThusLhe
1any should pay Rs. 5,000 as royalty of the new process to
Janaxumu,um expected prof+t of Rs. 15,000,

pe 14.21
ny dealing with newly invented telephonic device isfaced with the
ynofselecting
the following strategies:
device itself,
i) manufacture the
a) to be paid on a royal1y basis by another manufacturer,
l) sell the righis for its invention for a lump-sum
The profit in thousands of rupees that can be expected in each case and
pnbabilities associated with the sales volume are shown in the following

Table 14.29

Event Sell the rights

Probability | Manufactureitself| Royalties
h demand 0.2 100 40 20

ediurn demand 25 20
0.3 30
demand 15 20
0.5 -10
decision tree.
lo) of
Represent the company's problem intheformadditional a

inormation:
Extenh e diagramfurther for the following
t or high, it
()
f the comp manyfacturesitselfand sales are medium
has pany
has version ofits telephone.
the developing a
pportunity of
new
60% chance of
i) om estimates that there is
past exper it
uCCese
lii) f the cost development.
Successful deducting
the returns(after mediun
Rs. 20 and and
of evelopment
devela is
10for high
ethe devel
Rs.
Rs. 35 and
2velopment cost) a r e
4and
rand respectively. 14.7.

The decision problem

is shown in Fig.
for the
ree diagram
654 Problems In Operations
EMV of chance node A Resea
Rs. 1,000 [0.2 x 100 +0.3 30
0-0520
=
x
= Rs. 1,000 [20 +9- 5]
24,000.
E M V of chance node B
Rs. 1,000 02x40+0.3 x
5-05x15
=
MAHUF ACTURE
-SELE_ AMEDIUN CENANO 0 = Rs. 1,000 [8 +7.5 + 7.5]
= 23,000.
LCW CEMANO -10
EMV of chance node 1
SrL ALL IHTS
Max. Rs. [24,000, 20,000,
=
23,000
HIGH CEMAND
0 2
0 = Rs. 24,000.
Thus the optimal decCision for the
sAL ISGEu OEN compan isto
manufacture the telephonic device itself to
LOW OENANO 15
maximum expected profit of Rs. 24.000
(b) The decision tree diagram with
Fig. 14.7 addiuonal information is shown
in Fig. 14.8,
DEVELOP CSSUCCESS
0-6 0035 = I)3

FAILURE
0.4 100-2080
HIGH DEMAND NOT DEVELOP 100
0-2

MANUFACTURE MEDIUM
TSELF N DEVELOP 5UCCESS
0-3 30+10 40
FAILURE 30-20 10
0-4
NOT DEVELOP 30
LOW DE MAND
SELL ALL RIGHTS 05 20
HIGH DEMAND
0-2

MEDIUM
ROYALTIES-(®- DEMAND - 25
0-3

LOW DEMAND 15
0-5

Fig. 14.8
EMV of chance node C Rs.
=
1,000 [O0.6 x 135+ 0.4 x 80]
= Rs. 1,13,000.
EMV of chance nodeD =
Rs. 1,000 [0.6 x40 +0.4x 10]
Rs. 28,000.
EMV of decision node 2 =
Max. Rs. [1,00,000; 1,13,000]

EMV of decision node 3

=Rs. 1,13,000.
=
Max. Rs. [28,000; 30,000]
EMV of chance node A
Rs. 30,000.
=
Rs.[0.2x1,13,000+ 0.3 x30,000- 5x10
=Rs.26,600.
 Games
gion1heorAndG 655

n o d e
B
- Rs. [0.2x40,000 +0.3 x 25,000+0.5x 15,000
e n a n c e = Rs. 23,000.
AYol

ENMVof
decesionnode
node 1 = Max. Rs.
[26,600; 20,000; 23,000]
Rs.26,600.
timum expected profit to the company is Rs. 26,600. To earn
the ompany shoul manufacture the telephonic device itself.
n e w version of
amounl, h e
develop the telephone if the demand is
uld
develop it when the demand is medium. This is shown by high
bold
diagram
tree
the
on
ALmple
1 4 , 2 2

company
is sidering whether to enter a new market. In case
A
textie
decides to enter this market, it must either instal a new plant with
pan)3.00,000
any or pay overtime wages to its workers, with overtime
fthe company enters market, there is 60% chance
ofRs. 1,00,000. If1
ahoutla

1.00,000. the
the installation of the new plant. Also, a random
his kholders approving
slockh
rket structure reveals that there are 40% chances of
ofcurrent marke
30%.chances ofachievin8 medium sales,
rales by the company,
cieinghigh chances of achieving no sales..
ances of
achieving low sales ana 10%
a medium level of
high level of
sales will yield Rs. 10,00,000 profil,
00,000 profit, a low level of sales in a Rs. 2,00,000
ina Rs.
will resul
uls
wil
the company a loss of Rs..5,00,000 apart from the
No sales will cause
utofthe equipment. which option
the above problem and suggest
Draw a decision tree for
boldbe selected.

olution
14.9.
The decision tree diagram is shown in Fig.
RS. 10,00,000
RSALE,04-

M. SALE,0.3 -RS. 6,00,000

RS. 2,00,000
NEW PLANT, 0.5 L. SALE,02
-RS. 3,00,000
-RS. 5,00, oco
N. SALE,0.1 .

ENTER

NOTENTER RS.

-RS. 10,00,000
H.SALE,0.4

ENTER RS. 6,00,000

M. SALE,O0.3
PAY
RS. 2,00, 000
OVER TIME WAGESB L.
RS.1,00,000,0.4 X SALE,02
-RS. 5,00,000

N. SALE,0.1

Fig. 14.9
656

node A
= Rs. [0.4 x 0.000+0.3 x6
10,00,000+0.3 x6,00,000
EMV of hance +0.2 x 2,00,000 +0.1(-5,00,000
1,00,000 [4 + 1.8 +0.4-C.51
Rs.
= Rs. 5,70,000.

node B = Rs. 5,70,000.

Likewise, EMVof chance
plant is built
node 1 if new
EMV of decision = Rs. [0.6(5,70,000- 3,00,0001
Rs. 1,62,000.
are paid
EMV of decision node
1 if ovcrtime wages
Rs. 10.4(5,70,000-1,00,000)
Rs. 1,88,000.
does not enter the market D.
EMV of docision node l if the company use overtime to pet
Rs.0.
shouldenter the market and
an
The company
1,88,000.
opümun
expected prolit of Rs.

Example 14.23
An oil company has recently acquired rights in a certain area to condim
surveys and test drilling to lead to lifting ol. The area is considered to kaw
good potentialforfinding oil. At the outset, the company has a choice to ondus
further geological tests or to carry out a drilling programme immediately O
the known conditions thecompany estimates that there iS a 70: 30chanae
further tests showing a 'success'. Whether the tests show the possibliy d
ultimate success or not, or even if no tests are undertaken at all, the company
could still pursue its drilling programme or allernatively consider sellingi
rights to drill the area. Thereafter, however, if it carries ou the driling
programme, the likelihood of final success or failure is considered dependen!
on ils
foregoing stages. Thus if successful tests have been carried oul, the
expectationofsuccess in drilling is given as 80: 20.Ifthe tests indicate failure,
then the expectation of success in drilling is given as 20: 80. f no tesis haut
been carried out at al, the expectation of success in drilling is givenas5.4
Costs and revenues have been estimated for all possible outcomes and ue
present value is given below.
Outcomes Net present value (Rs. millions)
Success With prior tests 100
Withou prior tests 120
Failure With prior tests -50
W.unout prior tests
-40
Sale of exploitation rights
Prior tests show
Prior tests show
'success 65
failure
Withou prior tests
15
45
 h e o r yA n dC a m e s

657

ision tree diagram to represent the above information. Evaluate

Drawadec
rderadvise the management of the company on iis best course

decision
tree diagram is shown in Fig. 14.10.
The
Rs. 120
S,0.55

DRILL
F,0-45 RS. 40

SELLRIGHIS-Rs, 45
.Rs. 100
S,0.8

DRILL

F, 0.2 - R s . 50

5 ,0.7 SELL AIGHTF-RS. 65 R S . 10o

TEST S,0-2
DRILL
O. F,08 - RS. 50

SELLRIGHTSS 15
Fig. 14.10

EMV of chance node C =Rs. [100 x0.8 50x0.2] Rs. 70. =

EMV of chance node D =Rs. [100 x 0.2- 50x0.8] = - Rs. 20.

EMV of decision node2 max.

of Rs. (70, 65) Rs. 70. =

EMV of decision node 3 =max. of Rs. (- 20,15) =Rs. 15.

48.
Rs. [120 x 0.55- 40 x 0.45] Rs,
=
EMV of chance node A
Rs. 53.50.
EMV of chance node B R s . [70x0.7 + 15 x0.3] =

= m a x . of Rs. (48, 45, 53.50)

Rs. 53.50. =
EMV of decision node 1
these are successtul,
Dest Course of action is to carry out tests first. If
be C
b
corm the company should sell
the
should drill; if they result in failure,
th

SThe Theory of
Games
tis a ision situations. It is helpful
theory applicable to competitive
hen pwo OFor
NO with conflicting objectives
try
make more individuals or organisations
one decision
maker
decision made by
ers the CIsions. In such situations decision makers.

Oryis
decision made
maur lobya one or more of
of
the remaining
situations such
as two players
wide variety planning
ong
to win at chess, candidates fighting an election,twoenemies
windIe
based on
cs, firCness,
ucs, firms struggli their market
etc. It is shares,
S E n g tO maintain