10/16/2017

Table of contents

 Decision making

 Steps in making decisions

Unit 17: Risk & Decision Making
 Types of environment for Decision
making

 Decision trees

Risk & Decision Making 1 Risk & Decision Making 2

Introduction- Decision Theory Definitions in Decision Theory

 Each one of us is concerned with making decisions of  Decision-maker: Refers to an individual or a group of
one kind or another. individuals responsible for making a choice of an
appropriate course of action from the available
 Some of these are really difficult to make because of courses of action.
the complexity of a decision situation.
 Courses of action: Also called actions, acts, alternatives
 This complexity led many decision-makers to analyse or strategies. For a problem, all possible courses of
action should be included.
the process of decision-making.
 States of nature: Sometimes called outcomes or
 Need for decision theory which consists of several events. Decision-maker must develop an exhaustive
methods. list of possible future events. However, the decision-
maker has no control over the occurrence of
 Help decision-makers select wisely the best course specific event.
of action amongst several alternatives.
Risk & Decision Making 3 Risk & Decision Making 4

Definitions in Decision Theory

Decision Making & Management
 Payoff: Is the effectiveness associated with a
 Management is essentially a decision making
particular combination of a course of action process.
and state of nature. Payoffs may be evaluated
in terms of Profit, Cost or Opportunity Cost.  Every aspect of management function is
decided by decisions.
 Payoff Table: Lists the states of nature and a  Decision making therefore involves choice of
set of given courses of action. For each the best alternative among two or more
combination of states of nature and course of alternatives to achieve the objectives.
action, the payoff is calculated.
 OR techniques are helpful in making decisions.

Risk & Decision Making 5 Risk & Decision Making 6

1
 10/16/2017

Steps in Decision Making Decision Theory Models

 Identify the problem  Decision theory models can be classified on the
basis of the degree of certainty and can be
 Collect the needed information classified as follows:
 Formulate available alternatives (Courses of action) and ◦ Decision-making under certainty.
outcomes (States of nature)
◦ Decision-making under risk or probabilistic situation.
 Assess and weight the risk associated with each
alternative
◦ Decision-making under uncertainty.
 Select the best alternative
◦ Decision-making under conflict or theory of games.
 Implement the decision
Risk & Decision Making 7 Risk & Decision Making 8

Decision-Making under certainty Example 1

Certainty implies that all facts are known with certainty.

 An industrial enterprise can sell 500 units of
 Generally, decisions are made based on maximizing a product to another company at a price of
profit or minimizing cost. Rs. 40 per unit. The fixed cost are Rs.
10,000 and the variable cost/unit is Rs. 30. If
 Models constructed under certainty are called
deterministic models. For example: the enterprise has excess capacity, should
the order be accepted?
◦ Deterministic Inventory Models
◦ Game with saddle point
◦ Some of the replacement models
◦ Linear programming models

Risk & Decision Making 9 Risk & Decision Making 10

Decision-Making under risk Expected Monetary Value (EMV)

 Best strategy is selected on the basis of the
 Situations where alternatives are recognized but their
highest expected EMV.
resulting consequences are probabilistic.

 Decision-maker knows the probability of occurrence  EMV for a course of action is the sum of the
of each states of nature. products obtained by multiplying the payoff for
a given outcome by its probability value.
 Models are called probabilistic models. For example:
◦ If m1, m2,….,mn, are the payoffs corresponding to the
states of nature S1, S2,.....,Sn respectively. If the
◦ Queuing theory models corresponding probabilities of S1, S2,...,Sn are
◦ Simulation models p1, p2,...,pn , then the EMV is defined as:
◦ Probabilistic inventory models
◦ Some of the replacement models ◦ EMV = m1 p1+ m2 p2+….mn pn = ∑mp
◦ Genetic programming
Risk & Decision Making 11 Risk & Decision Making 12

2
 10/16/2017

Expected Opportunity Loss (EOL) Example 2

 Another decision criterion of decision theory is called Expected
Opportunity Loss (EOL).
 A proprietor of a food-stall has invented a
new item of food delicacy which he calls:
 Opportunity loss represents the amount of profit that is lost
because the most profitable course of action is not taken. WHIM. He has calculated that the cost of
manufacture is Rs. 1 per piece and that
 To calculate EOL, we must find the Conditional Opportunity Loss
(COL). because of its novelty and quality it would
 COL of the optimal act being zero, the COL of any other act is
be sold for Rs. 3 per piece. It is however
the difference between the payoff of the optimal act and the perishable and any good unsold at the end
action taken, and obviously will always be positive. of the day are a loss. He expects the
 When payoffs are replaced by their corresponding opportunity demand to be variable and has drawn up the
losses, we get a Loss Table.
following probability distribution expressing
his estimates:
Risk & Decision Making 13 Risk & Decision Making 14

Example 2 Example 2- Solution

No. of pieces
demanded
10 11 12 13 14 15  There are 6 courses of action. Let Si
Probability 0.07 0.10 0.23 0.38 0.12 0.10 represents the courses of action.
 Formulate: (i=1,2,....,6).
◦ The courses of action ◦ S1 : manufacture 10 pieces of WHIM.
◦ The states of nature ◦ S2 : manufacture 11 pieces of WHIM.
◦ A payoff table. ◦ S3 : manufacture 12 pieces of WHIM.
 Calculate the expected profit or loss for each ◦ S4 : manufacture 13 pieces of WHIM.
level of manufacture. ◦ S5 : manufacture 14 pieces of WHIM.
 How many pieces should be manufactured so ◦ S6 : manufacture 15 pieces of WHIM.
that his net expected profit is maximum?
Risk & Decision Making 15 Risk & Decision Making 16

Example 2- Solution Example 2- Solution

 There are 6 states of nature. Let Oj  Pay off table
represents the states of nature. (j=1,2,....,6). Production (Courses of Action) Probabilit
◦ O1 : demand is 10 pieces of WHIM. Demand S1 S2 S3 S4 S5 S6 y

◦ O2 : demand is 11 pieces of WHIM. O1 :10 20 19 18 17 16 15 0.07

O2 :11 20 22 21 20 19 18 0.10
◦ O3 : demand is 12 pieces of WHIM. O3 :12 20 22 24 23 22 21 0.23

◦ O4 : demand is 13 pieces of WHIM. O4 :13 20 22 24 26 25 24 0.38

O5 :14 20 22 24 26 28 27 0.12
◦ O5 : demand is 14 pieces of WHIM.
O6 :15 20 22 24 26 28 30 0.10
◦ O6 : demand is 15 pieces of WHIM.  E.g: O1 = 10 & S2 =11
◦ = (10*3) – (11*1)= 19

Risk & Decision Making 17 Risk & Decision Making 18

3
 10/16/2017

Example 2- Solution Decision-Making under uncertainty

 The expected profit for various levels are:
◦ E(S1)
 More than one state of nature exists here,
=(20*0.07)+(20*0.1)+(20*0.23)+(20*0.38)+(20*0.12)+(20*0.1 but the decision-maker lacks the knowledge
)=Rs. 20 about the probabilities of their occurrences.
◦ E(S2) = Rs. 21.79
◦ E(S3) = Rs. 23.28  Called uncertainty models
◦ E(S4) = Rs. 24.08
◦ E(S5) = Rs. 23.74  A few decision criteria are available which
◦ E(S6) = Rs. 23.04 can be helpful to the decision maker.

 The proprietor should manufacture 13 pieces

for maximizing profit which is Rs. 24.08
Risk & Decision Making 19 Risk & Decision Making 20

Uncertainty Models- Decision Criteria Uncertainty Models- Decision Criteria

 Maximaxi (optimistic) criterion  Hurwicz criterion (Criterion of realism)
◦ Decision maker find the maximum possible outcome for ◦ Compromise between Maximaxi and Maximini.
each alternative and then chooses the alternative which
gives maximum of maximum outcomes. ◦ Weighted average for each alternative is determined
as follows:
 Maximini (pessimistic) criterion  Weighted average outcome for alternative i =
◦ Decision maker find the minimum possible outcome for α * Max outcome for alternative i + (1- α ) * Min outcome for alternative i
each alternative and then chooses the alternative which  Where α is the degree of optimism
gives maximum of minimum outcomes. ◦ Select the alternative corresponding to the maximum
of these weighted outcomes.
 Equal probability criterion (Laplace)
◦ Decision maker find the average of all outcomes for each
strategy and then chooses the strategy corresponding to  Minimaxi regret criterion
the maximum of average outcomes. ◦ Decision maker chooses the alternative
corresponding to the minimum of maximum regrets.
Risk & Decision Making 21 Risk & Decision Making 22

Example 3 Example 3..cont

 A toy company is bringing out a new type of toy. The  Take the optimal decision under each of the
company is attempting to decide whether to bring out a
full, partial, or minimal product line. The company has three following decision criteria:
levels of product acceptance. Management will make its
decision on the basis of expected profit from the first year
of production. The relevant data are shown in the following
◦ Maximaxi criterion
table:
◦ Maximini criterion

◦ Laplace criterion

◦ Develop an opportunity loss table and hence

determine the minimaxi regret criterion.
Risk & Decision Making 23 Risk & Decision Making 24

4
 10/16/2017

Example 3..solution Example 3..solution

 The maximum profit (in Rs.000’s) for each  The minimum profit (in Rs.000’s) for each
product line is product line is

◦ Full 80 ◦ Full -25

◦ Partial 70 ◦ Partial -10
◦ Minimal50 ◦ Minimal 0

 Maximum of maximum outcome is Rs. 80,000.  Maximum of minimum outcome is Rs. 0.

 The company should go for full product line  The company should go for minimal product
under maximaxi criteria. line under maximaxi criteria.
Risk & Decision Making 25 Risk & Decision Making 26

Example 3..solution Example 3..solution

 Product
Anticipated 1st year Profit (Rs.000’s)
 Opportunity loss table as follows
Acceptance Product Line
◦ Regret in each cell = Max value in that row – Entry
Full Partial Minimal
in that cell
Good 80 70 50 Anticipated 1st year Profit (Rs.000’s)
Fair 50 45 40 Product
Acceptance Product Line
Poor -25 -10 0 Full Partial Minimal
Average Good 0 10 30
35 35 30
outcome
Fair 0 5 10
Poor 25 10 0
 Maximum of average outcome is Rs. 35,000 Max regrets 25 10 30

 Minimum of maximum regrets is Rs. 10,000.

 The company should go for full or partial  The company should go for partial product line
product line under Laplace criteria under minimax criteria.
Risk & Decision Making 27 Risk & Decision Making 28

Decision Trees Decision Trees

 In a multistage decision problem, an outcome of  Is made of nodes, branches, probability
one decision affects further decisions. estimates and payoffs.

 A graphical representation of the multistage  The decision tree has 2 types of nodes:
decision problem can be made by using a ◦ Square represents decision point. Numbers of
decision tree. alternatives are available at decision point where
decisions are taken to maximize the return.
 A decision tree is a simple graphic device which
◦ Circle represents chance of an event.
enables the decision maker to understand more
clearly the alternative options along with risks
associated with each of them. Decision tress are useful for finding the optimal
alternative.
Risk & Decision Making 29 Risk & Decision Making 30

5
 10/16/2017

Example 4 Example 4- cont

 The General Ford Motors Corporation (GFMC) is planning the
introduction of a brand new SUV: the Vector. There are two options for Production Costs ($)
production. One is to build the Vector at the company’s existing plant in
Indiana, sharing production time with its line of minivans that are currently Moderate Sales Strong sales
being produced there. If sales of the Vector are just moderate, this will
work out well as there is sufficient capacity to produce both types of Shared Plant in Indiana 16,000 24,000
vehicles at the same plant. However, if sales of the Vector are strong, this Dedicated Plant in Georgia 22,000 20,000
option would require the operation of a third shift, which would lead to
significantly higher costs.  The amortized annual cost of plant construction and other
associated fixed costs for the Georgia plant would total $400
 A second option is to open a new plant in Georgia. This plant would have million per year (regardless of sales volume). The fixed costs for
sufficient capacity to meet even the largest projections for sales of the adding Vector production to the plant in Indiana would total
Vector. However, if sales are only moderate, the plant would be $200 million per year (regardless of sales volume).
underutilized and therefore less efficient
◦ Draw a decision tree indicating the revenue for each possibility.
 This is a new design, so sales are hard to predict. However, GFMC
predicts that there would be about a 60% chance of strong sales (annual ◦ Using the decision tree to determine which production option
sales of 100,000 vehicles), and a 40% chance of moderate sales (annual maximizes the expected annual profit, considering fixed costs,
sales of 50,000 vehicles). The average revenue per Vector sold is $30,000. production costs, and sales revenues.
Production costs per vehicle for the two production options depend
upon sales, as indicated in the table below.
Risk & Decision Making 31 Risk & Decision Making 32

Example 4-Solution Example 4-Solution

 If the Indiana plant is used and sales are strong, the revenue
will be:
100,000 * (30,000 – 24,000) = 600 M

 If the Indiana plant is used and sales are moderate, the

revenue will be:
50,000 * (30,000 – 16,000) = 700 M

 If the Georgia plant is used and sales are strong, the

revenue will be:
100,000 * (30,000 – 20,000) = 1,000 M

 If the Georgia plant is used and sales are moderate, the

revenue will be:
50,000 * (30,000 – 22,000) = 400 M
Risk & Decision Making 33 Risk & Decision Making 34

Example 4-Solution
◦ Expected revenue at node 2: (600*0.6) + (700*0.4)
= 640 M

◦ Expected revenue at node 3: (1,000*0.6) + (400*0.4)

= 760 M

 There is a fixed cost at Indiana and Georgina is 200

M and 400 M respectively.

 Subtracting the fixed cost, the profit at Indiana and

Georgina is 440 M and 360 M respectively.

◦ At decision point 1, the decision to share the plant

in Indiana has a higher expected profit of 440 Million
Risk & Decision Making 35