1. Introduction to Decision Analysis
Decision analysis studies the process of making difficult decisions
The objective is to update, model, and document the intuition of managers
A structured approach to decision making is especially critical in group decision making
Key importance in the case of resource decisions (operations strategy)

2. Why are decisions difficult?
Complexity
Simply keeping all the issues in mind at one time is nearly imposible
Uncertainty
Making a decision is especially difficult when you are not sure about one decision variable’s state
Multiple objectives
Different perspectives lead to different conclusions

3. Example
Winter of 1985:
The Oregon Department of Agriculture (ODA) faced an infestation of gypsy moth in Lane County in Western Oregon
Forest industry representative call for an aggressive eradication campaign

4. Alternatives Use BT, a bacterial insecticide
target specific
ecologically safe
reasonably effective
Use Orthene, a chemical spray
registered as acceptable for home garden use
questions about its ultimate ecological effects
possible danger to humans
Possible to use both

5. Opinions Forestry officials: Environmentalists: Others Others...
Argue Orthene is more potent and is necessary to ensure eradication
Environmentalists:
Orthene potentially too dangerous
Others
Already too late anyway
Others...
Still time but decision/implementation has to be made now

6. Subjective Judgments
In decisional contexts such as the one faced by ODA
objective data is lacking
a procedure determining an « optimal » decision derived from objective data is of little use...
personal statements about uncertainty and value become important inputs (no choice...)
Discovering and finalizing these judgments is a key issue in decision analysis
Instead of criticizing them, we will learn how to better assess and use them

7. The Decision Analysis Process
Identify the decision situation and understand objectives
Identify alternatives
Decompose and model the problem
model of problem structure
model of uncertainty
model of preferences
Choose the best alternative
Sensitivity analysis
Is further analysis needed? yes/no?
Implement the chosen alternative

8. Requisite Decision Models
A model can be considered requisite only when no new intuitions emerge about the problem
or when it contains everything that is essential for solving the problem

9. Elements of a Decision Problem
Values and objectives
Decisions to make
Uncertain events
Consequences

10. Values and Objectives
Value: used in its general sense « things that matter to you »
Objective: Specific thing that you want to achieve
The set of objectives taken up together make up the values
Values are the reason for making the decision in the first place
They define the decision context

11. Objectives for Boeing’s Supercomputer
Cost
Five-year costs
Cost of
improved
performance
Management
Issue
Vendor Health
US Ownership
Commitment
to supercomputer
Performance
Speed
Throughput
Memory Size
Disk Size
On-site performance
User Needs
Installation date
Roll in/Roll out
Ease of Use
Software
compatibility
Mean time
between failures
Operational
Needs
Square footage
Water cooling
Operator tools
Telecommuni-
-cations
Vendor support

12. Decision to Make
Given a decisional context, one (or several) decision(s) has to be made
In some cases, several decisions may have to be made in a sequence
Time
Decision 1
Decision 2
Decision 3

13. Uncertain Events
Uncertain events are either linked to chance or are linked to a probability distribution
Uncertain events have outcome
It is important to position uncertain events appropriately between decisions
Decision 1
Decision 2
Decision 3
Time

14. Consequences
After the last decision has been made and the last uncertain event has been resolved, the decision maker’s fate is finally determined
Time
Decision 1
Decision 2
Decision 3
Consequence

15. Example Consequence: Cost $ Environmental damage PR damage Accident
Weather
for
Cleanup
Weather
Accident
Cost
Location
Environmental
Damage
Cause
Consequence:
Cost $
Environmental
damage
PR damage
Accident
Management
Decisions
Policy
Decision
Time

16. Making Choices Decision Trees Example: Texaco vs. Pennzoil
Decision trees and expected value
certainty equivalent

17. Decision Trees
Decision trees are a graphical representation of a decision problem
Large return
Venture succeeds
Invest
Funds lost
Venture fails
Do not invest
Typical return earned
on less risky investment

18. Decision Tree Safety Cost Evacuate Safe High Storm hits Miami Forecast
Danger
Low
Stay
Safe
Low
Storm misses Miami

19. Cash Flows and Probabilities
To each branch of the tree, we can attach
a probability
and/or, a cash flow
or any measure replacing monetary values for a specific problem

20. Case Study: Texaco vs. Pennzoil
In early 1984, Pennzoil and Getty Oil agreed to the terms of a merger
Before the signature of the formal agreement, Texaco offered Getty a substantially better price , and Gordon Getty (majority stockholder) defected on Pennzoil and sold to Texaco

21. Case Study: Texaco vs. Pennzoil
Pennzoil felt this was unfair practice and filed a lawsuit against Texaco, alleging that Texaco had interfered illegally in the the Pennzoil-Getty negotiations
Pennzoil won the case in late 1985 and was awarded $11.1 billion – the largest settlement in the US at this point in time
Texaco appealed and the settlement was reduced by $2 billion – but interest and penalty got the amount back to $10.3 billion

22. Case Study: Texaco vs. Pennzoil
Kinnear, Texaco’s CEO, announced that Texaco would file for bankruptcy if Pennzoil obtained court permission to secure the judgment by filing liens against Texaco’s assets
Kinnear promised to fight the case all the way to the Supreme Court

23. Texaco’s Offer
In April 1987, just before Pennzoil started filing liens, Texaco offered to pay Pennzoil $2billion to settle the entire case
Liedtke, chairman of Pennzoil, announced that his advisors estimated that a settlement of 3-5 billions would be fair
What should Liedtke do?

24. Decision Tree 2 5 10.3 5 10.3 5 3 Settlement Amount ($billion)
Accepts $2 billion 2
Texaco accepts $5 billion 5
Counteroffer
$5 billion
10.3
Texaco
refuses
counteroffer
Final Court Decision 5
10.3
Note: Several estimates are included in this decision tree and decision have in
effect, already been made:
If Pennzoil counteroffer, it will be for a settlement of $5billion
Texaco could counter-offer with another value than $3billion!
The court judgment is any value between 0 and 10.3 billion – the value of 5
is a compromise judgment – this, also, is an estimation
Final Court Decision
Texaco
counteroffers
$3 billion 5
Refuse
Accepts $3 billion 3

25. Subjective Probabilities
In the decision tree, we are missing probability estimates of the each event
For this lecture, we will take these probability values for granted

26. Decision Tree 2 5 10.3 5 10.3 5 3 Settlement Amount ($billion)
Accepts $2 billion 2
Texaco accepts $5 billion 5
(0.17)
(0.2)
10.3
Counteroffer
$5 billion
Texaco
refuses
counteroffer
Final Court Decision
(0.5) 5
(0.3)
(0.50)
(0.2)
10.3
(0.33)
Final Court Decision
(0.5)
Texaco
counteroffers
$3 billion 5
Refuse
(0.3)
Accepts $3 billion 3

27. Expected Monetary Value
Computing an expected monetary value is a way of selecting among risky alternative
Computing expected values bring the problem back to a « certainty equivalent »
What is the expected value of the court judgment?

28. Expected Value of the Court Judgment
EV = 0.2 * * * 0
EV = $ 4.56 billion
(0.2)
10.3
Final Court Decision
(0.5) 5
(0.3)
It is possible to reduce the tree with this certainty
equivalent

29. Reduced Tree 2 5 4.56 4.56 Eliminated 3 Accepts $2 billion
Texaco accepts $5 billion 5
(0.17)
Counteroffer
$5 billion
Texaco refuses counteroffer
4.56
(0.50)
(0.33)
4.56
Texaco
counteroffers
$3 billion
Refuse
Accepts $3 billion
Eliminated 3

30. Expected Monetary Value of the Counteroffer
What is the expected monetary value of Pennzoil $5 billion counter offer:
EV = P(Texaco accepts) * 5 + P(Texaco refuse) * P(Texaco counteroffers) * 4.56
EV = 4.63
Liedtke should not accept the $2 billion offer,
and should counter-offer $5 billion. If Texaco
refuses, then the matter should be taken to
court

31. Reducing the Decision Tree
In practice, we do not reduce the decision tree but report expected values on the nodes

32. Resolved Decision Tree
Accepts $2 billion 2
Texaco accepts $5 billion 5
(0.17)
Counteroffer
$5 billion
(0.2)
10.3
Texaco
refuses
counteroffer
Final Court Decision
(0.5)
4.56 5
(0.3)
4.63
(0.50)
(0.2)
10.3
(0.33)
Final Court Decision
(0.5)
Texaco
counteroffers
$3 billion
4.56 5
Refuse
(0.3)
Accepts $3 billion 3

33. Suggested Homework
Problem S2-10, p. 70
Problem S2-13, p. 71