Lecture 06 Utility

• Introduction and Meaning of Utility

• Utility Function
• Attitude to Risk: Risk Neutral, Risk Avoiding, or Risk Seeking
• Expected Utility Approach to Decision Making
Expected Monetary Value Criterion?
Using EMV criterion, should buy collision damage waiver insurance for a car?
Insurance cost = $10 per year.

Accident No Accident
(Probability=0.001) (Probability=0.999)
Buy Insurance $10 $10
Don’t Buy Insurance $5,000 $0

EMV=(5,000)(0.001) +(0)(0.999)=5
5<10
Not purchase the insurance

Why people often purchase insurance? This is called risk aversion. 2

 What is wrong with EMV criterion?
• Flaw of average
1) If the decision was repeated for many times, the EMV decision would maximize
total payoff.
2) The EMV criterion may not be appropriate if the decision is a one-time
opportunity with substantial risks.

• For the most part, the pain of a loss outweighs the pleasure of a gain.

3
 Utility Theory

• Utility is a measure of the total worth or relative desirability of a particular outcome;

it reflects the decision maker’s attitude toward a collection of factors such as profit,
loss, and risk. (Anderson et al. Quantitative Methods for Business.)

• Utility is especially relevant where payoffs can be extremely high or low

• Utility function U(M)=u where payoff is M, and utility value is u

Different decision makers may have different utility functions.

Utility functions should be increasing in payoff.

4
Certainty Equivalents Method to
Develop a Utility Function
Lottery

p Payoff X
ve1
ti
e rna Payoff Y
Alt 1-p
Alte
rnati
ve 2 Payoff Z
Sure

p, X and Y are given.

The decision maker can choose Z that makes him/her indifferent between Z for
sure and the lottery.
5
Example. Given only the maximum
payoff and minimum payoff.
Step 1. Highest payoff 5000: U(5000) = 1
Lowest payoff -2000: U(-2000) = 0

Payoff Utility
5000 1

-2000 0

6
Step 2. Suppose p=0.5.

0.5 5000
Payoff Utility e1
Payoff Utility
tiv
a 5000 1
5000 1 e rn -2000
Alt 0.5
0 0.5
Alte -2000 0
-2000 0 rnati
ve 2 Payoff Z
Sure

What value of would you take?

If
0.5*U(5000)+0.5*U(-2000)=0.5*1+0.5*0=0.5
7
 0.5 5000 Payoff Utility
Payoff Utility tive1 5000 1
rn a
5000 1 e 0 2000 0.75
Alt 0.5
0 0.5 0 0.5
Alte
-2000 0 rnati -2000 0
ve 2 Payoff Z
Sure

What value of would you take?

Suppose that you reply

0.5*U(5000)+0.5*U(0)=0.5*1+0.5*0.5=0.75
8
 0.5 5000 Payoff Utility
Payoff Utility
ve1 5000 1
5000 1 ati
rn 2000
Alt
e
0.5 3000 0.875
2000 0.75
2000 0.75
0 0.5 Alte
rnati
ve 2 Payoff Z 0 0.5
-2000 0 Sure -2000 0

What value of would you take?

Suppose that you reply

0.5*U(5000)+0.5*U(2000)=0.5*1+0.5*0.75=0.875
9
Function Curve

10
Equivalent Lottery Method to
Develop a Utility Function
Gamble

p Payoff X
ve1
ti
e rna Payoff Y
Alt 1-p
Alte
rnati
ve 2 Payoff Z
Sure

X, Y, and Z are given.

The decision maker can choose p that makes him/her indifferent between Z for
sure and the gamble.
Then, U(Z)=p. 11
 Example. Given a set of payoffs.
Suppose that we want to Step 1. Highest payoff 5000: U(5000) = 1
determine utility values at Lowest payoff -2000: U(-2000) = 0
the following payoffs.
Payoff Utility Payoff Utility
5000 5000 1
3000 3000
2000 2000
1000 1000
0 0
-2000 -2000 0

12
Step 2. Go through the other values one by one.
Play a gamble.

p 5000
Payoff Utility e1
Payoff Utility
tiv
5000 1 rn a 5000 1
Alt
e
1-p -2000
3000 3000 0.6
Alte 2000
2000 rnati
ve 2 3000
1000 Sure 1000
0 0
-2000 0 -2000 0
What value of would you take?

Suppose that you reply

U(3000)=0.6
13
 p 5000
Payoff Utility e1
Payoff Utility
tiv
5000 1 rn a 5000 1
Alt
e
1-p -2000
3000 0.6 3000 0.6
2000 Alte 2000 0.5
rnati
ve 2 2000
1000 Sure 1000
0 0
-2000 0 -2000 0
What value of would you take?

Suppose that you reply

U(2000)=0.5
14
 Exponential Utility Function
where the value of is set by the decision-maker
• is the decision maker’s risk tolerance
• Fit a risk-averse individual
• M and R have the same sign
• A small value of means a great aversion to risk
would cause the curve to bend sharply.

Utility 1.00
Payoff R=5000 R=1000 0.80 R=1000
5000 0.63 0.99 0.60 R=5000
3000 0.45 0.95
1000 0.18 0.63 0.40
500 0.10 0.39 0.20
20 0.00 0.02 0.00
0 0 0 0 0 0 0 0 0 0 0
50 10
0
15
0
20
0
25
0
30
0
35
0
40
0
45
0
50
0 15
Three Typical Risk Attitudes
Example.
• A Risk neutral
• A Risk avoider
Option #1 50% to get $5000 and 50% to get 0.
• A Risk taker
EMV=5000*0.5+0*0.5=2500
Option #2 100% to get $2,500

risk neutral: #1 or #2

risk avoider: #2

risk taker: #1
16
 Risk Neutral Decision
Maker
Makes choice based on expected payoff

p X
ve1
ti
rna
Alt
e
1-p Y
Alte
rnati
ve 2 Payoff Z
Sure
Given p, X, and Y, what value of would you take?
From its definition, Z=EMV=pX+(1-p)Y
U(Alternative 1)= pU(X)+(1-p)*U(Y)
U(Alternative 2)= U(Z)
U(Z)=pU(X)+(1-p)*U(Y)
17
 Utility Function of Risk Neutral Decision Maker
The utility function is always a linear function.

Utility
Utility
U(X) (maximum payoff, 1)
1
?

U(Y)

0
Y X Minimum Maximum
Payoffs Payoffs

18
 Utility Function of Risk Neutral Decision Maker
Example

Utility
U(X)
U($100)=60 and U($0)=10 for a player.
? This player is risk neutral.
What is U($40)?
U(Y)

Solution. Y=0. X=100. U(0)=10. U(100)=60.

x=40.
Y X Payoffs
Linear function of linking two points:
U($40)=30

19
 A Risk avoider
A Risk avoider has a concave utility function (decreasing marginal value)
Prefers a guaranteed payoff versus a gamble with the same expected payoff

p X
Utility
ve1
ati
rn
Alt
e
1-p Y U(X)
U(EMV)
Alte
rnati
ve 2 EMV=pX+(1-p)Y
Guaranteed U(Alternative 1)
U(Y)

Which alternative do you prefer?

Risk avoider: Alternative 2. Y EMV X Payoffs
U(EMV)>U(Alternative 1)= pU(X)+(1-p)*U(Y)
20
 Utility Function of A Risk Avoider
Utility Utility

U(X) Concave (maximum payoff, 1)

U(EMV)
1

U(Alternative 1)
U(Y)

0
Y EMV X Payoffs Minimum Maximum
Payoffs

21
 A Risk Taker
A Risk taker has a convex utility function (increasing marginal value)
Prefers a gamble with the same expected payoff than a guaranteed
payoff of the same amount

p X
Utility
ve1
ati
rn
Alt
e
1-p Y U(X)

Alte
rnati
ve 2 EMV=pX+(1-p)Y
Guaranteed U(Alternative 1)
U(Y) U(EMV)

Which alternative do you prefer?

Risk taker: Alternative 1.
Y EMV X Payoffs
U(EMV)<U(Alternative 1)= pU(X)+(1-p)*U(Y)
22
 Utility Function of A Risk Taker
Utility Utility

U(X) Convex (maximum payoff, 1)

1

U(Alternative 1)
U(Y)
U(EMV)

0
Y EMV X Payoffs Minimum Maximum
Payoffs

23
 Risk Attitudes
Summary
• A Risk neutral decision maker has a
linear utility function. Utility
• Makes choice based on expected payoff (max payoff,1)
risk avoider
• A Risk avoider has a concave utility
function (decreasing marginal value)
• Prefers a guaranteed payoff versus a risk
neutral
gamble with the same expected payoff
risk taker
• A Risk taker has a convex utility
function (increasing marginal value)
• Prefers a gamble with the same
0
expected payoff than a guaranteed Payoffs
payoff of the same amount
24
Risk Behaviour

• Most individuals are risk avoiders for payoffs in some range, risk neutral for some
other range, and risk takers for yet another range of payoff
• This explains why same individual will purchase both insurance and a lottery ticket
Utility

risk avoider Risk avoider Risk taker

Accident Lotto Max

(0.001) $5
risk
neutral Insurance $10
risk taker Don’t Buy $5,000

25
Payoffs
Example
A decision maker is indifferent between a sure payoff of $40 and a 30% chance
at receiving $100 or 70% chance of receiving $0. If U($100) = 1 and U($0) = 0,
what is U($40)? And are they risk averse, risk neutral, or risk seeking?

Utility
Solution.
U($40)=1*0.3+0*0.7=0.3 (maximum payoff, 1)
EMV=100*0.3+0*0.7=30 1

0.3

0
30 40 100
Payoffs 26
Exercise
In a game, a player a 30% chance at receiving $100 and 70% chance of receiving $0.
Suppose that U($100)=60 and U($0)=10 for this player; this player is risk averse.
For this player, U($30) is impossible to be
A. 22
B. 26
C. 30
D. 35

27
Expected Utility (EU) Approach

• Determine the utility of each possible payoff

• Calculate the Expected Utility(instead of Expected Payoff)
• Chose alternative with highest expected utility

28
Example
Consider this three-state, three-decision problem with the following payoff table.

Payoff (unit:$1,000) S1 S2 S3
D1 100 40 -60
D2 50 20 -30
D3 20 20 -10
probability 0.1 0.3 0.6

If using the EMV criterion, we have

EV(D1) = (100)(0.1) + (40)(0.3) + (-60)(0.6) = -14
EV(D2) = (50)(0.1) + (20)(0.3) + (-30)(0.6) = -7
EV(D3) = (20) (0.1) + (20)(0.3) + (-10)(0.6) = 2
The optimal decision is D3 29
Example (cont.) Decision Makers with Different Utilities
Payoff M Utility
100 1
50 0.94
40 0.9
20 0.8
-10 0.6
-30 0.4
-60 0
Best decision is D3 and expected utility is 0.68.

Payoff (unit:$1,000) S1 S2 S3 Expected utilities

D1 100 40 -60 (1.0)(0.1) + (0.9)(0.3) + (0)(0.6) = 0.37
D2 50 20 -30 (0.94)(0.1) + (0.8)(0.3) + (0.4)(0.6) = 0.574
D3 20 20 -10 (0.8)(0.1) + (0.8)(0.3) + (0.6)(0.6) = 0.68
probability 0.1 0.3 0.6
30
ExercisePayoff
100
M Utility
1
50 0.58
40 0.50
20 0.35
-10 0.18
-30 0.10
-60 0

Payoff (unit:$1,000) S1 S2 S3 Expected utilities

D1 100 40 -60
D2 50 20 -30
D3 20 20 -10
probability 0.1 0.3 0.6

31
Practical Issues with Utility
• Pluses
1) principled and practical model of decision maker preferences
2) can express tolerance for different levels of risk
3) easy to work with once utility function is determined
• Minuses
1) Does not always conform to empirical observations
 behavioural economics has shown that preferences often depend on how a question is asked
2) Difficult to determine the utility function
 lottery method requires decision maker to answer many questions correctly
 may be hard to assess for intangible or highly subjective criteria

32