Problem Set 2

Behavioral Finance
Summer 2012
Prof. Dr. Alexander Klos

Problem 1: Expected Utility and Bayes’ Law

Erwin owns a share in X Inc. and has no further wealth. Having a concave utility function, he
behaves expected utility maximizing. The price of his share strongly depends on the upcoming
annual general meeting. If the CEO announces a favorable earnings development, Erwin will be
able to sell his share for 400 € on the following day. If the CEO announces an adverse earnings
development, Erwin`s share will drop to 100 € only (ignoring transaction costs and assuming for
the sake of simplicity that only these two states may occur). The market (and Erwin, too) estimates
a 50% probability for a favorable announcement and a 50% probability for an adverse
announcement. The current stock market quote is 256 €.
a.) Erwin considers selling his share immediately (at the current quote) or waiting one day and
selling after the general meeting. Can you help him with that decision without knowing
precisely his utility function (just concavity)? If YES, how? If NO, why not?

Some additional definitions and thoughts on expected utility theory:

Def.: Looking at a lottery a, the certainty equivalent CE(a) represents the

safe consequence for which the decision maker is indifferent between CE(a)
and the lottery that has to be evaluated. It hence holds: u(CE(a)) = EU(a).

The difference between the expected value and the certainty equivalent is
called the risk premium.

Nutzen
Utility

1u(a )
2

EU(a)

u(a1)

0
xmin a1 CE(a)
SÄ(a) EV(a)
EW(a) a2 xmax Ausprägung
Consequences

RP(a)

1
 It holds:
Utility function RP(= EV – CE) Risk attitude
linear: uRN =0 Risk neutral
concave: uRS >0 Risk averse
convex: uRF <0 Risk seeking

See Eisenführ/Weber/Langer, 2010, for more details (textbook on decision

analysis available in German and English).

Solution for part a: Concavity means risk aversion, so his certainty equivalent
will be greater than 0. This means that if the Expected Value of any lottery is
not greater than the riskless alternative, he will not participate in the lottery
but choose the riskless option instead. In the present case, the Expected

0.5 ∙ 100€ 0.5 ∙ 400€ 250€

Value of holding the shares is

while the riskless option (selling the shares immediately) leads to a sure gain
of 256€. Since 256€ > 250€, a risk averse investor will decide to sell his
shares before the meeting.

b.) More precisely, assume that Erwin`s utility function on wealth x is √ . Given this
utility function, please confirm your answer to part a.).

0.5 ∙ √100 0.5 ∙ √400 15

√256 16

16 > 15 => he will choose the riskless alternative of selling the shares before
the meeting.

c.) While Erwin is still contemplating parts a.) and b.), a stock analyst and buddy of Erwin
informs him that according to an analyst report, X Inc. will announce a favorable earnings
development. However, Erwin`s longstanding experience tells him that his buddy`s

2
 predictions are not always correct: Indeed, given a favorable earnings development his
buddy correctly predicts a favorable earnings announcement. However, given an adverse
earnings announcement the analyst wrongly predicts a favorable earnings announcement,
with a probability of 1/3, too. Does the new information change Erwin`s answers in parts
a.) and b.)?

Bayesian updating
F: Favorable earnings announcement
A: Adverse earnings announcement
PF: Prediction of favorable earnings announcement
PA: Prediction of adverse earnings announcement

P(F) = 1/2, P(A) = 1/2, P(PF|F) = 1, P(PF|A) = 1/3

P(F|PF) = ?

1 1
! !"|" ∙ ! " 1∙2 3
! "|!" 2
! !"|" ∙ ! " ! !"|$ ∙ ! $ 1 1 1 4 4
1∙2 3∙2
6

The new (a posterior) probability given the favorable earnings projection by

the stock analyst for a favorable announcement is 75%. Thus, the Expected
Value of holding the shares rises to
EV(ℎ ℎ ℎ )=0.25×100€+0.75×400€=325€
Thus, without knowing the precise utility function no statement can be made
concerning Erwin’s preferences.

Given his function from part b.), his expected utility of holding the shares
changes to
EU(ℎ ℎ ℎ )=0.25∙√100+0.75∙ √400=17.5>16,
so he will choose to hold the shares.

Problem 2: Expected Utility and Bayes’ Law (Exam Question SS11)

Suppose that there is a stock with a market price of 100€ which can either go up to 110.25€ or go
down to 90.25€ in the next year. The riskless asset in the market generates a return of 2.01%.
Erwin, an EUT decision maker with a wealth level of 100€ and an utility function of √ , is
indifferent between the two investment opportunities.
a.) What is the probability of the stock going up next year?

p = probability of the stock going up

Final wealth position riskless asset: 102.01 €
Expected Utility has to be the same (indifference) EU(riskless) = EU(stock)

102.01 √102.01 10.1

3
 & '( ) ∙ √110.25 1 * ) ∙ √90.25 ) ∙ 10.5 1 * ) ∙ 9.5

& '( 10.1 , 10.5) 9.5 * 9.5) 10.1

-..-/0.1
,) 0.6
-..1/0.1

b.) An analyst who had the same expectations about the stock as Erwin observed a signal that
the stock will go down in the next year. He uses Bayesian updating to get to an updated
(posterior) probability of the stock going up of 45%. How much trust does the analyst have
in the signal he observed, i.e. what are the likelihoods of that signal, if the probability of
observing the signal at all is 50%? (Please use an a priori probability of the stock going up of
67.5% - this is not the correct answer - if you did not answer part 1.)

Value-definition: U=Up, D=Down, S=Signal, NS=No Signal

P(NS) = 0.5, P(S) = 0.5, P(U) = 0.6, P(D) = 0.4, P(U|S) = 0.45, P(D|S) = 0.55

Bayes:
! &|2 ∙ ! 2 ! 2|& ∙ ! &
! 2|& , ! &|2
! & ! 2

0.55 ∙ 0.5
! &|2 0.6875
0.4
5 6|7 ∙5 7 ..81∙..1
! &| =0.375
5 6 ..9

Problem 3: (µ, σ)-preferences

a.) Show that a quadratic expected utility function implies (µ, σ)-preferences for a risk-averse
decision maker.

;
Consider : -: * ;: with a1>0 and a2>0 (concave utility
function)

The expected value of gamble that can be represented by a random variable

< is given by
: :< - :< * ; : < ;.

;
[Recall from your statistics courses that = > ?> * > @
;
>; * ? > @ ]
;
It therefore holds that AB ≔ < * E B @; F
D?: < ; H * EB
G: ;

Then I ?: < @J - EK * ; AB
; ;
EB , i.e. the expected utility is a
function of the expected value and the variance.

4
 *
- - ;
b.) Calculate the certainty equivalent of a lottery l (1000, 1/2; 0, 1/2) for a (µ, σ)-decision
9,... -88,...,...
maker with the utility function u(x)= at the wealth levels 1,000€ and
10,000€. (Note: The certainty equivalent CE represents the safe consequence for which the
decision maker is indifferent between CE and the lottery that has to be evaluated.)

u(CE)=EU(l)

Lottery l (asset integration!): (2,000, 1/2; 1,000, 1/2)

1 1
∙ 2,000 ∙ 1,000
2 2
1 1 1 1
∙M 2,000 * 2,000; N
2 6,000 144,000,000 2
1 1
∙ 1,000 * 1,000; 0.232638889
6,000 144,000,000

u(x) = 0.232638889

1 1
* ;
0.232638889
6,000 144,000,000
144,000,000
, ;* 33,500,000.016 0
6,000

Solving the quadratic equation (see e.g.

http://en.wikipedia.org/wiki/Quadratic_equation):

144,000,000 144,000,000 ;
* 6,000 * 6,000
* P QR S * 33,500,000.016
-/;
2 2

-/; 12,000 P T 12,000 ; * 33,500,000.016

-/; 12,000 P T110,500,000
-/; 12,000 P 10,511.90
- 1,488.10
; 22,511.90

CE=1,488.10 with asset integration

CE=488.10 without asset integration

Certainty Equivalent (1,000€)≈488.10

Certainty Equivalent (10,000€)≈418.86 (see EXCEL-Sheet)

c.) Are (µ, σ)-preferences “reasonable” from a descriptive point of view?

(µ, σ)-preferences imply increasing absolute risk aversion, i.e. the risk
premium of a lottery increases with wealth (see b.))
5
 Most people do not behave in such a way.

Problem 4: (µ, σ)-preferences and Diversification

a.) Stock A has a return standard deviation of 20% and stock B has a return standard deviation
of 10%. Stock A has an expected return of 20% and stock B has an expected return of 10%.
Plot the efficient portfolio frontier under the assumption that the two stocks returns are
uncorrelated with each other. Show points A and B on the plot. (For these plots, it would
probably be easiest to use a spreadsheet to print the plots.)

See Excel Sheet

b.) Show on the plot the minimum variance portfolio. Does it contain positive quantities of
both A and B?

See Excel Sheet

c.) If the interest rate is 0%, roughly draw the tangency line on the plot. Do you think that the
optimal portfolio will contain positive quantities of both stocks? More of A or more of B?

See Excel Sheet

d.) Now redo a.) to c.) under the assumption that stock B has a high correlation, of 0.9, with
stock A. After answering a.), b.), and c.), describe in words the difference that the
correlation of 0.9 makes.

Diversification benefits at the portfolio level are reduced

e.) As an investor, which world would you rather live in, the world described by a.) to c.), or
the world described by d.)? (Assume that the assets A and B are the only assets that there
are to invest in.) Explain your answer.

Only less attractive portfolios (in terms of mean and variance) are achievable
with a correlation of 0.9.

Problem 5: Expected Utility Theory

Harry is expected utility maximizer. About his utility function u is nothing known but the fact that u
is strictly monotonically increasing and that Harry’s certainty equivalent for the lottery L is 70 €.
0.5 100 €
L=
0.5 50 €
Use the information about Harry’s preferences to conclude which of the following lotteries (a or b)
Harry will prefer. If necessary, draw a picture to support your reasoning.

6
 70 € 90 €
0.6 0.2
a= 0.2 50 € b= 0.5 40 €
0.2 10 € 0.3 0€

We use the independence axiom to substitute the outcome 70€ of lottery a

by L and receive:
30% 100€
a’ ---- 50% 50€
20% 10€

Comparing a’ with b we see that a’ is stochastically dominating (draw risk

profile), this is sufficient for increasing utility functions to conclude that
a~a’>b.