Scribd
Upload
40 views5 pages

Definition and Applications

1. The document discusses the exponential utility function and how it can represent different attitudes towards risk - risk aversion, risk neutrality, and risk preference. 2. It provides formulas for calculating the risk-adjusted value of random variables that follow different probability distributions like binomial, Poisson, normal, and gamma. 3. Examples are given that apply these concepts, such as determining the optimal bid in a bidding competition and analyzing investment decisions for risky projects and stock portfolios.

Uploaded by

Anirban Bardhan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
40 views5 pages

Definition and Applications

1. The document discusses the exponential utility function and how it can represent different attitudes towards risk - risk aversion, risk neutrality, and risk preference. 2. It provides formulas for calculating the risk-adjusted value of random variables that follow different probability distributions like binomial, Poisson, normal, and gamma. 3. Examples are given that apply these concepts, such as determining the optimal bid in a bidding competition and analyzing investment decisions for risky projects and stock portfolios.

Uploaded by

Anirban Bardhan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

Risk Adjusted Value Of a Random Variable

Under Exponential Utility


The exponential utility function of x with risk parameter r has the form
1
U(x) 1 e rx
r

This can represent all three attitudes towards risk:


Case 1. For r > 0, U(x) is strictly concave function and represents aversion towards
risk. Larger r values represent greater risk aversion in the sense that all risk premiums
are larger.
Case 2. For r = 0, U(x) = x. This represents risk neutrality, also known as expectedvalue decision making. [The result U(x) = x can be found as a limit or by series
expansion of the exponential, as shown below.
rx 2 rx 3 rx 4 ......
e rx 1 rx
2!
3!
4!
U(x) x

rx 2 r 2 x 3 r 3 x 4

...... x when r 0
2!
3!
4!

Case 3. For r < 0, U(x) is strictly convex. All risk-adjusted values exceed the
corresponding expected payoffs and hence the function encodes a positive preference
for risk.
The above represented in the form of graphs as shown below:

U(x), r=-0.5

U(x),r0

U(x), r=0.5
8.00
7.00
6.00
5.00

U(x), r=-0.5

U(x),r0

4.00
3.00
2.00

U(x), r=0.5

1.00
0.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50

0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00

0.00
0.27
0.57
0.91
1.30
1.74
2.23
2.80
3.44

0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00

0.00
0.24
0.44
0.63
0.79
0.93
1.06
1.17
1.26

2.25
2.50
2.75
3.00

4.16
4.98
5.91
6.96

2.25
2.50
2.75
3.00

1.35
1.43
1.49
1.55

The risk-adjusted value V of a random payoff X has a simple direct formula in the case of
exponential utility. Let X be the random pay-off from a lottery and let W be the initial wealth.
Then the definition of V is
U(W+V) = E{U(W+X)}
For the exponential utility this is

1
1

1 e r(W V) E 1 e r(W X)
r
r

1 1 rW rV 1 1 rW rX
e e e Ee
r r
r r

Therefore,

e rV E e rX
Algebraic operation results

1
V ln E e rX
r

Solving for V we get


This is a useful formula because it eliminates the need to compute the expected utility first
and then transform it to risk-adjusted value. Furthermore, in this formula risk-adjusted value
V is not a function of initial wealth. Two properties unique to exponential utility function are:
1. It is the only utility function that changes in wealth do not change the risk-adjusted
value assigned to return to lotteries (projects).
2. It is the only utility function such that the risk-adjusted value of a set (portfolio) of
independent lotteries (projects) is the sum of their individual risk-adjusted values.
The exponential utility function is the simplest model of risk aversion. It is not likely to be
entirely satisfactory for every situation; however, it is useful to begin a study with the models
than are simplest, both conceptually and mathematically.
The formula for risk-adjusted value applies to both continuous and discrete random variables.
Special-purpose formulae can be given for important analytic probability distributions.
1. Binomial Probability Density Function
Consider a lottery based upon binomial distribution. The lottery pays Rs. A per success. There
are n trials, each with random probability p of success. The number of successes is the
random variable X, which has the binomial probability function
f X | n, p

n!
p X (1 p) n X ; forX 0,1,2,3,..., n
X! (n - X)!

For this the risk adjusted value is

n
ln 1 - p - pe rA
r

2. Poisson Probability Density Function


Consider next a lottery based upon Poisson probability distribution. Again, the payoff will be
Rs. A per success. The number of successes is the random variable X, which has the Poisson
probability function
X e
f X |
; forX 0,1,2,3,...,
X!
For this the risk adjusted value is

1 - e rA
r

3. Normal Probability Density Function


Next, consider a lottery where the payoff X can be any number, positive or negative, and its
probability density is normal:
X 2 ; for - X
f X 2 2 exp
2 2

For this the risk adjusted value is


V

1 2
r
2

4. Gamma Probability Density Function


Next, consider a lottery where the payoff X cannot be negative and has gamma distribution:
f X | a, b

a b X b-1e aX
; forX 0
(b- 1)!

The risk-adjusted value is

b
r
ln 1
r
a

A BIDDING EXAMPLE (DISCRETE)


Firm A is engaged in a bidding competition with one other Firm B to obtain a large
government contract. The contract is to be awarded solely on the basis of lowest sealed bid
submitted. The Firm A wants to bid Rs. X without knowing Bs bid of Rs. Based on the past
behavior in similar situations, the probability of Bs bid Rs. Y is 0.2 for all values of Y = 2.9,
3.0, 3.1, 3.2, and 3.3 crores (0 probability for other values of Y). The cost of performing the
contract is Rs. 3.0 crores. Firm As bid of Rs. X can be any value; but is best restricted to the
same set of possible values of Y (each one paise lower to resolve ties in As favor). What
should be the optimum value of X when As risk parameter r = 10?
A BIDDING EXAMPLE (CONTINUOUS)
In the past lowest competitors bid of Rs. X on a project can be described by a density
function f(X). Given that Firm As estimated cost is Rs. C, obtain an expression for the
expected profit of a bid of Rs. X by Firm A. Remember that the lowest bidder is awarded the
contract. Given that

2.9 X 3.3
elsewhere

2.5
0

f X

And that C = Rs. 3 crores, what is the optimum bid X of Firm A should submit to maximize
expected profit? Is this bid different when one tries to optimize the risk-adjusted expected
profit where risk parameter r = 10?
ANSWER: Let Rs. X be the competitors bid amount, and Rs. B be the bid amount that Firm
A wants to submit. If this bid wins then the actual profit P(B,X) is given by
X B (Firm A loses)
0
B - C; X B (Firm Awins)

P B, X

Let P(B) be the expected profit of Firm A, then


P B P B, X f X dx
3.3

2.9

0 f X dx
B

3.3

2.9

B - Cf X dx

B - C 2.5 dx 2.5 B - C 3.3 B


3.3

For C = 3, this gives P(B) = 2.5(B-3)(3.3-B) = 2.5{-B2 + 6.3B 9.9}. The optimal value of
dP B
0
dB
B is obtained by solving the equation
; which gives -2B + 6.3 = 0.
That is, Bo = 3.15; P(Bo) = 0.05625; and the probability of winning under the optimum bid is

3.3

2.5dx 2.5 3.3 3.15 0.375

3.15

Now, the risk-adjusted expected profit V(B) where risk parameter r = 10 is given by

1
V B ln
r

e r0 f X dx e r BC f X dx
3.3

2.9

V B 0.1 ln 2.5 B 2.9 e 10 B3 3.3 B

dV B
0
dB

Using EXCEL one can search the optimal value of B0 that makes
and it is found to
be B0 = 3.10737, which is less than 3.15 that maximized the expected profit only; V (B 0) =

0.038126148; and probability of winning under this bid = 2.5(3.3-3.10737) = 0.482.


A RISK SHARING EXAMPLE
Consider a large, risky project with an initial cost of Rs. 10.5 crores. A study of the possible
outcomes indicates that the expected return is Rs. 13 crores and that the probability
distribution of profit is normal with a variance of 20. Should one invest in the project with
risk factor r=0.5? Analyze.

A PORTFOLIO EXAMPLE
Suppose an investor has an amount Rs. C of capital to invest in two common stock
investment opportunities that are not independent in a probability sense. The investment will
be held for 1 year. The unit return on investment in stock i is (1+X i), where Xi is a random
variable. Let c1 and c2 be the amounts to be invested in stocks 1 and 2 respectively. If the
initial capital is not all invested, the remainder is deposited in a bank for one year and will
return 1+I per Re, where I is the riskless interest rate. Suppose, X 1 and X2 are jointly normally
distributed with means 1 and 2 respectively; standard deviations 1 and 2 respectively and
their correlation coefficient is . Formulate the problem of selection of the amounts c 1 and c2
to be selected optimally. Analyze the problem for the following situation:
Stock 1.
Stock 2.
Interest Rate
Risk Aversion Level
Initial Capital

1 = 0.068, 1 = 0.03
2 = 0.056, 2 = 0.02
I =0.04
r = 0.001
C = Rs. 50,000

Note here that the value of has been left unspecified in order to study the effect of
correlation. Note further that one can borrow capital, if required, which carries fixed interest
rate of I = 0.04.

You might also like