Beruflich Dokumente
Kultur Dokumente
Utility
Hakon Tretvoll
Oce: B4 - 099
Email: hakon.tretvoll@bi.no
(Please put GRA 6543 at the start of the subject)
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 1 / 29
Learning Objectives:
We Want to Understand:
1. What is expected utility: Intuition and formal denition
2. The concept of risk aversion
3. Deriving risk premiums
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 2 / 29
Introduction
Axiom 1: COMPARABILITY
x y, or y x, or x y
Axiom 2: TRANSITIVITY
If x y, and y z, then x z
Axiom 4: MEASURABILITY
If x y z, then there exists a unique
such that y G(x, z; )
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 4 / 29
Axioms of Cardinal Utility
Axiom 5: RANKING
Assume x y z and x u z. There exist
1
and
2
such that y G(x, z;
1
) and u G(x, z;
2
).
If
1
>
2
, then y u
Additional Assumption
More is preferred to less MU(W) > 0
Ranking:
In general:
E[U(W)] =
i
p
i
U(w
i
)
Investors seek to maximize E[U(W)]
Discrete: E[ x] =
N
i =1
p
i
x
i
Continuous: E[ x] =
xp( x)d x
i =1
p
i
(
X
i
+ a) = E[
X] + a
(2) E[a
X] =
N
i =1
p
i
(a
X
i
) = aE[
X]
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 9 / 29
Uncertainty and Risk
Denition
The actuarial value of a gamble is its expected outcome.
Denition
A fair gamble (or fair lottery/bet) has an expected outcome of zero.
Denition
A risk averse investor will not accept a fair gamble.
Examples:
1. Coin toss: gain $1 if heads, lose -$1 if tails
2. Gamble x: gain $16 if heads, gain $4 if tails
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 10 / 29
Uncertainty and Risk (2)
Consider gamble x: investor has $10 and gains or loses $6 with equal
probability (a fair gamble)
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 11 / 29
Uncertainty and Risk (3)
(W) = U(W)/W
(W) < 0
Proof.
The utility from not gambling and keeping your $10 U(10).
This must be greater than the expected utility from the gamble:
U(10) > 0.5 U(16) + 0.5 U(4) or:
U(10) U(4) > U(16) U(10)
We can also prove the concavity in reverse: Concavity implies you will
not accept a fair game
Consider gamble x when you have wealth $10. You have to take it, or
pay to avoid it.
If wealth is reduced to W
0
, then the level of utility is U(W
0
)
Figure 3
U(W
0
) = E[U(W
0
+ x)]: Take a Taylor series approximation of
both sides around the points x = 0 and = 0. Assume higher order
terms are insignicant
LHS:
U(W
0
) U(W
0
) + U
(W
0
) ()
(ignore higher order terms)
RHS:
E[U(W
0
+ x)] E
U(W
0
) + U
(W
0
) x +
U
(W
0
)
2
x
2
Note: E[U(W
0
)] = U(W
0
) this is known and non-random
E(x
2
) =
2
x
because
2
x
E[(x) E(x)]
2
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 18 / 29
Measure of Risk Aversion (3)
(W
0
)
(W
0
) () = U(W
0
) +
2
x
2
U
(W
0
)
Solve for :
=
2
x
2
(W
0
)
U
(W
0
)
Since
2
x
2
> 0, the sign of the RP is determined by the term
U
(W
0
)
U
(W
0
)
(W
0
) > 0 and U
(W
0
) < 0
the RP is always positive
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 19 / 29
Measure of Risk Aversion (4)
Note: the amount you are willing to pay to avoid a fair bet depends
on:
1. riskiness of outcome,
2
x
2. shape of utility function U
(W
0
) and U
(W
0
)
3. initial wealth W
0
(W
0
) is very large, but unwilling to
pay a high premium if you are very poor (U
(W
0
) will also be high).
(W)
U
(W)
2. Relative risk aversion (RRA) - measures RA towards a change in wealth
by a certain fraction: RRA =
U
(W)
U
(W)
W
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 20 / 29
Exercise: Relative risk aversion
A risk averse investor will have a concave utility function over wealth
(see next slide...)
(W) = 0
RN
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 22 / 29
Examples of Utility Functions (2)
Examples (and conditions that must hold for risk averse investor)
Example (CWS):
U = ln(W), W = $20000
Two risks:
1. 50-50 chance of gaining/losing $10;
2. 80% chance of losing $1000, 20% chance of losing $10,000.
(W)
U
(W)
Variance:
2
x
=
i
p
i
(X
i
E(X
i
))
2
= 0.5(20010 20000)
2
+
0.5(19990 20000)
2
= 100
(W) = 1/W, U
(W) = 1/W
2
Thus
U
(W)
U
(W)
=
1
W
=
1
20000
Thus
=
100
2
1
20000
= 0.0025
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 25 / 29
Comparing RA for Small and Large Risks (3)
p
i
U(W
i
)
= 0.5 U(20010) + 0.5 U(19990)
= 9.903487
(W)
U
(W)
2
x
=
i
p
i
(X
i
E(X
i
))
2
= 0.8(19000 17200)
2
+
0.2(10000 17200)
2
= 12, 960, 000
=
12, 960, 000
2
1
20000
= 324
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 27 / 29
Comparing RA for Small and Large Risks (5)
p
i
U(W
i
)
= 0.8 U(19000) + 0.2 U(10000)
= 9.7238
Exercises:
1. Work through the proof that expected utility can be use to rank risky
alternatives (pg. 48, CWS)
2. 3.3, 3.4, 3.5, 3.8, 3.9
Hakon Tretvoll The Theory of Investor Choice Under Uncertainty: Utility 29 / 29