Expected Utility and Risk Aversion

1.1: Preferences 1.2: Risk Premia 1.3: Portfolio Choice 1.
4: Summary
Expected Utility and Risk Aversion
George Pennacchi
University of Illinois
George Pennacchi University of Illinois
Expected utility and risk aversion 1/ 56
1.1: Preferences 1.2: Risk Premia 1.3: Portfolio Choice 1.4: Summary
Introduction
Expected utility is the common framework for modeling investor
choices
We rst analyze conditions individual preferences must satisfy
to be consistent with expected utility functions.
We then consider the link between utility and risk aversion
and risk premia for particular assets.
Finally, we look at how risk aversion aects the choice
between a risky and riskfree asset.
Preferences when Returns are Uncertain
Economists typically analyze the price of a good using supply
and demand. We can do the same for assets.
The main dierentiator between assets is the future payo.
Risky assets have uncertain payos, therefore a theory of
demand for assets must specify investor preferences over
dierent, uncertain payos.
Consider relevant criteria for ranking preferences. One
possible measure is the average payo.
Criterion: Expected Payo
Suppose an asset oers a single random payo at a particular
future date, and this payo has a discrete distribution with n
possible outcomes (x
1
, ..., x
n
) and corresponding probabilities
(p
1
, ..., p
n
), where
n
i =1
p
i
= 1 and p
i
_ 0.
Then the expected value of the payo (or, more simply, the
expected payo) is x = E [x] =
n
i =1
p
i
x
i
.
What is wrong with this?
St. Petersburg Paradox, Nicholas Bernoulli, 1713
Peter continues to toss a coin until it lands heads. He
agrees to give Paul one ducat if he gets heads on the very rst
throw, two ducats if he gets it on the second, four if on the
third, eight if on the fourth, and so on.
If the number of coin ips taken to rst obtain heads is i , then
p
i
=
_
1
2
_
i
and x
i
= 2
i 1
. Thus, Pauls expected payo equals
x =
1
i =1
p
i
x
i
=
1
2
1 +
1
4
2 +
1
8
4 +
1
16
8 + ... (1)
=
1
2
(1 +
1
2
2 +
1
4
4 +
1
8
8 + ...
=
1
2
(1 + 1 + 1 + 1 + ... =
St. Petersburg Paradox
What is the paradox?
Daniel Bernoulli (1738) explained this paradox using expected
utility. His insight was that an individuals utility from
receiving a payo can dier from the size of the payo:
people cared about the expected utility of an assets payos.
Instead of valuing an asset as x =

n
i =1
p
i
x
i
, its value, V,
would be
V = E [U (x)] =

n
i =1
p
i
U
i
where U
i
is the utility associated with payo x
i
.
He hypothesized that U
i
is diminishingly increasing in wealth.
Criterion: Expected Utility
Von Neumann and Morgenstern (1944) derived conditions on
an individuals preferences that, if satised, would make them
consistent with an expected utility function.
Dene a lottery as an asset that has a risky payo and
consider an individuals optimal choice of a lottery from a
given set of dierent lotteries. The possible payos of all
lotteries are contained in the set x
1
, ..., x
n
.
A lottery is characterized by an ordered set of probabilities
P = p
1
, ..., p
n
, where of course,
n
i =1
p
i
= 1 and p
i
_ 0. Let a
dierent lottery be P
= p
1
, ..., p
n
. Let ~, -, and ~
denote preference and indierence between lotteries.
Expected Utility
Specically, if an individual prefers lottery P
to lottery P,
this can be denoted as P
~ P or P - P
.
When the individual is indierent between the two lotteries,
this is written as P
~ P.
If an individual prefers lottery P
to lottery P or she is
indierent between lotteries P
and P, this is written as

P
_ P or P _ P
.
Important: all lotteries have the same payo set x
1
, ..., x
n
,
so we focus on the (dierent) probability sets P and P

Expected Utility Axioms 1-3
There exists an expected utility function V(p
1
, ..., p
n
) i the
following hold:
Axioms:
1) Completeness
For any two lotteries P
and P, either P
~ P, or P
- P, or
P
~ P.
2) Transitivity
If P
_ P
and P
_ P, then P
_ P.
3) Continuity
If P
_ P
_ P, there exists some ` [0, 1] such that

P
~ `P
+ (1 `)P, where `P
+ (1 `)P denotes a
compound lottery; namely, with probability ` one receives the
lottery P
and with probability (1 `) one receives the lottery P.

Expected Utility Axioms 4-5
4) Independence
For any two lotteries P and P
, P
~ P if and only if for all `

(0,1] and all P
:
`P
+ (1 `)P
~ `P + (1 `)P
Moreover, for any two lotteries P and P

y
, P ~ P
y
if and only if for
all ` (0,1] and all P
:
`P + (1 `)P
~ `P
y
+ (1 `)P
5) Dominance
Let P
1
be the compound lottery `
1
P
z
+ (1 `
1
)P
y
and P
2
be the
compound lottery `
2
P
z
+(1 `
2
)P
y
. If P
z
~ P
y
, then P
1
~ P
2
if
and only if `
1
`
2
.
Discussion
The rst three axioms are analogous to those used to establish
a real-valued utility function in consumer choice theory.
Axiom 4 (Independence) is unique, but is necessary for our
denition of probability (being linear) to work.
To understand its meaning, suppose P
~ P. By Axiom 4,
the choices between `P
+ (1 `)P
and `P + (1 `)P
and tossing a coin with probability (1 `) of landing tails,

in which we get P
, and a probability ` of landing heads,

in which case we get P
or P, should be consistent.
Discussion: Machina (1987)
Not so fast:
Consider lotteries over x
1
, x
2
, x
3
= $0, $1m, $5m and two
lottery choices:
C1: P
1
= 0, 1, 0 vs P
2
= .01, .89, .1
C2: P
3
= .9, 0, .1 vs P
4
= .89, .11, 0
Which do you choose in C1? In C2?
Discussion: Machina (1987) contd
Experimental evidence suggests P
1
~ P
2
, P
3
~ P
4
. This
violates Axiom 4:
P
1
~ .11$1m + .89$1m
P
2
~ .11(
1
11
$0m +
10
11
$5m) + .89$1m
P
3
~ .11(
1
11
$0m +
10
11
$5m) + .89$0m
P
4
~ .11($1m) + .89$0m
What now? Expected utility is still the dominant paradigm,
but behavioral nance (CH. 15) considers alternatives.
Deriving Expected Utility: Axiom 1
Given preferences satisfying the preceding axioms, lets show
that the choice between any two (or more) lotteries depends
on which has the higher (highest) expected utility.
Dene an "elementary" or "primitive" lottery, e
i
, which
returns outcome x
i
with probability 1 and all other outcomes
with probability zero, that is, e
i
= p
1
, ...,p
i 1
,p
i
,p
i +1;
...,p
n
= 0, ..., 0, 1, 0, ...0 where p

i
= 1 and p
j
= 0 \j ,= i .
Without loss of generality, assume that the outcomes are
ordered such that e
n
_ e
n1
_ ... _ e
1
. This follows from the
completeness axiom for this case of n elementary lotteries
Deriving Expected Utility: Axiom 3, Axiom 4
From the continuity axiom, for each e
i
, there exists a
U
i
[0, 1] such that
e
i
~ U
i
e
n
+ (1 U
i
)e
1
(2)
and for i = 1, this implies U
1
= 0 and for i = n, this implies
U
n
= 1.
Now a given arbitrary lottery, P = p
1
, ..., p
n
, can be viewed
as a compound lottery over the n elementary lotteries, where
elementary lottery e
i
is obtained with probability p
i
.
P ~ p
1
e
1
+ ... + p
n
e
n
By the independence axiom, and equation (2), the individual
is indierent between lottery, P, and the following lottery:
p
1
e
1
+ ... + p
n
e
n
~ p
1
e
1
+ ... + p
i 1
e
i 1
+ p
i
[U
i
e
n
+ (1 U
i
)e
1
]
+p
i +1
e
i +1
+ ... + p
n
e
n
(3)
where the indierence relation in equation (2) substitutes for
e
i
on the right-hand side of (3).
By repeating this substitution for all i , i = 1, ..., n, the
individual will be indierent between P and
p
1
e
1
+ ... + p
n
e
n
~
_
n
i =1
p
i
U
i
_
e
n
+
_
1
n
i =1
p
i
U
i
_
e
1
(4)
Now dene =
n
i =1
p
i
U
i
. Thus, P ~ e
n
+ (1 )e
1
Similarly, we can show that any other arbitrary lottery
P
= p
1
, ..., p
n
~
e
n
+ (1
)e
1
, where
=
n
i =1
p
i
U
i
.
We know from the dominance axiom that P
~ P i
,
implying
n
i =1
p
i
U
i

n
i =1
p
i
U
i
.
So we can dene the function
V (p
1
, ..., p
n
) =
n
i =1
p
i
U
i
(5)
which implies that P
~ P i V (p
1
, ..., p
n
) V(p
1
, ..., p
n
).
Deriving Expected Utility: The End
The function in (5) is known as von Neumann-Morgenstern
expected utility. It is linear in the probabilities and is unique
up to a linear monotonic transformation.
The intuition for why expected utility is unique up to a linear
transformation comes from equation (2). Here we express
elementary lottery i in terms of the least and most preferred
elementary lotteries. However, other bases for ranking a given
lottery are possible.
For U
i
= U(x
i
), an individuals choice over lotteries is the
same under the transformation aU(x
i
) + b, but not a
nonlinear transformation that changes the shape of U(x
i
).
St. Petersburg Paradox Revisited
Suppose U
i
= U(x
i
) =
_
x
i
. Then the expected utility of the
St. Petersberg payo is
V =
n
i =1
p
i
U
i
=
1
i =1
1
2
i
_
2
i 1
=
1
i =1
2
(
1
2
(i +1)
=
1
i =2
2
i
2
=
1
i =2
_
1
_
2
_
i
= 2
2
2
+ 2
3
2
+ ...
=
1
i =0
_
1
_
2
_
i
1
1
_
2
=
1
1
1
p
2
1
1
_
2
=
1
2
_
2
~
= 1.707
A certain payment of 1.707
2
~
= 2.914 ducats has the same
expected utility as playing the St. Petersburg game.
St. Petersburg 2: Petersburg Harder
The St. Petersburg game has innite expected payo because
the probability of winning declines at rate 2
i
, while the
winning payo increases at rate 2
i
.
In a super St. Petersberg paradox, we can make the
winning payo increase at a rate x
i
= U
1
(2
i 1
) to cause
expected utility to increase at 2
i
. For square-root utility,
x
i
= (2
i
2)
2
= 2
2i 2
; that is, x
1
= 1, x
2
= 4, x
3
= 16, and so
on. The expected utility of super St. Petersburg is
V =
n
i =1
p
i
U
i
=
1
i =1
1
2
i
_
2
2i 2
=
1
i =1
1
2
i
2
i 1
= (6)
Should we be concerned that if prizes grow quickly enough,
we can get innite expected utility (and valuations) for any
chosen form of expected utility function?
Von Neumann-Morgenstern Utility
The von Neumann-Morgenstern expected utility can be
generalized to a continuum of outcomes and lotteries with
continuous probability distributions. Analogous to equation
(5) is
V (F) = E [U (x)] =
_
U (x) dF (x) =
_
U (x) f (x) dx (7)
where F (x) is the lotterys cumulative distribution function
over the payos, x. V can be written in terms of the
probability density, f (x), when x is absolutely continuous.
This is analogous to our previous lottery represented by the
discrete probabilities P = p
1
, ..., p
n
.
Risk Aversion
Diminishing marginal utility results in risk aversion: being
unwilling to accept a fair lottery. Why?
Let there be a lottery that has a random payo, -, where
- =
_
-
1
with probability p
-
2
with probability 1 p
(8)
The requirement that it be a fair lottery restricts its expected
value to equal zero:
E [-] = p-
1
+ (1 p)-
2
= 0 (9)
which implies -
1
,-
2
= (1 p) ,p, or solving for p,
p = -
2
, (-
1
-
2
). Since 0 < p < 1, -
1
and -
2
are of
opposite signs.
Risk Aversion and Concave Utility
Suppose a vN-M maximizer with current wealth W is oered
a fair lottery. Would he accept it?
With the lottery, expected utility is E [U (W +-)]. Without
it, expected utility is E [U (W)] = U (W). Rejecting it implies
U (W) E [U (W +-)] = pU (W + -
1
) +(1 p)U (W + -
2
)
(10)
U (W) can be written as
U(W) = U (W + p-
1
+ (1 p)-
2
) (11)
Substituting into (10), we have
U (W + p-
1
+ (1 p)-
2
) pU (W + -
1
)+(1p)U (W + -
2
)
(12)
which is the denition of U being a concave function.
Risk Aversion = Concavity
A function is concave if a line joining any two points lies
entirely below the function. When U(W) is a continuous,
second dierentiable function, concavity implies U
00
(W) < 0.
Risk Aversion = Concavity
To show that concave utility implies rejecting a fair lottery, we
can use Jensens inequality which says that for concave U()
E[U(~ x)] < U(E[~ x]) (13)
Therefore, substituting ~ x = W +- with E[-] = 0, we have
E [U(W +-)] < U (E [W +-]) = U(W) (14)
which is the desired result.
Risk Aversion and Risk Premium
How might aversion to risk be quantied? One way is to
dene a risk premium as the amount that an individual is
willing to pay to avoid a risk.
Let denote the individuals risk premium for a lottery, -.
is the maximum insurance payment an individual would pay to
avoid the lottery risk:
U(W ) = E [U(W +-)] (15)
W is dened as the certainty equivalent level of wealth
associated with the lottery, -.
For concave utility, Jensens inequality implies 0 when - is
fair: the individual would accept wealth lower than her
expected wealth following the lottery, E [W +-], to avoid the
lottery
Risk Premium
For small - we can take a Taylor approximation of equation
(15) around - = 0 and = 0.
Expanding the left-hand side about = 0 gives
U(W )
~
= U(W) U
0
(W) (16)
and expanding the right-hand side about - gives
E [U(W +-)]
~
= E
_
U(W) +-U
0
(W) +
1
2
-
2
U
00
(W)
(17)
= U(W) + 0 +
1
2
o
2
U
00
(W)
where o
2
= E
_
-
2
is the lotterys variance.

Risk Premium contd
Equating the results in (16) and (17) gives
=
1
2
o
2
U
00
(W)
U
0
(W)
=
1
2
o
2
R(W) (18)
where R(W) = U
00
(W),U
0
(W) is the Pratt (1964)-Arrow
(1971) measure of absolute risk aversion.
Since o
2
0, U
0
(W) 0, and U
00
(W) < 0, concavity of the
utility function ensures that must be positive
An individual may be very risk averse (U
00
(W) is large), but
may be unwilling to pay a large risk premium if he is poor
since his marginal utility U
0
(W) is high.
U
//
(W) and U
/
(W)
Consider the following negative exponential utility function:
U(W) = e
bW
, b 0 (19)
Note that U
0
(W) = be
bW
0 and
U
00
(W) = b
2
e
bW
< 0.
Consider the behavior of a very wealthy individual whose
wealth approaches innity
lim
W !1
U
0
(W) = lim
W !1
U
00
(W) = 0 (20)
Theres no concavity, so is there no risk aversion?
R(W) =
b
2
e
bW
be
bW
= b (21)
Absolute Risk Aversion: Dollar Payment for Risk
We see that negative exponential utility, U(W) = e
bW
,
has constant absolute risk aversion.
If, instead, we want absolute risk aversion to decline in wealth,
a necessary condition is that the utility function must have a
positive third derivative:
0R(W)
0W
=
0
U
00
(W )
U
0
(W )
0W
=
U
000
(W)U
0
(W) [U
00
(W)]
2
[U
0
(W)]
2
(22)
R(W) =U(W)
The coecient of risk aversion contains all relevant
information about the individuals risk preferences. Note that
R(W) =
U
00
(W)
U
0
(W)
=
0 (ln [U
0
(W)])
0W
(23)
Integrating both sides of (23), we have
_
R(W)dW = ln[U
0
(W)] + c
1
(24)
where c
1
is an arbitrary constant. Taking the exponential
function of (24) gives
e
R
R(W )dW
= U
0
(W)e
c
1
(25)
Its beginning to look like we can recover U(W)
R(W) =U(W) contd
Integrating once again, we obtain
_
e
R
R(W )dW
dW = e
c
1
U(W) + c
2
(26)
where c
2
is another arbitrary constant.
Because vN-M expected utility functions are unique up to a
linear transformation, e
c
1
U(W) + c
2
reects the same risk
preferences as U(W).
Relative Risk Aversion: Percentage Payment for Risk
Relative risk aversion is another frequently used measure of
risk aversion and is dened simply as
R
r
(W) = WR(W) (27)
It describes demand for risk as a fraction of wealth.
Let us now examine the coecients of risk aversion for some
utility functions that are frequently used in models of portfolio
choice and asset pricing. Power utility can be written as
U(W) =
1
, < 1 (28)
implying that R(W) =
(1)W
2
W
1
=
(1)
W
and, therefore,
R
r
(W) = 1 .
Hence, this form of utility is also known as constant relative
risk aversion.
Logarithmic Utility: Constant Relative Risk Aversion
Logarithmic utility is a limiting case of power utility. Since
utility functions are unique up to a linear transformation, write
the power utility function as
1
=
W
Next take its limit as 0. Do so by rewriting the

numerator and applying LHpitals rule:
lim
!0
W
= lim
!0
e
ln(W )
1
= lim
!0
ln(W)W
1
= ln(W)
(29)
Thus, logarithmic utility is equivalent to power utility with
= 0, or a coecient of relative risk aversion of unity:
R(W) =
W
2
W
1
=
1
W
and R
r
(W) = 1.
HARA: Power, Log, Quadratic
Hyperbolic absolute-risk-aversion (HARA) utility is a
generalization of all of the aforementioned utility functions. It
can be written as
U(W) =
1
_
cW
1
+ ,
_
(30)
s.t. ,= 1, c 0,
W
1
+ , 0, and , = 1 if = .
Thus, R(W) =
_
W
1
+

_
1
. Since R(W) must be 0, it
implies , 0 when 1. R
r
(W) = W
_
W
1
+

_
1
.
HARA utility nests constant absolute risk aversion ( = ,
, = 1), constant relative risk aversion ( < 1, , = 0), and
quadratic ( = 2) utility functions.
Another Look at the Risk Premium
A premium to avoid risk is ne for insurance, but we may also
be interested in a premium to bear risk.
This alternative concept of a risk premium was used by Arrow
(1971), identical to the earlier one by Pratt (1964).
Suppose that a fair lottery -, has the following payos and
probabilities:
- =
_
+c with probability
1
2
c with probability
1
2
(31)
How much do we need to deviate from fairness to make a
risk-averse individual indierent to this lottery?
Risk Premium v2
Lets dene a risk premium, 0, in terms of probability of
winning p:
0 = Prob(win) Prob(lose) = p (1 p) = 2p 1 (32)
Therefore, from (32) we have
Prob(win) = p =
1
2
(1 + 0)
Prob(lose) = 1 p =
1
2
(1 0)
We want 0 that equalizes the utilities of taking and not taking
the lottery:
U(W) =
1
2
(1 + 0)U(W + c) +
1
2
(1 0)U(W c) (33)
Risk Aversion (again)
Lets again take a Taylor approximation of the right side,
around c = 0
U(W) =
1
2
(1 + 0)
_
U(W) + cU
0
(W) +
1
2
c
2
U
00
(W)
(34)
+
1
2
(1 0)
_
U(W) cU
0
(W) +
1
2
c
2
U
00
(W)
= U(W) + c0U
0
(W) +
1
2
c
2
U
00
(W)
Rearranging (34) implies
0 =
1
2
cR(W) (35)
which, as before, is a function of the coecient of absolute
risk aversion.
Risk Aversion (again)
Note that the Arrow premium, 0, is in terms of a probability,
while the Pratt measure, , is in units of a monetary payment.
If we multiply 0 by the monetary payment received, c, then
equation (35) becomes
c0 =
1
2
c
2
R(W) (36)
Since c
2
is the variance of the random payo, -, equation (36)
shows that the Pratt and Arrow risk premia are equivalent.
Both were obtained as a linearization of the true function
around - = 0.
Simple Beginning
Lets consider the relation between risk aversion and an
individuals portfolio choice in a single period context.
Assume there is a riskless security that pays a rate of return
equal to r
f
and just one risky security that pays a stochastic
rate of return equal to r .
Also, let W
0
be the individuals initial wealth, and let A be the
dollar amount that the individual invests in the risky asset at
the beginning of the period. Thus, W
0
A is the initial
investment in the riskless security.
Denoting the individuals end-of-period wealth as
~
W, it
satises
~
W = (W
0
A)(1 + r
f
) + A(1 +~r ) (37)
= W
0
(1 + r
f
) + A(~r r
f
)
Single Period Utility Maximization
A vN-M expected utility maximizer chooses her portfolio by
maximizing the expected utility of end-of-period wealth:
max
A
E[U(
~
W)] = max
A
E [U (W
0
(1 + r
f
) + A(~r r
f
))] (38)
Maximization satises the rst-order condition wrt. A:
E
_
U
0
_
~
W
_
(~r r
f
)
_
= 0 (39)
What about second order condition? <2-
E
_
U
00
_
~
W
_
(~r r
f
)
2
_
_ 0 (40)
is satised because U
00
_
~
W
_
_ 0 from concavity.
Obtaining A
+
from FOC
If E[~r r
f
] = 0, i.e., E [~r ] = r
f
, then we can show A=0
satises FOC.
When A=0,
~
W = W
0
(1 + r
f
) and, therefore,
U
0
_
~
W
_
= U
0
(W
0
(1 + r
f
)) are nonstochastic. Hence,
E
_
U
0
_
~
W
_
(~r r
f
)
_
= U
0
(W
0
(1 + r
f
)) E[~r r
f
] = 0.
So what about E[~r r
f
] 0?
A = 0 would not satisfy the rst-order condition, because
E
_
U
0
_
~
W
_
(~r r
f
)
_
= U
0
(W
0
(1 + r
f
)) E[~r r
f
] 0 when
A = 0.
When E[~r ] r
f
0, the FOC is satised only when A 0.
Why must A > 0?
Let r
h
denote a realization of ~r r
f
, and let W
h
be the
corresponding level of
~
W
Also, let r
l
denote a realization of ~r < r
f
, and let W
l
be the
corresponding level of
~
W.
Then U
0
(W
h
)(r
h
r
f
) 0 and U
0
(W
l
)(r
l
r
f
) < 0.
For U
0
_
~
W
_
(~r r
f
) to average to zero for all realizations of
~r , it must be that W
h
W
l
so that U
0
_
W
h
_
< U
0
_
W
l
_
due
to the concavity of the utility function.
Why? Since E[~r ] r
f
0, the average r
h
is farther above r
f
than the average r
l
is below r
f
. To preserve FOC, the
multipliers must satisfy U
0
_
W
h
_
< U
0
_
W
l
_
to compensate,
which occurs when W
h
W
l
and which requires that A 0.
How does A change wrt W
0
?
Well use implicit dierentiation to obtain
dA(W
0
)
dW
0
:
Dene f (A, W
0
) = E
_
U
_
W
__
and let
v (W
0
) = max
A
f (A, W
0
) be the maximized value of expected
utility when A, is optimally chosen.
Also dene A(W
0
) as the value of A that maximizes f for a
given value of the initial wealth parameter W
0
.
Now take the total derivative of v (W
0
) with respect to W
0
by
applying the chain rule:
dv(W
0
)
dW
0
=
@f (A;W
0
)
@A
dA(W
0
)
dW
0
+
@f (A(W
0
);W
0
)
@W
0
.
However,
@f (A;W
0
)
@A
= 0 since it is the rst-order condition for a
maximum.
0
contd
The total derivative simplies to
dv(W
0
)
dW
0
=
@f (A(W
0
);W
0
)
@W
0
.
Thus, the derivative of the maximized value of the objective
function with respect to a parameter is just the partial
derivative with respect to that parameter.
Second, consider how the optimal value of the control
variable, A(W
0
), changes when the parameter W
0
changes.
We can derive this relationship by taking the total derivative
of the FOC 0f (A(W
0
) , W
0
) ,0A = 0 with respect to W
0
:
@(@f (A(W
0
);W
0
)=@A)
@W
0
= 0 =
@
2
f (A(W
0
);W
0
)
@A
2
dA(W
0
)
dW
0
+
@
2
f (A(W
0
);W
0
)
@A@W
0
0
contd
Rearranging the above gives us
dA(W
0
)
dW
0
=
0
2
f (A(W
0
) , W
0
)
0A0W
0
,
0
2
f (A(W
0
) , W
0
)
0A
2
(41)
We can then evaluate it to obtain
dA
dW
0
=
(1 + r
f
)E
_
U
00
(
~
W)(~r r
f
)
_
E
_
U
00
(
~
W)(~r r
f
)
2
_
(42)
The denominator of (42) is positive because of concavity.
Therefore, the sign of
dA
dW
0
depends on the numerator.
Implications for
dA
dW
0
with DARA
Consider the case of an individual with absolute risk aversion
that is decreasing in wealth. This implies
R
_
W
h
_
6 R (W
0
(1 + r
f
)) (43)
where, as before, R(W) = U
00
(W),U
0
(W).
Multiplying both terms of (43) by U
0
(W
h
)(r
h
r
f
), which is
a negative quantity, the inequality sign changes:
U
00
(W
h
)(r
h
r
f
) > U
0
(W
h
)(r
h
r
f
)R (W
0
(1 + r
f
)) (44)
Then for A > 0, we have W
l
6 W
0
(1 + r
f
). If absolute risk
aversion is decreasing in wealth, this implies
R(W
l
) > R (W
0
(1 + r
f
)) (45)
Implications for
dA
dW
0
with DARA
Multiplying (45) by U
0
(W
l
)(r
l
r
f
), which is positive, so
that the sign of (45) remains the same, we obtain
U
00
(W
l
)(r
l
r
f
) > U
0
(W
l
)(r
l
r
f
)R (W
0
(1 + r
f
)) (46)
Inequalities (44) and (46) are the same whether the
realization is ~r = r
h
or ~r = r
l
.
Therefore, if we take expectations over all realizations of ~r , we
obtain
E
_
U
00
(
~
W)(~r r
f
)
_
> E
_
U
0
(
~
W)(~r r
f
)
_
R (W
0
(1 + r
f
))
(47)
The rst term on the right-hand side is just the FOC.
Implications for risk-taking with ARA/RRA
Inequality (47) reduces to
E
_
U
00
(
~
W)(~r r
f
)
_
> 0 (48)
Thus, DARA = dA,dW
0
0: amount invested A increases in
initial wealth.
What about the proportion of initial wealth? To analyze this,
dene
j =
dA
dW
0
A
W
0
=
dA
dW
0
W
0
A
(49)
which is the elasticity measuring the proportional increase in
the risky asset for an increase in initial wealth.
Implications for risk-taking with RRA
Adding 1
A
A
to the right-hand side of (49) gives
j = 1 +
(dA,dW
0
)W
0
A
A
(50)
Substituting dA,dW
0
from equation (42), we have
j = 1+
W
0
(1 + r
f
)E
_
U
00
(
~
W)(~r r
f
)
_
+ AE
_
U
00
(
~
W)(~r r
f
)
2
_
AE
_
U
00
(
~
W)(~r r
f
)
2
_
(51)
Collecting terms in U
00
(
~
W)(~r r
f
), this can be rewritten as
Implications for risk-taking with RRA
j = 1 +
E
_
U
00
(
~
W)(~r r
f
)W
0
(1 + r
f
) + A(~r r
f
)
_
AE
_
U
00
(
~
W)(~r r
f
)
2
_
(52)
= 1 +
E
_
U
00
(
~
W)(~r r
f
)
~
W
_
AE
_
U
00
(
~
W)(~r r
f
)
2
_
(53)
The denominator in (53) is positive for A 0 by concavity.
Therefore, j 1 so that the individual invests proportionally
more in the risky asset with an increase in wealth if
E
_
U
00
(
~
W)(~r r
f
)
~
W
_
> 0.
Can we relate this to the individuals risk aversion?
Implications for risk-taking with DRRA
Consider an individual whose relative risk aversion is
decreasing in wealth.
Then for A > 0, we again have W
h
> W
0
(1 + r
f
). When
R
r
(W) = WR(W) is decreasing in wealth, this implies
W
h
R(W
h
) 6 W
0
(1 + r
f
)R (W
0
(1 + r
f
)) (54)
Multiplying both terms of (54) by U
0
(W
h
)(r
h
r
f
), which is
a negative quantity, the inequality sign changes:
W
h
U
00
(W
h
)(r
h
r
f
) > U
0
(W
h
)(r
h
r
f
)W
0
(1+r
f
)R (W
0
(1 + r
f
))
(55)
For A > 0, we have W
l
6 W
0
(1 + r
f
). If relative risk aversion
is decreasing in wealth, this implies
W
l
R(W
l
) > W
0
(1 + r
f
)R (W
0
(1 + r
f
)) (56)
Multiplying (56) by U
0
(W
l
)(r
l
r
f
), which is positive, so
that the sign of (56) remains the same, we obtain
W
l
U
00
(W
l
)(r
l
r
f
) > U
0
(W
l
)(r
l
r
f
)W
0
(1+r
f
)R (W
0
(1 + r
f
))
(57)
Inequalities (55) and (57) are the same whether the
realization is ~r = r
h
or ~r = r
l
.
Therefore, if we take expectations over all realizations of ~r , we
obtain
E
_
~
WU
00
(
~
W)(~r r
f
)
_
> E
_
U
0
(
~
W)(~r r
f
)
_
W
0
(1+r
f
)R(W
0
(1+r
f
))
(58)
The rst term on the right-hand side is just the FOC, so
inequality (58) reduces to
E
_
~
WU
00
(
~
W)(~r r
f
)
_
> 0 (59)
Hence, decreasing relative risk aversion implies j 1 so an
individual invests proportionally more in the risky asset as
wealth increases.
The opposite is true for increasing relative risk aversion: j < 1
so that this individual invests proportionally less in the risky
asset as wealth increases.
Risk-taking with ARA/RRA
This sections main results:
Risk Aversion Investment Behavior
Decreasing Absolute
@A
@W
0
0
Constant Absolute
@A
@W
0
= 0
Increasing Absolute
@A
@W
0
< 0
Decreasing Relative
@A
@W
0

A
W
0
Constant Relative
@A
@W
0
=
A
W
0
Increasing Relative
@A
@W
0
<
A
W
0
Summary
We now know
Why expected utility is useful
What conditions preferences must satisfy to be represented
using vN-M expected utility functions
How the specication of U(W) aects risk aversion
How the ARA/RRA aect the choice between risky and
risk-free assets.

Expected Utility and Risk Aversion

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Expected Utility and Risk Aversion

Uploaded by

Copyright:

Available Formats

1.1: Preferences 1.2: Risk Premia 1.3: Portfolio Choice 1.

and P, this is written as

George Pennacchi University of Illinois

_ P, there exists some ` [0, 1] such that

and with probability (1 `) one receives the lottery P.

~ P if and only if for all `

Moreover, for any two lotteries P and P

and tossing a coin with probability (1 `) of landing tails,

, and a probability ` of landing heads,

= 0, ..., 0, 1, 0, ...0 where p

is the lotterys variance.

Next take its limit as 0. Do so by rewriting the

You might also like