CH Pres 02

2.1: Assumptions 2.2: Indi¤erence 2.3: Frontier 2.4: R f 2.5: Hedging 2.
6: Summary
Mean-Variance Analysis
George Pennacchi
University of Illinois
George Pennacchi University of Illinois

Mean-variance analysis 1/ 52
2.1: Assumptions 2.2: Indi¤erence 2.3: Frontier 2.4: R f 2.5: Hedging 2.6: Summary
Introduction
How does one optimally choose among multiple risky assets?
Due to diversi…cation, which depends on assets’return

covariances, the attractiveness of an asset when held in a
portfolio may di¤er from its appeal when it is the sole asset
held by an investor.
Hence, the variance and higher moments of a portfolio need

to be considered.
Portfolios that make the optimal tradeo¤ between portfolio

expected return and variance are mean-variance e¢ cient.

Mean-Variance Utility
What assumptions are needed for investors to care about only

mean and variance (and not skewness, kurtosis...)?
Suppose a utility maximizer invests initial date 0 wealth, W0 ,
in a portfolio.
Let Rep be the gross random return on this portfolio, so that
~ = W0 R
the individual’s end-of-period wealth is W ep .
For short-hand, write U(W ~ ) = U W0 R ep ) since

ep as just U(R
~ is completely determined by R
given W0 , W ep .
Next express U(Rep ) by expanding it around the mean E [R
ep ]:

Taylor Series Approximation of Utility
ep ) = U E [R
U(R ep ] + R
ep e p ] U 0 E [R
E [R ep ]
2
ep
+ 21 R ep ]
E [R ep ] + :::
U 00 E [R
n
1
+ n! ep
R ep ]
E [R ep ] + :::
U (n) E [R (1)
If the utility function is quadratic, (U (n) = 0, 8 n 3), then

the individual’s expected utility is
h i 2
E U(R e p ) = U E [R ep ] + 1 E R e p E [R ep ] ep ]
U 00 E [R
2
ep ] + 1 V [R
= U E [R ep ]U 00 E [R
ep ] (2)
2

Alternative Assumptions for Mean-Variance
Quadratic utility, such as U (W ) = aW bW 2 , is problematic

a
due to a “bliss point” at W = 2b after which utility declines
in wealth.
Instead, let utility be a general increasing and concave
function but restrict the risky asset probability distribution.
Claim: If individual assets are multi-variate normally
distributed, utility of wealth depends only on portfolio mean
and variance.
Why? Note that the return on a portfolio is a weighted
average (sum) of the returns on the individual assets.
Then since sums of normals are normal, if the joint
distributions of individual assets are multivariate normal, then
the portfolio return is also normally distributed.
Centered Normal Moments
Let a random variable, X , be distributed N ; 2 . Its

moment generating function is:
1
m(t) = E (e tX ) = exp t+ 2 2
t (3)
2
Centralized (multiply by exp( t))
1 2 2
cm(t) = exp t (4)
2
Then we have following moments

Centered Normal Moments
1 2t2
ep 1 d exp 2
E [R ] = =0 (5)
dt
t=0
ep d 2 exp 12 2t2
E [R ]2 = = 2
dt 2
t=0
ep d3 exp 21 2t2
E [R ]3 = =0
dt 3
t=0
ep d 4 exp 21 2t2
E [R ]4 = =3 4
dt 4
t=0
:::

Normal Distribution of Returns
Soh moments are either

ni
zero or a function of the variance:
e
E Rp E [Rp ]e = 0 for n odd, and
h ni n=2
E R e p E [R
ep ] n!
= (n=2)! 1 e
2 V [Rp ] for n even.
Therefore, in this case the individual’s expected utility equals
h i 1 2
ep )
E U (R = e p ] + 1 V [R
U E [R ep ]U 00 E [R
ep ] + 0 + ep ]
V [R ep ]
U 0000 E [R
2 8
n=2
1 1 e ep ] + :::
+0 + ::: + V [R p ] U (n) E [R (6)
(n=2)! 2
which depends only on the mean and variance of the portfolio

return.

Caveats
Is a multivariate normal distribution realistic for asset returns?

ep are normally distributed, the gross
If individual assets and R
return will be negative with positive probility because the
normal distribution ranges over the entire real line.
This is a problem since most assets have limited liability,

implying Rep should be non-negative.
Later, in a continuous-time context, we can assume asset

returns are instantaneously normal, which allows them to be
log-normally distributed over …nite intervals.

Preference for Return Mean and Variance

Therefore, assume U is a general utility function and asset
returns are normally distributed. The portfolio return R~ p has
2
normal probability density function f (R; Rp ; p ), where we
de…ne Rp E [R ~ p ] and 2p V [R~ p ].
Expected utility can then be written as
h i Z 1
E U Rp = e U(R)f (R; Rp ; 2p )dR (7)
1
Consider an individual’s indi¤erence curves. De…ne e

x
~p Rp
R
p
,
h i Z 1
E U Rp =e U(Rp + x p )n(x)dx (8)
1
where n(x) f (x; 0; 1). (e

x is a standardized normal)
Mean vs Variance cont’d
Taking the partial derivative with respect to Rp :

h i
@E U R ep Z 1
= U 0 n(x)dx > 0 (9)
@ Rp 1
since U 0 is always greater than zero.

Taking the partial derivative of equation (8) with respect to
2 and using the chain rule:
p
h i h i
ep
@E U R @E U ep
R Z 1
1 1
2
= = U 0 xn(x)dx
@ p 2 p @ p 2 p 1
(10)

Risk and Utility
While U 0 is always positive, x ranges between 1 and +1.

Take the positive and negative pair +xi and xi . Then
n(+xi ) = n( xi ).

Risk and Utility cont’d
Comparing the integrand of equation (10) for equal absolute

realizations of x, we can show
U 0 (Rp + xi p )xi n(xi ) + U 0 (Rp xi p )( xi )n( xi )

0 0
= U (Rp + xi p )xi n(xi ) U (Rp xi p )xi n(xi )
0 0
= xi n(xi ) U (Rp + xi p) U (Rp xi p) <0 (11)
because
U 0 (Rp + xi p) < U 0 (Rp xi p) (12)
due to the assumed concavity of U.

Risk and Utility cont’d

Thus, comparing U 0 xi n(xi ) for each positive and negative pair,
we conclude that
h i
@E U R ep Z 1
1
= U 0 xn(x)dx < 0 (13)
@ 2p 2 p 1
which is intuitive for risk-averse individuals.

An indi¤erence curve is the combinations of Rp ; 2p that
h i
satisfy the equation E U R ep = U , a constant. Higher U
denotes greater utility. Taking the derivative
h i h i
h i @E U R ep @E U R ep
dE U R ep = d 2p + d Rp = 0
@ 2p @ Rp
(14)
Mean and Variance Indi¤erence Curve

h i
ep = 0, we obtain:
Rearranging the terms of dE U R
h i h i
@E U R ep @E U R ep
d Rp
= = >0 (15)
d 2p @ 2p @ Rp
@E [U (Rep )] ep )]
@E [U (R
since we showed @ p2 < 0 and @ Rp
> 0.
Hence, each indi¤erence curve is positively sloped in Rp ; 2p

space. They cannot intersect because since we showed that
utility is increasing in expected portfolio return for a given
level of portfolio standard deviation.

Mean and Standard Deviation Indi¤erence Curve

As an exercise, show that the indi¤erence curve is upward
sloping and convex in Rp ; p space:

Tangency Portfolios
The individual’s optimal choice of portfolio mean and variance

is determined by the point where one of these indi¤erence
curves is tangent to the set of means and standard deviations
for all feasible portfolios, what we might describe as the “risk
versus expected return investment opportunity set.”
This set represents all possible ways of combining various
individual assets to generate alternative combinations of
portfolio mean and variance (or standard deviation).
The set includes ine¢ cient portfolios (those in the interior of
the opportunity set) as well as e¢ cient portfolios (those on
the “frontier” of the set).
How can one determine e¢ cient portfolios?
Mean/Variance Optimization
Given the means and covariances of returns for n individual
assets, …nd the portfolio weights that minimize portfolio
variance for a given portfolio expected return (Merton, 1972).
Let R = (R1 R2 ::: Rn )0 be an n 1 vector of the assets’
expected returns, and let V be the n n covariance matrix
whose i; j th element is ij .
V is assumed to be of full rank. (no redundant assets.)
Next, let ! = (! 1 ! 2 ::: ! n )0 be an n 1 vector of portfolio
weights. Then the expected return on the portfolio is
P
Rp = ! 0 R = ni=1 ! i Ri (16)
and the variance of the portfolio return is
2 0 Pn Pn
p = ! V! = i =1 j =1 ! i ! j ij (17)
Mean/Variance Optimization cont’d
The constraint on portfolio weights is ! 0 e = 1 where e is

de…ned as an n 1 vector of ones.
A frontier portfolio minimizes the portfolio’s variance subject
to the constraints that the portfolio’s expected return equals
R p and the portfolio’s weights sum to one:
min 12 ! 0 V ! + Rp ! 0 R + [1 ! 0 e] (18)
!
The …rst-order conditions with respect to !, , and , are
V! R e = 0 (19)
0
Rp !R = 0 (20)
0
1 !e = 0 (21)


Solving (19) for ! , the portfolio weights are
1 1
! = V R+ V e (22)
Pre-multiplying equation (22) by R 0 and e 0 respectively:

1 1
Rp = R 0! = R 0V R + R 0V e (23)
0 0 1 0 1
1 = e! = eV R+ eV e (24)
Solving equations (23) and (24) for and :
Rp
= 2
(25)
&
& Rp
= 2
(26)
&
Here e 0V 1 R, & R 0V 1 R, and e 0V 1e are scalars.

The denominators & 2 are positive. Since V is positive
1
de…nite, so is V . Therefore, the quadratic form
0
R &e V 1 R &e = 2 & 2 2 & + & 2 = & & 2
is positive.
0
But since & R V 1 R is a positive quadratic form, then
& 2 must also be positive.
Substituting for and in equation (22), we have
Rp 1 & Rp 1
! = 2
V R+ 2
V e (27)
& &


Collecting terms in R p , the portfolio weights are:
! = a + bR p (28)
&V 1e V 1R V 1R V 1e
where a 2 2
and b .
& &
Based on these weights, the minimized portfolio variance for
given R p is
2
p = ! 0 V ! = (a + bR p )0 V (a + bR p ) (29)
2
Rp 2 Rp + &
= 2
&
2
1 Rp
= + 2
&
Mean/Variance Frontier
Equation (29) is a parabola in 2, R p space with its minimum

p
R 0V 1e 1 1
at R p = Rmv = and 2 = .
e 0V 1e mv e 0V 1 e

Substituting R p = into equation (27) and multiplying by

shows that this minimum variance portfolio has weights
! mv = 1 V 1 e = V 1 e= e 0 V 1 e .
An investor whose utility is increasing in expected portfolio
return and is decreasing in portfolio variance would never
choose a portfolio having R p < Rmv .
Hence, the e¢ cient portfolio frontier is represented only by
the region R p Rmv .
Next, let us plot the frontier in p , R p space by taking the
square root of both sides of equation (29):

Asymptotes
s
2
1 Rp
p = + 2
&
which is a hyperbola in p , R p space. Di¤erentiating, this
hyperbola’s slope can be written as
@R p & 2
= p (30)
@ p Rp
The hyperbola’s e¢ cient (ine¢ cient) upper (lower

q ) arc
& 2
asymptotes to the straight line R p = Rmv + p
q
& 2
(R p = Rmv p ).

E¢ cient Frontier

Two Fund Separation

We now state and prove a fundamental result:
Theorem
Every portfolio on the mean-variance frontier can be replicated by
a combination of any two frontier portfolios; and an individual will
be indi¤erent between choosing among the n …nancial assets, or
choosing a combination of just two frontier portfolios.
The implication is that if a security market o¤ered two mutual

or “exchange-traded” funds, each invested in a di¤erent
frontier portfolio, any mean-variance investor could replicate
his optimal portfolio by appropriately dividing his wealth
between only these two funds. (He may have to short one.)
Two Fund Separation: Proof

Proof: Let R1p , R2p and R3p be the expected returns on three
frontier portfolios. Invest a proportion of wealth, x, in
portfolio 1 and the remainder, (1 x), in portfolio 2 such that:
R3p = x R1p + (1 x)R2p (31)
Recall that the weights of frontier portfolios 1 and 2 are
! 1 = a + b R1p and ! 2 = a + b R2p , respectively. Hence, the
investment’s portfolio weights are
x! 1 + (1 x)! 2 = x(a + b R1p ) + (1 x)(a + b R2p()32)
= a + b(x R1p + (1 x)R2p )
= a + b R3p = ! 3
which shows that it is frontier portfolio 3.
Zero Covariance Portfolios

Frontier portfolios have another property. Except for the
minimum variance portfolio, ! mv , for each frontier portfolio
there is another frontier portfolio with which its returns have
zero covariance:
!1 0V !2 = (a + bR 1p )0 V (a + bR 2p ) (33)
1
= + 2
R 1p R 2p
&
Equating this to zero and solving for R 2p in terms of Rmv ,
& 2
R 2p = 2
(34)
R 1p
& 2
= Rmv 2
R 1p Rmv
Zero Covariance cont’d
Note that if R 1p Rmv > 0 so that frontier portfolio ! 1 is

e¢ cient, then by (34) R 2p < Rmv : frontier portfolio ! 2 is
ine¢ cient.
We can determine the relative locations of these zero
covariance portfolios by noting that in p , R p space, a line
tangent to the frontier at the point 1p ; R 1p is of the form
@R p
Rp = R0 + p = 1p p (35)
@ p
@R p
where @ p p = 1p
is the slope of the hyperbola at point
1p ; R 1p and R 0 is the tangent line’s intercept at p = 0.


Using (30) and (29), we can solve for R 0 by evaluating (35)
at the point 1p ; R 1p :
@R p & 2
R 0 = R 1p p = 1p 1p = R 1p 1p 1p
@ p R 1p
" 2#
& 2 1 R 1p
= R 1p + 2
R 1p &
& 2
= 2
(36)
R 1p
= R 2p
The intercept of the line tangent to ! 1 is the expected return
of its zero-covariance counterpart, ! 2 .

E¢ cient Frontier with a Riskless Asset

Assume there is a riskless asset with return Rf (Tobin, 1958).
Now the constraint ! 0 e = 1 does not apply because 1 ! 0 e is
the portfolio proportion invested in the riskless asset. Note
that the portfolio’s expected return equals
Rp = Rf + ! 0 (R Rf e) (37)
The variance of the return on the portfolio is still ! 0 V !.
Thus, the individual’s optimization problem is changed to:
min 12 ! 0 V ! + Rp Rf + ! 0 (R Rf e) (38)
!
Similar to the previous derivation, the solution to the …rst

order conditions is
1
! = V (R Rf e) (39)

p R R R R
Here 0
f
= & 2 pR +f R f2
, and the variance
(R R f e ) V 1 (R R f e) f
of the frontier portfolio in terms of ! is

2 Rp Rf 1
p = ! 0V ! = 0 (R R f e)0 V V
R Rf e V 1 (R R f e)
Rp Rf 1
0 V (R R f e)
R Rf e V 1 (R R f e)
(R p R f )2 (R p R f )2
= 0 = (40)
R Rf e V 1 (R R f e) & 2 R f + R f2
Taking the square root of (40) and rearranging:

1
R p = Rf & 2 Rf + Rf2 2
p (41)
which indicates that the frontier is now linear in p, R p space.

Two Fund Separation: Rf < Rmv
When Rf 6= Rmv , an even stronger separation principle

obtains: any frontier portfolio can be replicated with one
portfolio that is located on the “risky asset only” frontier and
another portfolio that holds only the riskless asset.
Let us prove this result for the case Rf < Rmv . We assert that
1
the e¢ cient frontier line R p = Rf + & 2 Rf + Rf2 2 p
can be replicated by a portfolio consisting of only the riskless
asset and a portfolio on the risky-asset-only frontier that is
determined by a straight line tangent to this frontier whose
intercept is Rf .
If we show that the slope of this tangent is
1
& 2 Rf + Rf2 2
, the assertion is proved.
Two Fund Separation: Rf < Rmv

Let R A and A be the expected return and standard deviation
of return, respectively, of this tangency portfolio. Then the
results of (34) and (35) allow us to write the tangent’s slope
as
" #
R A Rf & 2
Slope = 2
Rf = A
A Rf
" #
2 Rf & Rf2
= = A (42)
Rf
Furthermore, we can use (29) and (34) to write
2
2 1 RA
A = + 2
&
Two Fund Separation: Rf < Rmv cont’d

We then substitute (34) where R 1p = Rf for R A
1 & 2
2
A = + 2
3
Rf
Rf2 2 Rf + &
= 2
(43)
2
Rf
Substituting the square root of (43) into (42):
" #
R A Rf 2 Rf & Rf2 Rf
= 1(44)
A Rf Rf2 2 Rf + & 2
1
= Rf2 2 Rf + & 2
which is the desired result.

An Important Separation Result

This result implies that all investors choose to hold risky
assets in the same relative proportions given by the tangency
portfolio ! A . Investors di¤er only in the proportion of wealth
allocated to this portfolio versus the risk-free asset.

Level of Risk-free Return
Rf < Rmv is required for asset market equilibrium. If

Rf > Rmv , the e¢ cient frontier
1
R p = Rf + & 2 Rf + Rf2 2 p is always above the
risky-asset-only frontier, implying the investor short-sells the
tangency portfolio on the ine¢ cient risky asset frontier and
invests the proceeds in the risk-free asset.

Level of Risk-free Return

If Rf = Rmv the portfolio frontier is given by the asymptotes
of the risky frontier. Setting Rf = Rmv in (39) and
premultiplying by e:
R p Rf 1
! = V (R Rf e) (45)
& 2 Rf + Rf2
1 R p Rf
e 0! = e 0V (R e)
& 2 Rf + Rf2
R p Rf
e 0! = ( ) =0
& 2 Rf + Rf2
which shows that total wealth is invested in the risk-free asset.
However, the investor also holds a risky, but zero net wealth,
position in risky assets by short-selling particular risky assets
to …nance long positions in other risky assets.
Example with Negative Exponential Utility

Given a speci…c utility function and normally distributed asset
returns, optimal portfolio weights can be derived directly by
maximizing expected utility:
~)= ~
bW
U(W e (46)
where b is the individual’s coe¢ cient of absolute risk aversion.

Now de…ne br bW0 , which is the individual’s coe¢ cient of
relative risk aversion at initial wealth W0 . Equation (46) can
be rewritten:
~)= ~ =W 0
br W ~p
br R
U(W e = e (47)
where R~ p is the total return (one plus the rate of return) on

the portfolio.
Example with Negative Exponential Utility cont’d

We still have n risky assets and Rf as before. Now recall the
properties of the lognormal distribution. If x~ is a normally
distributed random variable, for example, x~ N( ; 2 ), then
z~ = e ex is lognormally distributed. The expected value of ez is
+ 21 2
E [~
z] = e (48)
~ p = Rf + ! 0 (R
From (47), we see that if R ~ Rf e) is normally
~ is lognormally distributed. Using
distributed, then U W
equation (48), we have
h i
f = b r [R f +! 0 (R R f e)]+ 21 b r2 ! 0 V !
E U W e (49)
The individual chooses portfolio weights to maximize expected

utility:
h i
f = max b r [R f +! 0 (R R f e)]+ 21 b r2 ! 0 V !
maxE U W e (50)
! !
Since expected utility is monotonic in its exponent, this

problem is equivalent to
max ! 0 (R Rf e) 1 0
2 br ! V ! (51)
!
The n …rst-order conditions are

R Rf e br V ! = 0 (52)
Solving for !, we obtain
1 1
! = V (R Rf e) (53)
br

Comparing (53) to (39), note that
1 Rp Rf
= 0 (54)
br R Rf e V 1 (R Rf e)
so that the greater is br , the smaller is R p and the proportion
of wealth invested in risky assets.
Multiplying both sides of (53) by W0 , we see that the

absolute amount of wealth invested in the risky assets is
1
V 1 (R Rf e)
W0 ! = (55)
b
implying that with constant absolute risk aversion the amount
invested in the risky assets is independent of initial wealth.
Cross-hedging (Anderson & Danthine, 1981)
Consider a one-period model of an individual required to trade

a commodity in the future and wants to hedge the risk using
futures contracts.
Assume that at date 0 she is committed to buy (sell) y > 0
(y < 0) units of a risky commodity at date 1 at the spot price
p1 . As of date 0, y is deterministic, while p1 is stochastic.
There are n …nancial securities (futures contracts) where the

date 0 price of the i th …nancial security is pis0 , and its risky
date 1 price is pis1 .
Let si denote the amount of the i th security purchased at date
0, where si < 0 indicates a short position.

Cross-hedging (Anderson & Danthine, 1981) cont’d

De…ne n 1 quantity and price vectors s [s1 ::: sn ]0 ,
p0s [p10
s ::: p s ]0 , and p s
n0 1
s ::: p s ]0 . Also de…ne
[p11 n1
p s p1s p0s as the n 1 vector of security price changes.
Thus, the date 1 pro…t from securities trading is p s 0 s
De…ne the moments E [p1 ] = p1 , Var [p1 ] = 00 , E [p1s ] = p1s ,
E [p s ] = p s , Cov [pis1 ; pjs1 ] = ij , Cov [p1 ; pis1 ] = 0i , and the
(n + 1) (n + 1) covariance matrix of the spot commodity
and …nancial securities is
00 01
= 0 (56)
01 11
where 11 is an n n matrix whose i; j th element is ij , and

01 is a 1 n vector whose i th element is 0i .

The end-of-period pro…t (wealth) of the …nancial operator,
W , is
W = p s 0 s p1 y (57)
Assuming constant absolute risk aversion (CARA) utility, the
problem is to choose s in order to maximize:
1
maxE [W ] 2 Var [W ] (58)
s
Substituting in for the operator’s expected pro…t and variance:

max p s 0 s p1 y 1
2 y2 00 + s0 11 s 2y 01 s (59)
s
The …rst-order conditions are

ps 11 s y 0
01 =0 (60)


Solving for s, the optimal …nancial security positions are
1 1 s 1 0
s = 11 p +y 11 01 (61)
1 1 1
= 11 (p s1 p0s ) + y 11
0
01
First consider y = 0. This can be viewed as a trader who has

no requirement to hedge.
If n = 1 and p1s > p0s (p1s < p0s ), the speculator buys (sells) the
security. The size of the position is adjusted by the volatility of
the security ( 111 = 1= 11 ), and the level of risk aversion .
For the general case of n > 1, expectations are not enough to
decide to buy/sell. All of the elements in 111 need to be
considered to maximize diversi…cation.

For the general case y 6= 0, the situation faced by a hedger,
the demand for …nancial securities is similar to that of a pure
speculator in that it also depends on price expectations.
In addition, there are hedging demands, call them s h :
1
sh y 11
0
01 (62)
This is the solution to the variance-minimization problem, yet

in general expected returns matter for hedgers.
From (62), note that when n = 1 the pure hedging demand
per unit of the commodity purchased, s h =y , is
sh Cov (p1 ; p1s )

= (63)
y Var (p1s )

For the general case, n > 1, the elements of the vector

1 0
11 01 equal the coe¢ cients 1 ; :::; n in the multiple
regression model:
p1 = 0 + 1 p1s + 2 p2s + ::: + n pns + " (64)
where p1 p1 p0 , pis pis1 pis0 , and " is a mean-zero

error term.
An implication of (64) is that an operator might estimate the
hedge ratios, s h =y , by performing a statistical regression using
a historical time series of the n 1 vector of security price
changes. In fact, this is a standard way that practitioners
calculate hedge ratios.

Summary
A multivariate normal distribution of individual asset returns is

su¢ cient for mean-variance optimization to be valid.
Two frontier portfolios are enough to span the entire

mean-variance e¢ cient frontier.
When a riskless asset exists, only one frontier portfolio

(tangency portfolio) and the riskless asset is required to span
the frontier.
Hedging can be expressed as an application of mean-variance

optimization.


CH Pres 02

Uploaded by

Copyright:

Available Formats

CH Pres 02

Uploaded by

Document Information

Original Description:

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

CH Pres 02

Uploaded by

Copyright:

Available Formats

2.1: Assumptions 2.2: Indi¤erence 2.3: Frontier 2.4: R f 2.5: Hedging 2.

George Pennacchi University of Illinois

How does one optimally choose among multiple risky assets?

Due to diversi…cation, which depends on assets’return

Hence, the variance and higher moments of a portfolio need

Portfolios that make the optimal tradeo¤ between portfolio

George Pennacchi University of Illinois

What assumptions are needed for investors to care about only

For short-hand, write U(W ~ ) = U W0 R ep ) since

George Pennacchi University of Illinois

Taylor Series Approximation of Utility

If the utility function is quadratic, (U (n) = 0, 8 n 3), then

George Pennacchi University of Illinois

Alternative Assumptions for Mean-Variance

Quadratic utility, such as U (W ) = aW bW 2 , is problematic

Centered Normal Moments

Let a random variable, X , be distributed N ; 2 . Its

Centralized (multiply by exp( t))

Then we have following moments

George Pennacchi University of Illinois

Centered Normal Moments

George Pennacchi University of Illinois

Normal Distribution of Returns

Soh moments are either

which depends only on the mean and variance of the portfolio

George Pennacchi University of Illinois

Is a multivariate normal distribution realistic for asset returns?

This is a problem since most assets have limited liability,

Later, in a continuous-time context, we can assume asset

George Pennacchi University of Illinois

Preference for Return Mean and Variance

Consider an individual’s indi¤erence curves. De…ne e

where n(x) f (x; 0; 1). (e

Mean vs Variance cont’d

Taking the partial derivative with respect to Rp :

since U 0 is always greater than zero.

George Pennacchi University of Illinois

Risk and Utility

While U 0 is always positive, x ranges between 1 and +1.

George Pennacchi University of Illinois

Risk and Utility cont’d

Comparing the integrand of equation (10) for equal absolute

U 0 (Rp + xi p )xi n(xi ) + U 0 (Rp xi p )( xi )n( xi )

George Pennacchi University of Illinois

Risk and Utility cont’d

which is intuitive for risk-averse individuals.

Mean and Variance Indi¤erence Curve

Hence, each indi¤erence curve is positively sloped in Rp ; 2p

George Pennacchi University of Illinois

Mean and Standard Deviation Indi¤erence Curve

George Pennacchi University of Illinois

The individual’s optimal choice of portfolio mean and variance

Mean/Variance Optimization cont’d

The constraint on portfolio weights is ! 0 e = 1 where e is

The …rst-order conditions with respect to !, , and , are

George Pennacchi University of Illinois

Mean/Variance Optimization cont’d

Pre-multiplying equation (22) by R 0 and e 0 respectively: