Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Problem set 2: Solutions

Damien Klossner∗
damien.klossner@epfl.ch
Extranef 128

March 9, 2015

Problem 1 (10 points)

Let Rπ denote the return on a portfolio located on the minimum-variance frontier for
risky assets only and suppose that π is different from the global minimum-variance
portfolio. Show that there is a portfolio z(π) also located on the minimum-variance
frontier for risky assets only, which has the property that Cov(Rπ , Rz(π) ) = 0. Show
that E[Rz(π) ] = (C − BE[Rπ ])/(B − AE[Rπ ]), where A, B, and C are the con-
stants defined in the slides. Hint: First show that the covariance between the return
on a minimum-variance portfolio with mean m1 and one with mean m2 is equal to
1
∆ (Am1 m2 − B(m1 + m2 ) + C).

Solution

You have seen in the lecture that the composition of any minimum-variance portfolio
(of risky assets only), with expected value m1 , has the form (slide 7):

w1 = λ1 Σ−1 1 + γ1 Σ−1 µ

This document is partly based on the notes prepared by Ilya Kolpakov for an earlier version of this
course. I would like to thank him for sharing his notes with me.

1
where

C − m1 B m1 A − B
λ1 = , γ1 =
∆ ∆
A = 10 Σ−1 1 > 0, 0 −1
B=1Σ µ

C = µ0 Σ−1 µ > 0, ∆ = AC − B 2 > 0.

Since portfolio 2 is composed of risky assets only, w2 satisfies

E [R2 ] = w20 µ = m2

V ar [R2 ] = w20 Σw2

w20 1 = 1.

(with R2 denoting the realized return of portfolio 2). So the covariance between any
two portfolios, 1 and 2, on the minimum-variance frontier is equal to:

Cov [R1 , R2 ] = w10 Σw2 = λ1 10 Σ−1 + γ1 µ0 Σ−1 Σw2




= λ1 10 w2 +γ1 µ0 w2
|{z} | {z }
1 m2
C − m1 B m1 A − B
= + m2
∆ ∆
1 
= Am1 m2 − B (m1 + m2 ) + C . (1)

Therefore, for any portfolio π, we can create a portfolio z(π) such that its expected
value E Rz(π) will force expression (1) to be equal to zero.
 

h i
0 = Cov Rπ , Rz(π)
 
0 = AE[Rπ ]E[Rz(π) ] − B E[Rπ ] + E[Rz(π) ] + C

(B − AE[Rπ ]) E[Rz(π) ] = C − BE[Rπ ]


C − BE[Rπ ]
E[Rz(π) ] = (2)
B − AE[Rπ ]

Since π is not the global minimum-variance portfolio, we know that (slide 12)

E[Rπ ]A − B
= γπ 6= γmin = 0.

So B − AE[Rπ ] 6= 0 and portfolio z(π) exists.

2
Problem 2 (10 points)

Let Rmin denote the return on the global minimum-variance portfolio of risky assets.
Let R be the return on any risky asset or portfolio of risky assets, efficient or not. Show
that
Cov(R, Rmin ) = V ar(Rmin ). (3)

Hint: Consider a portfolio consisting of a fraction w in this risky asset and a fraction
(1 − w) in the global minimum-variance portfolio. Compute the variance of the return
on this portfolio and realize that the variance has to be minimized for w = 0.

Solution using the hint

If R = Rmin then (3) clearly holds. So from now on we assume

ρ (R, Rmin ) < 1 (4)

where ρ (R, Rmin ) denotes the correlation between the returns of each portfolio.
Consider a portfolio suggested in the hint. On one hand, its variance V (w) cannot
be lower than the variance of the global minimum-variance portfolio V (0)

V (w) ≥ V (0), ∀w. (5)

On the other hand, V (w) is equal to

V (w) = w2 V ar(R) + (1 − w)2 V ar(Rmin ) + 2w(1 − w)Cov(Rmin , R).

The first two derivatives of V (w) are

V 0 (w) = 2wV ar(R) − 2(1 − w)V ar(Rmin ) + 2(1 − 2w)Cov(Rmin , R)

V 00 (w) = 2V ar(R) + 2V ar(Rmin ) − 4Cov(Rmin , R) > 0,

where the inequality holds because of (4). Therefore, V : < → < is strictly convex, and
arg min{V (w) | w ⊂ <} either is empty or contains a single point. But we know by
(5) that V (w) attains a minimum at w = 0, so V (0) is the unique global minimum of
V on <. Moreover, < is open, so the global minimum is an unconstrained minimum,
which can be identified by the first-order condition V 0 (w) = 0 and is attained at w = 0.

V 0 (0) = −2V ar(Rmin ) + 2Cov(Rmin , R) = 0

Cov(Rmin , R) = V ar(Rmin )

3
Direct solution
1 −1
Let wmin be the global minimum-variance portfolio wmin = A Σ 1. Furthermore, let
w be any portfolio of risky assets with return R. We have w0 1 = 1 and

0 1 0 −1 1
Cov(Rmin , R) = wmin Σw = 1 Σ Σw = = V ar(Rmin ), (6)
A A

where the last equality has been derived during the lecture (slide 11).

Problem 3 (30 points)

The optimal mean-variance portfolio is a complex function of estimated means, volatil-


ities, and correlations of asset returns. There are many parameters to estimate. Opti-
mized mean-variance portfolios can blow up when there are tiny errors in any of these
inputs.
In this exercise you will run a horse-race between four different portfolio strategies,
each of which is a special case of the full mean-variance strategy. Diversification is
common to all the strategies, but they build a diversified portfolio in different ways.
This leads to very different performance.
The four portfolios strategies are:

• The tangency portfolio (T).

• The minimum variance portfolio (MV). This is a special case of full mean-variance
analysis that does not estimate means; it implicitly assumes that all assets have the
same mean.

• The risk parity portfolio (RP) with weights equal to the inverse of the standard
deviations of returns. This is also a special case of mean-variance analysis that does
not estimate means or correlations; it implicitly assumes that all assets have the
same mean and all assets are uncorrelated.

• The equally weighted portfolio (EW). This has nothing to estimate. It is also a spe-
cial case of mean-variance analysis; it implicitly assumes that all assets are identical.

As you move from full-blown mean-variance analysis to equal weights, you estimate
fewer parameters and thus there are fewer things that can go wrong with the mean-

4
variance optimization. The extreme case is the equal weights, which require no analysis
of data.
The Excel file totalReturns.xlsx has monthly total return indices for four asset
classes: US Treasuries (from Bank of America Merrill Lynch bond indices), US corporate
bonds (from Bank of America Merrill Lynch bond indices), US stocks (from MSCI),
global stocks (from MSCI). It also contains 1-month T-bill rates, which is the relevant
risk-free rate over the sample period. The sample period starts in 1978.
For the four different portfolio strategies, track the performance from January 1988
to the end of the sample period. Implement the strategies at time t using data for
a ten-year window. Therefore, the first portfolios are formed at the end of December
1987 using returns from January 1978 to December 1987. The portfolios are held for
one month (i.e, until end of January 1988). The next portfolios are formed at the end
of January 1988 using returns from February 1978 to January 1988. The portfolios are
held for one month (i.e, until end of February 1988), and so on. Use one-month T-bills
as the risk-free rate.

(a) For each of the four portfolio strategies, plot the time series of the portfolio weights.

(b) Compute mean and standard deviation of portfolio excess returns as well as the

Sharpe ratio (annualize by multiplying by 12).

(c) Interpret the differences in performance.

Solution

The portfolio weights for each of the four strategies are shown in Figure 1 while return
statistics are in Table 1.
The compositions of both the tangency and minimum variance portfolios are rather
erratic and include sizable long/short positions in Treasury/corporate bonds. As can be
seen from Figure 1, the tangency portfolio is long corporate bonds/short treasuries in
the first half of the sample, while the minimum variance portfolio has a reverse position
in bonds for almost the whole sample. The weights of the risk parity portfolio are much
more stable.

5
Performance-wise, the tangency (T) portfolio is greatly outperformed by the mini-
mum variance (MV) portfolio whose Sharpe ratio is twice higher. Since MV does not
rely on estimated expected returns (which are used by T), it is plausible that T’s inferior
performance is due to the noise in the expected returns estimates.
The risk parity (RP) portfolio, which does not use correlation estimates, performs
a bit better than the MV portfolio.
The equally weighed (EW) portfolio performs worse than the RP and MV portfolios,
suggesting that variance estimates contain valuable information which should not be
neglected.

T MV RP EW

Mean excess return (monthly) 0.22% 0.26% 0.35% 0.41%


Excess return vol. (monthly) 2.25% 1.28% 1.57% 2.28%
Sharpe ratio (annualized) 0.34 0.70 0.78 0.63

Table 1: Mean and standard deviation of returns

Problem 4 (20 points)

In spite of the evidence above, consider an investor engaging in mean-variance optimiza-


tion. Use the data from the previous exercise and base your analysis on the full sample
period.

(a) Draw the risky-asset-only minimum-variance frontier for

(a) US stocks and global stocks

(b) US stocks, global stocks, and US corporate bonds

(c) All four asset classes

(In the (σ, µ) diagram, limit µ to the interval between 0.002 and 0.012 and σ to the
interval between 0 and 0.07.)

(b) Suppose you target a monthly expected return of 0.0100. In the case with all four

6
T MV
15 3

10 2

5 1

0 0

−5 −1

−10 −2
200 300 400 200 300 400

RP EW
0.5 1
Gov bonds
0.8 Corp bonds
0.4
0.6 US stocks
0.3 Global stocks
0.4
0.2
0.2

0.1 0
200 300 400 200 300 400

Figure 1: Portfolio weights.

asset classes, compute the minimum-variance portfolio that achieves this expected
return. What is its return standard deviation on a monthly basis?

(c) Consider an investor who has mean-variance utility U = µp − a2 σp2 and a risk aversion
coefficient a of 10. What is the optimal portfolio? What is its expected return and
return standard deviation on a monthly basis? Draw the corresponding indifference
curve.

(d) Suppose now that a riskless asset is also available for investment. Assume that it has
a monthly return of 0.0042 (approximately the average T-bill rate over the sample
period). Draw the new mean-variance efficient frontier.

(e) What is now the optimal portfolio of the investor? What is its expected return and

7
return standard deviation on a monthly basis? Does the investor lend or borrow?
How much? Draw the corresponding indifference curve.

Solution

(a) With A, B, C, and ∆ as defined in class (slide 7), the volatility σ(µp ) of a minimum-
q
1

variance portfolio with expected return µp is equal to σ(µp ) = ∆ Aµ2p − 2µp + C .
Frontiers are shown graphically in Figure 2.

−3
x 10
12
Stocks
11 Stocks and corporates
Stocks, corporates, and Treasuries
10

9
Expected Return

2
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Standard Deviation

Figure 2: Minimum variance frontiers for stocks (US and global), stocks and US corpo-
rate bonds, and all four asset classes.

(b) We use the formula derived in class for the composition of a minimum-variance
portfolio with expected return µp :

w = λΣ−1 1 + γΣ−1 µ

8
where

C − µp B µp A − B
λ = , γ=
∆ ∆

and A, B, C, and ∆ are as defined in class (slide 7). Setting µp = 0.01 and using
the sample means and covariances of the four returns series as our estimations for
the vector µ and the matrix Σ, we obtain w = (−0.2750, 0.3827, 1.0080, −0.1156)0 .
The return standard deviation of this portfolio can be calculated using (slide 7):
s
p Aµ2p − 2Bµp + C
σp = λ + γµp = = 0.0424.

(c) For the special case of an investor maximizing a mean-variance utility, we derived
closed-form formulas for all these quantities during class (slide 34). Setting a = 10
and using the sample moments obtained from the dataset as our estimations for the
vector µ and the matrix Σ, we obtain

1 − B/a −1 1
w∗ = Σ 1 + Σ−1 µ = (0.9089, −0.2175, 0.2355, 0.0732)0 (7)
A a
1 − B/a C
µ∗ = B+ = 0.0074
r A a
1 − B/a µ∗
σ∗ = + = 0.0173.
A a

The indifference curve is plotted in Figure 3.

(d) Using the formulas derived in class (slide 22 and 23) with R0 = 0.0042, we obtain:

Σ−1 (µ − R0 1)
wtan = = (0.9324, −0.2294, 0.2201, 0.0769)0 (8)
B − AR0
C − BR0
µtan = = 0.0073
B − AR0
s
C − 2R0 B + R02 A
σtan = = 0.0173.
(B − AR0 )2

When there is a riskless asset, the minimum-variance set is a pair of straight lines
p
in (σ, µ)-space, with intercepts R0 and slopes ± C − 2R0 B + R02 A. The minimum-
variance set is plotted in Figure 4.

9
−3
x 10
12

11

10

9
Expected Return

2
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Standard Deviation

Figure 3: The indifference curve, the efficient frontier and the investor’s optimal portfolio
when only the four risky assets are available.

(e) For the special case of mean-variance utility, the optimal portfolio can be derived in
closed-form (slide 38). Setting a = 10 and using the sample moments obtained from
the dataset as our estimations for the vector µ and the matrix Σ, we obtain
1 −1
w∗∗ = Σ (µ − R0 1) = (1.0057, −0.2475, 0.2374, 0.0830)0 (9)
a
0
µ∗∗ = R0 + w∗∗ (µ − R0 1) = 0.0075
p
σ∗∗ = 0 Σ−1 w
w∗∗ ∗∗ = 0.0183.

0 1 = 1.0786 so that the investor borrows 7.86% of what he invests in the


We have w∗∗
tangency portfolio. The indifference curve is plotted in Figure 4.

10
−3
x 10
12

11

10

9
Expected Return

2
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Standard Deviation

Figure 4: The indifference curve, the efficient frontier and the investor’s optimal portfolio
in case a riskless asset is available for investment.

Case Treasuries US Corp. Bonds US Stocks Global Stocks Risk-free E[R] Std. Dev.

b) −0.2750 0.3827 1.0080 −0.1156 0.0100 0.0424


c) 0.9089 −0.2175 0.2355 0.0732 0.0074 0.0173
d) 0.9324 −0.2294 0.2201 0.0769 0.0073 0.0173
e) 1.0057 −0.2475 0.2374 0.0830 −0.0786 0.0075 0.0183

Table 2: Summary statistics for the different portfolios considered in Problem 4.

11

You might also like