Please do not quote or distribute without permission

Andr F. Perold, October 2008 Page 1

Risk Stabilization
and Asset Allocation*
A d F P ld Andr F. Perold
Harvard Business School and HighVista Strategies
Prepared for QUAFAFEW
October 21, 2008
*Based on Risk Stabilization and AssetAllocationbyAndr F. Perold, manuscript, March 2008
Purpose of Asset Allocation
To make efficient tradeoffs between broad
asset class risks and expected returns
These tradeoffs usually are expressed These tradeoffs usually are expressed
through means of a policy portfolio, e.g.
60/40 equities/bonds
Policy portfolios are static asset allocations
Stock Market Volatility (VIX)
50%
60%
70%
80%
0%
10%
20%
30%
40%
1990 1993 1996 1999 2002 2005 2006 2007 2008
The Changing Stock-Bond Correlation
0.2
0.4
0.6
0.8
0.6
0.4
0.2
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Global Equity Market Correlation
0 50
0.75
1.00
0.00
0.25
0.50
Oct03 Apr04 Oct04 Apr05 Oct05 Apr06 Oct06 Apr07 Oct07 Apr08 Oct08
EMEAFE EMR3000 EAFER3000
Time Varying Risk
When asset class risks and correlations are
time varying, the risk of a static allocation is
also time varying y g
A static allocation thus is a very noisy
representation of portfolio risk
Please do not quote or distribute without permission
Andr F. Perold, October 2008 Page 2
Stable Risk Policies
Vary the asset allocation to maintain stable
portfolio risk
Example:
Reduce the exposure to global equities when equity Reduce the exposure to global equities when equity
market risk is high
Increase the exposure to global equities when equity
market risk is low
Basic questions:
Is risk forecastable?
What is the relationship between risk and expected
return?
Theoretical Relationship Between
Expected Return and Risk
Optimal exposure
= Risk Tolerance x Risk Premium/Variance
In equilibrium, exposure = constant
Theoretically, therefore
Risk premium = proportional to variance
U.S. Equities: Risk vs Contemp. Return
(Intra-Year Volatility, 1926-2008)
y = -0 46x + 0 19
20%
40%
60%
y = -0.46x + 0.19
R = 0.05
-60%
-40%
-20%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80%
R
e
t
u
r
n
Volatility
U.S. Equities: Risk vs Future Return
(Intra-Year Volatility, 1926-2007)
20%
40%
60%
y
e
a
r
y = -0.02x + 0.12
R = 0.00
-60%
-40%
-20%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80%
R
e
t
u
r
n
n
e
x
t

Volatility
Persistence in Volatility
(Intra-Year Volatility, 1926-2008)
y = 0.66x + 0.06
R = 0 44
50%
60%
70%
80%
t
y
e
a
r
R = 0.44
0%
10%
20%
30%
40%
0% 10% 20% 30% 40% 50% 60% 70% 80%
V
o
la
t
ilit
y

n
e
x
t
Volatility
Stylized Facts
Volatility varies significantly
Volatility is predictable
Volatility seems only weakly related to Volatility seems only weakly related to
expected return
Please do not quote or distribute without permission
Andr F. Perold, October 2008 Page 3
Return

P
Efficient Frontier
Risk
RF

Sharpe
Ratio
Changing Volatility: An Example
15%
20%
25%
a
t
i
l
i
t
y
0%
5%
10%
1 2 3
Time Period
V
o
l
a
Changing Volatility (cont.)
15%
20%
r
n
0%
5%
10%
0% 5% 10% 15% 20% 25%
Risk
R
e
t
u
r
Application to Active Management
Actively managed fund
= Index + long/short portfolio
Market neutral hedge fund g
= Cash + long/short portfolio
Example: 40 best buys and 70 best shorts
$long = $short
How much exposure to the long/short portfolio?
The Model
Allocation x
t
to a risky portfolio and 1-x
t
to
a riskless asset
Risk Premium is proportional to
t
a
Risk Premium is proportional to
t
a = 2: Proportional to variance
a = 1: Proportional to standard deviation
a = 0: No relation to risk
Expected Return vs. Sigma
15%
20%
a =2
0%
5%
10%
0% 8% 16% 24% 32%
Sigma
a =0
a =1
Please do not quote or distribute without permission
Andr F. Perold, October 2008 Page 4
The Model
Allocation x
t
to a risky portfolio and 1-x
t
to a
riskless asset
Risk Premium is proportional to
t
a
p p
t
a = 2: Proportional to variance
a = 1: Proportional to standard deviation
a = 0: No relation to risk
Log
t
is normally distributed with variance = v
2

t
is a forecast of
t
with correlation b
Strategies
Static policy allocation
X
t
= constant
Stable risk policy Stable risk policy
X
t
= */
t
Optimal strategy
X
t
= constant/
t
(2-a)
Main Finding
S Stable Allocation
Sharpe Ratio Policy
S x exp (a-2)
2
b
2
v
2
Optimal Strategy
S x exp (3/2-a)b
2
v
2
Stable Risk
Sharpe Ratio Improvement
1.0
1.5
2.0
2.5
Optimal Strategy
-1.5
-1.0
-0.5
0.0
0.5
0.0 0.5 1.0 1.5 2.0 2.5
Expected return parameter "a"
Risk Stabilization
Economic Significance
v b Static Allocation
Sharpe Ratio
Sigma
0.5 0.8 0.5 10%
Increase in Sharpe Ratio Increase in Portfolio Expected Return
a
0.0
0.5
1.0
1.5
2.0
Optimal Strategy
0.38
0.20
0.08
0.02
0.00 -
Stable Risk
0.27
0.17
0.08
0.00
0.08
Optimal Strategy
1.89%
0.99%
0.42%
0.10%
0.00%
0
-
Stable Risk
1.36%
.87%
0.42%
0.00%
0.38%
Investing when risk varies over time
Static Policy Allocation
Random risk
Lowest Sharpe Ratio (if returns vary slowly with risk)
Stable Risk Policy
Random exposure
Average Sharpe Ratio
Dont need to know expected return
Optimal Strategy
Random risk and exposure
Highest Sharpe Ratio
Need to know expected return