Portfolio Performance Measurement
Eric Zivot
December 8, 2009
1 Investment Styles
1.1 Passive Management
• Believe that markets are in equilibrium
— Assets are correctly priced
• Hold securities for relatively long periods with small infrequent changes
• Hold surrogates for market portfolio known as index funds
— Low cost diversified portfolios (e.g. Vanguard Index Funds)
— motivated by portfolio theory and CAPM: efficient portfolios are com-
binations of T-Bills and a market index portfolio
• Do not try to create portfolios to “actively” beat the returns on index
funds
1.2 Active Management
• Markets are not always in equilibrium
— Some securities are “mis-priced”
• Buy under-priced (positive “alpha”) assets and sell over-priced (negative
“alpha”) assets
• Active managers often “tweak” a benchmark (index) portfolio
weight in weight in active
Security
benchmark active port position
MSFT .05 .10 +.05
GM .02 -.05 -.07
.. .. .. ..
• Active management strategies
— individual stock selection
— sector selection (e.g. utility, technology)
— asset class selection (stocks, bonds, real estate)
• Most mutual funds are actively managed.
— management fees can vary substantially from fund to fund
— fee is often a percentage of assets under management
2 Evaluating Investment Performance
Q: Is it worthwhile to “pay” for active management of portfolios?
Key Concepts
• Actively managed portfolios should be compared with passive (index) bench-
marks of a similar risk class
• Superior past performance could be luck or could be skill
• Often very little historical data to evaluate managed portfolios
— Statistical analysis is difficult
2.1 Risk Adjusted Measures of Performance
Observe returns on active portfolio and benchmark over some time horizon (e.g.
5 years of monthly data)
• Does the managed portfolio exhibit superior performance adjusted for risk?
• How to rank different actively managed portfolios?
Measures of risk
• Market risk (portfolio beta, β p, from SI model or CAPM)
• Total risk (portfolio standard deviation, σ p)
Ex Post (Historical) measures
1 XT 1 XT
μ̂p = Rp,t, r̂f = rf,t
T t=1 T t=1
⎛ ⎞1/2
1 T
X
σ̂ p = ⎝ (Rp,t − μ̂p)2⎠
T − 1 t=1
d
cov(R p,t, RM,t)
β̂ p =
vd
ar(RM,t)
Types of Performance Measures
• Average return difference adjusted for risk
ave return on active portfolio -
ave return on risk adjusted benchmark
• Risk adjusted reward/risk ratio
average excess return
risk measure
2.1.1 Performance Measures Based on Market Risk
Idea: Under CAPM, market risk is captured by β and expected returns are
captured by the Security Market Line (SML)
μp,CAP M = rf + β p(μM − rf )
Jensen’s alpha
Risk-adjusted return difference
α̂∗p = μ̂p − μ̂p,CAP M
Computation: use linear regression to estimate the excess returns SI model
Rp,t − rf = α∗p + β p(RMt − rf ) + εpt, εpt ∼ iid N (0, σ 2ε )
Statistical evaluation:
H0 : α∗p = 0 (no superior performance) vs. H1 : α∗p 6= 0
Information Ratio
dp =
α̂∗p
IR
σ̂ ε
Statistical evaluation: Use bootstrap to compute standard error and confidence
interval
2.1.2 Performance Measures Based on Total Risk
Idea: Efficient portfolios are combination of T-bills and tangency portfolio.
Under CAPM, the tangency portfolio is the market portfolio
Sharpe ratio
μ̂p − r̂f
SRp =
σ̂ p
= excess return per unit portfolio risk
Statistical evaluation:
H0 : SRp = SRM (no superior performance) vs H1 : SRp 6= SRM
Evaluate H0 using bootstrap
R Package for Performance Evaluation
PerformanceAnalytics