Quantitative Methods
Add-
Add-on (Modul 4): Fund performance analysis
(Application of Simple & Multiple Linear Regression)
Fall Semester 2010
Performance Analysis
Fall Semester 08
Slide 2 Fund performance analysis
•There are various measures for the (risk-adjusted)
performance of mutual funds, e.g., the Sharpe ratio
measures excess return vs. volatility:
rp − rf
SRp =
σp
SRp Sharpe-Ratio of portfolio p
rp mean of realized returns of portfolio p
rf risk-free rate
σp standard deviation of portfolio returns
•The Sharpe ratio of various funds can be compared directly
•But it does not explain sources of (out-) performance, e.g.,
if the manager has "selection" and/or "timing" skills
Performance Analysis
Fall Semester 08
Slide 3 Jensen's alpha (1)
•Measure obtained from CAPM
•If market fluctuations are the only systematic source of risk,
then in the following equation beta measures the exposure
to market risk and the alpha coefficient
E (ri ) − rf = αi + [E (rM ) − rf ] βi
should be equal to zero (where ri is the return of asset i
and rM is the return of the market portfolio)
•In practice, the coefficients are estimated from time series
of ri,t , rf,t and rM,t using the regression
ri ,t − rf ,t = αi + ( rM ,t − rf ,t ) βi + ε i ,t
• αi measures the systematic portion of the average excess
return on asset i which is not due to market risk
Performance Analysis
Fall Semester 08
Slide 4 Jensen's alpha (2)
•A positive alpha would indicate that an asset outperformed
the market after adjusting for risk (or there is a contribution
to its return which is independent of the market)
•In the context of active portfolio management, αi measures
the contribution to the portfolio return in excess of the
market (after adjusting for risk) which is due to the
manager's selection skills
•When mutual fund data are analyzed, we would not expect
that most mutual fund managers can beat the market, i.e.,
have positive alphas
•Jensen's alpha allows only for comparison with a bench-
mark; the alphas of different portfolios cannot be compared
Performance Analysis
Fall Semester 08
Slide 5 Case study: Funds characteristics
Analyze the performance of five mutual funds for US stocks:
•UBSEUSI: Invests broadly, holds mainly large stocks
•UBSUSRI: 40 to 60 selected value stocks
•UBUSAGB: Large caps with focus on growth
•UBSESCI: Small caps with focus on growth
•SBCSMUI: Mid caps with focus on growth
UBSEUSI UBSUSRI
Performance Analysis
Fall Semester 08
Slide 6 Funds characteristics (cont.)
UBUSAGB
UBSESCI SBCSMUI
Performance Analysis
Fall Semester 08
Slide 7 Funds performance since 01/2005
Performance Analysis
Fall Semester 08
Slide 8 Estimation of market model
•The model is estimated with returns from the last 5 years
(except UBUSAGB which was initiated in Oct 2004)
•The excess market return that is used as explanatory
variable is taken from Fama/French (see later), they also
provide the risk-free rate (obtained from 1M T-bill data)
r_mean R^2 S.E. alpha t-value beta t-value
UBSEUSI 0.0021057 0.931 0.00728 -0.00258 -2.70500 0.91702 27.90015
UBSUSRI 0.0027877 0.797 0.01294 -0.00172 -1.01106 0.88078 15.07793
UBUSAGB 0.0041328 0.840 0.01490 0.00067 0.30117 1.12673 15.02660
UBSESCI 0.0042107 0.766 0.01909 -0.00186 -0.74242 1.18694 13.77672
SBCSMUI 0.0071249 0.818 0.01502 0.00153 0.77538 1.09464 16.14346
Note: Observation periods 5 years (except UBUSAGB 45 months), mean returns are per month
•Interpretation: …
•Note: The funds follow (various) benchmarks different
from the one used here
Performance Analysis
Fall Semester 08
Slide 9 Fama/French three-factor model
•CAPM uses a single factor to compare excess returns of an
asset or portfolio with excess returns of the whole market
•It is known that this oversimplifies the market complexity
•Typically two classes of stocks tend to perform better than
the market:
– Small caps
– Stocks with high book-to-market (btm) ratio ("value stocks")
•Fama/French (1993) added two factors to the CAPM to
reflect a portfolio's exposure to these classes
•Their model has a higher R2 than the CAPM
(≈ 0.9 vs 0.8 for diversified portfolios)
Performance Analysis
Fall Semester 08
Slide 10 Fama/French three-factor model
•They construct these factors from the returns on portfolios
of NYSE, Amex & NASDAQ stocks1):
– SMB ("small minus big"): portfolio long in small caps, short in
large caps
– HML ("high minus low"): portfolio long in stocks with high
btm ratio, short in stocks with low btm ratio
– The market factor is the return on a value-weighted
portfolio of the stocks used in the analysis
•Formally, let rs – rb the return on the portfolio of small caps
in excess of the return on the portfolio of large caps; like-
wise rH – rL is the return on the portfolio of high btm firms
in excess of the return on the portfolio of low btm firms
E (ri ) − rf = α i + [E (rM ) − rf ] βi * + E (rS − rB ) bs + E (rH − rL ) bv
1) The portfolio returns and factors can be downloaded from
http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html
Performance Analysis
Fall Semester 08
Slide 11 Comparison with CAPM
• Earlier result (based on CAPM with single factor):
R^2 S.E. alpha t-value beta t-value
UBSEUSI 0.931 0.00728 -0.00258 -2.70500 0.91702 27.90015
UBSUSRI 0.797 0.01294 -0.00172 -1.01106 0.88078 15.07793
UBUSAGB 0.840 0.01490 0.00067 0.30117 1.12673 15.02660
UBSESCI 0.766 0.01909 -0.00186 -0.74242 1.18694 13.77672
SBCSMUI 0.818 0.01502 0.00153 0.77538 1.09464 16.14346
• Result for Fama/French model with three factors:
R^2 S.E. alpha t-value beta t-value SMB t-value HML t-value
UBSEUSI 0.945 0.00659 -0.00292 -3.20787 0.96913 29.63923 -0.16514 -3.75791 0.05259 1.00885
UBSUSRI 0.870 0.01054 -0.00365 -2.50298 0.98277 18.79487 -0.28802 -4.09855 0.33621 4.03277
UBUSAGB 0.906 0.01170 0.00275 1.52902 1.09835 17.01182 -0.06983 -0.76556 -0.55360 -5.19197
UBSESCI 0.891 0.01327 -0.00020 -0.10771 0.96940 14.73057 0.68236 7.71522 -0.26542 -2.52966
SBCSMUI 0.866 0.01310 0.00198 1.09612 0.97371 14.98822 0.39231 4.49333 -0.06147 -0.59343
Note: Observation periods 5 years (except UBUSAGB 45 months), mean returns are per month
• Observations / interpretation: …
• What do the coefficients tell about the funds characteristics?
