Confirming Pages

Portfolio Performance
Evaluation
Chapter

18
Learning Objectives:

LO18-1 Compute risk-adjusted rates of return, and use them to evaluate investment performance.

LO18-2 Determine which risk-adjusted performance measure is appropriate in a variety of invest-

ment contexts.

LO18-3 Apply style analysis to assess portfolio strategy.

LO18-4 Decompose portfolio returns into components attributable to asset allocation choices ver-
sus security selection choices.

LO18-5 Assess the presence and value of market-timing ability.

I
n previous chapters, we derived predictions We show the problems with these approaches
for expected return as a function of risk. In when you try to apply them in a real and com-
this chapter, we ask how we can evaluate plex world. Finally, we examine evaluation pro-
the performance of a portfolio manager cedures used in the field. We show how
accounting for portfolio risk. Adjusting average overall investment results are decomposed
returns for risk presents a host of issues and attributed to the underlying asset alloca-
because the proper measure of risk may not be tion and security selection decisions of the
obvious and risk levels may change along with portfolio manager.
portfolio composition. We finally turn to two specific forms of active
We begin with conventional approaches to management: market timing based solely on
Related websites for this
chapter are available at
risk adjustment. These use the risk measures macroeconomic factors, and security selection
[Link]/bkm. developed earlier to rank investment results. based on microeconomic forecasting.

596

bod34698_ch18_595-[Link] 596 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 597

18.1 RISK-ADJUSTED RETURNS

Investment Clients, Service Providers,
and Objectives of Performance Evaluation
Individual households as well as institutional money managers must decide whether to use
passive or active management. Passive management involves (1) capital allocation between passive management
cash (almost-risk-free vehicles such as money market funds) and the chosen risky portfolio Holding a well-diversified
constructed from one or more index funds or ETFs. portfolio without attempting
Still, the concept of passive management is not completely unambiguous. At one extreme, to search out security
passive investors will commit to capital allocation with a fixed risky portfolio and change their mispricing.
allocations only infrequently in response to significant changes in circumstances or risk tolerance.
cash
At the other extreme, they will adjust portfolio weights based on estimates of risk derived, for
Shorthand for virtually
example, from VIX (the volatility indexes, discussed in Chapter 16) or from other sources.
risk-free money market
Alternatively, households and institutional endowments may choose active management,
securities.
in which case they usually become clients of professional portfolio managers.1 The dividing
line between passive and active management is the forecasting of future rates of return on active management
asset classes and/or individual assets. Such forecasting is more difficult than estimation of risk Attempts to achieve returns
by an order of magnitude. The reason for this is quite subtle and is lost on many professional higher than commensurate
as well as novice investors. Competition among the vast number of investors means that secu- with risk by forecasting broad
rity prices generally reflect publicly available information. Thus, successful forecasting of markets and/or by identifying
future prices and rates of return requires differential private information. To estimate risk, on mispriced securities.
the other hand, investors can freely and quite easily use publicly available information, making
these estimates a commodity. Accordingly, we call active managers those who forecast returns
in conjunction with risk to construct optimal portfolios. A few professionals restrict their
activity to market timing (switching between risky portfolios and cash), some concentrate on
asset allocation only, and most engage in both asset allocation and security selection.2
Both clients and professionals are interested in performance evaluation. Clients need to know
whether their chosen professionals produce adequate net-of-fee returns. Professionals need to
shore up their methodology and maintain qualified staff with adequate compensation to com-
pete in the market for these services. Lapses in performance can cost them dearly as evidence
shows that funds under management flow quickly from underachievers to superperformers.
Performance evaluation of a portfolio is difficult because of the great volatility of asset
returns. A portfolio’s average return over an evaluation period is inadequate to measure perfor-
mance. To begin with, the average return realized over any particular period may not represent
the expected return. Surely, luck (good or bad) should not be allowed to dominate the evalua-
tion process. Even when the average return does approximate expected return, it still would be
invalid as a measure of performance because it ignores risk—we expect higher-risk invest-
ments to outperform lower-risk ones in average to boom markets and to underperform in bear
markets. Hence, we must estimate portfolio risk to determine the adequacy of the average
return. Since volatility generates statistical errors in estimates of both expected return and risk,
we must remain skeptical of the evaluation process.

Comparison Groups
The simplest and most popular way to adjust returns for portfolio risk is to compare rates of
return with those of other investment funds with similar risk characteristics. For example,
high-yield bond portfolios are grouped into one “universe,” growth stock equity funds are
grouped into another universe, and so on. Then the average returns of each fund within the

1
Households and institutional endowments that conduct active management in-house become their own clients. The
adage that a lawyer who represents himself has a fool for a client doesn’t necessarily apply here.
2
Many professional managers are prohibited from extensive market timing by a prospectus or contract that fixes a
range of allowed weights in cash instruments.

bod34698_ch18_595-[Link] 597 09/08/12 7:35 PM

 Confirming Pages

598 Part SIX Active Investment Management

FIGURE 18.1

Universe comparison: Rate of return (%)

Periods ending December 31,
2012 30
The Markowill Group
S&P 500
25

20

15

10

1 quarter 1 year 3 years 5 years

universe are ordered, and each portfolio manager receives a percentile ranking depending on
comparison universe relative performance within the comparison universe, the collection of funds to which per-
The set of portfolio managers formance is compared. For example, the manager with the ninth-best performance in a uni-
with similar investment styles verse of 100 funds would be the 90th percentile manager: Her performance was better than
that is used to assess relative 90% of all competing funds over the evaluation period.
performance. These relative rankings usually are displayed in a chart like Figure 18.1. The chart summarizes
performance rankings over four periods: one quarter, one year, three years, and five years. The top
and bottom lines of each box are drawn at the rate of return of the 95th and 5th percentile man-
agers. The three dotted lines correspond to the rates of return of the 75th, 50th (median), and
25th percentile managers. The diamond is drawn at the average return of a particular fund, the
Markowill Group, and the square is drawn at the average return of a benchmark index such as the
S&P 500. This format provides an easy-to-read representation of the performance of the fund
relative to the comparison universe.
This comparison with other managers of similar investment groups is a useful first step in
evaluating performance. Even so, such rankings can be misleading. Consider that within a par-
ticular universe some managers may concentrate on particular subgroups, so that portfolio char-
acteristics are not truly comparable. For example, within the equity universe, one manager may
concentrate on high-beta stocks. Similarly, within fixed-income universes, interest rate risk can
vary across managers. These considerations suggest that we need more precise risk adjustment.

Basic Performance-Evaluation Statistics

Performance evaluation relies on the index model discussed in Sections 6.5 and 7.2 and on the
CAPM of Section 7.1. The single-index model equation applied to a portfolio P is

RPt 5 bP RMt 1 aP 1 ePt (18.1)

where RPt 5 rPt 2 rf t is portfolio P ’s excess return over cash equivalents during period t, rft is
the return on cash, and RMt is the excess return on the market index. bP is the portfolio’s sensi-
tivity to the market index, hence its measure of systematic risk, and bPRMt is the component of

bod34698_ch18_595-[Link] 598 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 599

return that is driven by the market. The extra-market or nonsystematic component, aP 1 ePt ,
includes the portfolio alpha plus zero-mean noise, e , called the residual, which is uncorrelated
with RM . Thus, the expected excess return of the portfolio for some evaluation period is
E(RP) 5 bPE(RM) 1 aP (18.2)

We measure expected returns over the period (unfortunately, with sampling error) by average
return.
The CAPM hypothesis is that the market portfolio is mean-variance efficient. The index
model uses an index portfolio, M, to proxy for the theoretical market portfolio, and hence it is
the benchmark passive strategy against which competing portfolios are measured. The CAPM
hypothesis is that the alpha of all securities and competing portfolios is zero. A professional
who claims to outperform the index must produce a positive alpha; the validity of the CAPM
doesn’t preclude some professionals from doing so, as long as the totality of investments that
exhibit positive alpha is not large relative to aggregate wealth in the economy.
What about portfolio risk? As noted above, beta measures systematic risk since the variance
of the market-driven return component is
Var(bP RMt ) 5 b P2 sM2 (18.3)

2
and the term sM is the same for all portfolios. The extra-market component of return contributes
the quantity Var(eP) to portfolio variance. The standard deviation of the residual return e, which
we will denote here as se , is called residual risk or residual SD . The variance of the return on P is
thus the sum of the variances (since the systematic and residual components are uncorrelated):
s 2P 5 b 2P sM
2
1 s 2e (18.4)

We may now prepare the statistics that are used for performance evaluation of a portfolio
P from a sample of observations over an interval of T periods (usually months). The procedure
includes the following steps:
1. Obtain the time series of RPt for portfolio P, and RM t for the benchmark M.
2. Compute the arithmetic averages of the series RP , RM . These are taken as estimates of
the expected returns of portfolios P and M for the evaluation period.
3. Compute the standard deviations of returns for portfolios P and M, sP and sM . These
serve as estimates of the total risk of P and M.
4. Run a regression of RPt on RMt to obtain estimates of P ’s beta, alpha, residual SD, and
correlation with the benchmark. Check the significance statistics to see that the sample
is reasonable. In particular, if the beta coefficient estimate is not significant, the sample
may be insufficient for the performance-evaluation statistics discussed below.
5. Recall from Equation 18.2 that the regression intercept is P ’s alpha, aP 5 RP 2 bP RM .
6. Recall from Equation 18.4 that the standard error, or residual standard deviation, of the
regression is se 5 SQRT(s 2P 2 bP2 s M 2
).
Table 18.1 presents performance-evaluation statistics for two professionally managed port-
folios, P and Q, the benchmark, M, and cash. Notice that P is aggressive with a beta of 1.25.
Q might be a hedge fund, not completely market-neutral (which would entail a beta of zero),
but still with a defensive beta of .5. Thus, most of the volatility of Q is due to its residual SD.

Performance Evaluation of Entire-Wealth Portfolios

Using the Sharpe Ratio and M-Square
Consider The Diabetes Foundation, a small charity, whose board has decided to invest its
endowment in one of the three portfolios of Table 18.1. In this case, the total risk of the cho-
sen portfolio will also be the endowment’s risk. Accordingly, the familiar Sharpe ratio, Sharpe ratio
Reward-to-volatility ratio;
R ratio of portfolio excess
S5 (18.5)
s return to standard deviation.

bod34698_ch18_595-[Link] 599 09/08/12 7:35 PM

 Confirming Pages

600 Part SIX Active Investment Management

TABLE 18.1 Performance of two managed portfolios, P and Q, the benchmark portfolio, M, and
cash equivalents

Portfolio P Portfolio Q Benchmark Cash

Average return 13.6 9.5 10.4 4

Average excess return (%) 9.60 5.50 6.37 0
Standard deviation (%) 24.1 18.0 18.5 0
Beta (pure number) 1.25 0.50 1.0 0
Alpha (%) 1.6 2.3 0 0
Residual SD (%) 6.79 15.44 0 0
Correlation with benchmark 0.96 0.51 1 0
Sharpe ratio 0.398 0.306 0.344 0
M-square (%) 1.00 20.72 0 0
Treynor measure 7.68 11.00 6 0
Information ratio 0.24 0.15 0 0

which measures risk by total volatility (SD), must determine the choice. Table 18.1 shows that
the Sharpe ratio of portfolio P (.398) is highest; hence P would be the charity’s choice. Notice
that P ’s average return is sufficiently large to compensate for the fact that it is the highest SD
portfolio; conversely, although Q is the least volatile, its Sharpe ratio is the lowest.
The Sharpe ratio has a clear interpretation, namely, the incremental return an investor may
expect for every increase of 1% of standard deviation. It is the slope of the capital allocation
line supported by that portfolio. But should investors consider the difference in Sharpe ratios
between portfolio P and the benchmark portfolio M (.398 2 .344 5 .054) large? That is
harder to interpret and leads us to a variant on the Sharpe ratio.
Imagine a portfolio with the same standard deviation as the benchmark, sM. Then the differ-
ence between the Sharpe ratios of the portfolio and the benchmark would be the difference in
their risk premiums divided by that common standard deviation. Put differently, ranking portfo-
lios with a common volatility by Sharpe ratio will be equivalent to ranking them very simply by
risk premium—estimated from the sample excess returns, as in Equation 18.5. This makes com-
parison of portfolios with equal standard deviation easy to interpret.
Can we transform P to an equivalent portfolio with the same standard deviation as the
benchmark, sM , without affecting its Sharpe ratio? Yes, we can: Recall that the slope of P ’s
CAL is the Sharpe ratio of all portfolios on that line. Therefore, we just choose the portfolio on
CALP that has standard deviation sM . All portfolios on CALP are mixtures of portfolio P with
risk-free borrowing or lending. When we invest a weight w in P and 1 2 w in the risk-free
asset, we just slide up (when w . 1) or down (when w , 1) the CAL. Call P * the portfolio
created by mixing P with the risk-free asset in just the right proportion to make the standard
deviation match that of the benchmark. In other words, portfolio P * is portfolio P with just the
right amount of leverage to make the standard deviation match that of the benchmark.
We form P * by choosing w 5 sM /sP because this makes the SD of P * equal to wsP 5 sM .
The risk premium of portfolio P * therefore can be written in terms of the Sharpe ratio of P:
sM
RP * 5 wRP 5 R 5 sMSP
sP P
Similarly, the risk premium of the benchmark can be written in terms of its Sharpe ratio:
M-square (M2) RM
Return difference between a RM 5 sM 5 sMSM
sM
managed portfolio leveraged
to match the volatility of a The difference between the risk premium of P *, the leverage-adjusted version of P, and the
passive index and the return benchmark is known as M-square (after Leah and Franco Modigliani)3 and is written M 2.
on that index.
3
The M-square measure was developed independently by Graham and Harvey (1997) and by Modigliani and
Modigliani (1997).

bod34698_ch18_595-[Link] 600 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 601

FIGURE 18.2
M 2 of portfolio P

14

CALP
12
CALM

10 P
Average excess return (%)

P∗
8
7.37%
M2 5 7.37 2 6.37 5 1%
6.37% M2 5 (.398 2 .344)18.5 5 1%
6 M

4
SP 5 .398

2
SM 5 .344 18.5% 24.1%

0
0 5 10 15 20 25 30 35 40
Standard deviation (%)

M 2 5 RP * 2 RM 5 sM (SP 2 SM) (18.6)

M-square is the rate-of-return differential between P * and M, a legitimate and

easy-to-interpret performance measure because the portfolios are volatility-matched.
Equation 18.6 shows us that it also is a simple transformation of the difference between
their Sharpe ratios. Table 18.1 and Figure 18.2 show this vividly.
Suppose that instead of investing all its funds in P, the endowment had invested only
sM/sP 5 18.5/24.1 5 .7676 or 76.76% of its funds, with the remainder placed in risk-free
assets. The average excess return on this P * would have been .7676 3 9.6% 5 7.37%. Thus
M-square of P is 7.37% 2 6.37% 5 1%. We also could find this measure directly from the
difference in Sharpe ratios: 18.5% 3 (.398 2 .344) 5 1%. (For practice, verify that the
M-square of portfolio Q is .72%.)

Performance Evaluation of Fund of Funds

Using the Treynor Measure
We have used the Sharpe ratio, or its variant M-square, to choose between an actively man-
aged portfolio competing with a passive benchmark as the sole risky position for an endow-
ment. But some funds are so large that they engage several managers to run risky component
portfolios. For example, CalPERS (the California Public Employee Retirement System) is
a large pension fund with around $220 billion to invest in September 2011. Like many
large plans, it uses a fund of funds approach, allocating the endowment among a number of fund of funds
professional managers (funds) based in part on performance. This requires a different per- Mutual funds or hedge funds
formance measure. that invest in other funds.
To see why, suppose CalPERS considers two managers, and it so happens that both
establish portfolios with average returns, beta, and residual standard deviation equal to those
of portfolio Q in Table 18.1. If Q were considered as the sole investment portfolio, the
endowment fund would reject it because it has a negative M-square of 2.72%. But if the
residuals of the two Q-like portfolios are uncorrelated, the residual SD of an equally weighted
portfolio of the two would be only "2 3 (1/2 3 15.44)2 5 10.92%. In turn, the total

bod34698_ch18_595-[Link] 601 09/08/12 7:35 PM

 Confirming Pages

602 Part SIX Active Investment Management

portfolio SD would be "(bsM)2 1 Var(e) 5 "(.5 3 18.5)2 1 10.922 5 14.31%. With the
same average return as Q, the Sharpe ratio of the combined portfolio is 5.5/14.31 5 .384,
and its M-square is positive: 18.5(.384 2 .344) 5 .74%. The improvement is due to the ben-
efits of diversification that arise when we combine the two funds into an equally weighted
portfolio. With residual risk lessened due to diversification, the trade-off of excess return to
total volatility is enhanced.
This exercise suggests that for a fund of funds, where residual risk can be largely diversified
away, we should compare average excess return to nondiversifiable or systematic, rather than
total, risk. Since beta measures systematic risk, Treynor (1965) proposed the following mea-
sure, since named after him:

R
T5 (18.7)
b
Treynor measure As Table 18.1 demonstrates, the Treynor measure can differ from Sharpe’s, suggesting that
Ratio of portfolio excess the proper performance measure depends on the role of the risky position in the investor’s
return to beta. overall portfolio.

Performance Evaluation of a Portfolio Added to the

Benchmark Using the Information Ratio
Now we consider yet another scenario, one in which an endowment considers adding a posi-
tion in an actively managed portfolio to an already existing passive portfolio. The Central
State University endowment has so far been a passive investor. Presented with a positive alpha
achieved by the managers of P and Q, the board decides to add a position in just one of the
portfolios. Which should it choose? To answer, we must choose the portfolio which, when
combined with the benchmark, generates the higher Sharpe ratio. In Section 6.5 we saw that
information ratio the key to this problem is the information ratio.
Ratio of alpha to the standard We can calculate the Sharpe ratio of the optimized portfolio from the Sharpe ratio of the
deviation of diversifiable risk. benchmark M, and the information ratio of the added portfolio using Equation 6.17, which
we repeat here:

aP 2
SO 5 SQRT BS 2M 1 ¢ ≤ R (18.8)
sP

aP
where is the information ratio of portfolio P . Table 18.1 indicates that the information
sP
ratio of P, .24, is higher than that of Q, .15. Equation 18.8 tells us that the Sharpe ratio of
the optimized portfolio using P will be .42, but only .38 using Q—still better than the
benchmark’s .34. Once again, we see that the role of the evaluated portfolio in the investor’s
complete portfolio determines the choice of performance measure. A different measure can
lead to different judgment of superiority. The following table summarizes our conclusions:

Performance Measure Definition Application

Sharpe Excess return When choosing among portfolios

Standard deviation competing for the overall risky
portfolio
Treynor Excess return When ranking many portfolios
Beta that will be mixed to form the
overall risky portfolio
Information ratio Alpha When evaluating a portfolio to be
Residual standard deviation mixed with the benchmark
portfolio

bod34698_ch18_595-[Link] 602 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 603

The Relation of Alpha to Performance Measures

Alpha, also known as the Jensen measure after Michael Jensen, who first proposed it, appears Jensen measure
everywhere in performance evaluation; why then, did we not present it as a performance The alpha of an investment.
measure? To answer, we must see its relation in the three performance measures.
When short sales are allowed, a negative alpha is as good as, or even better than, a positive
one when constructing the optimal portfolio. A negative, or short, position in a negative-alpha
stock will turn the alpha positive. If the stock’s beta is positive, a negative position in it will also
reduce systematic and therefore overall risk. Thus, a negative-alpha, positive-beta stock serves
double duty in optimizing a portfolio.4 However, if short sales are prohibited, a negative-alpha
stock is no better than a zero-alpha one because it must be ignored.
The foregoing not withstanding, when it comes to performance evaluation, we judge
ex-post (after the fact) portfolio returns. We cannot directly observe the manager’s ex-ante
(before the fact) expectations. We know that the manager must have judged the alpha positive
ex-ante. But forecasting errors or just bad luck could have driven an ex-ante positive-alpha
portfolio into negative-alpha territory. Thus, in performance evaluation, a negative realized
alpha must be taken to indicate below-average performance.
The relation of the Jensen measure to the Sharpe and Treynor measures can be gleaned
from Equation 18.2. Substituting the right-hand side of the equation for the average excess
return and employing some manipulation, we find that the Sharpe measure is

RP bP R M aP sP
SP 5 5 1 ; bP 5 r
sP sP sP sM
aP
SP 5 SM r 1 (18.9)
sP
aP
SP 2 SM 5 SM (r 2 1) 1
sP
where r is the correlation between the excess return of P and the benchmark. First, observe
that alpha alone does not determine which portfolio has a larger Sharpe ratio. The standard
deviation of P and its correlation with the benchmark are also important. Thus positive alpha
is not a sufficient condition for a managed portfolio to offer a higher Sharpe measure than the
passive benchmark.
While it is not sufficient, a positive alpha is necessary to obtain a higher Sharpe ratio than
the benchmark’s SM, because SM (r 2 1) is negative. Superior performance in this context is
a stiff challenge because, to achieve a positive alpha, it is necessary to construct a portfolio
that is different from the benchmark. But this, in turn, will increase residual risk (which low-
ers the correlation coefficient) and offset the improvement in alpha. Notice that portfolio Q
has a larger alpha, 2.3%, than P , 1.6%. Moreover, its ratio of alpha to standard deviation
(2.3/18 5 .128) is far greater than P ’s (1.6/24.1 5 .066). Despite all this, Q ’s Sharpe ratio is
smaller because its correlation coefficient with M is low (.51) compared with P ’s (.96).
The Treynor measure, which measures performance of a portfolio within a fund of funds,
also is related to the portfolio alpha via Equation 18.2 as follows:

RP bPRM 1 aP aP
TP 5 5 5 RM 1
bP bP bP
bM 5 1 TM 5 RM (18.10)

aP
TP 2 TM 5
bP

4
Since a negative-beta stock is a rarity, negative alpha is generally better than a positive one when short sales
are allowed.

bod34698_ch18_595-[Link] 603 09/08/12 7:35 PM

 Confirming Pages

604 Part SIX Active Investment Management

aP
RM is common to all portfolios; therefore, the relative rank of TP is determined by the ratio .
bP
Thus here, too, a positive alpha is necessary but not sufficient to rank alternative active portfo-
lios; we also need to know beta.
Finally, to complete the list, a positive alpha is needed to increase the Sharpe ratio of any
portfolio to which the measured one is added. At the stage of constructing a portfolio, negative-
alpha securities are useful because one can take a short position in them. But in performance
evaluation we are asking whether the realized return on a manager’s portfolio suggests we should
employ the manager in the future to construct a piece of our overall portfolio. We are willing to
do so only if the manager’s forecasts have translated to realized returns with a positive alpha.
However, here, too, alpha alone cannot rank portfolios, since a portfolio with lower alpha but
also lower residual risk still can be judged of better overall performance (Sharpe ratio). We can be
sure, though, that a negative alpha indicates inferior performance by all performance measures.

CONCEPT
c h e c k 18.1 Consider the following data for a particular sample period when returns were high:

Portfolio P Market M
Average return 35% 28%
Beta 1.2 1.0
Standard deviation 42% 30%

Calculate alpha and the three performance measures for portfolio P and the market. The T-bill
rate during the period was 6%. By which measures did portfolio P outperform the market?

Alpha Capture and Alpha Transport

In the next chapter, we will see that many hedge funds seek positive alpha with zero beta.
alpha capture They wish to obtain abnormal returns without taking a stance on the direction of the broad
Construction of a positive- market. Even if a portfolio is relatively underpriced, it may still suffer losses if it falls with the
alpha portfolio with all market. The solution is to hedge out the market exposure of the portfolio by selling either the
systematic risk hedged away. stock index or stock-index futures. This long stock–short index strategy provides a market
alpha transfer or alpha neutral position while maintaining the positive alpha and is therefore called alpha capture.
transport With the captured alpha, you can establish any desired sensitivity to particular market sectors
Establishing alpha while
using index products such as ETFs.
using index products both to This last procedure is called alpha transfer or alpha transport, because you transfer alpha
hedge market exposure and from the sector where you find it to the market sector in which you seek exposure. Finding
to establish exposure to alpha requires skill. By contrast, beta, or market exposure, is a “commodity product” that can
desired sectors. be supplied cheaply through index funds.

EXAMPLE 18.1 Zeta, a portfolio manager, established a positive-alpha portfolio P with a positive exposure to the
market index: bPM 5 1.3. Now she wishes to transfer the alpha. Her objective is a portfolio that is
Alpha Capture market neutral but with positive exposure to the health care sector. In other words, she wants to
and Transport “transport” her positive-alpha portfolio from a broad market exposure to a narrow health care expo-
sure, a sector she believes will outperform. Her goal is a zero-net-investment position with a beta of
zero on the market index but with a beta of .5 on a health care sector index.
We call Zeta’s final portfolio Z, which will be constructed from positions in the original positive-alpha
portfolio P, the market index portfolio M, the health care index portfolio H, and the risk-free asset F.
Zeta will first isolate alpha by neutralizing P’s market beta. She will then use a health care sector index
portfolio to establish her desired exposure to health care. In the end, she wants her final portfolio Z to
have a zero beta on the broad market, b ZM 5 0, and a beta of .5 on health care, b ZH 5 .5.
Zeta’s statistical analysis implies that a health care exchange-traded fund, XLV, has a market
beta, bXLV 5 .9. Therefore, as she establishes exposure to the health care portfolio, she will also
(continued)

bod34698_ch18_595-[Link] 604 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 605

take on market exposure, and this too must be hedged away. Therefore, as Table 18.2 shows, EXAMPLE 18.1
she must take a position in the market index sufficiently large to offset the beta of portfolio P as
well as the additional market exposure created by her position in the health care ETF. The hedging Alpha Capture
strategy that creates pure exposure to the health care sector is similar to the hedging of factor and Transport
exposures that we encountered in the discussion of the arbitrage pricing theory (see Tables 7.5 (concluded)
and 7.9).5

An important issue that is often lost when evaluating ex-post alpha is statistical signifi-
cance. After all, even if the true alpha is zero, you expect to estimate a positive alpha in roughly
50% of the evaluated portfolios (and a negative alpha in the other 50%). Given capital market
volatility, it is fair to expect that even truly nonzero alphas often would be statistically insig-
nificant. We would be more inclined to believe a nonzero alpha of a portfolio manager is a real
phenomenon if it persists over time. Take a look at Figure 8.8 of Chapter 8. Unfortunately, the
graph suggests that persistence of alpha is mostly found in negative-alpha portfolios, and little
is evident in portfolios of positive alpha.

Performance Evaluation with a Multi-Index Model

The Fama-French (FF) three-factor model discussed in Section 7.4 has almost completely
replaced the single-index model in academic performance evaluation, and has been gaining
“market share” in the investment services industry.6 Evidence in favor of augmenting the mar-
ket index with the size (SMB) and value (HML) portfolios is compelling. How should this
affect performance evaluation?
Expanding Equation 18.1 to include the size and value factors, we have,7

RPt 5 bP RMt 1 bSMB rSMBt 1 bHML rHMLt 1 aP 1 ePt

RPt 5 bP RMt 1 bSMB r SMBt 1 bHML r HMLt 1 aP (18.11)

Equation 18.11 states that expected return is determined by betas on three factors, not just
by beta relative to the market index. Notice that the index portfolio, M, has zero alpha; if you
regress RM on the three right-hand-side portfolios, it will be completely explained by the first
factor, RM, and so it will have an intercept of zero. The same applies to SMB and HML, and
thus any portfolio formed from one or more index portfolios will have zero alpha.

TABLE 18.2 Alpha capture and transfer to the health care sector

Portfolio Weight* In Asset Contribution to Excess Returns

wP 5 1 P wP (a P 1 b PM RM 1 eP) 5 a P 1 1.3 RM 1 eP
wXLV 5 .5 XLV w XLV R XLV 5 .5(.9 R M 1 e XLV) 5 .45 RM 1 .5e XLV
wM 5 2 b P 2 .5b XLV M w M R M 5 2 1.75 R M
5 21.75
wF 5 21 2 .5 1 1.75 Risk-free 0

0 Portfolio Z a P 1 e P 1 .5e XLV

*If P ’s alpha is negative, then reverse the sign of wP and adjust the signs of wM and wF.

5
However, in this application, portfolio Z is not an arbitrage portfolio; it is not likely to be even approximately well
diversified. The idea is to hedge all systematic exposures except for that to health care specific risk.
6
The three FF factors (market, SMB, and HML) sometimes are augmented by a momentum portfolio (long in recent
losers and short in recent gainers) and/or by a liquidity portfolio (long in liquid and short in illiquid stocks).
7
Notice that we replace uppercase R (which usually denotes an excess return relative to the risk-free rate) with lower-
case r for the SMB and HML factors because these portfolios already are excess returns, for example, small-stock
returns over large-stock returns. These are zero-net-investment portfolios (for example, long small stocks and short
large stocks), and thus have an opportunity cost of zero rather than r f.

bod34698_ch18_595-[Link] 605 09/08/12 7:35 PM

 Confirming Pages

E X C E L Performance Measures
APPLICATIONS

The Excel model “Performance Measures” calculates all of the performance measures discussed in this chap-
ter. The model available on our website is built to allow you to compare eight different portfolios and to rank
Please visit us at them on all measures discussed in this chapter.
[Link]/bkm
A B C D E F G H I J K
1 Performance Measurement
2
3
4
Average Standard Beta Unsystematic Sharpe Treynor Jensen 2 2 Appraisal
5 M T
6 Fund Return Deviation Coefficient Risk Ratio Measure Alpha Measure Measure Ratio
7 Alpha .2800 .2700 1.7000 .0500 0.8148 .1294 -.0180 -.0015 -.0106 -0.3600
8 Omega .3100 .2600 1.6200 .0600 0.9615 .1543 .0232 .0235 .0143 0.3867
9 Omicron .2200 .2100 0.8500 .0200 0.7619 .1882 .0410 -.0105 .0482 2.0500
10 Millennium .4000 .3300 2.5000 .2700 1.0303 .1360 -.0100 .0352 -.0040 -0.0370
11 Big Value .1500 .1300 0.9000 .0300 0.6923 .1000 -.0360 -.0223 -.0400 -1.2000
12 Momentum Watcher .2900 .2400 1.4000 .1600 0.9583 .1643 .0340 .0229 -.0243 0.2125
13 Big Potential .1500 .1100 0.5500 .0150 0.8182 .1636 .0130 -.0009 .0236 0.8667
14 S&P Index Return .2000 .1700 1.0000 .0000 0.8235 .1400 .0000 .0000 .0000 0.0000
15 T-Bill Return .06 0
16
17 Ranking by Sharpe
18 Return S.D. Beta Unsy. Risk Sharpe Treynor Jensen M2 T2 Appraisal
19 Millennium .4000 .3300 2.5000 .2700 1.0303 .1360 -.0100 .0352 -.0040 -0.0370
20 Omega .3000 .2600 1.6200 .0600 0.9615 .1543 .0232 .0235 .0143 0.3867
21 Momentum Watcher .2900 .2400 1.4000 .1600 0.9583 .1643 .0340 .0229 .0243 0.2125
22 S&P Index Return .2000 .1700 1.0000 .0000 0.8235 .1400 .0000 .0000 .0000 0.0000
23 Big Potential .1500 .1100 0.5500 .0150 0.8182 .1636 .0130 -.0009 .0236 0.8667
24 Alpha .2800 .2700 1.7000 .0500 0.8148 .1294 -.0180 -.0015 -.0106 -0.3600
25 Omicron .2200 .2100 0.8500 .0200 0.7619 .1882 .0410 -.0105 .0482 2.0500
26 Big Value .1500 .1300 0.9000 .0300 0.6923 .1000 -.0360 -.0223 -.0400 -1.2000
27
28 Ranking by Treynor
29 Return S.D. Beta Unsy. Risk Sharpe Treynor Jensen 2 2 Appraisal
M T
30 Omicron .2200 .2100 0.8500 .0200 0.7619 .1882 .0140 -.0105 .0482 2.0500
31 Momentum Watcher .2900 .2400 1.4000 .1600 0.9583 .1643 .0340 .0229 .0243 0.2125
32 Big Potential .1500 .1100 0.5500 .0150 0.8182 .1636 .0130 -.0009 .0236 0.8667
33 Omega .3100 .2600 1.6200 .0600 0.9615 .1543 .0232 .0235 .0143 0.3867
34 S&P Index Return .2000 .1700 1.0000 .0000 0.8235 .1400 .0000 .0000 .0000 0.0000
35 Millennium .4000 .3300 2.5000 .2700 1.0303 .1360 -.0100 .0352 -.0040 -0.0370
36 Alpha .2800 .2700 1.7000 .0500 0.8148 .1294 -.0180 -.0015 -.0106 -0.3600

Excel Questions
1. Examine the performance measures of the funds included in the spreadsheet. Rank the funds by the five
performance measures. Are the rankings across funds consistent? What explains these results?
2. Which fund would you choose if you were considering investing the entire risky portion of your portfolio?
What if you were considering adding a small position in one of these funds to a portfolio invested in the
market index?

In a three-factor security market as described by Equation 18.11, the market index is no

longer the single efficient portfolio. Instead, we can use as our benchmark (default) portfolio
the combination of factor portfolios that maximizes the Sharpe ratio. Once this benchmark is
identified, the Treynor-Black method can be deployed: First identify the optimal active port-
folio based on alpha values from security analysis; then mix the active portfolio with the said
benchmark to find the optimal risky portfolio. This implies that the information ratio from
the multifactor equation is the appropriate performance measure for an active portfolio to be
added to the multifactor benchmark.
Failure to recognize the multi-index equation (when valid) in favor of a misspecified single-
index equation can lead one to overestimate performance. Apparent alpha values that reflect
the impact of omitted factors will be mistaken for superior performance.
Recent research by Cremers, Petajisto, and Zitzewitz (2010) shows that indexes such as the
S&P 500 and Russell 2000 demonstrate significant nonzero alphas when evaluated using the
FF model even when a momentum factor is added. The problem in finding adequate passive
benchmarks tells us that performance evaluation is really (after more than 40 years) still in its
infancy and our inferences should elicit some healthy skepticism.
606

bod34698_ch18_595-[Link] 606 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 607

18.2 STYLE ANALYSIS

Style analysis was introduced by Nobel Laureate William Sharpe (1992). The popularity of the
concept was aided by a widely cited study (Brinson et al., 1991) concluding that 91.5% of
the variation in returns of 82 mutual funds could be explained by the funds’ asset allocation to
bills, bonds, and stocks. Later studies that considered asset allocation across a broader range of
asset classes found that as much as 97% of fund returns can be explained by asset allocation alone.
Sharpe considered 12 asset class (style) portfolios. His idea was to regress fund returns on
indexes representing a range of asset classes. The regression coefficient on each index would
then measure the implicit allocation to that “style.” Because funds are barred from short posi-
tions, the regression coefficients are constrained to be either zero or positive and to sum to
100%, so as to represent a complete asset allocation. The R-square of the regression would then
measure the percentage of return variability due to style choice rather than security selection.
Finally, in this regression there is no intercept, and residuals are not constrained to sum to zero.
This sum equals the total return from security selection. This feature allows us to track the cumu-
lative residual and observe how return from security selection evolves over time.
To illustrate the approach, consider Sharpe’s study of the monthly returns on Fidelity’s
Magellan Fund over the period January 1985 through December 1989, shown in Table 18.3.
While there are 12 asset classes, each one represented by a stock index, the regression coeffi-
cients are positive for only four of them. We can conclude that the fund returns are well
explained by only four style portfolios. Moreover, these four style portfolios alone explain
97.3% of the variance of returns.
The proportion of return variability not explained by asset allocation can be attributed to
security selection within asset classes. For Magellan, this was 100 2 97.3 5 2.7%. To evaluate
the average contribution of stock selection to fund performance we track the residuals from the
regression, displayed in Figure 18.3. The figure plots the cumulative effect of these residuals; the
steady upward trend confirms Magellan’s success at stock selection in this period. Notice that
the plot in Figure 18.3 is far smoother than the plot in Figure 18.4, which shows Magellan’s
performance compared to a standard benchmark, the S&P 500. This reflects the fact that the
regression-weighted index portfolio tracks Magellan’s overall style much better than the S&P
500. The performance spread is much noisier using the S&P as the benchmark.

TABLE 18.3 Sharpe’s style portfolios for the Magellan fund

Regression Coefficient*

Bills 0
Intermediate bonds 0
Long-term bonds 0
Corporate bonds 0
Mortgages 0
Value stocks 0
Growth stocks 47
Medium-cap stocks 31
Small stocks 18
Foreign stocks 0
European stocks 4
Japanese stocks 0
Total 100
R-squared 97.3%

*Regressions are constrained to have nonnegative coefficients and to have coefficients that sum to 100%.
Source: William F. Sharpe, “Asset Allocation: Management Style and Performance Evaluation,”
Journal of Portfolio Management, Winter 1992, pp. 7–19. Used with permission of Institutional
Investor, Inc., [Link]. All Rights Reserved.

bod34698_ch18_595-[Link] 607 09/08/12 7:35 PM

 Confirming Pages

608 Part SIX Active Investment Management

FIGURE 18.3

Fidelity Magellan Fund 30

cumulative return difference:
Fund versus style 25
benchmark
Source: William F. Sharpe, 20
“Asset Allocation: Manage-
ment Style and Performance
Evaluation,” Journal of 15
Portfolio Management, Winter
1992, pp. 7–19. Figure 17, 10
p. 18. Used with permission
of Institutional Investor, Inc.,
5
[Link]. All
Rights Reserved.
0
1986 1987 1988 1989 1990

FIGURE 18.4
Fidelity Magellan Fund 12
cumulative return difference:
Fund versus S&P 500 10
Source: William F. Sharpe,
“Asset Allocation: Manage- 8
ment Style and Performance
Evaluation,” Journal of 6
Portfolio Management, Winter
1992, pp. 7–19. Figure 16, 4
p. 17. Used with permission
of Institutional Investor, Inc.,
2
[Link]. All
Rights Reserved.
0

22
1986 1987 1988 1989 1990

Of course, Magellan’s consistently positive residual returns (reflected in the steadily increas-
ing plot of cumulative return difference) is hardly common. Figure 18.5 shows the frequency
distribution of average residuals across 636 mutual funds. The distribution has the familiar
bell shape with a slightly negative mean of 2.074% per month.
Style analysis has become very popular in the investment management industry and has
spawned quite a few variations on Sharpe’s methodology. Many portfolio managers utilize
websites that help investors identify their style and stock selection performance. The nearby
box shows that style analysis is at the heart of recent debates about the investment perfor-
mance of hedge funds.

18.3 MORNINGSTAR’S RISK-ADJUSTED RATING

The commercial success of Morningstar, Inc., the premier source of information on mutual
funds, has made its Risk Adjusted Rating (RAR) among the most widely used performance
measures. The Morningstar five-star rating is coveted by the managers of the thousands of
funds covered by the service.

bod34698_ch18_595-[Link] 608 09/08/12 7:35 PM

 Confirming Pages

On the MARKET FRONT

WHAT’S IT ALL ABOUT, ALPHA? However, it is also possible to take the opposite tack. This type of
analysis gives managers no credit for choosing the systematic fac-
Too many notes. That’s what Emperor Joseph II famously said to tors—the betas—that drive their portfolios. Yes, these betas could
Mozart on seeing his opera “The Marriage of Figaro.” But surely to often have been bought for very low fees. But would an investor have
think of a musical work as just a series of notes is to miss the magic. been able to put them together in the right combination?
Could the same be said about fund management? It is the fash- It is as if a diner in Gordon Ramsay’s restaurants were brave
ion these days to separate beta (the systematic return delivered by enough to tell the irascible chef: “This meal was delicious. But chemical
the market) from alpha (the manager’s skill). Investors are happy to analysis shows it is 65% chicken, 20% carrot, 10% flour and 5% milk.
pay high fees for the skill, but regard the market return as a com- I could have bought those ingredients for £1.50. Why should I pay
modity. Distinguishing the two is, however, sometimes difficult. £20?” The chef’s reply, shorn of its expletives, might be: “The secret is
A fund manager might beat the market because of luck or reck- in the mixing.” This debate matters because people are now trying to
lessness, rather than skill, for example. Suppose he has packed his replicate the performance of hedge funds with cloned portfolios.
portfolio with oil stocks and then profits when the price of crude There are two potential criticisms of the cloned approach. One is
rises. More generally, alpha skeptics often attribute abnormal returns that it will simply reproduce all the systematic returns that hedge
to “style bias,” such as [the manager who favors stocks with an funds generate and none of their idiosyncratic magic. However, this
energy focus. Popular style biases are often based on factors that “magic” is hard to pin down, and even if it does exist, it may be worth
seem to have predicted past alpha, such as firm size.] But should the no more than the fees hedge funds charge.
skeptics be biased against style bias? After all, the only portfolio The second criticism is that the clones will always be a step
utterly free of bias would be one that included the entire market. behind the smart money. You cannot clone a hedge fund until you
Academics have entered this debate, trying to pin down the factors know where it has been. But by then it may have moved on.
that drive a fund’s performance. Bill Fung and Narayan Naik of London Mozart might have sympathized. His operas were more than the
Business School have come up with a seven-factor model which, they sum of his notes. But even if the great composer had no peers, he
say, can explain the bulk of hedge-fund performance. After allowing for has had plenty of imitators.
these factors, the average fund of hedge funds has not produced any
alpha in the past decade, except during the dot-com bubble. This SOURCE: Excerpted from The Economist, March 22, 2007. © The Economist
approach suggests the whole idea of alpha might be an illusion. Newspaper Limited, London. Used with permission via Copyright Clearance Center.

FIGURE 18.5
90 Average tracking error, 636
mutual funds, 1985–1989
80
Source: William F. Sharpe,
70 “Asset Allocation: Manage-
ment Style and Performance
60 Evaluation,” Journal of
Portfolio Management, Winter
50 1992, pp. 7–19. Figure 18,
p. 18. Used with permission
40
of Institutional Investor, Inc.,
30 [Link]. All
Rights Reserved.
20

10

0
21.00

20.50

0.00

0.50

1.00

Average tracking error (%/month)

Morningstar calculates a number of RAR performance measures that are similar, although not
identical, to the standard mean-variance measures (see Chapter 4 for a more detailed discussion).
The most distinct measure, the Morningstar Star Rating, is based on comparison of each fund to
a peer group. The peer group for each fund is selected on the basis of the fund’s investment

609

bod34698_ch18_595-[Link] 609 09/08/12 7:35 PM

 Confirming Pages

610 Part SIX Active Investment Management

FIGURE 18.6

Rankings based on Sharpe ratio

Morningstar’s category percentile in category
RARs and excess return 1 ⫹⫹⫹ ⫹
⫹⫹⫹⫹
⫹ ⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹ ⫹⫹ ⫹ ⫹⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹ ⫹
⫹ ⫹⫹
⫹
⫹
⫹⫹ ⫹⫹
⫹⫹
Sharpe ratios ⫹ ⫹
⫹⫹⫹
⫹
⫹
⫹⫹⫹
⫹ ⫹⫹⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹ ⫹
⫹
⫹⫹⫹⫹⫹⫹⫹⫹⫹⫹
⫹ ⫹⫹
⫹ ⫹⫹⫹
⫹
⫹ ⫹⫹
⫹⫹ ⫹⫹
⫹
⫹⫹⫹ ⫹
⫹⫹⫹⫹ ⫹⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹⫹
⫹⫹ ⫹⫹
Source: William F. Sharpe 0.8 ⫹⫹ ⫹⫹⫹
⫹⫹⫹⫹⫹⫹
⫹⫹
⫹
⫹⫹⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹⫹ ⫹⫹⫹ ⫹ ⫹
⫹ ⫹⫹ ⫹⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹⫹ ⫹⫹⫹
⫹ ⫹⫹⫹⫹⫹
⫹ ⫹⫹
⫹
⫹⫹⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹⫹⫹
⫹
⫹⫹ ⫹ ⫹⫹
⫹ ⫹⫹
(1997), “Morningstar ⫹⫹⫹⫹⫹
⫹⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹⫹⫹⫹ ⫹ ⫹
⫹ ⫹⫹ ⫹⫹⫹ ⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹ ⫹ ⫹
Performance Measures,” www. ⫹⫹⫹⫹⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹ ⫹ ⫹
0.6 ⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹⫹
⫹
⫹⫹ ⫹ ⫹
⫹⫹⫹ ⫹⫹⫹
⫹
⫹⫹⫹
⫹⫹
⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹⫹⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹⫹⫹⫹⫹
[Link]/+ wfsharpe/art/ ⫹⫹⫹⫹⫹
⫹⫹
⫹⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹ ⫹⫹
⫹ ⫹⫹ ⫹⫹⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹⫹ ⫹
stars/[Link]. Used with ⫹⫹⫹⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹⫹⫹⫹⫹ ⫹
⫹⫹⫹⫹ ⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹
⫹⫹⫹
⫹
⫹⫹⫹⫹⫹⫹
⫹⫹ ⫹⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹⫹⫹⫹⫹ ⫹
permission. 0.4 ⫹⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹⫹ ⫹
⫹
⫹⫹ ⫹ ⫹ ⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹⫹
⫹⫹
⫹⫹
⫹ ⫹
⫹⫹ ⫹⫹⫹⫹⫹⫹⫹
⫹
⫹⫹
⫹⫹
⫹ ⫹
⫹⫹
⫹
⫹ ⫹⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹⫹
⫹
⫹
⫹
⫹
⫹⫹⫹
⫹
⫹
⫹⫹⫹
⫹⫹⫹⫹
⫹ ⫹
⫹⫹⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹⫹⫹
⫹⫹⫹
⫹ ⫹⫹ ⫹⫹ ⫹⫹⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹ ⫹
0.2 ⫹ ⫹⫹ ⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹⫹⫹
⫹
⫹ ⫹ ⫹
⫹⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹ ⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹⫹
⫹⫹⫹
⫹⫹⫹⫹⫹ ⫹ ⫹
⫹⫹ ⫹⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹⫹
⫹
⫹⫹
⫹⫹ ⫹ ⫹
⫹
⫹
⫹⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹
⫹⫹
⫹
⫹⫹
⫹
⫹⫹⫹⫹⫹ Category RAR
⫹
0 percentile in
0 0.2 0.4 0.6 0.8 1 category

universe (e.g., international, growth versus value, fixed-income) as well as portfolio characteristics
such as average price-to-book value, price–earnings ratio, and market capitalization.
Morningstar computes fund returns (adjusted for loads) as well as a risk measure based on
fund performance in its worst years. The risk-adjusted performance is ranked across funds in
a style group, and stars are awarded based on the following table:

Percentile Stars

0–10 1
10–32.5 2
32.5–67.5 3
67.5–90 4
90–100 5

The Morningstar RAR method produces results that are similar but not identical to that of
the mean/variance-based Sharpe ratios. Figure 18.6 demonstrates the fit between ranking by
RAR and by Sharpe ratios from the performance of 1,286 diversified equity funds over the
period 1994–1996. Sharpe notes that this period is characterized by high returns that contrib-
ute to a good fit.

18.4 RISK ADJUSTMENTS WITH CHANGING

PORTFOLIO COMPOSITION
One potential problem with risk-adjustment techniques is that they all assume that portfolio
risk, whether it is measured by standard deviation or beta, is constant over the relevant time
period. This isn’t necessarily so. If a manager attempts to increase portfolio beta when she
thinks the market is about to go up and to decrease beta when pessimistic, both the standard
deviation and the beta of the portfolio will change over time. This can wreak havoc with our
performance measures.

EXAMPLE 18.2 Suppose the Sharpe measure of the passive strategy (investing in a market-index fund) is .4. A port-
folio manager is in search of a better, active strategy. Over an initial period of, say, four quarters, he
Risk Measurement executes a low-risk or defensive strategy with an annualized mean excess return of 1.5% and a stan-
with Changing dard deviation of 3.4%. This makes for a Sharpe measure of .44, which beats the passive strategy.
Portfolio Composition
(continued)

bod34698_ch18_595-[Link] 610 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 611

Over the next period of another four quarters, this manager finds that a high-risk strategy is optimal, EXAMPLE 18.2
with an annual mean excess return of 8.75% and standard deviation of 20%. Here again the Sharpe
measure is .44. Over the two years, our manager maintains a better-than-passive Sharpe measure. Risk Measurement
Figure 18.7 shows a pattern of (annualized) quarterly returns that is consistent with our description with Changing
of the manager’s strategy over two years. In the first four quarters, the excess returns are 23%, 5%, 1%, Portfolio Composition
and 3%, consistent with the predicted mean and SD. In the next four quarters, the excess returns are
(concluded)
29%, 27%, 25%, and 28%, also consistent with predictions for the higher-volatility period. Thus,
each year exhibits a Sharpe measure of .44.
But if we treat the eight-quarter sequence as a single measurement period instead of two indepen-
dent periods, the portfolio’s mean and standard deviation over the full period are 5.125% and 13.8%
respectively, resulting in a Sharpe measure of only .37, apparently inferior to the passive strategy!

What went wrong in FIGURE 18.7

Example 18.2? Sharpe’s ratio
Annualized returns (%) Portfolio returns. In the first
does not recognize the shift
30 four quarters, the firm follows
in the mean from the first a low-risk, low-return policy.
four quarters to the next as a 25 In the next four quarters, it
result of a strategy change. shifts to a high-risk, high-
Instead, the difference in 20 return policy.
mean returns in the two years
adds to the appearance of vol- 15
atility in portfolio returns.
The change in mean returns 10
across time periods contrib-
5
uted to the variability of
returns over the same period. 0 Quarter
Unfortunately, an outside
observer cannot tell that –5
policy changes within the
sample period are the source –10
1 2 3 4 5 6 7 8
of some of the return vari-
ability. Therefore, the active
strategy with shifting means
appears riskier than it really is, which biases the estimate of the Sharpe measure downward.
When assessing the performance of actively managed portfolios, it is important to keep
track of portfolio composition and changes in portfolio mean return and risk. We will see
another example of this problem when we turn to market timing.
Another warning: When we address the performance of mutual funds selected because they
have been successful, we need to be highly cautious in evaluating their track records. In par-
ticular, we need to recognize that even if all managers were equally skilled, a few “winners”
would emerge by sheer chance each period. With thousands of funds in operation, the best-
performing funds will have been wildly successful, even if these results reflect luck rather than
skill. The nearby box addresses this issue. Another manifestation of selection bias arises when
we limit a sample of funds to those for which returns are available over an entire sample
period. This practice implies that we exclude from consideration all funds that were closed
down over the sample period. The ensuing bias is called survivorship bias. It turns out that survivorship bias
when even a small number of funds have failed, the upward bias in the performance of surviv- Upward bias in average
ing funds can be substantial. Most mutual fund databases now include failed funds so that fund performance due to the
samples can be protected from survivorship bias. failure to account for failed
funds over the sample period.

Performance Manipulation
Imagine a manager whose performance is measured over two-year return periods, as in
Example 18.2, for which the Sharpe ratio in each year is .44. Now we are at the end of the

bod34698_ch18_595-[Link] 611 09/08/12 7:35 PM

 Confirming Pages

612 Part SIX Active Investment Management

first year, when the manager’s portfolio has returned the aforementioned four annualized
quarterly rates of 23%, 5%, 1%, and 3%, providing the assumed Sharpe ratio of .44.
At this point the manager identifies the high-risk but better-than-passive strategy of the
example; but he recognizes that following this strategy, he can expect a losing Sharpe ratio of
.37. But what if the portfolio is “de-levered” by shifting 5/6 of its value into bills. With this shift
to safety (and zero excess return), the four second-year returns would be 21.5%, 4.5%, 4.17%,
and 21.33%. Because the Sharpe ratio is invariant to shifts between the risky portfolio and
risk-free asset, the second-year Sharpe ratio is still .44. Despite this, the eight-quarter Sharpe
ratio is now evaluated at .47.8 So far, little damage has been wrought. But given that the pro-
spectuses of many funds promise general investment strategies (for example, investing in
equity), they may not allow significant investments in bills. In that event, managers may reduce
risk by moving into low-beta stocks. This distorts their security selection decisions, and inves-
tors end up with less-than-optimal portfolios due to the manipulation of the Sharpe ratio.
This type of manipulation is only one in a menu that includes investments in derivatives.
These strategies also can be used to manipulate the other performance measures discussed
earlier in the chapter. A manipulation-free performance measure exists, but since it hasn’t yet
penetrated the industry, we leave it for future consideration.

18.5 PERFORMANCE ATTRIBUTION PROCEDURES

Rather than focus on risk-adjusted returns, practitioners often want simply to ascertain which
decisions resulted in superior or inferior performance. Superior investment performance
depends on an ability to be in the “right” securities at the right time. Timing and selection
ability may be considered broadly, such as being in equities as opposed to fixed-income securi-
ties when the stock market is performing well. Or it may be defined at a more detailed level,
such as choosing the relatively better-performing stocks within a particular industry.
Portfolio managers constantly make both broad-brush asset market allocation decisions as
well as more detailed sector and security allocation decisions within markets. Performance
attribution studies attempt to decompose overall performance into discrete components that
may be identified with a particular level of the portfolio selection process.
Attribution analysis starts from the broadest asset allocation choices and progressively focuses
on ever-finer details of portfolio choice. The difference between a managed portfolio’s perfor-
mance and that of a benchmark portfolio may be expressed as the sum of the contributions to
performance of a series of decisions made at the various levels of the construction process. For
example, one common attribution system decomposes performance into three components:
(1) broad asset market allocation choices across equity, fixed-income, and money markets;
(2) industry (sector) choice within each market; and (3) security choice within each sector.
To illustrate this method, consider the attribution results for a hypothetical portfolio. The
portfolio invests in stocks, bonds, and money market securities. The portfolio return over the
period is 5.34%. An attribution analysis appears in Tables 18.4 through 18.7.
The first step is to establish a benchmark level of performance against which performance
bogey ought to be compared. This benchmark is called the bogey. It is the portfolio designed to fit
The benchmark portfolio an the fund prospectus in the absence of any active choice, in other words, the fund’s passive or
investment manager is default portfolio. A fund’s prospectus likely dictates a “normal” range for investment weights,
compared to for performance say 50%–70% in equities, 20%–40% in bonds, and 0%–10% in bills. We will use the middle of
evaluation. these ranges to form the bogey, with a neutral asset allocation of 60% in equities, 30% in
bonds, and 10% in bills, as shown in Table 18.4.
The prospectus also likely specifies benchmarks for each asset class. For instance, the
S&P 500 often serves as the benchmark for equity investments, and the Barclays Capital
U.S. Aggregate Bond Index may be used for fixed income. Table 18.4 shows the returns on

8
The two-year Sharpe ratio can be higher than that of each of the two individual years in part because of the bias
correction to the SD (having to do with degrees of freedom). Instead of multiplying each annual estimate of SD by
4/3, we multiply the two-year estimate by “only” 8/7.

bod34698_ch18_595-[Link] 612 09/08/12 7:35 PM

 Confirming Pages

On the MARKET FRONT

THE MAGELLAN FUND AND MARKET
TABLE 1
EFFICIENCY: ASSESSING THE
PERFORMANCE OF MONEY PROBABILITY DISTRIBUTION OF NUMBER OF
SUCCESSFUL YEARS OUT OF THIRTEEN FOR THE
MANAGERS BEST-PERFORMING MONEY MANAGER
Fidelity’s Magellan Fund outperformed the S&P 500 in eleven of the
thirteen years ending in 1989. Is such performance consistent with Managers in Contest
the efficient market hypothesis? Casual statistical analysis would
Winning Years 50 100 250 500
suggest not.
If outperforming the market were like flipping a fair coin, as would 8 0.1% 0 0 0
be the case if all securities were fairly priced, then the odds of an
9 9.2 0.9 0 0
arbitrarily selected manager producing eleven out of thirteen winning
years would be only about 0.95%, or 1 in 105. The Magellan Fund, 10 47.4 31.9 5.7 0.2
however, is not a randomly selected fund. Instead, it is the fund that 11 34.8 51.3 59.7 42.3
emerged after a thirteen-year “contest” as a clear winner. Given that 12 7.7 14.6 31.8 51.5
we have chosen to focus on the winner of a money management 13 0.8 1.2 2.8 5.9
contest, should we be surprised to find performance far above the
Mean winning
mean? Clearly not.
Once we select a fund precisely because it has outperformed all
years of best
other funds, the proper benchmark for predicted performance is no performer 10.43 10.83 11.32 11.63
longer a standard index such as the S&P 500. The benchmark must
be the expected performance of the best-performing fund out of a
sample of randomly selected funds. Viewed in this context, the performance of Magellan is still
Consider as an analogy a coin flipping contest. If fifty contestants impressive but somewhat less surprising. The simulation shows that
were to flip a coin thirteen times, and the winner were to flip eleven out of a sample of 50 managers, chance alone would provide a
heads out of thirteen, we would not consider that evidence that the 43.3% probability that someone would beat the market at least
winner’s coin was biased. Instead, we would recognize that with fifty eleven out of thirteen years. Averaging over all 10,000 trials, the
contestants, the probability is greater than 40% that the individual mean number of winning years necessary to emerge as most reliable
who emerges as the winner would in fact flip heads eleven or more manager over the thirteen-year contest was 10.43.
times. (In contrast, a coin chosen at random that resulted in eleven Therefore, once we recognize that Magellan is not a fund chosen
out of thirteen heads would be highly suspect!) at random, but a fund that came to our attention precisely because it
How then ought we evaluate the performance of those managers turned out to perform so well, the frequency with which it beat the
who show up in the financial press as (recently) superior performers. market is no longer high enough to be considered a violation of mar-
We know that after the fact some managers will have been lucky. ket efficiency. Indeed, using the conventional 5% confidence level,
When is the performance of a manager so good that even after we could not reject the hypothesis that the consistency of its perfor-
accounting for selection bias—the selection of the ex post winner— mance was due to chance.
we still cannot account for such performance by chance? The other columns in Table 1 present the frequency distributions
of the winning number of successful coin flips (analogously, the num-
SELECTION BIAS AND PERFORMANCE ber of years in which the best-performing manager beats an efficient
BENCHMARKS market) for other possible sample sizes. Not surprisingly, as the pool
of managers increases, the predicted best performance steadily gets
Consider this experiment. Allow fifty money managers to flip a coin better. By providing as a benchmark the probability distribution of the
thirteen times, and record the maximum number of heads realized by best performance, rather than the average performance, the table
any of the contestants. (If markets are efficient, the coin will have the tells us how many grains of salt to add to reports of the latest invest-
same probability of turning up heads as that of a money manager ment guru.
beating the market.) Now repeat the contest, and again record the
winning number of heads. Repeat this experiment 10,000 times. SOURCE: Alan J. Marcus, “The Magellan Fund and Market Efficiency.” The
When we are done, we can compute the frequency distribution of Journal of Portfolio Management, Fall 1990, pp. 85–86. Used with permission
the winning number of heads over the 10,000 trials. of Institutional Investor, Inc., [Link]. All Rights Reserved.
Table 1 (column 1) presents the results of such an experiment
simulated on a computer. The table shows that in 9.2% of the con-
tests, the winning number of heads was nine; in 47.4% of the trials
ten heads would be enough to emerge as the best manager.
Interestingly, in 43.3% of the trials, the winning number of heads
was eleven or better out of thirteen.

613

bod34698_ch18_595-[Link] 613 09/08/12 7:35 PM

 Confirming Pages

614 Part SIX Active Investment Management

TABLE 18.4 Performance of the managed portfolio

Bogey Performance and Excess Return

Return of Index
Component Benchmark Weight during Month (%)

Equity (S&P 500) .60 5.81

Bonds (U.S. Aggregate Index) .30 1.45
Cash (money market) .10 0.48
Bogey 5 (.60 3 5.81) 1 (.30 3 1.45) 1 (.10 3 .48) 5 3.97%
Return of managed portfolio 5.34%
–Return of bogey portfolio 3.97
Excess return of managed portfolio 1.37%

these benchmarks for the relevant period. The neutral asset allocation, along with the
returns on the benchmark indexes, generates the bogey return shown in Table 18.4, 3.97%.
The table also records the actual portfolio return, 5.34%. The difference between actual and
bogey returns, 1.37%, is the excess return of the managed portfolio. We next try to measure
the relative contributions of asset allocation versus security selection decisions to this
advantageous performance.

Asset Allocation Decisions

The managed portfolio is actually invested in the equity, fixed-income, and money markets
with weights of 70%, 7%, and 23%, respectively. The portfolio’s performance could be due to
the departure of this weighting scheme from the benchmark 60/30/10 weights and/or to
superior or inferior results within each of the three broad markets.
To isolate the effect of the manager’s asset allocation choice, we measure the performance
of a hypothetical portfolio that would have been invested in the indexes for each market with
the actual weights of 70/7/23. This return measures the effect of the shift away from the
benchmark 60/30/10 weights without allowing for any effects attributable to active manage-
ment of the securities selected within each market.
Superior performance relative to the bogey is achieved by overweighting investments in
markets that outperform the bogey and by underweighting poorly performing markets. The
contribution of asset allocation to superior performance equals the sum over all markets of the
excess weight in each market times the return of the index for each sector.
Table 18.5A demonstrates that asset allocation contributed 31 basis points to the portfo-
lio’s excess return of 137 basis points. The major contribution of asset allocation to superior
performance in this period comes from the heavy weighting of the equity market when the
equity market has an excellent return of 5.81%.

Sector and Security Selection Decisions

If .31% of the excess performance can be attributed to advantageous asset allocation across
markets, the remaining 1.06% then must be attributable to sector selection and security
selection within each market. Table 18.5B details the contribution of the managed portfolio’s
sector and security selection to total performance.
Panel B shows that the equity component of the managed portfolio has a return of 7.28%
versus a return of 5.81% for the S&P 500. The fixed-income return is 1.89% versus 1.45% for
the Aggregate Bond Index. The superior performance in both equity and fixed-income markets
weighted by the portfolio proportions invested in each market sums to the 1.06% contribution
to performance attributable to sector and security selection.
Table 18.6 documents the sources of the equity market performance by each sector within
the market. The first three columns detail the allocation of funds within the equity market

bod34698_ch18_595-[Link] 614 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 615

TABLE 18.5 Performance attribution

A. Contribution of Asset Allocation to Performance

(1) (2) (3) (4) (5) 5 (3) 3 (4)

Actual Benchmark Index Contribution to
Weight in Weight in Excess Return Performance
Market Market Market Weight (%) (%)

Equity .70 .60 .10 5.81 .5810

Fixed-income .07 .30 2.23 1.45 2.3335
Cash .23 .10 .13 0.48 .0624
Contribution of asset allocation .3099
B. Contribution of Selection to Total Performance

(1) (2) (3) (4) (5) 5 (3) 3 (4)

Portfolio Index Excess
Performance Performance Performance Portfolio Contribution
Market (%) (%) (%) Weight (%)

Equity 7.28 5.81 1.47 .70 1.03

Fixed-income 1.89 1.45 0.44 .07 0.03
Contribution of selection within markets 1.06

TABLE 18.6 Sector allocation within the equity market

(1) (2) (3) (4) (5) 5 (3) 3 (4)

Beginning-of-Month
Weights Contribution
Difference in Sector of Sector
Sector Portfolio S&P 500 Weights Return (%) Allocation (%)

Basic materials 0.0196 0.083 2.0634 6.9 20.437

Business services 0.0784 0.041 .0374 7.0 0.262
Capital goods 0.0187 0.078 2.0593 4.1 20.243
Consumer cyclical 0.0847 0.125 2.0403 8.8 20.355
Consumer noncyclical 0.4037 0.204 .1997 10.0 1.997
Credit sensitive 0.2401 0.218 .0221 5.0 0.111
Energy 0.1353 0.142 2.0067 2.6 20.017
Technology 0.0195 0.109 2.0895 0.3 20.027
Total 1.0000 1.000 .0000 1.290

compared to their representation in the S&P 500. Column (4) shows the rate of return of each
sector, and column (5) equals the product of the difference in the sector weight and the sec-
tor’s performance.
Note that good performance derives from overweighting well-performing sectors such as
consumer noncyclicals, as well as underweighting poorly performing sectors such as technol-
ogy. The excess return of the equity component of the portfolio attributable to sector allocation
alone is 1.29%. As the equity component of the portfolio outperformed the S&P 500 by 1.47%,
we conclude that the effect of security selection within sectors must have contributed an addi-
tional 1.47 2 1.29, or .18%, to the performance of the equity component of the portfolio.
A similar sector analysis can be applied to the fixed-income portion of the portfolio, but we
do not show those results here.

bod34698_ch18_595-[Link] 615 09/08/12 7:35 PM

 Confirming Pages

E X C E L Performance Attribution
APPLICATIONS

The Excel model “Performance Attribution” that is available on our website is built on the example that
appears in Section 18.5. The model allows you to specify different allocations and to analyze the contribution
Please visit us at sectors and weightings for different performances.
[Link]/bkm
A B C D E F
1 Chapter 18
2 Performance Attribution
3 Solution to Question Contribution
4 Bogey Portflio Weight Return on to Portfolio
5 Component Index Benchmark Index Return
6 Equity S&P500 0.6 5.8100% 3.4860%
7 Bonds Aggregate Index 0.3 1.4500% 0.4350%
8 Cash Money Market 0.1 0.4800% 0.0480%
9
10 Return on Bogey 3.9690%
11
12 Contribution
13 Managed Portfolio Portfolio Actual to Portfolio
14 Component Weight Return Return
15 Equity .75 6.5000% 4.8750%
16 Bonds .12 1.2500% 0.1500%
17 Cash .13 0.4800% 0.0624%
18
19 Return on Managed 5.0874%
20
21 Express Return 1.1184%
22
23
24 Contribution of Asset Allocation
25 Actual Weight Benchmark Excess Market Performance
26 Market in Portfolio Weight Weight Return Contribution
27 Equity .75 .6 .15 5.8100% .8715%
28 Fixed Income .12 .3 -.18 1.4500% -.2610%
29 Cash .13 .1 .03 0.4800% .0144%
30 Contribution of
31 Asset Allocation .6249%

Excel Questions
1. What would happen to the contribution of asset allocation to overall performance if the actual weights
had been 70/17/13 in the three markets rather than 75/12/13? Explain your result.
2. Show what would happen to the contribution of security selection to performance if the actual return on
the equity portfolio had been 7.5% instead of 6.5% and the return on the S&P 500 had been 6.81%
instead of 5.81%. Explain your result.

Summing Up Component Contributions

In this attribution period, all facets of the portfolio selection process were successful.
Table 18.7 details the contribution of each aspect of performance. Asset allocation across the
major security markets contributes 31 basis points. Sector and security allocation within
those markets contributes 106 basis points, for total excess portfolio performance of 137
basis points.
The sector and security allocation of 106 basis points can be partitioned further. Sector
allocation within the equity market results in excess performance of 129 basis points, and
security selection within sectors contributes 18 basis points. (The total equity excess perfor-
mance of 147 basis points is multiplied by the 70% weight in equity to obtain the contribution
to portfolio performance.) Similar partitioning could be done for the fixed-income sector.

CONCEPT
c h e c k 18.2 a. Suppose the benchmark weights had been set at 70% equity, 25% fixed-income, and 5%
cash equivalents. What then would be the contributions of the manager’s asset allocation
choices?
b. Suppose the S&P 500 return had been 5%. Recompute the contribution of the manager’s security
selection choices.

616

bod34698_ch18_595-[Link] 616 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 617

TABLE 18.7 Portfolio attribution: summary

Contribution
(basis points)

1. Asset allocation 31.0

2. Selection
a. Equity excess return
i. Sector allocation 129
ii. Security selection 18
147 3 .70 (portfolio weight) 5 102.9
b. Fixed-income
excess return 44 3 .07 (portfolio weight) 5 3.1
Total excess return
of portfolio 137.0

18.6 MARKET TIMING

Pure market timing involves switching funds between the risky portfolio and cash in response market timing
to forecasts of relative performance. To evaluate the potential of a pure market-timing strategy, A strategy that moves funds
consider the fortunes of three families of investors who had $1 to invest on December 1, 1926. between the risky portfolio
Their heirs counted their blessings 82 years later, in 2008. The investment history of the families and cash, based on forecasts
included the Great Depression, a major bear market in 2008 (when the S&P 500 lost 39%), and of relative performance.
seven other recessions in between. The families differed wildly in their investment strategy:
1. Family A invested solely in a money market or cash equivalents.
2. Family B invested solely in stocks (the S&P 500 portfolio), reinvesting all dividends.
3. Family C switched, every month, 100% of its funds between stocks and cash, based on its
forecast of which sector would do better next month.
While the strategies of families A and B are straightforward, that of family C is worth
pondering. Try asking friends: “What would it take to be a perfect market timer?” Many
would venture that to accomplish perfect timing, the timer would need to be able to forecast
the rate of return on stocks at the start of every month. But actually, you wouldn’t need the
precise rate of return: “All” the perfect timer would have to know is whether stocks will out-
perform cash! You might think that such elementary knowledge wouldn’t be worth all that
much. But examine Table 18.8, computed from the actual return history on cash and stocks.
The first panel of Table 18.8 provides the punch line: After 82 years, $1 returned $20 to the
cash fund of family A, and most of those nominal profits were undone by inflation over the
period. Despite the Great Depression and recessions of varying severity, the stock fund of fam-
ily B outdid the cash fund by a factor of more than 80, ending up with $1,626. But the gains
to the perfect-timing family C would have been otherworldly indeed (as was the family’s
power of prediction); the timing fund starting with $1 would have ended with $36.7 billion.

CONCEPT
Use annual rates of return from the Online Learning Center ([Link]/bkm) to repli-
c h e c k 18.3
cate Table 18.8 for the 1926–2008 period for a market timer who could perfectly forecast only
once a year, rather than every month. Why is the performance of the annual timer not as good
as that of the monthly timer?

These results have some lessons for us. The first has to do with the power of compounding.
This effect is particularly important as ever more funds under management represent pension
savings. The horizons of pension investments may not be as long as 82 years, but they are
measured in decades, making compounding an important factor.

bod34698_ch18_595-[Link] 617 09/08/12 7:35 PM

 Confirming Pages

618 Part SIX Active Investment Management

TABLE 18.8 Performance of cash, stocks, and perfect-timing strategies

I. Family fund as of the end of 2008

Family/Strategy

A. Cash B. Stocks C. Perfect Timing

Final proceeds $20 $1,626 $36,699,302,473

II. Annualized monthly rate-of-return statistics (%)

Geometric average 3.71 9.44 34.54

Arithmetic average 3.71 11.48 35.44
Minimum monthly rate* 20.03 228.73 20.03
Maximum monthly rate† 1.52 41.65 41.65
Average excess return 0.00 7.77 31.73
Standard deviation 3.54 19.38 12.44

*Occurred in September 1931.

†
Occurred in April 1933.
Both extreme values occurred during the Great Depression.

The second is a huge difference between the end value of the all-safe asset strategy ($20) and
of the all-equity strategy ($1,626). Why would anyone invest in safe assets? By now you know
the reason: risk. The annual standard deviation of the equity strategy was 19.38%. The high
standard deviation of the return on the equity portfolio is commensurate with its significantly
higher average return. The higher average excess return reflects the long-term risk premium.
Is the return premium on the perfect-timing strategy also a risk premium? It can’t be:
Because the perfect timer never does worse than either bills or the market, the extra return
cannot be compensation for the possibility of poor returns; instead it is attributable to supe-
rior analysis. The value of superior information is reflected in the tremendous ending value of
the portfolio. This value does not reflect compensation for risk.
Consider how you might choose between two hypothetical strategies. Strategy 1 offers a
sure rate of return of 5%; strategy 2 offers an uncertain return that is given by 5% plus a ran-
dom number that is equally likely to be either 0% or 5%. The results for each strategy are:

Strategy 1 (%) Strategy 2 (%)

Expected return 5 7.5

Standard deviation 0 2.5
Highest return 5 10.0
Lowest return 5 5.0

Clearly, strategy 2 dominates strategy 1, as its rate of return is at least equal to that of
strategy 1 and sometimes greater. No matter how risk averse you are, you will always prefer
strategy 2 to strategy 1, even though strategy 2 has a significant standard deviation.
Compared to strategy 1, strategy 2 provides only good surprises, so the standard deviation
in this case cannot be a measure of risk.
You can look at these strategies as analogous to the case of the perfect timer compared with
either an all-equity or all-cash strategy. In every period, the perfect timer obtains at least as
good a return, in some cases better. Therefore, the timer’s standard deviation is a misleading
measure of risk when you compare perfect timing to an all-equity or all-cash strategy.

Valuing Market Timing as an Option

Merton (1981) shows that perfect market timing can be viewed as a call option on the market
index in this sense: Investing 100% in T-bills plus holding a call option on the equity

bod34698_ch18_595-[Link] 618 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 619

portfolio will yield returns FIGURE 18.8

identical to those of the
Rate of return Rate of return of a perfect
portfolio of the perfect
market timer
timer who invests 100% in
either the safe asset or the
equity portfolio, whichever
will yield the higher return.
The perfect timer’s return
is shown in Figure 18.8.
The rate of return is rf
bounded from below by the
risk-free rate, rf .
To see how timing abil-
ity can be treated as an rM
rf
option, suppose the market
index currently is at S0 and
a call option on the index
has exercise price of X 5 S0(1 1 rf ). If the market outperforms bills over the coming period,
ST will exceed X; it will be less than X otherwise. Now look at the payoff to a portfolio consist-
ing of this option plus S0 dollars invested in bills.

Payoff to Portfolio

Outcome: ST " X ST + X

Bills S0(1 1 rf ) S0(1 1 rf )

Option 0 ST 2 X
Total S0(1 1 rf ) ST

The portfolio returns the risk-free rate when the market return is less than the risk-free rate
and pays the market return when the market beats bills. This represents perfect market timing.
Consequently, the value of perfect-timing ability must equal the value of the call option.
Valuation of the call option embedded in market timing is relatively straightforward using
the Black-Scholes formula. Set S0 5 $1 (to find the value of the call per dollar invested in the
market), use an exercise price of X 5 (1 1 rf ) (for example, the current risk-free rate is about
.12%), and a volatility of s 5 18% (about the historical annual standard deviation of the S&P
500). For a once-a-month timer, T 5 1@12 . According to the Black-Scholes formula, the call
option conveyed by market-timing ability is worth 2.1% of assets, and this is the monthly fee
one could presumably charge for such services. Annualized, that fee is about 28%, similar to
the excess return of the market timer in Table 18.8. Less frequent timing would be worth less
(see Concept Check 18.3). If one could time the market only on an annual basis, then T 5 1
and the value of perfect timing would be about 7.2% per year.

The Value of Imperfect Forecasting

But managers are not perfect forecasters. While managers who are right most of the time pre-
sumably do very well, “right most of the time” does not mean merely the percentage of the time
a manager is right. A Tucson, Arizona, weather forecaster who always predicts “no rain” may be
right 90% of the time, but this “stopped clock” strategy does not require any forecasting ability.
Neither is the overall proportion of correct forecasts an appropriate measure of market
forecasting ability. If the market is up two days out of three, and a forecaster always predicts a
market advance, the two-thirds success rate is not a measure of forecasting ability. We need to
examine the proportion of bull markets (rM . rf ) correctly forecast as well as the proportion of
bear markets (rM , rf ) correctly forecast.
If we call P1 the proportion of correct forecasts of bull markets and P2 the proportion for
bear markets, then P1 1 P2 2 1 is the correct measure of timing ability. For example, a

bod34698_ch18_595-[Link] 619 09/08/12 7:35 PM

 Confirming Pages

620 Part SIX Active Investment Management

forecaster who always guesses correctly will have P1 5 P2 5 1 and will show ability of 1
(100%). An analyst who always bets on a bear market will mispredict all bull markets (P1 5 0),
will correctly “predict” all bear markets (P2 5 1), and will end up with timing ability of
P1 1 P2 2 1 5 0. If C denotes the (call option) value of a perfect market timer, then
(P1 1 P2 2 1)C measures the value of imperfect forecasting ability.
The incredible potential payoff to accurate timing versus the relative scarcity of billionaires
suggests that market timing is far from a trivial exercise and that very imperfect timing is the
most that we can hope for.

CONCEPT
c h e c k 18.4 What is the market-timing score of someone who flips a fair coin to predict the market?

Measurement of Market-Timing Performance

In its pure form, market timing involves shifting funds between a market-index portfolio and
cash equivalents, such as T-bills or a money market fund, depending on whether the market as a
whole is expected to outperform cash. In practice, most managers do not shift fully between cash
and the market. How might we measure partial shifts into the market when it is expected to
perform well?
To simplify, suppose the investor holds only the market-index portfolio and T-bills. If the
weight on the market were constant, say, .6, then the portfolio beta would also be constant, and
the portfolio characteristic line would plot as a straight line with a slope .6, as in Figure 18.9A.
If, however, the investor could correctly time the market and shift funds into it in periods
when the market does well, the characteristic line would plot as in Figure 18.9B. The idea is
that if the timer can predict bull and bear markets, more will be shifted into the market when
the market is about to go up. The portfolio beta and the slope of the characteristic line will be
higher when rM is higher, resulting in the curved line that appears in Figure 18.9B.

FIGURE 18.9

Characteristic lines rP – rf
A: No market timing, beta is
Slope = .6
constant
B: Market timing, beta (A)
increases with expected
market excess return
rM – rf

rP – rf Steadily
increasing
slope

(B)

rM – rf

bod34698_ch18_595-[Link] 620 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 621

Treynor and Mazuy (1966) tested to see whether portfolio betas did in fact increase prior
to market advances, but they found little evidence of timing ability. A similar test was imple-
mented by Henriksson (1984). Overall, 62% of the funds in his study had negative point
estimates of timing ability.
In sum, empirical tests to date show little evidence of market-timing ability. Perhaps this
should be expected; given the tremendous values to be reaped by a successful market timer, it
would be surprising to uncover clear-cut evidence of such skills in nearly efficient markets.

• The appropriate performance measure depends on the investment context. The Sharpe SUMMARY
measure is most appropriate when the portfolio represents the entire investment fund. The
Treynor measure is appropriate when the portfolio is to be mixed with several other assets,
allowing for diversification of firm-specific risk outside each portfolio. The information
ratio may be used when evaluating a portfolio to be mixed with the passive index portfolio.
• The shifting mean and variance of actively managed portfolios make it harder to assess
performance. A typical example is the attempt of portfolio managers to time the market,
resulting in ever-changing portfolio betas and standard deviations.
• Common attribution procedures partition performance improvements to asset allocation,
sector selection, and security selection. Performance is assessed by calculating departures of
portfolio composition from a benchmark or neutral portfolio.
• Active management has two components: market timing (or, more generally, asset alloca-
tion) and security analysis.
• The value of perfect market-timing ability is enormous. The rate of return to a perfect
market timer will be uncertain, but the risk cannot be measured by standard deviation,
because perfect timing dominates a passive strategy, providing only “good” surprises.
• Perfect-timing ability is equivalent to having a call option on the market portfolio. The
value of that option can be determined using valuation techniques such as the Black-
Scholes formula.
• The value of imperfect market timing depends on the sum of the probabilities of the true
outcome conditional on the forecast: P1 1 P2 2 1. Because the value of perfect timing equals
that of the implicit call option C, imperfect timing can be valued by: (P1 1 P2 2 1)C.

active management, 597 comparison market timing, 617 KEY TERMS

alpha capture, 604 universe, 598 passive management, 597
alpha transfer or fund of funds, 601 Sharpe ratio, 599
alpha transport, 604 information ratio, 602 survivorship
bogey, 612 Jensen measure, 603 bias, 611
cash, 597 M-square (M 2), 600 Treynor measure, 602

Decomposition of the variance of a portfolio, P: s 2P 5 b 2P sM

2
1 s 2e KEY FORMULAS
Performance measures:
R
Sharpe ratio: S 5
s
[Link]/bkm

M 2 of portfolio P relative to its Sharpe ratio: M 2 5 RP * 2 RM 5 sM (SP 2 SM)

R
Treynor measure: T 5
b
The information ratio of an incremental portfolio and the overall Sharpe ratio:
aP 2
SO 5 SQRT BS 2M 1 ¢ ≤ R
sP

bod34698_ch18_595-[Link] 621 09/08/12 7:35 PM

 Confirming Pages

622 Part SIX Active Investment Management

The relation of alpha to other performance measures:

RP bP RM aP sP
Sharpe ratio: SP 5 5 1 bP 5 r
sP sP sP sM
aP
SP 5 SM r 1
sP
aP
SP 2 SM 5 SM (r 2 1) 1
sP

Treynor measure:

RP bP RM 1 aP aP
TP 5 5 5 RM 1
bP bP bP
bM 5 1 TM 5 RM
aP
TP 2 TM 5
bP

Performance evaluation in a multi-index model:

RPt 5 bP RMt 1 bSMB rSMBt 1 bHML rHMLt 1 aP 1 ePt

and
R Pt 5 bP RMt 1 bSMB rSMBt 1 bHML rHMLt 1 aP

PROBLEM SETS Select problems are available in McGraw-Hill’s

Connect Finance. Please see the Supplements
section of the book’s frontmatter for more information.

Basic
1. The finance committee of an endowment has decided to shift part of its investment in an
index fund to one of two professionally managed portfolios. Upon examination of past
performance, a committee member proposes to choose the portfolio that achieved a
greater alpha value. (LO 18-1)
a. Do you agree? Why or why not?
b. Could a positive alpha be associated with inferior performance? Explain.
2. The board of a large pension fund noticed that the alpha value of the portfolio of one of
its contract managers has recently increased. Should the fund increase the allocation to
this portfolio? (LO 18-1)
3. Could portfolio A show a higher Sharpe ratio than that of B and at the same time a lower
M 2 measure? Explain. (LO 18-2)
4. Two portfolio managers use different procedures to estimate alpha. One uses a single-
[Link]/bkm

index model regression, the other the Fama-French model. Other things equal, would
you prefer the portfolio with the larger alpha based on the index model or the FF
model? (LO 18-2)

Intermediate
5. Based on current dividend yields and expected capital gains, the expected rates of return
on portfolios A and B are 11% and 14%, respectively. The beta of A is .8, while that of B
is 1.5. The T-bill rate is currently 6%, while the expected rate of return of the S&P 500

bod34698_ch18_595-[Link] 622 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 623

index is 12%. The standard deviation of portfolio A is 10% annually, while that of B is
31%, and that of the index is 20%. (LO 18-2)
a. If you currently hold a market-index portfolio, would you choose to add either of these
portfolios to your holdings? Explain.
b. If instead you could invest only in bills and one of these portfolios, which would you
choose?
6. Evaluate the timing and selection abilities of the four managers whose performances are
plotted in the following four scatter diagrams. (LO 18-5)

rP – rf

rP – rf

(A) rM – rf

(B)
rM – rf

rP – rf rP – rf

(D)
(C) rM – rf rM – rf

7. Consider the following information regarding the performance of a money manager in a

recent month. The table presents the actual return of each sector of the manager’s portfo-
lio in column (1), the fraction of the portfolio allocated to each sector in column (2), the
benchmark or neutral sector allocations in column (3), and the returns of sector indexes in
column (4). (LO 18-4)

(1) (2) (3) (4)

Actual Actual Benchmark Index
Return Weight Weight Return

Equity 2.0% 0.70 0.60 2.5% (S&P 500)

Bonds 1.0 0.20 0.30 1.2 (Aggregate Bond Index)
Cash 0.5 0.10 0.10 0.5

a. What was the manager’s return in the month? What was her over- or
[Link]/bkm

underperformance?
b. What was the contribution of security selection to relative performance?
c. What was the contribution of asset allocation to relative performance? Confirm that
the sum of selection and allocation contributions equals her total “excess” return rela-
tive to the bogey.
8. Conventional wisdom says one should measure a manager’s investment performance over
an entire market cycle. What arguments support this contention? What arguments con-
tradict it? (LO 18-1)

bod34698_ch18_595-[Link] 623 09/08/12 7:35 PM

 Confirming Pages

624 Part SIX Active Investment Management

9. Does the use of universes of managers with similar investment styles to evaluate relative
investment performance overcome the statistical problems associated with instability of
beta or total variability? (LO 18-3)
10. During a particular year, the T-bill rate was 6%, the market return was 14%, and a port-
folio manager with beta of .5 realized a return of 10%. Evaluate the manager based on
the portfolio alpha. (LO 18-1)
11. Bill Smith is evaluating the performance of four large-cap equity portfolios: funds A, B, C,
and D. As part of his analysis, Smith computed the Sharpe ratio and the Treynor measure
for all four funds. Based on his finding, the ranks assigned to the four funds are as follows:

Fund Treynor Measure Rank Sharpe Ratio Rank

A 1 4
B 2 3
C 3 2
D 4 1

The difference in rankings for funds A and D is most likely due to: (LO 18-2)
a. A lack of diversification in fund A as compared to fund D.
b. Different benchmarks used to evaluate each fund’s performance.
c. A difference in risk premiums.

Use the following information to answer Problems l2–16: Primo Management Co. is
looking at how best to evaluate the performance of its managers. Primo has been hearing
more and more about benchmark portfolios and is interested in trying this approach. As
such, the company hired Sally Jones, CFA, as a consultant to educate the managers on the
best methods for constructing a benchmark portfolio, how best to choose a benchmark,
whether the style of the fund under management matters, and what they should do with
their global funds in terms of benchmarking.
For the sake of discussion, Jones put together some comparative two-year perfor-
mance numbers that relate to Primo’s current domestic funds under management and a
potential benchmark.

Weight Return

Style Category Primo Benchmark Primo Benchmark

Large-cap growth .60 .50 17% 16%

Mid-cap growth .15 .40 24 26
Small-cap growth .25 .10 20 18

As part of her analysis, Jones also takes a look at one of Primo’s global funds. In this
particular portfolio, Primo is invested 75% in Dutch stocks and 25% in British stocks.
The benchmark invested 50% in each—Dutch and British stocks. On average, the British
stocks outperformed the Dutch stocks. The euro appreciated 6% versus the U.S. dollar
[Link]/bkm

over the holding period, while the pound depreciated 2% versus the dollar. In terms of the
local return, Primo outperformed the benchmark with the Dutch investments but under-
performed the index with respect to the British stocks.
12. What is the within-sector selection effect for each individual sector? (LO 18-4)
13. Calculate the amount by which the Primo portfolio out- (or under-) performed the
market over the period, as well as the contribution to performance of the pure sector
allocation and security selection decisions. (LO 18-4)

bod34698_ch18_595-[Link] 624 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 625

14. If Primo decides to use return-based style analysis, will the R2 of the regression
equation of a passively managed fund be higher or lower than that of an actively
managed fund? (LO 18-3)
15. Which of the following statements about Primo’s global fund is most correct? Primo
appears to have a positive currency allocation effect as well as: (LO 18-4)
a. A negative market allocation effect and a positive security allocation effect.
b. A negative market allocation effect and a negative security allocation effect.
c. A positive market allocation effect and a negative security allocation effect.
16. Kelli Blakely is a portfolio manager for the Miranda Fund (Miranda), a core large-cap
equity fund. The market proxy and benchmark for performance measurement purposes
is the S&P 500. Although the Miranda portfolio generally mirrors the asset class and
sector weightings of the S&P, Blakely is allowed a significant amount of leeway in man-
aging the fund. Her portfolio holds only stocks found in the S&P 500 and cash.
Blakely was able to produce exceptional returns last year (as outlined in the table
below) through her market-timing and security selection skills. At the outset of the
year, she became extremely concerned that the combination of a weak economy and
geopolitical uncertainties would negatively impact the market. Taking a bold step,
she changed her market allocation. For the entire year her asset class exposures aver-
aged 50% in stocks and 50% in cash. The S&P’s allocation between stocks and cash
during the period was a constant 97% and 3%, respectively. The risk-free rate of
return was 2%. (LO 18-1)

One-Year Trailing Returns

Miranda Fund S&P 500
Return 10.2% 222.5%
Standard deviation 37% 44%
Beta 1.10 1.00

a. What are the Sharpe ratios for the Miranda Fund and the S&P 500?
b. What are the M2 measures for Miranda and the S&P 500?
c. What is the Treynor measure for the Miranda Fund and the S&P 500?
d. What is the Jensen measure for the Miranda Fund?
17. Go to [Link]/bkm and link to the material for Chapter 18, where you will
find five years of monthly returns for two mutual funds, Vanguard’s U.S. Growth Please visit us at
Fund and U.S. Value Fund, as well as corresponding returns for the S&P 500 and the [Link]/bkm
Treasury-bill rate. (LO 18-2)
a. Set up a spreadsheet to calculate each fund’s excess rate of return over T-bills in each
month.
b. Calculate the standard deviation of each fund over the five-year period.
c. What was the beta of each fund over the five-year period? (You may wish to review
the spreadsheets from Chapters 5 and 6 on the Index model.)
d. What were the Sharpe, Jensen, and Treynor measures for each fund?

Challenge
[Link]/bkm

18. Historical data suggest the standard deviation of an all-equity strategy is about 5.5% per
month. Suppose the risk-free rate is now 1% per month and market volatility is at its
historical level. What would be a fair monthly fee to a perfect market timer, according to
the Black-Scholes formula? (LO 18-5)
19. A fund manager scrutinizing the record of two market timers comes up with this infor-
mation: (LO 18-5)

bod34698_ch18_595-[Link] 625 09/08/12 7:35 PM

 Confirming Pages

626 Part SIX Active Investment Management

Number of months that rM . rf 135

Correctly predicted by timer A 78
Correctly predicted by timer B 86
Number of months that rM , rf 92
Correctly predicted by timer A 57
Correctly predicted by timer B 50

a. What are the conditional probabilities, P1 and P2, and the total ability parameters for
timers A and B?
b. Using the data given in this problem, and the historical data in the previous problem,
what is a fair monthly fee for the two timers?

CFA Problems
1. A plan sponsor with a portfolio manager who invests in small-capitalization, high-growth
stocks should have the plan sponsor’s performance measured against which one of the
following? (LO 18-3)
a. S&P 500 Index.
b. Wilshire 5000 Index.
c. Dow Jones Industrial Average.
d. Russell 2000 Index.
2. The chairman provides you with the following data, covering one year, concerning the
portfolios of two of the fund’s equity managers (manager A and manager B). Although
the portfolios consist primarily of common stocks, cash reserves are included in the calcu-
lation of both portfolio betas and performance. By way of perspective, selected data for
the financial markets are included in the following table. (LO 18-1)

Total Return Beta

Manager A 24.0% 1.0

Manager B 30.0 1.5
S&P 500 21.0
Lehman Bond Index 31.0
91-day Treasury bills 12.0

a. Calculate and compare the alpha of the two managers relative to each other and to the
S&P 500.
b. Explain two reasons the conclusions drawn from this calculation may be misleading.
3. Carl Karl, a portfolio manager for the Alpine Trust Company, has been responsible since
2015 for the City of Alpine’s Employee Retirement Plan, a municipal pension fund.
Alpine is a growing community, and city services and employee payrolls have expanded in
each of the past 10 years. Contributions to the plan in fiscal 2020 exceeded benefit pay-
ments by a three-to-one ratio.
The plan’s board of trustees directed Karl five years ago to invest for total return over the
[Link]/bkm

long term. However, as trustees of this highly visible public fund, they cautioned him that
volatile or erratic results could cause them embarrassment. They also noted a state statute that
mandated that not more than 25% of the plan’s assets (at cost) be invested in common stocks.
At the annual meeting of the trustees in November 2020, Karl presented the following
portfolio and performance report to the board.

bod34698_ch18_595-[Link] 626 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 627

ALPINE EMPLOYEE RETIREMENT PLAN

At Cost At Market
Asset Mix as of 9/30/20 (millions) (millions)

Fixed-income assets:
Short-term securities $ 4.5 11.0% $ 4.5 11.4%
Long-term bonds and mortgages 26.5 64.7 23.5 59.5
Common stocks 10.0 24.3 11.5 29.1
$41.0 100.0% $39.5 100.0%

INVESTMENT PERFORMANCE
Annual Rates of
Return for Periods
Ending 9/30/20

5 Years 1 Year

Total Alpine Fund:

Time-weighted 8.2% 5.2%
Dollar-weighted (Internal) 7.7% 4.8%
Assumed actuarial return 6.0% 6.0%
U.S. Treasury bills 7.5% 11.3%
Large sample of pension funds
(average 60% equities, 40% fixed income) 10.1% 14.3%
Common stocks—Alpine Fund 13.3% 14.3%
Average portfolio beta coefficient 0.90 0.89
Standard & Poor’s 500 Stock Index 13.8% 21.1%
Fixed-income securities—Alpine Fund 6.7% 1.0%
Salomon Brothers’ Bond Index 4.0% 211.4%

Karl was proud of his performance and was chagrined when a trustee made the follow-
ing critical observations:
a. “Our one-year results were terrible, and it’s what you’ve done for us lately that
counts most.”
b. “Our total fund performance was clearly inferior compared to the large sample of other
pension funds for the last five years. What else could this reflect except poor manage-
ment judgment?”
c. “Our common stock performance was especially poor for the five-year period.”
d. “Why bother to compare your returns to the return from Treasury bills and the actuar-
ial assumption rate? What your competition could have earned for us or how we would
have fared if invested in a passive index (which doesn’t charge a fee) are the only rele-
vant measures of performance.”
[Link]/bkm

e. “Who cares about time-weighted return? If it can’t pay pensions, what good is it!”
Appraise the merits of each of these statements and give counterarguments that Karl
can use. (LO 18-2)
4. A portfolio manager summarizes the input from the macro and micro forecasts in the fol-
lowing table: (LO 18-2)

bod34698_ch18_595-[Link] 627 09/08/12 7:35 PM

 Confirming Pages

628 Part SIX Active Investment Management

MICRO FORECASTS
Residual Standard
Asset Expected Return (%) Beta Deviation (%)
Stock A 20 1.3 58
Stock B 18 1.8 71
Stock C 17 0.7 60
Stock D 12 1.0 55

MACRO FORECASTS
Asset Expected Return (%) Standard Deviation (%)

T-bills 8 0
Passive equity portfolio 16 23

a. Calculate expected excess returns, alpha values, and residual variances for these stocks.
b. Construct the optimal risky portfolio.
c. What is Sharpe’s measure for the optimal portfolio and how much of it is contributed
by the active portfolio? What is the M 2?

WEB master
Morningstar has an extensive ranking system for mutual funds, including a screening pro-
gram that allows you to select funds based on a number of factors. Open the Morningstar
website at [Link] and click on the Funds link. Select the Mutual Fund
Quickrank link from the right-side menu. Use the Quickrank screener to find a list of the
funds with the highest five-year returns. Repeat the process to find the funds with the high-
est 10-year returns. How many funds appear on both lists?
Select three of the funds that appear on both lists. For each fund, click on the ticker sym-
bol to get its Morningstar report and look in the Risk Measures section.
1. What is the fund’s standard deviation?
2. What is the fund’s Sharpe ratio?
3. What is the standard index? What is the best-fit index?
4. What are the beta and alpha coefficients using both the standard index and the best-fit
index? How do these compare to the fund’s parameters?
Look at the Management section of the report. Was the same manager in place for the entire
10-year period?

SOLUTIONS TO 18.1 Sharpe: (r 2 rf)/s

CONCEPT SP 5 (35 2 6)/42 5 .69
c h e c k s SM 5 (28 2 6)/30 5 .733
[Link]/bkm

Jensen (or alpha): r 2 [rf 1 b(rM 2 rf)]

aP 5 35 2 [6 1 1.2(28 2 6)] 5 2.6%
aM 5 0
Treynor: (r 2 rf )/b
TP 5 (35 2 6)/1.2 5 24.2
TM 5 (28 2 6)/1.0 5 22

bod34698_ch18_595-[Link] 628 09/08/12 7:35 PM

 Confirming Pages

Chapter 18 Portfolio Performance Evaluation 629

18.2 Performance attribution

First compute the new bogey performance as
(.70 3 5.81) 1 (.25 3 1.45) 1 (.05 3 .48) 5 4.45%
a. Contribution of asset allocation to performance:

(1) (2) (3) (4) (5) 5 (3) 3 (4)

Actual Benchmark Index Contribution to
Weight in Weight in Excess Return Performance
Market Market Market Weight (%) (%)

Equity .70 .70 .00 5.81 .000

Fixed-income .07 .25 2.18 1.45 2.261
Cash .23 .05 .18 0.48 .086
Contribution of asset allocation 2.175

b. Contribution of selection to total performance:

(1) (2) (3) (4) (5) 5 (3) 3 (4)

Portfolio Index Excess
Performance Performance Performance Portfolio Contribution
Market (%) (%) (%) Weight (%)

Equity 7.28 5.00 2.28 0.70 1.60

Fixed-income 1.89 1.45 0.44 0.07 0.03
Contribution of selection within markets 1.63

18.3 Import the series of annual returns on T-bills and large stocks (S&P 500).
a. Compute the return to the perfect timer. You can use the Excel function 5 max (stock
return, bill return) to select the greater of the two returns each year.
b. Use Excel functions to estimate average and SD.
c. Generate the wealth-index series. Set the wealth index at the end of 1925 to 1.
Because the rates of return are expressed in percentages, the index value at the
end of 1926 5 1 1 rate(1926)/100. For the following years, index 5 previous
index 3 (1 1 this year’s return/100).
d. The wealth index for 2008 is the terminal value of the fund per $1 invested at the
beginning of 1926.
e. The geometric average equals: Terminal value^(1/82) 2 1. Notice that this calcula-
tion results in a return expressed as a decimal, not percent.
f. The performance of the annual timer is not as good as the monthly timer. The
annual timer may switch funds between the market and T-bills only once per year.
He cannot advantageously move funds between the market and bills across months
within each year. Someone who can time perfectly will always be better off when
allowed to make more frequent allocation choices.
18.4 The timer will guess bear or bull markets randomly. One-half of all bull markets will
[Link]/bkm

be preceded by a correct forecast, and, similarly, one-half of all bear markets will be
preceded by a correct forecast. Hence, P1 1 P2 2 1 5 ½ 1 ½ 2 1 5 0.

bod34698_ch18_595-[Link] 629 09/08/12 7:35 PM