Chapter 34
PORTFOLIO PERFORMANCE
EV
ALUATION
LALITH P. SAMARAKOON, University of St. Thomas, USA
TANWEER HASAN, Roosevelt University, USA
Abstract indicate the extent to which the portfolio has out-
performed or under-performed, or whether it has
The portfolio performance evaluation involves the performed at par with the benchmark.
determination of how a managed portfolio has per- The evaluation of portfolio performance is im-
formed relative to some comparison benchmark. portant for several reasons. First, the investor,
Performance evaluation methods generally fall into whose funds have been invested in the portfolio,
two categories, namely conventional and risk- needs to know the relative performance of the
adjusted methods. The most widely used conven- portfolio. The performance review must generate
tional methods include benchmark comparison and and provide information that will help the investor
style comparison. The risk-adjusted methods adjust to assess any need for rebalancing of his invest-
returns in order to take account of differences in risk ments. Second, the management of the portfolio
levels between the managed portfolio and the bench- needs this information to evaluate the perform-
mark portfolio. The major methods are the Sharpe ance of the manager of the portfolio and to deter-
ratio, Treynor ratio, Jensen’s alpha, Modigliani mine the manager’s compensation, if that is tied
and Modigliani, and Treynor Squared. The risk- to the portfolio performance. The performance
adjusted methods are preferred to the conventional evaluation methods generally fall into two cate-
methods. gories, namely conventional and risk-adjusted
methods.
Keywords: performance; evaluation; standard devi-
ation; systematic risk; conventional methods;
34.2. Conventional Methods
benchmark comparison; style comparison; risk-
adjusted measures; Sharpe measure; Treynor 34.2.1. Benchmark Comparison
measure; Jensen measure; alpha; Modigliani-Mod-
igliani measure; Treynor squared The most straightforward conventional method
involves comparison of the performance of an in-
34.1. Introduction vestment portfolio against a broader market index.
The most widely used market index in the United
The portfolio performance evaluation primarily States is the S&
P 500 index, which measures the
refers to the determination of how a particular price movements of 500 U.S. stocks compiled by
investment portfolio has performed relative to the Standard &Poor’s Corporation. If the return
some comparison benchmark. The evaluation can on the portfolio exceeds that of the benchmark
618 ENCYCLOPEDIA OF FINANCE
index, measured during identical time periods, Reilly and Norton (2003) provide an excellent
then the portfolio is said to have beaten the bench- disposition of the use of benchmark portfolios and
mark index. While this type of comparison with a portfolios style and the issues associated with their
passive index is very common in the investment selection. Sharpe (1992), and Christopherson (1995)
world, it creates a particular problem. The level have developed methods for determining this
of risk of the investment portfolio may not be the style.
same as that of the benchmark index portfolio.
Higher risk should lead to commensurately higher 34.3. Risk-adjusted Methods
returns in the long term. This means if the invest-
ment portfolio has performed better than the The risk-adjusted methods make adjustments to
benchmark portfolio, it may be due to the invest- returns in order to take account of the differences
ment portfolio being more risky than the bench- in risk levels between the managed portfolio and
mark portfolio. Therefore, a simple comparison of the benchmark portfolio. While there are many
the return on an investment portfolio with that such methods, the most notables are the Sharpe
of a benchmark portfolio may not produce valid ratio (S), Treynor ratio (T), Jensen’s alpha (a),
results. Modigliani and Modigliani (M 2 ), and Treynor
Squared (T 2 ). These measures, along with their
34.2.2. Style Comparison applications, are discussed below.
A second conventional method of performance
34.3.1. Sharpe Ratio
evaluation called ‘‘style-comparison’’ involves com-
parison of return of a portfolio with that having a The Sharpe ratio (Sharpe, 1966) computes the risk
similar investment style. While there are many in- premium of the investment portfolio per unit of
vestment styles, one commonly used approach total risk of the portfolio. The risk premium, also
classifies investment styles as value versus growth. known as excess return, is the return of the port-
The ‘‘value style’’ portfolios invest in companies folio less the risk-free rate of interest as measured
that are considered undervalued on the basis of by the yield of a Treasury security. The total risk is
yardsticks such as price-to-earnings and price- the standard deviation of returns of the portfolio.
to-book value multiples. The ‘‘growth style’’ port- The numerator captures the reward for investing in
folios invest in companies whose revenue and a risky portfolio of assets in excess of the risk-free
earnings are expected to grow faster than those of rate of interest while the denominator is the vari-
the average company. ability of returns of the portfolio. In this sense,
In order to evaluate the performance of a value- the Sharpe measure is also called the ‘‘reward-
oriented portfolio, one would compare the return to-variability’’ ratio. Equation (34.1) gives the
on such a portfolio with that of a benchmark Sharpe ratio:
portfolio that has value-style. Similarly, a growth-
rp rf
style portfolio is compared with a growth-style S¼ (34:1)
sp
benchmark index. This method also suffers from
the fact that while the style of the two portfolios where S is the Sharpe ratio, rp the return of the
that are compared may look similar, the risks of portfolio, rf the risk-free rate, and sp the standard
the two portfolios may be different. Also, the deviation of returns of the portfolio.
benchmarks chosen may not be truly comparable The Sharpe ratio for an investment portfolio can
in terms of the style since there can be many im- be compared with the same for a benchmark port-
portant ways in which two similar style-oriented folio such as the overall market portfolio. Suppose
funds vary. that a managed portfolio earned a return of
PORTFOLIO PERFORMANCE EV
ALUATION 619
20 percent over a certain time period with a stand- risk. In fact, in the above example, the portfolio
ard deviation of 32 percent. Also assume that dur- earned an excess return of 16 percent whereas the
ing the same period the Treasury bill rate was 4 market earned only 9 percent. These two numbers
percent, and the overall stock market earned a alone do not tell anything about the relative
return of 13 percent with a standard deviation of performance of the portfolio since the portfolio
20 percent. The managed portfolio’s risk premium and the market have different levels of market
is (20 percent 4 percent) ¼ 16 percent, while its risk. In this instance, the Treynor ratio for
Sharpe ratio, S, is equal to 16 percent=32 percent ¼ the managed portfolio equals (20 percent 4
0.50. The market portfolio’s excess return is (13 percent)=1.5 ¼ 10.67, while that for the market
percent 4 percent) ¼ 9 percent, while its S equals equals (13 percent 4 percent)=1.00 ¼ 9.00.
9 percent=20 percent ¼ 0.45. Accordingly, for each Thus, after adjusting for systematic risk, the man-
unit of standard deviation, the managed portfolio aged portfolio earned an excess return of 10.67
earned a risk premium of 0.50 percent, which is percent for each unit of beta while the market
greater than that of the market portfolio of 0.45 portfolio earned an excess return of 9.00 percent
percent, suggesting that the managed portfolio for each unit of beta. Thus, the managed portfolio
outperformed the market after adjusting for total outperformed the market portfolio after adjusting
risk. for systematic risk.
34.3.2. Treynor Ratio 34.3.3. Jensen’s Alpha
The Treynor ratio (Treynor, 1965) computes the Jensen’s alpha (Jensen, 1968) is based on the Cap-
risk premium per unit of systematic risk. The risk ital Asset Pricing Model (CAPM) of Sharpe
premium is defined as in the Sharpe measure. The (1964), Lintner (1965), and Mossin (1966). The
difference in this method is in that it uses the alpha represents the amount by which the average
systematic risk of the portfolio as the risk para- return of the portfolio deviates from the expected
meter. The systematic risk is that part of the total return given by the CAPM. The CAPM specifies
risk of an asset which cannot be eliminated the expected return in terms of the risk-free rate,
through diversification. It is measured by the par- systematic risk, and the market risk premium. The
ameter known as ‘beta’ that represents the slope of alpha can be greater than, less than, or equal to
the regression of the returns of the managed port- zero. An alpha greater than zero suggests that the
folio on the returns to the market portfolio. The portfolio earned a rate of return in excess of the
Treynor ratio is given by the following equation: expected return of the portfolio. Jensen’s alpha is
given by.
rp rf
T¼ (34:2)
bp a ¼ rp [rf þ bp (rm rf )] (34 :3)
where T is the Treynor ratio, rp the return of the where a is the Jensen’s alpha, rp the return of the
portfolio, rf the risk-free rate, and bp the beta of portfolio, rm the return of the market portfolio, rf
the portfolio. the risk-free rate, and bp the beta of the port-
Suppose that the beta of the managed portfolio folio.
in the previous example is 1.5. By definition, the Using the same set of numbers from the previ-
beta of the market portfolio is equal to 1.0. This ous example, the alpha of the managed portfolio
means the managed portfolio has one-and-half and the market portfolio can be computed as fol-
times more systematic risk than the market port- lows. The expected return of the managed port-
folio. We would expect the managed portfolio to folio is 4 percent þ 1.5 (13 percent 4 percent) ¼
earn more than the market because of its higher 17.5 percent. Therefore, the alpha of the managed
620 ENCYCLOPEDIA OF FINANCE
portfolio is equal to the actual return less the formed the market. The difficulty, however, is
expected return, which is 20 percent 17.5 percent that the differential performance of 0.05 is not an
¼ 2.5 percent. Since we are measuring the expected excess return. Modigliani and Modigliani (1997)
return as a function of the beta and the market measure, which is referred to as M 2 , provides a
risk premium, the alpha for the market is always risk-adjusted measure of performance that has an
zero. Thus, the managed portfolio has earned a economically meaningful interpretation. The M 2 is
2.5 percent return above that must be earned given by
given its market risk. In short, the portfolio has
a positive alpha, suggesting superior performance. M 2 ¼ rp rm (34:4)
When the portfolio is well diversified all three
where M 2 is the Modigliani-Modigliani measure,
methods – Sharpe, Treynor, and Jensen – will give
rp the return on the adjusted portfolio, rm the
the same ranking of performance. In the example,
return on the market portfolio.
the managed portfolio outperformed the market
The adjusted portfolio is the managed portfolio
on the basis of all three ratios. When the portfolio
adjusted in such a way that it has the same total
is not well diversified or when it represents the
risk as the market portfolio. The adjusted portfolio
total wealth of the investor, the appropriate meas-
is constructed as a combination of the managed
ure of risk is the standard deviation of returns of
portfolio and risk-free asset, where weights are
the portfolio, and hence the Sharpe ratio is the
specified as in Equations (34.5) and (34.6).
most suitable. When the portfolio is well diversi-
fied, however, a part of the total risk has been sm
wrp ¼ (34:5)
diversified away and the systematic risk is the sp
most appropriate risk metric. Both Treynor ratio wrf ¼ 1 wrp (34:6)
and Jensen’s alpha can be used to assess the per-
formance of well-diversified portfolios of secur- where wrp represents the weight given to the man-
ities. These two ratios are also appropriate when aged portfolio, which is equal to the standard de-
the portfolio represents a sub-portfolio or only a viation of the market portfolio (sm ) divided by the
part of the client’s portfolio. Chen and Lee standard deviation of the managed portfolio (sp ).
(1981, 1986) examined the statistical distribution wrf is the weight on the risk-free asset and is equal
of Sharpe, Treynor, and Jensen measures and to one minus the weight on the managed portfolio.
show that the empirical relationship between The risk of the adjusted portfolio (sp ) is the
these measures and their risk proxies is dependent weight on the managed portfolio times the stand-
on the sample size, the investment horizon and ard deviation of the managed portfolio as given in
market conditions. Cumby and Glen (1990), Equation (34.7). By construction, this will be equal
Grinblatt and Titman (1994), Kallaberg et al. to the risk of the market portfolio.
(2000), and Sharpe (1998) have provided evidence
sm
of the application of performance evaluation sp ¼ wrp sp ¼ sp ¼ sm (34:7)
sp
techniques.
The return of the adjusted portfolio (rp ) is com-
34.3.4. Modigliani and Modigliani Measure puted as the weighted average of the returns of the
managed portfolio and the risk-free rate, where the
The Sharpe ratio is not easy to interpret. In the weights are as in Equations (34.5) and (34.6)
example, the Sharpe ratio for the managed port- above:
folio is 0.50, while that for the market is 0.45. We
concluded that the managed portfolio outper- rp ¼ wrf rf þ wrp rp (34:8)
PORTFOLIO PERFORMANCE EV
ALUATION 621
The return on the adjusted portfolio can be bm
wrp ¼ (34:10)
readily compared with the return on the market bp
portfolio since both have the same degree of risk.
wrf ¼ 1 wrp (34:11)
The differential return, M 2 , indicates the excess
return of the managed portfolio in comparison to where wrp represents the weight given to the
the benchmark portfolio after adjusting for differ- managed portfolio, which is equal to the beta of
ences in the total risk. Thus, M 2 is more meaning- the market portfolio ( bm ) divided by the beta
ful than the Sharpe ratio. of the managed portfolio ( bp ). wrf is the weight on
In the example, the standard deviation of the risk-free asset and is equal to one minus the
the managed portfolio is 32 percent and the weight on the managed portfolio. The beta of the
standard deviation of the market portfolio is 20 adjusted portfolio ( bp ) is the weight on the man-
percent. Hence, the wrp ¼ 20=32 ¼ 0:625, and aged portfolio times the beta of the managed port-
wrf ¼ 1 0:625 ¼ 0:375. The adjusted portfolio folio, and this will be equal to the risk of the market
would be 62.5 percent invested in the managed portfolio as shown in the following equation:
portfolio and 37.5 percent invested in Treasury
bills. Now the risk of the adjusted portfolio, bm
bp ¼ wrp bp ¼ bp ¼ bm (34:12)
sp ¼ 0:625 32 percent ¼ 20 percent, is the bp
same as the risk of the market portfolio. The return
on the adjusted portfolio would be rp ¼ 0:375 The return of the adjusted portfolio (rp ) is com-
4 percent þ 0:625 20 percent ¼ 14 percent. puted as the weighted average of the returns of the
The M 2 ¼ 14 percent 13 percent ¼ 1 percent. managed portfolio and the risk-free rate, where the
Thus, on a risk-adjusted basis, the managed port- weights are as determined above in equations
folio has performed better than the benchmark by (34.10) and (34.11):
1 percent. rp ¼ wrf rf þ wrp rp (34:13)
34.3.5. Treynor Squared The return on the adjusted portfolio can be readily
compared with the return on the market portfolio
Another performance measure, called T 2 analo- since both have the same level of market risk. The
gous to M 2 , can be constructed. This is a deviant differential return, T 2 , indicates the excess return
of the Treynor measure, and the rationale is the of the managed portfolio in comparison to the
same as that of M 2 . T 2 is defined as benchmark portfolio after adjusting for differences
in the market risk.
T 2 ¼ rp rm (34:9)
In the example, the beta of the managed port-
where T 2 is the Treynor-squared measure, rp the folio is 1.5. Hence, wrp ¼ 1:0=1:5 ¼ 0:67 and
return on the adjusted portfolio, and rm the return wrf ¼ 1 0:67 ¼ 0:33. The adjusted portfolio
on the market portfolio. would be 67 percent invested in the managed port-
The adjusted portfolio is the managed portfolio folio and 33 percent invested in Treasury bills.
adjusted such that it has the same degree of sys- The beta of the adjusted portfolio, sp ¼ 0:67
tematic or market risk as the market portfolio. 1:5 ¼ 1:00, which is equal to the beta of the
Since the market risk or beta of the market port- market portfolio. The return on the adjusted
folio is equal to one, the adjusted portfolio is con- portfolio would be rp ¼ 0:33 4 percent þ 0:67
structed as a combination of the managed 20 percent ¼ 14:72 percent. T 2 ¼ 14:72 percent
portfolio and risk-free asset such that the adjusted 13:00 percent ¼ 1:72 percent. Thus, after adjust-
portfolio has a beta equal to one. The weights are ing for market risk, the managed portfolio
specified as in equations below. has performed better than the benchmark by 1.72
622 ENCYCLOPEDIA OF FINANCE
percent. T 2 is a better measure of relative perform- Kallaberg, J.G., Lin, C.L., and Trzcinka, C. (2000).
ance when the market risk of a managed portfolio ‘‘The value added from investment managers: an
examination of funds of REITs.’’ Journal of Finan-
is the relevant risk metric.
cial Quantitative Analysis, 35: 387–408.
Lintner, J. (1965). ‘‘The valuation of risk assets and the
selection of risky investments in stock portfolios and
REFERENCES
capital budgets.’’ Review of Economics and Statistics,
47: 13–47.
Chen, S.N. and Lee, C.F. (1981). ‘‘The sampling relation-
Modigliani, F. and Modigliani, L. (1997). ‘‘Risk-
ship between Sharpe’s performance measure and its
adjusted performance: how to measure it and why.’’
risk proxy: sample size, investment horizon and mar-
Journal of Portfolio Management, 23: 45–54.
ket conditions.’’ Management Science, 27: 607–618.
Mossin, J. (1966). ‘‘Equilibrium in the capital market.’’
Chen, S.N. and Lee, C.F. (1986). ‘‘The effects of the
Econometrica, 34: 768–783.
sample size, the investment horizon and market con-
Reilly, F.K. and Norton, E.A. (2003). Investments, 6th
ditions on the validity of composite performance
edn. Mason, Ohio: Thompson-Southwestern.
measures: A generalization.’’ Management Science,
Sharpe, W.F. (1964). ‘‘Capital asset prices: a theory of
32: 1410–1421.
market equilibrium under conditions of risk.’’ Jour-
Christopherson, J.A. (1995). ‘‘Equity style classifica-
nal of Finance, 19: 425–442.
tions.’’ Journal of Portfolio Management, 21: 32–43.
Sharpe, W.F. (1966). ‘‘Mutual fund performance.’’
Cumby, R.E. and Glen, J.D. (1990). ‘‘Evaluating the
Journal of Business, 39(1): 119–138.
performance of international mutual funds.’’ Journal
Sharpe, W.F. (1992). ‘‘Asset allocation: management
of Finance, 45: 497–521.
style and performance measurement.’’ Journal of
Grinblatt, M. and Titman, S. (1994). ‘‘A study of
Portfolio Management, 18: 7–19.
monthly mutual fund returns and performance
Sharpe, W.F. (1998). ‘‘Morningstar’s risk-adjusted rat-
evaluation techniques.’’ Journal of Financial and
ings.’’ Financial Analysts Journal, 54: 21–33.
Quantitative Analysis, 29: 419–444.
Treynor, J.L. (1965). ‘‘How to rate management of
Jensen, M.C. (1968). ‘‘The performance of mutual
investment funds.’’ Harvard Business Review, 43:
funds in the period 1945–1964.’’ Journal of Finance,
63–75.
23: 389–416.
