Evaluation of Portfolio Performance
The contents of this presentation are based on Reilly and Brown (2009)
Dr. Vinodh Madhavan
Associate Professor
Amrut Mody School of Management
Ahmedabad University
Contact Details:
[email protected] Portfolio Manager
What is the required from a portfolio manager ?
There are two desirable attributes of a portfolio manager’s
performance.
The ability to derive above-average returns for a given risk class.
The ability to diversify the portfolio completely to eliminate all
unsystematic risk, relative to the portfolio benchmark.
Superior risk-adjusted returns mat be derived through either superior
timing or superior selection
Superior Timing
An equity portfolio manager who does a good job predicting the ups and
downs of the equity market can adjust his/her portfolio such that the
portfolio constitutes
High beta stocks during rising markets
Low beta stocks during market downturns
Similarly, a fixed income portfolio manager with superior timing ability
would alter the duration of the portfolio based on anticipated interest rate
changes
The portfolio manger would increase the duration of the portfolio if
he/she anticipates fall in the interest rates.
The portfolio manger would reduce the duration of the portfolio if
he/she anticipates rise in the interest rates.
Superior Security Selection & Diversification
An alternative strategy to earn above-average risk-adjusted returns, a
portfolio manager may consistently choose undervalued stocks or
bonds.
In this case, the portfolio manager does not rely on market timing
ability to derive above-average risk-adjusted returns.
When it comes to diversification, a portfolio manager would aim to
diversify his/her portfolio as the market does not reward
unsystematic/unique risks.
So, the risk of a fully diversified portfolio held by an investor would be
closer to the risk of an overall market portfolio
Risk-Adjusted Returns vs. Diversification
It may be noted that superior risk-adjusted returns and diversification may
not go hand-in-hand.
In other words, a portfolio may yield above-average returns, partly due to a
conscious attempt by a portfolio manager to limit diversification.
The portfolio manager may do so, in his/her quest to achieve above-
average returns, even if it comes at the risk of suboptimal diversification.
Not all measures of portfolio performance consider both these facets.
Some portfolio performance techniques consider both the facets but do
not differentiate between them
Treynor Measure
Treynor (1965) developed the first composite measure of portfolio
performance that included
Risk produced by general market fluctuations
Risk resulting from unique fluctuations in the portfolio securities.
𝑅𝑖 − 𝑅𝐹𝑅
𝑇𝑖 =
𝛽𝑖
where
𝑅𝑖 is the average rate of return of portfolio i during a specified time
period
𝑅𝐹𝑅 is the average risk-free rate of return during the same time period
𝛽𝑖 is the slope of the fund’s characteristic line during that period.
Sharpe Measure
Sharpe (1966) proposed a composite portfolio measure in the lines of
CAPM, that deals specifically with Capital Market Line (CML; 𝑟𝑖 vs.
𝜎𝑖 )
𝑅𝑖 − 𝑅𝐹𝑅
𝑆𝑖 =
𝜎𝑖
This performance measure is similar to Treynor measure.
However, it considers the total risk of the portfolio (𝜎𝑖 ) as opposed to
the Treynor measure that considers only the systematic risk (𝛽𝑖 ) of the
portfolio.
It measures the risk premium earned per unit of total risk.
Jensen’s portfolio performance measure
The Jensen measure (Jensen, 1968) is a one-period measure based on
CAPM
𝐸 𝑅𝑗 = 𝑅𝐹𝑅 + 𝛽𝑗 𝐸 𝑅𝑀 − 𝑅𝐹𝑅
Where
𝐸(𝑅𝑗 ) is the expected return of security/portfolio j
𝐸 𝑅𝑀 is the expected return of the market
𝛽𝑗 is the systematic risk of security/portfolio j
Unlike Treynor and Sharpe measures, the Jensen measure deals with time
series.
Jensen’s portfolio performance measure
Jensen’s measure can be empirically validated using the following
framework.
𝑅𝑗𝑡 = 𝑅𝐹𝑅𝑡 + 𝛽𝑗 𝑅𝑚𝑡 − 𝑅𝐹𝑅𝑡 + 𝑒𝑗𝑡
The above framework postulates that the realized return of a security
during a given time period is a linear function of the risk-free rate of
return during the period, plus a risk-premium that depends on the
systematic risk of the security or portfolio plus a random error term
𝑅𝑗𝑡 − 𝑅𝐹𝑅𝑡 = 𝛽𝑗 𝑅𝑚𝑡 − 𝑅𝐹𝑅𝑡 + 𝑒𝑗𝑡
If the market is at equilibrium, then the asset price in the market is the
same as the intrinsic value of the asset.
Jensen’s portfolio performance measure
If that is the case, there is no marginal utility (additional value) for
security selection. In the real world, assets are bound to be undervalued
or overvalued for variety of reasons.
Superior portfolio managers have the capability to consistently select
undervalued assets that earn higher risk premium than what is implied
by CAPM model.
Consequently, in the real world, wherein the market is at a dis-
equilibrium, the CAPM empirical framework is bound to contain an
intercept – one that captures the activeness of portfolio management.
𝑅𝑗𝑡 − 𝑅𝐹𝑅𝑡 = 𝛼𝑗 + 𝛽𝑗 𝑅𝑚𝑡 − 𝑅𝐹𝑅𝑡 + 𝑒𝑗𝑡
Jensen’s portfolio performance measure
A superior manager is bound to have a significant 𝛼, while an inferior
manager’s returns that consistently fall short of expectations based on
CAPM, is bound to yield negative 𝛼.
To sum up, 𝛼 captures the degree of portfolio returns that is attributable
to manager’s ability to derive above-average risk-adjusted returns
Unlike Treynor and Sharpe measures that deal with average returns
(𝑅𝑖 ; 𝑅𝑚 ; 𝑅𝐹𝑅), Jensen measure deals with time-dependent returns (𝑅𝑗𝑡 ;
𝑅𝑚𝑡 ; 𝑅𝐹𝑅𝑡 )
For instance, a yearly Jensen’s framework would entail employment of
different annual RFRs at different points of time (t). This is in contrast
to Treynor and Sharpe measures that use only average rates of return.
Jensen’s portfolio performance measure
Lastly, Jensen’s alpha measure framework may be extended in the
lines of multifactor models.
𝑅𝑗𝑡 − 𝑅𝐹𝑅𝑡 = 𝛼𝑗 + 𝑏𝑗1 𝐹1𝑡 + 𝑏𝑗2 𝐹2𝑡 + ⋯ + 𝑏𝑗𝑘 𝐹𝑘𝑡 + 𝑒𝑗𝑡
where
𝐹𝑘𝑡 represents the period t return of Kth common risk factor.
Information Ratio (IR)
This statistic measures a portfolio’s average return in excess of a benchmark
portfolio; divided by the standard deviation of the excess return.
𝑅𝑗 − 𝑅𝑏
𝐼𝑅𝑗 =
𝜎𝐸𝑅
A Sharpe ratio is a special case of IR, wherein the risk-free rate (RFR) is the
benchmark portfolio.
Should excess return be estimated with historical data using single-factor
Jensen’s alpha framework, then IR measure may translate as follows
𝛼𝑗
𝐼𝑅𝑗 =
𝜎𝑒
Where 𝜎𝑒 is the standard of the single-factor regression.
Information Ratio (IR)
Lastly Information Ratio based on periodic returns measured T
times per year could be annualized as follows.
𝑇𝛼𝑗
𝛼𝑗
𝐴𝑛𝑛𝑢𝑎𝑙𝑖𝑧𝑒𝑑 𝐼𝑅 = = 𝑇 = 𝑇 × 𝐼𝑅
𝑇𝜎𝑒 𝜎𝑒
Sortino Measure
Sortino measure is a risk-adjusted investment performance statistic that
differs from Sharpe ratio in two ways.
It measures the portfolio’s average return in excess of a user-specified
minimum acceptable return threshold (This could be, but does not
always have to be, risk free rate of return)
It measures just the downside risk (DR) in the portfolio as opposed to
Sharpe ratio that takes into account total risk.
𝑅𝑖 − 𝜏
𝑆𝑇𝑖 =
𝐷𝑅𝑖
Where 𝜏 is the minimum acceptable return threshold specified for the
time period.
𝐷𝑅𝑖 is the downside risk coefficient for portfolio i during the specified
time period
Sortino Measure
Downside Risk (DR) may quantified using Semi-Derivation ( a
variant of the traditional standard deviation)
1 2
𝑆𝑒𝑚𝑖 − 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 = 𝑅𝑖𝑡 − 𝑅𝑖
𝑛
𝑅<𝑅
Where n is the number of instances the portfolio returns falls below
the expected return.
Attribution Analysis
Attribution Analysis attempts to identify the source of investors’ overall
portfolio performance.
It compares the manager’s total return to the return of a pre-determined
benchmark and decomposes the same into an allocation effect and a
selection effect.
Allocation effect = 𝑖 (𝑤𝑎𝑖 −𝑤𝑏𝑖 × (𝑅𝑏𝑖 − 𝑅𝑏 )]
where
𝑤𝑎𝑖 is the proportion of investment in ith asset class (markets segment) by
the portfolio manager
𝑤𝑏𝑖 is the proportion of investment in ith asset class (markets segment) in
the benchmark policy portfolio
Attribution Analysis
Allocation effect = i (wai −wbi × (R bi − R b )]
R bi is the investment return of ith market segment in the benchmark
policy portfolio
R b is the total return of policy portfolio
Allocation effect measures manager’s decision to over-weight or under-
weight a particular market segment (𝑤𝑎𝑖 − 𝑤𝑏𝑖 ) in terms of that
segment’s return performance relative to the overall return of the
benchmark (𝑅𝑏𝑖 − 𝑅𝑏 )
Selection Effect = i (wai × (R 𝑎i − R bi )]
R ai is the investment return of ith market segment in the manager’s
portfolio
Attribution Analysis
Selection Effect = i (wai × (R 𝑎i − R bi )]
Selection effect measures the manager’s ability to form specific
market segment portfolios that generate superior returns relative
how comparable market segment is defined in the benchmark
portfolio (𝑅𝑎𝑖 − 𝑅𝑏𝑖 )
This, in turn, is weighed by the manager’s actual market segment
investment proportions (𝑤𝑎𝑖 )
