Title: INDEX MODELS
1INDEX MODELS
2Outline of the Chapter
- Single factor model
- Common factor for the macroeconomic changes
- Single-index model
- SP 500 index is employed as the common factor
- Advantages and disadvantages of the single-index
model - Estimate a singel-index model
- Constructing portfolios with single-index models
- Practical Aspects
- Index vs Markowitz
- Tracking Portfolios
3A Single-Factor Security Market
- The method of finding the optimal risky portfolio
in the last chapter (Markowitz Model) depends on
the quality of estimates of expected security
returns, and the covariance matrix. - The Markowitz Model has two main difficulties
- The number of estimates for the expected returns
and covariances matrix is very high even for the
portfolios with small number of securities - The errors in the estimation of correlation
coefficients - In order to solve the problem of estimating
covariance and correlation uncertainty is
decomposed into system-wide versus firm-specific
sources.
4A Single-Factor Security Market (Continued)
- The rate of return on any security, i, can be
divided into two as the sum of its expected
return and the unanticipated components
(unexpected return) - riE(ri)ei
- E(ei)0
- s(ei) measures the uncertainty about the security
return - We assume that the security returns are affected
by one or more common variables - We assume in this chapter that security returns
are affected from a single factor - Single-factor security market
5A Single-Factor Security Market (Continued)
- Assume that there is a common factor, m
- m a macroeconomic variable that affects all
firms - riE(ri)m ei
- Where m is the uncertainty about the economy as a
whole and ei is the uncertainty about the
specific firm - m measures unanticipated macro surprises
- E(m)0 (over time surprises will average out to
0) - Sandard deviation sm
- has no subscript, common factor for all
securities - ei measures the firm-specific surprises
6A Single-Factor Security Market (Continued)
- m and ei are uncorrelated
- since ei is firm specific, independent of shocks
to the common factor that affect the entire
economy - s2i s2m s2(ei)
- The variance of ri is the total of two
uncorrelated sources, systematic and firm
specific - Cov(ri ,rj)Cov(mei,mej) s2m
- m generates correlation across securities
- all securities respond to the same macroeconomic
news - Firm-specific surprises (ei) are uncorrelated
across firms - m is uncorrelated with any of the firm-specific
surprises
7A Single-Factor Security Market (Continued)
- The sensitivity of the securities to the
macroeconomic shocks differ - riE(ri)ßim ei
- Each firm assigned a sensitivity coefficient to
macro conditions (ßi) - We obtain single-factor model
- Systematic risk of the security i is determined
by its beta coefficient
8A Single-Factor Security Market (Continued)
- Total risk of the security i
- s2i ß2i s2m s2(ei)
- The systematic risk of the security i ß2i s2m
- The unsystematic risk of the security i s2(ei)
- Cov(ri ,rj)Cov(ßimei, ßjmej) ßi ßjs2m
- The covariance between any pair of securities
also determined by their betas - Equivalent beta securities gives equivalent
market positions - Show the same sensitivity to the changes in the
macroeconomy
9A Single-Factor Security Market (Continued)
- There is a linear relationship between security
returns and the comon factor - However these models do not identify the common
factor - We are looking for a variable that we can employ
as a common factor in the specified models - This variable should be observable so that the
volatility and the sensitivity of the securities
returns to variation of this common factor can be
estimated
10The Single-Index Model
- The valid proxy for the common factor model is
the rate of return on a broad index of securities
such as the SP 500 - Single-Index Model
- Similar to single-factor model
- Uses market index as a proxy for the common
factor - The regression equation
- M market index (SP 500)
- RM excess return
- RMrM-rf
- sM standard deviation
11The Single-Index Model (Continued)
- Ri excess return of a security i
- Riri-rf
- We can estimate the sensitivity (beta)
coefficient of a security on the index using a
single-variable linear regression
12The Single-Index Model (Continued)
- The regression equation
- Ri(t)aißiRM(t)ei(t)
- ai intercept
- The security is expected excess return when the
market excess return is 0. - ßi the slope
- The security is sensitivity to the index
- The amount by which the security is return tends
to increase or decrease for every 1 increase or
decrease in the return on the index - ei residual
- Firm-specific surprise in the security return
13The Single-Index Model (Continued)
- The Expected Return-Beta Relationship
- E(Ri)aißiE(RM)
- E(ei)0
- ßiE(RM)
- Part of the security is risk premium is due to
the risk premium of the index (market) - The market risk premium is multiplied by the
relative sensitivity of the individual security - Systematic risk premium
- ai
- Nonmarket premium
- For example a maybe larger if you think a
security is underpriced and so offers an
attractive expected return
14The Single-Index Model (Continued)
- Risk and Covariance in the Single-Index Model
- Ri(t)aißiRM(t)ei(t)
- E(Ri)aißiE(RM)
- Total risksystematic riskfirm-specific risk
- s2i ß2i s2M s2(ei)
- Covarianceproduct of betasmarket index risk
- Cov(ri ,rj)ßi ßjs2M
- Correlationproduct of correlations with the
market index - Corr(ri ,rj) ßi ßjs2M/sisj
- ßi s2Mßjs2M/sisMsjsM Corr(ri ,rM)Corr(rj
,rM)
15The Single-Index Model (Continued)
- Different than Markowitz model we only need to
estimate a, ß, and s(e) for the individual
securities and risk premium (RM) and variance of
the market index (s2M) - The Set of Estimates Needed for the Single-Index
Model - ai
- The stocks expected return if the market is
neutral, if the markets excess return, rM-rf, is
zero - ßi (rM-rf)
- The component of return due to movements in the
overall market - ßi is the securitys responsiveness to market
movements
16The Single-Index Model (Continued)
- ei
- The unexpected component of return due to
unexpected events that are relevant only to this
security (firm-specific) - ß2i s2M
- The variance attributable to the uncertainty of
the common macroeconomic factor - s2(ei)
- The variance attributable to firm-specific
uncertainty - The advantages of the single-index model
- Decrease the number of inputs to be estimated for
the model - Suggests a simple way to compute covariances
17The Single-Index Model (Continued)
- Disadvantage of the single-index model
- The model divides the uncertainty into two as
macro and micro - Ignores the other sources of dependence in stock
returns - Events that affect the firms in the same industry
without affecting the broad macroeconomy - Ignores the correlation between residuals (ei) of
the firms in the same industry - Multi-index models models which includes
additional factors to capture those extra sources
of cross-security correlation
18The Single-Index Model (Continued)
- The Index Model and Diversification
- Sharpe
- First suggested the index model
- Offers insight into portfolio diversification
19The Single-Index Model (Continued)
- As the number of stocks included in this
portfolio increases, the part of the portfolio
risk attributable to nonmarket factors become
ever smaller - Risk attributable to nonmarket factors is
diversified away - Risk attributable to market factors remains
- The portfolios variance
- s2P ß2P s2M s2(eP)
20The Single-Index Model (Continued)
- ß2P s2M systematic risk component
- component that depends on marketwide movements
- depends on the sensitivity coefficients of
individual securities - this part of risk depends on portfolio beta
and variance of market return and persist
regardless of diversification - s2(eP) nonsystematic risk component
- attributable to firm-specific component
ei - as number of stocks increase the
firm-specific risk component cancells out, so
there is smaller nonmarket risk - diversifiable risk
21The Single-Index Model (Continued)
- As diversification increase the total variance
of a portfolio approaches the systematic variance - The variance decrease because of the decrease in
firm-specific risk
22Estimating the Single-Index Model
- Example
- Estimate the equation
- Ri(t)aißiRM(t)ei(t)
- Data
- Six large US corporations and their stocks
- Hewlett-Packard, Dell (IT), Target, Wal-Mart
(retail), British Petroleum, and Royal Dutch
(energy) - SP 500 portfolio
- T-bills rate
- April 2001-March 2006 (monthly data-60
observations)
23Estimating the Single-Index Model (Continued)
- Lets look at the index model regression for
Hewlett-Packard (HP) - RHP(t)aHPßHPRSP500(t)eHP(t)
- Equation describes the linear dependence of HPs
excess return on changes in the state of the
economy (excess return of the SP500 index
portfolio) - The regression estimates a straight line with the
intercept (aHP) and slope (ßHP) - It is called the security characteristic line
(SCL) for HP
24Estimating the Single-Index Model (Continued)
- The HP returns generally follow those of the
index but with much larger swings - The swings in HPs excess return suggests a
greater-than-average sensitivity to the index - Beta greater than 1
25Estimating the Single-Index Model (Continued)
26Estimating the Single-Index Model (Continued)
- Regression Statistics
- Multiple R the correlation of HP with the SP
500 - R-square the variation in the SP 500 excess
returns explains what portion of the variation in
the HP series - Adjusted R-square corrects for an upward bias in
R-square - Standard error the standard deviation of the
residual (s(eHP))
27Estimating the Single-Index Model (Continued)
- Analysis of Variance (ANOVA)
- Sum of squares (SS) of the regression the
portion of the variance of the HPs return that
is explained by the SP 500 return - MS for the residual the variance of the
unexplained portion of HPs return, portion of
return that is independent of the market index - s2(eHP)
- SS(residual)/58
- Square root is equal to standard error of the
regression
28Estimating the Single-Index Model (Continued)
- Total SS of the regression/(n-1) the estimate of
the variance of the dependent variable - s2HP
- Note annualized standard deviationmonthly
standard deviation(12)1/2 - R-square explained to total variance
- the explained (regression) SS/total SS
- Estimate of a
- The average value of HPs return net of the
impact of market movements - The nonmarket component of HPs return is the
actual return minus the return attributable to
market movements
29Estimating the Single-Index Model (Continued)
- Rfirm-specificRfsRHP-ßHPRSP500
- The intercept term (a) is insignificant
- The value of the intercept is not statistically
different than 0 - Estimate of ß
- The beta coefficient for the HP indicates high
market sensitivity (greater than market beta,
which is equal to 1) - The slope term (ß) is significant
- The slope is statistically different than 0
- The slope is statistically different than 1
30Estimating the Single-Index Model (Continued)
- Firm-specific risk
- The annual standard deviation of HPs residual is
26.6 and the standard deviation of systematic
risk is ßs(SP 500)27.57 - ß2HPs2SP500SS Regression/n-1
- HPs firm specific risk is as large as its
systematic risk, a comon result for individual
stocks
31Portfolio Construction and the Single-Index Model
- How to construct portfolios by employing
single-index models - The biggest advantage of the single-index model
is the simple framework it provides for
macroeconomic and security analysis when
preparing the input list (needed estimates) to
have efficient optimal portfolio - The Markowitz model requires estimates of risk
premiums for each security - The estimate of expected return depends on both
macroeconomic and individual-firm forecasts
32Portfolio Construction and the Single-Index Model
(Continued)
- Different analysts may have different forecasts
related to macroeconomy and have different
expected return estimates for the same securities - The underlying assumption for market-index risk
and return often are not explicit in the analysis
of individual securities - The single-index model helps to seperate sources
of return variation (macroeconomic and
firm-specific sources of variation) and makes it
easier to ensure consistency across analysts
33Portfolio Construction and the Single-Index Model
(Continued)
- Steps in the single-index model
- Macroeconomic analysis is used to estimate the
risk premium and risk of the market index - Statistical analysis is used to estimate the beta
coefficients of all securities and their residual
variances - The portfolio manager uses the estimates for the
market-index risk premium and the beta
coefficient of a security to establish the
expected return of that security absent any
contribution from security analysis - Security-specific expected return forecasts
(security alphas) are derived from securtiy
valuation models
34Portfolio Construction and the Single-Index Model
(Continued)
- Thus, the risk premium on a security (ßiE(RM)),
risk premium derived from the securitys tendency
to follow the market index, is not subject to
security analysis. However any expected return
beyond this benchmark risk premium (the security
alpha) would be due to some nonmarket factor and
is subject to security analysis - Alpha is very important for the portfolio manager
- Tells whether a security is a good or bad buy
- There may be many other securities with the same
systematic component (beta coefficient) - The positive alpha security provides premium and
should have a higher weight in the portfolio. On
the other hand the negative alpha security is
overpriced and should have less weight in the
portfolio
35Portfolio Construction and the Single-Index Model
(Continued)
- Assume a portfolio manager plans to compile a
portfolio from a list of n actively researched
firms and a passive market index portfolio. - The input list for n actively researched firms
and a passive market index portfolio - Risk premium on the SP 500 portfolio
- Estimate of the standard deviation of the SP 500
portfolio - n sets of estimates for beta coefficients, stock
residual variances, and alpha values
36Portfolio Construction and the Single-Index Model
(Continued)
- Ri(t)aißiRM(t)ei(t)
- E(Ri)aißiE(RM)
- Cov(ri ,rj)ßi ßjs2M
- If the SP 500 index is treated as the market
index, - it will have a beta of 1 (its sensitivity to
itself) - no firm-specific risk
- an alpha of 0 (there is no nonmarket component in
the expected return of the market) - covariance of any security i with the index is
ßis2M (since ßM1)
37Portfolio Construction and the Single-Index Model
(Continued)
- Optimal risky portfolio with single-index model
- Generate (n1) expected returns by employing
estimates of beta and alpha coefficients and risk
premium of the index portfolio - Construct the covariance matrix by employing the
estimates of the beta coefficients, residual
variances and variance of the index portfolio - Then maximize the Sharpe Ratio for the
constructed portfolio (find the steepest CAL) to
get the optimum risky portfolio
38Portfolio Construction and the Single-Index Model
(Continued)
- The optimal risky portfolio is composed of an
active portfolio (A) and a passive portfolio (M) - A composed of n analysed securities (formed as a
result of active security analysis) - M the (n1)th asset used for diversification
39Portfolio Construction and the Single-Index Model
(Continued)
40Portfolio Construction and the Single-Index Model
(Continued)
- The Sharpe ratio of an optimally constructed
risky portfolio will exceed that of the index
portfolio (the passive strategy) - S2P S2MaA/s(eA)2
- The contribution of active portfolio (when held
in its optimal weight, wA) to the Sharpe ratio
of the overall risky portfolio is determined by
the ratio of its alpha to its residual standard
deviation - aA/s(eA) is called information ratio
- Measures the extra return we can obtain from
securtiy analysis (aA) compared to the
firm-specific risk
41Portfolio Construction and the Single-Index Model
(Continued)
- The information ratio reveals that the positive
contribution of the security to its portfolio is
made by its addition to the nonmarket risk
premium (its alpha) and its negative contribution
is made by increasing the portfolio variance by
its firm-specific risk (residual variance)
42Practical Aspects of Portfolio Management with
the Index Model
- Index vs Markowitz model
- The Markowitz model allows flexibility in our
modelling of asset covariance structure but the
problem is the errors in the covariance estimates - The single-index model is more practical and
helps us to seperate macro and security analysis
43Practical Aspects of Portfolio Management with
the Index Model (Continued)
- The single-index model provides a benchmark for
security analysis - A portfolio manager who has no info about the
security will assume alpha value as 0 and
forecast a risk premium for the security (ßiRM) - If we restate the model in terms of total returns
- E(rHP)rfßHPE(rM)-rf
- A portfolio manager who has a forecast for the
market index, E(rM), and observes the risk-free
T-bill rate, rf,can use the model to determine
the benchmark expected return for any stock
44Practical Aspects of Portfolio Management with
the Index Model (Continued)
- The beta coefficient, the market risk, s2M, and
the firm-specific risk, s2(e), can be estimated
from historical SCLs, from regressions of
security excess returns on market index excess
returns
45Practical Aspects of Portfolio Management with
the Index Model (Continued)
- Tracking Portfolios
- Example
- A portfolio manager believes that she identified
an underpriced portfolio (positive alpha) - RP0.041.4RSP500eP
- The alpha value of P is 4 and the beta is 1.4
- The downfall of the market will affect the value
of the portfolio negatively if she buys this
portfolio even if it is relatively underpriced,
since the beta coefficient is larger than 1
46Practical Aspects of Portfolio Management with
the Index Model (Continued)
- She would like to take a position that takes
advantage of her teams analysis but is
independent of the performance of the overall
market - She constructed a tracking portfolio (T)
- A tracking portfolio for portfolio P is a
portfolio designed to match the systematic
components of Ps return - The tracking portfolio must have the same beta on
the index portfolio as P and as little
nonsystematic risk as possible - This procedure is also called beta capture
47Practical Aspects of Portfolio Management with
the Index Model (Continued)
- T will have a levered position in SP 500 to have
a beta of 1.4 - T includes position of 1.4 in the SP 500 and
-0.4 in T-bills. - T is constructed from the index and T-bills so
alpha is 0 - Buy portfolio P and short sell portfolio T (you
offset the systematic risk) - RCRP RT (0.041.4RSP500eP)-1.4RSP500
- 0.04eP
- C is still risky due to residual risk, eP, the
systematic risk has been eliminated - If P is well diversified the remaining
non-systematic risk will be small as well - The aim is achieved. The manager can take the
advantage of the 4 alpha without taking on
market risk