Title: Module 1 The investment setting and Modern portfolio Theory
1Module 1The investment settingand Modern
portfolio Theory
2(No Transcript)
3(No Transcript)
4Portfolio Management
- Purpose maximization of wealth by reaching a
heuristic Reward-to-risk - How? Allocate, Select and Protect
- Illustration realized and expected wealth?
- Realized wealth Expected wealth Error
- Heuristic Reward to risk Allocation
Selection protection - It always starts with the Policy
- Ask the right question!? what risk? ?Thus, what
allocation? - Set the right allocation target in terms of
objectives, constraints and weight range
monitoring
5Choose a Portfolio strategy Passive or Active
Asset allocation Security Selection
Active (for pros) Market timing Stock/Bond picking
Passive (for ind.) Fixed weights Indexing
- No matter what, an investment strategy is based
on four decisions - What asset classes to consider for investment
- What normal or policy weights to assign to each
eligible class - The allowable allocation ranges based on policy
weights - What specific securities to purchase for the
portfolio - Most (85 to 95) of the overall investment
return is due to the first two decisions, not the
selection of individual investments
6First, set the rules the policy statement
- TOTAL RETURN INCOME YIELD CAPITAL GAIN YIELD
- Objectives Think in terms of risk and return to
find the best weightsi.e., - Capital preservation (high income, low capital
gain)? Low to moderate risk - Balanced return (Balanced capital gains and
income reinvestment)?moderate to high risk - Pure Capital appreciation (high capital gains,
low to no income)?High risk - Constraints - liquidity, time horizon, tax
factors, legal and regulatory constraints, and
unique needs and preferences - Management - Define an allowable allocation
ranges based on policy weights - Selection - Define guideline to pick securities
to purchase for the portfolio (optional)
7(No Transcript)
8Objectives ?Age/Risk Matrix
Risk tolerance/ Time Horizon 0-5years (C/B/S) 6-10 (C/B/S) 11 (C/B/S)
Higher 10/30/60 0/20/80 0/0/100
Moderate 20/40/40 10/40/50 10/30/60
Lower 50/40/10 30/40/30 10/50/40
- C stands for CASHi.e. money market securities
- B stands for Bondsi.e. corporate, municipal or
treasury securities - S stands for Stocksi.e. value, growth,
international equity securities - Color code
- Capital preservation
- Balanced return
- Capital appreciation
- Where do you fit?
- ?Retirement simulation comes next
9In Sum (1),
- Active strategy will likely outperform a passive
strategy - Asset allocation is more powerful than stock
selection - Everything starts with the policy statementi.e.,
(1) types and weights of assets to include in
portfolio and (2) guidelines on how to manage
the allocation and selection process. - Risk of a strategy depends on the investors
goals and time horizon - Over long time periods sizable allocation to
equity will improve results - Over short time periods sizable allocation to
fixed income securities will shield against
uncertainty
10Example Case
- Mr. Bob is 70 years of age, is in excellent
health pursues a simple but active lifestyle, and
has no children. He has interest in a private
company for 90 million and has decided that a
medical research foundation will receive half the
proceeds now it will also be the primary
beneficiary of his estate upon his death. Mr. Bob
is committed to the foundation s well-being
because he believes strongly that , through it, a
cure will be found for the disease that killed
his wife. He now realizes that an appropriate
investment policy and asset allocation are
required if his goals are to be met through
investment of his considerable assets. Currently
the following assets are available for building
an appropriate portfolio - 45 million Cash (from the sale of the private
company interest, net of 45 million gift to the
foundation) - 10 million stocks and bonds (5 million each)
- 9 million warehouse property not fully leased)
- 1 Million Bob residence
- Build a policy statement for Mr. Bob!
11Objectives (return)
- Large liquid wealth from selling interest in the
private company - Income from leasing warehouse
- Not burdened by large or specific needs for
current income nor liquidity. - He has enough spendable income.
- He will leave his estate to a Tax-exempted
foundation - He has already offered a large gift to the
foundation - Thus, an inflation-adjusted enhancement of the
capital base for the benefit of the foundation
will the primary minimum return goal. - He is in the highest tax bracket (not mentioned
but apparent) - Tax minimization should be a collateral goal.
12Objectives (risk)
- Unmarried, Childless, 70 years old but in good
health - ? Still a long actuarial life (10), thus long
term return goal. - Likely free of debt (not mentioned, but neither
the opposite) - Not skilled in the management of a large
portfolio - Yet, not a complete novice since he owned stocks
and bonds prior to his wifes death. - His heirthe foundationhas already received a
large asset base. - ?Long term return goal with a portfolio bearing
above average risk.
13Constraints
- Time--Two things (1) long actuarial life and (2)
beneficiary of his estatethe foundation has a
virtually perpetual life - Taxes highest tax brackets, investment should
take this into consideration tax-sheltered
investments. - Unique circumstances Large asset base, a
foundation as a unique recipient? some freedom in
the building of the portfolio
14Adapted Strategy
- Majority in stocks (shield against inflation,
above average risk tolerance, and no real income
or liquidity needs) - He already has 15 in real estate (house
warehouse)? no more needed, diversification
effect achieved. - Additional freedom Non-US stocks? additional
diversification - ? Target 75 equity (including Real Estate)
- Fixed Income used to minimize income taxesi.e.,
municipal and treasury securities. No need to
look for YIELD nor downgrade the quality of the
issues used. - Additional freedom Non-US fixed-income?
additional diversification effect. - ? Target 25 in fixed income
15Proposed Allocation
Current Proposed Range
Cash / Money Market 70 0 0-5
US Stocks--LC 30 30-40
US StocksSC 15 15-25
Non US Stocks 15 15-25
Total 7.5? 60 60-80
Real Estate 15 15 10-15
US Fixed Income 15 10-20
Non-US Fixed Income 10 5-15
Total Fixed Income 7.5? 25 15-35
16What is Investments?
- Purpose maximization of portfolio wealth through
adequate Portfolio management - Fair Reward-to-risk? Ask the right question!
- Optimal portfolio management Allocation
Selection Risk protection
17(No Transcript)
18Historical Return
- The additional cents on the dollar invested
- R(profitadditional cash flows)/initial
investment - Over a period of timeaverage return
- Average returnS(all returns)/nb of observations
- Why do returns matter?
- does not mean muchalone
- Cross-comparison between markets
- Are normally distributed
19Historical Risk
- We need to think in terms of estimates in an
uncertain world - Estimateaverage return /- some volatility
- Uncertainty or volatility of returns
- Standard deviation of returns
- Measured in
- What does it mean?
20Example Historical Returns and Risks
- Computation of Monthly Rates of Return
21Variance (Standard Deviation) of Expected Returns
for an Individual Investment
Standard deviation is the square root of the
variance Variance is a measure of the variation
of possible rates of return Ri, from the expected
rate of return E(Ri)
- where Pi is the probability of the possible rate
of return, Ri
22Example Expected Return and Risk
Variance ( 2) .00050 Standard Deviation (
) .02236
23Covariance and Correlation
- Covariance is a measure of the degree to which
two variables move together relative to their
individual mean values over time - The correlation coefficient is obtained by
standardizing (dividing) the covariance by the
product of the individual standard deviations
(Correlation coefficient varies from -1 to 1)
24Example case
- Look at the following data () and tell the
difference between an arithmetic and a geometric
average. Also, which of these asset classes is
the most attractive? What is the chance that you
will break-even if you invest in stocks only?
What is the chance that you make more than 10
return (arithmetic) if you invest in corporate
bonds only? If you hold only a real estate
portfolio, what would be the rational for adding
other classes?
Ar. Average Geo. Average Standard dev.
Stocks 10.28 8.81 16.9
T-bills 6.54 6.49 3.2
LT Gov Bonds 6.1 5.91 6.4
LT Corp Bonds 5.75 5.35 9.6
Real Estate 9.49 9.44 3.5
Correlations Stocks T-bill LT Gov Bonds LT Corp. Bonds Real estate
Stocks 1
T-bills 0.11 1
LT Gov Bonds 0.27 0.89 1
LT Corp Bonds 0.33 0.46 0.76 1
Real Estate 0.29 0.06 -0.02 0.08 1
25Solution
- The arithmetic average assumes the presence of
simple interest, while the geometric average
assumes compounding (interest-on-interest).
Ranking is best accomplished by using the
coefficient of variation (standard
deviation/arithmetic mean, multiplied by 100) - 1 - Real Estate 36.88
- 2 - Treasury Bills 48.93
- 3 - Long Gov't Bonds 104.92
- 4 - Common Stocks 164.37
- 5 - Long Corp. Bond 166.96
- To get the probability of success or failure,
compute a Z-score to be transformed into a
probability - P(Rstockgt0)gt z(Rstock 0)/standard deviation
stock 10.28/16.90.608 - P(Rstockgt0)P(0.608) 72.9
- P(Rcorp bondsgt10)gt z(Rbond 10)/standard
deviation bond (5.75-10)/9.6-0.44 - P(Rcorp bondsgt10) P(-0.44)1-P(0.44)1-0.6733
- It seems at first that government bonds offer
less return and more risk than real estate.
However, real estate and government bonds will
provide a good combination as they do not
fluctuate in a similar fashion, so that the
variability of the portfolio is less than the
variability of the individual investments (the
correlation coefficient applicable to this pair
of investments is known and is slightly negative.
26- A note on real estate investments a risk
component of low liquidity is not included in the
standard deviationso, Real estate might not be
so much a better investment.
27Risk and Return
- How to compare assets?
- Coefficient of variation measure of relative
risk - CV Total risk/return
- CS 1.56
- SCS 1.91
- CB 1.41
- TB 1.64
- Rf 0.84
- Which one do you pick?
- What is the problem here?
28Portfolio effect
- Portfolio Return is the weighted average return
of each asset in the portfolio - Portfolio Risk is not the weighted average risk
of each asset in the portfolio. - Portfolio risk has to do with each assets weight
and risk, but also the degree to which they move
together (?)
29Portfolio Standard Deviation Formula (with
covariance)
- Any asset of a portfolio may be described by two
characteristics - The expected rate of return
- The expected standard deviations of returns
- The correlation, measured by covariance, affects
the portfolio standard deviation - Low correlation reduces portfolio risk while not
affecting the expected return
30Mathematical Explanation
31Summary Portfolio effect
- Portfolio return (RP)
- Average return of all securities
- Portfolio risk (sP)
- Average risk of all securities
- Minus
- the propensity of those securities to be
unrelated (returnwise!)
32Portfolio risk and returnin English
- Portfolio return
- (weighted) average assets return
- Portfolio risk
- (weighted) average assets risk
- (weighted) average assets prices propensity to
move in opposite direction - Or
- Portfolio risk
- (weighted) average assets risk
- - Benefits from diversification
33(No Transcript)
34Example Combining Stocks with Different Returns
and Risk
1 .10 .50
.0049 .07 2
.20 .50 .0100 .10
- Case Correlation Coefficient
Covariance - a 1.00
.0070 - b 0.50
.0035 - c 0.00
.0000 - d -0.50
-.0035 - e -1.00
-.0070
35Combining Stocks with Different Returns and Risk
- Assets may differ in expected rates of return and
individual standard deviations - Negative correlation reduces portfolio risk
- Combining two assets with -1.0 correlation
reduces the portfolio standard deviation to zero
only when individual standard deviations are equal
36Example Combining Stocks with Different Returns
and Risk
1 .10 .07 2
.20 .1
37Portfolio Risk-Return Plots for Different Weights
E(R)
2
With two perfectly correlated assets, it is only
possible to create a two asset portfolio with
risk-return along a line between either single
asset
Rij 1.00
1
Standard Deviation of Return
38Portfolio Risk-Return Plots for Different Weights
E(R)
f
2
g
With uncorrelated assets it is possible to create
a two asset portfolio with lower risk than either
single asset
h
i
j
Rij 1.00
k
1
Rij 0.00
Standard Deviation of Return
39Portfolio Risk-Return Plots for Different Weights
E(R)
f
2
g
With correlated assets it is possible to create a
two asset portfolio between the first two curves
h
i
j
Rij 1.00
Rij 0.50
k
1
Rij 0.00
Standard Deviation of Return
40Portfolio Risk-Return Plots for Different Weights
E(R)
With negatively correlated assets it is
possible to create a two asset portfolio with
much lower risk than either single asset
Rij -0.50
f
2
g
h
i
j
Rij 1.00
Rij 0.50
k
1
Rij 0.00
Standard Deviation of Return
41Portfolio Risk-Return Plots for Different Weights
Exhibit 7.13
E(R)
f
Rij -0.50
Rij -1.00
2
g
h
i
j
Rij 1.00
Rij 0.50
k
1
Rij 0.00
With perfectly negatively correlated assets it is
possible to create a two asset portfolio with
almost no risk
Standard Deviation of Return
42Numerous Portfolio Combinations of Available
Assets
43Efficient Frontier for Alternative Portfolios
44Efficient Frontier In Practice (all equity
markets of the world 1981-2001)
45Application 9 different Institutional efficient
Benchmarks
Asset Allocation and cultural Differences
- Mindset, Social, political, and tax environments
- U.S. institutional investors average 45
allocation in equities - In the United Kingdom, equities make up 72 of
assets - In Germany, equities are 11
- In Japan, equities are 24 of assets
46(No Transcript)
47Example What is the use of an efficient set?
- Goal find an optimal mix (weight) so that the
ratio of compensation for risk to risk (or reward
to risk) is optimal for your level of risk
tolerance. - Your inputs Expected returns, standard
deviations and correlations (for each asset
class) - Your output Optimal weight in each asset class
(how much should you put in each asst class?)
48Can we do better than the efficient set?
- Imagine two portfolio (1) a risky best of the
best portfolio with an expected return of Rm and
a standard deviation of sm and (2) a riskless
portfolio of t-bills with an expected of Rf and a
standard deviation close to zero. - You allocate Wrf in the riskless portfolio and
(1-Wrf) in the risky (best of the best portfolio) - The standard deviation and expected return of
this portfolio shall be - sp(1-Wrf) x sm or Wrf1- sp/sm, then
- RpWrf x Rf (1-Wrf) x Rp replace Wrf by 1-
sp/sm - RP Rf (Rm Rf) /sm x sp?Capital Market
Line (CML) - Rp intercept slope x sp
49What does it mean?
50Example Illustration of the separation theorem
Wrfgt0
Wrflt0
- How is the concept of leverage included in the
CML?
Wrf0
51It means that
- We know how to get the composition of the
best-of-the-best portfolio (M)? It has the
highest reward to risk i.e., (Rm Rf)/sm - Then, we know how to get Rm and sm
- Finally, for the risk we are willing to take
(indifference curve? policy statement), we can
find our optimal asset allocation by mixing the
best of the best portfolio with cash! - Cool (I mean sweeeeet) huh?
- Application efficient frontier analysis
52Example The CML and the allocation process
- Step 1 Estimate the expected returns, standard
deviations and correlations for each asset
classes - Step2 Find the the allocation mix as well as the
expected return and standard deviation of the
M--i.e., M should be optimized so that its
reward-to-risk (Rm-Rf)/sigma_M is optimal - Step 3 Choose an allocation for your utility for
risk using the CML. You will invest Wrf in the
risk-free rate and (1-Wrf) in M. - ? Application Say that Rf has an expected return
of 5 M has a standard deviation of 20 and an
expected return of 30 and it consists of an
allocation of 40 in foreign stocks, 15 in
foreign bonds, 20 in domestic stocks and 25 in
domestic bonds. What is the expected return, risk
and allocation in the following 2 cases (1) you
are willing to accept 15 risk and (2) you are
willing to accept 25 risk.
53Solution
- For a standard deviation of 15
- Allocation
- Wrf1-(15/20)25
- W(foreign stocks) (1-25) x 4030
- W(foreign bonds)(1-25) x 15 11.25
- W(domestic stocs)(1-25) x 2015
- W(domestic bonds)(1-25) x 2518.75
- Expected return 25 x 5 75 x 3023.75
- For a standard deviation of 25
- Allocation
- Wrf1-(25/20)-25 (borrow 25 at the risk free
rate and use it to invest in M) - W(foreign stocks) (125) x 4050
- W(foreign bonds)(125) x 15 18.75
- W(domestic stocs)(125) x 2025
- W(domestic bonds)(125) x 2531.25
- Expected return -25 x 5 125 x 3036.25
54The selection process Intuitive Risk for
individual assets
- Return(asset) expected unexpected
- Risk (return) 0 market risk business risk
- The trick if you hold many securities, the
particularities of each security becomes
irrelevantthus, in a well diversified portfolio
business-specific risk is irrelevant!
55A Risk Measure for the individual assets a more
rigorous Explanation
- Because all individual risky assets are part of
the M portfolio, an assets rate of return in
relation to the return for the M portfolio may be
described using the following linear model
where Rit return for asset i during period
t ai constant term for asset i bi slope
coefficient for asset i RMt return for the M
portfolio during period t random error
term
56Variance of Returns for a Risky Asset
57Measuring Components of Risk
- ?i2 bi2 ?m2 ?2(ei)
- where
- ?i2 total variance
- bi2 ?m2 systematic variance
- ?2(ei) unsystematic variance
58Examining Percentage of Variance
- Total Risk Systematic Risk Unsystematic Risk
- The proportion of systematic risk is Systematic
Risk/Total Risk R2 - so ßi2 ? m2 / ?i2 R2
- The proportion of diversifiable risk is then
measured by (1- R2 )
59Risk Reduction with Diversification
St. Deviation
Unique Risk s2(eP)s2(e) / n
bP2sM2
Market Risk
Number of Securities
60(No Transcript)
61Risk and Return for a single asset
- The higher the risk, the greater the expected
return. - RiReal rate Inflation premium Risk premium
- Ririsk free rate compensation for risk
- Compensation for riskrisk premiumcompensation
for a high systematic risk
62Risk that matters
- If only market risk matters, then the risk
premium of a security should be related (somehow)
to the market risk premium! - Lets assume that those risk premiums are
proportional - security risk premiumß x market risk premium
- This ß is a multiplier which has to do with the
relative risk premium of a security to the market
risk premiumit is a relative Market (systematic)
Risk
63SML
- RiRF RRP, then
- Security risk premium (Ri- RF)
- Market risk premium (Rm- RF)
- If security risk premiumß x market risk premium
- Then, (Ri- RF) ß x (Rm- RF)
- That is,
- Ri RF ß x (Rm - RF)
- This is also known as the SML (market
equilibrium), a component of the CAPM - As a result, any securitys return can calculated
using ß, RRF, and Rm
64Formal derivation of CAPM
65Some things to know about beta
- What is betas formula
- Ans Beta COV(I,M)/VAR(M)
- What is the market relative risk (ß)?
- Ans 1
- What does a ß of 2 mean?
- Ans twice more risky that the market
- What does a ß of 1 mean?
- Ans negatively correlated with the market
- What is the beta of a risk-free security?
- Ans 0
- How do we get ß?
- Ans regression or above formula
- What is the ß of a portfolio?
- Ans The weighted average of each stocks beta
66Graph of SML
- What if the observed returns are different from
the theoretical returns? - The Alpha-strategy consists of finding securities
with abnormal excess return.
67Example A simple illustration
E(RA) 0.06 0.70 (0.12-0.06) 0.102
10.2 E(RB) 0.06 1.00 (0.12-0.06) 0.120
12.0 E(RC) 0.06 1.15 (0.12-0.06) 0.129
12.9 E(RD) 0.06 1.40 (0.12-0.06) 0.144
14.4 E(RE) 0.06 -0.30 (0.12-0.06) 0.042
4.2
68Comparison of Required Rate of Return to
Estimated Rate of Return
69Plot of Estimated Returnson SML Graph
.22 .20 .18 .16 .14 .12 Rm .10 .08 .06 .04 .02
C
SML
A
E
B
D
.20 .40 .60 .80
1.20 1.40 1.60 1.80
-.40 -.20
70Question How can we use the SML to select
underpriced securities?
- According to the SML
- Ri-Rf 0 ß x (Rm-Rf)
- In a regression format (Ri-Rf) a ß x (Rm-Rf)
e - Then (alpha strategy)
- if a is not significantly different from 0,
security is fairly priced - if a is significantly greater 0, security is
underpriced - if a is significantly smaller than 0, security is
overpriced
71Illustration Alpha Strategy
x A
x B
a
72Alpha-strategy example
- SML (Ri Rf) alpha beta x (Rm-Rf) e
- Example EXTR
Alpha Beta R2
EXTR (t-stat) 0.019 (2.28) 1.47 (4.57) 0.25
What is the amount of diversifiable risk in EXTR
if its standard deviation is 20? Ans 20 x
(75)1/2 17.32
73In Sum,
- What is the difference between the CML and SML?
Why are the measures of risk different? - RP Rf (Rm Rf) /sm x sp ?CML (allocation)
- Ri Rf (Rm-Rf) x si/ sm x ri,m ?SML
(selection) - Ri Rf (Rm-Rf) x ßi,m
74Lets conclude and summarize now
- Develop an investment policy statement
- Identify investment needs, risk tolerance, and
familiarity with capital markets - Identify objectives and constraints
- Investment plans are enhanced by accurate
formulation of a policy statement - ALLOCATION determine the market/sector weights
- Asset allocation determines long-run returns and
risk, which success depends on construction of
the policy statement - (1) EFFICIENT FRONTIER and (2) CML
- CML EFFICIENT FRONTIER when T-Bill is included
in the efficient set - SELECTION determine undervalued securities
- Actual (observed of predicted) Return Vs. SML
(fair) return - Alpha Analysis Is the SML significantly
violated? - Optimal allocation between selected securities
with the efficient frontier
75Example
- You are a new analyst with a small mutual fund.
You have been assigned to make a presentation to
clients on portfolio management. You need to
discuss two main issues (1) diversification
(relationship between portfolio variance and
correlation) and (2) asset pricing (difference
between the CML and SML). Be ready to answer both
issues.
76Problems With CAPM
- 1. Beta coefficients are not stable for
individual securities. - Performance evaluation depends upon the choice
of the market proxybenchmark error. - T-bills are not exactly risk-free
- Unpleasantries have been neglected (taxes and
transaction cost)
77Empirical Tests of the CAPM
- A theory should not be judged on the basis of the
reasonableness of its assumptions. Instead, the
test of any model is how well it explains the
observed facts, and then how well it can predict. - There are two main questions about the CAPM that
need to be answered - 1) How stable is beta, and
- 2) Is the risk -return relationship linear as
suggested by the SML? That is, how well do the
returns conform to the equation of the security
market line?Positive linear between beta and
risky assets returns? E(Ri) RFR bi(Rm -
RFR) - Specific questions might include
- Does the intercept approximate the RFR that
prevailed during the test period? - Was the slope of the line positive and was it
consistent with the slope implied by the risk
premium (Rm - RFR) that prevailed during the test
period?
78Stability of Beta
Individual stocks versus portfolios?Individual
stock betas have been found to be generally more
volatile over time whereas portfolio betas were
stable. In fact, Levy and Blume concluded that
portfolio betas are much more stable than betas
for individual stocks. Also, betas have a
tendency to regress toward one. Tole observed
greater stability of betas for larger portfolios
and that benefits are realized in having over a
100 securities in a portfolio. Time period?Basel
found that the stability of beta estimates rose
with larger estimation periods. Roenfeldt,
Grienpentrog and Pflamm found that 48 month betas
were not good for estimating subsequent 12-month
betas, but were good for estimating 24, 36, and
48-month betas. Trading volume? Carpenter and
Upton contended that that volume adjusted betas
were slightly better than ordinary predictions.
79Comparability of Published Estimates of Beta
- Merrill Lynch?60 monthly returns with SP 500
- Value Line? 260 weekly returns with NYSE
Composite - Ri RFR biRm Et ? badjusted Merrill
Lynch 0.127 badjusted Value Line R2 .55
- It is often convenient to use the betas published
by Value Line or Merrill Lynch. The computation
of the betas, however, use different data.
Merrill Lynch uses 60 monthly observations and
SP 500 for market proxy while. Value Line uses
260 weekly observations and NYSE composite series
as the market proxy. Both services use the same
regression model and follow slightly different
adjustment procedures to deal with regression
tendencies. Given these relatively minor
differences, you would expect the published betas
to be quite similar. - Statman compared these published betas for both
individual stocks and portfolios and found small
but significant differences between the two beta
estimates. For 195 individual stocks he found
the relationship with an R-square value of .55.
Reilly and Wright also found that difference was
due to the alternative time intervals? The
securitys market value affects both the size and
the direction of the interval effect. Therefore,
when estimating beta or using published sources,
you must consider the interval used and the
firms relative size.
80Relationship Between Systematic Risk and Return
The ultimate question regarding CAPM ? is there a
positive linear relationship between systematic
risk and the rates of return on risky assets?
Sharpe and Cooper (1972)?Positive relationship
between risk/return...not completely linear.
Black, Jensen, and Scholes (1972), Fama and
Macbeth (1973) ?Positive linear relationship
between monthly excess returns and betas. Douglas
(1969)? Systematic risk measures not significant
in explaining returns. Effect of Skewness on
Relationship investors prefer stocks with high
positive skewness that provide an opportunity for
very large returns Effect of Size, P/E, and
Leverage size, and P/E have an inverse impact on
returns after considering the CAPM. Financial
Leverage also helps explain cross-section of
returns ?Effect of Book-to-Market Value. For EX,
Fama and French found that Size and
book-to-market ratios explain returns on
securities. Also, Beta is not a significant
variable when other variables are included?Study
results support multifactor models)
81Summary of CAPM Risk-Return Empirical Results
Early studies supported CAPM but not without some
doubts regarding the higher-than-expected
intercept. Subsequently, the efficient markets
literature has provided extensive evidence that
both size and the P/E ratio were variables that
could help explain cross-sectional variation
addition to beta. Recent studies also have found
that financial leverage and the book-to-market
value of equity ratio (i.e., B/P ratio) have
explanatory power regarding returns beyond beta.
The Fama-French study concluded that between 1963
and 1990, beta was not related to average returns
on stocks when included with other variables, or
when considered alone. Moreover, the two
dominant variables were size and the book value
to market value ratio, which was even stronger
than size although both variables were
significant. Subsequent studies have supported
the Fama-French results. However, more recent
studies that have used longer time periods for
computing betas, and adjusting for expected
returns have provided support for the original
CAPM.
82The Market Portfolio Theory Vs Practice
Recall that the CAPM market portfolio is assumed
to include all the risky assets in the economy.
This would include art, real estate, stocks,
bonds, jewels, etc. In practice it is not
possible to choose a proxy that contains all
risky assets, so the SP 500, the NYSE composite
index, or some other market indicator must be
used as a proxy. Most academicians recognize
this problem and conclude that it is not
serious. Richard Roll concluded that the use of
indexes as a market proxy has a very serious
implication for tests of the model and especially
for using the model when evaluating portfolio
performance. Roll refers to it as a benchmark
error. With respect to tests of the CAPM, Roll
contends that prior to testing, an analysis is
required of whether the proxy for the market
portfolio is mean variance efficient (on the
Markowitcz efficient frontier) and whether it is
the true market portfolio.
83Differential CML Using Market Proxy That Is
Mean-Variance Efficient
84Differential Performance Based on an Error in
Estimating Systematic Risk
85Differential SML Based on Measured Risk-Free
Asset and Proxy Market Portfolio
86Relaxing the Assumptions
- The CAPM made a series of assumptions that may
appear to be unrealistic. One might ask what
happens to the results of the CAPM if we change
the assumptions upon which it was developed. - The assumptions that we will examine are
- Differential lending and borrowing rates
gtLending and borrowing at the risk-free rate? - Liquidity Risk Premium?
- What effect will transaction costs and taxes have?
87Differential Borrowing Lending Rates
- One of the assumptions of the CAPM was that
investors could both borrow and lend unlimited
amounts at the risk-free rate. - It seems reasonable to assume that investors can
lend unlimited amounts at the risk-free rate by
buying government securities (e.g., T-bills). - However, it is questionable whether investors can
borrow at the risk-free rate since it is usually
below the prime rate and most investors cant
even borrow at the prime rate.
88Investment Alternatives When the Cost of
Borrowing Is Higher Than the Cost of Lending
If borrowing rate (Rb) gt lending rate (RFR),
there are two different lines going to the
Markowitz efficient frontier. The line segment
RFR - F represents the investments opportunities
available we an investor lends at the RFR and
invests in Portfolio F. Opportunities on this
line beyond point F are unavailable because you
cannot borrow at the risk-free rate. Assuming
that borrowing takes place at Rb, the portfolio
that can be purchased with borrowed funds is
Portfolio K. Notice that Portfolio K is the
point of tangency between the borrowing rate, Rb,
and the efficient frontier. Due to the
differential lending and borrowing rates, the CML
becomes RFR - F - K - Rb.
89Zero-Beta Model? coping with Rfr
- If the market portfolio is mean-variance
efficient, then we can find several portfolios
that have no systematic risk (beta 0) and the
assumption that a risk-free rate exists is not
necessary. - From among the zero-beta portfolios, we can
select the one with the smallest variance. This
will not affect the CML but it will allow the
construction of a linear SML where the intercept
is the minimum variance zero-beta portfolio. - The combination of this zero-beta portfolio and
the market portfolio will be a linear
relationship in return and risk described by this
equation. - where, E(Ri), E(Rz), and E(Rm) are the expected
returns for security i, the zero-beta portfolio,
and the market portfolio, respectively
90Security Market Line With a Zero-Beta Portfolio
- Absence of a risk-free asset
- Combinations of portfolios on the efficient
frontier are efficient - All frontier portfolios have companion portfolios
that are uncorrelated - Returns on individual assets can be expressed as
linear combinations of efficient portfolios
Return
Zero beta
Risk
91CAPM Liquidity
- Liquidity
- Illiquidity Premium
- Research supports a premium for illiquidity
f (ci) liquidity premium for security i f (ci)
increases at a decreasing rate
92Illiquidity and Average Returns
Average monthly return()
Bid-ask spread ()
93Security Market with Transaction Costs
The assumption of no transaction costs in the
development of the original CAPM was so that
investors will buy or sell mispriced securities
until they plot on the SML line. When transaction
costs exist investors will not correct all
mispricing. Instead they will only correct
mispricing to the point that the excess return
due to the correction process is exactly offset
by the cost. As shown below, transaction costs
result in the SML becoming a band of securities.
Transaction costs also affect the level of
diversification that investors are willing to pay
to achieve. The additional transactions needed
to extract the last 10 of non-systematic risk in
a portfolio may be more expensive than the
benefits derived.
94GeneralizationWhat is a factor model?
- Factor models are used to operationalize
portfolio theory for equity portfolios - To construct portfolios that meet specific
objectives - To measure and control portfolio risk
- To evaluate portfolio performance
- They are all statistically based and designed to
reduce the portfolio selection problem - They require additional theoretical structure to
justify different approaches
95Types of Factor Models
- Single Factor Models
- Capital Asset Pricing Model (CAPM) and Market
Model - Multi-factor Models
- Macroeconomic Factor Models prespecify
macroeconomic indicators such as interest rates,
inflation, exchange rates (e.g. Arbitrage Pricing
Theory, APT, and BIRR) - Statistical Factor Models extract from
historical time-series and cross-section of stock
returns (e.g. Principal Components or Factor
Analysis) - Fundamental Factor Models use firm-level
attributes and market data, like P/B, P/E,
industry, momentum, trading activity, etc. (e.g.
Fama-French, Carhart, BARRA, Vestek)
96Examples of Factor Models
- E(Ri) rf
- ?i E(Rm) rf
- ?i,F1 E(RF1) rf ?i,F2 E(RF2) rf .
?i,FK-1 E(RFK-1) rf ?i,FK E(RFK) rf - ?i,m E(Rm) rf ?i,SMB ESMB ?i,HML
EHML ?i,Mom EMomentum
97Single Factor Model CAPM (Review)
- The CAPM is a centerpiece of modern finance that
gives predictions about the relationship between
risk expected return - Based on original work of Harry Markowitz by Bill
Sharpe, John Lintner Jan Mossin in mid-1960s - Portfolio Theory (Markowitz, 1952) how an
investor could select an optimum portfolio - CAPM (Sharpe, Lintner, Mossin, 1964) predicts
how aggregate of investors will behave how
prices will be set so that markets clear
98CAPM Aggregation (review)
- Begins with simplistic assumptions for
hypothetical world of investors builds into
comprehensive model - Investors are price takers
- Same one-period investment horizon (myopic)
- Fixed quantities of assets and all marketable
- No taxes, transactions costs, regulations, etc
- Risk-averse utility-maximizing Markowitz
investors - All investors analyze securities in same way with
same probabilistic forecasts for each - Since they must all choose the same optimal risk
(tangency) portfolio, this portfolio must be M,
the market portfolio, for market to clear (share
prices set to balance supply/demand)
99Portfolio Problem 3 ? CAPM (Review)
- Portfolio Problem 3 considers two steps
- Choose one optimal risk portfolio along efficient
frontier same for all investors, market M - Now, each investor (different wealth, different
risk tolerances) choose optimal capital
allocation with riskfree security along CAL for
portfolio M, or capital market line - CAPM equilibrium
- share prices balance supply and demand
- no pressure for prices to change
- rational investors passively hold market
portfolio
100CAPM Key implication (review)
- There is a linear relationship between expected
return for a security and the market risk
premium - E(Ri) rf ?i E(Rm) rf
- Beta, ?i, is a standardized measure of covariance
with market portfolio M ?i
Cov(Ri,RM)/?m2 - Expected return for stock equals riskfree return
PLUS market risk premium scaled by beta of stock
101Capital Asset Pricing Model (review)
Security Market Line
M
E(Rm)
Market Risk Premium E(Rm) - rf
rf
102The Market Model (review)
- CAPM implies that security prices are governed by
their market risks and NOT their firm-specific
risks - How to decompose total risk of stock into two
components? A single-factor model Market model - Based on simple statistical regression framework
using T historical returns Rit ?i ?i
Rmt ?it
103Regression Approach Excel (review)
US Technology Index
US Utilities Index
104Regression Approach - Excel
US Technology Index
US Utilities Index
105Decomposing Risk (Review)
- If IBM stock has an annual standard deviation of
21, where as that of the SP 500 is only 16,
then we can decompose total risk of IBM given its
beta of 1.15 -
- 212 1.15 x 162 102
- 0.0441 0.0339 0.0102
- R-squared Proportion of variance due to market
0.0339/0.0441 76.9
Beta of IBM x Market Variability
Specific Risk IBM
Variability of IBM
106Macroeconomic Factor Models
- Early tests of the market model suggested
significant systematic, but extra-market factors - B. King, JB 1966, used statistical procedures to
identify strong extra-market, but within-industry
comovements - F. Black, M. Jensen and M. Scholes, 1972,
uncovered significant systematic patterns in
betas by market size - S. Ross in 1976 developed the Arbitrage Pricing
Theory (APT) as an alternative to the CAPM which
proposed a set of macroeconomic factors which
pervasively explain stock returns and which are
priced (matter for expected return relative to
risk)
107Arbitrage Pricing Theory (APT)
- Three major assumptions
- Capital markets are perfectly competitive
- Investors always prefer more wealth to less
wealth with certainty - The stochastic process generating asset returns
can be expressed as a linear function of a set of
K factors or indexes. - Assumptions Not Required for APT
- Quadratic utility function.
- Normally distributed security returns.
- Market portfolio that contains all risky assets
and is mean variance efficient.
108EXAMPLE CAPM VS APT
- CAPM Assumptions
- Investors want to invest on Efficient Frontier
(mean variance efficient) - Investors can lend and borrow at the risk free
rate - Investors have similar expectations
- Investors have the same one-period time horizon
- Investments are infinitely divisible ?Investors
can buy any amounts of any security - No taxes and no transaction costs
- No changes in interest rates (no inflation or it
is fully anticipated) - Capital markets are in equilibrium? investments
are properly priced for their risk levels
- APT Assumptions
- Capital Markets are perfectly competitive
- Investors prefer more wealth than less wealth
with certainty - Returns can be described linearly as a function
of K factors - It does not require
- Homogenous expectations and quadratic utility
function - Normality distributed returns
- A M portfolio that is mean-variance efficient and
contains all risky assets
109Arbitrage Pricing Theory (APT)
- All security returns are pervasively affected by
several macroeconomic factors
- where
- the expected return on an asset with zero
systematic risk where
the risk premium related to each of the
common factors - for example the risk premium
related to interest rate risk
bi the pricing relationship between the risk
premium and asset i - that is how responsive
asset i is to this common factor K?Different
stocks have different sensitivities (b
coefficients) to different factors, e.g. banks
and inflation, auto companies and national income
growth, utilities and interest rates,
etc.Multi-factor models can quantify these
multiple sources of risk
110Arbitrage Pricing Theory (APT)
- Multiple factors expected to have an
impact on all assets
- Multiple factors expected to have an impact on
all assets - Inflation
- Growth in GNP
- Major political upheavals
- Changes in interest rates
- And many more.
- Contrast with CAPM insistence that only beta is
relevant
111Example Two Stocks and a Two-Factor Model
- changes in the rate of inflation. The risk
premium related to this factor is 1 percent for
every 1 percent change in the rate
percent growth in real GNP. The average risk
premium related to this factor is 2 percent for
every 1 percent change in the rate
the rate of return on a zero-systematic-risk
asset (zero beta boj0) is 3 percent
112Example of Two Stocks and a Two-Factor Model
- the response of asset X to changes in the rate
of inflation is 0.50
the response of asset Y to changes in the rate
of inflation is 2.00
the response of asset X to changes in the
growth rate of real GNP is 1.50
the response of asset Y to changes in the
growth rate of real GNP is 1.75
113Example of Two Stocks and a Two-Factor Model
- .03 (.01)bi1 (.02)bi2
- Ex .03 (.01)(0.50) (.02)(1.50)
- .065 6.5
- Ey .03 (.01)(2.00) (.02)(1.75)
- .085 8.5
114Arbitrage Pricing Line
- No arbitrage implies
- E(Ri) rf ?i,F1 E(RF1) rf ?i,F2
E(RF2) rf . ?i,FK E(RFK) rf - How many factors?
- Which factors included?
- Is market portfolio a factor?
E(Ri)-rf
bF2
bF1
115Statistical Factor Models
- Early implementation of multi-factor models
inspired by theoretical development of Ross APT
as green light for purely statistical approach - Process Take variance-covariance matrix of
security returns and find extract factors that
maximally explain the variances - Principal Components Analysis Roll and Ross,
1980 - Factor Analysis - Chen, 1983 Connor and
Korajczyk, 1988 - Problem Economic interpretation of purely
statistical artifacts, e.g. F1, F2, F3, etc.
116Prespecified Macro Factors
- Economic intuition drove researchers to develop a
multi-factor macroeconomic model with
prespecified factors (no empirical justification) - Growth rate in monthly industrial production
- Default premium (Yield spread of Baa Aaa
corporates) - Term premium (Long-short bond yield spread)
- Expected and unexpected changes in CPI
- Major contributions Chen, Roll and Ross (1986),
Fama French (1988, 89), Ferson Harvey (1991) - Burmeister, Ibbotson, Roll Ross built the BIRR
model (www.birr.com) Salomon Smith Barney RAM
(Risk Attribute Model) model is competitor from
equity strategy group
117Comparing BIRR RAM
- RAM factors
- Economic Growth changes in industrial
production - Business Cycle default spread as proxy
- Long-term Interest Rates change in 10-year
Treasury note yield - Short-term Interest Rates changes in 1-month
Treasury bill - Inflation Shock unexpected changes
- US Dollar fluctuations in trade-weighted index
of currencies
- BIRR factors
- Confidence Risk tilted toward safe haven
stocks - Time Horizon Risk growth stocks have higher
positive exposure vs income stocks - Inflation Risk unexpected changes in inflation
- Market Timing Risk market risk, but as a
residual to other four - Business Cycle Risk unexpected changes in
growth of business activity
118Using Macroeconomic Factors
- Each stock has an estimated factor risk exposure
using multivariate regression - What is expected excess return for Reebok stock?
- 0.73(2.59) 0.77(-0.66) -0.48(-4.32)
4.59(1.49) 1.50(3.61) 15.71 - Compare projected return with different risk
exposures
BIRR Model for Reebok
Factor Market Risk Factor Sensitivity Price of
Risk Confidence 0.73 2.59 Time
Horizon 0.77 -0.66 Inflation Risk -0.48 -4.32 Bu
siness Cycle 4.59 1.49 Market Timing 1.50 3.61
119Fundamental Factor Models
- These models use company and industry attributes
and market data as factors, e.g. P/E, P/B ratios,
projected earnings growth, trading activity and
even technical factors, like price momentum - These are extra-market risk factors that
represent systematic comovements in historical
stock returns (e.g. Fama/French, 1992/93,
Carhart, 1997) - Commercially available fundamental factor models
are BARRA, Vestek Systems and Wilshires Atlas
120Alternative Risk Decomposition
Total Risk Variability
Specific Risk Diversifiable Risk
Market Risk Beta
Unique Risk Firm Specific Risk
Extramarket Risk Common Factor Risk
Macro Factor Risk GDP, Inflation, etc
Micro Factor Risk Size, P/E, P/B etc
Industry Risk
121Examples
- Fama-French (1993) Three-factor model
- E(Ri) rf ?i,m E(Rm) rf ?i,SMB ESMB
?i,HML EHML - SMB is return spread between small-cap and
large-cap triciles - HML is return spread between high P/B and low P/B
triciles - Carhart (1997)
- E(Ri) rf ?i,m E(Rm) rf ?i,SMB ESMB
?i,HML EHML ?i,Mom EMomentum - Momentum is return spread of stocks between
highest and lowest quintiles of returns
performance over (-1mo, -7mo)
122BARRA Model
- BARRA E2 model has 13 risk indices and 55
industry groups, but some models have 85 factors - Some of the factors are very complex in
construction, but detailed, thereby marketing
transparency - 1. Variability in markets (VIM) 2. Success
(momentum) - 3. Size 4. Trading activity (turnover)
- 5. Growth (in earnings) 6. Earnings/price
ratio - 7. Book/price ratio 8. Earnings variability
- 9. Financial leverage 10. Foreign income
- 11. Labor intensity 12. Yield
- 13. LOCAP (extension of size)
123An 87-Factor Model
Low Valuation (9 indices)
Small Firm Size (8 indices)
Earnings Variability (5 indices)
Growth Orientation (9 indices)
Factor Based Risk Estimates
Market Variability (9 indices)
Financial Risk (9 indices)
Industry Classification (39 indices)
124New Developments
- Chan, Karceski Lakonishok 1998 sought to
identify which factors best capture systematic
return covariation using a common dataset
(NYSE/Nasdaq, 1963-93, monthly stock returns) - Fundamental factors P/B, P/E, P/CF, D/P, Size
- Technical factors 1-month, 6-month, 60-month
momentum - Macroeconomic factors Inflation, default
premium, term premium, slope, industrial
production - Statistical factors Principal component
analysis up to 4 factors - Market factor equally- and value-weighted
indexes - Random factors
- CKL find that fundamental and technical factors
are stronger than macro variables and most
statistical factors
125Global and local risk factors
- Evidence shows international diversification
(Direct stock purchases, American depository
receipts, Mutual Funds--Open-end funds and
Closed-end funds, WEBS, Trusts) is beneficial - Possible to expand the efficient frontier above
domestic only frontier - Possible to reduce the systematic risk level
below the domestic only level - Then,
- What are the risks involved in investment in
foreign securities? - How do you measure benchmark returns on foreign
investments? - Are there benefits to diversification in foreign
securities?
126What do we learn?
- Evidence of priced factors beyond market beta
is becoming clearer, so it is more appropriate to
use a multi-factor model for risk measurement - Which factors are relevant? Buyer beware! Factor
models can be a license to data mine - Risk analysis should focus on understanding
active management bets in portfolio and risk
control of overall portfolio