CFAspace

Provided by APF
Academy of Professional Finance 专业金融学院

CFA Level II
Portfolio Management:
An Introduction to Multifactor Models
Lecturer: Nan Chen
 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Reading: Analysis of Active Portfolio Management Yes

Reading: Economics and Investment Markets Yes

Reading: The Portfolio Management Process No
Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
Joshua White Scenario
(This case was adapted from the CFA mock exams)
Joshua White is an equity portfolio manager for Eastwood Investments, who currently uses the
Capital Asset Pricing Model (CAPM) to evaluate securities and mean-variance portfolio optimization
to construct equity portfolios. White is meeting with two assistant portfolio managers, Stephen
Butler and Deb Miller . Butler and Miller have been asked to do some research on ways to improve
on the methods currently being used by Eastwood to evaluate securities and develop portfolios.

White begins the meeting by outlining some issues relating to the CAPM. He makes the following
statements:

Statement 1
“One of the reasons I am uncomfortable using the CAPM is that it makes some very restrictive
assumptions such as :
*investors pay no taxes on returns and no transaction costs on trades,
*investors have unique views on expected returns, variances and correlations of securities, and
*investors can borrow and lend at the same risk-free rate of interest.”

Statement 2
“We are also faced with a problem that our mean-variance optimization models can generate
unstable minimum-variance efficient frontiers. Consequently, we face considerable uncertainty
regarding recommendations we make to our clients on asset allocation. I attribute the instability to
our use of:
A short sales constraint, and
Historical betas.”

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
Butler suggests that multifactor models provide a better way to model stock returns. He states that
“there are two ways to model stock returns using the following multifactor model:
Ri=ai+bi1F1+bi2F2+…+bikFk+ ɛi

Model1: In this model, stock returns (Ri) are determined by surprises in economic factors such as GDP
growth and the level of interest rates.

Model2: Here, stock returns (Ri) are determined by factors that are company attributes such as PE
ratios and market capitalization.

While the interpretation of the intercept ai is similar for both models, the factor sensitivities bi are
interpreted differently in the two models.”

Miller notes that a multifactor Arbitrage Pricing Model (APT) provides a much better basis than the
CAPM for calculating expected portfolio returns and evaluating portfolio risk exposures. In order to
illustrate the advantages of the multifactor APT model, Miller provides information for two portfolios
Eastwood currently manages. The information is provided below in the exhibit. The current risk-free
rate is 2%.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
Miller makes the following statement:

Statement 3
“We can tell from the exhibit that Portfolio A is structured in such a manner that it will benefit from an
expanding economy and improving confidence because the factor sensitivities for confidence risk and
business cycle risk exceed the factor sensitivities for the benchmark. Portfolio B has very low factor
sensitivities for confidence risk and business cycle risk but moderately high exposure to inflation risk,
therefore Portfolio B can be referred to as a factor portfolio for inflation risk.”

White wants to examine how active management is contributing to portfolio performance. Miller
responds with the following statement:

Statement 4
“Our models show that Portfolio A has annual tracking error of 1.25% and an information ratio of 1.2
while Portfolio B has an annual tracking error of 0.75% and an information ratio of 0.87.”

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
1. Which assumption of the CAPM is most likely incorrect in White’s statement 1? The assumption regarding:
A. Borrowing and lending
B. Taxes and transaction costs
C. Expected returns, variances and correlations

2. Is White’s statement 2 most likely correct?

A. Yes.
B. No, she is incorrect about the short sales constraint.
C. No, she is incorrect about the use of historical betas.

3. With regard to the statement on multifactor models, Butler is most likely incorrect with respect to the:
A. Intercept value ai.
B. Factor sensitivities bi.
C. Description of the factors.

4. Based on the information in the exhibit, the expected return for portfolio A is closest to:
A. 8.4%
B. 10.2%
C. 12.2%

5. Is Miller’s Statement 3 most likely correct?

A. Yes.
B. No, she is incorrect about Portfolio A.
C. No, she is incorrect about Portfolio B.

6. Based on Statement 4 by Miller, an appropriate conclusion is that the portfolio that has benefited the
most from active management is:
A. Portfolio B because of tracking error.
B. Portfolio A because of the information ratio.
C. Portfolio B because of the information ratio.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Arbitrage Pricing Theory (APT)

 Three Assumptions of Arbitrage Pricing Theory (APT):

 Unsystematic risk can be completely diversified away.
 Returns are derived from a multifactor model.
 No arbitrage opportunities exist.
• An arbitrage opportunity is defined as an investment opportunity that bears no
risk, no cost, and yet provides a profit.
 Arbitrage Pricing Model:
• Equilibrium relationship between expected returns for well diversified portfolios
and their multiple sources of systematic risk.

Expected return Risk premium Sensitivity of

on Portfolio P associated with Portfolio P to
each risk factor each risk factor

 Compare CAPM and APT Model:

 The CAPM can be considered a special restrictive case of the APT
in which there is only one risk factor (market risk factor).
 The APT does NOT require that one of the risk factors is the market risk factor,
which is a major advantage of the arbitrage pricing model.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Arbitrage Pricing Theory (APT)

 Calculate expected returns from the Arbitrage Pricing Model:

EX: A two-factor APT model was adopted by Eastwood Investment Firm. Calculate the
expected return for one of the firm’s portfolios using the following data:

The risk-free rate equals 5%.

Factor 1 Factor 2

Factor Risk Premiums 1.5 2.0

Factor Sensitivities 0.0300 0.0125

the expected return on the Portfolio:

E(Rp) = 0.05 +1.5×0.0300+2.0×0.0125=12%

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Arbitrage Pricing Theory (APT)

 How to exploit an arbitrage opportunity:

EX: Determine whether an arbitrage opportunity exists from the data below:
Assume Eastwood Investment Firm uses a single factor model to evaluate assets.
Information related to portfolios A, B, and C are provided:

Portfolio Expected Return Beta

A 10% 1.0
B 20% 2.0
C 13% 1.5

By allocating 50/50 between portfolios A and B, we can obtain a portfolio D with
the same beta as that of portfolio C, because 0.5(1)+0.5(2) = 1.5
=>Now portfolio D and portfolio C has the same beta, i.e. the same risk.
E(RD)=0.5(0.1)+0.5(0.2) = 15%
=> Despite the same risk, portfolio D has higher expected return than portfolio C;
portfolio D is undervalued.
By purchasing portfolio D and short-selling portfolio C , an arbitrage profit could be
exploited.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Statistical Factor Models:

 Use statistical methods to explain asset returns.
 Major weakness: the statistical factors do not lend themselves
well to economic interpretation.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Macroeconomic Factor Models

 A two-factor macroeconomic model:

 Describe the two-factor macroeconomic model:

Systematic Risk Factors FGDP: surprises in the GDP rate “F’s” ( factor surprises):
(Priced) the differences between the predicted value of
FQS: surprise in the credit quality spread the factor and the realized value.
Unsystematic Risk : ℇi : the part of the return that CANNOT be
can be diversified away explained by the model.
(NOT priced)
Factor Sensitivities bi1 for retail stocks: larger “b’s”: sensitivities of the stock to the surprises;

bi1 for food stocks : smaller

Expected return E(Ri): intercept

 We should be able to compute a stock return using a macroeconomic

factor model.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Fundamental Factor Models
 An example of fundamental factor model:

 Describe the fundamental factor model:

FP/E: rate of return associated with the PE factor

FSIZE: rate of return associated with the size factor

Factor sensitivities:
For macroeconomic factor models: regression slopes;
If bi1=2, Stock i has a P/E that is 2 standard deviations above For fundamental factor models: standardized attributes.
the mean.

Intercept ai: NOT interpreted as expected return

EX: The P/E ratio for Stock HT is 15.20, the average P/E ratio for all stocks is 11.90, and
the standard deviation of P/E ratios is 6.30. Calculate the standardized sensitivity of
Stock HT to the P/E factor.

Standardized Sensitivity of Stock HT to the P/E factor

=(15.20-11.90)/6.30
=0.52
Interpretation: the P/E ratio for the stock is 0.52 standard deviation above the
average stock P/E.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Fundamental Factor Models

 Compare
Macroeconomic Factor Model Fundamental Factor Model

Sensitivities Slope estimates from regression Calculated from the attribute data (ex:
P/E); not estimated
Interpretation of Factors are surprises in the Factors are rates of return associated
factors macroeconomic variables with each factor
Number of factors Small in number Large in number

Intercept term Equals the stock’s expected return Does not equal the stock’s expected
return; Has no economic interpretation.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Application of Multifactor Models

 Active Return = RP –RB

 Active Risk / Tracking Error / Tracking Risk = s (RP –RB)

 Information Ratio = demonstrates a manager’s consistency in generating active return.

Q1 Q2 Q3 Q4 Average Active Risk Information

Active Return Ratio
Manager A 0.6% 0.5% 0.5% 0.4% 0.5% 0.000816 6.12
Manager B 8% 5% -3% -8% 0.5% 0.073258 0.07

 Information Ratio vs. Sharpe Ratio

RP  RB RP  RF
IR  VS. SR 
s( RP  RB ) P

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Application of Multifactor Models – Return Attribution

 Active Return = RP –RB

 Active Return= Factor Return + Security Selection Return

 Factor Return: arising from the manager’s decision to take on factor
exposures that differ from those of the benchmark.

 Security Selection Return: arising from the manager choosing a

different weight for specific securities compared to the weight of those
securities in the benchmark.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Application of Multifactor Models – Risk Attribution

 Active Risk / Tracking Error / Tracking Risk = s (RP –RB)

 Active Risk Squared = Active Factor Risk + Active Specific Risk

 Active Factor Risk: risk from active factor tilts attributable to under or
overweighting particular industries relative to the portfolio’s benchmark.

 Active Specific Risk: risk from active asset selection attributable to

deviations of the portfolio’s individual asset weightings versus the
benchmark’s individual asset weightings.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Application of Multifactor Models – Risk Attribution

EX: A fund analyst is analyzing the performance of three actively managed mutual funds
using a two-factor model. The results of risk decomposition are shown below:
Fund Active Factor Active Specific Active Risk Squared
Size Factor Style Factor Total Factor
A 6.25 12.22 18.47 3.22 21.69
B 3.20 0.80 4.00 12.22 16.22
C 17.85 0.11 17.96 19.7 37.66
Which fund assumes the highest level of active risk, the highest level of style factor risk
as a proportion of active risk, the highest level of size factor risk as a proportion of active
risk, and the lowest level of active specific risk as a proportion of active risk?
Proportional contributions of various sources of active risk as a proportion of active
risk squared:
Fund Active Factor Active Active Risk
Specific
Size Factor Style Factor Total Factor
A 29% 56% 85% 15% 4.7%
B 20% 5% 25% 75% 4.0%
C 47% 0% 48% 52% 6.1%

Fund C assumes the highest level of active risk; Fund A assumes the highest level
of style factor risk; Fund C assumes the highest level of size factor risk; Fund A
assumes the lowest level of active specific risk.
Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com
 Application of Multifactor Models – Portfolio Construction

 Passive Management:
Managers seeking to track a benchmark can construct a tracking portfolio.
 Tracking Portfolio is a portfolio having factor sensitivities that are matched to
those of a benchmark or other portfolio.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Application of Multifactor Models – Portfolio Construction
 Active Management
 Factor Portfolio for a particular factor has a sensitivity of 1 for that factor and a
sensitivity of 0 for all other factors.
 Factor portfolios can be used to hedge that factor risk (offset it) or speculate on it.
Risk Factor Factor  This portfolio is a factor portfolio for business cycle risk.
Sensitivities
 Speculate: The portfolio manager would take a long
Confidence Risk 0
position in the factor portfolio to place a bet on an
Time Horizon Risk 0 increase in real business activity.
Inflation Risk 0  Hedge: The portfolio manage would take a short
Business Cycle Risk 1 position in the factor portfolio to hedge an existing
positive exposure to business cycle risk.
Market Timing Risk 0

EX: The table below provides the factor exposures of three portfolios based on Carhart Model:
Portfolio Risk Factor
RMRF SMB HML WML Which strategy would be most appropriate if the
manager expects that:
A 1.00 0.00 0.00 0.00
B 0.00 1.00 0.00 0.00 A. RMRF will be higher than expected.

C 1.20 0.00 0.20 0.80 B. Large cap stocks will outperform small cap stocks.

A. go long on Portfolio A, which is constructed to be a pure bet on RMRF factor.

B. go short on Portfolio B, which is constructed to be a pure bet on SMB factor.
Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com
 Application of Multifactor Models – Portfolio Construction

 Rules-based active management (alternative indices):

 These strategies routinely tilt toward factors such as size, value, quality, or
momentum when constructing portfolios.
 These strategies introduce biases in the portfolio relative to market capitalization-
weighted indexes.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Framework
Reading Changes

Reading: An Introduction to Multifactor Models Yes

Arbitrage Pricing Theory (APT)

Multifactor Models

Macroeconomic Factor Models

Fundamental Factor Models

Application of Multifactor Models

Benefits in Considering Multiple Risk Factors

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Benefits in Considering Multiple Risk Factors
 By including more risk factors, multifactor models enable
investors to
 zero in on risks that the investor has a comparative advantage in
bearing and
 Avoid the risks that the investor is incapable of absorbing.

 University endowments
 Comparative advantage – business cycle risk and liquidity risk
 Comparative disadvantage – inflation risk

 CAPM Framework vs. Multifactor Model:

 Under the CAPM framework, investors choose a combination of the
market portfolio and the risk-free asset.
 A multifactor approach offer a richer combination of SMB, HML, and
WML factors in addition to the market factor, and thus can help
investors achieve better-diversified and possibly more efficient
portfolios.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
Joshua White Scenario
(This case was adapted from the CFA mock exams)
Joshua White is an equity portfolio manager for Eastwood Investments, who currently uses the
Capital Asset Pricing Model (CAPM) to evaluate securities and mean-variance portfolio optimization
to construct equity portfolios. White is meeting to two assistant portfolio managers, Stephen Butler
and Deb Miller . Butler and Miller have been asked to do some research on ways to improve on the
methods currently being used by Eastwood to evaluate securities and develop portfolios.

White begins the meeting by outlining some issues relating to the CAPM. He makes the following
statements:

Statement 1
“One of the reasons I am uncomfortable using the CAPM is that it makes some very restrictive
assumptions such as :
*investors pay no taxes on returns and no transaction costs on trades,
*investors have unique views on expected returns, variances and correlations of securities, and
*investors can borrow and lend at the same risk-free rate of interest.”

Statement 2
“We are also faced with a problem that our mean-variance optimization models can generate
unstable minimum-variance efficient frontiers. Consequently, we face considerable uncertainty
regarding recommendations we make to our clients on asset allocation. I attribute the instability to
our use of:
• A short sales constraint, and
• Historical betas.”

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
Butler suggests that multifactor models provide a better way to model stock returns. He states that
“there are two ways to model stock returns using the following multifactor model:
Ri=ai+bi1F1+bi2F2+…+bikFk+ ɛi

Model1: In this model, stock returns (Ri) are determined by surprises in economic factors such as GDP
growth and the level of interest rates.

Model2: Here, stock returns (Ri) are determined by factors that are company attributes such as PE
ratios and market capitalization.

While the interpretation of the intercept ai is similar for both models, the factor sensitivities bi are
interpreted differently in the two models.”

Miller notes that a multifactor Arbitrage Pricing Model (APT) provides a much better basis than the
CAPM for calculating expected portfolio returns and evaluating portfolio risk exposures. In order to
illustrate the advantages of the multifactor APT model, Miller provides information for two portfolios
Eastwood currently manages. The information is provided below in the exhibit. The current risk-free
rate is 2%. =2%+0.81*4.5%-0.15*(-1.2%)+1.23*5.2%=12.2%

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
Miller makes the following statement:

Statement 3
“We can tell from the exhibit that Portfolio A is structured in such a manner that it will benefit from an
expanding economy and improving confidence because the factor sensitivities for confidence risk and
business cycle risk exceed the factor sensitivities for the benchmark. Portfolio B has very low factor
sensitivities for confidence risk and business cycle risk but moderately high exposure to inflation risk,
therefore Portfolio B can be referred to as a factor portfolio for inflation risk.”

White wants to examine how active management is contributing to portfolio performance. Miller
responds with the following statement:

Statement 4
“Our models show that Portfolio A has annual tracking error of 1.25% and an information ratio of 1.2
while Portfolio B has an annual tracking error of 0.75% and an information ratio of 0.87.”

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

 Item Set Example
1. Which assumption of the CAPM is most likely incorrect in White’s statement 1? The assumption regarding:
A. Borrowing and lending
B. Taxes and transaction costs
C. Expected returns, variances and correlations

2. Is White’s statement 2 most likely correct?

A. Yes.
B. No, she is incorrect about the short sales constraint.
C. No, she is incorrect about the use of historical betas.

3. With regard to the statement on multifactor models, Butler is most likely incorrect with respect to the: Answer:A
A. Intercept value ai.
B. Factor sensitivities bi.
C. Description of the factors.

4. Based on the information in the exhibit, the expected return for portfolio A is closest to: Answer:C
A. 8.4%
B. 10.2%
C. 12.2%

5. Is Miller’s Statement 3 most likely correct? Answer:C

A. Yes.
B. No, she is incorrect about Portfolio A.
C. No, she is incorrect about Portfolio B.

6. Based on Statement 4 by Miller, an appropriate conclusion is that the portfolio that has benefited the Answer:B
most from active management is:
A. Portfolio B because of tracking error.
B. Portfolio A because of the information ratio.
C. Portfolio B because of the information ratio.

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com

Academy of Professional Finance 专业金融学院 Copyright © CFAspace.com