Title: Bus 315 Ch'8 Factor Models and Arbitrage Pricing Theory
1Bus 315 Ch.8 Factor Models and Arbitrage
Pricing Theory
2CAPM downside
- To use CAPM we need to find the market
portfolio (tangency portfolio) on efficient
frontier - To utilize Markowitz portfolio theory, one needs
too many inputs (for calculating efficient
frontier). For example, with 50 securities one
would need 1225 estimates of covariances, 50
variances and expected returns). With of
stocks in TSX (1600), one would need to estimate
1.3 million parameters - Inputs may not be accurate and thus the
calculated portfolios may not be efficient.
Moreover, optimization may lead to nonsensical
results because of poor estimates(e.g. negative
variance)
3Returns and systematic risk
- Covariances between returns are caused by some
common macroeconomic factors (interest rate, GDP
change etc.) - Unexpected changes in these factors lead to
unexpected changes in returns - We can group all relevant factors in one economic
indicator (for example, GDP growth) - Therefore, we can specify the return on any asset
as - E(R) is the expected return on the asset at the
beginning of the period, m is the impact of
unanticipated macroeconomic events, e is the
impact of firm specific events
4Single Factor Model
- Different stocks have different sensitivities to
macroeconomic factors, Fs. These sensitivities
will be denoted by factor betas. - F denotes the unanticipated component in factor
(e.g. unexpected inflation or unanticipated GDP
growth) - We can assume that the return on a broad market
index (TSX) can be a good factor. - If we use a stock index (TSX) as a factor then
this model is called a single-index model
5Single Index Model
- With single index model the (excess) return on
any stock can be decomposed in three parts - If we rewrite the equation in terms of excess
returns, it will look like - Return on an index may be viewed as a sum of
predicted part (risk-free rate) and unanticipated
part (market premium)
6Estimating inputs with factor model
- Now, in case of 50 securities we need to
estimate only 50 expected returns, 50 sensitivity
betas, 50 firm-specific variances, and market
variance
7Estimating inputs Example
- Suppose that the index model for stocks A and B
is estimated with the following results - Calculate the standard deviation of each stock
and the covariance between them
8The Single Index Model
- Advantages
- Reduces the number of inputs
- Easier for security analysts to specialize to
calculate covariance one simply needs to know the
sensitivity of the stock to a factor. No need for
inter-industry knowledge. - Drawback
- The simple separation of factors (macro versus
micro risk) rules out important risk sources
(such as industry events) - Estimation
- Run regression of historical excess returns of a
security on the market (e.g. SP/TSX composite)
excess returns - The regression equation is called Security
Characteristic Line (SCL)
9Security Characteristic Line
10Interpreting the regression results
11Index Model and Diversification
- If we create a portfolio of n equally weighted
stocks (w1/n), then the non-systematic part of
the portfolio will fade away - To see it, recall that
- The portfolio variance is
- As n increases, the portfolio variance approaches
the systematic variance
12Risk Reduction with Diversification
13Tracking Portfolios
- Suppose you find a portfolio that you consider to
be mispriced - You can replicate the sensitivity to the factor
by creating a tracking portfolio (140 in TSX
Index and -40 (borrowing) at risk-free rate) - Tracking portfolio matches the systematic risk
(has the same beta) - If you take long position in portfolio P and
short the tracking portfolio you can earn alpha
of 4 irrespective of the market movements.
Still, the non-systematic risk would be present.
14Multifactor Models
- The single index model uses just one factor as
explanation for systematic risk - However, there might be other factors too
- Multifactor models use other factors in addition
to market return - Estimate a beta or factor loading for each factor
using multiple regression - The way we find what factors to include is purely
empirical, there is no theory that say which ones
to include - Some factor ideas change in expected inflation,
change in unanticipated inflation, change in
industrial production (GDP), credit spread, yield
spread.
15Multifactor Model Equation
- For example, the multifactor (2-factor) model may
look like this - Ri E(ri) ßGDP (GDP) ß IR (IR) ei
- Ri Return for security i
- ß GDP Factor sensitivity for GDP
- ß IR Factor sensitivity for Interest Rate
- ei Firm specific events
16Arbitrage Pricing Theory (APT)
- Arbitrage - arises if an investor can construct a
zero investment portfolio with a sure profit - Law of one price says that if two assets are the
same in terms of economic aspects, they should
cost the same - Since no investment is required, an investor can
create large (infinite) positions to secure large
levels of profit - In efficient markets, profitable arbitrage
opportunities will quickly disappear since
arbitrage is a stronger argument than dominance
of one security over another - To get to the APT one needs the following
assumptions - Security returns can be described by a factor
model - There are enough securities to diversify away
firm-specific risk - Arbitrage opportunities are not allowed
17APT and Well-Diversified Portfolios
- For well-diversified portfolios (ones with
negligible non-systematic risk), this should be
true - If the portfolio are well diversified, and have
the same exposure to the factor risk, they should
have the same return. Otherwise, arbitrage is
possible. - This means that the risk premiums, E(r)-rf,
should be proportional to the factor betas
18Disequilibrium Example
19APT and CAPM Compared
- APT applies to well diversified portfolios and
not necessarily to individual stocks - With APT it is possible for some individual
stocks to be mispriced - not lie on the SML - APT is more general in that it gets to an
expected return and beta relationship without the
assumption of the market portfolio - APT can be extended to multifactor models
20A Multifactor APT
- A factor portfolio is a portfolio constructed so
that it would have a beta equal to one on a given
factor and zero on any other factor - These factor portfolios are the benchmark
portfolios for a multifactor security market line
for an economy with multiple sources of risk
21Where Should we Look for Factors?
- The multifactor APT gives no guidance on where to
look for factors - Chen, Roll and Ross
- Returns a function of several macroeconomic and
bond market variables instead of market returns - Fama and French
- Returns a function of size and book-to-market
value as well as market returns
22Bus 315 Ch.9 Market efficiency
23Efficient Market Hypothesis (EMH)
- Can prices be predicted?
- Do security prices reflect information ?
- Any information that can be used to predict stock
performance must already be reflected in stock
prices - The notion that stocks prices fully reflect all
available information is called Efficient Market
Hypothesis (EMH)
24Illustration of Efficient Market Hypothesis
- Suppose there is rule for predicting stock
prices. The forecast of favourable future prices
would increase current price as everybody would
tart buying this stock - Example suppose that current stock price is
100, the stock is expected to pay dividend of 3
next year and have price 107, and according to
CAPM, this stock return should be 10. If you had
a model that could predict that future price to
be 110, what would happen to todays price?
25Market Efficiency and Competition
- As long as analysis of new information generates
higher return, investors will spend time and
money to research and uncover new information - Competition among analysts will make it hard to
predict future prices - Stock prices fully and accurately reflect
publicly available information - Once information becomes available, market
participants analyze it - Competition assures prices reflect information
- Why look at market efficiency
- Implications for business and corporate finance
- Implications for investment
26Random Walk and the EMH
- Random Walk - stock price change is
unpredictable E(Pt1)E(Pt)et - Actually stock prices follow a sub-martingale
- Expected price is positive over time
- Positive trend and random around the trend
- Why are price changes random?
- Prices react to information
- Flow of information is random
- Therefore, price changes are random
27Random Walk with Positive Trend
28Forms of the EMH
- Definition of EMH poses two questions
- What is all available information?
- What fully reflect all available information
means - Depending on the answers to these questions we
will have different versions of market
efficiency. - Forms of efficiency hypothesis differ by the
definition of all available information - Weak Semi-strong Strong
- The phrase prices fully reflect all available
information means that at market prices
investments in assets have NPV0.
29Forms of the EMH (2)
- Weak all past market trading information
- Past prices, trading volume, interest rates
- Testing patterns of stock returns (correlation
etc.) - Weak form of EMH is supported by the data
- Semi-strong all publicly available information
regarding the prospects of a firm - Firm fundamental data, financial statements
- Testing event studies
- Semi-strong form of EMH is supported by the data
- Strong all information relevant to the firm
- All info, including insider info
- Testing
- Strong-form of EMH has mixed evidence
30Types of Stock Analysis
- Technical Analysis - using prices and volume
information to predict future prices (Ch.16) - Weak form efficiency technical analysis
- Technical trading rules are not consistently
profitable (SP 500 index behaviour is similar
to coin flipping) - Fundamental Analysis - using economic and
accounting information to predict stock prices
(Ch.14-15) - Semi strong form efficiency fundamental
analysis
31Implications of EMH
- Trust market prices
- Buying and selling assets are zero NPV
activities, thus only risk implied return is
provided - Market prices give best estimate of securities
values - Firms receive fair value for their securities
when issued - Information can be extracted from knowing prices
- There are no financial illusions
- Market price reflects value only from an assets
payoff - It is not easy to trick the market
32Market Efficiency and Portfolio Management
- Active Management
- Security analysis Timing -gt Not justified
according to EMH - Passive Management
- Buy and Hold Index Funds
- Even if the market is efficient a role exists for
portfolio management - Diversification (even if markets are efficient,
diversification is still required) - Appropriate risk level (different investors will
require different levels of risk depending on
investors age - Tax considerations (some investors prefer capital
gains to income because of their tax rate)
33Tests of Market Efficiency
- 1) Examine prices and returns over time
34Tests of Market Efficiency (2)
- 2) Returns are adjusted to determine if they are
abnormal - Market Model approach
- a. Rt a bRmt et
- (Expected Return)
- b. Excess Return (Actual - Expected)
- et Actual - (at btRmt)
35Tests of Market Efficiency (3)
- 3) Market Model approach
- Cumulate the excess returns over time
36Issues in Examining the Results
- Magnitude Issue
- Difficult to measure abnormal return of say,
0.01, because of market volatility - Selection Bias Issue
- Only schemes that couldnt bring abnormal returns
are reported - Lucky Event Issue
- Sometimes its just coincidence
- Nice example create 8 newspapers to report
market direction for 3 years. One paper will
perfectly predict it by construction! - Model for adjusting stock returns may be
misspecified
37What Does the Evidence Show?
- Returns over short horizons
- Very short time horizons small magnitude of
positive trends - 3-12 month some evidence of positive momentum
- Filter rules
- Returns over long horizons pronounced negative
correlation - Evidence on Reversals
- Winners become losers and vice versa
- Anomalies Exist
38Anomalies (tests of semistrong EMH)
- Small Firm Effect (January Effect)
- Returns for small firms are abnormally hogh in
January - Neglected Firm
- Returns on firms with less research are higher
- Book to Market Ratios
- High book-to-market stocks have higher returns
- Post-Earnings Announcement Drift
- Abnormal returns even after announcement
39Explanations of Anomalies
- Risk Premiums or market inefficiencies ?
- Anomalies or data mining ?
- Behavioural explanations
- These are assertions that investors exhibit
systematic departures from rationality when
making complicated decisions - Such departures from rationality cannot be
exploited because there are limits to arbitrage
40Behavioural Explanations
- Information processing
- Forecasting errors (overweight recent events)
- Overconfidence (90 of drivers in Sweden report
to be better-than-average) - Conservatism (update beliefs too slowly)
- Sample size neglect (sample size is not taken
into account) - Behavioural biases
- Framing (decision will depend on how choices are
framed) - Mental accounting (different mental accounts
and attitudes toward them) - Regret avoidance (regret more when decision is
unconventional)
41Mutual Fund and Professional Managers Performance
- Some evidence of persistent positive and negative
performance - Potential measurement error for benchmark returns
- Style changes
- Superstar phenomenon
- Some managers made fortunes with superior
performance