Lecture 12: Factor Pricing

Prof. Markus K. Brunnermeier

Factor Pricing Slide 12-1

 Overview
1. ICAPM – multiperiod economic model (last lecture)
2. Asset Pricing Theory (APT) – static statistical model
 Merits of Factor Pricing
 Exact Factor Pricing and Factor Pricing Errors
 Factor Structure and Pricing Error Bounds
 Single Factor and Beta Pricing (and CAPM)
 (Factor) Mimicking Portfolios
 Unobserved Factor Models
 Multi-period outlook
3. Empirical Factor Pricing Models
 Arbitrage Pricing Theory (APT) Factors
 The Fama-French Factor Model + Momentum
Factor Pricing Slide 12-2
 The Merits of Factor Models
• Without any structure one has to estimate
 J expected returns E[Rj] (for each asset j)
 J standard deviations
 J(J-1)/2 co-variances
• Assume that the correlation between any two assets is
explained by systematic components/factors, one can
restrict attention to only K (non-diversifiable) factors
 Advantages: Drastically reduces number of input variables
Models expected returns (priced risk)
Allows to estimate systematic risk
(even if it is not priced, i.e. uncorrelated with SDF)
Analysts can specialize along factors
 Drawbacks: Purely statistical model (no theory)
(does not explain why factor deserves compensation: risk vs. mispricing )
relies on past data and assumes stationarity
Factor Pricing Slide 12-3
 Factor Pricing Setup …
• K factors f1, f2, …, fK
 E[fk]=0
 K is small relative to dimension of M
 fk are not necessarily in M
• F space spanned by f1,…,fK,e
• in payoffs

 bj,k factor loading of payoff xj

Factor Pricing Slide 12-4
 …Factor Pricing Setup
• in returns

• Remarks:
One can always choose orthogonal factors Cov[fk, fk‟]=0
Factors can be observable or unobservable

Factor Pricing Slide 12-5

 Factor Structure
• Definition of “factor structure:”

• ) risk can be split in systematic risk and

idiosyncratic (diversifiable) risk

Factor Pricing Slide 12-6

Exact vs. Approximate Factor Pricing
• Multiplying (1) by kq and taking expectations

• Rearranging

• Exact factor pricing:

 error: j = 0 (i.e. j s orthogonal to kq )
 e.g. if kq 2 F
Factor Pricing Slide 12-7
 Bound on Factor Pricing Error…
• Recall error
 Note, if 9 risk-free asset and all fk 2 M, then …
• If kq 2 F, then factor pricing is exact
• If kq F, then
 Let‟s make use of the Cauchy-Schwarz inequality
(which holds for any two random variables z1 and z )

 Error-bound

Factor Pricing Slide 12-8

Error-Bound if Factor Structure Holds
• Factor structure ) split idiosyncratic from systematic risk
• ) all idiosyncratic risk j are linearly independent and
span space orthogonal to F. Hence,
• Note
• Error

• Pythagorean Thm: If {z1, …, zn} is orthogonal system in

Hilbert space, then
 Follows from def. of inner product and orthogonality
Factor Pricing Slide 12-9
Error-Bound if Factor Structure Holds
Applying Pythagorean Thm to
implies

Multiply by …… and making

use of

RHS is constant for constant max[ 2( j)].

) For large J, most securities must have small pricing error
• Intuition for Approximate Factor Pricing:
Idiosyncratic risk can be diversified away
Factor Pricing Slide 12-10
 One Factor Beta Model…
• Let r be a risky frontier return and set
f = r – E[r] (i.e. f has zero mean)
 q(f) = q(r) – q(E[r])
_
• Risk free asset exists with gross return of r
_
 q(f)_= 1 – E[r]/r
• f and r span E and hence kq 2 F
) Exact Factor Pricing
Factor Pricing Slide 12-11
 …One Factor Beta Model
• Recall
_ _
 E[rj] = _r - j r q(f) _
 E[rj] = r - j {E[r] - r}

• Recall

 j = Cov[rj, f] / Var[f] = Cov[rj, r] / Var[r]

• If rm 2 E then CAPM
Factor Pricing Slide 12-12
 Mimicking Portfolios…
• Regress on factor directly or on portfolio that mimics factor
 Theoretical justification: project factor on M
 Advantage: portfolios have smaller measurement error
• Suppose portfolio contains shares 1, …, J with j
J
j = 1.
• Sensitivity of portfolio w.r.t. to factor fk is k = j j jk
• Idiosyncratic risk of portfolio is = j j
 2 ( ) = j 2 ( j)
 diversification

Factor Pricing Slide 12-13

 …Mimicking Portfolios
• Portfolio is only sensitive to factor k0 (and
idiosyncratic risks) if for each k k0 k= j
jk=0, and k0= j jk0 0.
• The dimension of the space of portfolios
sensitive to a particular factor is J-(K-1).
• A portfolio mimics factor k0 if it is the portfolio
with smallest idiosyncratic risk among portfolios
that are sensitive only to k0.
Factor Pricing Slide 12-14
Observable vs. Unobservable Factors…
• Observable factors: GDP, inflation etc.
• Unobservable factors:
 Let data determine “abstract” factors
 Mimic these factors with “mimicking portfolios”
 Can always choose factors such that
• factors are orthogonal, Cov[fk, fk‟]=0 for all k k‟
• Factors satisfy “factor structure” (systemic & idiosyncratic risk)
• Normalize variance of each factor to ONE
) pins down factor sensitivity (but not sign, - one can always change sign of factor)

Factor Pricing Slide 12-15

 …Unobservable Factors…
• Empirical content of factors
Cov[ri,rj] = k ik jk 2(fk)
 2(rj) = k jk jk 2(fk)+ 2( j)
 (fk)=1 for k=1,L,K. (normalization)
In matrix notation
• Cov[r,r„] = k k‟ k
2(f )
k + D,
– where k =( 1k,…, Jk).
• = B B‟ + D,
– where Bjk= jk, and D diagonal.
– For PRINCIPAL COMPONENT ANALYSIS assume D=0
(if D contains the same value along the diagonal it does affect
eigenvalues but not eigenvectors – which we are after)
Factor Pricing Slide 12-16
 …Unobservable Factors…
• For any symmetric JxJ matrix A (like BB‟), which is semi-
positive definite, i.e. y‟Ay ¸ 0, there exist numbers 1 ¸ 2
¸…¸ lambdaJ ¸ 0 and non-zero vectors y1, …, yJ such that
 yj is an eigenvector of A assoc. w/ eigenvalue j, that is A yj = j yj
 j
J yij yij‟ = 0 for j j‟
 j
J yij yij = 1
 rank (A) = number of non-zero „s
 The yj „s are unique (except for sign) if the i „s are distinct
• Let Y be the matrix with columns (y1,…,yJ), and
let the diagonal matrix with entries i then

Factor Pricing Slide 12-17

 …Unobservable Factors
• If K-factor model is true, BB' is a symmetric positive
semi-definite matrix of rank $K.$
 Exactly K non-zero eigenvalues 1,…, k and associated
eigenvectors y1,…,yK
 YK the matrix with columns given by y1,…,yK K the diagonal
matrix with entries j, j=1,…, K.
 BB'= K
Hence,

• Factors are not identified but sensitivities are (except for sign.)
• In practice choose K so that k is small for k>K.
Factor Pricing Slide 12-18
 Why more than ONE mimicking
portfolio?
• Mimic (un)observable factors with portfolios
[Projection of factor on asset span]

• Isn‟t a single portfolio which mimics pricing kernel

sufficient ) ONE factor
• So why multiple factors?
 Not all assets are included (real estate, human capital …)
 Other factors capture dynamic effects
[since e.g. conditional unconditional. CAPM]
(more later to this topic)

Factor Pricing Slide 12-19

 Overview
1. ICAPM – multiperiod economic model
2. Asset Pricing Theory (APT) – statistical model
 Merits of Factor Pricing
 Exact Factor Pricing and Factor Pricing Errors
 Factor Structure and Pricing Error Bounds
 Single Factor and Beta Pricing (and CAPM)
 (Factor) Mimicking Portfolios
 Unobserved Factor Models
 Multi-period outlook
3. Empirical Factor Pricing Models
 Arbitrage Pricing Theory (APT) Factors
 The Fama-French Factor Model + Momentum
Factor Pricing Slide 12-20
APT Factors of Chen, Roll and Ross (1986)

1. Industrial production
(reflects changes in cash flow expectations)
2. Yield spread btw high risk and low risk corporate bonds
(reflects changes in risk preferences)
3. Difference between short- and long-term interest rate
(reflects shifts in time preferences)
4. Unanticipated inflation
5. Expected inflation (less important)
Note: The factors replicate market portfolio.

Factor Pricing Slide 12-21

 Fama-MacBeth 2 Stage Method
• Stage 1: Use time series data to obtain estimates for
each individual stock‟s j

(e.g. use monthly data for last 5 years)

Note: is just an estimate [around true j]

• Stage 2: Use cross sectional data and estimated js to

estimate SML

b=market risk premium

Factor Pricing Slide 12-22
CAPM esting Fama French (1992)
• Using newer data slope of SML b is not significant (adding size and B/M)
• Dealing with econometrics problem:
 s are only noisy estimates, hence estimate of b is biased
 Solution: Portfolio
• Standard Answer: Find instrumental variable
• Answer in Finance: Derive estimates for portfolios
– Group stocks in 10 x 10 groups
sorted to size and estimated j
– Conduct Stage 1 of Fama-MacBeth for portfolios
– Assign all stocks in same portfolio same
– Problem: Does not resolve insignificance
• CAPM predictions: b is significant, all other variables insignificant
• Regressions: size and B/M are significant, b becomes insignificant
 Rejects CAPM
Factor Pricing Slide 12-23
Book to Market and Size

Factor Pricing Slide 12-24

 Fama French Three Factor Model
• Form 2x3 portfolios book/market
Size factor (SMB)
• Return of small minus big
Book/Market factor (HML)
• Return of high minus low
• For …
s are big and s do not vary much
• For …
(for each portfolio p using time series data)
s are zero, coefficients significant, high R2.
Factor Pricing Slide 12-25
 Fama French Three Factor Model
• Form 2x3 portfolios book/market
Size factor (SMB)
• Return of small minus big
Book/Market factor (HML)
• Return of high minus low
• For …
s are big and s do not vary much
• For …
(for each portfolio p using time series data)
ps are zero, coefficients significant, high R2.
Factor Pricing Slide 12-26
 Book to Market as a Predictor of Return

25%
Annualized Rate of Return

20%

15%

10%
Value
5%

0%
1 2 3 4 5 6 7 8 9 10
High Book/Market Low Book/Market
 Book to Market Equity of Portfolios Ranked by Beta

1
Book to Market Equity

0.9

0.8

0.7

0.6

0.5
0.6 0.8 1 1.2 1.4 1.6 1.8

Beta
 Adding Momentum Factor
• 5x5x5 portfolios
• Jegadeesh & Titman 1993 JF rank stocks
according to performance to past 6 months
Momentum Factor
Top Winner minus Bottom Losers Portfolios

Factor Pricing Slide 12-29

 Monthly Difference Between Winner and
Loser Portfolios at Announcement Dates

1.0%
Monthly Difference Between Winner and

0.5%
Loser Portfolios

0.0%
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

-0.5%

-1.0%

-1.5%

Months Following 6 Month Performance Period

 Cumulative Difference Between Winner and
Loser Portfolios at Announcement Dates

5%
Cumulative Difference Between
Winner and Loser Portfolios

4%
3%

2%
1%

0%

-1% 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

-2%

-3%
-4%

-5%

Months Following 6 Month Performance Period

Morgan Stanley‟s Macro Proxy Model
• Factors
 GDP growth
 Long-term interest rates
 Foreign exchange (Yen, Euro, Pound basket)
 Market Factor
 Commodities or oil price index
• Factor-mimicking portfolios (“Macro Proxy”)
 Stage 1: Regress individual stocks on macro factors
 Stage 2: Create long-short portfolios of most and least
sensitive stocks [5 quintiles]
• Macro Proxy return predicts macro factor

Factor Pricing Slide 12-32

 Haugen’s view: The Evolution of Academic Finance

The Old Finance

1930’s 40’s 50’s 60’s 70’s 80’s 90’s beyond

Modern Finance

Modern Finance
Theme: Valuation Based on Rational Economic Behavior
Paradigms: Optimization Irrelevance CAPM EMH
(Markowitz) (Modigliani & Miller) (Sharpe, Lintner & Mossen) (Fama)
Foundation: Financial Economics
Factor Pricing Slide 12-33
 Haugen’s view: The Evolution of Academic Finance

The Old Finance The New Finance

1930’s 40’s 50’s 60’s 70’s 80’s 90’s beyond

Modern Finance

The New Finance

Theme: Inefficient Markets
Paradigms: Inductive ad hoc Factor Models Behavioral Models
Expected Return Risk

Foundation: Statistics, Econometrics, and Psychology

Factor Pricing Slide 12-34