Asset Pricing Models

What is the price of a stock?

Discounted value of all the future cash flows!

Expected return R (discount rate) depends on:

I. future cash flows
II. time value of money,
III. riskiness of the cash flows
Jonathan Berk Peter Demarzo
Stanford University
2
Hold a stock or a portfolio? Diversification

Var(R) = σ2

3
 Which Portfolio? MV frontier: Market Portfolio

Mean-Variance Framework

Efficient
Frontier

4
How the return of a stock is decided? CAPM

5
 CAPM Lessons

1. Don’t hold an individual asset, hold the market

2. The average investor holds the market

3. Assets losing more in bad times (with high beta) require high risk
premiums

6
 Tests of Asset Pricing Models and Pricing Anomalies

 Time-series test: 𝛼𝛼𝑖𝑖 = 0 for each stock i

 Cross-section test: 𝛾𝛾0𝑡𝑡 = 0

 Fama-MacBeth two-step regression:

Step 1: produces estimates bi of βi

Step 2:

 Pricing Anomalies: 𝛂𝛂𝐢𝐢 ≠ 𝟎𝟎, 𝜸𝜸𝟎𝟎𝟎𝟎 ≠ 𝟎𝟎

7
 How to Identify Pricing Anomalies (new factors)

Portfolios Analysis (the simplest method)

1. Sort stocks into 10 portfolios every month/year based on a signal

2. Calculate the average returns and alphas of these portfolios, and

check if there is a difference

8
 Portfolio Sorting

 Fama and French (1992): sort by Size and Book-to-Market ratio every year.

 Pastor and Stambaugh (2003): sort by liquidity betas every year.

9
 Risk and Return
40

35

30

25
Mean/Stdev (%)

20 Mean
Stdev
15

10

0
Market Small 1 Large 10 Growth 1 Value 10 Losers 1 Winners 10

Size Value Momentum

10
 Factor: Definition

 A factor is a variable which influences the returns of assets

 Exposure to factor risk over the long run yields a risk premium

 The premium does not come for free: it is compensation for bearing losses during
bad times. Being exposed to the factor results in holding risk that other investors
seek to avoid.

Pricing Factors

 Size (SMB), value (HML), profitability (RMW), investment (CMA) …

 Fama French 3-Factor (5 Factor) Model

11
 Other Anomalies (will be introduced in Lecture 2)

 Value, Momentum and Reversals

 Low-Vol Anomalies

1. Idiosyncratic volatility puzzle

2. Betting against beta

 Liquidity Risk

 Seasonality

Hou, Xue and Zhang (2019) summarize 452 anomalies in the literature

12
 How to deal with new anomalies?
1. Industry: trade on it, and hope it will last

2. Academia: try to find an explanation to it, and have a new model

(either a risk-based explanation or a behavioral-based explanation)

Understanding the reason behind an anomaly helps industry to know

a. When does it work

b. Whether it will last

David Booth Cliff Asness

Funders of Dimensional Fund and AQR Capital

are both former PhD students of

13 Eugene Fama
Mutual Fund Performance Puzzle
 Mutual Fund Industry

 Mutual funds manage about 100 trillion dollars of assets

worldwide (as of 2022)

 Fund managers are among the highest paying jobs

 Do fund managers outperform the market?

15
 Mutual Fund Performance Puzzle

Abnormal Returns (in % per year) Period

Gross (before fees) Net (after fees)

Wermers (2000) 0.71** -1.16*** 1975–1994

(2.79) (2.96)
Fama and French (2010) -0.05 -1.00*** 1984–2006
(-0.15) (-3.02)
Berk and van Binsbergen (2015) 0.78** -0.12 1977–2011
(>1.96) (-0.31)

Eugene Fama Russ Wermers Jonathan Berk Jules van Binsbergen

Chicago Booth Uni. of Maryland Stanford Uni. Wharton

Skill measures: Risk-adjust alphas Style-adjust alphas Value Added

16
 Literature on Fund Performance

 “Old consensus” in the academic literature:

– Active investors as represented by mutual funds have no skill based on
returns after fees: Jensen (1968), Fama (1970), Carhart (1997)

 Paradox: Fund managers earn a lot. Are investors all stupid?

Berk and Green (2004): No. Managers outperform, but they do not share their profits with investors.

 “New consensus” in the academic literature

– The average mutual fund outperforms before fees, but not after fees. The
average hides significant cross-sectional variation across good/bad managers:
Wermers (2000), Berk and van Binsbergen (2015)

 Consistent with Efficiently Inefficient Markets --- Lasse Pedersen

If you think everyone else is stupid, you're probably the stupid one. 17
 Efficient Market Hypothesis --- Eugene Fama (1960s)

 Efficient markets hypothesis (EMH)

– states that markets are efficient, with market prices reflecting all
available information at any given time

1. Weak form – states that stock prices fully reflect all information
contained in past prices and volumes of trading

2. Semi-strong form – suggests security prices adjust rapidly reflecting

all available public information

3. Strong form – implies share prices reflect all information, public and
private, and no one can earn excess returns

18
 Joint Hypothesis Problem

 Testing for market efficiency is difficult, or even impossible. Any attempts

to test for market (in)efficiency must involve asset pricing models so that
there are expected returns to compare to real returns.

 When real returns differ from expected returns (implied by the model)

Either

 Market price is wrong (market is inefficient)

or

 Model is wrong (market is efficient)

19
 Tests of Asset Pricing Models and Pricing Anomalies

 Time-series test: 𝛼𝛼𝑖𝑖 = 0 for each stock i

 Cross-section test: 𝛾𝛾0𝑡𝑡 = 0

 Fama-MacBeth two-step regression:

Step 1: produces estimates bi of βi

Step 2:

 Pricing Anomalies: 𝛂𝛂𝐢𝐢 ≠ 𝟎𝟎, 𝜸𝜸𝟎𝟎𝟎𝟎 ≠ 𝟎𝟎

20
Efficiently Inefficient Market
 Market Efficiency

 Market efficiency: at the heart of financial economics

 Nobel Prize 2013 awarded to Eugene Fama, Lars Hansen, and Robert Shiller

Inefficient!
Efficient!
 Efficient Markets?

Markets cannot be fully efficient

1. If they were, there would be NO incentive to collect information (Grossman-

Stiglitz, 1980)

2. Logically impossible that both market for asset management and asset markets fully
efficient
– Asset market efficient  no one should pay for active management

3. Clear evidence against market efficiency

Failure of the Law of One Price, e.g.
– Asset management: Closed-end fund discount, ETFs
– Stocks: Siamese twin stock spreads
– Bonds: Off-the-run vs. on-the-run bond spreads
– FX: Covered interest-rate parity violations Completely
Perfectly
– Credit: CDS-bond basis Inefficient Efficient

Not subject to “joint hypothesis problem”

EFFICIENCY-O-METER
 Inefficient Markets?

Market prices cannot be completely divorced from fundamentals

1. Money managers compete to buy low and sell high

2. Free entry of managers and capital

3. If markets were completely divorced from fundamentals

– Making money should be very easy

– But, professional managers hardly beat

the market on average

Completely
Perfectly
Inefficient Efficient

EFFICIENCY-O-METER

24
 Efficiently Inefficient Markets

Markets are efficiently inefficient

 Markets must be
– inefficient enough that active investors are compensated for their costs
– efficient enough to discourage additional active investing

 Investment implications
– some people must be able to beat the market

Market Efficiency Investment

Implications
Efficiently
Efficient market hypothesis Passive investing Inefficient
Inefficient market Active investing Completely
Inefficient
Perfectly
Efficient

Efficiently inefficient markets Active investing by those

with comparative EFFICIENCY-O-METER

advantage 25
Flow-Performance Sensitivity Puzzle
 Flow-Performance Sensitivity Puzzle

Fact 1: Investors chase best performing funds (Chevalier and Ellison 1997)

27
 Flow-Performance Sensitivity Puzzle
Fact 2: Best performing funds do not continue to outperform (Carhart 1997)

28
29
 Flow-Performance Sensitivity Puzzle
Facts:

1. Investors chase best performing funds (Chevalier and Ellison 1997)

2. Best performing funds do not continue to outperform (Carhart 1997)

Are investors insane?

Berk and Green (2004):

 No, because more capital automatically leads to lower % return.

i.e., Decreasing Returns to Scale

Berk and van Binsbergen (2015):

 Skill of fund managers should not be measured by % returns, but by dollar

amount of money that they can earn (value added).

If you think everyone else is insane, you're probably the one that is insane.
30
 Decreasing Returns to Scale

Manager A
AUM
(Asset Under Management) $ 1,000,000 $ 1,000,000,000
Abnormal Returns 10% ?

31
 Decreasing Returns to Scale

Manager A
AUM
(Asset Under Management) $ 1,000,000 $ 1,000,000,000
Abnormal Returns 10% 1%

Potentially because of
(1) the price impact of large trades
(2) limited number of investment ideas

32
 Berk and Green (2004)

 There is decreasing returns to scale in fund return (gross alpha decreases)

 The ability to identify mispricing (alpha) is in scarce supply

(i.e., the stock market is relatively efficient)

 Investors competitively supply capital to skilled managers (net alpha = 0)

(i.e., the asset management market is fully efficient)

0
 The economic rents (revenue) go to the managers who create them, not to the
investors who invest in them.

33
 Value Added Measure

Manager A Manager B
AUM
(Asset Under Management) $ 1,000,000 $ 1,000,000,000
Abnormal Returns 10% 2%
Abnormal Return / Revenue
(in Dollar) $ 100,000 $ 20,000,000

34
 Value Added Measure in Berk and van Binsbergen (2015 & 2017)

 Gross alpha is decreasing with fund size q

 Value added is the product of gross alpha and fund size q

 A fund chooses the optimal amount to invest

 The maximum value added is

= f q (revenue in $)
 Gross alpha is

35
 Size, value added, and gross alpha

36
-- from Berk and van Binsbergen (2017)
 Size, value added, and gross alpha

37
-- from Berk and van Binsbergen (2017)
 Why Gross and Net Alphas do not measure skill?

Manager A Manager B
AUM
(Asset Under Management) $ 1,000,000 $ 1,000,000,000
Gross Alphas 10% 2%
Fees 10% 2%
Net Alphas 0% 0%
Abnormal Return / Revenue
(in Dollar) $ 100,000 $ 20,000,000

38
 Summary: Why Gross and Net Alphas do not measure skill?

 Net alpha measures how much fund investors get

0
 Gross alpha always equals to the fee that funds charge (in equilibrium)

= f
 When q > q*=a/2b, fund manager should index the excess money

= f ≠

-- from Berk and van Binsbergen (2017)

39
 Persistence of Value Added

Sorted by: past compensation (fees*AUM) past value added (gross alpha*AUM)
-- from Berk and van Binsbergen (2015)
40
 Using Vanguard Index Funds as the Benchmark

 Use investable alternative investment opportunity as the benchmark (i.e. Vanguard)

 Factors do not take transaction costs into account (i.e., momentum)

 Use factors to evaluate funds when they were not even introduced? (i.e., 3 factors
first introduced in 1992, momentum factor in 1997)
41
-- from Berk and van Binsbergen (2015)
 Using Vanguard Index Funds as the Benchmark

Table: Net alphas of all funds (in bp/month)

-- from Berk and van Binsbergen (2015) 42

 Average Value Added of Mutual Funds is Significantly Positive

million $s / month

-- from Berk and van Binsbergen (2015)

43
 Howtotoread
How read
an an article?
article?

Most Important: Story

Read Abstract + Introduction
 Research Question
 Methodology
 Main Findings
 Contributions
Search for the contents you are interested in
 e.g. Tables, equations
Read some paragraphs word by word only if you want to
 replicate their analyses

Example: Pastor and Stambaugh (2003)

44