HEC Paris — M1 — Jan-Mar 2025

Advanced Finance

[#1] Market Efficiency

Johan Hombert, Daniel Schmidt

1 / 29
 Agenda

• CAPM (recap)

• α — risk or mispricing?

• Other factor models

• Market efficiency

• How can mispricing persist? (Limits of arbitrage)

• Sources of mispricing/inefficiencies:
I Behavioral biases
I Institutional constraints
I Information frictions (biased beliefs, limited attention)

2 / 29
The Capital Asset Pricing Model (CAPM)

William F. Sharpe

Nobel Prize in Economics 1990

3 / 29
 The Capital Asset Pricing Model (CAPM)

• CAPM = equilibrium model that determines an asset’s expected

return as a function of its risk

• Equilibrium means:
1. Investors choose optimal portfolios
2. Markets clear (demand = supply)

• The CAPM has profoundly changed asset management and is widely

used for capital budgeting decisions

• Key assumptions:
I 1 riskfree asset, N risky assets in positive net supply
I Investors have common beliefs about expected payoffs and risks
I Investors are rational and have standard preferences
I No frictions

4 / 29
 CAPM — key results

• Investors hold combination of riskfree asset and market portfolio M

• M contains all risky assets, where the weight of asset i is given by:

#shares outstandingi × pricei

wi =
∑N
j =1 #shares outstandingj × pricej

• The expected return of asset i is given by:

Cov (ri , rM )
E (ri ) = rf + β i × (E (rM ) − rf ) with βi =
| {z } Var (rM )
asset i ’s risk premium | {z }
beta captures i ’s exposure
to market risk

⇒ E (ri ) does not depend on σi as investors diversify away

i’s idiosyncratic risk

5 / 29
 CAPM — quiz

1. What is β f ?

2. What is β M ?

3. Suppose the company Foreclosure Inc (FI) has a market beta of

β FI = −0.2. What is E (rFI ) if rf = 3% and E (rM ) = 8%?

4. Under the CAPM, does it happen that an investor holds short

positions in individual stocks?

[NB: a short sale position means that the investor has shorted the stock (i.e.,
he/she borrowed and then sold the stock)]

5. Suppose stock A has E (rA ) = 10% and σA = 10%, and stock B has
E (rB ) = 10% and σB = 20%. Does anyone hold stock B?

6. Suppose stock A’s market cap is twice the market cap of stock B.
What does this imply for investors’ portfolio weights in A and B?

6 / 29
 CAPM — Capital Market Line (CML)
• Under the CAPM, efficient portfolios lie on the CML spanned by the
risk-free asset and the market portfolio M
𝐸 𝑟
Capital Market Line (CML)

𝑀
𝐸 𝑟
Slope of CML:
𝐸 𝑟 𝑟
𝜆
𝜎

𝜎 𝜎

E (rP )−rF
• Define Sharpe ratio λP = σP
• All efficient portfolios satisfy λP = λM
7 / 29
 CAPM — Security Market Line (SML)
• Under the CAPM, all stocks (and portfolios) lie on the SML

• Define αi = ri − rf − β i × (E (rM ) − rf )
• If the CAPM holds true, then E (αi ) = 0 for all stocks and portfolios

[NB: we don’t expect αi = 0 every period, but rather αi = 0 on average]

8 / 29
 Alpha

• Stock A’s return was too high given its risk (β A ) ⇒ αA > 0
• A appears to be under valued (return too high ⇒ price too low)
• Investors want to buy A, thereby pushing up the price until αA = 0

[See Problem 1 of Problem Set]

9 / 29
 Estimating Alpha

• Consider time-series data, t = 1, ..., T

rit : return on asset (or portfolio) i in period t (e.g., month)

rmt : market return in period t

rft : riskfree return in period t (known at t − 1)

• Run the following regression, called the "market model" for asset i:

rit − rft = αi + β i × (rmt − rft ) + eit

Cov (rit −rft ,rmt −rft )

⇒ βi = Var (rmt −rft )
is asset i’s market beta

⇒ αi should not be statistically different from zero

[See Problem 2 of Problem Set]

10 / 29
 Three reasons for finding Alpha
1. Statistical fluke (i.e., pure luck)
I sample may have been too small/unrepresentative
I alpha disappears if sample is extended

⇒ Always need to back-test and test out-of-sample any trading idea

2. Wrong risk model
I CAPM only has one risk factor: the Market Factor
I CAPM only holds under specific assumptions
I different assumptions yield other models (possibly with >1 factors)

⇒ What looks like α may in fact be a compensation for risk

3. Mispricing
I when not 1. and 2., then there is a genuine trading opportunity
I α > 0 ⇒ asset is under valued ⇒ investors want to buy
I α < 0 ⇒ asset is over valued ⇒ investors want to short-sell
⇒ α should decrease as investors trade on (exploit) the mispricing
11 / 29
 Other risk models

• Other popular risk models:

1. Fama/French 3-Factor Model

rit − rft = αi + βM SMB

i (rmt − rft ) + β t SMBt + βHML
t HMLt + eit

SMBt : return on small-minus-big portfolio in period t

HMLt : return on high-minus-low book-to-market portfolio in period t

2. Carhart 4-Factor Model

rit − rft = αi + βM SMB

i (rmt − rft ) + β t SMBt + βHML
t HMLt + βWML
t WMLt + eit

WMLt : return on winner-minus-loser portfolio in period t

• Data on factor returns available at Kenneth French’s data library

12 / 29
 Risk or mispricing?

• One-factor market model is implied by theory (i.e., the CAPM and

its underlying assumptions)

• Instead, three-factor and four-factor models are empirical models:

I small stocks tend to outperform big stocks ⇒ SMB ("Size Factor")

I high book-to-market stocks tend to outperform low book-to-market

stocks ⇒ HML ("Value Factor")

I past-winner stocks tend to outperform past-loser stocks ⇒ WML

("Momentum Factor")

⇒ Open question whether factor returns are compensation for risk

or an anomaly (i.e., come from mispricing )

13 / 29
 Market Efficiency (?) – The Case of NVIDIA

• On Jan 27, 2025, NVIDIA dropped 17%, wiping almost $600 billion
from its market cap (due to DeepSeek’s claim that their powerful AI
models need far fewer chips)
• Overdue market correction or overreaction?
14 / 29
 Market Efficiency

Eugene Fama
Nobel Prize in Economics 2013

"I’d compare stock pickers to astrologers but I don’t want to bad

mouth astrologers."
"What I like to remind people is that active management is a
zero-sum game before costs."
15 / 29
 Efficient Market Hypothesis (EMH)
Asset prices reflect all available information
⇔ impossible to "beat the market" on a risk-adjusted basis
1. Weak form: asset prices reflect information in past prices
I Counterexample: return momentum (Jegadeesh and Titman,
1993)—past winners outperform past losers

2. Semi-strong form: asset prices reflect all public information

I Counterexample: post-earnings announcement drift (Bernard and
Thomas, 1989)—stocks with positive (negative) earnings surprises
continue to outperform (underperform) over the next trading days

3. Strong form: asset prices reflect all public and private information
I Grossman-Stiglitz Paradox: if this were true, smart investors couldn’t
profit from trading on information they have painstakingly collected;
but how could this information be in the price then?

[See Problem 3 of Problem Set]

16 / 29
 Semi-strong Efficiency
• Stock prices react to public news (e.g., earnings announcements)

• But prices drift in direction of the news after the announcement

(the so-called post-earnings announcement drift, PEAD)
⇒ PEAD is inconsistent with semi-strong efficiency
17 / 29
 Market Efficiency — quiz
Yesterday, pharmaceutical company BigPharma Inc announced excellent
results for its clinical trials of a coronavirus vaccine. This represents very
good news for future earnings of BigPharma Inc.

1. If markets are efficient, is it a good idea to buy BigPharma Inc?

Yes or no?

2. If markets are inefficient, is it a good idea to buy BigPharma Inc?

Yes or no?

E-commerce company Mekong Inc announces that its sales are up 25%
compared to last year. Upon the announcement, Mekong Inc’s stock
price drops 5%.

3. The price reaction shows that the stock market is inefficient.

True or false?

18 / 29
 Efficient Market Hypothesis (EMH)
• EMH is a benchmark against which to compare market outcomes
• Inefficiencies do exist, but they are rare and difficult to exploit

I Alphas for active U.S. mutual funds over 1984-2006

(Source: Fama and French, 2010)
⇒ Aggregate mutual fund portfolio is close to market portfolio M
⇒ Alphas net of fees are negative!
19 / 29
 Anomalies
• Market inefficiencies give rise to arbitrage opportunities
(a.k.a. as anomalies or α)
• Academics and investors have found many anomalies

• Examples of positive-α portfolios:

I Long high-profitability, short low-profitability stocks (Novy-Marx,
2013)
I Long low-idiosyncratic volatility, short high-idiosyncratic volatility
stocks (Ang et al., 2006)
I Long low-beta, short high-beta stocks (Frazzini and Pedersen, 2014)

I ...

• Anomaly returns (α) decrease after publication (McLean and Pontiff,

2016)
⇒ Arbitrage opportunities do exist!
⇒ But they are (partly) competed away after discovery
20 / 29
 Hedge funds as arbitrageurs
• Hedge funds are in the business of finding α

• Invest their clients’ money and charge performance fees

• Hedge fund managers earn a ton of money! (top earners in 2019)

⇒ Hedge funds can only make money when arbitrage opportunities exist
⇒ High earnings suggest that skill to find and exploit α is rare
21 / 29
 Limits to arbitrage

• Anomalies (α) can only persist when arbitrage is limited

I otherwise arbitrageurs (e.g., hedge funds) would trade until α =0

• Reasons why arbitrage is limited:

I Arbitrage is not riskfree; mispricing could widen before it disappears

(e.g., GameStop)

I Fund investors/lenders may withdraw their money after initial losses

I May force funds to unwind their positions at the worst possible time,
exacerbating the losses (Shleifer and Vishny, 1997)

[See Problem 4 of Problem Set]

22 / 29
 GameStop short squeeze

• Example for the limits of arbitrage: GameStop short squeeze

400 300

Short squeeze commences as hedge funds

GameStop closing price

Melvin Capital and Citron Research are

forced to cover their short position
200

GameStop is talked up on social media;

retail investors on Robinhood heavily buy
100 0

01oct2020 20nov2020 09jan2021 28feb2021

⇒ Hedge funds had to cover short positions at huge losses

23 / 29
 Three sources of mispricing
1. Behavioral biases
I individual investors succumb to behavioral biases
(e.g., overconfidence, disposition effect, cognitive dissonance)

⇒ "Noise" trading pushes prices away from fundamentals

2. Institutional constraints
I many institutional investors face constraints
(e.g., on short-selling, borrowing, which assets they can hold)

⇒ Price distortions arise as constraints become binding

3. Information frictions
I attention is limited

I systematic deviations from rational expectations

⇒ Lead to patterns of under- and overreaction

24 / 29
 Example for behavioral bias: overconfidence

• Individual investors are overconfident

I "the investor’s chief problem—and even his worst enemy—is likely to

be himself" (Benjamin Graham)

I "93% of American drivers rate themselves as better than the

median" (Svenson, 1981)

I self attribution bias: profits are due to skill, losses due to bad luck

⇒ Individual investors trade too much (Barber and Odean, 2001)

25 / 29
Example for institutional constraints: betting against beta
• Many institutional investors have leverage constraints
• Only way to have high expected returns is to buy high-beta stocks;
excess demand for high-beta stocks (Frazzini and Pedersen, 2014)

𝐸 𝑟
SML as predicted by theory (CAPM)

𝛼
SML as found in the data

⇒ High-beta (low-beta) stocks have α < 0 (α > 0)

26 / 29
Example for information friction: sticky expectations
• People have "sticky" expectations: they put too much weight on
their priors and update too litte
• Even prevalent among professionals such as financial analysts:

⇒ Investors underreact to good (bad) news about companies’ profits

⇒ High-profitability stocks have α > 0 (Landier et al., 2019)
27 / 29
Example for information friction: perceptual blindness

Watch this video and answer the questions.

28 / 29
 Market efficiency — summary

• If markets were efficient, there would be no α!

• If you find α, there are three possibilities:

1. it is spurious

2. the risk model is not correct

3. it is mispricing (anomaly)

• Mispricings can occasionally persist because of:

1. limits to arbitrage

2. behavioral biases/institutional constraints/information frictions

• Mispricings/anomalies are rare (as they are chased by smart

investors such as hedge funds)

⇒ Don’t try to swim with the sharks! (More on this in Lecture 2)

29 / 29