Summary Advanced Behavioral Finance
Lecture 1: Prospect Theory & Dividends
Traditional Finance: according to the rational expectations hypothesis investors and
managers are generally rational, the market is efficient and asset price therefore
incorporate all available information. Behavioral Finance: tries to explain why prices
do not always reflect the fundamental value. The idea of this hypothesis is that not
all agents are rational. Furthermore, it tries to explain finance by incorporating
psychology. The main areas are:
1. Asset pricing: aggregate stock market, bond market and real estate.
Performance of various stock market strategies and bubbles.
2. Investor behavior: portfolios that investors hold, trading activity over time
3. Corporate finance: security issuance, capital structure, investment decisions,
M&A activity
An agent is considered to be rational if he makes decisions based on the following
different criteria:
1. Decisions are based on current assets
2. Decisions are based on possible consequences of the choices (no framing-
effect)
3. Agents make choices which are consistent with the expected utility
framework:
a. Consider different possible future outcomes
b. Decide how good/bad each outcome will make him feel
c. Weight each outcome by its probability and sum up; how affects it, his
overall wealth?
4. When agents receive new information they update their expectation
correctly, as described by Bayes law (no over- or under-reaction)
According to behavioral finance not all agents are rational. Some will fail to
incorporate all available information correctly. Their choices are normatively
unacceptable. Moreover, there are limits to arbitrage: rational agents cannot always
correct the irrationality of other investors. Recall, traditional finance argues that in
the existence of irrational behavior arbitrage possibilities occur, and rational agents
will use these and correct the irrational behavior. Irrational agents will be removed
from the market. But, the irrational agents can still have a substantial and long
impact and the problem will increase over time. To conclude, there is behavioral
bias (psychology):
Irrational decisions are not random
Systematic form of irrationality that creates a mispricing
Belief formation – how agents form expectations
Decision making/preferences – how agents evaluate risk decisions.
Limits to arbitrage: even when asset is mispriced, strategies to correct mispricing
can be costly and risk and therefore unattractive. Engaging in arbitrage can mean
facing substantial losses.
, 1. Fundamental risk: risk that assets held by investors loses value. Investors
need to find a perfect hedge/substitute which is usually impossible/difficult.
2. Noise trader risk: mispricing worsens further in short run, due to investor
sentiment. It can force arbitrageurs to liquidate their positions prematurely.
3. Implementation costs: costs that makes execution of arbitrage costly;
transaction, short-sale and information costs. Horizon and synchronization
risk.
Sufficient conditions to limit Arbitrage
Arbitrageurs are risk averse and have short horizons
o Arbitrageurs cannot afford to be patient
o Creditors and investors evaluate the arbitrageur based on his returns
o Forced disclosure of a short position
Noise trader risk systematic
o Investor sentiment causing undervaluation of an assets relative to
others, can also cause the mispricing to increase in the short term
Noise trader risk is really important, systematic so can move. This happens on
markets which are not correlated, arbitrageurs cannot afford to be patient. If
creditors become impatience, they will force you to close your position. So imagine
you need to borrow stocks for a short position, if creditors lose their faith, they want
his stock back and you will lose your money.
To identify mispricing we need to know fundamental values. But, what is the true
asset pricing model? Joint hypothesis problem: any test of mispricing is a joint test
of mispricing of an asset pricing model. Two possible options: 1) mispricing 2) wrong
model. There are only a few cases where mispricing can be measured beyond any
doubt.
Twin shares: Dual listed companies
Two corporations function as a single operating business. The fundamental risk is
zero and implementation costs small. But, how about the noise trader risk? For
example, we have Royall Dutch Shell (60% Royall Dutch, 40% Shell). When we go
short in Royall Dutch and long in Shell; we are able to get a huge return of 10%.
There is no fundamental risk because cash flows follow the same pattern.
Index inclusions of stocks: when a stock is added to the index, the index jumps
even though the fundamental value has not changed. It should not happen because
there is no information about increase profits. In this case, investors face
fundamental risk and noise trader risk.
Closed end funds: mutual funds issuing a fixed numbers of shares, traded on
exchange, fund share prices differ from the Net Asset Values. You mainly face noise
trader risk in this case.
Bubbles: during bubbles there were significant short-sale constraints; when these
disappear the bubbles burst. During dotcom bubble there was limited short-selling,
high implementation costs. During housing bubble no short selling possibilities at
all.
, Irrational agents do not make random decisions and therefore can create
mispricing. We can separate this in two blocks:
1. Beliefs; how agents form their expectations
a. Representativeness problem (Kahnemevent & Tversky): agents
determine whether an event or sample is representative based on:
i. Similarity of the sample to the parent population
ii. Reflection of randomness
b. Very small samples: people want to see 50-50 distribution, but if there
is no perfect pattern it is not reprehensive, like the second distribution:
(Boy Girl Boy Girl; Boy Boy Girl Boy Girl Girl).
c. Reflection of randomness
2. Decision making: how agents evaluate risk decisions.
Two biases generated by representativeness
Sample size neglect; belief in law of small numbers where people tend to see
trend to quickly in financial data even though the sample size is too small.
They forget that the pattern is random.
Base rate neglect; description belongs to a philosophy student but in the end
the probability that it is a philosophy student is low, because small faculty.
Therefore, people expect that it is an economic student for example although
they know it is an philosophy student. People expect that a small sample will
reflect the properties of the model that generate it. Agents tend to infer the
model on the basis of too few data points.
Price earnings ratio is hard to explain. If P/E is 20 and you expect it to follow the
pattern, it will take 20 years before you get back what you have invested. The PE is
high today because high PE expectation in the future. The problem is that this idea
has been rejected: Robert Shriller argues that if PE reflects high PE in the future it
should be able to make an estimation of PE in the future; however this fails. Maybe
because PE is high they expect that risk level/ interest will be low in the future. The
model that is left is the Habit Model. Talks about changes in the risk aversion of
investors; if positive news on the stock market and prices go up, investors are
suddenly richer are less worried that investors they have reduce their life-style.
Meaning that the investor becomes less risk-averse and therefore the stock market
even rise further. The behavioral model argues the same but has a different starting
point. A subset of rational investors follows the Habit Model; based on
representativeness. There is mean-reverting behavior, investors going up and down.
Investors might not be correct but they think they are, because they were for 1 or 2
years, explaining bubbles. The Habit Model argues that if PE is high, they expect low
returns next year; if PE low they are more risk-averse so demand more return.
Behavioral model argues the opposite, supported with historical data. If investors
observe a high PE they expect a high PE in the future too.
Overconfidence
Over-optimism: people overestimate own abilities. Systematic planning
fallacy, thinking you’re done on Monday but instead it is Wednesday.
(Malmendier and Tate).