BEHAVIORAL FINANCE

1
BEHAVIORAL FINANCE

• Portfolio Management
• Ali Nejadmalayeri

2
Perfect Markets

• In perfect markets, prices converge to their
  equilibrium level because clever investors
  would arbitrage out the mispricings
• Imagine gasoline is selling too high in Reno
• 2.22/g gt 0.70 (crude) 1.30 (processing)
• You would import crude from Venezuela
• Contract out refining, ship the gasoline and sell
  the gas at a discount outlet for 2.10 and still
  make a profit!
• If your competitors are rational, they drop the
  price. Eventually the price will hover around
  2.00! THIS IS ARBITRAGE AT WORK!

3
Limited Arbitrage

• If your competitors are stubborn, and their
  customers for some reason are too loyal, and
  etc., then they would never drop the price, you
  would never react to that, and price stays high
  for good!
• Irrational market participants can create
  limited arbitrage opportunities they money
  making machine doesnt work all the time!
• WHY ARE PEOPLE IRRATONAL?
• Beliefs
• Preferences

4
Psychology of Decision Making

• Cognitive science shows that people possess
  certain biases
• Overconfidence
• Optimism (Wishful Thinking)
• Representativeness
• Conservatism
• Belief Perseverance
• Anchoring
• Availability Biases

5
Overconfidence

• People are overconfident about their judgment.
• Confidence bounds of estimates is much smaller
  than what it should be!

6
Optimism

• Most people have rosy views about their abilities
• Most people surveyed think they are above average
  driver, conversationalist, and humorist!

7
Representativeness

• Kahneman and Tversky (1974) show in determining
  what are the odds of certain person belonging to
  certain group, people use the representativeness
  heuristics.
• Linda is 31 years old, outspoken, and very
  bright. She majored in philosophy. As a student
  she was deeply concerned with issues of
  discrimination and social justice, and also
  participated in anti-war demonstrations.
• A Linda is a bank teller
• B Linda is bank teller and a feminist activist

8
Bayes Law Broken

• One needs to apply the Bayes Law
• With representativness people put too much in
  weight on p(descriptionstatement B)
• Makes assessment bases far too small of a sample!

9
Conservatism

• When data is representative of a model, people
  overweight data, but if data is not
  representative of a model, people react too
  little, too slowly.
• Urn 1 contains 3 blue 7 red balls,
• Urn 2 contains 7 blue 3 red balls
• In 12 random draws with replacement from the one
  of the urns yields 8 red ones and 4 blue ones,
  what is the probability that draws were made from
  the first urn?

10
Belief Perseverance

• One people form an opinion, they cling on to it
  too tightly and for too long!
• People dont search for evidence that contradicts
  the idea
• Galileo vs. Vatican
• Even when there is evidence, they treat it as
  exception to the rule
• Anomalies to the market efficiency!

11
Prelude to Anchoring

• What percentage of United Nations countries are
  from Africa?
• If you know that is not 5?
• If you know that is not 60?

12
Anchoring

• There are 191 countries in the United Nations out
  of which 54 countries are in Africa, so 28.27 is
  the answer!
• Kahneman and Tversky (1974) show that in forming
  estimates, people often start with some initial,
  possibly arbitrary, and then adjust away from it.

13
Availability Biases

• In determining probability of an event, people
  search for similar events in the their own (or
  their close friends) history.
• Whats the likelihood of getting less than B in
  my classes?
• Whats the likelihood of loosing on Intel?
• Whats the likelihood of getting whip-lashed by a
  teenager?

14
Preferences

• In perfect markets, participants make decisions
  based on the expected utility
• Evidence show that people do not follow expected
  utility framework
• A (1000 with ½ odds) vs. B (500 with 1/1)
• C (-1000 with ½ odds) vs. B (-500 with 1/1)
• Prospect theory attempt to explain the world
  rather than dictate how it should be

15
Experiment 1

• You all start the semester with an implicit grade
  of zero. Then you work your way up to the
  final grade.
• Now, for listening to my boring lecture next
  time, I offer two choices
• 10 points if I see you next time But theres a
  50 chance you might be out of town
• Five points right now to give me your word!

16
Experiment 2

• Imagine that you all start the semester with an
  implicit grade of 100. Then you can only loose
  points if you mess up!
• Now, for listening to my exciting lecture next
  time, I offer two choices
• Loose 10 points if I dont see you next time
  Knowing theres a 50 chance I might be out of
  town
• Loose five points right now for telling me that
  you dont give a hoot for my lectures!

17
Loss Aversion

• People do not necessarily hate risk rather they
  despise losses.
• People often do mental accounting meaning that
  they treat risky opportunities separately without
  considering the overall impact in their wealth.
• Consider (heads 2000 tails 500)
• Play the above five or six times?
• Play the above five times but you dont know your
  wins (losses). Would you play the sixth time?

18
Review of Probability Theory

• Using Excel
• Imagine that we are given the following data in
  an Excel sheet
• We now want to use the data to answer some simple
  probability questions.

19
Questions

• What is the likelihood of a negative return for
  each of the states?
• Now, imagine that you want to find what the
  likelihood of ending up in state 2 is if the ABC
  has lost value.
• Can we do the previous problem at the end of each
  day?

20
Question 1

• First you need to sort the data based on the
  state, then count the number of observations with
  less than zero data and divide that by the total
  number of observations in each state.
• For state 1
• COUNTIF(B1B7,"lt0")/COUNT(B1B7) 2/7 0.2857
• For sate 2
• COUNTIF(B8B10,"lt0")/COUNT(B8B10) 2/3
  0.6667
• Note that in the language of probability theory
  the aforementioned are conditional probabilities
  P( returnlt0 state1) 0.2857 and P( returnlt0
  state2) 0.6667.

21
Question 2

• This is inverse probability problems and
  Reverent Bayes Theorem has to be used
• In our case then we have P( returnlt0 state2)
  2/3 0.667.
• Also, P(returnlt0) COUNTIF(B1B10,"lt0")/COUNT(B1
  B10) 4/10
• and P(state2) COUNTIF(A1A10,"2")/COUNT(A1A10)
  3/10
• So the P(state2 returnlt0) 0.6667 ? 0.30 /
  0.40 0.50
• Can we verify this? Sure. First we need to sort
  the data based on returns. Then count the number
  of times that state 2 occurs when returns are
  negative. You find that out of 4 times that
  returns are negative, two times state 2 occurs so
  P(state2 returnlt0) 2/4 0.50!

22
Question 3

• Sure, however, note that at the end of each day,
  you only have up to that day worth of
  information. So every day your estimates of
  different probabilities changes.