Professor Anne Sibert
MSc Finance
Spring 2013
 Bankers
rashly bet that the US house bubble
would continue long after economists
predicted its demise.
 Managers of insurance companies and
pension funds did not exercise due
diligence when they purchased
collateralised debt obligations and
mortgage-backed securities that they did
not understand.
 Bankers were over confident and they took
on too much risk.
 “We
were hitting on all 99 cylinders, so
you have to ask yourself, What can we do
better? And I just can’t decide what that
might be … Everyone says that when the
markets turn around, we will suffer. But
let me tell you we are going to surprise
some people this time around. Bear
Stearns is a great place to be.” James E.
Cayne, Chairman and CEO of Bear
Stearns, 2003.
 Odean
(1998), Barber and Odean (2001),
Biais, Hilton, Mazurier and Pouget (2005)
demonstrate that overconfidence can
lead to excess trading and lower profits.
 Daniel, Hirshleifer and Subrahmanyam
(1998), Scheinkman and Xiong (2003)
and Burnside, Han, Hershleifer and Wang
(2011) show that it can lead to asset price
anomalies such as overreactions, excess
volatility and bubbles.
 An
individual has been described by a
neighbour as follows: “Steve is very shy
and withdrawn, invariably helpful but
with little interest in people or the world
of reality. A meek and tidy soul, he has a
need for order and structure and a
passion for detail.” Is Steve more likely to
be a librarian or a farmer?
Kahneman, D., Thinking, Fast and Slow, London, Allen Lane, 2011.
 Humans
have evolved to form
impressions, make judgements and
invent explanations quickly: Fast and
frugal heuristics
 A snap judgement: “This is dangerous”
may be more apt to keep us alive than a
slower and more finely nuanced thought.
 But, the heuristics that promote speed
may cause systematic biases.
 The
representativeness heuristic is where we
judge something by comparing it to our mental
picture of a category.
 An example from Myers (2008): “Linda, who is
31, single, outspoken, and very bright. She
majored in philosophy in college. As a student
she was deeply concerned with discrimination
and other social issues, and she participated in
anti-nuclear demonstrations. Based on that
description, would you say it is more likely that
• Linda is a bank teller
• Linda is a bank teller and active in the feminist
movement
 Most
people say that the answer is b. (Mellor et
al, 2001) But that cannot possibly be right!
 Some other examples: (Myers, 2008)
• Do more people live in Iraq or Tanzania? People
usually answer according to how readily Iraqis and
Tanzanians come to mind.
• Vivid, easy to imagine events (shark attacks) seem
more likely than hard-to-picture events.
• People are quick to infer general conclusions from a
single striking event: People switched from air travel
to car travel after 11 September 2001.
A person’s mental life can be described by the
metaphor of two internal agents. The first
produces that fast and frugal heuristics; the
second produces slow and deliberative thought.
Call these System 1 and System 2, respectively.
 System 1 can be very good. A team of fire
fighters is trying to douse a fire in a kitchen when
the chief, without knowing why, heard himself
yell to get out. Immediately after the floor
collapsed. Only later did he realise that the fire
had been unusually quiet and his ears unusually
hot. The heart of the fire was in the basement
below.

A
bat and a ball cost $1.10.
 The bat costs one dollar more than the
ball.
 How much does the ball cost?
 Do the math:
 Let bat := price
of the bat
ball := price of the ball
bat + ball = 1.10
bat = ball + 1.00
This implies that 2 x ball + 1.00 = 1.10 or that
2 x ball = 10 cents or that ball = five cents.
Thus the bat cost 1.05.
 Probably everyone who got this right
immediately thought of the answer that the
bat cost a dollar and the ball ten cents, but
they managed to resist it.
Another example from Kahneman (2011)
ANN
APPROACHED
THE BANK
 Going
back to the introverted Steve. Our
System 1 thinking associated Steve’s
personality traits with those of a
librarian.
 Unless our System 2 overrules our
System 1 we will say that Steve is more
likely to be a librarian.
 There are far more farmers than
librarians. It is more likely that Steve is a
farmer.
A
study of the incidence of kidney cancer
in 3,141 counties of the United States
reveals a remarkable pattern. The
counties in which the incidence is lowest
are mostly rural, sparsely populated and
located in traditionally Republican states
in the Midwest, the South and the West.
What do you make of this?
 The
counties in which the incidence of
kidney cancer is highest are mostly rural,
sparsely populated and located in
traditionally Republican states in the
Midwest, the South and the West. What do
you make of this?
 Suppose
an urn is filled with marbles.
One fifth of the marbles are red and fourfifths are blue.
 Suppose we consider a scenario where
we draw three marbles. The probability
that all are red is 1/125. In a scenario
where we draw seven marbles the
probability that they are all red is
1/78,125.
 Also
known as the “Lake Woebegone
Effect”.
 Overconfidence is pervasive.
Fischhoff (1977) found that when people claimed
to be 100 percent confident, they were right 70 –
80 percent of the time.
 Most of us are sure that we are better drivers
than average. Svenson (1891), for example, found
that 80 percent of survey respondents claimed to
be in the top 30 percent of all drivers.

 Daniel
Kahneman and Amos Tversky
(1979) asked people to fill in the gaps in
statements such as “I fell 98 percent
certain that the air distance between New
Delhi and Beijing is more than ___ miles
but less than ___ miles.”
 About 30 percent of the time the answers
lay outside the range they felt 98 percent
confident about.
Kahneman, D. a nd Tversky, A. (1797), “Intuitive Prediction: Biases and Corrective
Procedures,” Managnement Science 12, 313-327.
 Overconfidence
may lead to a failure to
look for disconfirming evidence.
 In
psychology experiments, the failure of
subjects to look for disconfirming
evidence has “raised more doubts over
human rationality than any other
psychological tasks.” Oaksford and
Chator (1993)
 Wason
gave participants in his test a
sequence: 2,4,6. He asked participants to
guess the rule that generated the sequence.
 The rule was: Any three ascending
numbers.
 The subjects were allowed to discover the
rule by generating sequences of three
numbers. They were then told whether there
sequence conformed to the rule or not.
 Participants
typically formed a belief
about what the rule was: say, counting by
twos.
 They then tested this by seeking for
confirming evidence, rather than
attempting to look for evidence that
might disprove their theories.
 There
is a sizable literature documenting
that men are more confident of their
abilities at activities that are perceived as
traditionally male.
 Using data on 35,000 households from a
large brokerage firm, Barber and Odean
(2001) argue that men are substantially
more overconfident than women in
financial markets.
 Overconfidence
is not related to ability: it
is found even when women perform as
well or better than men.
 Using
undergraduate and graduate
students, Lundeberg et al (1994) found
that most students were over confident,
but women were less overconfident then
men, “who tended to show high degrees
of confidence when wrong.”
 Campbell
and Sedikides (1999) document the
importance of the self-serving bias: a tendency
to attribute success to one’s own ability and
failure to external factors.
 Beyer
(1990) and others, however, find that this
effect is stronger in men than in women. Thus, if
confidence helps produce successful
outcomes, there is more likely to be strong
feedback loop in confidence in men than in
women.
 In
a fascinating and innovative study, Coates
and Herbert (2008) consider the effect of
testosterone on both risk aversion and
overconfidence.
 They explain that testosterone – a steroid
hormone -- rises in athletes preparing for a
contest; it rises further in winners and falls in
losers.
 Because increased testosterone increases
confidence and risk taking, this produces a
positive feedback loop in the winners as it
improves their chances of winning again.
 They hypothesised that something similar
might happen in financial market participants.




To test this, they took samples of testosterone levels of
17 male traders on a typical London trading floor
(which had 260 traders, only four of whom were
female) twice a day for eight days.
(They controlled for food, medications and whether the
traders had received any important news from outside
work.)
They found that testosterone was significantly higher
on days when traders made more than their daily onemonth average profit.
Furthermore, higher levels of testosterone led to
greater profitability – presumably because of greater
confidence and risk taking.




The authors hypothesise that if raised testosterone were to
persist for several weeks the elevated appetite for risk
taking might have important behavioural consequences and
there might be cognitive implications as well.
Testosterone, they say, has receptors throughout the areas of
the brain that neuroeconomic research has identified as
contributing to irrational financial decisions.
They also point to a study where testosterone was
administered to a group of subjects playing a gambling task
and it led to irrational behaviour: subjects preferred options
with high variance and a negative expected return to
options with a low variance and a positive expected return.
They speculate that steroid feedback loops may help
explain why bankers behave irrationally when caught up in
bubbles.



ρ
The announced
accuracy ρ is
increasing in
competency
(although not
necessarily as a
straight line) up to
some point where
experts they are
perfectly accurate.
This point can be at
one half or at one.
1
1/2
1/2
π*
π
1



 Let
y = f(x). Then f-1(y) = f-1(f(x)) = x.
Differentiating yields f-1’(f(x)) f’(x) = 1
and thus f-1’(y) = 1/ f’(x). The derivative
of the inverse function is the reciprocal of
the derivative of the original function.
 Differentiating f-1’(f(x)) f’(x) = 1 yields
f-1’’(f(x)) f’(x)2 + f-1’(f(x)) f’’(x) = 0. This
implies f-1’’(y) = - f-1’(y) f’’(x)/ f’(x)2 =
f’’(x)/ f’(x)3 .



 Relatively
incompetent experts would be
willing to say that they are more confident
than they are to be perceived as more
competent than they are.
 But, relatively competent experts avoid
being imitated (and thus be perceived as
less competent than they are) by saying
they are more competent than they are.
 It is less costly for a competent expert to
exaggerate his competency because he is
less likely to be wrong. So, relatively
competent experts avoid being imitated.
 There
is an upper bound on how
confident you can be.
 So there is a region where relatively
competent experts cannot separate
themselves.