Chance - Misunderstanding risk

advertisement
Don M. Chance, PhD, CFA
William H. Wright, Jr. Endowed Chair for Financial Services
Louisiana State University
CFA South Africa Investment Conference 2009
9 September, 2009
Johannesburg, South Africa
A Most Difficult Decision
•
What would you do if you found a
suspicious growth inside a non-vital
organ?
• The doctor is unable to identify whether the
lump is malignant or not without surgery
• The doctor gives you a choice:
• Remove the lump only, examine for the
presence of cancer, and return for a second
more extensive surgery if cancer is found
• Remove the entire organ, which you can live
without, a more drastic surgery
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 2 of 42
Making Decisions Under Risk
Not just in the financial world, but in life
in general
 We constantly make decisions under
risk
 Rarely are we ever presented certainty
 How we make those decisions tells us a
lot about the human appetite for risk and
how we so often make risk mistakes

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 3 of 42
What is Risk?
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 4 of 42
Elements of the Definition of Risk?
Exposure to an event
 Uncertain outcomes
 (Usually) undesirability of one or more
outcomes

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 5 of 42
Measuring Risk

Measures of risk
 Probability or odds
 Standard deviation
 Worst case scenario
 In investing
○ Maximum drawdown
○ Beta
○ Value-at-Risk
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 6 of 42
Probability as a Measure of Risk
We do not understand it
 We misuse it
 All too often we assume the bell curve
and oftentimes we ignore the tails (the
Black Swan effect)

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 7 of 42
A Simple (and somewhat silly)
Example: The Monty Hall Problem
Behind these doors are three prizes: a car and
two worthless prizes. A contestant chooses one
door.
Door # 1
9 September, 2009
Door #2
Don M. Chance, Ph.D., CFA
Door #3
p. 8 of 42
The Monty Hall Problem (cont.)
Let’s say the contestant chooses Door # 3.
Monty then opens door # 1 to reveal one of the
two worthless prizes and offers the contestant the
opportunity to switch doors.
$0
Door #2
Door #3
(the contestant’s choice)
Should the contestant switch? Most people
say it doesn’t matter.
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 9 of 42
But it matters a lot!
You double your chance of winning by
switching.
Odds of winning by not switching: 1-in-3
Odds of winning by switching: 2-in-3
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 10 of 42
A Tougher and More Serious Situation
(described by Gerd Gigerenzer in the book Calculated Risks.)




One in every one-hundred 40-year old women has
breast cancer
A mammogram detects 90% of all breast cancers
and gives a false positive 9% of the time
A 40-year old women gets a positive mammogram.
What is the probability she has breast cancer?
A sample of doctors believed it was very high (8090%)
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 11 of 42
10,000 40-year old women
100 with BC
90 have
+ m/m
9,900 without BC
10 have
– m/m
891 have
+ m/m
Actual Probability =
9 September, 2009
9,909 have
– m/m
90
 0.092
90 + 891
Don M. Chance, Ph.D., CFA
p. 12 of 42
Using Misleading Risk Information to
Sell a Product
A pharmaceutical company advertises that a new cholesterollowering drug reduces heart attack risk by 22%.
(Also described in Calculated Risks)
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 13 of 42
1,000 people with high
cholesterol
1,000 people with high
cholesterol
Receive new drug
32 die of
heart
attack
968 do not
die of heart
attack
Receive placebo
41 die of
heart
attack
959 do not
die of heart
attack
22% fewer heart attacks
(9/41 = 22%)
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 14 of 42
The Correct Interpretation
9 out of 1,000 people were helped by the drug.
In other words, of every 111 people who took the
drug (1,000/9 = 111), one person was helped.
Would you take or recommend a drug in which
111 people have to take it for one person to
benefit?
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 15 of 42
O. J. Simpson: What He Taught Us
About Risk
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 16 of 42
The Defense Claim
100,000 battered women in
U. S
40 murdered by their
batterer
99,960 not murdered by
their batterer
Odds that a battering led to a murder =
40/100,000 = 0.04% or 1-in-2,500
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 17 of 42
The True Picture
(not argued by the prosecution)
100,000 battered women in
U. S
40 murdered by their
batterer
5 murdered by someone
else
Odds that a battered and murdered woman was
killed by her batterer = 40/45 = 89%
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 18 of 42
The Weather

We are continuously confronted with risk
information about the weather
 What does x% probability of rain mean?
 What about hurricane prediction paths?
 Why is the weather more predictable than
most other uncertainties in life?
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 19 of 42
Political Polls
What does it mean to say that 51% favor
candidate A and 49% favor candidate B?
 What does a sample of 1,000 people out of
a population of 100 million registered voters
mean?
 Why does the public opinion of politicians
change so quickly?

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 20 of 42
Jury Duty
The American judicial system is terribly
flawed in that it prohibits use of all
information related to the risk
 “Innocent until proven guilty”

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 21 of 42
Voting





Voting is an amazingly low-risk situation, yet we
often act like it is a high-risk one
Elected officials are typically elected by a fairly
uninformed populace
They constantly clamor for an even more
uninformed populace to vote
Which is better? Large turnout of uninformed
voters or small turnout of informed voters
Ballot measures
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 22 of 42
Mental Health and Emotional
Instability
Identifying serial killers in advance
 Examples, true and false

 Virginia Tech shooting
 Imams on a plane
 Suspicious people
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 23 of 42
Predictors

Predictors are notoriously pessimistic






Airlines
Restaurants
Hospitals
Why?
And what about the weather forecaster?
What about economists?
 Granville, Kaufman, Roubini


Politicians running for office against an incumbent
Predictors are almost useless, not because they
cannot predict but because they engage in selfprotecting bias
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 24 of 42
Evaluating Predictors
We too often succumb to the small
sample fallacy
 It takes an incredibly large sample to
adequately evaluate predictive ability

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 25 of 42
What does it Mean to Make a
Mistake in Making a Decision
Under Risk?
Given the choices A and B, you choose
A
 Assume that A works out poorly and it is
apparent that B would have worked out
better
 Was that a mistake?

 Maybe
 Maybe not, but perhaps a lesson can be
learned (e.g., WMD in Iraq)
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 26 of 42
Common Risk Mistakes
Confusing ex post with ex ante
 Inability to distinguish luck from skill or lack
of it
 Fearing risk less the further back bad
outcomes occur
 Overweighting highly improbable events and
underweighting highly probable events

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 27 of 42
Common Risk Mistakes
(continued)
Trusting government for risk information
 Being irrational about risk
 Assuming that if a “bad” outcome occurs, it
would have been better to have made a
different decision
 Does the reward justify the risk?
 Is the cost of risk reduction worth it?

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 28 of 42
Common Risk Mistakes
(continued)
Expecting some decision makers to be
100% correct
 Paying too much attention to predictions

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 29 of 42
The Loneliness of the Decision
Maker





Sports
Killing a product: the Coca-Cola case
Decisions made by doctors
Presidents and leaders of countries
In understanding the decision maker, we
need to appreciate the Heisenberg
Uncertainty Principal (of social science)
 Obama: “I was right about the war”
 McCain: “I was right about the surge”

Changing one’s mind (flip-flopping)
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 30 of 42
Risk Information – So Misleading
Violence in Iraq? Violence in the U. S.
 A new source of energy
 School violence
 Which is riskier?

 a house with a gun or a house with a
swimming pool
 Lawn mowers or beds
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 31 of 42
How We (All too Often) Cope with
Risk
Litigation (mostly in the U. S.)
 Making someone else take the risk (as
in sports)
 Taking the safe route
 Relying too much on others, such as the
government, to make our decisions
 Overstating the amount of information
you have

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 32 of 42
Using Risk Ignorance to our
Advantage
A personal story from my teen years
 D-Day: 6 June, 1944

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 33 of 42
Viewing the Risk as Though it
were Someone Else’s

Torture
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 34 of 42
Making a Tough Decision Under
Risk: The Prisoner’s Dilemma
Two people are arrested and charged with
the same crime
 The suspects are put into different rooms
and separately interrogated
 Each is given an offer

 Testify against the other and receive immunity
(cooperate)
 If only one cooperates, that person is free and the other
is punished
 If both cooperate, each is given a lighter sentence
 If neither cooperates, both will be punished
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 35 of 42
The Prisoner’s Dilemma
Prisoner B
Prisoner A
Cooperate
Do not Cooperate
Cooperate
A: light punishment
B: light punishment
A: freedom
B: severe punishment
Do not Cooperate
A: severe punishment
B: freedom
A: punishment
B: punishment
Assuming that each prisoner has no idea what the
other will do, each should independently decide to
cooperate. Why?
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 36 of 42
How do We Manage Risk?
Panic
 Forecast
 Decide on how much risk we want, how
much risk we have, the cost of changing
the level of risk, and whether an
adjustment is economically justified

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 37 of 42
Making Good Decisions under
Risk
Some times we have too many choices
 Get as much information as possible
 Decide which risks are worth taking and
which are not
 For risks not worth taking, evaluate the
cost of eliminating the risk and do so if
supported by cost-benefit analysis
 Be moderately pessimistic

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 38 of 42
What Should We do After the
Fact?
Do not assume that a decision that
turned out adversely was a bad risk
management decision
 A correct risk management decision was
one that would be made again if the
circumstances were identical
 Do not evaluate the decisions of others
without putting ourselves in their position

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 39 of 42
What Does This Mean for Investment
Managers
Use probability and risk quantification
but be sure you use it carefully
 Respect the past but don’t obsess over
it
 The financial media knows no more than
anyone else (and oftentimes a lot less)
 Investment decisions that look bad after
the fact can actually be good decisions

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 40 of 42
Risk is Good
Do you really want to know the future?
 What would a financial world be like if
there were no risk?

9 September, 2009
Don M. Chance, Ph.D., CFA
p. 41 of 42
Thank you for having me here to speak.
If you have any questions later:
dchance@lsu.edu
Feel free to email me for a copy of this
presentation.
9 September, 2009
Don M. Chance, Ph.D., CFA
p. 42 of 42
Download