You are on page 1of 57

BEPP 305/805:

Risk Management, Lecture 1


Professor Jeremy Tobacman
January 16, 2014
Goals
Individuals and firms face risks in nearly all
decisions that they make.

Provide an introduction to decision making in a
world with uncertainty.

How should individuals, and managers of firms, make
decisions involving risk?

What are the typical mistakes made in decisions
involving risk?

2
Why study risk management?
As an individual, you face risks in many aspects
of your life.

Managers of firms make many decisions that
involve risks, and the consequences can be large.

A lesson from the recent financial crisis: the
failure to properly manage risk can result in
disaster.
3
Lessons from the financial crisis

The crisis spurred a remarkable degree of
reflection and activity throughout the community.
The unifying theme is a focus on risk
management: the risks of a particular product or
financial service, the risks to a firm, and the
systemic risks to society as a whole.

- Retiring HBS Dean Jay Light on the recent
developments in the curriculum
4

I believe that a CEO must not delegate risk
control. Its simply too important

Warren Buffet

5

"Named must your fear be before banish it you
can.

Yoda

6
Structure of the course

1. Optimal decision making under risk (Tobacman)

2. Barriers to risk management (Wang)

3. Corporate risk management (Nini)
7
Module I in one slide

Why is it important to account for risks?

How is risk measured in practice?

What is the optimal way to make decisions
under risk?
8
Module II in one slide

Barriers to risk management

Market impediments
Information and incentive problems

Psychological impediments
People dont always behave optimally
9
Module III in one slide

Corporate risk management

When firms SHOULD NOT manage risk
When firms SHOULD manage risk
Strategies for corporate RM
Managing liability risk
10
Overview of the syllabus

Course structure and requirements
Prerequisites
Course grading
Policies for dropping/withdrawing
Expectations
Policies for exams
11
Grading

Three exams, one for each module.

Problem sets, posted on Canvas
Work in teams but write your own solutions
Graded on a complete/incomplete system
You can skip turning in one problem set with no penalty
Module I due dates: 1/24, 1/31, 2/7 at 5:00pm

Survey questions will also be posted on Canvas

Problem sets and survey answers are worth 10% of your
grade
12
Slides and notes

Slides will be posted on Canvas

Notes summarizing certain aspects of the course
material will be posted on Canvas periodically,
generally after the material is covered in class
13
One slide study guide

Primary resources:
Lectures
Notes posted to Canvas
Problem sets

Readings are intended to be references
14
About me
Assistant Professor in BEPP since 2008

Ph.D. in Economics from Harvard

Research on household finance for the poor
Consumer credit in the US
Microinsurance against rainfall risk in India
Behavioral economics
15
My info

Office: 1409 SH-DH
Email: tobacman@wharton.upenn.edu
Office hours: Tuesdays 4:30-5:30pm or by appt

16
TAs for the course
Banruo (Rock) Zhou
banruo@wharton.upenn.edu
Ella Zhang
zhq@sas.upenn.edu
Neil Iyer
neiliyer@wharton.upenn.edu

17
Practice Sessions
Neil (1/21 & 2/4) - 4:30pm
Rock (1/21 & 2/4) - 7:30pm
Ella (1/22 & 2/5) - 4:30pm

Attend the most convenient one
Optional but awesome
Rooms TBA
18
Probability Theory
Rest of the lecture

1. Define what we mean by risk
2. Build up concepts of probability theory
3. Some methods for measuring risk
a. Variance as a measure of risk
b. Value at Risk
c. Mean-variance criterion


20
An example
A person retires at age 70, with a total of $1
million

She expects to live for another 25 years

How much can this person consume per year?
21
An example
A person retires at age 70, with a total of $1
million

She expects to live for another 25 years

How much can this person consume per year?

Assume a real interest rate of 2% per year:
Approximately $50.22k per year
22
Dollars remaining (in thousands)
23
But, there is uncertainty!
What if the person lives longer than 25 years?

What if the interest rate falls?

Calculations based on averages can be
misleading
Need to account for risk
24
Another example
A manager wants to estimate inventory costs for
the business, based on inventory amount.
If demand is lower than inventory: Unsold units
spoil, entailing a $50 cost per unit.
If demand exceeds inventory: Extra units must be
air-freighted in, at a cost of $150 per unit.

Monthly demand is, on average, 5,000 units per
month.





25
Understanding probabilities is crucial
Given average monthly sales of 5,000, what are
expected inventory costs if the manager decides
to have monthly inventory of 5,000?

Zero?

More than Zero?

Cannot be determined?

Source of this example: Harvard Business Review

26
Understanding probabilities is crucial
Expected inventory costs are greater than zero, if
there is any variation in demand from month to
month.

Using averages can be very misleading!

The appropriate method is to consider the whole
probability distribution for demand, not just the
average of the distribution.

27


The flaw of averages
28
One of many other examples
In 1997, the U.S. Weather Service forecast that
North Dakotas rising Red River would crest at
49 feet.
Official in Grand Forks made flood management
plans using this single number, an average.


29
One of many other examples
In 1997, the U.S. Weather Service forecast that
North Dakotas rising Red River would crest at
49 feet.
Official in Grand Forks made flood management
plans using this single number, an average.

The river crested above 50 feet, breaching the
dikes.
50,000 people were forced from their homes,
and there was $2 billion in property damage.

30
What is risk?
Very broadly, risk involves uncertainty
Many possible outcomes

Most decisions involve some degree of
uncertainty
31
Examples of risk
Individuals
Labor income, mortality, injuries, asset returns

Firms
Input costs, borrowing costs, demand, regulation

Governments
Unemployment, social security costs, business
cycles, wars, commodity prices

32
How can we model risk?
Answer: Probability theory

Provides us a way to think about what the most
likely outcome is

and gives us a way to model the range of
possible outcomes
33
Some concepts
Sample space
Set of all possible things that can happen

Probability distribution
Relative chance that each state can occur

Random variable
Function that assigns outcomes to each state
34
A simple example: a coin flip
Sample space
{H,T}

Probability distribution
{, }

Random variable, some examples
X = Number of heads
X(H)=1. X(T) = 0
Y = Number of tails
Y(H)=0, Y(T) = 1

35
Another example: two coin flips
Sample space
{HH, HT, TH, TT}

Probability distribution
{, , , }

Random variables
X = number of heads
X(HH) =2, X(HT) = 1, X(TH) = 1, X(TT) = 0
Y = proportion of heads
Y(HH) = 1, Y(HT)= , Y(TH) = , Y(TT) = 0
36
Probability distribution
Outcomes x
1
x
2
x
3
x
4
x
5
x
6

P
r
o
b
a
b
i
l
i
t
y

p
1

p
2

p
3

p
4

p
5

p
6

37
Properties of random variables

Expected value
Measure of the central tendency

Variance and standard deviation
Measures of dispersion

38
Expected value (mean)

Weighted average of outcomes


E X | | = p
1
x
1
+ p
2
x
2
+...+ p
n
x
n

=
=
n
i
i
x
i
p
1
39
Variance

Expected squared deviation from the mean


( ) ( ) | | ( ) | | ( ) | |
( ) | | { }
2
2 2 2
...
2 2 1 1
X E X E
X E
n
x
n
p X E x p X E x p X Var
=
+ + + =
40
Standard deviation

Square root of the variance




Same units as X
SD X ( ) = Var X ( )
41
Variance and SD as measures of risk

Var and SD measure the expected dispersion
between outcomes and the average outcome

Higher when
Outcomes can deviate a lot from expected value
Probability of extreme deviations is high

Lets think about whether these are good
measures of risk
42
Example
Investment A
$0 with probability 2/3
$9M with probability 1/3

Investment B
$-5M with probability 0.2
$5M with probability 0.8

Which is the riskier investment?
43
Example
Profit from Investment A
$0 with probability 2/3
$9M with probability 1/3

Profit from Investment B
$-5M with probability 0.2
$5M with probability 0.8

Mean = 3M
Variance = 18
2




Mean = 3M
Variance = 16
2

44
Asymmetry
Var and SD potentially measure risk, but they
miss something:

Large losses are more risky than large gains

Extreme example
Worst case scenario
45
Another way to quantify risk

Value at Risk (VaR)

Question:
What is the minimum loss under
exceptionally bad outcomes?
46
Value at Risk (VaR)
Minimum loss in the bottom p% of outcomes
Focus on the left tail of the distribution
Usually 1% or 5% for a given time interval


P
r
o
b
a
b
i
l
i
t
y

1%
2%
3%
5%
VaR at 10% is -4
7.5%
VaR at 1% is -10

VaR at 5% is -6
Profit -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20
47
Example
Profit from Investment A:
-$1M with probability 0.1%
$0 with probability 49.9%
$2 with probability 50%

Profit from Investment B:
-$1 with probability 20%
$1 with probability 80%

Which investment is riskier?
48
Example
Profit from Investment A:
-$1M with probability 0.1%
$0 with probability 49.9%
$2 with probability 50%

Profit from Investment B:
-$1 with probability 20%
$1 with probability 80%

Which investment is riskier?
VaR at 1% = $0
VaR at 5% = $0
VaR at 10%= $0


VaR at 1% = -$1
VaR at 5% = -$1
VaR at 10%= -$1
49
Mean-Variance Criterion
Balancing expected tendency and variance

aE(X)-bVar(X)
a>0, b>0

An investment in X might be preferred to Y if:
a[E(X)]-b[Var(X)] > a[E(Y)]-b[Var(Y)]
What does this say?
50
Mean-Variance Criterion
Basis of Markowitzs (1950) portfolio theory
1990 Nobel Prize
Often used in practical applications

Prior to Markowitz, portfolios were chosen on
the basis of E(X) alone, without regard for
Var(X)!

We will study the properties of the mean-
variance criterion later in the course
51
Review of concepts: An example
Random variable: damages from an automobile
accident

Possible Outcomes for Damages Probability
$0 0.50
$200 0.30
$1,000 0.10
$5,000 0.06
$10,000 0.04

52
Expected value
Possible Outcomes for Damages Probability
$0 0.50
$200 0.30
$1,000 0.10
$5,000 0.06
$10,000 0.04

EV = .5(0) + .3(200) + .1(1,000)
+ .06(5,000) + .04(10,000)
= $860

53
Variance
Possible Outcomes for Damages Probability
$0 0.50
$200 0.30
$1,000 0.10
$5,000 0.06
$10,000 0.04

Variance = .5(0-860)
2
+ .3(200-860)
2
+ .1(1,000-860)
2
+ .06(5,000-860)
2
+ .04(10,000-860)
2

= 4,872,400 ($
2
)

54
Standard deviation
Possible Outcomes for Damages Probability
$0 0.50
$200 0.30
$1,000 0.10
$5,000 0.06
$10,000 0.04

SD = (Variance)
1/2
= (4,872,400)
1/2
= 2,207 ($)

55
Practical concerns
Where do these probabilities come from?

We need a way to translate past observations
into probabilities
Statistics
56
Summary of todays class
Inference based on samples averages can be
quite misleading
Need to account for risk

Probability theory allows us to model risks
A measure of a typical observation (mean)
Measures of expected dispersion (variance and SD)
(Imperfect) measures of risk
57