Uncertainty

Lecture 9

Reading: Perlo¤ Chapter 17

August 2015

1 / 68
Introduction

Everything is uncertain.
We don’t know what state of nature we will be in tomorrow.
We looked at consumer choice over consumption bundles.
Now we will look at consumer choice over a variety of risky
alternatives.

2 / 68
Introduction

How much insurance should I buy?

What is the optimal portfolio given my attitude towards risk?
Should I cross the street when the light is red?

3 / 68
Outline

Probability Overview - How can we measure risk?

Decision making under uncertainty - The decisions people make
depend on their attitudes towards risk.
Avoiding risk - People can take steps to avoid risk altogether.
Behavioral economics of risk - Do people behave like we have just
described?

4 / 68
Probability Overview

We don’t know what state of nature will be in tomorrow.

But we might know the relative likelihood of di¤erent states of nature.
Tomorrow is a roll of the dice, it is a random outcome.

5 / 68
Probability Overview

The gender of the next person you meet, your grade on an exam, the
number of times your computer crashes all have an element of
randomness.
An outcome is a mutually exclusive result of the random process.
Your computer might crash once, it might never crash... these are
outcomes.

6 / 68
Probability Overview

The probability of an outcome is the proportion of time that

outcome occurs in the long run
The probability of rolling a 4 is 16 . If we rolled a dice 1,000,000,000
times, we would get 4 16 of the time
We could also express it as the probability of event X is

all events favorable to X

Pr[X ] =
all possible events

7 / 68
Probability Overview

A random variable is a numerical summary of a random outcome.

If my computer crashes the variable takes a 1, if not it takes a 0 for
example.
The roll you get from a dice is a random variable.

8 / 68
Probability Overview

A discrete random variable takes on only a discrete set of values,

like 0, 1, 2, 3.
A continuous random variable takes on a continuum of possible
values (any point on some measure of the real line).
Roll of the dice is a discrete random variable, the amount of rain is a
continuous random variable.

9 / 68
Probability Overview

A probability distribution is the list of all possible values of the

random variable and the probability they will occur.
A cumulative probability distribution is the probability that the
random variable is less than or equal to some value.

10 / 68
Probability Overview

EXAMPLE
What is the probability distribution and cumulative probability
distribution for a dice role?

11 / 68
Probability Overview

12 / 68
Probability Overview

The expected value of a random variable Y is the long run average

value the random variable will take over many repeated trials
It is the weighted average of all possible values (also called
expectation or mean)
If Y is discrete
∞
E [Y ] = ∑ pi xi
i =1

If Y is continuous
Z∞
E [Y ] = xf (x )dx
∞

13 / 68
Probability Overview

Lets say X is the roll of a die, which has 6 possible outcomes all
equally likely

1 1 1 1 1 1
E [X ] = 1 + 2 + 3 + 4 + 5 + 6 = 3.5
6 6 6 6 6 6
If you were to roll a fair die 1,000,000,000,000 times, on average you
would get a value of 3.5

14 / 68
Probability Overview

EXAMPLE
I give you £ 100 if you ‡ip a heads and £ 0 if you ‡ip a tails. What is
your expected payo¤ from this game?

15 / 68
Probability Overview
The variance of some random variable X is a measure of how spread
out the distribution is. If the variance is high You are less likely to
get a value close to the mean

n
Var [X ] = ∑ pi (xi E [X ])2
i =1
Var [X ] = E [X 2 ] (E [X ])2

Or if continuous
Z
Var [X ] = (x E [X ])2 f (x )dx

The standard deviation of

pX is just the square root of the
variance S.D.[X ] = σx = σ2x
16 / 68
Probability Overview

17 / 68
Decision Making Under Uncertainty

Remember in chapter 3 we learned about preferences over bundles of

goods
People now have preferences over lotteries
People will pick the lottery that gives them the highest utility.
You don’t know exactly what utility you will get, however.
People maximize their expected utilities
The expected utility of some action is the probability weighted sum
of utilities you get associated with each possible outcome.

18 / 68
Decision Making Under Uncertainty

Suppose with you the following lottery.

With probability 12 , you win £ 10 and with probability 1
2 you win £ 1
Your utility function is U = 10 w where w is wealth.

19 / 68
Decision Making Under Uncertainty

Your expected utility is

EU = Pr[heads ] U (heads ) + Pr[tails ] U (tails )

1 1
EU = 10(10) + 10(1) = 55
2 2

20 / 68
Decision Making Under Uncertainty

EXAMPLE
Suppose you have a dice roll and you win the value of the dice roll (if
you roll a 6 you get $ 6).
Write down an equation for expected utility of an individual who has
a utility function U = ln w (don’t actually calculate the number).

21 / 68
Decision Making Under Uncertainty

Two people might face the same lottery, but they will get a di¤erent
expected utility.
I buy lottery tickets, many people don’t.
The shape of your utility function determines your attitude over risk.

22 / 68
Decision Making Under Uncertainty

A fair bet is a wager with an expected value of 0.

You pay $ 10 to play the lottery and the expected value of the ticket
is $ 10. This is a fair bet. On expectation you do not lose money.

23 / 68
Decision Making Under Uncertainty

EXAMPLE
You win the value of the dice roll (roll a 6 get $ 6 for example)
If you have to pay £ 4 to play this game, is that a fair bet?

24 / 68
Decision Making Under Uncertainty

Suppose I o¤er you a fair bet.

People who are risk neutral are indi¤erent between taking this bet or
not
People who are risk loving will always take this bet
People who are risk averse will not.

25 / 68
Decision Making Under Uncertainty

Let’s look at the risk averse people …rst.

If somebody is risk averse, they have a diminishing marginal utility to
wealth.
Their utility function is concave.
The extra utility they get from one more dollar is less than they got
from the last.

26 / 68
Decision Making Under Uncertainty

1
Suppose somebody’s utility function is U = W 2 .
This person can play the following game.
You win $2 with probability 21 or you win $ 0 with probability 1
2
The expected value of this game is $1

27 / 68
Decision Making Under Uncertainty

The person’s expected utility from this lottery is

1 1
EU = .5 (2) 2 + .5 (0) 2 = .7
1
The person’s utility of having the $1 for sure is (1) 2 = 1.
This person gets a higher utility from $1 for sure than a bet whose
expected payo¤ is $1.
This person is risk averse.

28 / 68
Decision Making Under Uncertainty

The risk premium is the amount that a risk averse person will pay to
avoid taking a risk.
In the previous example, we know the lottery gives us an expected
utility of .7.
To …nd the risk premium, we need to …nd the amount of money we
would be willing to give up to eliminate risk altogether.
The lottery in this case has an expected value of $1, how much of
that $1 would I give up?

29 / 68
Decision Making Under Uncertainty

1
(X ) 2 = .7
X = .49

I am indi¤erent between playing the lottery with an expected value of

$1 and having $0.49 cents for sure.
The risk premium is $0.51. I would give up $0.51 to avoid risk
altogether.

30 / 68
Decision Making Under Uncertainty

EXAMPLE
1
With probability 4 the car turns out to be bad and is worth $400.
3
With probability 4 the car turns out to be good and is worth $1600.
What is the expected payo¤ from buying this car?
1
If this person has a utility function U = W 2 , what is the expected
utility of buying this car?
What is the risk premium? What is the most this person would be
willing to pay for this car?

31 / 68
Decision Making Under Uncertainty

This is what a risk-averse person’s utility function looks like.

This person would give up 14 dollars to have no risk at all.

32 / 68
Decision Making Under Uncertainty

Now let’s look at a risk neutral person.

A risk-neutral person has a constant marginal utility over wealth, her
utility function is a straight line.
One extra dollar is worth just as much to her no matter how much
money she has.

33 / 68
Decision Making Under Uncertainty

The utility she gets from the expected value of the bet is exactly
equal to the expected utility.
A risk neutral person will take the bet with the highest expected value.

34 / 68
Decision Making Under Uncertainty

Suppose somebody has a utility function U = W .

This person can play the following game.
You win $2 with probability 21 or you win $ 0 with probability 1
2
The expected value of this game is $1

35 / 68
Decision Making Under Uncertainty

Will this person take the bet?

The expected utility of the bet is
1 1
2+ 0 = 1
2 2
The utility of having $1 for sure is exactly the same.
The presence of risk does nothing.

36 / 68
Decision Making Under Uncertainty

Now let’s look at a risk loving person.

A risk loving person has an increasing marginal utility of wealth, they
will always take a fair bet.
Their utility function is convex.
This person has a negative risk premium.
The expected utility of some bet is higher than the utility she gets
from the utility of the expected value.

37 / 68
Decision Making Under Uncertainty

38 / 68
Decision Making Under Uncertainty

To summarize
Somebody is risk averse if U (E [X ]) > E [U (X )]
Somebody is risk averse if U (E [X ]) = E [U (X )]
Somebody is risk loving if U (E [X ]) < E [U (X )]

39 / 68
Decision Making Under Uncertainty

EXAMPLE
If you ‡ip a heads I give you £ 10, if you ‡ip a tails I give you £ 5.
What is the most each of the following people would be willing to pay
for this game?
1
1 U = (W ) 2
2 U=W
3 U = (W )2

Draw each person’s utility function showing the expected value,

expected utility and risk premia.

40 / 68
Decision Making Under Uncertainty

The curvature of the utility function is what determines somebody’s

risk aversion
One commonly used measure of this curvature is the Arrow-Pratt
measure of risk aversion

d 2 U (W )/dW 2
ρ (W ) =
dU (W )/d (W )

41 / 68
Decision Making Under Uncertainty

The numerator of this expression is the second derivative.

If utility is concave, this will be negative
ρ is positive if somebody is risk averse and it is negative if they are
risk loving.

42 / 68
Decision Making Under Uncertainty

EXAMPLE
What is the Arrow-Pratt measure of risk aversion for the following
utility functions?

1
U = W2
U = 2W
U = W2

43 / 68
Avoiding Risk

There are a lot of ways people can avoid or decrease risk

One such way is through diversi…cation.
Diversi…cation simply means putting your eggs into di¤erent baskets.
The probability that all of the baskets break at the same time is low.

44 / 68
Avoiding Risk

Another way to deal with risk is through insurance.

Insurance is just contingent consumption.
You can move some of your payo¤ from the good state of the world
to the bad state of the world.
Make the bad outcome less bad.

45 / 68
Avoiding Risk

If your house doesn’t burn you keep £ 1,000,000

If your house burns down you get $ 0
Insurance would be something like buying a plan that gives you
£ 900,000 if your house doesn’t burn down and £ 100,000 if
your house does burn down.

46 / 68
Avoiding Risk

Suppose that for every $1 of insurance you buy, you get $4 if your
house burns down.
The probability your house burning down is .1 and the probability it
doesn’t is .9.
By being able to purchase insurance, your expected utility changes
from the …rst expression to the second expression.

EU = 0.9U (1, 000, 000) + 0.1(0)

EU = 0.9 U (1, 000, 000 x ) + 0.1 U (4x x)

To …nd out exactly how much insurance you would by, you would
maximize this with respect to x.

47 / 68
Avoiding Risk

If our utility function were U = ln W

EU = 0.9 ln(1, 000, 000 x ) + 0.1 ln(4x x)

maximise this with respect to x and solve for x.

48 / 68
Avoiding Risk

dEU 0.9 0.1

= + =0
dx x 1, 000, 000 x
0.1 0.9
=
x 1, 000, 000 x
0.9x = 100, 000 0.1x
x = 100, 000

49 / 68
Avoiding Risk

EXAMPLE
You can insure your bike worth $500 against theft.
The probability your bike gets stolen is 12 .
For every $1 of insurance you buy you get $2 if your bike is stolen.
If your utility function is U = W 1/2 , how much insurance should you
buy?
What is the intuition behind this result?

50 / 68
Avoiding Risk

In the previous example, insurance was actuarially fair.

A policy is actuarially fair if the insurance company does not make
any pro…t.
Expected pro…t for the insurance company in the previous example for
every $1 of insurance bought is
1 1
E (π ) = ($1) ($1 $2) = 0
2 2

51 / 68
Avoiding Risk

EXAMPLE
Your house burns down with probability 1%.
For every £ 1 of insurance you buy the insurance company gives you
£ 50 back.
Is this actuarially fair?

52 / 68
Behavioral Economics of Risk

Is this how people actually behave towards risk though?

Many people make decisions that are inconsistent with expected
utility theory.

53 / 68
Behavioral Economics of Risk

One common fallacy is the gambler’s fallacy

People believe that past events a¤ect current, independent outcomes.
If you ‡ip a fair coin 100 times and they all come up heads, the
probability the next coin is heads is still .5.
But many people believe that tails is more likely because it is "due."

54 / 68
Behavioral Economics of Risk

People often overestimate their probability of winning.

In one survey, gamblers estimated their chance of winning a bet is
45% when the objective probability was 20%.

55 / 68
Behavioral Economics of Risk

Many people put excessive weight on outcomes they consider to be

certain.
A group of subjects were asked to choose between two options in an
experiment.

56 / 68
Behavioral Economics of Risk

Option A: You receive $4,000 with probability .8 and $0 with probability

.2
Option B: You receive $3,000 with certainty
80% of people choose option B

57 / 68
Behavioral Economics of Risk

They were then given a separate set of options

Option C: You receive $4,000 with probability .2 and $0 with probability
.8
Option D: You receive $3,000 with probability .25 and $0 with
probability .75
65% of people choose C

58 / 68
Behavioral Economics of Risk

These choices are inconsistent

Choosing B over A implies the expected utility of B is greater than A
U (3, 000) > .08U (4000)
U (3,000 )
Or U (4000 )
> .08

59 / 68
Behavioral Economics of Risk

Choosing C over D implies that .2U (4000) > .25U (3, 000)
U (3,000 )
This implies U (4000 )
< .08
This is inconsistent with expected utility theory

60 / 68
Behavioral Economics of Risk

The way you present the same lottery can change people’s
preferences.
The Avian Flu will kill 600 people, the government is proposing two
programs to combat it
Program A: 200 people will be saved
Program B: There is a 1/3 probability that 600 people will be saved and
a 2/3 probability nobody will be saved.
72% opted for program A

61 / 68
Behavioral Economics of Risk

The Avian Flu will kill 600 people, the government is proposing two
programs to combat it
Program C: 400 people will die
Program D: There is a 1/3 probability that nobody dies and a 2/3
probability that 600 people will die.
78% opted for program D

62 / 68
Behavioral Economics of Risk

A and C are the same and B and D are the same!

Framing is a huge issue in economics.

63 / 68
Behavioral Economics of Risk

Prospect theory is an alternative approach to decision making under

uncertainty.
Empirically, it is a better match of people’s behaviors.
People treat gains and losses di¤erently in prospect theory.
Small changes are evaluated di¤erently than big changes.

64 / 68
Behavioral Economics of Risk

The basic tenants of prospect theory are

1 Everything is evaluated according to some reference point.

2 People are more sensitive to small gains than to large gains. For
example the di¤erence between $1 and $2 is larger than $1001 and
$1002.
3 People hate taking losses more than they like taking gains.

65 / 68
Behavioral Economics of Risk

66 / 68
Summary

What is the expected value of a random variable?

What is expected utility?
What are the di¤erent attitudes towards risk and who will take a fair
bet?
What is the risk premium?

67 / 68
Summary

What is fair insurance?

What are inconsistencies with expected utility theory in the real world?
What is prospect theory?

68 / 68