ECONOMICS

CLASS NOTES

Gregory Lypny
 CONTENTS

About These Notes v

Optional Textbook Reading .......................................................................................v

I. ECONOMICS
1 BUILDING ECONOMIC MODELS ...................................................................................... 2
Reading ......................................................................................................................5
2 UTILITY THEORY .............................................................................................................6
Income, Wealth, and Budgets ....................................................................................7
Utility Functions and Indifference Curves .................................................................9
Optimal Consumption .............................................................................................11
Reading ....................................................................................................................13
3 EQUILIBRIUM ..................................................................................................................14
The First Welfare Theorem of Economics .............................................................. 16
Reading ....................................................................................................................18
4 INFORMATION ................................................................................................................ 19
Rational Expectations .............................................................................................. 20
Take a Random Walk...............................................................................................20
Bubbles and Crashes ................................................................................................22
Is Private Information Reflected in Prices? .............................................................. 23
Reading ....................................................................................................................25
5 PRODUCTION ..................................................................................................................26
Reading ....................................................................................................................29
6 GOVERNMENT INTERVENTION ...................................................................................... 30
Taxing and Subsidizing Business Loans ...................................................................30
Subsidized Daycare ..................................................................................................33
Reading ....................................................................................................................35
7 MARKET FAILURE ..........................................................................................................36
The Lemons Problem ...............................................................................................37
Moral Hazard...........................................................................................................37
Externalities.............................................................................................................. 38
Public Goods ............................................................................................................38
Reading ....................................................................................................................40

!ii
 8 INTERNATIONAL MARKETS............................................................................................ 41
Gains from Specialization and Trade ...................................................................... 41
Exchange Rates ........................................................................................................43
Reading ....................................................................................................................47
9 INTERTEMPORAL MARKETS ..........................................................................................48
Bonds ........................................................................................................................48
The Term Structure of Interest ...............................................................................50
Optimal Consumption .............................................................................................53
Real Investment ........................................................................................................55
Equilibrium Interest Rate ........................................................................................57
Permanent Income Hypothesis ................................................................................60
Reading ....................................................................................................................63
10 MARKETS FOR RISK.......................................................................................................64
States and Dates .......................................................................................................64
Description of the Economy ....................................................................................66
Pareto Optimum ...................................................................................................... 67
Portfolios and Trade .................................................................................................68
Reading ....................................................................................................................70

II. FINANCIAL ECONOMICS

11 PORTFOLIO THEORY...................................................................................................... 72
Portfolio Return........................................................................................................73
Portfolio Risk ............................................................................................................74
Portfolios of Two Risky Assets .................................................................................76
Portfolios of More Than Two Risky Assets .............................................................79
Three Ways of Looking at Diversification ...............................................................84
Adding a Risk-Free Asset to Portfolios .....................................................................87
Choosing a Portfolio .................................................................................................91
Reading ....................................................................................................................92
Homework ................................................................................................................93
Answers ....................................................................................................................95
12 CHOICE UNDER CERTAINTY........................................................................................102
Optimal Consumption ...........................................................................................102
Demand for Consumption .....................................................................................104
Real investment ......................................................................................................105
The Equilibrium Interest Rate ...............................................................................108
Reading ..................................................................................................................110
Homework ..............................................................................................................111
Answers ..................................................................................................................113

!iii
 13 CHOICE UNDER UNCERTAINTY................................................................................... 118
Expected Utility Theory ........................................................................................118
Working with Utility Functions Instead of Indifference Curves ............................ 121
Coefficient of Risk Aversion ..................................................................................124
Pricing a Risky Asset ..............................................................................................125
Reading ..................................................................................................................127
Homework ..............................................................................................................128
Answers ..................................................................................................................130
14 EQUILIBRIUM IN CAPITAL MARKETS .......................................................................... 131
Time-State Preference Model ................................................................................131
The Capital Asset Pricing Model ...........................................................................138
Homework (coming soon) ......................................................................................144
Answers (coming soon) ...........................................................................................145
15 DERIVATIVE SECURITIES (ALMOST) .............................................................................146
Forwards and Futures .............................................................................................146
Options...................................................................................................................148

APPENDICES (COMING SOON)

1 TIME VALUE OF MONEY (COMING SOON) .................................................................. 150
Future Value ...........................................................................................................151
Compounding Interest More Than Once a Year ..................................................152
Present Value ..........................................................................................................154
Present Value of a Cash Flow Series......................................................................157
Present Value of an Annuity ..................................................................................158
Present Value of a Growing Annuity .....................................................................159
Examples ................................................................................................................160
2 RATES OF RETURN (YOU TELL ME).............................................................................. 161
3 MATRIX ALGEBRA (NOT YET) ...................................................................................... 162
4 STATISTICS (I DON’T KNOW) ........................................................................................163
FILLER TEXT ........................................................................................................................164

!iv
 A BOUT T HESE N OTES

These are draft class notes for the topics that we will be discussing in Analysis of Markets and Fi-
nance Theory I. Neoclassical economics and financial economics is emphasized. We will deal with
behavioural economics in our class discussions.
If you intend to print these notes, I suggest printing one topic at a time because I will be posting
updates during my writing of them.

Gregory Lypny
Friday, January 5, 2018

Optional Textbook Reading

Some of you may want to consult a textbook from time to time. One or more of the following
books may appear in the Readings section that is at the end of each chapter. The cited textbook
chapter numbers are approximate and depend on the edition of the book you consult. The topic
you are looking up, however, will take you to the right place.
Hal Varian, 2014, Intermediate Microeconomics: A Modern Approach, W.W. Norton and Company. New
York. ISBN 0-393-97830-3. Any edition is fine. Referred to as Varian. [COMM 220 and
FINA 385]
Bodie, Z.; A. Kane; A.J. Marcus; S. Perrakis; and P.J. Ryan. Investments. McGraw-Hill Ryerson,
Toronto. Any edition is fine. Referred to as Bodie et. al. [FINA 385]
Fabozzi, Frank, J.; Edwin H. Neave, and Guofu Zhou, Financial Economics, John Wiley & Sons, Inc.
Hoboken, New Jersey. Referred to as Fabozzi et. al. Any edition is fine. [FINA 385]

!v
!vi
 I. E CONOMICS

This part covers introductory microeconomics for courses such as Analysis of Markets
(COMM 220), and serves as a review for intermediate courses in financial economics such as Fi-
nance Theory I (FINA 385).

!1
 1
BUILDING ECONOMIC MODELS

Twelve students take part in an experiment. Each is randomly assigned to be a buyer or a

seller of an otherwise worthless metal token. You couldn’t even use for a body piercing. Buyers are
told that if they buy a token, they can turn it in at the end of the experiment and be paid its re-
demption value. The redemption value is different for each buyer, and is known only to the buyer.
Sellers are given one token each, called an endowment, and told that if they sell their token, they
can keep the sale price less the token’s cost. The cost is different for each seller, and is known only
to the seller.

Buyer Redemption Value Seller Cost

B3 18 S2 6

B2 17 S4 8

B5 15 S5 11

B1 14 S3 12

B4 11 S1 16

B6 8 S6 19

The buyers and sellers trade by submitting their offers privately to Raymond, a skinny,
bearded guy wearing Buddy Holly glasses and a Keep on Truckin’ t-shirt that he bought in 1968.
Raymond sorts the buyers’ bids from highest to lowest and the sellers’ asks from lowest to highest,
and if there is a price for which the number of tokens that would be bought is equal to the num-
ber that would be sold, Raymond will call out that price as the equilibrium or market-clearing price.
This type of market is called a call market because the offers are batched. All buyers who bid at
least the market-clearing price will receive a token supplied by the sellers who asked no more than
the market-clearing price. What is your hypothesis for the equilibrium price and quantity in this
market?
I think that not more than four tokens will be traded at a price of about $13. It may not
happen the very first session of the experiment because the subjects may need some time to learn,
especially since redemption values and costs are private information. I do know that it would be
strange if the number of tokens traded were greater than four (can you say why?). Four tokens at
about $13 should be what we observe on average.

!2
 Raymond draws the supply and demand schedules like this. If buyers want to earn some
cash, they will bid less than their redemption values. I don’t know how much lower; too low, and
p
S6 ( $19)
B3 ( $18)
18 B2 ( $17)
S1 ( $16)
CS = $12.00 B5 ( $15)
B1 ( $14)
14 e
13 S3 ( $12)
12 S5 ( $11) B4 ( $11)

PS = $15.00
S4 ( $8) B6 ( $8)

S2 ( $6)
6

q
1 2 3 4 5 6
there is a risk of slipping to the right of the equilibrium, which is unknown to buyers and sellers,
and earning nothing. If sellers are greedy too, they will ask more than cost, but not too much
more. So, it looks like greed is necessary to put us in the neighbourhood of e.
The $13 equilibrium price needs a word or two because it is, after all, being announced by
Raymond. It isn’t the only possible equilibrium price. You can see from the graph that Buyer 1
and Seller 3, at the margin, determine the equilibrium. Buyer 1 will buy a token as long as the
price is no higher than $14, and Seller 3 will give one up as long as the price is no less than $12.
So that fourth token would trade at any price from $12 to $14, and the equilibrium price is not
unique. But Raymond’s job as auctioneer is to call out one equilibrium price, if it exists, so he has
to have a rule to deal with multiple equilibrium prices, and his rule, which he announced to
everyone at the beginning of the experiment, is to use the mid-point of the marginal bid and ask.
It didn’t have to be split down the middle; that’s just the way Raymond’s head works.
The world is a complicated place. There’s a lot going on all of the time. Economic models,
like all scientific theories, are simplified versions of some small bit of reality. A theory sets out the
conditions or variables that are necessary to answer the question being asked. What should be the
equilibrium in a call market when information about cost and redemption value are private? Simple is usually
best. Being able to explain or predict something with just two variables is better than needing
three. A theory is not intended to duplicate reality but capture enough of we want to explain or
predict. When a variable important to a theory cannot be measured because it is unobservable, an
assumption has to be made about it. What assumptions were made to arrive at the hypotheses in
the tokens experiment? Greed is one: people prefer having more to less. Another is that people
place offers without thinking that their offers could somehow influence the equilibrium. They be-
have as if they are price-takers. The tokens experiment has been done many times, and the hypoth-
esis is strongly supported. That’s a brownie point for economics because, in this simple market,
goods flow from the people who value them least (the lowest cost sellers) to those who value them

!3
most (the ones with the biggest redemption values). The value of trade to the people in this mar-
ket is reflected in the consumers’ and producers’ surpluses shown as CS and PS in the graph.
They are better off than they would be without trade.
Where the model fails in practice is its sensitivity to the good being traded. When the good
is held for personal use and not specifically for trade, far fewer units than predicted are traded,
and correspondingly, the gap in the price offers of the buyers and sellers is much wider. What if
each of 50 sellers were endowed with an I-Love-Concordia coffee mug, and everyone, including
50 buyers, knew that the mugs were selling for $13.89 in the university bookstore?

!4
Reading
Kahneman, Daniel, 2011, Thinking, Fast and Slow, Doubleday Canada, chapter 27.
Kahneman, Daniel; Jack L. Knetsch, and Richard H. Thaler, 1991, “The Endowment Effect,
Loss Aversion, and Status Quo Bias,” Journal of Economic Perspectives, 5(1):193-206.
“Behaviourists at the Gate,” The Economist, May 10, 2003.

!5
 2
UTILITY THEORY

Would you keep the I-Love-Concordia coffee mug if you had been given one as a seller in
the experiment in chapter 1. Or would you try to sell it? Would you want to buy one if you were
assigned the role as buyer? Economics is all about choices, and it assumes that we make choices
that are best for us, that give us the greatest satisfaction. The tokens version of the experiment was
designed so that greed would motivate the decision to buy or sell and the offers to be made. The
prediction of four tokens traded at around $13 followed from the assumption that people are
greedy. Assuming greed is not a big leap, but then there’s philanthropy, altruism, apathy, vindic-
tiveness, and self-destructive behaviour, among others, that do not fit so neatly in the best-for-me
mould. The assumption of greed is not enough to form a hypothesis about equilibrium price and
quantity, or whether there would any trade at all, in the coffee mug version of the experiment be-
cause everyone was told the price of the mug in the university bookstore, and everyone knew that
the price was public information. A broader assumption had to be made: people must differ in
their preference for mugs. Some sellers might offer their mug for sale because they prefer cash,
and some buyers might want a mug. We can get a rough idea about someone’s tastes and whether
they are greedy after the fact, that is, by observing the actual choices they make. But it is much
harder to do before the fact; tastes and greed are unobservable in that sense. That’s why, Utility
Theory, the economic paradigm of preferences, uses mathematical functions to represent tastes
and greed in an economically meaningful way and to avoid having to measure them directly. In
this way, choices as well as responses to changes in prices and income can then be predicted.

!6
Income, Wealth, and Budgets
Consuming the things that you have chosen is what gives you satisfaction. Call these things x
and y. x and y are flows of goods, services, or any activity that is consumed.1 Groceries per week,
haircuts per year, and time spent listening to the blues are all consumption flows; combinations of
these are sometimes called consumption bundles. Your income is a flow too. Income together with
the prices of goods defines your budget, which determines how much you can consume. Suppose
your weekly income is $100, the price of x is $5, the price of y is $2, and you don’t happen to own
y

50.0
$100 = $5x + $2y ⟺ m = px x + p y y
px m 5
y=- x+ ⟺ y = - x + 50
py py 2

a
30.0

e
12.5
b
5.0
x
8 15 18 20
any x or y. You could consume 20 units of x per week if you spent all of your income on x, 50
units of y if you spent all of it on y, or any combination of the two that doesn’t cost more than
$100.
All of the bundles whose cost equals your income lie on a budget line.2 Bundles a, b, and e
are just some of the many bundles that are affordable, and you would chose the one that you pre-
5 $5 p
ferred to all others on the line. The slope of the budget line is −2
" =− = − x because x
2 $2 py
p
costs two and half times y. The price ratio, " x , and not the individual prices, is what is important in
py
making choices. That the price of x is $5 or $5,000 doesn’t mean much unless you know the prices
of other things that might be consumed along with or instead of x. (I’ll leave it to you to review
how the budget line changes position if one or both prices or income changes. You’ve done it be-
fore.)
If you have no income but are endowed with 15 x and 12½ y then your wealth is the market
value of your endowment at market prices—whatever you own or whatever you’re entitled to—and
that’s $100 ($75 worth of x and $25 of y). Your wealth is $100 everywhere along the budget line

1 To keep it short, I’ll use goods to mean goods, services, or activities.

2It is a line if there are only two goods or only two are being considered. It is more generally called a bud-
get constraint or income constraint.

!7
or wealth constraint, so long as prices stay the same. Wealth, unlike income, is a stock; it is a value
at a point in time. A warehouse inventory, a firm’s total assets, the number of bottles of beer in
your fridge, my collection of Beatles records, and my daughter’s 143 Beanie Babies (which she
insists we store for her indefinitely) are all stocks.
The budget line looks exactly the same whether you have an income (but no endowment)
and are deciding how to spend it, or you have an endowment of goods (but no income) and are
deciding whether to trade off some of one good to get more of the other. The only difference is
that both the x and y intercepts of the wealth line change even when only one of the prices has
y

65.0 W = $130
$100 = $5x + $2y ⟺ W = px x + p y y
p W 5
50.0 W = $100 y=- x x + ⟺ y = - x + 50
py py 2

35.0 W = $140
a
30.0

e
12.5

5.0 b
x
8 15 18 20
changed. This is because the wealth line always passes through the endowment point. This makes
sense because, no matter how much the price of goods changes, you can always afford what is
already yours. Can you say how prices must have changed for the initial wealth line ($100) to be-
come the wealth line at $140 or $130?

!8
Utility Functions and Indifference Curves
Your budget fixes the choices that you can afford to make. Your tastes set out the choices that
you want to or are willing to make. Imagine listing all of the bundles of x and y that give you exactly
the same satisfaction as your endowment of 15 x and 12½ y. This set of bundles is called an indif-
y

u( x, y) = x + y = 7.40852
g
33

i
19 h
16 e
12.5

x
4 7 15 17
ference curve, and it is how tastes are modelled in economics. Your tastes for x and y might be repre-
sented by the indifference curve shown in the next figure.The value of the function at e is about
7.4. This number is called utility and can be taken as a measure of how satisfied you are with 15
x’s and 12½ y’s. All of the other points on the indifference curve are bundles that imply utility of
7.4.
Bundles that a person prefers must lie on higher indifference curves. If someone offered you
bundle g in return for e, you would take it. The utility of g is higher (I’ll let you do the math), but
it’s not like you had a higher numerical value of utility in mind when you made the trade. You
simply prefer g to e. Utility is not a real, physical, measurable thing like temperature, gravity, or the
number of hairs on your head. It is just a ranking that is implied by the choices we make. Actual
tastes, as mentioned before, are difficult to measure before a choice is made or if a choice is not
observed. I don’t think that many people believe that our tastes can truly be captured by mathe-
matical functions, but if the functions are chosen carefully, they have plausible economic interpre-
tations. A person like you with square root utility would choose g over e. Someone else with tastes
represented by another function might not.
So what economic meaning is in the shape of indifference curves? They slope downward
because they represent the tradeoff that a person is willing to make between two goods. Take
away some of my x, and you’ll have to give me a certain amount of y in return to leave me as well
off as I was before. The tradeoff always exists if both goods are “good” in sense that more is pre-
ferred to less. It is not the case if one of the goods is a “bad” like secondhand cigarette smoke or
construction noise. The slope of an indifference curve, ignoring the negative sign, is what really
defines a person’s tastes because it is the most y that a person is willing to give up in order to gain

!9
one more x or the least y they demand in return for giving up one x. The absolute value of the
slope is called the marginal rate of substitution. Think of it as a psychological price ratio defined by
personality or tastes. The marginal rate of substitution at e for the square root utility function is
about 0.91: one more or one less x is worth 0.91 y to you if you currently have 15 x’s and 12½ y’s.
The marginal rate of substitution at a is about 1.3 and at b, 2.3. Why does a unit of x increase in
psychological value moving from right to left along an indifference curve? Relative scarcity. You
have less x compared to y at b than you have at a than at e. It makes sense that we value something
y

dy Δy
MRS = ≃
dx Δx

u( x, y) = x + y = 7.40852

b ( MRS = 2.31319)
26.7542

a ( MRS = 1.34278)
18.0306
e ( MRS = 0.912871)
12.5

0 x
0 5 10 15
more highly as it becomes scarcer relative to other stuff. Economics handles this by using convex
functions for indifference curves.
The functions chosen for indifference curves also never flatten out completely because that
would imply that there comes a point when you have so many x’s that you are unwilling to pay
anything for one more (MRS is zero). That wouldn’t sit well with greed assumption. If you prefer
more to less, then one more slice of pizza always provides you with some positive utility, no matter
how small, and it does not matter that you have just finished your 17th slice.

!10
 We have different tastes. I wish that high-waisted, pleated pants would come back in style. I
bet that you do not. The marginal rate of substitution is the only way to tell one person from an-
other, theoretically that is. The next figure shows Lucy and Ricky’s indifference curves along with
y

you: u( x, y) = x + y = 7.40852
g Lucy: u( x, y)= 4 ln( x ) + 2 ln( y) = 15.8837
33

i
19
h
16
e
12.5 Ricky: u( x, y)= 2 ln( x ) + 5 ln( y) = 18.0447

x
0 4 7 15 17
yours. They are log utility people in this example. All three of you have the same endowment.
Who places a bigger value on the next unit of x?

Optimal Consumption
Utility theory brings together budgets and utility functions in a simple mathematical opti-
mization: people make choices that maximize their utility subject to their budgets. English translation: people
make the best choices for themselves that they can afford. As a person whose tastes are represent-
ed by a square root utility function, you would choose bundle C*, about 5.7 x and 35.7 y, because
y

u* = 8.3666 u( x, y)= x + y

ue = 7.40852
50.
40 250 p
C* ( , ) : MRS = x = 2.5
7 7 py
35.7

e: MRS = 0.912871
12.5

x
5.7 15 20.

!11
this brings you the highest utility. I’ll let you check that utility at the optimum is 8.3666. There is
no other bundle on the budget line that gives higher utility because the indifference curve that
passes through C* is tangent to the budget line; it touches the line at just that one point. Moving
the indifference curve any higher would place it above the budget line, and those bundles are not
affordable. Because the optimum is a tangency, it is determined by equating the slope of the indif-
ference curve to the slope of the budget line. Because the negative signs of the slopes cancel when
they are set equal to one another, C* is chosen so that the marginal rate of substitution is equal to
the price ratio, which is constant at 2½-to-1.

px $5
C* : M R S = =
py $2

The marginal rate of substitution is the rate at which you are willing to trade x for y, and the price
ratio is the rate at which the market is willing to trade them. You cannot make yourself better off
when you’ve reached the point where the value you place on the next unit of x is the same value
that the market places on it. How did you get there? If utility theory says that people maximize
utility, they are always at C*. Why would anyone be anywhere else? If there was a change in prices
or income or some other event that moved them away from C*, they would quickly do whatever
was necessary (like trade) to get back to C*. That implies that we really wouldn’t find people at e
for any appreciable length of time. C* is your current e as far as anyone can tell. But not many
would understand utility theory if they were told that C* simple is because it is, so the explanation
almost always includes a story of how a person trades from their endowment to move up along
the budget line until they reach C*. At e, your marginal rate of substitution is less than the price
ratio, which means that you value the next unit of x less than the market does. So you sell some of
your x’s in return for 2½ y’s each. As you climb up the budget line, your marginal rate of substitu-
tion increases because x is becoming relatively scarcer to you. The bliss point is reached at C*
when your marginal rate of substitution is exactly equal to the price ratio, and you stop selling.

!12
Reading
Camerer, Colin F. and Ernst Fehr, 2006, “When Does ‘Economic Man’ Dominate Social Behav-
ior,” Science, 311:47-52.
Camerer, Colin F. and Richard H. Thaler, 1995, “Anomalies: Ultimatums, Dictators, and Man-
ners,” Journal of Economic Perspectives, 9(2):209-219.
Kahneman, Daniel and Richard H. Thaler, 2006, “Utility Maximization and Experienced
Utility,” Journal of Economic Perspectives, 20(1):221-234.
Thaler, Richard, H., 2000, “From Homo Economicus to Homo Sapiens,” Journal of Economic Per-
spectives, 14(1):133-141.
“Blatant Benevolence and Conspicuous Consumption,” The Economist, August 4, 2007.
“Happiness (and How to Measure It),” The Economist, December 23, 2006.
“Economics Discovers Its Feelings,” The Economist, December 23, 2006.
“Prison Breakthrough,” The Economist, August 20, 2016
[optional] Chapters 1 to 5 in Varian.

!13
 3
EQUILIBRIUM

What can be said about the welfare of society as a whole if, as economic theory assumes,
each of us looks out for number 1? Go to school or not, get married or not, get a job or live in
your parents’ basement for the rest of your life. All of these me-first decisions inevitably involve us
interacting with each other, affecting each other, and while not always apparent, take in place in
countless different markets.
What then does it mean for a society to be as well off as can be? One gauge of improved
welfare is a Pareto improvement, which means that at least one person is made better off without
anyone else being made worse off. An allocation is Pareto optimal if it is impossible to make
someone better off without making others worse off.

An allocation is Pareto optimal if any of the following is true

there is no way to make everyone better off.

there is no way to make one person better off without making someone else worse off.

there are no gains from trade.

Here is how to draw a Pareto optimum. Pretend that the economy is inhabited by just two
people, Lucy and Ricky, and both have log utility functions of the form

u" (x , y) = α l n (x) + (1 − α)l n (y)

where 𝛼 = 0.4 for Lucy and 0.3 for Ricky. Lucy is endowed with 20 x and 60 y. Ricky is endowed
with 80 x and 20 y. This means that the aggregate endowment or supply of x and y in the econo-
my is 100 x and 80 y. The economy can be represented as a rectangle, called an Edgeworth box,
with the width equal to the aggregate endowment of x and the height equal to y. Lucy’s origin is
the lower-left corner and Ricky’s the upper-right. Lucy has some level of utility associated with
her endowment, with an indifference curve passing through it (the light blue one), and so must
Ricky with his (also light blue). Point e represents their endowments, read from Lucy’s perspective.
Lucy increases her utility if she can obtain a bundle that lies above her initial indifference curve.
For Ricky, down is up because his origin is the top-right corner of the Edgeworth box. He in-
creases his utility by moving down towards Lucy’s origin. The shaded lens-shaped area formed by
the intersection of their initial indifference curves, including the boundary, is the area of Pareto
improvement. A reallocation to any point in the shaded area from e would make both better off,
or at least one better off without hurting the either. Neither would object to such a reallocation.

!14
 y Ricky
80
Pareto Optimum
p
C* : MRSLucy =MRSRicky ⇔ x = 0.441176
e py
60
Give up
18.7059 y
C*
41.29
Get 42.4 x

0 x
0 20 62.4 100
Lucy

The allocation C* is the Pareto optimum. At this point neither Lucy nor Ricky can be made
better off without making the other worse off. If Lucy were to move to a higher indifference
curve, Ricky would have to pull back to a lower one. The same goes for Ricky. To get to C*, Lucy
bought some x from Ricky in return for some y. It makes sense that Lucy would buy x from Ricky
since her marginal rate of substitution at e is bigger than his (can you see that?). At the Pareto op-
timum, Lucy and Ricky place the same value on the next unit of x in terms of y, so there are no
mutually advantageous trades left. Mathematically, their indifference curves are tangent to one

Lucy Ricky

Taste parameter 𝛼 0.4 0.3

Endowment e {20,60} {80,20}

Optimum C* {62.4,41.2941} {37.6,38.7059}

MRS(e) 2 0.107143

MRS(C*) = price ratio 0.441176 0.441176

Utility(e) 3.6549 3.41162

Utility(C*) 3.88586 3.6473

another (the dashed grey line passing through e and C*), which means that their marginal rates of
substitution are equal. It is the condition that defines the equilibrium.

( py )
px
C* : M R SLucy = M R SRick y =

!15
 Their common marginal rate of substitution at the Pareto optimum must also be the equi-
librium price ratio. If money were to take the place of barter, the dollar prices of x and y, whatev-
er they might be, would have to be in the ratio 0.44 (see table or, again, the slope of the tangent
line passing through e and C* in the graph). Notice that the price ratio is not given but is, in fact,
determined jointly with the optimal allocation C*, and is called a general equilibrium—the values of
all important variables, such as consumption and the price ratio, are determined jointly. You are
not expected to be able to compute the Pareto optimum in this course (take my FINA 385 for
punishment), but you should be able to look at the figure or table above and interpret them.
When a market is allocationally efficient, the allocation of goods and services and the price
ratio together reflect an aggregation of everyone’s tastes. If production had also been included in
the economy, the allocation and price ratio would also be driven by producers, which could be
Lucy and Ricky, each choosing the mix of x and y to manufacture so as to maximize profits.3 You
can think of an allocationally efficient market as one in which each person derives the most bene-
fit—utility for consumers and profit for producers, which in turn becomes utility too—at the same
time that everyone else is doing the same. It is an equilibrium in the sense that there is no incen-
tive for anyone to change: Ricky is happiest with his consumption bundle (37.6, 38.7) at C*, and
Lucy hers (62.4, 41.3), in light of the price ratio, 0.44, and the price ratio is literally what it is be-
cause Lucy and Ricky are holding bundles for which their marginal rates of substitution are
equal. I know what you are thinking: this whole general equilibrium looks like a tautology. Well, it
is, but that is a problem of the economic paradigm that we’ll have to leave for another day.

The First Welfare Theorem of Economics

It looks like trade is one way for an economy to reach a Pareto optimum, although it isn’t
clear what kind of trade, or whether the optimum is unique or exits at all. But are there other
ways? Could a government or benevolent dictator come up with a formula to reallocate the en-
dowments of its citizens to a Pareto optimum—doing for the people what is best for the people?
Does a democratic system of majority vote do the trick? Lotteries? Rankings? No such luck. Pro-
fessor Kenneth Arrow figured out that there is no social decision rule that can guarantee a Pareto
optimum, and so we have Arrow’s Impossibility Theorem. The reason is pretty simple: governments,
rulers, or anyone on the outside, cannot know people’s tastes, so it would be a fluke if any social
decision rule they created could pick off a Pareto optimum. Makes you wonder how the millions
of shareholders in modern corporations agree to anything!
Trade saves the day though; but only one kind of trade or market guarantees a Pareto op-
timum, and that is pure competition. This is not competition as in destroy-my-competitors-no-
matter-what kind of competition but the kind we read about in textbooks where everyone is taken
to behave as if they were price-takers. This means that when Lucy is thinking about how many y’s
she’d be willing to pay Ricky for so many x’s, she does not think that her demand for x could
somehow influence its price in the bigger scheme of things. The same goes for Ricky. Of course,
the demand and supply of the millions of people like Lucy and Ricky, taken all together, do influ-

3The economy described here is sometimes called a pure exchange economy because there is trade but no pro-
duction.

!16
ence prices and quantities. But individually, each person must act as if they were inconsequential,
tiny fish in a big sea.
If Lucy and Ricky set out to maximize their utility and do so by trading as price takers, they
will arrive at a competitive equilibrium. The First Welfare Theorem of Economics tells us (we won’t
prove it) that competitive equilibria are Pareto optimal. So, if you know that a market is in equi-
librium in a supply-equals-demand sense and that it is a purely competitive—something harder to
do—then you also can conclude that it is also Pareto optimal. You don’t need to know anything
about individual tastes or motives. It is underlies Adam Smith’s idea of the invisible hand at work:
everyone acting in their own self interest makes for the best outcome for society as a whole. The
implication of The First Welfare Theorem of Economics is profound because it suggests that free
markets and economic freedom in general—private property and protection of property rights,
reasonably low taxes, rights to trade and do business—should be the default path for improving
societal welfare. What does that say about centralized economies?

!17
Reading
Farrell, Joseph, 1987, “Information and the Coase Theorem,” Journal of Economic Perspectives, 1(2):
113-129.
Roth, Alvin, E., October 2007, “The Art of Designing Markets,” Harvard Business Review, 118-126.
Peck, Don and Ross Douthat, “Does Money Buy Happiness?” The Atlantic, January-February
2003, 291(1):42.
“Optimising Romance,” The Economist, February 13, 2016.
“Say’s Law: Supply Creates Its Own Demand,” The Economist, August 10, 2017.
[optional] Chapters 6 to 9, 14 to 16, 30, and 32 in Varian.

!18
 4
INFORMATION

In the previous note, one x is valued at 0.441176 y’s in (general) equilibrium. Everything
about the economy is packed into that price ratio. When the price ratio changes, it signals change,
although what change is not always clear: technology, weather, demography, tastes, health. Take a
step back from the price ratio to the information about those things that might cause the price
ratio to change. Firms have to buy or rent their factors of production. For this they need financing
from investors in the form of equity and loans. While the relative prices of goods and services are
a signal to firms to allocate their factors of production profitably, existing and potential investors
also rely on signals from the prices of financial securities to help them know whether this is being
done; otherwise, investment capital wouldn’t necessarily flow to the firms earning the highest re-
turn for a given level of risk. If we are greedy, wouldn’t we be monitoring all kinds of information,
trying to figure out its effect on the prices of goods and services, and in turn, the effect on the
prices of stocks and bonds of the companies that put out those goods and services? You bet. Good
news about the future cash flow of a firm should translate into a rise in the price of its securities
and bad news a fall because greedy investors act on the news. How big a rise or fall? The present
value or discounted value of the change in future cash flow, such as dividends, that investors expect to
receive.4 The level of and change in security prices presumably acts to discipline firms to produce
efficiently.
How quickly should security prices adjust to new information? If enough of us jump up to
buy or sell a financial security when there is news, its price should change quickly, so quickly in
fact, that a stock that is underpriced at the moment of good news is never underpriced long
enough for anyone to profit from buying it. It is never really underpriced to begin with because
new information would be incorporated into the price instantly! (That’s economics channeling
physics.) That blindingly fast change in price should equal the present value of the expected
changes in future cash flow. Some investment companies locate their operations as close as possi-
ble to the stock exchange so that they can minimize the length of wire runs from their computers
to the stock exchange’s servers. Enough said. The price is always right, in theory at least.
A capital market is said to be informationally efficient if the price of its securities reflects all rel-
evant information about future cash flow. This is also called the efficient markets hypothesis. Nowhere
is there greater attention paid to the relationship between prices and information than in capital
markets, where investors, analysts, alchemists, and clairvoyants try to decipher the effect of all
sorts of information on the value of securities. (I remember watching a TV show on PBS about
corporations and capitalism in America, and a commentator said something to the effect that if it
is reported that a woman dies of breast cancer in America, the price of biotech stocks will go up. I

4 You’ll get to do some present value calculations and other time value of money calculations in the quizzes.

!19
thought it was Noam Chomsky who made the remark, but when I emailed Professor Chomsky
about it, he replied that he did not recall.) Informational efficiency is to capital markets what allo-
cational efficiency is to the markets for goods, services, and factors of production. In an informa-
tionally efficient capital market, you can’t make yourself wealthier trading securities in the same
way that Lucy and Ricky do not make themselves wealthier trading x and y in a pure exchange
economy. To get rich trading financial securities—and that does not mean hitting the jackpot now
and again because anyone can get lucky—you’d have to have valuable information that most oth-
ers do not have or have a better understanding of the available information than everyone else.

Rational Expectations
The assumption that people are soaking up and responding to information continuously also in-
cludes each of us taking into account how we think everyone else is motivated and responds to
information. A long time ago, an insightful little story about this appeared on the back cover of
Journal of Political Economy. A group of young people who, while taking a walk on a well-traveled
country road, come across a peach tree laden with ripe fruit. When one of them suggests that they
stop to pick some peaches, another quickly responds that they shouldn’t bother because if the
peaches were any good they would’ve already been picked. That’s rational expectations. I like peach-
es, and all the better if they are free. But wait, don’t most other people like free peaches? And
wouldn’t they pick them if they were free? Then why are those beautiful peaches still on the tree?
In utility theory, tastes are modelled by utility functions because tastes are unobservable. In
the same way, individual decision processes are unobservable, so it is assumed that people behave
as if they take into account every available scrap of information in making any decision, and that
they are frighteningly logical about it. Rational expectations takes this individual logic one step
further in assuming that we consider the effect of our choices on everyone else and everyone else’s
choices on us. The peaches never get picked. Assuming rational expectations to avoid having to
include a specification of individual decision processes in economic models implies that people are
assumed to behave as if they know the true model, that is, the economist’s model of the economy
(how convenient for economists), and that they behave as if everyone else knows the true model of
the economy (I know that you know that I know…). An implication of rational expectations is that
monetary policy is ineffective. A government that pursues an inflationary monetary policy by
printing money will be undone by unions who anticipate the inflation and, in turn, demand wage
increases to compensate for the erosion of real wages.

Take a Random Walk

An implication of the efficient markets hypothesis is that you can’t use the past to predict
the future: markets have no memory. That’s because news isn’t really news unless it is, itself, a sur-
prise. If news cannot be predicted, then our reaction to it must result in price changes that cannot
be predicted. If you used past changes in price as your information or news, and identified a pat-
tern in those prices that could predict the future ups and downs of prices, then there are probably

!20
other clever people who have found the same pattern.5 Everyone would exploit the pattern, buy-
ing when it signals a rise in price and selling when it signals a decline. The buying and selling
would cause the pattern to self-destruct, taking down trading gains along with it. That’s why it is
sometimes said that price changes in an informationally efficient market should resemble a random
walk, and markets where the past cannot be used to predict future is said to be weak-form efficient.
Here are three stocks, each of which happened to be worth $100 at the beginning of month
1. Twelve months later there are big differences in their prices. Some people will look at a price
chart like this and see patterns or trends. Some might say that the price of the brown stock has
Price
$214.073

200

180
$167.614
160

140

120 $122.542

100

80 Month
2 4 6 8 10 12 14
reached its peak and will soon start to fall. Others might say that the blue stock is trending up-
wards. What would they think about the green stock? Those who didn’t flunk their intro stats
course might compute the correlation of the month-to-month price changes of each stock, despite
the fact that they have only 11 price changes to work with (or they might go further back in time
to increase their sample size). They might find that one or two of them display positive serial correla-
tion: a rise in price is more likely to be followed by another rise than a fall. They might find that
one of the stock’s price changes is not serially correlated or may be negatively serially correlated
(what does that mean?).
If it were now the end of month 12, would you base your decision to buy one of the three
stocks on a pattern that you see? It wouldn’t matter because there are no patterns. They all follow
random walks, where each month there is an equal chance that price rises by 20 per cent or falls
by 10 per cent. The monthly changes are completely unpredictable. The only thing that can be
predicted is that if you buy any one of them and hold it long enough, your average return will be
five per cent over the long run.6

5 Don’t think that you are the only smart cookie out there.
6Since we are looking ahead, we can also say that the expected return is five per cent, a probability-weighted
average, 0.5 x 20% + 0.5 x (-10%) = 5%. Probability-weighted averages are called expected values.

!21
Bubbles and Crashes
If people have rational expectations and information about a company’s future dividends is
widely available, the price of the company’s stock should equal the present value of its expected
(future) dividends. Professor Vernon Smith ran a series of clever experiments that demonstrated
that a stock market bubble and crash could be created in a laboratory under controlled condi-
tions, and that the rational expectations hypothesis is violated (for shame).
Subjects took part in an oral double auction market, trading a stock with a ten-period life.
At the end of each period, a coin toss determined whether the stock paid a dividend of $15 or
$25, and all participants knew this. The rational expectations price each period, that is, the equi-
librium price implied by the efficient markets hypothesis, is simply the sum of expected dividends.
Since the expected dividend each period is $20, the predicted price for period 1 is $200 ($20 x 10
periods), $180 for period 2 ($20 x 9 periods), and so on down the blue steps in the figure. In real
life, the predicted equilibrium prices would be less than the sum of expected dividends because
the market would discount them at some positive interest rate to compensate for the time value of
money and the fact that the dividends are risky. But in an experimental market it is reasonable to
take the interest rate as zero because each period is short, say, 20 minutes, as opposed to months,
quarters, or years, and subjects face little risk since they do not have to put up any of their own
P

200 H0 assuming risk neutrality and time irrelevance

180
160
140 Actual trading prices

120
100
80
60
40 E DIVt+1 E DIVt+2 E DIVT
H0 :Pt = + +…+
20 1+r t+1 1+r t+2 1+r T

t
1 2 3 4 5 6 7 8 9 10
money to trade. Subjects are effectively assumed to treat the time between dividend payments as
irrelevant and ignore risk. Someone who does not care about risk is said to be risk neutral as op-
posed to risk averse, which is natural for real life situations. Risk neutrality implies that subjects care
only about the expected dividend and not how risky it is: they see no difference between a fifty-
fifty chance of receiving $15 or $25 or a fifty-fifty chance of $5 or $35. The actual average trad-
ing prices (brown line) sketch a classic stock market bubble and crash, where at some point the
price might be so high that even if all of the remaining dividends were $25, it would not be
enough to recoup the purchase price!

!22
 Professor Robert Shiller’s surveys of investors who experienced the stock market crash of
1987 and the boom and bust real estate markets of the mid to late 1980s are provocative. He finds
there is a marked tendency for investors to focus most closely on recent price changes as their
primary source of information about the future, resulting in herd behaviour. If everyone is watch-
ing price changes in order to guess what everyone else is thinking, who’s looking at economic fun-
damentals? This is hinted at in Professor Smith’s bubbles and crashes experiment in the com-
ments of subjects who reported that they were aware that the stock was overpriced but bought
anyway (or did not sell) because they were afraid of missing out on further possible price rises.
Anyone taken as an individual is tolerably sensible and reasonable — as a member of
a crowd, he at once becomes a blockhead.
—Friedrich Von Schiller, as quoted by Bernard Baruch

Is Private Information Reflected in Prices?

Suppose you took part in an experiment that had you trading shares of a stock in a contin-
uous, double oral auction market. The shares pay dividends in each of two periods, A and B. The
dividends are certain, so you know exactly what you will earn, but yours won’t necessarily be the
same as those paid to other investors (types I, II, or III). Information about dividends is also pri-
vate. You do not know what others will earn, and they do not know what you earn, and everyone
knows that.

Investor Period A Period B Total

I $60 $40 $100

II $90 $50 $140

III $40 $70 $110

What does economic theory have to say about the equilibrium price in period A and B? If a
price is an equilibrium price, there are no gains from trading at that price, and so there would be
no point in trading. Start in period B and work back to period A. The equilibrium price in B must
be $70 because, at a price of $70, type III investors are indifferent to buying and selling because
that is exactly the dividend they will earn. At any price below $70, type III would gladly buy from
I or II and pocket the difference between the purchase price and $70, and at any price above $70,
III would gladly sell, although neither I or II would buy. Because B is the last period, the equilib-
rium price is equal to the highest dividend anyone might receive, and that happens to be the divi-
dend paid to type III investors. Now slide back in time to period A. The price in A must be at least
$140 because type II investors can earn $140 simply by sitting back and collecting dividends. But
all investors can trade, and that means that those earning lower dividends can gain by selling to
those earning higher dividends, who can of course gain too. The equilibrium price in A must
then be $160 because the most that anyone (type II) can earn in A is $90 and the most that any-
one (type III) can earn in B is $70. The right to trade means that price reflects the biggest bene-
fit—cash flow in this case—that can be received at every point in time, but it does not matter who

!23
receives it. You can also think of the value of the right to trade as being $20 in this experiment,
the difference between the equilibrium price of $160 and $140, the biggest total cash flow that
any one investor type could earn from dividends alone. That’s neat because the right to trade is a
legal construct or freedom, and we’ve just demonstrated that is has value; it results in a Pareto im-
provement. A right has non-negative value.
It may be hard at first to wrap your head around this result because you can’t help but think
of real people, such as yourself, trading. How can the price be $160 in A if no one, not even in-
vestor III, knows that the biggest dividend to be paid in period B is $70? That’s where the finess-
ing assumption of rational expectations comes in. Economic theory assumes that, even though
dividends are private information, investors behave as if they do know each others’ dividends (I act
like I know even though I really don’t know). Without the rational expectations assumption, the
equilibrium price cannot be predicted. And with the assumption, the equilibrium price in each
period appears instantaneously. The moment the bell rings to start trading in period A, the bid
and ask prices must be $160; the market is in equilibrium from the get-go and no trade occurs.
The same goes for B. A market that is so informationally efficient that its prices even reflect pri-
vate information is said to be strong-form informationally efficient. If real-world markets were
strong-form efficient, insiders would not be able to enrich themselves at the expense of others, and
investment advisors and hedge funds would not do better than the rest of us (which, in fact, they
generally do not).
That is the theory. For experimental sessions, we wouldn’t be so demanding. As a subject in
the experiment, you are only human after all, and you cannot read the minds of other subjects.
We’d expect trade to occur and hypothesize that the average trading price in A is $160 and the
average in B $70, and that the trading prices should converge to these hypothesized values quickly
rather than slowly. It turns out that in actual sessions the price in period B moves to $70 pretty
quickly in the first run, but the period A price hangs around $140. If the experiment is repeated a
number of times, subjects learn through the feedback of repetition that the stock is worth $70 in
B and then incorporate this information into the period A price in subsequent runs, driving up the
price to $160. It takes time to learn just as Bill Murray did in the 1993 movie Groundhog Day. In
class, we’ll discuss how the addition of a forward market can make the prices in this experimental
market move more quickly to their efficient levels. (You can learn about forward markets in
“About Forwards and Futures.pdf ” and in the first part of chapter 15 of these notes.)

!24
Reading
De Bondt, Werner and Richard H. Thaler, “1989, Anomalies: A Mean-Reverting Walk Down
Wall Street,” Journal of Economic Perspectives, 3(1):189-202.
Forsythe, Robert; T.R. Palfrey, and C.R. Plott, 1982, “Asset Valuation in an Experimental
Market,” Econometrica, 50(3):537-568.
Shiller, Robert, 1990, “Speculative Prices and Popular Models,” Journal of Economic Perspectives,
4(2):55-65.
Thaler, Richard, 1987, ”Anomalies: The January Effect," Journal of Economic Perspectives, 1(1):
197-201.
Thaler, Richard, 1987, ”Anomalies: Seasonal Movements in Security Prices II: Weekend, Holiday,
Turn of the Month, and Intraday Effects," Journal of Economic Perspectives, 1(2):169-177.
“Bubble Bubble,” The Economist, March 25, 2000.
“To Infinity and Beyond,” The Economist, October 6, 2007.
“Dismal Science, Dismal Sentence,” The Economist, September 9, 2006.
[optional] Chapters 11 through 13 in Varian.

!25
 5
PRODUCTION

Just as a person’s decision to consume a certain mix of goods can be modelled with a utility
function, a firm’s decision to produce a certain mix of goods, meat pies (x) and sausage (y), can be
modelled using a function called a production possibility curve. For people it is a model of tastes; for
firms it is a model of technology and resources. A production possibility curve, like the one in the
figure, sets out the most of one good that a firm can be produce given the amount of the other
good to be produced. If the firm chooses to produce 365 meat pies then its technology permits it

a
954 b
890 ( 0.39 )
( 0.64 ) c
791
( 0.97 ) px $9
P* : MRT = =
640 py $6
2 y 2 ; g = 800, h = 1000 ( 1.5)
1= x +
g2 h2
d
381 $9,372
( 3.03 )

$7,881

x
240 365 490 615 740

to produce up to 890 kg of sausage (point b). The shape of a firm’s production possibility curve
depends on the current state of technology and the quantity and quality (or productivity) of its
factors of production—land, labour, and capital. Another firm producing the same meat pies and
sausage may have a production possibility curve that is flatter or more rounded or pushed out
more or less. Production possibility curves can represent the production tradeoffs for a person, a
firm, an industry, or a whole economy.
Like indifference curves, production possibility curves slope downward because they repre-
sent a tradeoff. Increasing production of meat pies to 490 from 365 means reducing production
of sausage to 791 kg.7 The absolute value of the slope of a production possibility curve is called
the marginal rate of transformation (MRT). It is an opportunity cost because it tells us how many units
of y must be given up to produce one more unit of x, or how many more units of y can be pro-

7 Don’t you just love Italian sausage on the barbecue? I barbecue all winter.

!26
duced if one less unit of x is produced. At point c, the cost of producing one more meat pie on top
of the 490 already produced is 0.97 kg of sausage, roughly one for one. But unlike indifference
curves, production possibility curves are usually drawn strictly concave to the origin. This is be-
cause the rate of swapping (transforming) one good for the other is diminishing—it becomes in-
creasingly costly to produce each additional unit of x as x production is increased. That makes
sense because producing more and more x means shifting factors out of y production and into x
production, and at some point the firm will be tapping into factors that are not as well suited for
making x as they are for making y. Some of the sausage makers are equally proficient at making
meat pies, others less so.8
As economics assumes that people are greedy, the people who own and run companies will
produce a mix of x and y that maximizes profits. At a price of $9 for meat pies and $6 per kg of
sausage, the firm will earn about $7,881 if it produces 240 meat pies and 954 kg of sausage (point
a). As a matter of fact, it will earn $7,881 if it produces anywhere along the line passing through
px $9 3
a, whose slope (ignoring the minus sign) is " = = . This is called an iso-profit or iso-income
py $6 2
line. But b is a better choice than a because profit or income with this mix is $8,622. Why the in-
crease? At a, producing one more x only costs four-tenths of a y. You earn $9 for an extra x and
$2.40 for the four-tenths of a y which brought in $6. As long as the price ratio is greater than the
marginal rate of transformation, you can earn more by shifting production out of y and into x.
Let’s see the math. When

px Δx
" > MRT =
py Δy

cross multiply gives

p
" x Δx > py Δy

which says that the profit or cash flow gained from one more x is greater than the profit foregone
from the 0.392 fewer y’s.

$9
" × 1 > $6 × 0.392

If the inequality was reversed, producing more y and less x would increase your profit (start at d,
for example). Profit must be maximized when the mix is chosen so that net profit or net cash flow
doesn’t change at all if production of x is increased or decreased by one unit. That point is P*, the
tangency of the production possibility curve and the iso-profit line, where profit is $9,372, right?
At P*,

p
" x Δx = py Δy

means that

8 This discussion would not apply to sushi and gluten-free wraps stuffed with kale because both of those
foods are just plain silly.

!27
 px
P* : M R T =
py

For those of you who are interested in cranking out the numbers or are seriously bored, the pro-
duction possibility curve in the figure is part of an ellipse with equation

x2 y2
1" = +
g2 h2

where g = 800 and h = 1,000.

Where’s the connection to financial markets in all of this? Well, if profits each period are as
big as they can be, the market value of the company—it’s assets—is as big as it can be, and share-
holders will thank you because the value of their shares are as big as they can be (lots of bigs
here). If not, say because the boss choose a production point other than P*, cash flow would be
smaller and share value lower. If investors have information about the company’s technology and
the competence of its managers, and can make comparisons to similar companies, then they’d
recognize that the company is not doing as well as it could be. Some will see the low share price as
a bargain if they figure that profits can be improved by giving the boss the boot and installing one
that knows that P* is best. The gains go to the investor activists. Welcome to the world of corpo-
rate finance.

!28
Reading
Montgomery, Cynthia, 1994, “Corporate Diversification,” Journal of Economic Perspectives, 8(3):163-
178.
“Beyond Shareholder Value,” The Economist, June 28, 2003. [Survey]
“How to Pay Bosses,” The Economist, November 16, 2002.
Chapters 18, 19, and 31 in Varian. [Optional]

!29
 6
GOVERNMENT INTERVENTION

Government intervention takes many forms but often involves manipulation of market
prices. Taxes reduce incomes, subsidies lower prices, sometimes for select groups. This results in a
misallocation of resources because prices will no longer signal the relative value that people place
on different goods and services. But government intervention is not a bad thing so long as its costs
do not outweigh its benefits (to state the obvious).

Taxing and Subsidizing Business Loans

Suppose that interest paid on business loans was not a tax-deductible expense, and that
lenders were on not taxed on interest earned from business loans. Suppose also that both borrow-
ers and lenders are taxed at a rate of 50 per cent, and that this rate applies to all of the other usu-
al types of income and expenses; it’s just that interest business loans is left out. Point e in the graph
is the equilibrium for business loans in this situation. The interest rate is assumed to be eight per
cent, and the quantity (dollars) of loans outstanding is 1,000. But then the government decides to
change the tax rules. Business loan interest becomes a deductible expense, and the banks will be

Td = Ts = 0.5
0.2

e′
0.16

0.12

e
0.08

0.04

0. q $ loans
0 500 1000 1500 2000

taxed on that same income. To borrowers, interest expense deductibility is a subsidy because it
reduces taxes payable. This type of subsidy is called an ad-valorem subsidy or value subsidy because
the amount depends on the price (the interest rate). It is also called a tax shield in this case. The tax
on interest earned by lenders is called an ad-valorem tax or value tax because, like the subsidy, the
amount depends on the price.

!30
 How will the change in tax rules affect the market for business loans? The demand for loans
will increase because interest deductibility makes the after-tax cost of a loan less than the before-
tax cost. Borrowers can now afford to pay a higher before-tax rate of interest for business loans.
You can find a point on the new demand schedule by asking how high the before-tax rate of in-
terest—eight per cent before the change—can rise without costing borrowers more than eight per
cent after tax. The answer is obviously 16 per cent because a borrower in a 50 per cent tax brack-
et who pays 16 cents in interest on every dollar borrowed gets a tax shield of eight cents, so they
are really only paying eight cents in interest. Let x be the before-tax rate of interest and Td (d for
demand) the borrowers’ tax rate, then

0.08 = x (1 − Td )
" 0.08
x= = 0.16
1 − 0.50

That makes sense because the subsidy or tax shield is eight cents on the dollar when the interest
rate is 16 per cent.

"
Tax Shield = Td × (r × q) = 0.50 × (0.16 × $1) = $0.08

The new demand curve must pass through ($1,000; 0.16) no matter what the rest of the curve
looks like (it doesn’t have to be a straight line like the one shown). Substitute different before-tax
interest rates in the numerator in the equation above, and you will see why the demand curve
shifts up more (or forward if you like) at higher interest rates and less at lower ones. That’s the
nature of an ad-valorem subsidy.
For lenders, having their interest income taxed is an increase in the cost of lending because,
after all, taxes are a cost of doing business. A bank facing a 50 per cent tax rate will now get to
keep only $50 of every $100 in interest earned, so the supply of loans will decrease. You can find
one point on the new supply curve by asking how high the interest rate would have to rise for
lenders to continue earning eight per cent after tax. The answer is 16 per cent again because a
lender in a 50 per cent tax bracket will have to pay eight cents in taxes to the government for
every 16 cents of interest earned. If Ts is the lenders’ tax rate (s for supply),

0.08 = x (1 − Ts )
" 0.08
x= = 0.16
1 − 0.50

The new supply curve, like the new demand curve, passes through ($1,000; 0.16), so the
new equilibrium must be ($1,000; .16). The market for business loans did not grow or shrink (q* is
still $1,000) because the after-tax return to lenders and the after-tax cost to borrowers did not
change (still eight per cent). The change in tax policy is neutral because it does not affect con-
sumers (borrowers) or producers (lenders). And if it is neutral for consumers and producers, it
must be neutral for the government, that third player lurking ominously in the background. The
new tax revenue of $80 is exactly offset by the new tax shield of $80. The government’s cash flow
has not changed. Pretty boring.

!31
 It should be obvious from the graph that the reason that the new tax rule is neutral is be-
cause borrowers and lenders face the same tax rate. When the two tax rates are the same, the de-
crease in the supply of loans is exactly offset by the increase in demand, and the equilibrium
quantity of loans does not change. For the policy to be neutral, the two tax rates do not have to be
50 per cent; they just have to be the same.
Things get interesting (well, if you’re an economist) if lenders and borrowers are in different
tax brackets. If lenders face a higher tax rate than borrowers, the supply of loans will decrease
“more” than the demand for loans will increase, and the loan market will shrink. If lenders face a
lower tax rate than borrowers, the opposite happens. Here is an example of the first case. Bor-
rowers are taxed at 35 per cent and lenders at 55 per cent. For the equilibrium quantity of loans
to stay put at $1,000, lenders require a interest rate of 17.78 per cent (point u), but borrowers will
not borrow that amount if the new interest rate is anything higher than 12.3 per cent (point l).
The new policy will cause supply to decrease “more” than demand increases, causing the new

r
Td = 0.35; Ts = 0.55
0.2
0.1778 u
0.16
0.1455 e′
l
0.1231

0.0945 d
e
0.08
0.0655 s
0.04

q $ loans
0 500 818 1000 1500 2000

equilibrium, e’, to lie to the left of the old one. I calculated the new equilibrium quantity of loans
to be $818 and the new equilibrium interest rate to be 14.5%. But I was able to find the intersec-
tion of demand and supply only because I knew their equations. In real life, we don’t know what
the supply and demand curves for loans looks like, and estimating them is a crap shoot. What we
do know, however, is the location of points l and u, and it is plain to see that the new interest rate
must lie somewhere between the rates at l and u, 12.3 per cent and 17.78 per cent. The finance
minister will not be impressed when you tell him that your best forecast of the effect of a policy
change puts the interest rate in a five per cent range, but at least you’re being honest about it. The
minister will tell the public whatever he or she wants it to hear.
Because the loan market has shrunk, loans must have lost some of their appeal to borrowers
and lenders compared to other forms of finance and investment. Look at points d and s, which lie
directly below the new equilibrium. Those borrowers who continue to borrower after the policy
change, now pay more for their loans after tax, 9.45 per cent (point d). And lenders who continue
to lend earn less, 6.55 per cent (point s) down from eight. If borrowers pay more and lenders earn

!32
less, the government must be generating a positive cashflow for itself. That cashflow is $23.80,
which comes from $65.45 in new tax revenue less $41.65 in new tax shields. I’ll leave it to you to
do the arithmetic to confirm those numbers and identify the area in the graph that represents the
$23.80.
Switch the tax rates of borrowers and lenders and rework the analysis. Explain it to a loved
one. They’ll either thank you for it or dump you for someone who wears interesting shoes.

Subsidized Daycare
When my daughter attended Concordia University’s Garderie Les Petit Prof in the early
1990s, I think my wife and I paid about $23 a day. Most registered daycares in Montreal at the
time were charging between $20 and $25. The picture shows a stylized equilibrium with a market-
clearing price of $25 and 100 children attending the various registered daycares in Montreal.

50

e
25 Subsidy = ?

Price ceiling e′
7

0 qkids
0 28 100 172 200

Then in 1997 the Quebec government set a price ceiling of five dollars a day, to help parents bal-
ance work and childrearing and to make daycare more accessible to those on low incomes. The
idea was that having access to daycare would allow single parents and stay-at-home parents of
two-parent families to find jobs and better provide for their families. More people working would
boost the economy through increased spending. The ceiling was raised to seven dollars a day in
2004.
No daycare can survive on seven dollars a day per child, and if fixing the price is all that the
government did, there would be a severe shortage of spots. The figure shows that at seven dollars,
there would be demand for 172 spots but only 28 would be offered. If there was some way to ac-
curately predict that 172 spots would be demanded at seven dollars a day, the government could
give daycares a subsidy that would guarantee that everyone who wanted a spot would get it.9 The

9This type of subsidy is called a quantity subsidy because the amount is fixed at some amount per child per
day.

!33
magic subsidy is a question mark. A smaller quantity subsidy would still leave a shortage, but the
government may be satisfied as long as the number of spots increases above the previous level of
100.
One consequence of fixing the price of daycare is that it discourages variety. Some daycares
are more costly to run, and so charge higher fees, because they offer better or more varied ser-
vices. I don’t know if the variety in daycare services runs from no-frills to Rolls Royce, but I am
sure there are some differences from one to another. The ones that are more costly to run and that
fall under the government’s jurisdiction had to cut costs by cutting back on services in order to be
viable. Fewer field trips, that sort of thing. Daycares look more the alike, especially when it comes
to “ratios,” the number of children supervised by each caregiver. Daycares now maximize their
revenue by taking in as many children allowable by law. As a matter of fact, the government re-
quires it (so I’m told, but I have to fact-check this). Previously a selling point of some daycares was
their low ratios; your child would receive more attention. Is there a benefit to this homogeniza-
tion? It is estimated that in 2011, 215,000 kids were being cared for at an expected subsidy of
$2.215 billion or $10,302 per child per year or $28 per child per day, and that 70,000 more moth-
ers were able to hold jobs, translating into an increase of 1.7 per cent in Quebec’s GDP.10 The
researchers generating those estimates argued that the benefits of the price ceiling and accompa-
nying subsidy to the Quebec economy outweighed the cost. Other provinces often pointed to the
Quebec model as an example of successful “nationalized” daycare.
Seven-dollar-a-day daycare has since been abandoned. What changed?

10Pierre Fortin, Luc Godbout, and Suzie St. Cerny, 2012, “Impact of Quebec’s Universal Low-Fee Child-
care Program on Female Labour Force Participation, Domestic Income, and Government Budgets,” Work-
ing Paper, University of Sherbrooke.

!34
Reading
“Pro-market, Not Pro-business,” The Economist, June 28, 2003.
“A Simple Lesson in Economics,” The Economist, February 1, 2003.
“The Avuncular State — Libertarian Paternalism,” The Economist, April 8, 2006. [Special report]
“A Blunt Instrument (The Minimum Wage),” The Economist, October 26, 2006.
“In Higher Education, A Free-for-all Does not Work,” The Economist, October 28, 2006. [Survey:
Lessons from the Campus]
Srigley, Ron, “Pass, Fail,” The Walrus, March 18, 2016.
Chapter 6 in Varian. [Optional]

!35
 7
MARKET FAILURE

A market fails when price does not reflect true value in the sense that a price ratio formed
with that price is not equal to peoples’ marginal rates of substitution. In the extreme, the market
may collapse or not exist to begin with because either there are no consumers willing buy at the
“wrong” price or there are no sellers willing to sell. To understand why a market might fail, we
need to look back at the First Welfare Theorem: A competitive equilibrium is Pareto optimal.
People (and firms) behaving as price takers is a sufficient condition for a Pareto optimum. But
there are a number of characteristics or defects that can cause an otherwise competitive market to
be, well, less than competitive, scuttling the optimum.
What are these characteristics? One is economies of scale. If a firm can lower its costs with-
out end simply by getting bigger, it will come to dominate a market and wield monopoly power.
Too much market power is seen as a bad thing for this very reason, and that is why we have laws
that protect competition. The second concerns who knows what. In real life, we do not all have
access to the same information; there is information asymmetry. Sellers usually know more about the
quality of their products and services than buyers do; people buying insurance know more about
the risks they face than insurers do. What does this unequal distribution of information mean for
the price of products and services? The third characteristic is the unavoidable by-products of our
consumption and production. These are called externalities. Smoke a cigarette (that’s consumption),
and the poor fella next to you suffers, but it’s not like he can ask you to compensate him for the hit
to his utility. Well, he can ask, but he probably won’t get very far. He may not have a right to clean
air, and even if he did, how would he enforce the right? The question of compensation, of course,
also applies to our primary production, not just inadvertent by-products. We want to be paid for
what we produce. But what if the thing that you produce can be freely consumed by anyone who
wants to—a book, a song, computer software—and you have no easy way to make them pay you
or to stop consuming if they won’t. Goods that are seemingly free to those who would consume
them are called public goods. Public goods include infrastructure, such as roads, intellectual proper-
ty, such as books, music, computer software, and scientific research, which may be produced pub-
licly or privately. The fifth cause of market failure is extreme inequality in the distribution of en-
dowments—income or wealth. It is possible for a Pareto optimum in the Edgeworth box to be one
in which Lucy is filthy rich, and Ricky is so poor that he can barely afford essentials like housing.
No one said that a Pareto optimum has to be fair.
This chapter will take a brief look at three of the five causes of market failure: information-
al asymmetries, externalities, and public goods. Many examples will be discussed in class, but you
can, of course, find examples for yourself. When markets are prone to failure, government inter-
vention may be justified.

!36
The Lemons Problem
Professor George Akerlof asked, “Why are all used cars lemons?” Information asymmetry was
the answer. Sellers of used cars know more about the quality of what they are selling than buyers
do. If buyers cannot easily tell the difference between a good used car and an otherwise identical
lemon, the price buyers will offer will be somewhere between the unknown true values of the two.
Anticipating a single price for all used cars, rather than a high price for good ones and a low price
for lemons, sellers don’t bother putting good used cars up for sale. Buyers, then, anticipating that
good used cars will be nowhere to be found, offer only the lower, lemon price. The middle price
never materializes because rational expectations is at work here. Where there could have been a
market for good used cars selling at the higher price and another market for lemons, selling at the
lower price, there is only a market for lemons. Good used cars suffer adverse selection: the bad has
driven out the good. The only way to establish a market for good used cars is for the owners to
help potential buyers distinguish between their good used cars and lemons. That takes money and
effort, both of which fall on sellers of good used cars.
Now, if you replace good used cars with bright students and lemons with, say, mediocre
students, both looking for jobs, how would you retell the story in a world of grade inflation? Enjoy.

Moral Hazard
You lock your garage every night so that your bicycle doesn’t get stolen. Would you be as
cautious if you had insurance that covers theft? Economic theory says no. Taking care of your
things requires the kind of effort you’d rather avoid. And why bother anyway, the bike is covered?
If the insurance premium that you paid reflects the risk of your bike getting stolen before you de-
cided to slack off, it is now too low. By being less diligent after being insured, you have tilted the
deal in your favour, and ripped off the insurance company. That is moral hazard. But have you real-
ly ripped off the insurance company? Insurers are not dumb. They anticipate that you, and
everyone else, may be less careful after buying insurance, so they jack up everyone’s premium in
anticipation of moral hazard. So even if you are not homo economicus and would never dream of
leaving your garage unlocked, you will pay for the possibility that you might. That’s rational ex-
pectations at work again in the face of asymmetric information. Good risks pay higher premiums
than they should because insurance companies do not know as much about your risks as you do,
and how you might purposely change those risks once you are insured. Does that mean that good
risks are adversely selected? Yes, in that they are charged too much. But is the the adverse selec-
tion so severe that many people decide to go uninsured, leaving a market of lemon risks? No, be-
cause the premium for a market of nothing but lemon risks might be astronomically high. There’s
a primitive and fairly effective fix for moral hazard, at least when it comes to insurance: deductibles.
If your $800 bike gets stolen, you’ll only be covered for $600. I wonder if something like a de-
ductible could have worked in financial markets to prevent the mortgage brokers, bankers, and
real estate agents from engaging in moral hazard and causing the sub-prime loan crisis.

!37
Externalities
Pollution is a by-product of manufacturing cars, just as it is a by-product of manufacturing
most things. Pollution hurts surrounding communities, sometimes even faraway ones. If car com-
panies had the right to pollute by law—a property right—and people affected by the pollution
could get together and negotiate with a car company to pollute less, say, by offering some amount
of money, then all would be well economically. People wouldn’t offer an amount that was greater
than the value they place on the reduction in pollution, which presumably would come from a
reduction in the number of cars produced or investment in cleaner manufacturing processes. And
the car company wouldn’t accept an amount that was less than the value of lost production or the
cost of a cleaner way of doing things. Mind you, the government could have given the people the
right to a minimum level of clean air rather than giving the car company the right to pollute. In
that case, the people have the property right, and the car company might offer to pay the com-
munity something in return for allowing it to increase production, if it were worthwhile. In either
case, a market is effectively putting a price on pollution. Professor Ronald Coase11 showed, in
what is known as The Coase Theorem, that the “socially optimal” level of production of something
like cars does not depend on who is allocated the property right for an externality, such as pollu-
tion, provided that the two sides can negotiate the outcome and compensate each other to their
mutual satisfaction. Make the property rights clear in law, and then let people work it out for
themselves.
In real life, property rights are not always clear, in which case pollution does not have a
price, and car companies probably produce more cars than is socially optimal because the costs to
society of pollution are not a cost of producing cars. But even if property rights are clear, negotia-
tion between the two sides is tricky. One car company, many people in the community. How do
the people in the community coordinate their effort to negotiate? If some people in the communi-
ty choose not to participate, they still enjoy the benefits of a reduction in pollution negotiated by
the neighbours. The incentive to free ride is strong. You probably come up against free riders every
time you have to do a group project for one of your courses.

Public Goods
What if you produce something for sale but people can consume it without paying. Your
work is treated as if it were free. All sorts of intellectual property fits the bill: music, art, books and
other writing, computer software, scientific research and inventions. All of these take a great deal
of effort or cost a lot to produce but can be consumed easily because one person’s consumption
does not affect another person’s consumption (my listening to a song never prevents you from lis-
tening to it), and it is difficult for the producer to exclude anyone from consuming. Without fair
compensation, these goods will be underproduced because the incentive is not there, and those
that are produced will be over-consumed. Even goods that are not free because they are paid for
with our tax dollars have the appearance of being free will be over-consumed. Just look at the
congestion on our roads.

11 Professor Coase died September, 2013, aged 104.

!38
 Goods that are over-consumed and underproduced because they have the appearance of
being free are called public goods. Copyright and patents are the imperfect fix for the problem that
producers of intellectual property have in extracting fair compensation. Both provide legal protec-
tion of ownership in the form of a monopoly for a fixed period. But enforcement of the property
right is still a problem. For public infrastructure such as roads, user-pay and pay-per-use are the
remedies. Hey, suburbanites, ready for a toll on the next Champlain bridge?

!39
Reading
Akerlof, George, 1970, “The Market for ‘Lemons’: Quality, Uncertainty, and the Market Mecha-
nism,” The Quarterly Journal of Economics, 84(3):488-500.
“Secrets and Agents,” The Economist, July 23, 2016.
Shulman, Seth, “Patent Absurdities,” 1999, The Sciences, 39(1):30-33.
Stavins, Robert, 1998, “What Can We Learn from the Grand Policy Experiment?: Lessons from
SO2 Allowance Trading,” Journal of Economic Perspectives, 12(3):69-88.
Srigley, Ron, “Pass, Fail,” The Walrus, March 18, 2016. (Also appearing in chapter 6.)
“Genetic Patent Singles Out Jewish Women,” Nature, July 2005.
“An Eye for an A,” The Economist, March 7, 2002.
“D-graded,” The Economist, September 1, 2001.
Crook, Clive, “A Matter of Degrees,” The Atlantic, November, 2006.
Chapters 35 and 36 in Varian. [Optional]

!40
 8
INTERNATIONAL MARKETS

International trade can make people better off when their country manufactures goods whose
costs are low compared to that of other countries. It’s that simple. As for currencies, translating
one to another isn’t rocket science, but the difference in interest rates between two countries sends
a message about how the market expects an exchange rate to change.

Gains from Specialization and Trade

Wine and wool sweaters are made in both England and Italy. Suppose that England can
produce as much as 30,000 litres of wine or as many as 2,000 sweaters in a year. Italy can produce
48,000 litres of wine or 2,400 sweaters. Both countries, of course, produce some mix of the two
goods. Suppose also that the number of litres of wine that each country must give up to produce
an extra sweater does not depend on the mix of sweaters and wine it is producing. That means
that one more sweater always costs 15 litres of wine in England and 20 in Italy. The opportunity

Wine
48 000

30 000 PPC England

PPC Italy

Sweaters
2000 2400

cost of a sweater in England is 15 litres of wine, and the opportunity cost in Italy is 20. Because
this about production, you could also say that the marginal rate of transformation of wine for
sweaters is 15 in England and 20 in Italy. Constant marginal rates of transformation mean that
England and Italy’s production possibility curves are straight lines.
If sweaters can be made more cheaply in England than Italy, wouldn’t there be a gain if
England made more of them and Italy fewer? You bet. If England makes an extra 100 sweaters, it

!41
will cost 1,500 litres of wine. If Italy makes 100 fewer, it will be able to increase its output of wine
by 2,000 litres. The number sweaters in the world wouldn’t change, but there’d be 500 more litres
of wine to drink. The gain is equal to five litres of wine per sweater, the difference between Eng-
land and Italy’s marginal rates of transformation.

UK Italy Gain

∆Wine in litres -1,500 2,000 500 litres

∆Sweaters 100 -100 0

MRT litres per sweater 15 20 5 litres

If there is a gain of 500 litres of wine by shifting production of 100 sweaters from Italy to
England, why not shift 200, 300, or more? The more England specializes in sweaters and Italy in
wine, the greater the gain to the world. The biggest gain is had if England stops making wine al-
together and specializes in nothing but sweaters, and Italy manufactures nothing but wine. Al-
though, this is true only because we assumed that their marginal rates of transformation are con-
stant. Normally as a country shifts more of its production to the good in which it has a compara-
tive advantage, the cost of doing so will gradually increase, so partial specialization normally rules
the day.
No one is going to enjoy a single drop of that 500 litres of found wine unless the two coun-
tries trade. England must export sweaters to Italy in return for vino. (Or Italy must export wine to
England in return for sweaters. It doesn’t matter how you say it.) Think about it. England is pretty
dry sitting on nothing but a big pile of sweaters, and everyone in Italy is shivering during the win-
ter. Brits gain as long as they can sell sweaters for more than it costs to make them, 15 litres of
wine each; and Italians are happy to buy English sweaters if price is less than the 20 litres of wine
it costs of make one at home.
If the two countries are friendly, they’ll certainly explore the possibility of trading with one
another. A world price ratio of litres of wine per sweater, known as the terms of trade in in the jar-
gon of international economics, will be determined by the tastes and incomes of the English and
Italians and their production possibilities. Suppose the terms of trade are 17, and that England
exports 1,000 sweaters to Italy. England will be ahead 2,000 litres of wine because Italy paid
17,000 litres for sweaters that cost 15,000 litres for England to make. Italy will be ahead 3,000
litres of wine because it would have cost 20,000 litres of wine to make those sweaters at home.
The gains to England and Italy have to add up to the gain to the world: 5,000 litres. It’s not hard
to see that the closer the terms of trade are to England’s cost (MRT), the smaller the gain to Eng-
land and the bigger the gain to Italy. And you would never expect to see a price ratio outside of
the two MRTs; otherwise, one of the two countries would be losing wine for every sweater bought
or sold.

!42
 Wine
48 000

CPC England
30 000

CPC Italy

PPC England

PPC Italy

Sweaters
1000 2000 2400

By specializing its production in the goods in which it has a comparative advantage and
then trading internationally, each country can consume along it consumption possibility curve resulting
in a Pareto improvement.12 All free trade agreements are based implicitly on this idea.

Exchange Rates
Suppose that the interest rate is three per cent in Canada and five per cent in France. What
do you make of the difference? Both rates are for one year and are risk-free. They are rates that
you would earn on safe government bonds.
Invest $1 million for a year here at home and your return is sure to be $1,030,000. If in-
stead you invest that $1 million in France, you’ll first have to buy Euros because French govern-
ment bonds are priced in Euros. Say that the Euro spot price is $1.3021. One million Canadian
dollars will buy 767,990 Euros. Those Euros, invested at the five per cent French risk-free rate, will
return €806,390 at the end of the year.
One million dollars is guaranteed to grow to CAD $1,030,000 or €806,390. What would
you do? Invest in Canada or France? It’s not clear because if you invest in France, you will have to
sell the Euros for Canadian dollars at the end of the year so that you can buy smoked meat in
Montreal. The decision would be easy if you knew today what the value of the Euro was going to
be a year from now. But you do not, and that uncertainty about the exchange rate makes the
French option risky even though the return on French government bonds is risk-free. If the Euro
doesn’t budge, you’ll bring home $1,050,000 (obvious, right?). If it falls by two cents, $1,033,872,
which is $3,872 more than you’d earn in Canada. But if it falls by three cents, you’ll be left with
$1,025,808, and it would have been better to have kept your money in Canada. Most people don’t
like risk; they need to be compensated to bear it. And if you’re like most people, you might not be
willing to invest in France even if you thought there was little chance of the Euro depreciating by
three cents.
Ignore risk for a minute—I know, not realistic at all—and pretend that all you care about is
earning as much as possible (what do we call that kind of attitude towards risk?). What would the

12 Can you pick off the gains to each country in the picture?

!43
spot price of the Euro have to be one year from now for you to break even? It would have to be
$1.2773 because selling €806,389.68 for $1.2773 each gives $1,030,000.

$1,030,000
" = $1.2771 × €806,389.68

It probably doesn’t surprise you that $1.2773 is about two per cent less than the current Euro spot
price of $1.3021. The interest rate in France is two percentage points higher than in Canada, so
the Euro would have to depreciate by about two per cent to cancel France’s interest rate advan-
tage. Time for some notation.

rd 0.03 Canada’s risk-free interest rate (d for domestic)

rf 0.05 France’s risk-free interest rate (f for foreign)

p0 CAD $1.3021 Spot price of the Euro today

p1 CAD ? Unknown spot price of the Euro one year from now

" 1̂
p CAD $1.2800 Your personal one-year forecast for the Euro

f0 CAD $1.2500 Today’s actual one-year forward price of the Euro

Now let’s work out the details. The total return on $1 million invested in Canada at three per cent
is

$1,000,000(1
" + rd ) = $1,000,000(1.03) = $1,030,000

Or buy CAD $1 million worth of Euros

$1,000,000 $1,000,000
" = = €767,990.17
p0 $1.3021

Invest them in France at five per cent for a total return of

"€767,990.17(1 + rf ) = €767,990.17(1.05) = €806,389.68

and then plan to sell those Euros in a year at a yet unknown spot price

"
€806,389.68 p1 = €767,990.17(1 + rf )p1

To calculate what the spot price a year from now would have to be to break even, equate the re-
turn in Canada with the return in France

$1,000,000
$1,000,000(1 + rd ) = (1 + rf ) p1
p0
"
1 + rd 1.03
∴ p1 = p0 = $1.3021 × = $1.2773
1 + rf 1.05

!44
The calculation is done using the three known current prices—the two interest rates and the spot
price of the Euro. If these three prices are equilibrium prices in the supply-equals-demand sense,
then every investor must have already decided to keep their money here or invest in France. Equi-
librium means that everyone is content keeping their money wherever they have decided to keep
it, and because of that, it must also mean that the market expects to break-even, E " ( p1) = $1.2773.
If this weren’t true, people would still be moving their money around to exploit some perceived
advantage and the three known prices would still be in flux; they would not be equilibrium prices.
The equilibrium relationship between the interest rates in two countries and the exchange rate is
known as uncovered interest rate parity. (rd, rf and E(p1) should really have * attached to them to indicate
that they are equilibrium prices.)

$1
$1(1 + rd ) = (1 + rf ) E ( p1)
p0

The interpretation is neat. If the interest rate in France is higher than the interest rate in Canada,
the market must be expecting the Euro to depreciate ($1.2773 is less than $1.3021). If France’s
interest rate is lower than Canada’s, a Euro appreciation is expected. The difference between the
two interest rates is a market forecast of the future exchange rate! To see the relationship in per-
centage terms, put the two interest rates on one side of the equal sign and the spot prices on the
other, and then subtract one from both sides

1 + rd E ( p1)
−1= −1
1 + rf p0

The difference in interest rates

1 + rd 1.03
" −1= − 1 = − 0.0190476 ≈ rd − rf = − 0.02
1 + rf 1.05

E ( p1)
predicts the expected percentage change in the exchange rate, " − 1, a depreciation of 1.9
p0
per cent. The simple difference, r" d − rf = -2 per cent is always a good approximation when the two
interest rates are not too far apart. You may earn two per cent more on French government bonds
but on average you will lose that amount when you convert Euros back to Canadian dollars.
Greed keeps the parity in uncovered interest rate parity. You are part of the reason that the
market expects the Euro to depreciate to $1.2773. You believe that the Euro will not fall by as
much. Your forecast is $1.28. Being the gutsy individualist that you are, you speculate by borrowing
$1 million in Canada at three per cent and use it to buy French government bonds paying five per
cent.13 Your expected profit is $2,178.79. Let p1̂ = $1.28 be your forecast for the Euro, which re-
places the expected price implied by the market

13 By gutsy, I mean risk neutral. Gutsy sounds so much less clinical, don’t you think?

!45
 $1
$1,000,000(1 + rd ) < (1 + rf ) p1̂
p0
" $1 $1.28
$1,000,000(1.03) < (1.05)
p0 $1.3021
$1,030,000 < $1,032,178.7881

The left-hand side is what you owe on your loan and the right is the expected return on your in-
vestment. The difference is your expected profit. Hope things work out for you. If the Euro falls
before $1.2773, you’ll be out of pocket. There are other speculators like you who believe the Euro
will end up even higher than $1.28. They sit above you on the demand curve for Euros. Still oth-
ers believe that the Euro will fall below $1.2773, and they speculate by borrowing Euros at five per
cent to invest in Canada at three. They are on the lower part of the supply curve for Euros. To-
gether, all of you speculators determine the equilibrium for the two interest rates and the current
spot price.
If there is also a forward market for currencies, covered interest rate parity is possible. The equi-
librium forward price of the Euro must to be equal to the expected future spot price or else it is
possible to make a completely riskless profit—just like printing money.

f 0* = E ( p1)
$1
$1(1 + rd ) = (1 + rf ) f 0*
p0

To see why the forward price must be equal to $1.2773, assume that it is temporarily out of
whack, say, "f0 = $1.25. A forward price of $1.25 means that you can contract now to buy or sell
Euros in one year at $1.25 no matter what the spot price turns out to be. I would borrow 767,990
Euros at five per cent to invest in Canada at three per cent and at the same time buy 806,390 Eu-
ros forward at $1.25 because that is how many Euros I will need to pay off my loan. That leaves
me a guaranteed profit of $22,012.90. Follow the inequality. It’s all there.

$1
$1,000,000(1 + rd ) > (1 + rf ) f0
p0
$1,000,000
" $1,000,000(1.03) > × 1.05 × $1.25
$1.3021
$1,000,000(1.03) > €767,990 × 1.05 × $1.25
$1,030,000 > $1,007,987.0978

Earn $30,000 with a loan that cost you about $8,000. Earning a riskless profit like this without
putting up any of your own money is, as you know, arbitrage. The two investments, Canada or
France, should be perfect substitutes for all investors—not just risk neutral investors—because they
are both one-year riskless investments. Perfect substitutes must have the same price or, saying the
same thing, yield the same return. If the forward exchange rate is $1.25, they do not, and an at-
tractive profit opportunity exists. Of course, everyone knows that the forward rate should be
$1.2773 just by looking at the interest rates and the spot price. So it should not be surprising that

!46
these sorts of opportunities are rare. The absence of arbitrage profit is another way of knowing
that the markets are in equilibrium.

Reading
Froot, Kenneth and Richard H. Thaler, 1990, “Anomalies: Foreign Exchange,” Journal of Economic
Perspectives, 4(3):179-192.
“McCurrencies,” The Economist, May 27, 2006. (Usually published every year.)

!47
 9
INTERTEMPORAL MARKETS

A capital market may be as simple as the opportunity to borrow and lend. Borrow to con-
sume more now at the cost of having less later; lend to consume more later by giving up some
now. Having the choice of shifting consumption across time leaves us better off. The price of con-
suming $1 worth of anything today is the rate of interest. That is why the rate of interest is called
the time value of money. The opportunity to borrow can also leave us better off by making us
wealthier when the money raised is used to finance real production that earns a higher return
than the rate of interest.

Bonds
A bond is a loan to a company. A financial institution, often an investment bank, lends
money to a company, which in turn promises to make periodic interest payments, usually semi-
annually, and a lump sum repayment of the principal at maturity. The investment bank chops up
the loan into many smaller loans, which it offers to sell to its clients, so that they can collect inter-
est payments and a proportional lump-sum payment at maturity. A $50 million loan to a company

Par or face value $1,000

Coupon rate (c) 0.05

Coupon (C) $50

Interest rate (r) 6.5%

Maturity (T years) 25

Compounding (m times a year) 2

Present value of coupons $613.79

Present value of par $202.07

Bond value (P) $815.86

might be distributed by the investment bank to its clients as 50,000 bonds, each one a loan of
$1,000 (the par value or face value of the bond), with interest payments, known as coupon payments,
computed on the face value. Some bonds pay only interest and others only a lump sum at maturi-
ty. Bonds are traded.

!48
 Retail investors like you and me can buy bonds in the secondary market. Bonds are risky.
The price of a bond at any time depends on the market’s expectation that the issuing company
can make the interest payments as they come due and the repay of the face value at maturity.
Changes in the price of a bond caused by changes in a company’s prospects is called default risk.
The price of a bond also changes when the interest rate or the yield on bonds of similar risk
changes. This is because alternative investments may have become more or less attractive.
Changes in the price of a bond caused by changes in interest rates is called interest rate risk.
The bond described in the table matures in 25 years and pays interest at a rate of five per
cent on a par value of $1,000. That’s 50 payments of $25 because the coupons are paid every
semi-annually. At the end of 25 years, a lump sum payment of $1,000 is made in addition to the
last coupon payment. The price of the bond today is $815.86, which is the sum of the present
value of the coupons, $613.79, and the present value of the par, $202.07, both cash flows having
been discounted at 6.5 per cent. The coupons are an annuity, and that’s how they are treated in
the first term of the bond valuation equation below. The present value of the face value is a
straight forward discounting of the single amount. Everything rides on that 6.5 per cent discount
rate. It is an opportunity cost, which in bond circles is called the required yield or yield-to-maturity.
Where does it come from? You can’t look it up. As an opportunity cost, it must be the rate that
could be earned on an investment of the same risk, say, the bonds of another company in a simi-
lar line of business. But how was the yield-to-maturity of the bonds of the other company deter-
mined? Going around in a circle.

−mT
1 − (1 + m)
r
C F
P = r + mT
(1 + m)
m r
m

For our bond,

$613.79
" + $202.07 = $815.86

The 6.5 per cent yield is really the bond’s price just as $815.86 is the bond’s price. There’s a
dollar price on the left-hand side of the equation and a price in per cent on the right, which
means that the equation is saying that $815.86 today is worth the same as a series of $25 dollar
payments and a lump-sum payment of $1,000, 25 years down the road, because that cash flow
has a price of 6.5 per cent. So take your pick: supply and demand for bonds with quantity on the
horizontal axis and dollar price on the vertical axis or per cent yield on the vertical axis. It’s all the
same. But lurking within the supply and demand schedules are investors’ tastes for risk and return,
and their expectations regarding the company’s future prospects and ability to make the pay-
ments.
Does buying a bond make you wealthier? You already know the answer to that: no if the
market is informationally efficient. In buying the bond, you are giving up $815.86 now for a fu-
ture cash flow whose value today is, to beat the point to death, $815.86. And of course, the same

!49
goes for the seller. It makes sense. In buying and selling financial securities, we are not producing
anything of value; we are only moving money across time. Then why do it? See section 3.

The Term Structure of Interest

Suppose the risk-free, four-year spot rate of interest is 9.75 per cent. It is the rate that you
might earn (might have earned many years ago) on an insured deposit, such as a guaranteed in-
vestment certificate or a treasury bond, if you were willing to tie up your money for the full four
years. One dollar will grow to $1.46344 in four years if the interest is compounded semi-annually.

2×4

( 2 )
0.0975
$1
" 1+ = $1.46344

Now, what if the risk-free, two-year spot rate of interest was 9.5 per cent? You could invest or lend
one dollar for two years and earn $1.20397.

2×2

( 2 )
0.095
" $1 1+ = $1.20397

Which would you choose: $1.46344 in four years or $1.20397 in two? The two investments can’t
be compared as because they don’t pay off at the same time. Anyone would naturally say, “It de-
pends on what I think I could earn in years 3 and 4 if I were to invest for two years and then rein-
vest my principal and interest for another two years.” But you can’t know now what the two-year
spot rate of interest will be in two year’s time. The two-year investment rolled over for another

9.75% $1.46344

E r * = 10.0003% $1.46344

9.5% $1.20397

t
0 2 4
two is risky, while the four-year investment is not. We’ll have to fudge this by assuming that you
and all other investors are risk neutral, so that all that anyone cares about is earning as much
money as possible at the end of four years, regardless of the risk. Now the investments can be
compared.
The two investments are equivalent if we expected the future two-year spot rate for years 3
and 4, E(r*), to be just a hair over 10 per cent.

2×2 2×2 2×4

" ( ) ( 2 ) ( 2 )
0.095 E (r*) 0.0975
$1 1 + 1+ = $1 1 + = $1.46344
2
⟹ E (r*) = 0.100003

We know three things: the short-term interest rate is 9.5 per cent; the long-term rate is 9.75 per
cent; and people act on their expectations. If the two interest rates are equilibrium interest rates,

!50
in the supply-equals-demand sense as always, then money is not being shifted from short-term to
long-term or vice versa. Those who expect the future two-year spot rate to be higher than 10 per
cent have invested short term; those who expect it to be lower than 10 per cent have invested or
lent long term. And those dollars that lie at the two intersection of the supply and demand sched-
ules represent investors who are indifferent between investing short term or long term. They must
expect the short-term investment rolled over to yield the same return as the long-term investment.
That's why the break-even rate, the equilibrium expected future spot interest rate (the word future
is redundant I think), is 10.0003 per cent.
This is neat because the two spot interest rates, which exist and which we can transact with,
tell us what the market expects will be the short-term spot rate in two year’s time, in this case,
10.0003 per cent. You can now say why the long-term interest rate is higher than the short-term
rate: the market expects the short-term rate to rise in the future. If the four-year rate was lower
than the two-year rate, the market must be expecting the short-term rate to fall. This is known as
the Expectations Theorem of the Term Structure. The term structure or yield curve refers to the shape of
the graph of interest rates y-axis against their terms, two years or four. In our example, the yield
curve is upward sloping, which is usual, but there is nothing unusual about a flat or even a down-
ward-sloping yield curve since the difference between the short and long rates is determined by
expectations.
If you thought that the two-year spot rate in years 3 and 4 would climb to 10.25 per cent
from its current 9.5 per cent, then you would take the chance by investing short-term and rolling

9.75% $1.46344

r = 10.25% > E r * $1.47042

9.5% $1.20397

t
0 2 4
over the investment for another term because the expected total return is 1.47042, and that’s
more than the $1.46344 you get by investing long term.

2×2 2×2

( 2 ) ( 2 )
0.095 0.1025
"$1 1 + 1+ = $1.47042

What if you don’t have a dollar to lend (darn those video lottery terminals)? No problem, borrow
it at the four-year rate. At the end of four years you will owe $1.46344 which you are expecting to
be able to repay from a return of $1.47042 for an expected profit of $0.00697287 or $6.97 per
$1,000. If the future two-year interest rate ends up even higher, your smile will be bigger; on the
other hand, you’ll be in trouble if it doesn’t climb high enough. What would your speculative
strategy be if you expected the future two-year spot rate to be 9.8 per cent?
If a forward market for borrowing and lending exists, then the forward rate of interest must
be equal to the equilibrium expected future spot rate implied by the expectations theorem

f" * = E (r*) = 0.100003

!51
because if it wasn’t, you and everyone else could earn a riskless profit using a strategy called arbi-
trage. To see this, suppose that, by some glitch, the actual two-year forward interest rate is 9.9 per
cent instead of its equilibrium value that I claim should be 10.0003 per cent. You would borrow
one dollar for two years at 9.5 per cent and invest it for four years at 9.75 per cent. You already
know that at the end of four years you will receive $1.46344. But how will you repay the loan
after just two years? That’s $1.20397 you need to come up with. The answer is to borrow for-

9.75% $1.46344

f = 9.9% < E r * $1.46065

9.5% $1.20397

t
0 2 4
ward. This means that today you arrange a two-year loan of $1.20397 to start in year 3, at the
forward rate of 9.9 per cent. What you have done is rolled over or refinanced your original loan at
a rate that is fixed in advance in your favour. At the end of year 4, you will owe $1.46065 for sure.

2×2 2×2

( 2 ) ( 2 )
0.095 0.099
$1
" 1+ 1+ = $1.46065

Earn $1.4634 and repay $1.46065 for an arbitrage profit of $0.00279382 per dollar played or
$2,793.82 per $1 million. Too good to be true? Pretty much, or at least too fleeting to be captured,
because if you can do it, so can everyone else. The efficient markets hypothesis tells us that arbi-
trage profits should not exist! That is why the forward price is always equal the equilibrium ex-
pected future spot price: f" * = E (r*).

!52
Optimal Consumption
Like most economic theories, utility theory tries to be one-size-fits-all. It can be used to de-
scribe someone’s choice of how much to spend now and in the future just as easily as we used it to
describe their choice of goods, at any given time, as we did in chapter 2. Good x becomes C0, for
current consumption or the dollars that you have today, and good y becomes C1, for future con-
sumption or the dollars you will have next year. You can’t consume dollars directly, of course, but
the more you have, the more real stuff you can consume, and it really doesn’t matter what the
stuff is. The graph shows Lucy’s intertemporal consumption optimum. Her tastes are represented
by a Cobb-Douglas utility function, u" = C0αC01−α with α " = 0.4. Dollars plotted against dollars may
seem strange at first, but they are different things because one is dollars today and the other is dol-
lars next year. (And speaking of one-size-fits all, there is nothing stopping us from extending the
dates to year 2, 3, 4, right up to when Lucy kicks the proverbial bucket, and even beyond if she
C1
W1
1270

C* : MRS = 1 + r = 1.07
762

Earn $562 in interest u* = 630.616

e ue = 380.731
200
Lend $525
W0
C0
0 475 1000 1187
wants to leave an inheritance to her pet hamster, Francesca.) We’ll assume that the future is cer-
tain despite the fact that it is not. This allows us to consider Lucy’s choice as involving only her
preference for consumption over time independent of her preferences for risk.
Lucy has an endowment of $1,000 today and $200 next year. At a seven per cent rate of
interest, her wealth is $1,186.92, which is where her budget line crosses the horizontal axis. The
budget line or wealth constraint is exactly the same as it was in chapter 2 except that it is now rep-
resents choices for consumption dollars rather than goods x and y, so

W
" 0 = px x + py y

becomes

W
" 0 = p0C0 + p1C1

!53
and the only thing that needs to be done is to interpret p0 and p1. If C0 is current consumption
measured in dollars, then p0 must be 1 because the value of one dollar today is, well, one dollar.
And if C1 is consumption next year but wealth is being measured in dollars today, then p1 must be
1
the present value of one dollar, " , where r is the risk-free rate of interest. Substituting the in-
1+r
terpreted prices into the wealth constraint gives

C1 $200
" 0 = C0 +
W = $1,000 + = $1,186.92
1+r 1.07

which says, very neatly, that your wealth is the present value of the cash flow stream to which you
are entitled over the course of your life. Now rearrange the equation so that it is in the standard
format for an equation for a line

" 1 = − (1 + r)C0 + (1 + r)W0 = − (1 + r)C0 + W1

C

The price ratio is "1 + r, and it is the absolute value of the slope of the budget line as before. Con-
suming one more dollar today means foregoing a dollar plus interest later and vice versa. Another
way to say this is that 1" + r is the price of current consumption. (1
" + r)W0 is the intercept on the
C1 or vertical axis, which is simply wealth measured in one year’s time or W1.
Lucy maximizes her utility by choosing to consume $474.77 today and $762 next year. She
gave up $525.23 this year, which at seven per cent, gives her an extra $562 next year for a total of
$762. It’s easy to see why Lucy is a lender and not a borrower. Her marginal rate of substitution
at his endowment point is less than the price ratio

M
" R Se < 1 + r

With her endowment Lucy places a lower value on consuming one more dollar today than the
market does. So she lends, moving up her budget line, and with each dollar lent, the dollars she
has to consume now are becoming relatively scarcer, pushing up her marginal rate of substitution.
She stops lending, having maximized her utility, when her marginal rate of substitution is exactly
equal to 1" + r because both she and everyone else values the next current dollar at exactly seven
per cent. That is why the optimum is defined by

C*
" : M R Se < 1 + r

A few things to note about the intertemporal consumption optimum. One is that Lucy’s
utility-maximizing choice is less extreme than her endowment in that $475 and $762 are closer to
one another than are $1,000 and $200. This is called consumption smoothing, and it makes sense:
most people prefer to have an income stream that is smooth (or, better still, smooth and increasing
over time) than one that jumps around. You can see that consumption smoothing is a direct result
of modelling with convex indifference curves. If Lucy’s endowment was e’ instead of e, she’d bor-
row to reach C*. No matter what specific mathematical function is used to represent a person’s
tastes, if it is convex, the consumption optimum will always lie between e and e’; the optimum is a
weighted average of the two and therefore “smoother.” The other thing to note is that Lucy is not

!54
wealthier for having chosen C*, and she is not wealthier for having lent. C* is worth $1,186.92 just
as e and every other point on the budget line is. But she is better off because she prefers C* to e.
Despite Scotiabank’s slogan, you’re richer than you think, a capital market by itself does not create
wealth. It gives us choice, and that is a good thing. For wealth creation, something investment in
real production is necessary.
The last thing to note is more subtle. Everyone else has presumably borrowed or lent to
reach their own consumption optimums, so everyone must be evaluating their next dollar to spend
(borrow to take that vacation?) at the market rate of interest. Utility theory implies that all of us
implicitly agree what the time value of money is. If that is true in real life—an empirical ques-
tion—then governments can justify using market interest rates as discount factors in cost-benefit
analysis. If not, those analyses don’t mean all that much.

Real Investment
To make herself wealthier, Lucy must produce something that is worth more than it costs.
Pretty basic, eh? Suppose that she can invest in any of the five projects shown in the table. Each is
an investment in real production. The projects are risk-free in keeping with our framework of

Project Cost Income ROI NPV at 7%

A $195 $179.40 -8% -$27.34

B $210 $390.60 86% $155.05

C $240 $261.60 9% $4.49

D $165 $173.25 5% -$3.08

E $190 $305.90 61% $95.89

looking at choices over time in the absence of risk. If Lucy takes project D, for example, she
would invest $165 out of her current endowment of $1,000 to earn a sure $173.25 in one year.
That’s a return on investment of five per cent. But she’d never invest in D because she can always
earn seven per cent just by lending her money. The value today at seven per cent of receiving
$173.25 in a year is only $161.92. That means that the net present value of project D is -$3.08.
By taking project D, Lucy make herself poorer by $3.08 because she’d be giving up $165 for
something that is worth only $173.25.

In c o m e 173.25
"N PV = − Cos t + = − $165 + = − $3.08
1+r 1.07

!55
 Being greedy, Lucy will take only those projects that increase her wealth, and that means
taking projects whose cash flows have positive net present value. She will invest a total of $640 in
projects B, C, and E, for a combined net present value of $255.42. Her wealth increases by
$255.42 to $1442.34 from $1,186.92. Lucy will also be better off if she increases her wealth this
C1

P*
1158
Repay
C*
926
762 Borrow
Earn
u* = 766.323
u = 630.616

e ue = 380.731
200
Invest
C0
360 475 577 1000 1187 1442
way. That seems obvious but best to check. You can do that by adding the five projects to Lucy’s
consumption diagram.
The projects are shown as green line segments, ordered from highest to lowest rate of re-
turn, starting at her endowment and going to the left. The first is project B with a rate of return
of 86 per cent. The cost of $210 ($1,000 left to $790 on the horizontal axis) and income of
$390.60 ($200 up to $591) on the vertical axis gives the line segment a absolute slope of 0.86.
Then comes E with a return of 61 per cent, C at five percent, and so on. Connecting the seg-
ments end to end creates a piece-wise curve that is sort of concave to the origin, and is called an
investment opportunity schedule. If the projects were infinitely divisible, the investment opportunity
schedule would be smooth and look just like a production possibility curve. And that’s because it is
a production possibility curve, except that it represents dollars today being transformed into dol-
lars tomorrow rather than combinations of goods.
To spot which projects have positive net present values, slide the budget line parallel and
upwards until it touches the investment opportunity schedule at just one point. This happens at
point P* (which would be a tangency if the investment opportunity schedule were smooth).
Projects to the right of P* have positive net present values and should be undertaken, and those to
the left should be passed over because they do not. That makes sense because the budget line has
a slope equal to one plus the rate of interest (ignoring the negative sign), and the slope of each line
segment representing each project has a slope equal to one plus the project’s return on investment.
The projects to the right of P* earn more than the rate of interest, which in a two-period world
means they have positive net present values. The horizontal intercept of the pushed-out budget
line, $1,442, is Lucy’s maximized wealth. The distance, $255, from the old intercept is the com-
bined net present values of projects B, C, and E.

!56
 Lucy’s investment choice has nothing to do with her taste for consumption over time. It is
driven by greed, and anyone who is greedy and who has the same opportunities would also choose
projects B, C, and E. That’s a good thing because if these investment opportunities belonged to a
corporation, there would be no disagreement among shareholders as to which ones management
should undertake. Greed and a shared gauge for profitability—the market rate of interest—gets
us around Arrow’s impossibility result. It allows shareholders to hire professional managers to do
the project picking for them, directing them with one simple rule: make us richer. (In the uncer-
tainty of the real world, making sure that managers actually do their job, and that they do not rob
shareholders blind is another story.)
Having made herself as wealthy as can be, Lucy then expresses her preference by choosing
the cash flow stream she wants to consume. That is $577 today and $926 next year, shown at C*.
After investing $640 in projects B, C, and E, Lucy has only $360 left from of her current endow-
ment of $1,000. Not a problem. With a capital market, she simply borrows $217 at seven per cent
to bring her cash on hand to $577. So there are two reasons that Lucy’s utility is higher than it
would be if she consumed her endowment: one is the increase in wealth from profitable real in-
vestments; the other is the choice to consume provided by the ability to borrow or lend (otherwise
she’d have to consume P* rather than C*). It is the existence of a market rate of interest, at which
everyone can borrow or lend, that allows Lucy to separate her investment decision and your con-
sumption choice. One does not interfere with the other, so Lucy can maximize her wealth, with-
out concern about what she wants to consume, and in that way, boost her utility. The separation
of consumption and investment is known as Fisher Separation after economist Irving Fisher. Fisher
Separation can be summarized like this:

P* : M R T = 1 + r
C* : M R S = 1 + r

MRT is the marginal rate of transformation or the absolute value of the slope of the investment
opportunity schedule, just as it is for production possibility curves. But here MRT is interpreted as
one plus the marginal return on investment. The condition MRT = 1 + r, says to keep investing
until the last penny invested earns exactly the rate of interest; in other words, chose all projects
with positive net present values.14 You are already familiar with the second condition, MRS = 1 +
r, so no need to discuss that.

Equilibrium Interest Rate

Up to this point we have been looking at intertemporal consumption in partial equilibrium.
Lucy achieved her preferred consumption stream by investing part of her current endowment and
borrowing given that the interest rate was seven per cent.
Where does the seven per cent interest rate come from? Equilibrium in the market for mon-
ey. The equilibrium interest rate is the one that equates the number of dollars potential lenders
are willing to lend with the number of dollars that potential borrowers are willing to borrow. You
can guess that the supply and demand for dollars today depends upon everyones’ endowments

14 Zero present values are okay too.

!57
(think of this as the distribution of income both now and in the future), their tastes for consump-
tion (‘high’ or ‘low’ marginal rates of substitution at the endowments), and individual real invest-
ment opportunities. And all three of these variables interact.
People whose current endowments are small compared to their future endowments will
have biggish marginal rates of substitution, and they will generally want to increase their current
consumption. If these same people comprise a big share of the aggregate current endowment of
the economy, the borrowing that they do from those whose current endowments are also compar-
atively small will occur at a higher interest rate than it otherwise would. One example of this is an
economy dominated by people facing lean times now compared to how they’ll be doing next year.
But we can’t talk about endowments without talking about tastes. You know that two people with
the same endowment will have different marginal rates of substitution if their tastes differ. That is,
after all, what it means to have different tastes. It is also possible that someone with a small current
endowment relative to their future endowment has a smaller marginal rate of substitution than
someone with a bigger current endowment relative to their future consumption. (Confused? Draw
some indifference curves.) The person with the smaller relative current endowment values more
current consumption less than the person with the bigger relative current endowment because
differences in their preferences (I don’t have much now, but it doesn’t matter because I really don’t value con-
suming more now). So, endowments and tastes necessarily interact.
The figure on the next page is an example of the determination of a three per cent rate of
interest in general equilibrium. Including real investment opportunities in the Edgeworth box is a
little advanced for this course, so the equilibrium is presented for a pure exchange economy. The
aggregate current endowment is $100, of which $16 belongs to Lucy and $84 to Ricky. The ag-
gregate future endowment is $102 with $70 going to Lucy and $32 to Ricky. Both Lucy and Ricky
have the same natural log utility functions noted in the figure, but value of their taste parameter,
α
" , differs. The economy is growing at a rate of two per cent.

!58
 At the endowment point, e, Lucy’s marginal rate of substitution is bigger than Ricky’s, so
Lucy borrows $21.71 from Ricky and they agree that the repayment should be $22.37. At the op-
timum, C*, Lucy’s marginal rate of substitution is equal to Ricky’s is equal to 1.03. That implies

C1 Ricky
102
u C0 , C1 = α ln C0 + ( 1 -α) ln C1

α = 0.44918 ( Lucy) , α = 0.5413 ( Ricky)

e
70

Lucy repays
$22.37
C* : ( MRSL = MRSR )
47.63 ⇔ p * = 1 + r * = 1.03
Lucy borrows
$21.71

C0
0
0 16 37.71 100
Lucy

an interest rate of three per cent.

Change any one of the parameters—individual endowments, the growth rate of the econ-
omy (aggregate endowment), or tastes—and the equilibrium changes. For example, had the value
of Lucy’s taste parameter, "α, been a little smaller, 0.415964 rather than 0.44918, the equilibrium
rate of interest would be minus three per cent!

!59
Permanent Income Hypothesis
Consumption smoothing is one result of utility maximization. Both Lucy and Ricky
smoothed their consumption. Life has its ups and downs but we prefer to take the edge off, and we
can do that if there is a capital market where we can borrow and lend. Does utility theory predict
that you will smooth your consumption if you live to be 88? Yes, all of the results for the two-peri-
od model still hold. This means that when you are just starting out you will probably be a borrow-
er, or net dissaver to use the jargon, because your income will be less than the smoothed amount
you prefer. By mid life, your income will have grown, and you will hopefully be a net saver
(lender). Income stops coming in or shrinks to a pension when you retire, but utility theory pre-
dicts that you will still want to maintain the lifestyle to which you are accustomed, so you will like-
ly become a net dissaver again and stay that way until you kick the proverbial bucket. This is the
essence of the Life Cycle Theory of Consumption and Savings: people want to be able to spend smoothly
over their lifetimes—think of it as maintaining a certain standard of living—and they adjust their
savings, cycling through borrowing and lending, in order to achieve this. Let’s look at the special
case of perfect consumption smoothing, where the optimal amount consumed each year is con-
stant, and then see how to relates to consumption smoothing in general, that is, smooth by in-
creasing every year.
A guy named Nigel will be the player in this example. Nigel is 20 years old. He just landed
his first job, with a starting salary of $100,000, which will grow by 1.5 per cent every year right up
until he retires at age 70. He owns a $300,000 house, a $75,000 investment portfolio, and a

Nigel in a nutshell

Preferences homothetic

Age now 20

Age at retirement 70
Age at death 88
Interest rate 0%
Annual income $100,000
Growth rate of income 1.5%
Assets
House $300,000
Portfolio $75,000
Beatles records $5,000

collection of Beatles records worth $5,000. Not bad for a 20-year old, eh? One other thing about
Nigel, and he knows this himself, is that he will live to be 88. Not the kind of thing most of us
would want to know, but we have assumed a world of perfect certainty.

!60
 What amount will Nigel consume each year if he maximizes his utility? This optimal con-
sumption is also known as permanent income because, by spending that amount each year, Nigel will
have drawn his wealth down to zero by the time he dies. The present value of the optimal con-
sumption stream is exactly equal to his wealth, as it must be because you saw in the two-period
case that Lucy’s optimal consumption lies on the budget line; it satisfies the wealth constraint. So,
to compute permanent income, we need an interest rate in order to figure out Nigel’s wealth, and
we need to know Nigel’s utility function. There are two conditions for permanent income to be
constant, that is, perfect smoothing: one is that the interest rate must be zero, so that there is no
incentive to push a few extra dollars into the future; and the other is that Nigel’s utility function
must be such that the enjoyment he gets from an extra dollar doesn’t depend on when he gets it.
For example,

" = C0αC01−α , α = 0.5

u

does the trick. Any utility function that is symmetric in this way would also work.
Now for Nigel’s wealth. His wealth is the market value of his assets plus the present value of
the income he will earn over his lifetime.

Wealth = Market value of assets plus the present value of income stream

The combined value of his house, portfolio, and record collection is $380,000, so his assets
are taken care of. His wage income is an annuity that will grow by 1.5 per cent a year. He will re-
ceive $100,000 in one year, $101,500 in two, $103,022 a year after that, and when he turns 70, his
50th and last pay cheque will be $207,413. The present value of his income stream is

1.0150 1.0151 1.0152 1.015N−1

[ (1 + r)1 (1 + r)2 (1 + r)3 (1 + r) N ]
PV
" (i n c o m e) = $100,000 + + +…+

Let m be Nigel’s $100,000 starting salary, g be the 1.5 per cent the growth rate, and N be the num-
ber of years he will work. The present value can then be written as

(1 + g)0 (1 + g)1 (1 + g)2 (1 + g) N−1

[ (1 + r)1 (1 + r) N ]
PV
" (i n c o m e) = m + + +…+
(1 + r)2 (1 + r)3

You can show that the series simplifies to

1+g N
1−(
1+r)
PV
" (i n c o m e) = m
r −g

which can be used to compute the present value of any annuity that grows at a fixed rate.

!61
 If the interest rate is zero, the equation for the present value of his income is even simpler

N
(1 + g ) − 1 1.01550 − 1
PV
" (i n c o m e) = m = $100,00 = $7,368,280
g 0.015

Nigel’s wealth is $7,748,280, $380,000 from his assets and the bulk, almost $7.4 million, from his
wages. He will spread this evenly over the 68 years he has left to live. That means he will consume
$113,945 each year. If the interest rate was positive, Nigel’s permanent income would be increas-
ing over the years, although he’d still be smoothing his consumption.15
The first thing to note about Nigel’s permanent income of $113,945 is that it is almost $14
thousand bigger than his first few pay cheques. As a matter of fact, he will have to borrow or sell
some of his assets to make up the difference until he is 29. His life cycle of consumption and sav-
$

207,413

Nigel's wage

113,945
100,000 His permanent income

Age
21 70 88
ings is shown in the chart. The timeline starts when Nigel is 21 (t = 1) rather than 20 (t = 0) be-
cause we assumed that cash flow occurs at the end of the year and that is when it is consumed.
Can you spot the times when he is a net saver and dissaver?
The other, and more important, thing to note about the permanent income hypothesis is
that if Nigel’s permanent income changes, he will increase or decrease his consumption by exactly
the change in permanent income. If he finds ten dollars lying on the sidewalk right now (happens
to me all the time), he will treat himself by spending about 15 cents more each year (can you fig-
ure that out?). He will not run down the street and spend the $10 at Double Pizza.

15 That calculation is left for a more advanced course.

!62
Reading
Thaler, Richard, 1990, “Anomalies: Savings, Fungibility, and Mental Accounts,” Journal of Econom-
ic Perspectives, 4(1):193-205.
“Why Johnny Can’t Save for Retirement,” Fortune, March 21, 2005.
Chapter 10 in Varian. [Optional]

!63
 10
MARKETS FOR RISK

A capital market not only allows us to shift consumption over time, it allows us to shift risk.
Both are done through trade, of course. Those who are willing to take on more risk buy it from
those who want to get rid of some. Risk-averse buyers will demand a higher rate of return for tak-
ing bigger risks, and risk-averse sellers decide how much return they are willing to give up to shed
risk. Markets for risk, such as those for bonds, stocks, and insurance, are no different from any
other in that they benefit society by providing choice.

States and Dates

Suppose that there is no risk now. Lucy and Ricky know everything about the present mo-
ment: their bank balances, pairs of clean socks in their dresser drawers, whether the salami in
their refrigerators is still fresh, and the condition of the shingles on their roofs. You know that you
can’t know everything even if there is no uncertainty, but there is an awful lot that is knowable.
That is period 0 in the intertemporal choice model in chapter 9. The future, however, is uncer-
tain, so consumption in period 1 must now be treated differently (likewise, periods 2, 3, 4 and so
on, if there are more than two periods in the model). Risk and time are intertwined. In a world
without risk, Lucy’s period 0 endowment might be $180 and her period 1 endowment $214. If
she chooses to consume her endowment, she knows right now that $214 will be waiting for her
next year for sure.
Not so in a risky world. Lucy’s current endowment is still a sure $180, but her future en-
dowment might be any amount. Probably not $1 billion and hopefully more than $3.57. How is
risk to be brought into the model? One way is to assume that future, risky consumption is a ran-
dom variable that behaves according to a known probability distribution. Probability distributions
are made up of two things: outcomes (consumption or income in our case) and the probabilities
of those outcomes occurring. Lucy, for example, might face a 0.4 chance of ending up with $160
next year and a 0.6 chance of $250. That there are only two outcomes is arbitrary. There could
be any number, say, 11 or 546. In real life I don’t think we can truly know what the outcomes are
or how many are possible, and we would struggle to attach probabilities to them. Economists refer
to the circumstances in which particular outcomes occur as states or states of nature. If there are
two, then they might be called state a and state b (like Thing 1 and Thing 2 from Dr. Seuss).
States are mutually exclusive: rain or shine, war or peace, boom or bust. State a has a 0.4
chance of occurring, and if it does, Lucy will have $160; state b has a 0.6 chance of occurring, in
which case Lucy will have $250. In a riskless model, Lucy’s endowment is written as

" " "e = (e0, e1) = ($180, $214)

!64
but in a time-state preference model, her endowment or any consumption bundle is written as

"C = (C0, C1) = (C0, (Ca, Cb; πa )) = ($180, ($160, $250; 0.4))

where C1 is now a random variable; πa is the probability of state a happening and having con-
sumption Ca; πb = 1 − πa is the probability of state b and having Cb. It is important to remember
that Lucy will never have $160 and $250 next year. Today she knows now that she will have either
$160 or $250 depending on what happens. Que sera sera.
Lucy’s wealth in a riskless world is the present value of her endowment computed at what-
ever the risk-free rate of interest happens to be

Contingencies Endowments Securities

Dates States Probs. Ps Lucy Ricky Canada M F

0 - 1 1 180 230 410 4.10 0.943

1 a 0.4 0.328 160 340 500 5.00 1.00

1 b 0.6 0.615 250 150 400 4.00 1.00

C1 $214
W
" 0 = C0 + = $180 +
1 + rF 1 + rF

In a risky world, her wealth is still the present value of her endowment but it is expected future
consumption, E " (C1), that must be discounted at a higher rate, k, to take account of the risk

E (C1) 0.4 × $160 + 0.6 × $250 $214

"W0 = C0 + = $180 + = $180 + , k > rF
1+k 1+k 1+k

In the time-state preference model, it is convenient to write wealth in terms of theoretical prices rather
than interest rates or discount factors. We can figure out the rates later.

W
" 0 = p0C0 + p1E (C1) = p0C0 + paCa + pbCb

1
Instead of having one price for future consumption, p1 = , there is a price today, pa and pb,
1+k
for consumption in each future state of nature. These are called pure state prices, primitive security
prices, or Arrow-Debreu prices, after the two economists who, independently, developed the model.
Take a moment to think about what a pure state price is. It is the price or value today of having
exactly one dollar in a particular future state and nothing in all of the other states. That means
that you can think of a pure state price as a present value factor that is attached to a particular
future event or, even easier, you can think of it as the market price of a gamble that pays one dol-
lar if a particular event occurs and nothing if anything else occurs.

• pa is the price today of a gamble that pays $1 in state a and $0 in state b

• pb is the price today of a gamble that pays $1 in state b and $0 in state a

!65
While pure state prices are theoretical—you can’t look them up anywhere—they do exist in real
markets various guises, some of which we will discuss in class. For now, here’s an example: the
price of a lottery ticket that pays $1 if you earn an A or better in my course and $0 if you don’t.
Trade that online.
In the economy we’ll be discussing in the rest of this chapter, pa = 0.328 and pb = 0.615.
Calculating the prices is left to an intermediate course. But here their interpretation is more im-
portant than their computation: A dollar next year in state a is worth 33 cents now and a dollar in
state b is 62 cents. You’ll be able to use them to calculate Lucy and Ricky’s wealth, as well as the
equilibrium risk-free rate of interest and the expected return on the stock market index.

Description of the Economy

Information about a two-period Canadian economy is given in the table. The states and
dates are in the first panel. The state probabilities and pure state prices are carried over from the
previous section. The state prices are in the rows labeled time period 1 to associate them with
states a and b, but they are time 0 or now prices. The Endowments columns shows endowments for
Lucy, Ricky, and Canada as the two of them combined. Since Lucy and Ricky are Canada’s only
inhabitants, the country is endowed with $410 today and will grow to $500 next year if state a
occurs or decline to $400 if it’s b. You can think of the aggregate endowment as Canada’s gross
domestic product. The last panel introduces two financial securities. M is a risky stock that pays a
dividend of $5.00 in state a and $4 in state b. M’s price is PM. M is like a stock market index or
super stock that includes all firms in the economy, so it pays out the aggregate endowment of the
economy. M’s payoff is one one-hundredth the aggregate endowment, so the supply of shares of
M must be 100. If M’s payoffs were $2.50 and $2, there would be 200 outstanding because the
payoffs have to be proportional to the aggregate endowment. Anyone holding nothing but shares
in M is bearing the same risk as the economy as a whole. F is a riskless security because it pays $1
no matter what happens. It is like a pure discount bond with price PF because it does not pay in-
terest, and its payoff is not related to the aggregate income of the economy. You can think of F as
a Treasury bill with a face value of $1. Since F is a debt security, it’s net supply must be zero—for
every dollar borrowed there is a dollar lent. By figuring out the equilibrium price of M and F, we
will know the expected return on risky stock and the risk-free rate of interest.
The figure on the next page shows the economy represented in an Edgeworth box. It in-
cludes the Pareto optimum, C* = (C* 0
, (C*
a , C*
b
)) = ($180, ($251.50, $201.20)). Notice there is no C0
axis. If it was drawn, the C0 axis would come straight out into your face from Lucy’s origin, be-
cause the Edgeworth box would be three dimensional. e" 0 = $180 could be shown on a third axis
but it isn’t necessary because the time-state preference model describes how Lucy and Ricky trade
their future endowment now, e, to end up at the Pareto optimum, C*, and that is also why optimal
current consumption is fixed at $180. What you are looking at in two dimensions is next year or
period 1, and the axes are the two outcomes. Even though the probabilities are not shown any-
where (you’ll have to remember them or look back at the table), every point in the Edgeworth box
is a distribution of future consumption or income. Endowment point e is the distribution ($160,

!66
 340 248.5 100 0
400 0

R
E u Ca , Cb = πa ( 1 -γ) ln Ca + πb ( 1 -γ) ln Cb

ick
y
Pa γ = 0.5, πa = 0.4, πb = ( 1 - πa )
= 0.533
Pb

e
250 150
C* : MRSL = MRSR
Cb 201.2 πa γ Cb Pa 198.8
= · · =
1 -πa 1 -γ Ca Pb
Lu
cy

0
0 100 160 251.5 500
Ca

$250; 0.4) for Lucy, and by subtracting from the aggregate endowment of the economy, ($340,
$150; 0.4) for Ricky. Canada is expected to grow by 7.3 per cent

E (E1) $440
E
" (g) = −1= − 1 = 0.0731707
E0 $410

Pareto Optimum
The Pareto optimum is a general equilibrium and is determined in the usual way: Lucy and
Ricky’s marginal rates of substitution over uncertain state consumption are equal, and this hap-
pens at a value of 0.533, which is the implied equilibrium price ratio. One dollar of consumption
in state a is worth 0.533 dollars of consumption in state b.

P*
a 0.328 ·
" : M R SL = M R SR =
C* = = 0.533
P*
b 0.615

Notice that you cannot compute the pure state prices individually from just the implied price ra-
tio. But having given you those values, pa = 0.328 and pb = 0.615, you can compute Lucy’s wealth
to be $386.23, Ricky’s as $433.77, making $820 the wealth of the nation. Why is a dollar of con-
sumption worth so much less in state a than in b? You can find two reasons in the table.

!67
 If wealth is the present value of a person’s income, then the price or market value of a se-
curity, in the same way, is the present value of the income it provides. The equilibrium price of
risky stock M is about $4.10 for an expected return of 7.3 per cent.16 The price of safe bond F is
$0.943, making the risk-free rate of interest 6.04 per cent. Letting Xs be the state income or pay-
off of any security, then

PM = Pa XaM + Pb XbM = 0.328 × $5 + 0.615 × $4 = $4.10

E (XM ) $4.40
∴ E (rM ) = −1= − 1 = 0.07317
PM $4.10
"
PF = Pa XF + Pb XF = XF (Pa + Pb ) = $1(0.328 + 0.615) = $0.943
X $1 $1
∴ rF = F − 1 = −1= − 1 = 0.0604454
PF PF $0.943

The difference between the expected return on the stock market index and risk-free bonds is
called the market risk premium. It is the extra return that the market compensates you with for bear-
ing risk. The risk premium is only 1.3 per cent in this economy. That’s tiny. A market risk premi-
um in the range of three to six per cent was typical during the 20th century. Tweak the parame-
ters of the time-state preference model all you want and you won’t be able to crank out realistic
risk premiums unless you make Lucy and Ricky pathologically risk averse. This anomaly, that real
life risk premiums are so different from what theory predicts, is known as the equity premium puzzle.

Portfolios and Trade

C* is an equilibrium because, mutually, Lucy and Ricky prefer it to any other distribution of
future consumption. At any other point in the Edgeworth box, their marginal rates of substitution
would differ from one another, and they would then do something to get to C*. How do they get
to C* if, for whatever reason, they are not there? The government could try to reallocate their en-
dowments. So could a benevolent monarch or dictator. But we haven’t included a government in
this model. Moving around the Edgeworth box means that Lucy and Ricky must be trading future
distributions of consumption. Since the only thing to trade in this economy is the two securities,
and there is no production, every distribution (point) must be a portfolio of shares of risky stock M
and safe bond F. You can see this by considering two references. One is the main diagonal, and
the other is the two dashed grey lines, one leaving Lucy’s origin at 45 degrees and the other leav-
ing Ricky’s at the same angle. The main diagonal or Market line is the set of all portfolios that are
comprised of nothing but shares of M (no bonds) because every distribution along the line pays
proportionally to the economy as a whole, $4 in state b for every $5 in a. For example, the point
(220, 160) on the main diagonal (the point is not shown in the figure) represents the distribution of
future income that Lucy would earn if she holds 44 shares of M because $220 = $5 x 44 and
$160 = $4 x 44. Ricky holds the other 56 shares at that point. Now consider the 45-degree line
starting from Lucy’s origin. It is her certainty line because for every point on it the income is the
same in both states. At (193, 193), for example, Lucy must own 193 bonds and no shares since the

16Notice that the expected growth rate of the economy is the same as the expected return on the market. It
does not have to be. It is true in our example because I gave Lucy and Ricky the same utility function.

!68
bond is guaranteed to pay $1. She has lent $193 and Ricky must be the borrower. Ricky would
likewise be in a riskless position on his certainty line. Lucy and Ricky cannot both be in riskless
positions; the economy is, after all, risky, and they make up the economy.
Any point not on the Market line or one of the two certainty lines must be an income dis-
tribution for a portfolio of shares of M and safe bonds. Lucy and Ricky’s shares add up to 100,
and their bonds add up to 0. Points e and C* are the two distributions that are important to the
time-state preference model. Working out the composition of these portfolios is as simple as solv-
ing two equations in two unknowns. At e, Lucy receives $160 in state a or $250 in state b.

{eb
ea = $160 = $5qM + $1qF
e" Lucy → qM = − 90, qF = 610
= $250 = $4qM + $1qF

Those amounts, $160 and $250, have to come from the payoffs of M and F. Lucy is short 90
shares. She borrowed those from Ricky. This is a risky loan because the amount she’ll have to re-
pay Ricky depends on which state prevails. If a then she has to pay him $450 but if b, only $360.
She used the loan to buy 610 bonds that will pay her $610 no matter what. Kind of strange, isn’t
it? Taking out a risky loan to buy something safe? 17 Strange but not impossible; this is a theoretical
model, and e was arbitrary. Ricky is long 90 shares and short 610 bonds (he borrowed $610). No
need to solve equations; it’s just bookkeeping. I’ll let you confirm that at the optimum Lucy holds
about 50.3 shares, Ricky holds 49.7, and there is no debt outstanding (C* is on the Market line in
the example). Could you have gotten a sense of the composition of Lucy’s e portfolio just by look-
ing at its location in the Edgeworth box? Move e around. What happens to its composition if it is
moved closed to Lucy’s certainty line? Beyond that, as it is moved closer to the Market line? Be-
yond the Market line to the lower-right of the lower-right?
Here’s some homework for you. Pick any point in the Edgeworth box and see if you can
figure out the composition of Lucy and Ricky’s portfolios at that point using the payoffs of securi-
ties M and F. You’ll find that you can because the following conditions are met: there are at least
as many securities as there are states of nature (M and F, a and b); the securities are not perfect
substitutes (you cannot create the payoffs of M by multiplying the payoffs of F by some constant);
and negative quantities are allowed (short selling). These three conditions are necessary for a com-
plete market, which is one where any distribution of payoffs can be created by combining existing
securities into portfolios. The upshot of this is that Lucy and Ricky can trade to any point in the
Edgeworth box that they prefer, and that includes C*. Choice is good.

17Owning a positive quantity of a security is called a long position. Owning a negative number is called a short
position or short selling, and it is the same as having borrowed.

!69
Reading
“Risk Management for the Masses,” The Economist, March 22, 2003.
“Doomsday Derivatives,” The Economist, October 19, 2002.
Chapters 11 through 13 in Varian. [Optional]

!70
 II. F INANCIAL E CONOMICS

This part covers intermediate financial economics for courses such as Finance Theory I
(FINA 385). It’s nasty stuff.

!71
 11
PORTFOLIO THEORY

In this chapter you'll learn how to form portfolios of risky assets and how to expand risk-
return choices by borrowing or lending at a risk-free rate of interest. Make sure to work through
all of the numerical examples. Some are based on real-world stock prices. You can find the data in
“Stock Prices.csv”. The file contains monthly closing prices in US dollars on 109 S&P big cap
stocks, for the 30 months, January 2009 to June 2011.

!72
Portfolio Return
If you invest $40 out of $100 at five per cent and $60 at eight per cent, the return on your
portfolio will be 6.8 per cent.

0.068
" = .40 × 0.05 + .60 × 0.08

You could invest $125 at eight per cent but, if your capital is $100, you’d have to borrow the $25
at five per cent or short sell $25 worth of the asset earning five per cent. The return on your port-
folio in that case would be 8.75 per cent.

0.0875
" = − 0.25 × .05 + 1.25 × 0.08

What’s true for the actual return on a portfolio must also be true for the expected return, so for an
investment in any two risky assets, the expected return on a portfolio is

E (rp ) = w1E (r1) + w2 E (r2 ), where w1 + w2 = 1

where the weights are the percentage of your wealth or capital invested in each. The weights ob-
viously have to add up to one unless you are a government, in which case they can be anything
you like. It doesn’t matter how many assets are in your portfolio—one, 50, 637, or some number
n.
n n

∑ ∑
E (rp ) = wi E (ri ), where wi = 1
i=1 i=1

The same expression can be written in matrix notation as

E (rp ) = w′ . r, where w′ . 1 = 1

′
" = ( w1 w2 … wn)′ is the vector of weights, r" = (E (r1) E (r2 ) … E (rn )) a vector of expected
w
returns, and 1 (a boldface one) is a vector of n ones.

!73
Portfolio Risk
The riskiness of a portfolio can be measured by the variance or standard deviation of its
returns, which in turn depends on the riskiness of the assets in it. Suppose that asset 1’s standard
deviation is 12 per cent and asset 2’s is 17 per cent. If their prices always moved in the same direc-
tion—they are perfectly positively correlated—the standard deviation of a portfolio of the two
would be, like expected return, just a simple weighted average of the standard deviations of the
two assets.

σ" p = w1σ1 + w2 σ2

Your 40-60 portfolio would have a standard deviation of 15 per cent.

" 0.15 = .4 × 0.12 + .6 × 0.17

But there are very few things in this world, particularly the returns on financial assets, that
are perfectly correlated, either positively or negatively. The price of the two assets may move in the
same direction most of the time, but sometimes the price of asset 1 will go up when the price of
asset 2 goes down or vice versa. The correlation of their returns is less than one. If their correla-
tion was, say, 0.8, then the standard deviation of your 40-60 portfolio would be less than 15 per
cent because some of the price declines of one will be, on occasion, cancelled partly, entirely, or
more than entirely, by the price increases of the other. The calculation of portfolio standard devi-
ation captures washing away of risk or diversification that results from return correlations, ρ" 12, being
different from +1,

σp2 = w12 σ12 + w22 σ22 + 2w1w2 ρ12 σ1σ2

= w12 σ12 + w22 σ22 + 2w1w2 σ12

σp = + σp2

where σ" 12 = ρ12 σ1σ2 is the covariance of the returns of the two assets. Your 40-60 portfolio would
have a standard deviation of 14.3 per cent if the correlation between the returns of assets 1 and 2
was 0.8 or, equivalently, their covariance was 0.01632.

"σp = + σp2 = + 0.0205416 = 0.143323

!74
 What if your portfolio has three assets? Use your intuition to write down the equation for its
variance. The pattern that you should see applies to a portfolio of any number of assets. In sum-
mation notation
n n
σp2 =
∑∑
wi wj σij
i=1 j=1

where as always the weights add to one and, for handy notation, σ" ii ≡ σi2 and σ" jj ≡ σj2 (the covari-
ance of an asset with itself is just its own variance). The same expression is elegant when written
in matrix notation,

σp2 = w′ . V . w

V is called the variance-covariance matrix or just the covariance matrix. For a two-asset portfolio

(σ21 σ22)
σ12 σ12
(σ21 σ22 )
σ11 σ12
V
" = or . The diagonal elements are variances, and the off-diagonal ele-

ments are covariances. V is symmetric because "σij = σji. The covariance matrix for assets 1 and 2
is

(0.01632 0.0289 )
0.0144 0.01632
V
" =

and the variance of your 40-60 portfolio can be computed as

(0.01632 0.02890) (0.6)

"σp2 = (0.4 0.6) 0.01440 0.01632 0.4
= 0.0205416

Don’t be lazy. Check my math by calculating the variance using matrix math.

!75
Portfolios of Two Risky Assets
A portfolio frontier shows how a portfolio’s risk and return depends on the investment in a
set of assets. The frontier described in this section is formed from two real-world stocks, Halibur-
ton (HAL) and Time Warner Cable (TWC), using their US monthly closing prices from Decem-
ber 2008 to June 2011. You can find their prices in the file “Stock Prices.csv” mentioned at the
beginning of this chapter. HAL and TWC’s estimated expected returns, variances, and covari-
ances are what you need to form portfolios for a frontier.

HAL (asset 1) TWC (asset 2)

Mean 0.0407485 0.0234194

Variance 0.00987177 0.0142828

Standard deviation 0.0993568 0.119511

Their covariance show up in the covariance matrix (makes sense),

r′ = (0.04074850.0234194)

(0.005942390.0142828 )
" V = 0.009871770.00594239

ρ12 = 0.500446

The table of portfolios on the next page was made using this information for investment
weights running from -20 per cent to 120 per cent in increments of 20 percentage points. For ex-
ample, when the investment in asset 1 is 40 per cent, the investment in asset 2 is 60 per cent. The
40-60 weighting makes for a portfolio with a three per cent expected return and a 9.8 per cent
standard deviation. The table could also have been made just as easily by choosing the expected
returns first, using those to figure out the weights, and then using weights to get the standard devi-
ations.

!76
 Investment Weights
"σp !E (rp ) HAL (asset 1) TWC (asset 2)

0.134573 0.0199536 -0.2 1.2

TWC 0.119511 0.0234194 0 1

0.106946 0.0268853 0.2 0.8

0.097845 0.0303511 0.4 0.6

0.0932279 0.0338169 0.6 0.4

mvp 0.0928084 0.0351989 0.679752 0.320248

0.0937593 0.0372827 0.8 0.2

HAL 0.0993568 0.0407485 1 0

0.109244 0.0442143 1.2 -0.2

a 0.119511 0.0469784 1.3595 -0.359504

There are two reference portfolios in the table. One is the minimum-variance portfolio or mvp for short.
It is the portfolio having the smallest variance or standard deviation. You can find the minimum-
variance portfolio for two-asset frontiers by solving for the investment weight in either stock that
minimizes portfolio variance. The mvp’s weight in asset 1 is
σ22 − σ12
" 1(mvp) =
w = 0.679752
σ11 + σ22 − 2σ12

The other reference portfolio in the table is a. Portfolio a has the same standard deviation as
TWC, but notice that it has a higher expected return (about twice that of TWC)? Anyone willing
to bear the risk of TWC but wanting the highest possible expected return would choose portfolio
a over holding TWC by itself. Portfolio a is called an efficient portfolio because there is no other port-
folio of HAL and TWC that offers a higher expected return for its level of risk.

!77
 The graph of expected portfolio return versus its standard deviation is a hyperbola. If vari-
ance is used instead of standard deviation, the frontier is a parabola. Both are pretty and are in-

E rp )

a
0.04698

0.04075 HAL
mvp
0.03520

TWC
0.02342

0.09281
σp
0.09936 0.11951
terpreted in much the same way. To see this algebraically, derive an expression for portfolio vari-
ance as a function of expected return, without the weights being there.
Portfolios on a two-asset frontier are unique. There is only one portfolio of HAL and TWC
that gives an expected return of 2.5 per cent, only one that gives 3.75 per cent, and only one that
gives 7.4 per cent. When you choose the weight in one asset, the weight in the other is one minus
the weight in the first. That is why a two-asset frontier always passes through the points represent-
ing the two assets.
All portfolios on the upward-sloping part of the frontier, including the mvp, are efficient be-
cause they offer the highest expected return for a given level of risk. You already know that portfo-
lio a is efficient but asset 2 (TWC) by itself is not. All efficient portfolios on this frontier are at least
68 per cent invested in HAL. If you wanted a higher expected return than HAL, you'd have to
take a short position in TWC.

!78
Portfolios of More Than Two Risky Assets
There is just one snag in working with portfolios of more than two risky assets: there are
many ways—infinitely many ways—to allocate your capital among them and get the exact same
expected return. It is a big difference, but it is the only difference between portfolios of two risky
assets and those with more than two. The problem then is how to choose the investment weights.
To work through this we’ll add the stocks of Apple Inc. (APPL), Anadarko Petroleum Corp.
(APC), and Covalon Technologies Ltd. (COV) to our set of risky assets. They will be assets 1, 2,
and 3. HAL and TWC are relabelled 4 and 5. Here is the summary of risk and return for the five
stocks. Their covariances and correlations will come later.

AAPL (asset 1) APC (asset 2) COV (asset 3) HAL (asset 4) TWC (asset 5)

Mean 0.0489129 0.0315206 0.0166714 0.0407485 0.0234194

Variance 0.00477222 0.0157596 0.00468611 0.0100405 0.014527

Std. Dev. 0.0684982 0.124478 0.0678774 0.0993568 0.119511

The only way to get an expected portfolio return of 3.75 per cent with just HAL and TWC in
your portfolio is to put 81.3 per cent in HAL and 18.7 per cent in TWC. But what if you decided
to make a portfolio of APPL, APC, and COV? Allocating 61.9 per cent to APPL, 5.9 per cent to
APC, and 32.2 per cent to COV would get you a 3.75 per cent expected return, but so would 54.9
per cent APPL, 21.1 per cent APC, and 24 per cent COV. There are countless ways to invest in
those three stocks for a 3.75 per cent return. This is true for all 16 combinations of three, four, or
all of the five stocks.

{{1,2,3}, {1,2,4}, {1,2,5}, <<10>> , {1,3,4,5}, {2,3,4,5}, {1,2,3,4,5}}

Add to those the 10 combinations of two stocks, and that’s a lot to choose from!
So, how should you invest? If, like most people, you don’t take on risk unnecessarily, you
should allocate your capital so that your portfolio has the smallest possible risk for the expected
return you have chosen. Turning that into a mathematical statement, the optimal investment is
one that minimizes portfolio variance with respect to the investment weights, for a given portfolio
expected return and the weights adding up to one.

Minimize σp2 ≡ w′ ⋅ V ⋅ w with respect to w

"
subject to w′ ⋅ r = E (rp ) and w′ ⋅ 1 = 1

I bet you recognized this as a constrained optimization problem from your high school calculus
class. You’re right. Turns out there is a unique solution for w and it is

w * = g + h ⋅ E (rp )

!79
w* works for portfolios of any number of risky assets, even the easy textbook case of just two.
Vectors g and h are computed as

D[ ( ]
1
g= B V−1 ⋅ 1) − A (V−1 ⋅ r)

D[ ( ]
1
h= C V−1 ⋅ r) − A (V−1 ⋅ 1)
"
A = r′ ⋅ V−1 ⋅ 1
B = r′ ⋅ V−1 ⋅ r
C = 1′ ⋅ V−1 ⋅ 1
D = BC − A 2

w* is linear in E
" (rp ). g is like an intercept, more precisely, a vector of intercepts, and h is a vector
of slopes. You can think of the expression for w* as a stack of lines.

g1 + h1 × E (rp )
g2 + h 2 × E (rp )
" * = g3 + h 3 × E (rp )
w
⋮
gn + h n × E (rp )

To compute g and h for our sample of five stocks, start with the vector of expected returns and
the covariance matrix.

"r′ = (0.0489129 0.0315206 0.0166714 0.0407485 0.0234194)

0.00469201 0.00347929 0.00165498 0.00283357 0.000803448

0.00347929 0.0154948 0.00333506 0.00772093 0.0035537
" = 0.00165498 0.00333506 0.00460735
V 0.00210297 0.000702788
0.00283357 0.00772093 0.00210297 0.00987177 0.00594239
0.000803448 0.0035537 0.000702788 0.00594239 0.0142828

Next compute the inverse of the covariance matrix,

286.555 −25.4972 −59.5777 −60.7008 18.4106

−25.4972 117.228 −40.1493 −80.4612 7.71844
" −1 = −59.5777 −40.1493 272.502 −12.6842 5.20969
V
−60.7008 −80.4612 −12.6842 226.632 −70.2325
18.4106 7.71844 5.20969 −70.2325 96.0222

Then A, B, C, and D,

A = 11.3172, B = 0.54648, C = -363.012, D = 70.2994

!80
And finally, g and h,

−0.400875 26.9247
0.0478901 −3.40596
g" = 1.29006 , h = −26.7738
−0.281744 9.26287
0.344673 −6.00783

Now you have an vector equation for computing the percentage investment in each of the five
stocks that minimizes the portfolio’s risk for any portfolio expected return that you chose.

−0.400875 26.9247
0.0478901 −3.40596
w
" * = g + h ⋅ E (rp ) = 1.29006 + −26.7738 ⋅ E (rp )
−0.281744 9.26287
0.344673 −6.00783

The optimal investment weights for your 3.75 per cent portfolio are

−0.400875 26.9247 0.608802

0.0478901 −3.40596 −0.0798334
"w * = 1.29006 + −26.7738 × 0.0375 = 0.286039
−0.281744 9.26287 0.0656134
0.344673 −6.00783 0.119379

and the portfolio’s standard deviation is about 5.4 per cent (" w′ ⋅ V ⋅ w = 0.0544174). The next
table shows the standard deviations of the 3.75 per cent portfolios we discussed earlier and a few
more. The portfolio in the first row uses w* as its weights. There is no other 3.75 per cent portfo-
lio that has a lower risk than that first one.

Investment Weights
!σp !E (rp ) AAPL APC COV HAL TWC

0.0544174 0.0375 0.608802 -0.0798334 0.286039 0.0656134 0.119379

0.0580537 0.0375 0.618732 0.059244 0.322024 0 0

0.0628525 0.0375 0.548726 0.211245 0.240029 0 0

0.0702113 0.0375 0.41208 0.32387 0.094359 0.09165 0.078041

0.0800822 0.0375 0.412 0.32387 -0.141302 0 0.405432

0.0830412 0.0375 0.31217 0.12734 0.463622 0.4743 -0.377432

0.0939667 0.0375 0 0 0 0.812539 0.187461

0.117803 0.0375 -0.274058 0.10956 0 1.1645 0

0.141205 0.0375 0 0.25497 -0.562507 0.4743 0.833237

!81
 Armed with a formula for the optimal investment weights, you can draw the five-asset fron-
tier. Choose a series of expected portfolio returns, calculate the investment weights for each, and
then use the weights to calculate each portfolio’s standard deviation. Give it a shot, and make it
pretty. Here’s the frontier with some reference points. Interpret them; calculate their investment
weights; and their standard deviations if not shown.
E rp )

b
0.07843

c a
0.04698 AAPL
HAL
0.03750
mvp mvp APC
0.03118
d TWC
0.02342
COV

0.06359
σp
0.05249 0.11951
Is the five-asset frontier a parabola in mean-variance space just like the two-asset frontier?
Yes. Because expected portfolio return is a function of the optimal weights and so is portfolio vari-
ance, the weights can be eliminated—as I encouraged you to do for the two asset case—by com-
bining the two expressions. Portfolio variance can then be written directly in terms of portfolio
expected return as

1
D(
σp2 = CE (rp )2 − 2A E (rp ) + B)

You can see that is is a parabola because variance (think of it as y) is a quadratic function of
expected return (x). Take the square root of both sides, and it is a hyperbola like the figure above.
This means you don’t have to calculate the weights if you don’t need them, for example, when all
you want to do is graph the frontier. But as a practical matter, you will want to know the weights
(How do I allocate my capital to get an expected return of eight per cent? The average monthly return on Lucy’s
portfolio was six per cent. How did she invest?).

!82
 The next table shows some of the portfolios that are in the graph. Make a table of portfo-
lios like this one both with and without calculating the investment weights. You’ll be glad you did.

Investment Weights
!σp !E (rp ) APPL APC COV HAL TWC

0.088168 0 -0.400875 0.0478901 1.29006 -0.281744 0.344673

0.0712058 0.01 -0.131628 0.0138305 1.02232 -0.189115 0.284595

0.0583069 0.02 0.137619 -0.0202291 0.75458 -0.0964867 0.224516

d 0.055366 0.0234194 0.229687 -0.0318756 0.663029 -0.0648129 0.203973

0.0525535 0.03 0.406866 -0.0542887 0.486842 -0.00385806 0.164438

mvp 0.0524855 0.0311759 0.438526 -0.0582937 0.45536 0.00703384 0.157374

0.0561855 0.04 0.676113 -0.0883483 0.219104 0.0887706 0.10436

c 0.0635943 0.0469784 0.864005 -0.112116 0.0322662 0.153411 0.0624347

0.067709 0.05 0.945361 -0.122408 -0.0486333 0.181399 0.0442815

b 0.119511 0.078425 1.7107 -0.219222 -0.809678 0.444696 -0.126491

0.067709 0.05 0.945361 -0.122408 -0.0486333 0.181399 0.0442815

!83
Three Ways of Looking at Diversification
Earlier in this chapter I described diversification in portfolios as a washing away of risk that
happens because the prices of assets do not always move in the same direction. Here are three
graphs that look more closely at that idea. I’m not going to say much about them because I want
you to think about them, especially how they relate to one another.
What would the portfolio frontier of HAL and TWC look like if the correlation between
their returns was not 0.500446? What if their correlation was -0.5 or +0.75 or something else?

E rp )

HAL
0.04075
ρ = -1 -0.5 0 +0.5 +0.75 +1

0.02342
TWC

σp
0.09936 0.11951

!84
 What if you graphed all 26 frontiers that are possible from our set of five stocks? There are
10 two-stock frontiers, 10 three-stock frontiers, five four-stock frontiers, and of course, one frontier
of all five. Where does the five-stock frontier sit in relation to all of the others, or, for that matter,
any one of the other frontiers in relation to the frontiers of portfolios with fewer stocks in them?
Looks like linguini, I know.
E rp )

0.08

0.06

AAPL
HAL
0.04
APC

TWC

0.02 COV

σp
0.02 0.04 0.06 0.08 0.10 0.12 0.14

!85
 What happens to the risk of a portfolio as the number of assets in it is increased? The final
graph shows average portfolio standard deviation of samples of portfolios containing the same
number of stocks. Each of the 20 samples contains 250 portfolios, randomly-selected from all 109
stocks in our data set.
σp

0.09

0.08

0.07

0.06 Equally -weighted portfolio of all 109 stocks

No. of stocks
1 2 3 4 5 10 15 20 in portfolio

!86
Adding a Risk-Free Asset to Portfolios
Suppose you don’t want to invest all of your capital in risky assets; maybe you want $25 out
of your capital of $100 to earn a guaranteed return, or at least as safe a return as can be. Or
maybe you’d like to invest $125 in risky assets by borrowing the extra $25, so long as the rate on
the loan is fixed. If that kind of safe borrowing and lending is possible, you are including a risk-
free asset in the mix. By borrowing you are short-selling the risk-free asset (a negative investment
weight) and by lending you are buying the risk-free asset (a positive investment weight). Nothing is
truly risk-free, of course. The closest thing we have is government bonds, particularly short-term
bonds, such as Treasury Bills. They’re pretty safe because most stable governments are able to
repay their debts since they have the power to tax. Inflation risk is usually low in the short run and
the risk of a disaster—the country being hit by a meteor or someone in a far off place pressing the
big red button is generally small too.
When risk-free borrowing and lending is possible, the elegant hyperbola or parabola that is
the minimum-variance frontier of risky assets becomes an equally elegant straight line.
Call the risk-free asset f. If you invest in any risky asset, k, with expected return E(rk) and

E rp )

0.05371 k→ M

k is any
risky asset

E rk - r f
f σk
0.0075
σp
0.07332
borrow or lend at rf, the expected return on your portfolio is a weighted average of the two, as it
must be for any two-asset portfolio,

E
" (rp ) = wf rf + wk E (rk )

But the standard deviation of your portfolio is simpler than it would be if both assets were risky. It
is now simply proportional to the standard deviation of k. That is because the standard deviation
of f is zero, and f and k are uncorrelated,

!87
 σp2 = wf2 σf2 + wk2 σk2 + 2wf wk σf k
" = wk2 σk2 because σf = σf k = 0
∴ σp = |wk|σk

Putting expected portfolio return and standard deviation together

E (rk ) − r f
" (rp ) = rf +
E σp
σk

E (rk ) − r f
gives a linear frontier with intercept rf and slope, , which is called the Sharpe Ratio.
σk
In the graph of our five-asset frontier, the risk-free rate is assumed to be 0.75 per cent. Each of
the light blue dashed lines is a linear frontier of portfolios of f and some risky asset k, which of
course could be a portfolio itself.
Now here’s where a bit of theory sneaks in again. If you like return but not risk, you’ll want
to choose k that maximizes the slope of the linear frontier because that gives portfolios of f and k
that are efficient. It’s another optimization problem from your calculus class. The optimal risky
asset k is portfolio M, the point where the linear frontier and the minimum-variance frontier are
tangent. M is a portfolio of all five risky assets, and has an expected return of 5.4 per cent. You
can do the math to find the tangency for the case where there are only two risky assets, such as for
HAL and TWC. This is sometimes presented in a footnote in finance textbooks as something that
is doable. It may be doable but it is ugly, and when there are more than two risky assets, it is just
plain sick. More on that in a moment.
Any portfolio on the optimal linear frontier is just an investment in M and some amount of
borrowing or lending according to your preference. Portfolios below M involve lending and above
M involve borrowing. What would be the expected return and standard deviation of your portfo-
lio if you invested 75 per cent of your capital in M and lent the rest at the risk-free rate? What if
you borrowed 30 per cent of your capital to lever an investment in M? I’ll leave that to you to
work out.
A nifty way to find the expected return of tangency portfolio M is

A ⋅ rf − B
E (rk=M ) =
C ⋅ rf − A

A, B, and C are the efficient set math parameters for frontiers of any number risky assets. The so-
lution exploits the fact that for any frontier portfolio z (next figure) there is an uncorrelated fron-
tier portfolio k

σ" kz = w*′
k ⋅ V ⋅ w*
z =0

The trick is to choose portfolio z to have an expected return equal to the risk-free rate, so that z
plays the role of surrogate intercept of the line, making M the tangency. Solving σ" k w = 0 for
" (rz ) = rf gives the solution above. I’d show you the proof, but I don’t want you to drop the course.
E

!88
Plugging in the values of A, B, and C for our five-asset frontier and a risk-free rate of 0.75 per cent
gives "E (rM ) = 0.05371, from which you can confirm the weights of the assets in M as shown in the
figure.
E rp )

1.0452
0.05371 k→ M
-0.135038
w M = -0.147914
0.215747
0.0220037

f z
0.0075
0.0751617
σp
0.07332
Notice that two of the asset weights in M are negative, -13.5 per cent for APC and -14.8 per
cent for COV? No problem, the math is right, but the negative weights tell us that the optimal
linear frontier in the example cannot represent an equilibrium in capital markets. That is because
risky assets are always in positive supply, and if all the other people like you, who prefer higher
return and lower risk, want to hold portfolio M, then the assets in M must be held (owned by or
invested in) by everyone. In that case, the weight of every asset in M must be positive. The reason
that not all of the assets weights are positive in our example is because it is just an example: the
risk-free rate of 0.75 per cent is not an equilibrium interest rate—it is made up—and the expected
returns and covariances of the five risky assets are not equilibrium values—they are estimated
from historical data. But for our purpose of learning how to form portfolios, we will turn a blind
eye to negative weights in M should they appear. We will assume that M contains all risky assets
that are available, even if we are working with only five, and that the risk and return of all of the
assets are equilibrium values. These assumptions, roughly speaking, let us refer to M as the Market
portfolio, the portfolio that everyone holds when they can borrow or lend at the same risk-free rate
of interest. And in that case, the name of the optimal linear frontier is elevated to the Capital Mar-
ket Line.
Capital Market Line

E (rM ) − r f
E (rp ) = rf + σp
σM

!89
 The Capital Market Line is shown, with more reference points than an acupuncture chart,
in the next figure. These reference points are for you to interpret. For example, what can you say
about CML portfolios g and h? What are their allocations of f and M? Can you calculate g’s ex-
pected return only knowing that it has the same standard deviation as TWC? Can you calculate
h’s standard deviation?
E rp )

g
0.08281 h
0.07843
b

M
0.05371
i a
0.04698 AAPL
c HAL
j
0.03750
mvp APC
0.03118
e mvp d TWC
0.02342
COV
f
0.00750
0.06359
σp
0.05249 0.07332 0.11951

!90
Choosing a Portfolio
We’re finally at the spot where we can talk about portfolio theory rather than just portfolio
math. Utility theory says that a person will make choices that maximize their utility, but for deci-
sions involving risk, they maximize expected utility (more on that in chapter 13). Suppose that
Lucy and Ricky have utility that is quadratic in random return,

( 2)
γ γ
u" (r) = 1− + (1 − γ) r + r 2
2

where Lucy’s risk aversion parameter, γ" , is 0.92 and Ricky’s is 0.79. I’ll let you show that, by taking
the expected value of this utility function, the expected utility function is

( )
γ γ γ
"E [u (r)] = 1 − + (1 − γ) E (r) − E (r)2 − σr2
2 2 2

Can you see from their expected utility function why a bigger value of γ" implies the person is
more risk averse? Expected utility is, conveniently, a function of expected return and variance of
return, which is exactly how we developed our portfolio math. Lucy and Ricky’s optimal portfo-
lios are show by the blue points and brown straddling M. The blue points are portfolios of the five
assets without risk-free borrowing or lending, while the brown points are the Market with some
borrowing or lending.
E rp )
γ γ 2
u( r )= 1 - + ( 1 -γ) r - r
2 2
γ = 0.79
0.0809

0.0692

0.0537 M

0.0402
γ = 0.92 mvp
0.0301

f
0.0075
σp
0.0358 0.0564 0.0733 0.1012 0.1165
Some work for you: derive expressions for the optimal portfolios on the five-asset frontier
and the Capital Market Line. To do that you’ll need to equate the marginal rate of substitution
for the indifference curve of the expected utility function,

δE (r)
M
" RS =
δσ

!91
to the slope of the five-asset frontier and the slope of the Capital Market Line. The expressions
you end up with will be simple. Work out the composition of all four portfolios from the informa-
tion in the figure even if you cannot work out the expressions for the optimal portfolios.
An implication of portfolio theory is that quadratic utility (mean-variance expected utility) is
a sufficient condition for mean-variance choice. English translation: you have to care only about
mean and variance in order choose only according to mean and variance. You like skewness? No
good. How about kurtosis (no, it’s not a type of facial rash)? Still no good. Quadratic utility, how-
ever, is not a necessary condition. If asset returns are jointly normally distributed, then mean-
variance portfolio choice holds no matter what kind of weird utility function you may have. So, do
you think asset returns are normally distributed?

Reading
Al-Shakfa, Omar and Gregory J. Lypny, 2011, “Islamic Investment and the Cost of Observance,”
Journal of Investing, 20(2):101-109.
Dimson, Elroy; Paul Marsh, and Mike Staunton, 2004, “Irrational Optimism,” Financial Analysts
Journal, 60(1):15-25.
Mandelbrot, Benoit and Nassim Taleb, July 2005, “How the Finance Gurus Get Risk All Wrong,”
Fortune, 99-100.
Varian, Hal, May 3, 2007, “Sometimes the Stock Does Better Than the Investor That [Who]
Buys the Stock,” New York Times. [My correction to the title.]
“Can Twitter Predict the Future?” The Economist, June 2, 2011.
“To Infinity and Beyond,” The Economist, October 6, 2007.
“Momentum in Financial Markets: Why Newton was Wrong,” The Economist, January 6, 2011.
[optional] Chapters 2, 3, 4, and 6 in Bodie et. al.
[optional] Chapter 13 in Fabozzi et. al.

!92
Homework
Form two-stock and five-stock portfolios using the data in “Data - Stock Returns.csv”. This
file contains monthly returns on a sample of Canadian stocks in the S&P/TSX index as at June
2014. Work with stocks ATD.B, BA, BNS, PJC.A, and YRI. For exercises 1 to 11 your in-sample
or estimation period is the first 60 months, January 2008 through December 2012. For exercise
12, use the last 12 months as your out-of-sample period. Assume that the risk-free rate of interest
is 0.18 per cent whenever your calculations require it.

1. Make a table that summarizes the returns on the five stocks. Your table should include means,
medians, standard deviations, annualized holding period returns, and the percentage of
months in which each stock’s return was negative. Interpret your table. What is the difference
between mean and median returns? What is the difference between mean return and holding
period return? Does a higher mean return go hand in hand with a higher holding period re-
turn?

2. Compute the covariance and correlation matrix of returns. Which companies’ returns are
most strongly correlated? Which are least correlated? Negatively correlated? Can you suggest
why?

3. Make a table of portfolios of ATD.B and BA. Let the investment weight in either stock go
from -120 per cent to 120 per cent in increments of 10 per cent. Include the minimum-vari-
ance portfolio as one of the portfolios. Report the investment weights, expected return, and
standard deviation for each portfolio. Explain your table. What are the investment weights?
What does it mean for an investment weight to be negative?

4. Plot the two-stock frontier. Explain it. Why does the frontier pass through the points repre-
senting a 100 per cent investment in ATD.B or BA? Must it?

5. What would the portfolio frontier for ATD.B and BA look like if the correlation of their re-
turns was different from what you computed in exercise 2? Plot your original frontier from
exercise 4 and portfolio frontiers for assumed correlations of -1, -0.5, 0.5, and +1 all in one
graph. Explain why the frontiers look the way they do. Is there practical advice that comes
from this?

6. Make a table of portfolios of all five stocks like the one you made in exercise 3. Include the
minimum-variance portfolio and portfolios that have the same expected returns as each of the
five stocks. Describe your table.

7. What is the expected return on the Market portfolio if you assume that the five stocks are the
entire market? Hint: use .

8. What is the composition of the Market portfolio?

9. Make a table like the one you made in exercise 6 but this time for portfolios that lie on the
Capital Market Line. Let the investment in the Market run from 0 per cent to 1.3 per cent in

!93
 increments of 10 per cent. Include portfolios that have the same expected returns as each of
the five stocks. Explain your table, making comparisons to your table from exercise 6.

10. Plot the Capital Market Line and the five-stock frontier. Describe your plot.

11. Your friend Nigel wants an efficient portfolio of the five stocks (not with the risk-free asset and
Market portfolio) that has the same expected return as Yamana Gold. What is the composi-
tion of the portfolio? What is its standard deviation?

12. How well did Nigel’s portfolio perform during the 12 months following the in-sample period?
Fill in the table and see.

!94
Answers
1. Couche Tard (ATD.B) performed best, earning an average monthly return of 2.1 per cent or
about 23 per cent a year. Yamana Gold (YRI) was the riskiest. All five stocks lost money 40 to
45 per cent of the time.

Monthly Return Performance, January 2008 through December 2012

Mean Median Std. Dev. HPR p.a. % Months Neg.

ATD.B 0.02133 0.00821 0.09009 0.22967 0.40

BA 0.00591 0.00535 0.03668 0.06468 0.38

BNS 0.00743 0.00521 0.05769 0.07120 0.47

PJC.A 0.00819 0.00385 0.06428 0.07663 0.45

YRI 0.01592 0.02983 0.14304 0.06975 0.45

2. Covariance matrix V and the correlation matrix 𝜌.

0.00811686 0.000266965 0.0000145114 0.00183428 −0.00206091

0.000266965 0.00134522 0.000341095 0.000343396 0.000195863
V
" = 0.0000145114 0.000341095 0.00332799 0.000620237 −0.000179134
0.00183428 0.000343396 0.000620237 0.00413209 0.00042017
−0.00206091 0.000195863 −0.000179134 0.00042017 0.0204606

1. 0.0807912 0.00279206 0.316729 −0.159922

0.0807912 1. 0.161208 0.145651 0.0373333
"ρ = 0.00279206 0.161208 1. 0.167256 −0.0217084
0.316729 0.145651 0.167256 1. 0.0456964
−0.159922 0.0373333 −0.0217084 0.0456964 1.

Couche Tard and Jean Coutu are the most strongly correlated. The least correlated are
Couche Tard and Bank of Nova Scotia. Couche Tard and Bank of Nova Scotia are negative-
ly correlated with Yamana Gold but Jean Coutu is not. Your thoughts?

!95
3. Portfolios of ATD.B and BA. It’s pretty easy to spot the minimum-variance portfolio.

Investment Weights
"σp "E (Rp ) ATD.B BA

0.129575 -0.012597 -1.20000 2.20000

0.111489 -0.009513 -1.00000 2.00000

0.093725 -0.006429 -0.80000 1.80000

0.076507 -0.003345 -0.60000 1.60000

0.060302 -0.000261 -0.40000 1.40000

0.046192 0.002824 -0.20000 1.20000

0.036677 0.005908 0.00000 1.00000

0.034857 0.007770 0.12077 0.87923

0.035652 0.008992 0.20000 0.80000

0.043716 0.012076 0.40000 0.60000

0.057144 0.015160 0.60000 0.40000

0.073034 0.018245 0.80000 0.20000
0.090094 0.021329 1.00000 0.00000

0.107768 0.024413 1.20000 -0.20000

4. Two-stock frontier.
E rp )

ATD.B
0.02133

mvp
0.00777
BA
0.00591

0.03486
σp
0.03668 0.09009

!96
5. The effect of the correlation between two stocks on the shape of their portfolio frontier.
E rp )

ATD.B
0.0213

0.0059 BA

σp
0.0367 0.0901
For the five-stock portfolios you needed to compute the frontier parameters

A = 9.0093; B = 0.109394; C = 1,104.41; D = 39.6486

and for the optimal portfolio weights

g = {-0.37991, 1.0288, 0.15962, 0.28598, -0.0944881}

h = {56.5372, -55.684, 4.52161, -22.6687, 17.2939}

!97
6. Five-stock portfolios.

Investment Weights
"σp "E (Rp ) ATD.B BA BNS PJC.A YRI

0.0525270 0.0000000 -0.3799100 1.0288000 0.1596200 0.2859800 -0.0944881

0.0450701 0.0018000 -0.2781430 0.9285660 0.1677590 0.2451770 -0.0633591

0.0343973 0.0050000 -0.0972238 0.7503780 0.1822280 0.1726370 -0.0080188

0.0323489 0.0059077 -0.0459031 0.6998310 0.1863320 0.1520600 0.0076795

0.0303353 0.0074294 0.0401267 0.6151000 0.1932130 0.1175660 0.0339947

0.0300908 0.0081576 0.0812953 0.5745520 0.1965050 0.1010600 0.0465875

0.0300914 0.0081929 0.0832957 0.5725820 0.1966650 0.1002570 0.0471994

0.0316230 0.0100000 0.1854620 0.4719580 0.2048360 0.0592937 0.0784506

0.0404186 0.0132705 0.3703700 0.2898400 0.2196240 -0.0148453 0.1350110

0.0470064 0.0150000 0.4681480 0.1935370 0.2274440 -0.0540497 0.1649200

0.0508105 0.0159150 0.5198780 0.1425880 0.2315810 -0.0747908 0.1807430

0.0693682 0.0200000 0.7508340 -0.0848826 0.2500520 -0.1673930 0.2513890

0.0757477 0.0213287 0.8259550 -0.1588700 0.2560600 -0.1975130 0.2743680

0.0938458 0.0250000 1.0335200 -0.3633030 0.2726600 -0.2807360 0.3378590

!98
7. The expected return on the Market portfolio is 1.3 per cent (0.0132705).

8. The composition of the Market portfolio is

{0.37037, 0.28984, 0.219624, -0.0148453, 0.135011}

Knowing the weights, the standard deviation of the Market portfolio is 0.0404186.

9. Portfolios on the Capital Market Line.

Investment Weights
"σp "E (Rp ) F M

0.006343 0.000000 1.15692 -0.15692

0.000000 0.001800 1.00000 0.00000

0.011276 0.005000 0.72103 0.27898

0.014474 0.005908 0.64189 0.35811

0.019836 0.007429 0.50923 0.49077

0.022402 0.008158 0.44575 0.55425

0.022527 0.008193 0.44267 0.55734

0.028894 0.010000 0.28513 0.71487

0.040419 0.013271 0.00000 1.00000

0.046513 0.015000 -0.15077 1.15077

0.049737 0.015915 -0.23054 1.23054

0.064131 0.020000 -0.58667 1.58667
0.068813 0.021329 -0.70251 1.70251

0.081750 0.025000 -1.02257 2.02257

!99
10. The five-stock frontier and the Capital Market Line.

E rp )

0.02737
0.02425
0.02133 ATD.B
YRI
0.01591 M
0.01327
PJC.A
0.00816 mvp BA
BNS
f
0.03009 0.07575
σp
0.04042 0.06881 0.09009 0.14304

!100
11. The composition of Nigel’s portfolio is

{0.519878, 0.142588, 0.231581, -0.0747908, 0.180743}

The standard deviation is 5.1 per cent (0.0508105), which is considerably less than YRI’s
standard deviation of 14.3 per cent (0.14304).

12. Performance of Nigel’s portfolio. In-sample is Jan 2008 to Dec 2012 (60 months); out-of-sam-
ple is Jan 2013 to Dec 2013 (12 months).

Mean Median Std. Dev. HPR p.a. Months Neg. (%)

In-sample 0.01592 0.00577 0.05081 0.16308 0.43
Out-of-sample 0.02152 0.02100 0.02968 0.28464 0.25

E rp )
ATD.B

PJC.A
Nigel
0.0215 BNS Nigel ATD.B YRI
0.0159
BA BA BNS
PJC.A
σp
0.0297 0.0508

In-sample: Jan 2008 to Dec 2012 ( 60 months)

Out-of-sample: Jan 2013 to Dec 2013 ( 12 months)
YRI

!101
 12
CHOICE UNDER CERTAINTY

In this chapter, we’ll look at the details of some of the results in Chapter 9. If you need to, review
Chapter 2, Utility Theory, and then sections “Optimal Consumption,” “Real Investment,” and
“Equilibrium Interest Rate” in chapter 9.

Optimal Consumption
Lucy, as you recall, lives in a riskless, two-period world. She is endowed with $1,000 today
and $200 next year. At an interest rate of seven per cent, her wealth is $1,186.92.

e1 $200
W
" 0 = e0 + = $1,000 + ≈ $1,186.916
1+r 1.07

She choses to consume $475 now and $762 next year by lending $525 at seven per cent to earn
$562 next year. How do we come up with those numbers?
C1
W1
1270

C* : MRS = 1 + r = 1.07
762

Earn $562 in interest u* = 630.616

e ue = 380.731
200
Lend $525
W0
C0
0 475 1000 1187
Lucy’s tastes for consumption are represented by a Cobb-Douglas utility function,
" (C0, C1) = C0αC11−α, with 𝛼 = 0.4. She maximizes her utility with respect to her current and future
u
consumption, subject to the constraint that she cannot consume more than her endowed wealth.
For any utility function, u" (C0, C1), this can be written

C1 e
" Ma x u (C0, C1) s.t. C0 + = e0 + 1 = W0
<C0 ,C1> 1+r 1+r

!102
The solution to the optimization problem is

u′(C0 )
C * : MRS = =1+r
u′(C1)

which says that utility is maximized when a person consumes so that their marginal rate of substi-
tution is equal to one plus the interest rate.
How does this relate to the graph? MRS is the absolute value of the slope of an indifference
curve

δC1
M
" RS =
δC0

which you can find by differentiating u separately by C

" 0 and C
" 1, and then taking the ratio of the
two derivatives
δu
δC1 δC0 u′(C0 )
" RS =
M =
δu
≡
δC0 u′(C1)
δC1

You can see that MRS is the ratio of marginal utilities for current and future consumption. It is a
“personal” price ratio, one at which a person is willing trade future dollars for dollars now. The
price ratio that the market trades future for current dollars is one plus the rate of interest; it is the
slope of the wealth constraint

δC1
" =1+r
δC0

When the two price ratios are equal a person is consuming at the point where they place the same
value as the market does on having one more dollar now. There are no transactions that can make
the person better off.
For Lucy,

α C1
"M R S =
1 − α C0

At her optimum,

α C1
" =1+r
1 − α C0

and then combining with the wealth constraint, gives

( 1 + r)
e
C*0
= αW0 = α e0 + 1
"
C*
1
= (1 − α)(1 + r)W0 = (1 − α)((1 + r)e0 + e1))

!103
for Lucy’s smoother consumption of $474.77 now and $762 next year.

Demand for Consumption

When we say demand for consumption, we usually mean current consumption. The de-
mand for consumption is a function of the interest rate—the price— for a given endowment and
taste parameters.

C = f (r, e, taste parameters)

Lucy’s demand function for consumption is the expression for her optimal current consumption
but with the interest rate allowed to vary.

( 1 + r)
e
" 0 = α e0 + 1
C

It can be written for any year looking one year ahead

( 1 + r)
et+1
Ct = α et +

for Lucy’s Cobb-Douglas utility function.

C0

480
478

475

471
468
466

455

r
0.02 0.07 0.12 0.17 0.22 0.45

!104
Real investment
If Lucy has opportunities to invest in real production, she will do so to make herself as
wealthy as possible and then consume according to her tastes in line with that greater wealth. In-
vestment followed by consumption: it is usually described as happening in that order but really the
idea is that it just happens—the decision, that is—all at once.
Her investment opportunities are represented by the concave investment opportunity
schedule or production function in the figure. It is a parabola

1
" 1 = g − hC02, g = $1,000, h =
C
$1,250

C1

1358
1270

P* : MRT = 1 + r = 1.07
642

Earn $442 on investment

e
200
NPV ( I* ) =*$82
Invest I* = $331 W0 W0
C0
0 669 1000 11871269

The concavity means that, going leftwards from e, the first dollar she invests earns the most, the
next a little less, the third still less, and so on. In other words, the absolute value of the slope of
her investment opportunity schedule (marginal rate of transformation or MRT) is one plus the
marginal return on investment.

δC1
M
" R T = 1 + ROI = = 2hC0
δC0

The MRT on the first dollar Lucy invests is 1.6 or a return of 60 per cent. The second dollar
earns 59.92 per cent and the third 59.84 per cent. You can find the MRT for any investment op-
portunity schedule written as an implicit function, T
" (C0, C1), the same way we found the marginal
rate of substitution for an indifference curve.
δT
δC1 δC0
" R T = 1 + ROI =
M =
δT
δC0
δC1

!105
 Step 1. Lucy maximizes her wealth by making sure that every dollar she invests either in-
creases or maintains her wealth; in other words, every dollar invested earns a non-negative net
present value (NPV ≥ $0).

( 1 + r)
C1
" Ma x W0 = C0 + s.t. T (C0, C1) = 0
<C0 ,C1>

The solution is point P*, where the wealth constraint is tangent to the investment opportunity
schedule as

P* : M R T = 1 + r

Solving for Lucy’s IOS,

M R T = 2hC0 = 1 + r
1+r
" ∴ C0 = = 668.75
2h
∴ P* = (668.75, 642.219)

She invests about $331 = $1,000 - $669 and increases her wealth by $82 to $1,269, the net
present value of the investment.

$ROI $442.219
N PV (I*) = − I* + = − $331.25 + = $82.0386
" 1+r 1.07
W*
0
= W0 + N PV (I*) = $1268.95

Step 2. Lucy chooses the consumption stream that maximizes her utility. You already
know how to do this. It is the solution to

α C1
C*
" : MRS = =1+r
1 − α C0

applied to Lucy’s new level of wealth. The details are in the figure on the next page.

!106
 The result for an exchange (borrowing and lending) and production (real investment) is
called Fisher Separation because optimal consumption is independent of optimal investment.
How you want to spend does not interfere with how you should invest.
Fisher Separation

P* : M R T = 1 + r
C* : M R S = 1 + r

C1

1358
1270

C* : MRS = 1 + r = 1.07
815
Earn $172 in interest P* : MRT = 1 + r = 1.07
642
u* = 674.204
Earn $442 on investment

e ue = 380.731
200
NPV ( I* ) =*$82
Lend $161 Invest I* = $331 W0 W0
C0
0 508 669 1000 11871269

!107
The Equilibrium Interest Rate
Let’s work out the three per cent equilibrium interest rate from Chapter 9. The equilibrium
is shown as a Pareto Optimum in an Edgeworth box. The solution is a general equilibrium be-
cause both the interest rate and the allocation of optimal consumption are determined jointly.
We’ll show that the interest rate and optimal consumption depend on tastes (culture, demography)
and endowments (the distribution of income) but do not depend directly on each other.

C1 Ricky
102
u C0 , C1 = α ln C0 + ( 1 -α) ln C1
α = 0.44918 ( Lucy) , α = 0.5413 ( Ricky)

e
70

Lucy repays
$22.37
C* : ( MRSL = MRSR )
47.63 ⇔ p * = 1 + r * = 1.03
Lucy borrows
$21.71

C0
0
0 16 37.71 100
Lucy

Step 1. Lucy and Ricky, acting as price takers, maximize their utilities. This means that
each seeks to borrow or lend an amount so that their marginal rate of substitution is equal to one
plus the interest rate. But the interest rate doesn’t exist, except in equilibrium, so Lucy and Ricky
are not “comparing,” for lack of a better term, their own marginal rate of substitution to an actu-
al interest rate but are comparing it implicitly to the other’s marginal rate of substitution because
they are, after all, transacting with one another. There are mutually-beneficial loans to be made as
long as their marginal rates of substitution differ; equilibrium exists when they are the same.

!108
 Since Lucy and Ricky have the same utility function, u" (C0, C1) = α ln(C0 ) + (1 − α)ln(C1),
the condition for utility maximization is the same for both

α C
" RS =
M ⋅ 1 =1+r
1 − α C0

implying each will consume

( 1 + r)
e
" 0 = αW0 = α e0 + 1
C*

" 1 = (1 − α)(1 + r)W0 = (1 − α)((1 + r)e0 + e1))

C*

according to the value of their individual taste parameter 𝛼 and their endowments.
Step 2. Solve for the equilibrium interest rate by imposing the market-clearing condition.
It’s just Supply = Demand, the bookkeeping that ensures everything adds up. The total of Lucy
and Ricky’s optimal current consumption from step 1, for example, must equal the total of their
current endowments.

e0L + e0R = C*
0
+ C*
0
L R

( 1 + r* )
e1L
( 1 + r* )
" e1R
= αL e0L + + αR e0R +

You could also apply the analogous market-clearing condition for future consumption, but you
don’t need to because one is all it takes to balance the books. Now the market-clearing condition
for the equilibrium interest rate.

αL e1L + αR e1R
1 + r* =
(1 − αL )e0L + (1 − αR )e0R

Notice that interest rate depends on tastes and endowments but not on final optimal consumption.
Evaluating r* with the parameter values in our example gives an equilibrium interest of three per
cent.
Step 3. Solve for the optimal allocation of consumption. Here is Lucy’s optimal current
consumption from

( 1 + r* )
e1L
C*
" 0 = αL e0L +
L

Substitute in the expression for r* to eliminate it and get

(1 − αL )e0L + (1 − αR )e0R
( )
C*
0L
= αL e0L + e1L
αL e1L + αR e1R

!109
which, like the interest rate, depends only on tastes and endowments. It evaluates to 37.7137 in
the example. Work out the other three consumption values (see table). You don’t need more fancy
equations to do it.

A Positive Interest Rate in a Pure Exchange Economy

Lucy Ricky

Taste parameter 𝛼 0.44918 0.5413

Endowment e {16, 70} {84, 32}

Optimum C* {37.7137, 47.6349} {62.2863, 54.3651}

MRS(e) 3.56771 0.449552

MRS(C*) = price ratio 1.03 1.03

Utility(e) 3.58555 3.98813

Utility(C*) 3.75866 4.06935

Reading
Benartzi, Shlomo and Richard H. Thaler, 2007, “Heuristics and Biases in Retirement Savings
Behavior,” Journal of Economic Perspectives, 21(3):81-104.
Hayek, Friedrich A., 1989, “The Pretence of Knowledge,” American Economic Review, 79(6):3-7.
Loewenstein, George and Drazen Prelec, 1992, “Anomalies in Intertemporal Choice: Evidence
and Interpretation,” Quarterly Journal of Economics, 107(2):573-597.
[optional] Chapters 2 and 3 in Fabozzi et. al.

!110
Homework
1. Lucy lives in a nice neighbourhood in a riskless, two-period, Fisher-type economy. This is
what we know about her situation.
The interest rate: r = 3.9 per cent
Endowment: e = ($240,000, $106,400)
31
Utility function: "u (C0, C1) = − e −αC0 − e −αC1, α =
10,000,000
C02 C12
Investment opportunity schedule: 1" = + , a = 250,000, b = 380,000
a2 b2
Fill in the table and draw a graph, as well-labelled as practical, showing the Fisher Separation
results. Explain it using plain language. Also draw a second graph showing Lucy’s demand for
current consumption as a function of the interest rate.

Without Real With Real

Investment Investment

MRS at C*

MRT at P* (1 + Marginal ROI) -

Optimal Production (P*) -

Optimal Investment (I*) -

Dollar ROI -

ROI -

NPV(I*) -

Wealth
Optimal Consumption (C*)

Utility at C*

Amount Lent (Borrowed)

Principal + Interest Earned (Repayed)

2. Read Loewenstein and Prelec (1992), and write a short description of the anomalies of in-
tertemporal choice discussed in the article. Concentrate on the ideas; ignore the math.

!111
3. Compute the equilibrium interest rate and consumption optimum for a two-person, two-peri-
od, riskless economy. Both Lucy and Ricky have Cobb-Douglas utility functions of the form

u" (C0, C1) = α ln(C0 ) + (1 − α)ln(C1)

Lucy’s α is 0.46 and Ricky’s 0.54. Her endowment is (16, 70) and his is (84, 33).
a. What is the growth rate of the economy?
b. What is the equilibrium rate of interest?
c. Is Ricky a borrower or lender in equilibrium? How much?
d. What is the consumption (Pareto) optimum? Illustrate the Pareto optimum in a drop-dead
gorgeous, well-labeled Edgeworth box diagram.
e. All things being equal, what would be the effect on the interest rate if the growth rate of
economy was higher?
f. All things being equal, what would be the effect on the interest rate if Ricky’s α was small-
er?

!112
Answers
1. Fisher Separation. Make sure you can interpret all of the figures. Math follows.

Without Real Investment With Real Investment

MRS at C* 1.039 1.039

MRT at P* (1 + Marginal ROI) - 1.039

Optimal Production(P*) - {$141,078, $313,713.}

Optimal Investment (I*) - $98,921.50

Dollar ROI - $207,313.00

ROI - 109.6%

NPV(I*) - $100,610.00
Wealth $342,406.00 $443,016.00

Optimal Consumption (C*) {$168,425, $180,766} {$219,692, $232,033}

Utility at C* -1.16426 -0.993179

Amount Lent $71,575.00 -$78,613.50

Interest Earned $74,366.50 -$81,679.40

The results graphed.

Lucy's consumption optimum

C1
460,293

355,760
P*Borrow
313,713
Repay
C*
232,033
Earn

e
106,400 Invest

240,000
C0
141,078 219,692 342,406 443,016

!113
Math for the Fisher-Separation result. Solve for the optimal production point P* and the op-
timal investment I* by equating the marginal rate of transformation to one plus the rate of
interest

b 2 C0
" : MRT =
P* ⋅ =1+r
a 2 C1

Plug that back into the equation for the IOS to get P*. Then solve for Lucy's maximized
wealth and find her consumption optimum.

C* : M R S = e α(C1−C0 ) = 1 + r
" 1
∴ C*1
= C*0
+ ln(1 + r)
α

Throw that into the wealth constraint to get

( 1+r )
1+r 1 ln(1 + r)
C*
0
= W0 − ⋅
2+r α

2+r ( 0 α )
1+r 1
C*
1
= W + ln(1 + r)

Lucy’s demand for consumption as a function of her wealth with and without real investment.
Can you tell which is which? Demand is downward sloping for “normal” interest rates but
turns upwards when interest rates are at stratospheric level.

Lucy's demand for current consumption

C0

240 000

220 000

200 000

180 000

160 000

140 000

120 000
r
0.5 1.0 1.5 2.0

!114
 Words with little economic mumbo jumbo. Lucy is worth $342,406 if she doesn't invest in her busi-
ness. This amount is the $240,000 she has today plus $102,406, which is the value today at
3.9 per cent of the $106,400 she has coming to her next year. Given her tastes and the fact
that next year will be leaner than this year for her, Lucy has a strong desire to have more cash
available next year, in fact, so much so that she values her next current dollar at -33.9 per
cent. This negative “personal rate of interest” means she'd pay someone to help her build a
bigger nest egg! No need for that though. She can make herself better off by investing in her
business—to increase her wealth obviously—and then borrowing or lending to spread her
cash more evenly between this year and next. Her real business opportunities are such that
she’ll get back about 4.21 dollars on the first dollar she invests, and earn more than the rate of
interest on every dollar after that up to $98,922 invested for a total income of $207,313—an
average return of almost 109.6 per cent! This makes her richer by $100,610. Lucy's tastes at
this higher level of wealth are such that she'd like to spend more next year without having to
cut back by much this year, so she'll borrow $78,613, yet still end up with a smoother spending
stream of $219,692 and $232,033.

2. There are three inconsistencies observed in intertemporal decisions. (1) Discount rates decline
with the time to be waited. An amount of money has a bigger implied present value if re-
ceived later (dynamic inconsistency & self control). (2) Discount rates decline with the total
amount of money. Small amounts are discounted at implausibly high rates (magnitude effects,
& absolute differences & mental accounts). (3) Discount rates are higher for gains than for
losses. People need to be paid a lot to wait for a reward but are unwilling to pay to delay a
fine (sign effects, reference points & debt aversion). The authors attribute these to the behav-
ioural tendencies indicated in parentheses above, all of which are associated with an overrid-
ing tendency to make decisions relative to context-dependent reference points that are not
accounted for in economic models. Loss aversion is used to explain our preference for increas-
ing consumption profiles, and costly self-control is used to explain our preference for increas-
ing income profiles, with our various decisions being further tempered by savouring and
dread. What policy implications does this have for the use of social discount rates? The an-
swer is that social discount rates become all but useless.

!115
3. The equilibrium interest rate.
a. The growth rate of the economy is three per cent.
b. The equilibrium rate of interest is 5.7953 per cent. Why is it different from the growth
rate?

αL e1,L + αR e1,R
1 + r* = = 1.0579526
(1 − αL )e0,L + (1 − αR )e0,R

c. Ricky lends $21.80 to Lucy.

d. The Pareto optimum or consumption optimum read from Lucy’s origin is {37.7961,
46.9407}. Here’s the expression for current consumption.

( 1 + r* )
e1,L
C*
0L
= αL WL = αL e0,L +

(1 − αL )e0,L + (1 − αR )e0,R
( )
= αL e0,L + e1,L = 37.7961
αL e1,L + αR e1,R

C1 Ricky
103
u(C 0 , C 1 ) = α ln(C 0 ) + (1-α) ln(C 1 )

α = 0.46 (Lucy), α = 0.54 (Ricky)

e
70

Lucy repays
$23.06
C* : ( MRSL = MRSR )
46.94 ⇔ p * = 1 + r * = 1.05795
Lucy borrows
$21.80

C0
0
0 16 37.8 100
Lucy

e. The interest rate would be higher if the growth rate was higher. See the expression for r*
above. A bigger e1 or smaller e0 for either Lucy or Ricky or both of them implies a higher
growth rate growth rate and higher r*. Does that make sense?

!116
f. The interest rate would be lower if Ricky’s α " was smaller. Once again, see the expression
for r*. It doesn't matter whose α we're talking about because, for both Lucy and Ricky, "α
gauges the preference for current consumption. A smaller α means a relatively weaker
preference for current consumption, which implies a lower interest rate.

!117
 13
CHOICE UNDER UNCERTAINTY

In the St. Petersburg gamble a coin is tossed until it comes up tails, and for every toss that it
comes up heads, a starting amount of money, say $1, is doubled. The prize is the amount to
which $1 has grown, "$2n, where n is the number of consecutive heads. Daniel Bernouli
(1700-1782), mathematician and physicist, considered the gamble a paradox because the expected
prize is infinite yet most people are not willing to pay more than a modest amount to play the
game.

N
1
"E ( pr i z e) = lim ∑ n $2n = $1 + $1 + $1 + . . . = ∞
N→∞
n=1
2

This led Bernouli to suggest that perhaps people place an expected psychological value, v, on the
gamble that is a concave function of the prize

N
1
f ($2n) < ∞,
N→∞ ∑ 2n
v" ( pr i z e) = lim
n=1

where f is a strictly concave function. If f is square root, for example, the expected psychological
value is only ⅓.
N
1 1
$2n =
∑
v" ( pr i z e) = lim
N→∞ 2n
n=1
3

Expected Utility Theory

Bernouli’s mathematical solution to the St. Petersburg paradox parallels economics’ par-
adigm of treating people as displaying diminishing marginal utility of consumption. Every slice of
pizza provides satisfaction but less satisfaction than the slice before. And so with money. Suppose
your wealth is $100 right now but there’s a 52 per cent chance that in the next instant you’ll lose
$72 and a 48 per cent chance that you’ll gain $78. The change in your wealth in the immediate
future is the gamble

" ΔW ∼ (ΔW1 = − $72, ΔW2 = $78; π1 = 0.52)

and, in this case, a fair gamble because it has an expected value of $0, or saying the same thing,
your expected wealth under the gamble is equal to your current wealth.

!118
 E (ΔW ) = π1 ⋅ ΔW1 + π 2 ⋅ ΔW2
" = 0.52( − $72) + 0.48($78) = $0
∴ E (W ) = W0 = $100

Your wealth is the distribution

" ∼ (W1, W2; π1) = ($28, $178; 0.52)

W

In economics, Bernouli’s expected psychological value becomes expected utility

E [u (W )] = π1u (W1) + π 2 u (W2 )

with the only change in the marginal rate of substitution compared to non-risky choices of apple
π1
and oranges being the appearance of the odds ratio, " ,
π2

π1 u′(W1)
MRS =
π 2 u′(W2 )

If wealth yields diminishing marginal utility, then risky outcomes like ($28, $178) must lie
on a convex indifference curve (point a in the figure), where here the person is assumed to have a

W2

204
EuW = π1 W1α + π2 W2α , α = 0.4
a
178

152

126
E ( W ) = π1 W1 + π2 W2
W * : MRS = 1 = 1.08333
π
100 π2

CE
80.53

48 6.30957

5.78607

W1
4 28 52 80.53 100 124 148 192.31

utility function of the form

u" (W ) = W α , α = 0.4

!119
and therefore expected utility

" [u (W )] = π1W1α + π 2W2α

E

All of the distributions of wealth on the indifference curve passing through a yield the same ex-
pected utility: 5.79. There is one distribution in particular on that initial indifference curve that
tells us the person is risk averse. That distribution is completely free of risk and is denoted by CE for
certainty equivalent. You can find the certainty equivalent at the intersection of an indifference curve
and a 45-degree line passing through the origin. The certainty equivalent in the example is
$80.53. How does the certainty equivalent tell us that the person is risk averse? It does so because
it is less than the person’s expected wealth of $100 under distribution a. Think about it. The per-
son is indifferent between risky distribution a and riskless amount CE

u (CE ) = E [u (W )]

In everyday language, CE is the answer to the question: what is the smallest sure amount you’d be
willing to accept if all risk was taken away from you? A risk averse person is always willing to ac-
cept a sure amount ($80.53) that is less than their expected wealth ($100) when faced with risk. Do
the math.

u (CE ) = E [u (W )]
∴ CE α = π1W1α + π 2W2α

" ∴ CE = α π1W1α + π 2W2α

0.4
∴ CE = 5.78607
∴ CE ≈ $80.53

What about the line passing through a?

" (W ) = π1W1 + π 2W2 = $100

E

It is not a wealth constraint or a budget line; it is line of distributions of equal expected wealth,
π1
and its slope is equal to the odds ratio, . Since W0 is $100, all of the distributions on this partic-
π2
ular line are fair gambles. Points lying above the line are distributions with expected wealth
greater than $100, and those below, less than $100. The distributions become less risky moving
down the line from the top left because the spread between W1 and W2 is smaller for the fixed
probabilities. Risk falls until there is no risk at all at W1 = W2 = $100, where a ray from the origin
at 45 degrees intersects the line, and then risk increases all the way down to the horizontal inter-
cept.
The line of equal expected wealth is not a constraint; the person represented in the figure
wouldn’t ordinarily be restricted to choosing only from among the distributions on the line. All
that can be said is that the outcomes the person currently faces are represented by a, and paired
with the fixed probabilities, that is the distribution of their wealth. Their preferred distributions
are the ones that lie above the indifference curve passing through a—expected utility is higher.

!120
Can you pick out areas in the graph matching the nine distribution classes in the table? Which are
preferred?

Risk Compared to Point a

Lower Same Higher

Higher
Expected Wealth
Same
Compared to Point a
Lower

Working with Utility Functions Instead of Indifference Curves

From this point on we’ll work with utility functions. They provide the same information as
indifference curves but are somewhat handier when dealing with risky prospects because all of the
outcomes (W1 and W2) can be shown on one axis and utility on the other. Here is our original ex-
" ∼ (W1, W2; π1) = ($28, $178; 0.52), shown on the utility function for
ample for risky wealth, W
α
" (W ) = W , α = 0.4, rather than point a on one of the utility function’s indifference curves. Once
u

uW

uW = Wα , α = .4
7.94636 u W2

6.30957 u[ E ( W )]
5.78607 E [ u( W )]

3.79196 u W1

Risk Premium
19.5

W0
W1 CE E(W ) W2
W
28 80.5 100 178

again, we know this is a fair gamble because current wealth is equal to expected wealth. The ex-
pected utility of the gamble can be found at the intersection of the line connecting u(W1) and
u(W2) and the vertical line through E(W). It is the red point on the right. You already know how to
calculate the certainty equivalent. The red point on the left picks off CE below ($80.53) because

!121
that sure amount yields the same utility as the expected utility of the gamble (the red point on the
right).
There are three ways that a concave utility tells us a person is risk averse:

• u[E (W )] > E [u (W )]:—Your utility of expected wealth is greater than the expected utility
of wealth. You’d rather have $100 for sure than a gamble paying $28 or $178 and ex-
pected payoff of $100.

• CE < E (W ):—Your certainty equivalent is less than your expected wealth. You are indif-
ferent between CE ($80.53) and the gamble. You would trade the risk for any sure
amount greater than or equal to CE.

• Risk Premium ≡ E (W ) − CE > 0:—Your risk premium for the gamble is positive. This is
just another way of looking at the certainty equivalent. You’d be willing to pay an
amount up to the risk premium in order to have the risk removed. Think of it as a per-
sonal valuation of an insurance premium.

!122
 The next graph illustrates the definition of risk aversion. Expected utility is higher (5.974)
under a safer gamble, "(W′1, W′2; π1) = ($40, $165; 0.52) with the same expected wealth as the
original gamble. The changes with the certainty equivalent and the risk premium follow suit. A
change in risk without a change in the expected outcome is called a mean-preserving change in spread.
For this safer gamble, it is a mean-preserving decrease in spread. Can you visualize in the graph

uW
uW = Wα , α = .4

5.97448

12.8

19.5

CE′
W1 W′1 CE E ( W ) W′2 W2
W
80.5
87.2
100

165
178
28
40

what would happen to expected utility with repeated applications of a mean-preserving decrease
in risk?

Risk aversion:—If faced with two or more gambles having the same expected payoff, a
risk averse person prefers the least risky.

!123
 Most risks we face are not fair gambles; expected wealth is bigger or smaller than its current
level. Consider taking on a gamble with payoffs (" ΔW1 = $24, ΔW2 = 40; 0.52) and a positive ex-
pected payoff of $31.68, which would increase your expected wealth to $131.68 from $100. Tak-
ing on the gamble would increase your expected utility to 6.662, so you clearly prefer the distribu-

uW
uW = Wα , α = .4

6.66225

14.6
W0 CE′
W1 W′1 CE E ( W ) E(W ′ ) W2 W′2
W
131.68
114.6
80.5

100

178

218
28

52

tion of your wealth with the gamble than without it. But would you be willing to pay to take on
this new risk? The answer is yes; you’d be willing to pay up to $14.60, the difference between CE’
and W0.

Coefficient of Risk Aversion

It’s hard to say that one person is more risk averse than another unless they have similar
wealth and you get to see them respond to a similar risk, for example, the amount each is willing
to pay for a risky investment. Economics uses a standardized measure of sorts in modelling to rep-
resent differences in risk aversion. ARA is the coefficient of absolute risk aversion, and RRA is the
coefficient of relative risk aversion.

u′′(W )
A R A(W ) = −
u′(W )
u′′(W )
R R A(W ) = − W = − W ⋅ A R A(W )
u′(W )

ARA is simply the ratio of second derivative of the utility function to the first derivative. Because
the second derivative of a concave function is negative, the negative sign in front of the fraction
ensures that ARA is positive for anyone who is risk averse. The intuition, although perhaps not a
straightforward intuition, behind the coefficient is this: it is the rate of decline in marginal utility
(the numerator) per unit of marginal utility (the denominator). A rougher way to say the same
thing is that it tells us how fast marginal utility is declining. The faster the decline, the more risk

!124
averse the person is given the risk that they face. Here’s how to interpret the coefficients for some-
1
one with log utility, u" (W ) = ln(W ). The coefficient of absolute risk aversion is " , which is de-
W
creasing in W; the person becomes less risk averse as they get wealthier, and they will put more of
their wealth in dollar terms at risk, for example, by investing in risky assets. The coefficient of rel-
ative risk aversion is 1, a constant. The person’s relative risk aversion is unchanged as their wealth
changes; they keep a constant percentage of their wealth at risk. So a log utility maximizer display
declining absolute risk aversion (puts more dollars = absolute at risk as they become wealthier) and
constant relative risk aversion (they keep a constant percentage of their wealth put at risk)

Pricing a Risky Asset

Here’s how expected utility theory is used to developed a rudimentary pricing model, that
is, one that tells us the return that is required by an investor depending on their level of risk aver-
sion and the riskiness of the asset.
Let an investor’s wealth be comprised of the market values of just two assets: risk-free debt
M and a risky stock S.

" 0 = M+S
W

If the person looks just one period ahead, say a year, their uncertain end-of-period income, y, is

y = rf M + r S
"
= rf W0 + S(r − rf )

where rf is the return on the safe bond and r is the risky return on the stock. The investor maxi-
mizes their expected utility with respect to their dollar investment S in the stock

" Max E [u (y)] = Max E [u (rf W0 + S(r − rf ))]

The condition for the maximum (called a first-order condition) is

d E [u (y)]
" = E [u′(y)(r − rf )] = 0
dS

The argument of the expectation E " [u′(y)(r − rf )] is the product of two terms, and can be written
using the definition of covariance as

"E [u′(y)]E [r − rf ] + cov[u′(y), (r − rf )] = 0

Solving for E(r) gives an expression for the return that the investor “requires”

cov[u′(y), (r − rf )]
E (r) = rf −
E [u′(y)]

!125
If the fraction in the second term is negative, the expression has the nice interpretation that re-
turn an investor requires is equal to the risk-free rate plus a positive risk premium, which depends
on the investor’s taste for risk and return. The denominator of the second term is positive because
marginal utility, u′
" (y), is always positive, so must be its expected value. That leaves the question of
whether the covariance in the numerator is negative. It is. When the return on the stock is higher
than the risk-free rate, that is, r" − rf > 0, income y will rise, and the marginal utility of income,
u′
" (y), will
For the specific case of quadratic utility, u" (y) = y − α y 2, α > 0, the risk premium can be
broken down into a risk aversion part and the riskiness of the stock. To get there, note the first
and second derivative of the utility function and its coefficient of absolute risk aversion

u′(y) = 1 − 2α y and u′′(y) = − 2α

" 2α
∴ A R A(y) =
1 − 2α y

Substitute the first derivative into the expression for required return

cov[1 − 2α y, (r − rf )]
E
" (r) = rf −
E [1 − 2α y]

Now substitute in the definition of income y

cov[1 − 2α (rf W0 + S(r − rf )), (r − rf )]

" (r) = rf −
E
E [1 − 2α (rf W0 + S(r − rf ))]

Additive constants like 1 and r" f W0 do not affect covariance, so they can be dropped, and multi-
plicative constants like 2"α, and S can be factored out, leaving

2α
E (r) = rf − cov[y, r − rf ]
1 − 2α E (y)
2α
" = rf − cov[S(r − rf ), r − rf ]
1 − 2α E (y)
2α
= rf − ⋅ S ⋅ cov[r − rf , r − rf ]
1 − 2α E (y)

2α
In the last line above, you can see A
" R A(y) = , solved at E(y), and cov(r
" − rf , r − rf ) is
1 − 2α E (y)
the variance of return of the stock (the risk-free rate in there does nothing because it is a
constant). The last line can then be written

E (r) = rf + A R A(y) ⋅ S ⋅ σr2

The risk premium on stock, demanded by a quadratic utility maximizer is proportional to their
coefficient of absolute risk aversion and the variance of the stock’s return. Tastes for risk and the
“quantity” of risk have been separated.

!126
Reading
Grether, David M. and Charles R. Plott, 1979, “Economic Theory of Choice and the Preference
Reversal Phenomenon,” American Economic Review, 69(4):623-638.
Kahneman, Daniel and Amos Tversky, 1979, “Prospect Theory: An Analysis of Decision Under
Risk,” Econometrica, 47(2):263-292.
Lypny, Gregory J., 1993, “An Experimental Study of Managerial Pay and Firm Hedging Deci-
sions,” Journal of Risk and Insurance, 60(2):208-229.
[optional] Chapter 9 in Fabozzi et al. and chapter 12 in Varian.

!127
Homework
1. There’s no two ways about it. Mostafa, an investment advisor, is going to be out some money.
He has been cited by the provincial securities regulator for dubious bookkeeping (we don’t
have a federal securities regulator in Canada, how lame). Mostafa is required by law to
present himself at the regulator's office today. He can either plead guilty and pay a fine of
$55,000 on the spot, in which case it is a sure loss, or take a chance by pleading not guilty to
argue his case. If he argues his case, there is a 63.2 per cent chance that the judge will rule
against him, and administrative and legal fees will raise his fine to $82,500, also payable right
away. As big fan of Wintersleep and a risk averse expected utility maximizer with utility func-
tion

1 α
u" (W ) = W , α = 0.4, β = 44
β

what will he do? Use at least four-decimal place precision in your calculations. Drawing a
generic utility diagram (you don't need to draw the actual function) with W on the horizontal
axis and u(W) on the vertical will help a lot.

Without Real Investment

Current wealth $700,000

Fine if he pleads guilty $55,000

Fine if he pleads guilty and loses his case $82,500

Probability of losing his case 0.632

a. What is Mostafa's expected utility if he pleads guilty? If he pleads not guilty?

b. What will Mostafa do? Would all risk averse people do the same?

2. You are a property insurer and one of your clients, Renata, whose current wealth is $1.2 mil-
lion, wants to insure her $590,000 house (which would be a dump in Toronto in today's mar-
ket). The chances of the house burning down in any given year are 1.3 in a thousand, and
Renata's utility function is u" (W ) = W . She doesn't face any other risks.
a. Will Renata buy insurance for $805? What is the most that she is willing to pay?
b. Is the amount that Renata is willing to pay for insurance relatively elastic or inelastic with
respect to her current wealth? What about with respect to the probability of her house
burning down? Recall from your introductory economics class that price elasticity of de-
mand or supply is the percentage change in quantity for a given percentage change in
% ΔQ
price, " . All you have to do is rename the variables. Q becomes the most Renata is
% ΔP
willing to pay for insurance, and depending on which elasticity you are computing, P is
either her current wealth, W0, or the probability of her house burning down, 𝜋. Compute

!128
 both elasticities, the first for a one per cent increase in her wealth and the other for a one
per cent increase in the probability of her house burning down.

3. Your current wealth is $1,400,000 and you have logarithmic preferences, u" (W ) = ln(W ). At
the moment, you don't face any risks, but an old acquaintance, Zelda, asks you to be her part-
ner in an investment where there is a 10 per cent chance that you will lose $100,000 and a 90
per cent chance that you will earn $250,000.
Are you interested in Zelda's offer? If so, how much are you willing to invest?

!129
Answers
1. Oh, Mostafa. When will you learn?
a. Mostafa’s expected utility is 4.79039 if he pleads guilty, which is his utility of $645,000
(his current wealth $700,000 less the sure loss of $55,000). If he pleads not guilty, his ex-
pected utility is 4.79674, which is the expected value of his utility if he loses, 4.70763, and
if he wins, 4.94979.
b. Mostafa will plead not guilty and take his chances before the judge because his expected
utility under the risky loss, 4.79674, is higher than it is under the sure loss, 4.79039. You
come to the same conclusion by noting that his certainty equivalent under the risky loss,
$647,139, is greater than his wealth if he pleads guilty and pays the fine, $645,000.
Whether another risk-averse person would take the same decision depends on their tastes
because wealth under the sure loss is less than expected wealth under the risky loss. Can
you show this with a utility diagram? Can you also show with the same diagram that if
wealth under the sure loss was bigger than expected wealth under the risky loss, any risk
averse person would choose the sure loss? Turns out that people do not behave that way.
See Kahneman and Tversky (1979) for experiments that illustrate the failure of expected
utility theory in this regard.

2. Renata has smoke detectors installed on every floor of her home.

a. Yes she will buy insurance from you at that premium because her expected utility insured,
1095.08, is higher than it is uninsured, 1095.04. Renata is willing to pay up to $895.35 for
insurance, which is her wealth of $1.2 million less her certainty equivalent of
$1,199,104.65.
b. If Renata was one per cent wealthier, the most she would be willing to pay for insurance
would drop by 0.2 per cent (-0.198204). This is considered to be inelastic because the per-
centage change is less than one. That utility theory implies a negative relationship be-
tween wealth and willingness to pay for insurance seems counter intuitive. You could also
compute point elasticity, that is, using calculus, and the answer is almost the same,
-0.201063, but the derivative is ugly18 . Renata is more sensitive to the probability of loss
but not by much. If the probability of her house burning down was one per cent higher,
she be willing to pay about one per cent more for insurance (0.999812). A one-for-one
change such as this, in either direction, is referred to as unitary elasticity. The answer is al-
most identical when calculated as a point elasticity (0.999813).

3. Zelda is always able to sniff out the good opportunities. You are interested her proposal be-
cause your expected utility with the investment, 14.2924, is greater than your utility for your
current wealth, 14.152. The most you are willing to invest is $211,127. This is the difference
between your certainty equivalent under the investment, $1,611,127, and your current wealth
of $1,400,000.

18 Ask me if you want to see the math.

!130
 14
EQUILIBRIUM IN CAPITAL MARKETS

Time-State Preference Model

Let’s work out the details for the equilibrium model that appears in Chapter 10. I’ve repeat-
ed a lot of the exposition from Chapter 10 but in somewhat more technical language. Much of
the information about the economy is in the Edgeworth diagram; more of it is in the table on the

340 248.5 100 0

400 0

R
E u Ca , Cb = πa ( 1 -γ) ln Ca + πb ( 1 -γ) ln Cb

ick
y
Pa γ = 0.5, πa = 0.4, πb = ( 1 - πa )
= 0.533
Pb

e
250 150
C* : MRSL = MRSR
Cb 201.2 πa 1 -γ Cb Pa 198.8
= · · =
1 -πa 1 -γ Ca Pb
Lu
cy

0
0 100 160 251.5 500
Ca

next page. We want to figure out optimal future consumption, the equilibrium interest rate, and
the equilibrium expected return on a risky stock. The future is next period (t = 1), and because the
future is uncertain, optimal consumption is not a single amount; it is a distribution with Ca to be
consumed if state a occurs and Cb if state b occurs.

!131
 Contingencies Endowments Securities

Dates States Probs. Ps Lucy Ricky Canada M F

0 - 1 1 180 230 410 4.10 0.943

1 a 0.4 0.328 160 340 500 5.00 1.00

1 b 0.6 0.615 250 150 400 4.00 1.00

Lucy and Ricky are assumed to have the same tastes, the log utility function shown in the dia-
gram. Making them twins is a big simplification, but that won’t detract from the insights that the
results will give. Lucy and Ricky are also assumed to hold the same beliefs about probability of
each state occurring.
Recall that a pure state price, Ps, is a present value discount factor, but more specific than a
regular present value factor. Instead of telling you the value today of some amount to be received
in the future, it tells you the value today of an amount received in the future if a particular state of
nature prevails. A dollar received in state a next period is worth 32.8 cents today, and a dollar re-
ceived in state b is worth 61.5 cents today. All market values in a time-state preference model de-
pend on those two equilibrium prices (or more than two if there are more than two states); they’re
to financial prices as carbon is to life. Lucy and Ricky’s wealth, the price of a riskless bond, the
price of a risky stock, and the value of any financial contract is derived from the state prices.
That’s why state prices are called pure or primitive.
The price ratio and the individual state prices. In equilibrium, Lucy and Ricky’s
marginal rates of substitution are equal, and that common value is the implied ratio of pure state
prices.

C* : M R SL = M R SR
π 1 − γ C* π 1 − γ C*
∴ a ⋅ ⋅ bL = a ⋅ ⋅ bR
πb 1 − γ C* aL π b 1 − γ C*
aR

Here’s where the assumption about Lucy and Ricky having the same tastes and beliefs will simpli-
fy the derivation. The probabilities cancel on both sides, and so does the common taste parameter,
γ" , leaving

C*
bL C*
bR
" =
C*
aL C*
aR

which says that in equilibrium Lucy and Ricky choose state b consumption to state a consumption
to be in the same ratio. That means the consumption optimum, C*, must be on the main diagonal
(Market line) of the Edgeworth box

C*
bL C*
bR Eb 400 4
" = = = =
C*
aL C*
aR Ea 500 5

Since the equilibrium price ratio is equal to their common MRS, it must be that

!132
 P*
a π 1 − γ Eb 0.4 1 − 0.5 400
" = a ⋅ ⋅ = ⋅ ⋅ ≈ 0.533
P*
b πb 1 − γ Ea 0.6 1 − 0.5 500

The price ratio is less than one; a state b dollar is worth more than a state a dollar. That makes
sense because state b is less prosperous ($400 < $500)19 and more likely to occur (a probability of
0.6). Both conditions support Pb being greater than Pa in equilibrium.
The individual equilibrium state prices can be extracted from the price ratio of each state at
Ps
t = 1 to t = 0 as because P0 = 1 and π0 = 1 by definition.
P0

P*a π 1 − γ E0 0.4 1 − 0.5 410

P*
a = = a ⋅ ⋅ = ⋅ ⋅ = 0.328
P0 π 0 1 − γ Ea 1 1 − 0.5 500
"
P* π 1 − γ E0 0.6 1 − 0.5 410
P*b
= b = b ⋅ ⋅ = ⋅ ⋅ = 0.615
P0 π 0 1 − γ Eb 1 1 − 0.5 400

Notice that the state prices do not add up to one; there’s no reasons that they should. And don’t
confuse them with the state probabilities. The prices are not probabilities but they depend on the
probabilities.20
Consumption optimum. Solving for C* means solving for the two unknowns, C* " a and
C*
" b . This requires two pieces of information or conditions. The first piece of information, which
we’ve already come across, is that C* sits on the Market line

C*
b Eb 4
" = =
C*
a Ea 5

The second piece of information is the condition that market value of Lucy and Ricky’s con-
sumption, that is, their wealth, must be equal to the market value of their endowments

" 0 = C0 + PaCa + PbCb = e0 + Pa ea + Pb eb

W

but current consumption is taken as given, "C0 = e0, which more compactly,

P
" aCa + PbCb = Pa ea + Pb eb

19 State b is subject to relatively greater scarcity.

20
From this point on, we’ll drop the * from the state prices, and it will be understood that they are equilibri -
um prices.

!133
Substituting the sits-on-the-Market-line condition into the wealth condition gives a simple and
elegant expression for Lucy or Ricky’s consumption optimum.

Pa ea + Pb eb
C*
a = Ea = θEa
Pa Ea + Pb Eb
P e + Pb eb
C*
b
= a a E b = θEb
Pa Ea + Pb Eb

Optimal state a and b consumption are proportional to the aggregate endowments. The propor-
tion "θ is a person’s market value share of the aggregate endowment of the country. Lucy con-
sumes 50.3 per cent ("θL = 0.503) of Canada’s aggregate state a endowment and 50.3 per cent of
the aggregate state b endowment. Ricky consumes 49.7 per cent ("θR = 0.497). Plugging in Lucy’s
values gives

C*
a = 0.503 × 500 = $251.50
"
C*
b
= 0.503 × 400 = $201.20

Wealth. Wealth in a riskless, two-period world (chapter 11)

C1
W
" 0 = C0 +
1 + rF

becomes

E (C1)
W
" 0 = C0 +
1+k

in a risky, two-period world, where k" > rf is some risk-adjusted discount rate; and that present val-
ue is written as

" 0 = C0 + PaCa + PbCb

W

in the time-state preference model. Lucy’s wealth, based on her endowment, is $386.23.

W
" 0 = 180 + 0.328 × 160 + 0.615 × 250 = $386.23

Check that the answer is the same if Lucy’s C* is used to compute her wealth; it was, after all, the
wealth condition we used to find C*! The two equations for wealth in a risky world suggest that
the connection between pure state prices and a “normal” discount rate or required k is

E (C1)
P
" aCa + PbCb =
1+k

Work out k for Lucy and Ricky at e and C*.

Prices and returns on a risky stock and a safe bond. It’s easy to work out the prices
of the risky stock and the safe bond: multiply their state payoffs by the state prices and add them
up. If Xs is the payoff of asset x in state s, then the price of x is

!134
 Px = Pa Xa + Pb Xb

and its expected return is

E (X ) π X + πb Xb
E (rx ) = −1= a a −1
Px Px

Plugging in the values for Lucy and Ricky’s economy,

PM = Pa XaM + Pb XbM = 0.328 × $5 + 0.615 × $4 = $4.10

E (XM ) $4.40
∴ E (rM ) = −1= − 1 = 0.07317
PM $4.10
"
PF = Pa XF + Pb XF = XF (Pa + Pb ) = $1(0.328 + 0.615) = $0.943
X $1 $1
∴ rF = F − 1 = −1= − 1 = 0.0604454
PF PF $0.943

The expected return on the market, say, the TSX, is 7.3 per cent and the risk-free rate of interest
is 6 per cent. The risk premium on the market is only 1.3 per cent. Historically, market risk pre-
miums have averaged from three to six per cent. No matter how much you play with the variables
in the time-state preference model—risk aversion parameters, endowments, risk—it is difficult to
generate risk premiums that are close to those observed in real life. This anomaly is known as the
equity premium puzzle.
Portfolio composition. I guess you could also call this security holdings. How many
shares and bonds is Lucy endowed with? How many does she hold at the consumption optimum?
Let qx be the number of units of asset x she holds. Her consumption (read income) in either state
comes is the number of shares of M she owns times their payoff and the number of bonds she
owns times their payoff. Her security holdings, qM and qf, are the solution to this system of equa-
tions.

Ca = qM XaM + qf XF
"
Cb = qM XbM + qf XF

For Lucy’s endowment,

ea = $160 = $5qM + $1qF

"
eb = $250 = $4qM + $1qF

She is short 90 shares (qM = − 90) and is long 610 bonds (qf = 610).21 Ricky must be long 190
shares because there are 100 shares of M outstanding, and he must be short 610 bonds (he’s a
borrower) because debt is in zero net supply. Confirm that at C* Lucy holds 50.3 shares (where
have you seen that number before?) and no bonds.
Complete markets. The system of equations representing state consumption as a func-
tion of security holdings can be written in matrix form as

21 Being long in bonds means lending, and short means borrowing.

!135
 "C = X . q

where

(Cb) ( XbM XbF ) ( F)

Ca XaM XaF qM
C
" = ; X= ; and q = q

The solution for q

" = X−1 . C
q

exists if the payoff matrix X is non-singular (has an inverse). This has a neat economic interpreta-
tion. If X has an inverse, it means that the payoffs of the securities are linearly independent: the
payoff of any one security cannot be replicated with some combination of the payoffs of the oth-
ers (if there are more than two). That is really saying that all of the securities are unique—that
there are no perfect substitutes. When is true, Lucy and Ricky then have maximum choice and
any point in the Edgeworth box can reached through trade. It is called a complete market. The nec-
essary conditions for a complete market are the following:

• There must be at least as many securities as there are states of nature (M and F for a and b)

• The payoffs of the securities must be linearly independent (X has an inverse)

• Short positions must be allowed (negative q’s)

!136
 Consider the seven alternative consumption points shown in red in the next figure. Are
Lucy and Ricky able trade M and F to arrive at these points?
340 248.5 100 0
400 0

R
ick
y
250 150
h j
m
g
Cb 201.2 i k 198.8
l

g { 234, 203 }
h { 202, 223 }
i { 266, 186 }
j { 222, 222 }
k { 287, 187 }
l { 289, 173 }
m { 262, 210 }
Lu
cy

0
0 100 160 251.5 500

" Ca

The answer is yes. Could they trade to those points if F was replaced by risky security H
whose payoff is $15 in state a and $12 in state b. The answer is no.

!137
The Capital Asset Pricing Model
Straight to the model because I have to get out and buy some groceries before the store
closes. The CAPM is represented by the Security Market Line (SML), which says that the re-
quired

E (rj ) = rf + βj (E (rm ) − rf )

(expected) return on any risk security j is equal to the risk-free rate plus a risk premium that is
proportional to the risk premium on the market, E (rm ) − rf.22 In the stylized graph of the security
market line below, the risk-free rate is two per cent, the expected return on the market is five per
cent, making the slope of the SML or the risk premium on the market three per cent. The risk

E( r)

S1
0.065

M
0.05
Underpriced
0.0425
S2
0.035
Overpriced
0.0275
F
0.02

S3
0.008

β
- 0.4 0.5 1 1.5

premium is "βj (E (rm ) − rf ), where beta is a measure of a security’s systematic or market risk.
σjm σj
βj = = ρjm
σm2 σm

Beta is the only thing that matters in the CAPM. You can think of it as a security’s contri-
bution to the risk of a well-diversified portfolio. The portfolio is the market portfolio (you can’t get
more diversified than that since the market includes all risky assets), and the security’s contribution
to the risk of the market portfolio is its covariance with the market, σ" jm. Dividing the covariance
by the variance of the market, σ" m2 , normalizes beta to risk contribution per unit of market risk.
Stock S1 has a beta of 1.5. Its return is positively correlated with the market but fluctuates 50 per
cent more widely then the market, so it earns a premium that is 50 per cent bigger, 4.5 per cent,
than the market’s three per cent.

22 Unlike the time-state preference model, the risk-free rate is taken as given, not determined, in the CAPM.

!138
 E
" (rS1) = 0.02 + 1.5 × 0.03 = 0.02 + 0.045 = 0.065

Stock S2, with a beta of 0.5, is also positively correlated with the market but half as volatile and so
earns a premium of only 1.5 per cent above the risk-free rate. Stock S3 commands a return that is
less than the risk-free rate, which might not make sense at first glance. After all, don’t all risk
averse investors require that expected returns be greater than the risk-free rate? S3 earns less than
the risk-free rate because it is negatively correlated with the market—its beta is -0.4—and there-
fore contributes a great deal to the diversification of the market portfolio. You can think of its low
required return as commensurate with a high price to reflect its important role in diversification.
A well-known example of an asset with a negative beta is gold. It’s value tends to rise in times of
turmoil, financial, political, or otherwise, as investors seek what they perceive to be a safe store of
value.
The expression for beta above shows it can also written as
σj
ρ
" jm
σm

which may be more intuitive with the correlation coefficient than covariance. A security is more
σj
or less risky that the market as " is greater than or less than one, but within a well diversified
σm
portfolio, this ratio is tempered by a security’s correlation with the market.
Why is the covariance or correlation between the security and the market in the expression
for beta? As the number of securities in a portfolio is increased, the decline in risk that we know
comes about occurs because the portfolio’s risk becomes more and more determined by the co-
variances between securities than their individual variances (often referred to as total risk). There
n (n − 1)
are " unique covariances in an n-stock portfolio but only n variances. If n approaches the
2
number of securities in the market, the product of the covariance matrix and the weight of each
security in the market portfolio23 yields the vector of covariances of each security with the market.

σ11 σ12 σ13 … σ1n w1 σ1m

σ21 σ22 σ23 … σ2n w2 σ2m
" σ31 σ32 σ33 … σ3n ⋅ w3 = σ31
⋮ ⋮ ⋮ ⋱ ⋮ ⋮ ⋮
σn1 σn2 σn3 … σnn wn σnm

That’s the mathematical connection between a security’s returns and its beta coming from diversi-
fication over many assets. It does not explain why people would hold only the market or, what
amounts to the same thing, why only market risk is relevant in determining a security’s return.

23By weight of each security in the market I mean market value weight, which is the dollar value of a com-
pany’s equity divided by the dollar value of the entire market. The investment weights that make the M or
market portfolio on the Captial Market Line in chapter 11 are market value weights, although not in the
example used in that chapter.

!139
 Another way to think about beta is to assume that the random return on a security is linear-
ly related to just one random variable, the return on the market, with of course, some error. You
could call it a one-factor model. You might write

r" j = αj + bj rm + ϵj

The slope "bj connects the return on security j to the return on the market. The porttion of the
security’s return that cannot be accounted for by the market is the error term, ϵ" j, which you might
interpret as being unqiue to the company and whose variance is often called unique risk or unsystem-
atic risk.24 Also suppose that you wanted an estimate of bj that minimized unique risk because that
estimate would give you the best fit between the security’s return and the return on the market.
Solve for the error term in your linear model and write its variance.

ϵj = rj − (αj + bj rm )
= rj − αj − bj rm
" ∴ var(ϵj ) = var(rj − αj − bj rm )
= var(rj − bj rm )
∴ σϵ2j = σj2 + bj2 σm2 − 2bj σjm

Now minimize the variance with respect to b" j

d σϵ2j
= 2bj σm2 − 2σjm = 0
" d bj
σjm
∴ bj =
σm2

What you just did was solve for the slope in a simple linear regression of r" j on r" m.

Derivation of the CAPM. Thinking about beta as the coefficient in a one-factor model
still doesn’t answer why people would hold only the market or, what amounts to the same thing,
why only market risk is relevant in determining a security’s return. That comes from utility theory
and the assumption that people are risk averse. Our proof of the CAPM follows directly from our
section “Pricing XXX” in chapter 13, but now we need to impose an equilibrium or market-clear-
ing condition, and for that, we’ll will need to consider many investors and many securities.

i = 1,2,3,…, N investors
"
j = 1,2,3,…, K securities, excluding the risk-free security F

The uncertain end-of-period income, yi, for any investor i is

24 The error is independent of the return on the market.

!140
 ∑
yi = rf Mi + rj Sij
j
"
∑
= rf Wi + Sij (rj − rf )
j

where rf is the risk-free rate, rj is the risky return on the security j, and Wi is the investors current
wealth. The investor maximizes their expected utility with respect to their dollar investment Sj in
each security

∑
"
Max E [u (yi )] = Max E [u (rf Wi + Sij (rj − rf ))] ∀ i
j

where ∀
" means “for all.”
There are K first-order conditions for a maximum, one for each security

d E [u (yi )]
" = E [u′(yi )(rj − rf )] = 0 ∀ j
d Sj

Using the definition of covariance gives

E
" [u′(yi )]E [rj − rf ] + cov[u′(yi ), (rj − rf )] = 0 ∀ j

or

cov[u′(yi ), (rj − rf )]
E (rj − rf ) + = 0 ∀j
E [u′(yi )]

Here’s where the story ends unless we make a the critical assumption that allows us to separate an
investors’ tastes for risk (their utility functions) from the return on the security inside the covari-
ance operator

Assume that investors have quadratic utility or that security returns are normally distributed.

Without subjecting you to the math, either assumption allows us to write the above expression as

E [u′′(yi )]
" (rj − rf ) +
E cov(yi , rj ) = 0 ∀ j
E [u′(yi )]

You’ll recognize the fraction in front of the covariance operator as the coefficient of absolute risk
aversion. Take the reciprocal of ARA and call it risk tolerance θ" i (makes sense, eh?)

u′(yi ) 1
" θi =
Let =
u′′(yi ) A R A(yi )

For the specific case of quadratic utility, u" (y) = y − α y 2, α > 0, the risk premium can be
broken down into a risk aversion part and the riskiness of the stock. To get there, note the first
and second derivative of the utility function and its coefficient of absolute risk aversion

!141
 θ" i E (rj − rf ) − cov(yij , rj ) = 0 ∀ i and j

Now for the market-clearing condition. Aggregate (add up) over all of the investors

∑
θ" m E (rj − rf ) − cov(ym, rj ) = θi E (rj − rf ) − cov(yij , rj ) = 0 ∀ j
i

where θ" m is the risk tolerance of the market (a kind of average of the risk tolerance of all in-
vestors) and ym is the income of all investors or the market. But the income of the market must be
equal to the return on the market times its value

y" m = rm ⋅ Vm

so

If θm E (rj − rf ) − cov(ym, rj ) = 0 ∀ j
then
" θm E (rj − rf ) − cov(rm ⋅ Vm, rj ) = 0 ∀ j
∴ θm E (rj − rf ) − Vm cov(rm, rj ) = 0 ∀ j

Almost there. Look at the last line above. If it is true for any security j, then it must be true for a
security that is the portfolio of all securities, the market

If θm E (rj − rf ) − Vm cov(rm, rj ) = 0 ∀ j
"
then θm E (rm − rf ) − Vm cov(rm, rm ) = 0

Combine the last two equations

θm E (rj − rf ) = Vm cov(rm, rj )
"
and θm E (rm − rf ) = Vm cov(rm, rm )

Divide the first by the second

θm E (rj − rf ) Vm cov(rm, rj )
=
θm E (rm − rf ) Vm cov(rm, rm )
"
cov(rm, rj )
∴ E (rj − rf ) = E (rm − rf )
cov(rm, rm )

Solve for E(rj) and bingo, the CAPM

cov(rm, rj )
E (rj ) = rf + E (rm − rf )
var(rm )
= rf + βj E (rm − rf )

Time for groceries.

!142
!143
Homework (coming soon)

!144
Answers (coming soon)

!145
 15
DERIVATIVE SECURITIES (ALMOST)

A derivative security is one whose payoff or cashflow depends on the value of an underlying
security. Derivative security D might be one that pays 10 cents for every point that stock index U
is above 1000 on a particular date. The price of D any time before that date, and how it changes
from day to day, presumably reflects the market’s expectation of U’s value on that date. That is an
ideal interpretation because D’s value could just as easily reflect everyone second-guessing every-
one else’s forecast for U—as in Keyne’s Beauty Contest—and turn out to be a poor predictor of
U. But that is true for any financial security, not just derivatives. Either way, you can see that the
market for D is a prediction market, and because of that, the underlying security does not have to
be a financial security at all. U could be any event that people have an interest in betting on, such
as cumulative rainfall in a particular province by a particular date.25 So it’s not surprising that de -
rivative securities are also called contingent claims. An insurance contract is probably the most com-
mon contingent claim.
The most common derivatives traded in financial markets are futures and forward contracts
and option contracts. Those are the ones we’ll consider here. But given that derivatives are bets on
future events, there are countless variations: derivatives on derivatives, exchanges on cashflows
called swaps. The list goes on and on.

Forwards and Futures

A forward contract is an agreement made now for a transaction that will take place later.
It’s that simple. Suppose it’s February 2nd and you know you will need 1000 euros on May 31st.
You could buy the euros now at the current spot price and stuff them in a drawer until May 31st,
or maybe deposit them in a euro account. 26 You could wait until May 31st and buy them at what-
ever the spot price is on that day, but of course you’d be bearing the exchange rate risk, and you
may not like that. Or you could buy them forward. If you can find a willing seller, the two of you
enter into a contract where the seller agrees to delivery 1000 euros to you on May 31st and you
agree to pay the price that the two of you have negotiated. That price is called the forward price.
Let’s say it is CAD $1.48. You have locked in the May 31st price of the euro. It doesn’t matter
what the spot price is on that day. If the spot price is $1.52, you’ll be feeling good about yourself;
if the spot price is $1.43, you’ll be feeling some regret. Either way, by buying forward, you have
eliminated the price uncertainty. You would not have bought forward at $1.48 if you believed
strongly that the spot price on May 31st was going to be below $1.48, and the seller would not

25 There’s a lot of particulars here.

26 Scotiabank offers a euro account. Other Canadian banks probably do too.

!146
have sold forward at $1.48 if they believed that spot price was going to be above $1.48. That’s
why you agreed to $1.48.
A forward contract is a contract. There is offer and acceptance, and that creates obligation:
you are obligated to pay the price on the expiry date, and the seller is obligated to deliver the eu-
ros, however or wherever you’ve agreed. No money or goods changes hands at the time the con-
tract is struck, so the contract itself has no value. You may feel a little uneasy about the forward
price in this case because it is not a market price; the interaction of one buyer and one seller is
best described as a transaction, not market trade. If there are a number of forward buyers and

$1.57 S

E t ST > F*t

* E
$1.48 = F t = E t ST

E t ST < F*t

$1.33 D
q€ ( billions )
50

sellers, however, the market forward price is the one that clears the market. The supply and de-
mand graph shows that the equilibrium forward price F* " at any time t as being equal to the ex-
pectation now of what the spot price S will be in T periods, that is, at the expiration of the con-
tract.

F*
t = Et (ST )

The forward buyers of the 50 billion euros in open interest in the contracts are people who
believe the future spot price will be higher than $1.48, and as a result, expect a gain or positive
consumer’s (or buyer’s) surplus. The forward sellers believe the opposite and so expect a positive
producer’s (or seller’s) surplus. Their coming together results in the $1.48 equilibrium price. But
how can a gain be expected in a transaction that involves no real production and where no money
changes hands now? It should not as long as the future spot price ends up equalling the forward
price on average.27 It’s not unreasonable. If the forward price systematically overshot or undershot
the future spot price, speculators and arbitragers would jump at the opportunity to exploit the
pattern.
futures as standardized forwards

27That is, if the forward price is an unbiased predictor of the future spot price. See Froot and Thaler
(1990), cited in chapter 7, for evidence.

!147
 Changes in forward prices are generally highly correlated with changes in the current spot
prices of their underlying assets.

Options

!148
A PPENDICES (C OMING SOON )

!149
APPENDIX 1

1
TIME VALUE OF MONEY (COMING SOON)

Intro here

Generic timeline

Amounts shown dimmed are not received or paid. Example.

Notation

!150
APPENDIX 1

Future Value
If you invest $100 at an interest rate of three per cent compounded annually, you’ll have $112.55
in four years. The $100 will grow to $103 the first year and then to $106.09 by the end of the sec-
ond. That extra nine cents in the second year is interest on interest already earned, likewise for the

$100 $103 $106.09 $109.273 $112.551

t
0 1 2 3 4

27.3 cents in the third year and 55.1 cents in the fourth. Compounding interest means interest is
paid on interest already earned. Here’s the math.

F V = $100(1.03)(1.03)(1.03)(1.03)
" = $100(1.03)4
= $112.55

The future value of C

" 0 earning an interest rate r compounded annually for t years is

F V = C0 (1 + r)t

FV ($)
5%4% 3% 2% 1%
200

180

160

140

120

100

years
14 18 23 35 70

!151
APPENDIX 1

Compounding Interest More Than Once a Year

An investment grows a little faster if interest is compounded more than once a year because
[COMPLETE]. If interest is paid semi-annually, for example, you’ll end up with $112.65 after

$100 $103.023 $106.136 $109.344 $112.649

t
0 0.5 1 1.5 2 2.5 3 3.5 4
$101.50 $104.568 $107.728 $110.984

four years instead of $112.55. There are eight six-month periods in four years, and in each of
those six-month periods, the value of the investment grows by 1.5 per cent.

2⋅4

( 2 )
.03
F V = $100 1 +
"
F V = $100(1.015)8
= $112.649

Divide the interest rate and multiply the number of years by m, the number of times that
interest is compounded during the year.
m⋅t

( m)
r
F
" V = C0 1 +

!152
APPENDIX 1

It takes a little more than 23 years to double your money if the interest rate is three per
cent.

3%, compounded (m) m Value after 23 years

Annually 1 $197.3587

Semi-annually 2 $198.3526

Quarterly 4 $198.8589

Monthly 12 $199.2000

Weekly 52 $199.3319

Daily 365 $199.3659

Continuously "∞ $199.3716

When interest is compounded continuously, it is like water flowing from a tap. This is how
you compute future value in that case.

" V = C0 ⋅ e r⋅t
F

!153
APPENDIX 1

Present Value
Present value is future value in reverse. It is the answer to a question such as, How much would I
have to set aside today in order to have $112.65 in fours years if the interest rate is three per cent
compounded semi-annually? You already know that the answer is $100.

2⋅4

( 2 )
.03
C0 1 + = $112.649

" $112.649
C0 =
2⋅4
(1 + 2 )
.03

C0 = $100 = PV

The present value of amount Ct, paid or received at the end of period t, is

Ct
PV = m⋅t
(1 + m)
r

PV graph here.

!154
APPENDIX 1

Suppose you’ll receive $230 in a year, $167 in two years, and $560 after four years, but
you’ll have to pay $280 in the third year. The interest rate is three per cent compounded semi-an-

PV = $621.64 $230 $167 -$280 $560

t
0 1 2 3 4

$222.54

$160.20

-$267.42

$533.40

nually. If you needed cash today you should not accept less than $621.64 for that series of pay-
ments or cashflow stream.

$230 $167 $280 $560

PV = + − +
2⋅1 2⋅2 2⋅3 2⋅4
(1 + 2 ) (1 + 2 ) (1 + 2 ) (1 + )
.03 .03 .03 .03
" 2
= 222.543 + 161.586 − 270.922 − 541.844
= $621.643

T
Ct
∑
PV = m⋅t
(1 + m)
r
t=1

Stuff

!155
APPENDIX 1

PV = $621.64 $230 $167 -$280 $560

t
0 1 2 3 4

$222.54

$160.20

-$267.42

$533.40

!156
APPENDIX 1

Present Value of a Cash Flow Series

Include sensitivity table

!157
APPENDIX 1

Present Value of an Annuity

!158
APPENDIX 1

Present Value of a Growing Annuity

!159
APPENDIX 1

Examples

Home Mortgages

Saving for Retirement

!160
APPENDIX 2

2
RATES OF RETURN (YOU TELL ME)

!161
APPENDIX 3

3
MATRIX ALGEBRA (NOT YET)

E (r1)

" (rp ) = wT . r = ( w1 w2
E … wn) E (r2 ) , where wT . 1 = 1
⋮
E (rn )

!162
 4
STATISTICS (I DON’T KNOW)

!163
 FILLER TEXT

Text to pad figures and tables before the writing is done. Taken from Wiki.

Neil Percival Young, OC OM[3][4] (born November 12, 1945), is a Canadian singer-song-
writer, musician, producer, director and screenwriter. After embarking on a music career in the
1960s, he moved to Los Angeles, where he formed Buffalo Springfield with Stephen Stills, Richie
Furay and others. Young had released two solo albums by the time he joined Crosby, Stills & Nash
in 1969, in addition to three as a member of Buffalo Springfield. From his early solo albums and
those with his backing band Crazy Horse, Young has recorded a steady stream of studio and live
albums, sometimes warring with his recording company along the way.
Young's often distorted electric guitar work, deeply personal lyrics[5][6][7] and signature
tenor singing voice[8][9] transcend his long career. Young also plays piano and harmonica on
many albums, which frequently combine folk, rock, country and other musical styles. Known to
rip up live set lists, Young often plays acoustic versions of songs in one show and electric versions
in others. His gritty guitar work, especially with Crazy Horse, earned him the nickname "Godfa-
ther of Grunge"[10] and led to his 1995 album Mirror Ball with Pearl Jam. More recently Young
has been backed by Promise of the Real.
Young directed (or co-directed) films using the pseudonym Bernard Shakey, including Jour-
ney Through the Past (1973), Rust Never Sleeps (1979), Human Highway (1982), Greendale
(2003), and CSNY/Déjà Vu (2008). He also contributed to the soundtracks of the films Phil-
adelphia (1993) and Dead Man (1995).
Young has received several Grammy and Juno awards. The Rock and Roll Hall of Fame
inducted him twice: as a solo artist in 1995 and in 1997 as a member of Buffalo Springfield.[11]
In 2000, Rolling Stone named Young the 34th greatest rock 'n roll artist.
He has lived in California since the 1960s but retains Canadian citizenship.[12] He was
awarded the Order of Manitoba on July 14, 2006,[4] and was made an Officer of the Order of
Canada on December 30, 2009.[3]

!164
"

!165