Rice University
ECO 501
——————————————————
Lecture Notes: Microeconomic
Theory I
——————————————————
Christian Roessler
Fall 2008
Contents
1 Preference 3
1.1 Consumption Set . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Rational Preference . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Utility Functions . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Utility 8
2.1 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Quasiconcavity . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3 Demand I: Utility Maximization Problem 15
3.1 Budgets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 The Utility Maximization Problem (UMP) . . . . . . . . . . . 16
3.3 Indirect Utility Function . . . . . . . . . . . . . . . . . . . . . 21
4 Demand II: Expenditure Minimization Problem 23
4.1 EMP and Hicksian Demand . . . . . . . . . . . . . . . . . . . 23
4.2 Expenditure Function . . . . . . . . . . . . . . . . . . . . . . . 26
4.3 Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5 Comparative Statics 30
5.1 Wealth E¤ects . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.2 Price E¤ects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3 Law of Demand . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.4 Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.5 Money Metric . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.6 Welfare Comparisons . . . . . . . . . . . . . . . . . . . . . . . 36
6 ChoiceBased Approach 41
6.1 Choice Structures . . . . . . . . . . . . . . . . . . . . . . . . . 41
6.2 Weak Axiom . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.3 Relationship with the Law of Demand . . . . . . . . . . . . . 47
6.4 Strong Axiom . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
7 Integrability 55
7.1 Slutsky and Hicks Compensation . . . . . . . . . . . . . . . . 55
7.2 Aside: Dot Product . . . . . . . . . . . . . . . . . . . . . . . . 57
7.3 Substitution Matrix . . . . . . . . . . . . . . . . . . . . . . . . 58
1
7.4 Substitution Matrix with Preference . . . . . . . . . . . . . . . 62
7.5 Integrability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
8 Aggregation 71
8.1 Aggregate Demand Function . . . . . . . . . . . . . . . . . . . 71
8.2 Representative Consumer . . . . . . . . . . . . . . . . . . . . . 76
8.3 Failure of the Weak Axiom . . . . . . . . . . . . . . . . . . . . 80
9 Expected Utility 84
9.1 Lotteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
9.2 Preference over Lotteries . . . . . . . . . . . . . . . . . . . . . 86
9.3 Expected Utility Theorem . . . . . . . . . . . . . . . . . . . . 88
9.4 Paradoxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
9.5 StateSpace Approaches . . . . . . . . . . . . . . . . . . . . . 98
10 Risk 100
10.1 Money Lotteries . . . . . . . . . . . . . . . . . . . . . . . . . . 100
10.2 Risk Attitude . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
10.3 Stochastic Dominance . . . . . . . . . . . . . . . . . . . . . . 107
11 Pro…t Maximization Problem 111
11.1 Production Set . . . . . . . . . . . . . . . . . . . . . . . . . . 111
11.2 Transformation Function . . . . . . . . . . . . . . . . . . . . . 114
11.3 Pro…t Maximization . . . . . . . . . . . . . . . . . . . . . . . 116
11.4 Law of Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
12 E¢ciency of Aggregate Supply 120
12.1 E¢cient Production . . . . . . . . . . . . . . . . . . . . . . . . 120
12.2 Cost Minimization . . . . . . . . . . . . . . . . . . . . . . . . 122
12.3 Aggregate Supply . . . . . . . . . . . . . . . . . . . . . . . . . 125
13 Partial Competitive Equilibrium 126
13.1 Competitive Equilibrium . . . . . . . . . . . . . . . . . . . . . 126
13.2 Partial Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . 129
13.3 The Long Run . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
14 Welfare Analysis 132
14.1 Pareto E¢ciency and Surplus . . . . . . . . . . . . . . . . . . 132
14.2 E¢ciency of Competitive Equilibrium . . . . . . . . . . . . . . 134
2
14.3 E¢cient Allocations through the Market Mechanism . . . . . 135
15 Externalities 136
15.1 Externalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
15.2 Ine¢ciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
15.3 Remedies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
15.4 Public Goods . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
16 Monopoly and Product Di¤erentiation 145
16.1 Monopoly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
16.2 Bertrand Price Competition . . . . . . . . . . . . . . . . . . . 148
16.3 Repetition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
16.4 Product Di¤erentation . . . . . . . . . . . . . . . . . . . . . . 153
17 Capacity Constraints 156
17.1 CapacityConstrained Pricing . . . . . . . . . . . . . . . . . . 156
17.2 Cournot Quantity Competition . . . . . . . . . . . . . . . . . 158
17.3 Competitive Limit . . . . . . . . . . . . . . . . . . . . . . . . 161
18 Precommitment and Entry 162
18.1 Precommitment . . . . . . . . . . . . . . . . . . . . . . . . . . 162
18.2 Entry Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . 166
18.3 Socially Optimal Entry . . . . . . . . . . . . . . . . . . . . . . 168
1 Preference
1.1 Consumption Set
« In the …rst part of these lectures, we consider the consumer decision
problem in a market economy, where goods are o¤ered at posted prices.
Most of our attention will be on the primitive approach that starts with
"rational" individual preferences and derives optimal choices for given
prices and endowments. We will also look at connections with the
"revealed preference" approach that makes assumptions directly about
choices.
« Let there be 1 di¤erent commodities. A consumption bundle is a list
3
r
1
. r
2
. . . . . r
1
of quantities of each commodity, represented by a vector
r =
_
¸
_
r
1
.
.
.
r
1
_
¸
_
in the commodity space R
1
.
« Note that R
1
includes consumption bundles that list negative quantities
of some goods. These can be interpreted as debts or giving some of one’s
endowment to others.
« Any two items that could sell at di¤erent prices should be modeled as
distinct commodities. Strictly speaking, the description of a commodity
may have to include detailed context information, such as "a diet coke
can from a vending machine at Grand Central Station in the summer
of 2008."
« The de…nition of the commodity space tells us what kind of object a
consumption bundle is. It may be the case that not all such objects
(i.e. not all vectors) can be feasibly consumed, or we may wish to
impose restrictions in a particular model. Physical constraints include:
we need time to consume, and time is limited; we cannot consume in
two places simultaneously; we have to consume enough of certain things
(food, shelter) to survive.
« A restriction we will impose for modeling purposes is that the set of
feasible consumption bundles, or the consumption set, is convex. That
is, if r and ¸ are feasible, then any mixture . = cr + (1 ÷c) ¸ with
c ¸ (0. 1) is also feasible. This assumption rules out indivisible com
modities.
Example. Suppose the commodities are time (in minutes) spent watch
ing television, and time (in minutes) spent in a rollercoaster car on "The
Beast" (Kings Island near Cincinnati), a ride which takes approximately …ve
minutes. It is possible to watch no television and ride "The Beast" once (call
this bundle r), and it is also possible to watch television for one hour and
not ride "The Beast" (call this bundle ¸). But one cannot watch television
4
for half an hour and ride "The Beast" for twoandahalf minutes (unless it
is a oneperiod model ....). Since this bundle (call it .) is a mixture of r and
¸, the consumption set for these two commodities is not convex.
« Any economic model is an idealized representation of reality. We will
make many assumptions that are in some sense too strong to ever hold
exactly. In some cases, they preserve the spirit of the problem, and
propositions for the idealized world are also informative about reality.
In others, the assumptions limit the applicability of the model to a more
speci…c problem. Convexity of the consumption set is a convenience
restriction that should not alter the thrust of our results, even though
many commodities are actually indivisible.
« Most of the time, we will assume that the consumption set A contains
all bundles with only nonnegative quantities:
A = R
1
+
=
_
r ¸ R
1
s.t. r
¹
_ 0 for / = 1. . . . . 1
_
,
which is convex.
1.2 Rational Preference
« The …rst step in analyzing individual choices from the consumption
set is to de…ne preference over its elements. For the moment, we can
think of A more generally as a choice set, where the elements may in
particular be consumption bundles. Or they may be something else
one can express a preference about, for example sports teams.
« A preference % is a binary relation on A: a subset of A A. We
interpret (r. ¸) ¸%, which is commonly denoted r % ¸, as "r is at least
as good as ¸."
« The preference % associates with every r ¸ A its betterthan (or upper
contour) set and worsethan (or lower contour) set:
% (r) = ¦c ¸ A s.t. c % r¦
 (r) = ¦/ ¸ A s.t. r % /¦ .
The intersection of the upper and lower contour sets is the indi¤erence
set at r:
~ (r) = ¦/ ¸ A s.t. r ~ /¦ .
5
« From % derive the strict preference and indi¤erence relations ~ and ~:
r ~ ¸ i¤ r % ¸ and not ¸ % r
r ~ ¸ i¤ r % ¸ and also ¸ % r.
« In almost all economic theory, preferences are assumed to be rational
in the following sense.
De…nition. Preference % is rational i¤
(i) % is complete: \r. ¸ ¸ A, r % ¸ or ¸ % r (or both).
(ii) % is transitive: \r. ¸. . ¸ A, r % ¸ % . == r % ..
« Completeness says that any two elements of A can be compared: one
is always preferred, or else they are indi¤erent. But it is never the case
that the agent could not say whether he prefers r or ¸ (or …nds them
indi¤erent), for example because he is seeking more information. This
scenario could be accommodated by statedependent preference, where
the state is de…ned by what the agent learns.
« Transitivity is most closely related to the usual notion of rationality: it
requires the agent to rank alternatives consistently and predictably. In
practice, people often violate transitivity unwittingly, especially when
the choice objects are complex and unfamiliar. However, most would
revise a stated preference when one points out to them that it is in
transitive.
« Moreover, a transitivity violator is exposed to "Dutch book" trades.
Suppose the agent initially has r. Since the preference is cyclical, e.g.
r ~ ¸ % . % r, and the agent should be willing to pay some nonzero
amount to exchange ¸ for r, one could o¤er . for r, then ¸ for ., then
r for ¸ (at a price). Which leaves the agent with the initial bundle r,
but he has made a payment. If these trades are repeated often enough,
they will bankrupt the agent.
Exercise 1 (MWG1.B.1, 1.B.2). Show: if %is rational, then \r. ¸. . ¸ A
(i) r ~ r, (ii) relations ~ and ~ are transitive, (iii) r ~ ¸ % . == r ~ ..
6
1.3 Utility Functions
« To apply tools from calculus in analyzing choices, it is useful to repre
sent the preference relation by a function.
De…nition. The function n : A ÷ R is a utility function that represents
preference relation % if \r. ¸ ¸ A,
r % ¸ == n(r) _ n(¸) .
« If n represents %, then any strictly increasing function , : R ÷R can
be composed with n to give a new function · : A ÷R (where · (r) =
, (n(r)) for all r ¸ A), with the property that · (r) _ · (r
0
) ==
n(r) _ n(r
0
). Thus · also represents %. Clearly, utility functions are
nonunique.
« On the other hand, a utility function may fail to exist. It never exists
for a preference that is not rational.
Proposition. Only a rational preference relation can be represented by a
utility function.
Proof. Suppose a utility function n : A ÷ R represents preference %
on A. Then % is complete: for any r. ¸ ¸ A, we have n(r) _ n(¸) or
n(¸) _ n(r), so r % ¸ or ¸ % r. And % is transitive: when r % ¸ % ., we
have n(r) _ n(¸) _ n(.), hence n(r) _ n(.) and then r % .. So existence
of a utility function implies that % is rational. The contrapositive, "if % is
not rational, then there does not exist a utility function," follows.
Exercise 2 (MWG 1.B.5). Show: if A is …nite and % is a rational pref
erence relation on A, then there exists a utility function n : A ÷ R that
represents %.
7
2 Utility
2.1 Continuity
« In this lecture, we discuss which properties of preferences ensure the
existence of the type of utility function to which the standard optimiza
tion techniques apply. When is % represented by a utility function n
that is twice di¤erentiable and has a (unique) maximum? Conditions
for di¤erentiability are fairly di¢cult to establish, so we focus on the
related, but weaker, property of continuity.
« In addition to continuity, di¤erentiability requires the absence of kinks
in the utility function.
De…nition. Preference relation % is continuous if r % ¸ whenever r and
¸ are the limits of sequences ¦r
a
¦
1
a=1
and ¦¸
a
¦
1
a=1
such that r
a
% ¸
a
for all
:.
« Informally, a continuous preference ranks r and ¸ the same as it ranks
objects that are very similar to r and ¸.
« In particular, suppose ¦¸
a
¦
1
a=1
converges to ¸, and ¸
a
% r for all :.
Then continuity requires ¸ % r. Hence the limit of every convergent
sequence in % (r), the upper contour set, is also in % (r), so that % (r)
is closed. By the analogous argument,  (r), the lower contour set, is
also closed.
« The converse is true, too. Thus, a continuous preference is equiva
lently de…ned by the closedness of all lower and upper contour sets.
Informally, r is preferred to ¸ if objects that are very similar to r are
preferred to ¸.
« In the context of functions, continuity preserves closeness, rather than
preference, under limits: whenever r is very close to ¸ in the domain,
, (r) is very close to , (¸) in the image.
De…nition. The function n : A ÷ R is continuous if , (r) is the limit of
sequence ¦, (r
a
)¦
1
a=1
whenever r is the limit of sequence ¦r
a
¦
1
a=1
_ A.
8
« The proof of existence of a continuous utility function is simpli…ed by
(but actually valid without) imposing a property called monotonicity
on preferences. Its de…nition presupposes that the consumption set is
A = R
1
+
.
« The notation r ¸¸ stands for r
¹
¸
¹
for / = 1. . . . . 1.
De…nition. Preference relation % is monotone if r ¸¸ == r ~ ¸.
« Monotonicity is a fairly strong assumption. It rules out bads, but this
is not a real problem, since bads can be relabeled as goods (e.g. replace
"waste" with "waste disposal"). Many things are, however, desirable
only up to a point (you may like some ice cream, perhaps a lot, but
not tons of it delivered to your home). Monotonicity welcomes more of
anything; it is inconsistent with limited wants.
Proposition. Any continuous rational preference relation can be represented
by a continuous utility function.
Proof. Assume that preference % is continuous (and monotone) on R
1
+
.
We will construct utility values and show that the resulting function repre
sents % and is continuous.
Let c = (1. . . . . 1) ¸ R
1
+
be a vector of 1s. Monotonicity implies: for every
consumption bundle r, there exists c _ 0 such that r % cc and c < · such
that cc % r (just let c = 0 and c large enough so that all quantities in
cc are greater than the corresponding quantities in r). This means that the
sets ¹
+
(r) = ¦c ¸ R
+
s.t. cc % r¦ and ¹
(r) = ¦c ¸ R
+
s.t. r % cc¦ are
nonempty. Given r, we have c ¸ ¹
(r) or c ¸ ¹
+
(r) for any c ¸ R
+
by completeness, so R
+
_ ¹
(r) ' ¹
+
(r). Moreover, ¹
(r) and ¹
+
(r)
are closed, since  (r) and % (r) are closed, so that preference must be
preserved under limits of sequences ¦c
a
c¦
1
a=1
.(i.e. sequences of bundles that
have equal quantities of all commodities). To sum, ¹
(r) and ¹
+
(r) are
nonempty, closed, and together cover R
+
, which is a connected set. Hence
¹
(r) ¨ ¹
+
(r) =
_
c ¸ R
+
s.t. r ~ cc
_
is nonempty for all r ¸ R
1
+
.
9
Monotonicity implies further that there is only one c such that r ~ cc,
since c
0
c ~ r for all c
0
c and r ~ cc for all c
00
< c. We may therefore
de…ne a function n : A ÷R
+
by
n(r) = c ¸ ¹
(r) ¨ ¹
+
(r)
for all r ¸ A. It remains to be shown that n represents % and is continuous.
Suppose r % ¸ and r = cc and ¸ = ,c. Then c _ , by monotonicity,
so n(r) _ n(¸) . Conversely, if n(r) _ n(¸), then cc _ ,c, so r % ¸ by
monotonicity.
Continuity of n requires: for any sequence ¦r
a
¦
1
a=1
with limit r, the se
quence ¦n(r
a
)¦
1
a=1
converges to n(r). Note that, because ¦r
a
¦
1
a=1
converges,
there exists for any a number :() such that r
a
÷r < for : _ :().
Hence ¦n(r
a
)¦
1
a=a(.)
lies in a compact set: namely, in the interval [c. c]
where c = 0 and c is the highest quantity of any commodity in ¦r
a
¦
1
a=a(.)
.
Therefore ¦n(r
a
)¦
1
a=1
must have a convergent subsequence.
Furthermore, every convergent subsequence has limit n(r), which is there
fore the limit of the sequence ¦n(r
a
)¦
1
a=1
. To see this, suppose there is a
convergent subsequence
_
n
_
r
n(a)
__
1
a=1
(: is an increasing function) that
converges to c ,= n(r). If e.g. c n(r), then cc ~ n(r) c by monotonicity,
and also ^ cc ~ n(r) c, where ^ c =
1
2
(c +n(r)) lies between c and n(r).
Because n
_
r
n(a)
_
÷ c ^ c, we have, for some `, n
_
r
n(a)
_
^ c for all
: `. Then r
n(a)
~ n
_
r
n(a)
_
c ~ ^ cc, which implies r % ^ cc, because % is
continuous and % (r) therefore closed. But r ~ n(r) c, which con‡icts with
^ cc ~ n(r) c. Analogously, c < n(r) leads to a contradicton.
Exercise 3 (MWG 3.C.2). Prove the converse: if a continuous utility
function represents %, then % is continuous.
Example. Lexicographic preferences are not continuous and do not admit
a utility representation. With lexicographic preferences, there exists an or
dering of commodities, so that r
(1)
¸
(1)
== r % ¸ (where (1) refers to
the highestpriority commodity), r
(1)
= ¸
(1)
and r
(2)
¸
(2)
== r % ¸, etc.
(As in alphabetic entries in a telephone book or dictionary.)
These preferences violate continuity: e.g. every bundle in the sequence
¦r
a
¦
1
a=1
, where r
a
(1)
=
1
a
and r
a
(2)
= 0 for all :, is preferred to ¸ such that
¸
(1)
= 0 and ¸
(2)
= 1, but the limit r = (0. . . . . 0) is worse than ¸.
10
Suppose n is a utility function representing lexicographic preference %.
Let r
(1)
= ¸
(1)
= ` and r
(2)
= 1. ¸
(2)
= 2. Whatever values n(r) and
n(¸) n(r) the utility function assigns to r and ¸, we can …nd a rational
number : (`) ¸ (n(r) . n(¸)). Lexicographic preference implies ` `
0
==
: (`) : (`
0
) (since r with r
(1)
= ` is preferred to ¸
0
with ¸
0
(1)
= `
0
, i.e.
` n(r) n(¸
0
) `
0
). Hence the function : maps onetoone from R (an
uncountable set) to Q (a countable set), a contradiction.
One can understand the argument intuitively by noting that, with lexi
cographic preferences, the indi¤erence sets are singletons, so that a di¤erent
utility value has to be assigned for every bundle, i.e. every point of R
1
. But
the utility values come from the (in some sense smaller) set R.
Exercise 4 (MWG 3.C.4). Find a preference relation that is not contin
uous but has a utility function that represents it.
2.2 Quasiconcavity
« Besides continuity (di¤erentiability), which allows us to apply calculus
in solving the utility maximization problem, we would like to know
whether the utility function has a maximum. This property (quasicon
cavity) relates to the convexity of preference.
De…nition. Preference relation.% is convex if \r ¸ A the upper contour
set % (r) is convex: ¸. . ¸% (r) == \c ¸ [0. 1]
c¸ + (1 ÷c) . ¸% (r) .
« I.e. % is convex if ¸ % r and . % r imply \c ¸ [0. 1], c¸ +(1 ÷c) . %
r.
« Preference relation % is strictly convex if ¸ % r and . % r imply
\c ¸ (0. 1), c¸ + (1 ÷c) . ~ r, provided ¸ ,= ..
11
« Convex preference can be interpreted as a desire for variety: if I like
two bundles equally, then I …nd a mixture of the two (a more balanced
bundle) more appealing. By the same token, I dislike extremes. If I
have a lot of one commodity and little of the others, I am willing to
trade aggressively (give up a lot) for a more balanced bundle. This
results in a "diminishing marginal rate of substitution": a high relative
valuation for commodities of which I have little, which decreases as I
acquire more.
Exercise 5 (MWG 3.C.1). Show that lexicographic preference is rational,
monotone and strictly convex.
De…nition. Utility function n is quasiconcave if \r ¸ A the upper contour
set % (r) = ¦c ¸ A s.t. n(c) _ n(r)¦ is convex.
« Quasiconcavity is weaker than concavity, which is easily apparent from
an equivalent de…nition: n is quasiconcave if \r. ¸ ¸ A and \c ¸ [0. 1]
n(cr + (1 ÷c) ¸) _ min ¦n(r) . n(¸)¦ .
(Why are the de…nitions equivalent? Suppose r % ¸, then convexity of
% (¸) implies cr + (1 ÷c) ¸ ¸% (¸), hence n(cr + (1 ÷c) ¸) _ n(¸).
Conversely, suppose r. ¸ ¸% (.), and note that n(cr + (1 ÷c) ¸) _
min ¦n(r) . n(¸)¦ _ n(.) implies cr + (1 ÷c) ¸ ¸% (.).)
« Of course, a convex function satis…es: \r. ¸ ¸ A and \c ¸ (0. 1)
n(cr + (1 ÷c) ¸) _ cn(r) + (1 ÷c) n(¸) _ min ¦n(r) . n(¸)¦ .
« Utility function n is strictly quasiconcave if \r. ¸ ¸ A and \c ¸ (0. 1)
n(cr + (1 ÷c) ¸) min ¦n(r) . n(¸)¦ .
provided ¸ ,= ..
« The graph of a convex function always lies above a line segment between
two of its points. The graph of a quasiconcave function only has to lie
above the lower of the two points. Figure 1 illustrates the concave and
quasiconcave cases in panels (a) and (b).
12
Figure 1: Concave and quasiconcave functions
« Quasiconcavity is su¢cient to guarantee that a local extremum is a
global maximum. If there existed a greater maximum or a minimum,
then the utility function would have a trough somewhere and increase
on both sides of it, resulting in a "gap" (nonconvexity) in the upper
contour set. However, the global maximum may not be unique if it is
part of a plateau.
Proposition. If a utility function represents a (strict) convex preference
relation, then it is (strictly) quasiconcave.
Proof. Consider a weakly convex preference relation; I demonstrate that
it implies quasiconcavity (the strict case is analogous). To show:
n(cr + (1 ÷c) ¸) _ n(r)
or
n(cr + (1 ÷c) ¸) _ n(¸)
for all r. ¸ ¸ A, with c ¸ (0. 1). With convex preference, cr +(1 ÷c) ¸ % r
whenever ¸ % r, and cr + (1 ÷c) ¸ % ¸ whenever r % ¸. If n represents
13
%, it must satisfy n(cr + (1 ÷c) ¸) _ n(r) (i.e. the …rst inequality) if
n(¸) _ n(r), and n(cr + (1 ÷c) ¸) _ n(¸) (i.e. the second inequality) if
n(r) _ n(¸).
Exercise 6 (MWG 3.B.1). Show: if % is monotone, then % is locally
nonsatiated.
Exercise 7 (MWG 3.C.5).
(a).Preference relation % is homothetic if \r. ¸ ¸ A, ¸ ¸r == ¸ ~ r,
and r ~ ¸ == \c _ 0, cr ~ c¸. Utility function n is homogeneous
of degree 1 if \c 0 n(cr) = cn(r). Show: a continuous preference is
homothetic i¤ it admits a utility function that is homogeneous of degree 1.
(b) Let c
1
= (1. 0. . . . . 0) denote the bundle that contains 1 unit of com
modity 1 and none of the other commodities. Preference relation.% is qua
silinear with respect to good 1 if \r. ¸ ¸ A, \c 0, r + cc
1
~ r, and
r ~ ¸ == \c ¸ R, (r +cc
1
) ~ (¸ +cc
1
). Show: a continuous preference
is quasilinear with respect to commodity 1 if it admits a utility function of
the form n(r) = r
1
+c(r
2
. . . . . r
1
). (No need to show a converse.)
Exercise 8 (MWG 3.C.6). Consider preferences in a twocommodity
world represented by the CES (constant elasticity of substitution) utility
function
n(r) = (c
1
r
j
1
+c
2
r
j
2
)
1¸j
.
Show:
(a) When j = 1, the graphs of the indi¤erence.sets are linear.
(b) As j ÷ 0, the CES utility function represents in the limit the same
preferences as the CobbDouglas utility function n(r) = r
c
1
1
r
c
2
2
.
(c) As j ÷ ÷·, the CES utility function has in the limit the same
indi¤erence sets as the Leontief utility function n(r) = min ¦r
1
. r
2
¦.
14
3 Demand I: Utility Maximization Problem
3.1 Budgets
« A consumer is constrained in her choices by a limited budget. If she
has wealth n, and commodities sell at market prices
j =
_
¸
_
j
1
.
.
.
j
1
_
¸
_
¸ R
1
++
.
then her chosen bundle r ¸ R
1
+
must satisfy
j r = j
1
r
1
+ +j
1
r
1
_ n.
« The set of bundles that meet the budget contraint, for given prices and
wealth, is called the budget set.
De…nition. The Walrasian budget set is the set of all bundles r ¸ R
1
+
a¤ordable with wealth n at prices j:
1
j,&
=
_
r ¸ R
1
+
s.t. j r _ n
_
.
« Notice that a decline in any price enlarges the budget set: more bundles
satisfy the constraint.
« In a twocommodity world, the budget set is the area under the budget
line j
1
r
1
+j
2
r
2
= n.
« More generally, the budget set is bounded by an (1 ÷1)dimensional
hyperplane, de…ned by j r = n.
« The budget set is convex: if j r _ n and j r
0
_ n, then \c ¸ [0. 1]
we have j r
00
= j cr + j (1 ÷c) r
0
_ n. It is also compact: closed
and bounded, since r
¹
_ n,j
¹
for / = 1. . . . . 1 (you can at most spend
all your wealth on one commodity)
Exercise 9 (MWG 2.D.2). For an individual who consumes amount r of
a commodity priced at j, and / hours of leisure, when the hourly wage is 1,
what is the Walrasian budget set?
15
3.2 The Utility Maximization Problem (UMP)
« Let the consumer preferences be rational, continuous, and locally non
satiated  which implies, from the previous lecture, that there is a
continuous and quasiconcave utility function representing it. The con
sumption set is A = R
1
+
.
« The "utility maximization problem" is one of two equivalent ways to
frame the consumer choice problem (the other is the "expenditure min
imization problem," which we will get to later on). In the UMP, the
consumer picks the mostpreferred bundle in her Walrasian budget set,
which gives utility
max
a0
n(r) s.t. j r _ n.
« By the Weierstrass theorem, a continuous realvalued function on a
compact, nonempty set has a maximum. Therefore the UMP has a
solution as long as preferences are continuous.
« A solution to the UMP is in principle a set of consumption bundles
in the budget set, all of which give maximal utility. This set depends
on the parameters of the budget constraint: commodity prices and
individual wealth. The map from prices and wealth to the set of utility
maximizing consumption bundles in the associated budget set is called
the Walrasian demand correspondence.
« In general, the solution can be stated in terms of KuhnTucker condi
tions: for / = 1. . . . . 1,
Jn (r)
Jr
¹
= `
Jq (r)
Jr
¹
+
1
¹=1
j
¹
J/
¹
(r)
Jr
¹
.
where ` _ 0 and the j
¹
_ 0 are Lagrange multipliers (shadow prices),
q (r) = j r ÷n _ 0.
is the budget constraint, and the
/
¹
(r
¹
) = ÷r
¹
_ 0.
/ = 1. . . . . 1 are the nonnegativity constraints. If the relevant con
straint is nonbinding, ` = 0, respectively j
¹
= 0 (i.e. the interior
…rstorder condition holds).
16
« The conditions reduce to
Jn(r)
Jr
¹
= `j
¹
÷j.
or equivalently
Jn (r)
Jr
¹
_ `j
¹
with equality if r
¹
0. More concisely we can write this as _n(r) _ `j
(with equalities where r
¹
0) in terms of the gradient vector
_n(r) =
_
¸
_
0&(a)
0a
1
.
.
.
0&(a)
0a
L
_
¸
_
.
« In a twocommodity world, where the budget constraint binds and both
goods are consumed, these conditions imply
Jn (r) ,Jr
1
Jn (r) ,Jr
2
=
j
1
j
2
.
which is incidentally the tangency condition familiar from diagrams in
intermediate micro books. The left side is the marginal rate of substi
tution, i.e. the slope of the indi¤erence curve (solve dn = `l
1
dr
1
+
`l
2
dr
2
= 0 for dr
2
,dr
1
). The right side is the slope of the bud
get line (found by totally di¤erentiating the budget line constraint, i.e.
j
1
dr
1
+j
2
dr
2
= 0, and solving for dr
2
,dr
1
.). To be tangent, the slopes
have to be equal. See Figure 2.
« Tangency is necessary for an interior optimum, but there are in…nitely
many points satisfying it for di¤erent levels of wealth. Therefore, the
budget constraint is needed to …x the solution.
Example. Consider preferences represented by the CobbDouglas utility
function n(r
1
. r
2
) = r
c
1
r
1c
2
. Note that, if n represents preferences, then so
must
~ n(r
1
. r
2
) = ln n(r
1
. r
2
) = cln r
1
+ (1 ÷c) ln r
2
17
Figure 2: Optimal choice at the point of tangency
represent them, since the logarithm is an increasing transformation. Because
~ n(r
1
. r
2
) is strictly increasing in r
1
and r
2
, the budget constraint
j
1
r
1
+j
2
r
2
_ n
must bind.
Thus, the constrained choice is the solution to the Lagrangean problem
max
a
1
,a
2
1(r
1
. r
2
. `) = cln r
1
+ (1 ÷c) ln r
2
+`(n ÷j
1
r
1
÷j
2
r
2
) .
which satis…es …rstorder conditions
J1(r
1
. r
2
. `)
Jr
1
=
c
r
1
÷`j
1
= 0
J1(r
1
. r
2
. `)
Jr
2
=
1 ÷c
r
2
÷`j
2
= 0
J1(r
1
. r
2
. `)
J`
= n ÷j
1
r
1
÷j
2
r
2
= 0.
The …rst two reduce to
c
1 ÷c
r
2
r
1
=
j
1
j
2
.
18
Thus
cj
2
r
2
= (1 ÷c) j
1
r
1
.
From the budget constraint (the third …rstorder condition), we have
n ÷j
1
r
1
= j
2
r
2
.
hence
r
1
= c
n
j
1
and r
2
= (1 ÷c)
n
j
2
.
« If preference is monotonic, the budget will always be exhausted, since
all commodities are valuable. But a weaker property, local nonsatia
tion, actually su¢ces.
De…nition. Preference relation.% is locally nonsatiated if \r ¸ A and
\ 0, ¬¸ ¸ A such that ¸ ÷r _ and ¸ ~ r.
« Local nonsatiation (which also has the e¤ect of ruling out thick indif
ference sets) is a more plausible property than monotonicity, since it
allows for limited wants.
De…nition. The Walrasian demand correspondence r (j. n) satis…es Walras’
law if \j ¸0,\n 0 and \r ¸ r (j. n), j r = n.
« Walras’ law is satis…ed if the consumer’s choice exhausts the budget.
« The demand correspondence is homogeneous of degree / if r (cj. cn) =
c
I
r (j. n).
Proposition. If n is a continuous utility function that represents locally
nonsatiated preference relation % on A = R
1
+
, then \j ¸ 0, \n 0,
\r ¸ r (j. n) the Walrasian demand correspondence is homogeneous of degree
zero and satis…es Walras’ law. If n is in addition quasiconcave, then r (j. n)
is convex, and if n is strictly quasiconcave, then r (j. n) is a singleton.
19
Proof. Homogeneity of degree zero follows from the fact that the budget
constraint is unchanged when prices and wealth are scaled by c:
cj r _ cn == j r _ n.
Walras’ law re‡ects nonsatiation. If j r < n for r ¸ r (j. n), then by
nonsatiation there exists in every neighborhood of r an r
0
such that r
0
~ r.
If we pick a su¢cienty small neighborhood, then j r
0
< n, so that r
0
is in
the budget set. But then r is not a utilitymaximizing choice, a construction.
If n is quasiconcave, then its upper contour sets are convex. Since all
elements of r (j. n) are equally preferred, r (j. n) is also the upper contour
set of any of its members: if r ¸ r (j. n), then
r (j. n) =% (r) = ¦c ¸ A s.t. n(c) _ n(r)¦ .
Hence r (j. n) inherits the convexity of % (r). If n is strictly quasiconcave
and r (j. n) has two distinct elements r and r
0
, then
n(r
00
) = n(cr + (1 ÷c) r
0
) min (n(r) . n(r
0
)) .
But since n(r) = n(r
0
) = min (n(r) . n(r
0
)), this implies r
00
(which is in the
budget set, because it is convex) is strictly preferred to both r and r
0
, which
contradicts r. r
0
¸ r (j. n) .
Exercise 10 (MWG 3.D.1). Verify that the above proposition holds for
the Walrasian demand function with CobbDouglas utility.
« The following is a natural generalization of the continuity concept for
pointvalued functions to setvalued correspondences. Informally it says
that solutions in one constraint set should still be solutions at a very
similar constraint set.
De…nition. The Walrasian demand correspondence is upper hemicontinuous
if \(j. n), r ¸ r (j. n) whenever (j. n) and r are limits of sequences ¦(j
a
. n
a
)¦
1
a=1
and ¦r
a
¦
1
a=1
such that r
a
¸ r (j
a
. n
a
) for all :.
Proposition. If n is a continuous utility function that represents locally
nonsatiated preference relation % on A = R
1
+
, then \j ¸ 0, \n 0 the
Walrasian demand correspondence is upper hemicontinuous.
20
Proof. Suppose sequences ¦(j
a
. n
a
)¦
1
a=1
and ¦r
a
¦
1
a=1
converge to (j. n)
and r, and we have r
a
¸ r (j
a
. n
a
) for all :, but r , ¸ r (j. n). Then there
exists r
0
in 1
j,&
such that n(r
0
) n (r). By continuity of n, there exists
also ¸ ¸ 1
j,&
such that n(¸) n(r). Since (j
a
. n
a
) converges to (j. n), it
must be the case for all su¢ciently large : that ¸ ¸ 1
j
n
,&
n. This implies
n(r
a
) _ n(¸), since r
a
is in the choice set for 1
j
n
,&
n. But as ¦r
a
¦
1
a=1
converges to r, this argument leads to n(r) _ n(¸), a contradiction.
« Of course, the result implies that, if the Walrasian demand correspon
dence is in fact a function, then this function is continuous.
3.3 Indirect Utility Function
« The value of the utility function at a solution to the UMP is the highest
it can attain on the budget set, i.e. for a particular set of prices and
wealth. The map from prices and wealth to the highest attainable
utility value is called the independent utility function.
« The indirect utility function · is quasiconvex if \ · the lower contour
set ¦(j. n) s.t. · (j. n) _ ·¦ is convex.
Proposition. If n is a continuous utility function that represents locally
nonsatiated preference relation % on A = R
1
+
, then \j ¸ 0, \n 0, the
indirect utility function is homogeneous of degree zero, i.e. · (cj. cn) =
· (j. n), strictly increasing in n and nonincreasing in j
¹
for / = 1. . . . . 1,
quasiconvex, and continuous in j and n.
Proof. The indirect utility function is homogeneous of degree zero because
the Walrasian demand correspondence is: since the scaling of prices and
wealth does not a¤ect the set of utilitymaximizing choices, it cannot a¤ect
the utility derived from them.
Since wealth increases and price decreases enlarge the budget set, the
best available choice can only improve, so that utility associated with it
cannot decrease. Because of local nonsatiation, which implies that the utility
maximizing choice lies in the boundary of the budget set, the indirect utility
function must strictly increase in wealth.
21
To establish quasiconvexity, suppose · (j. n) _ · and · (j
0
. n
0
) _ · and
consider· (j
00
. n
00
), the maximal utility attainable in the budget set de…ned
by prices
j
00
= cj + (1 ÷c) j
0
and wealth
n
00
= cn + (1 ÷c) n
0
.
For any r in this budget set,
cj r + (1 ÷c) j
0
r _ cn + (1 ÷c) n
0
.
so j r _ n or j
0
r _ n
0
must be true. In the …rst case, r ¸ 1
j,&
, so
n(r) _ · (j. n). In the second case, r ¸ 1
j
0
,&
0 , so n(r) _ · (j
0
. n
0
). Thus
n(r) _ ·, i.e. r ¸ ¦(j. n) s.t. · (j. n) _ ·¦, the lower contour set at ·.
Because r was arbitrary, all bundles in 1
j
00
,&
00 belong to the lower contour
set at ·, and because · was arbitrary, · is quasiconvex.
In the more restrictive case that n is strictly quasiconcave, r (j. n) is
continuous, and n(r (j. n)) is a composition of continuous functions, so that
it is also continuous.
Exercise 11 (MWG3.D.2). Verify that the above proposition holds for the
inverse utility function with the log transformation of CobbDouglas utility.
Exercise 12 (MWG 3.D.6). In a threecommodity setting, let preferences
be represented by the utility function
n(r) = (r
1
÷/
1
)
c
(r
2
÷/
2
)
o
(r
3
÷/
3
)
¸
.
(a) Why is there no loss of generality from imposing c + , + ¸ = 1?
Assume this for the remaining parts.
(b) What are the …rstorder conditions for the UMP? Derive Walrasian
demand and the indirect utility function.
(c) Verify that the general properties of Walrasian demand and indirect
utility functions hold in this case.
22
4 Demand II: Expenditure Minimization Prob
lem
4.1 EMP and Hicksian Demand
« In the UMP ("utility maximization problem"), an optimal choice max
imizes utility on a …xed budget set. Analogously, we can de…ne optimal
choice as minimizing expenditure on a …xed upper contour set (i.e. a
set in which all bundles yield at least a certain utility). This approach
is called the EMP ("expenditure minimization problem").
« In the EMP, a consumer picks the consumption bundle for which ex
penditure is
min
a0
j r s.t. n(r) _ n
= max
a0
(÷j r) s.t. n(r) _ n.
« We assume j ¸0 and n n (0) throughout (so that at least one com
modity must be consumed, and none in in…nite quantity). Further
more, we assume that n represents a continuous, locally nonsatiated
preference on R
1
+
, and is di¤erentiable.
« The set of consumption bundles that are solutions to the EMP at prices
j and required utility n is denoted as /(j. n) _ R
1
+
and called the
Hicksian demand correspondence (or function, if singlevalued).
Exercise 13 (MWG 3.E.3). Argue that a solution to the EMP exists if
j ¸0 and n(r) _ n for some r ¸ R
1
+
.
« In parallel to the UMP, we apply the KuhnTucker conditions: at a
solution r
, we have for / = 1. . . . . 1,
÷
J (j r
)
Jr
¹
= `
J (n ÷n(r
))
Jr
¹
+
1
¹=1
j
¹
J (÷r
¹
)
Jr
¹
.
where ` _ 0 and the j
¹
_ 0 are Lagrange multipliers (shadow prices).
If the utility constraint is nonbinding at r, i.e. n(r) n, we have
` = 0. If the /th nonnegativity constraint is nonbinding at r, i.e.
r
¹
0, then j
¹
= 0.
23
« Resolving the derivatives,
÷j
¹
= ÷`
Jn(r
)
Jr
¹
÷j
¹
.
i.e. (after rearranging and relabeling the multiplier
~
` = 1,`)
Jn (r
)
Jr
¹
_
~
`j
¹
with equality if r
¹
0 (so that j
¹
= 0).
« This is exactly analogous to the …rstorder conditions in the UMP. In a
twocommodity world, where the utility constraint binds (so that
~
` <
·) and both goods are consumed, the solution r
is again characterized
by the tangency condition:
Jn (r
) ,Jr
1
Jn (r
) ,Jr
2
=
j
1
j
2
.
« The only di¤erence is that the remaining constraint that …xes r
is now
not the budget constraint, but the utility constraint n(r
) = n.
Example. Reconsider preferences represented by the CobbDouglas utility
function n(r
1
. r
2
) = r
c
1
r
1c
2
. The utility constraint has to bind, since expen
diture j r is strictly decreasing in r
1
and r
2
. (If it did not bind, you could
slightly lower consumption of one or both of the goods and attain a lower
expenditure within the utility constraint.) Moreover, n(r) = n(0) if r
1
or
r
2
is zero, so they must be strictly positive for r to attain n n (0).
The constrained choice therefore satis…es the "tangency" condition, which
specializes to
c
1 ÷c
r
2
r
1
=
j
1
j
2
.
given the CobbDouglas marginal utilities.
From the utility constraint, we have
r
c
1
r
1c
2
= n.
hence
r
2
=
_
r
2
r
1
_
c
n
24
and
r
2
=
_
1 ÷c
c
j
1
j
2
_
c
n and r
1
=
_
c
1 ÷c
j
2
j
1
_
1c
n.
« Notice that r
1
and r
2
are functions of prices and the required utility
(not of wealth) in the EMP. The wealth that is needed to attain n is
allowed to vary. In this sense, Hicksian demand is also called com
pensated demand, because if prices increase, expenditure is implicitly
adjusted as needed in order to keep utility constant. But the consump
tion bundle r may change so as to make the increase in expenditure as
small as possible.
« Hicksian demand has properties that correspond to those of Walrasian
demand. The Hicksian demand correspondence is said to have no excess
utility if \r ¸ /(j. n), n(r) = n.
Proposition. If n represents continuous, locally nonsatiated preferences
on R
1
+
, and j ¸ 0, then the Hicksian demand correspondence /(j. n) is
homogeneous of degree zero in j, has no excess utility, is convex if preference
is convex, and singlevalued if preference is strictly convex.
Proof. Homogeneity of degree zero follows from the fact that the upper
contour set at n (the constraint set), and therefore any solution to EMP, is
una¤ected by scaling prices:
r ¸ /(cj. n) == r ¸ /(j. n) .
(It is only the expenditure that changes, not the bundle that minimizes it 
the scaling leaves relative prices unaltered.)
No excess utility re‡ects continuity of the utility function, which is in
herited from preferences. If there were a solution r ¸ /(cj. n) such that
n(r) n, then we could construct a scaleddown bundle r
0
= cr with
c ¸ (0. 1) that satis…es j r
0
< j r (since r
0
¸r) and n(r
0
) _ n (for c close
enough to 1, by continuity). But this means r
0
, and not r, can be a solution
to EMP at n, a contradiction.
Let r. r
0
¸ /(j. n), so that j r = j r
0
. If preference is convex (utility
quasiconcave), then upper contour sets are convex, so
r
00
= cr + (1 ÷c) r
0
25
attains utility n. Moreover,
j r
00
= j cr +j (1 ÷c) r
0
= cj r + (1 ÷c) j r
0
= j r.
It follows that r
00
also minimizes expenditure and belongs to /(j. n). If
preference is strictly convex (utility strictly quasiconcave), then r
00
~ r (and
r
00
~ r
0
since /(j. n) satis…es no excess utility, so that r
0
~ r). Continuity
implies there exists , ¸ (0. 1) such that ,r
00
~ r, i.e. n(,r
00
) n(r). But
since ,r
00
¸ r
00
and j r
00
= j r, we have j ,r
00
< j r, which implies
that r does not minimize expenditure on the constraint set, i.e. r , ¸ /(j. n),
a contradiction. Hence two such elements r. r
0
of /(j. n) cannot exist, and
/(j. n) must be a singleton.
4.2 Expenditure Function
« The value of j r
at a solution r
to the EMP is denoted as c (j. n)
and called the (minimum) expenditure.
« The expenditure function is homogeneous of degree one in j if \c 0,
c (cj. n) = cc (j. n).
Proposition. If n represents a continuous, locally nonsatiated preference
on R
1
+
, and j ¸ 0, then the expenditure function c (j. n) is homogeneous
of degree one in j, strictly increasing in n and nondecreasing in j
¹
for
/ = 1. . . . . 1, concave in j, and continuous in j and n.
Proof. Since scaling up the price does not change the constraint set
(i.e. the upper contour set at n), it does not a¤ect the solution r
. Hence
c (cj. n) = cj r
= cc (j. n), so that the expenditure function is homoge
neous of degree one in j.
If c (j. n) were not strictly increasing in n, then there would exist solutions
r
0
and r
00
, respectively at n
0
and n
00
n
0
, such that j r
0
_ j r
00
0. (We
maintain n n(0), so r
0
. r
00
,= 0.) Continuity implies there is a bundle
r = cr
00
for some c ¸ (0. 1) that satis…es n(r) n
0
and j r < j r
00
_ j r
0
(since r ¸r
00
). But then r
0
cannot be a solution to EMP.
Suppose c (j. n) were strictly decreasing in j
¹
for some /. I.e. if we
compare expenditure at two price vectors j
0
and j
00
di¤ering only in that
j
00
¹
_ j
0
¹
, then c (j
00
. n) < c (j
0
. n). Since the constraint set is not a¤ected
26
by the price di¤erence, the same r is a solution with both j and j
0
, so
c (j
00
. n) = j
00
r _ j
0
r = c (j
0
. n), a contradiction.
If r
00
solves EMP at n with prices j
00
= cj +(1 ÷c) j
0
for c ¸ [0. 1], then
c (j
00
. n) = j
00
r
00
= cj r
00
+ (1 ÷c) j
0
r
00
_ cc (j. n) + (1 ÷c) c (j
0
. n) .
since r
00
is available at prices j and j
0
(so expenditure with prices j and j
0
at the minimizing bundles r and r
0
cannot be larger then at r
00
). Hence the
expenditure function is concave in j.
We do not prove continuity.
« Concavity follows from the fact that, if j is increased to j
0
and r kept
…xed, expenditure (to maintain utility level n) increases linearly to
c (j
0
. n). The consumer can always attain n at an expenditure no greater
than c (j
0
. n), but may be able adjust r to r
0
to attain n at reduced
expenditure. Since this is true for all j, c (j
0
. n) increases less than
linearly in j, thus is concave.
Exercise 14 (MWG 3.E.2). Con…rm that the general properties of the
Hicksian demand function and the expenditure function hold with Cobb
Douglas preferences.
Exercise 15 (MWG 3.E.6). For the CES (constant elasticity of substitu
tion) utility function
n(r) = (r
j
1
+r
j
2
)
1¸j
.
derive the Hicksian demand function and the expenditure function, and verify
their general properties in this case.
Exercise 16 (MWG 3.E.7). If preferences are quasilinear with respect to
the …rst good, show that the Hicksian demand functions for the remaining
goods are invariant to n, and …nd the form of the expenditure function.
27
4.3 Duality
« The connection between Walrasian and Hicksian demand is that UMP
and EMP have the same solutions when wealth n in the UMP is …xed
at the level of minimized expenditure j r
= c (j. n) in the EMP,
and (equivalently) when utility n in the EMP is …xed at the level of
maximized (or indirect) utility n(r
) = · (j. n) in the UMP. I.e.
/(j. n) = r (j. c (j. n)) and r (j. n) = /(j. · (j. n)) .
Example. Compare the Walrasian and Hicksian demands for CobbDouglas
preferences:
r
&
1
= c
n
j
1
and r
&
2
= (1 ÷c)
n
j
2
and
r
I
1
=
_
c
1 ÷c
j
2
j
1
_
1c
n and r
I
2
=
_
1 ÷c
c
j
1
j
2
_
c
n.
Equating either the r
1
s or r
2
s gives
n =
_
j
1
c
_
c
_
j
2
1 ÷c
_
1c
n.
Since
n(r
&
) =
_
c
n
j
1
_
c
_
(1 ÷c)
n
j
2
_
1c
=
_
c
j
1
_
c
_
1 ÷c
j
2
_
1c
n = · (j. n)
in the UMP, we see that r
&
= r
I
if n = n(r
&
). On the other hand, since
j r
I
= j
1
_
c
1 ÷c
j
2
j
1
_
1c
n +j
2
_
1 ÷c
c
j
1
j
2
_
c
n
=
_
_
c
1 ÷c
_
1c
+
_
1 ÷c
c
_
c
_
j
c
1
j
1c
2
n
=
_
j
1
c
_
c
_
j
2
1 ÷c
_
1c
n = c (j. n) .
in the EMP, r
&
= r
I
if n = j r
I
.
28
« The UMP and EMP are generally equivalent in the following sense.
Proposition. If n represents continuous, locally nonsatiated preferences
on R
1
+
, and j ¸0, then:
(i) if r
solves the UMP at wealth n 0, then r
solves the EMP at
n = n(r
), and c (j. n) = n;
(ii) if r
solves the EMP at utility n n (0), then r
solves the UMP at
n = c (j. n), and n(r
) = n.
Proof. (i) Suppose r
is a solution to UMP at n, but not to EMP at
n = n(r
). Let instead r
0
be a solution to EMP at n = n(r
), so that
j r
0
< j r
and n(r
0
) _ n. Local nonsatiation implies that there exists r
00
such that n(r
00
) n (r
0
) su¢ciently close to r
0
for j r
00
< j r
to still hold.
But then r
00
¸ 1
j,&
and n(r
00
) n(r
), so that r
was not a solution to
UMP. By contradiction, r
solves EMP, and therefore c (j. n) = j r
= n.
(ii) Conversely, suppose r
is a solution in EMP at n n(0), but not
in UMP at n = j r
. Let instead r
0
be a solution to UMP at n, so that
n(r
0
) n(r
) and j r
0
_ j r
. By continuity, we can scale r
0
down to
r
00
= cr
0
, with c ¸ (0. 1) su¢ciently close to 1, such that n(r
00
) n (r
) still
holds. (Note that n n(0) implies r
,= 0 and r
0
,= 0, so j r
0.) Since
r
00
¸ r
, we have j r
00
< j r
, and r
cannot be a solution to EMP. By
contradiction, r
solves UMP, and therefore n(r
) = n.
Exercise 17 (MWG 3.E.9). Using the equivalence of the UMP and EMP,
show that the general properties of the indirect utility function (homogeneous
of degree zero, strictly increasing in n, nonincreasing in prices, quasiconvex,
continuous in j and n) imply the general properties of the expenditure func
tion (homogeneous of degree one in j, strictly increasing in n, nondecreasing
in prices, concave in j, continuous in j and n), and vice versa.
Exercise 18 (MWG 3.E.10). Using the equivalence of the UMP and
EMP and the properties of indirect utility and expenditure functions, show
that properties of the Walrasian demand function for continuous, locally
nonsatiated preferences (homogeneous of degree zero, Walras’ law, convex /
singlevalued if preferences are convex / strictly convex) imply properties of
the Hicksian demand function (homogeneous of degree zero in j, no excess
utility, convex / singlevalued if preferences are convex / strictly convex),
29
and vice versa. (Note there is a typo in the book  you are to show that
Proposition 3.D2 implies Proposition 3.E3, not 3.E4).
« The relationship between UMP and EMP is just a special case of a far
more general theory of duality. In this connection, duality means that
a constrained maximization problem can be expressed as a constrained
minimization problem, swapping objective and constraint.
5 Comparative Statics
5.1 Wealth E¤ects
« In this lecture, we examine the behavior of the demand function as
prices and wealth change. We assume for now that demand is single
valued (i.e. preferences are strictly convex).
« The wealth e¤ect for the /th good is Jr
¹
(j. n) ,Jn.
« If the wealth e¤ect is positive, i.e. Jr
¹
(j. n) ,Jn _ 0, we say that the
good is normal. If the wealth e¤ect is negative, i.e. Jr (j. n) ,Jn < 0,
we say that the good is inferior.
« Goods are inferior when there are higherquality, costlier substitutes
for the agent to switch to as wealth increases. (E.g. supermarket bread
vs. fresh bread from a bakery.)
« The total wealth e¤ect is given by the derivative vector with respect to
n:
1
&
r (j. n) =
_
¸
_
0a
1
(j,&)
0&
.
.
.
0a
L
(j,&)
0&
_
¸
_
¸ R
1
.
5.2 Price E¤ects
« The price e¤ect for the /th good is Jr
¹
(j. n) ,Jj
¹
.
« Typically, we expect the price e¤ect to be negative. A good for which
the price e¤ect is positive, so that the agent consumes more of it after
its price increases, is called a Gi¤en good.
30
« A Gi¤en good is similar to an inferior good  actually, it is just a very
inferior good. As price increases, the agent’s budget set shrinks, and if
the wealth e¤ect (having more of an inferior good when you are poorer)
is su¢ciently powerful, we have a Gi¤en good.
« The total price e¤ect is given by the derivative matrix with respect to
j:
1
j
r (j. n) =
_
¸
_
0a
1
(j,&)
0j
1
0a
1
(j,&)
0j
L
.
.
.
.
.
.
.
.
.
0a
L
(j,&)
0j
1
0a
L
(j,&)
0j
L
_
¸
_
.
« We consider now some relationships between price and wealth e¤ects.
Recall that Walrasian demand is homogeneous of degree zero. An im
pliciation is that changes to the consumption bundle in response to
simultaneous (and proportionate) increases in prices and wealth can
cel. This merely re‡ects the invariance of the budget constraint.
Proposition. If Walrasian demand r (j. n) is homogeneous of degree zero,
then \j and \n,
1
j
r (j. n) j +1
&
r (j. n) n = 0.
Proof. By zerohomogeneity, \c 0,
r (cj. cn) ÷r (j. n) = 0.
i.e.
_
¸
_
r
1
(cj. cn)
.
.
.
r
1
(cj. cn)
_
¸
_
÷
_
¸
_
r
1
(j. n)
.
.
.
r
1
(j. n)
_
¸
_
=
_
_
0
0
0
_
_
This is true, in particular, for c = 1. Totally di¤erentiating with respect to
c, we have for / = 1. . . . . 1,
1
I=1
Jr
¹
(cj. cn)
Jj
I
j
I
+
Jr
¹
(cj. cn)
Jn
n = 0.
which corresponds to the claim in matrix notation when c = 1.
31
« Walras’ law implies that an increase in prices (while wealth remains
…xed) must be accompanied by an o¤setting decrease in consumption,
and that an increase in wealth (while prices remain …xed) brings with
it a corresponding increase in consumption.
« To understand how the following results relate to this intuition, no
tice that a price increase for all commodities would increase the re
quired wealth in proportion r (j. n), and therefore the decrease in re
quired wealth from reduced consumption, j1
j
r (j. n), has to be of the
same magnitude. Similarly, an increase in available wealth has to be
matched by the increase in required wealth from greater consumption,
j 1
&
r (j. n).
Proposition. If Walrasian demand r (j. n) satis…es Walras’ law, then \j
and \n,
r (j. n)
T
+j
T
1
j
r (j. n) = 0
T
and
j
T
1
&
r (j. n) = 1.
Proof. Di¤erentiating Walras’ law,
j r (j. n) = n.
…rst with respect to prices, we have for / = 1. . . . . 1
r
¹
(j. n) +
1
I=1
j
I
Jr
I
(j. n)
Jj
¹
= 0.
which corresponds to the …rst claim in matrix notation.
Di¤erentiating Walras’ law with respect to wealth, we have for / =
1. . . . . 1
1
¹=1
j
¹
Jr (j. n)
Jn
= 1.
which corresponds to the second claim in matrix notation.
32
Exercise 19 (MWG 2.E.3). If r (j. n) is homogeneous of degree zero, i.e.
\c 0, r (cj. cn) = r (j. n), and satis…es Walras’ law, show that
j 1
j
r (j. n) j = ÷n.
Interpret.
Exercise 20 (MWG 2.E.4). If r (j. n) is homogeneous of degree one with
respect to n, i.e. \c 0, r (j. cn) = cr (j. n), and satis…es Walras’ law,
show that
¹&
(j. n) = 1 for / = 1. . . . . 1. Interpret.
Exercise 21 (MWG 2.E.6). In the case of the demand function
r
1
(j. n) =
j
2
j
1
+j
2
+j
3
n
j
1
r
2
(j. n) =
j
3
j
1
+j
2
+j
3
n
j
2
r
3
(j. n) =
j
1
j
1
+j
2
+j
3
n
j
3
.
verify the three derivative conditions for a demand function r (j. n) that is
homogeneous of degree zero and satis…es Walras’ law.
5.3 Law of Demand
« The "law of demand" refers to the intuitive property that a price in
crease should reduce demand for a commodity. This statement is, how
ever, not generally valid for Walrasian demand.
« Exceptions are the Gi¤en goods. They arise because a price increase
e¤ectively makes the agent poorer and may lead to an increase in the
consumption of relatively cheap commodities.
« In the Hicksian de…nition of demand, a price increase is accompanied
by an increase in expenditure, so it does not "impoverish," and the
Gi¤en e¤ect is absent. Hence the two de…nitions of demand di¤er in
comparative statics.
« Hicksian demand is said to obey the "compensated law of demand,"
i.e. (singlevalued) demand for any commodity diminishes if its price
increases.
33
Proposition. If n represents continuous, locally nonsatiated preferences
on R
1
+
, and the Hicksian demand correspondence /(j. n) is singlevalued for
all j ¸0, then \j
0
. j
00
,
(j
00
÷j
0
) (/(j
00
. n) ÷/(j
0
. n)) _ 0.
Proof. Simply observe that
j
00
/(j
00
. n) _ j
00
/(j
0
. n)
j
0
/(j
0
. n) _ j
0
/(j
00
. n) .
since /(j
00
. n) minimizes expenditure when prices are j
00
, and /(j
0
. n) mini
mizes expenditure when prices are j
0
. Subtracting j
0
/(j
0
. n) on the right
and j
0
/(j
00
. n) on the left of the …rst inequality preserves signs and gives
the result.
5.4 Elasticity
« The partial derivatives are not unitfree measures. E.g. Jr
¹
(j. n) ,Jj
¹
would depend on the currency in which we quote prices: clearly, a $1
increase in the price of petrol would cause a greater response in demand
than a U1 increase (which translates into about one cent). Moreover,
Jr
¹
(j. n) ,Jj
¹
depends in this case on whether petrol is sold by gallon
or by liter (about a quarter gallon).
« An alternative way to describe price and wealth e¤ects is in terms of
elasticities, which are unitfree.
De…nition. The price elastisticity of demand r
¹
(j. n) with respect to the
/th good is:
¹I
(j. n) =
Jr
¹
(j. n)
Jj
I
j
I
r
¹
(j. n)
.
If / = /,
¹I
(j. n) is called "ownprice elasticity." If / ,= /,
¹I
(j. n) is
called "crossprice elasticity." The wealth elastisticity of demand r
¹
(j. n) is:
¹&
(j. n) =
Jr
¹
(j. n)
Jn
n
r
¹
(j. n)
.
34
« Elasticities can be interpreted as the approximate relative percent change
in two variables (this relationship holds exactly only in the limit, when
the change is very small): e.g. the percent change in r
¹
(j. n) in re
sponse to a small percent change in j
¹
.
Exercise 22 (MWG 2.E.8). Demonstrate that the price elasticity of de
mand r
¹
(j. n) with respect to the /th good can be expressed as
¹I
(j. n) =
J ln (r
¹
(j. n))
J ln j
I
.
and derive a corresponding expression for
¹&
(j. n).
« In terms of elasticities, we can restate the relationship between the
responses in demand for commodity / to changes in prices and wealth
(as implied by homogeneity) as:
1
I=1
¹I
(j. n) +
¹&
(j. n) = 0.
5.5 Money Metric
« In the remainder, we consider the welfare implications of price changes.
First, we derive a quantitative measure of welfare change in terms of the
expenditure function. Since data availability is normally a signi…cant
constraint, we are also interested in conditions under which a price is
welfareimproving that are based on minimal information (such as the
initial price and demand and the new price).
« Let the consumer’s preference relation be rational, continuous, locally
nonsatiated and the expenditure and indirect utility functions di¤eren
tiable. Suppose wealth is …xed at n 0, and the price changes from
j
0
to j
1
.
« If consumer’s preferences are known, then she is better o¤ after the
price change if and only if · (j
1
. n) ÷· (j
0
. n) _ 0.
35
« Due to the duality of the UMP and EMP, we can express the indirect
utility at j
1
and j
0
in terms of the expenditure required to attain it.
Recall that the expenditure function c (j. n) is strictly increasing in n.
Hence, at given prices, the expenditure function is itself an indirect
utility function when n is evaluated at n = · (j. n).
« I.e. we have, for any …xed price j ¸0,
c ( j. · (j. n)) _ c ( j. · (j
0
. n
0
)) == · (j. n) _ · (j
0
. n
0
) .
and c ( j. · (j. n)) ÷c ( j. · (j
0
. n
0
)) is a meaningful quantitative measure
of welfare change. It is the change in wealth needed to buy the optimal
bundle when the budget set changes from 1
j,&
to 1
j
0
,&
0 .
« The expenditure function evaluated at n = · (j. n) is called the money
metric (indirect utility function). It is independent of the utility repre
sentation we choose for the agent’s preference, as all utility representa
tions select the same consumption bundle at given prices and wealth.
Hence it is unique up to the choice of j.
« Note that because indirect utility is decreasing in prices, so is the money
metric. (Expenditure increases in its price arguments, but those prices
are …xed in the money metric.)
« Clearly, the agent is better o¤ at prices j
1
than at price j
0
if and only
if c ( j. · (j
1
. n)) ÷c ( j. · (j
0
. n)) _ 0 (for any indirect utility function),
i.e. if a bundle that yields the maximal utility attainable at prices j
1
is
more expensive than a bundle that yields the maximal utility attainable
at prices j
0
.
5.6 Welfare Comparisons
« Let n
0
= · (j
0
. n) and n
1
= · (j
1
. n). Then we can de…ne the change
in welfare by
c
_
j
0
. n
1
_
÷c
_
j
0
. n
0
_
= c
_
j
0
. n
1
_
÷c
_
j
1
. n
1
_
.
since c (j
0
. n
0
) = c (j
0
. · (j
0
. n)) = n = c (j
1
. · (j
1
. n)) = c (j
1
. n
1
)
through the equivalence of UMP and EMP.
36
« This is the additional wealth the agent would have needed at the old
prices j
0
in order to attain the level of utility n
1
that is available under
the new prices j
1
.
« The following is also a plausible way to de…ne the change in welfare
from the money metric:
c
_
j
1
. n
1
_
÷c
_
j
1
. n
0
_
= c
_
j
0
. n
0
_
÷c
_
j
1
. n
0
_
.
This is what the agent has to spend at the new prices j
1
in order to
maintain the original utility n
0
that was available at the old prices j
0
.
« Note, however, that if there were wealth e¤ects, then these measures
would not coincide. Consider a price change in good 1 only. If j
0
1
j
1
1
,
so that n
1
n
0
, then n
1
is associated with greater wealth at given
prices. If good 1 is normal, this means more of it is demanded at
optimum. If good 1 is inferior, less of it is demanded. Hence the chosen
bundles and their associated expenditures would depend on how n is
…xed.
Exercise 23 (MWG 3.I.3). Welfare change as measured by
1\
_
j
0
. j
1
. n
_
= c
_
j
0
. n
1
_
÷c
_
j
1
. n
1
_
is called the "equivalent variation" (EV), and
C\
_
j
0
. j
1
. n
_
= c
_
j
0
. n
0
_
÷c
_
j
1
. n
0
_
is called the "compensating variation" (CV). (Respectively, these tell us the
change in wealth that is required to maintain utility at n
1
and n
0
after a
price change.) Suppose the price of good / falls (other price remained …xed),
giving the new price vector j
1
_ j
0
. Demonstrate that C\ (j
0
. j
1
. n)
1\ (j
0
. j
1
. n) if good / is inferior.
Exercise 24 (MWG 3.I.5). Suppose n(r) is quasilinear with respect to
the …rst good (and j = 1 is …xed). Show that C\ (j
0
. j
1
. n) = 1\ (j
0
. j
1
. n)
for any prices j
0
and j
1
, at all wealth levels n.
Exercise 25 (MWG3.I.6). Let there be a population of consumers indexed
by i = 1. . . . . 1, with utility functions n
i
(r) and individual wealths n
i
. For
37
any change in prices from j
0
to j
1
such that
i
C\
i
(j
0
. j
1
. n) 0, show that
it is possible to compensate everyone for lost utility. I.e. there exist wealth
levels ¦n
0
i
¦
1
i=1
such that
i
n
0
i
_
i
n
i
and ·
i
(j
1
. n
0
i
) _ ·
i
(j
0
. n
i
) for all i.
« Absent wealth e¤ects, the two de…nitions of welfare change are equiv
alent. We can use the fact that, for / = 1. . . . . 1,
Jc (j. n)
Jj
¹
= /
¹
(j. n)
(by the envelope theorem, since c (j. n) = j /(j. n) is locally invariant
to changes in its minimizer /(j. n)) to derive
c
_
j
0
. n
_
÷c
_
j
1
. n
_
=
1
¹=1
_
j
0
`
0
/
¹
(j. n) dj
¹
÷
1
¹=1
_
j
1
`
0
/
¹
(j. n) dj
¹
=
1
¹=1
_
j
0
`
j
1
`
/
¹
(j. n) dj
¹
.
« This is the change in consumer surplus as a result of the price change.
See Figure 3 for illustrations of the consumer surplus from good 1 before
and after a price cut.
« What if only partial information is available about demand, such as
the initial price j
0
and choice r
0
= r (j
0
. n), and we wish to evaluate
the impact of a change in prices to j
1
?
Proposition. If preference is locally nonsatiated, then the agent is strictly
better o¤ under (j
1
. n) than under (j
0
. n) if (j
1
÷j
0
) r
0
< 0.
Proof. By Walras’ law, j
0
r
0
= n, so (j
1
÷j
0
) r
0
< 0 implies j
1
r
0
< n.
But then r
0
is in the interior of the budget set at j
1
, and by local nonsatiation
there exists a bundle r ¸ 1
j
1
,&
that is strictly preferred to r
0
.
Exercise 26 (MWG 3.I.12). Extend this test of welfare improvement to
changes in prices and wealth from (j
0
. n
0
) to (j
1
. n
1
), where it is now not
necessarily the case that n
1
= n
0
. (No need to do this for equivalent and
compensating variation.)
38
Figure 3: Change in consumer surplus
« The converse is not always true: (j
1
÷j
0
) r
0
0 does not imply
that the agent is worse o¤ after the price change. To appreciate the
di¤erence, look at Figure 4, where the two scenarios are depicted in
price space.
« The set of prices that keep expenditure constant at c (j
0
. r
0
) is drawn
as a convex curve, because the expenditure function is concave in each
price. (I.e. keeping j
1
…xed, expenditure increases at a diminishing
rate as j
2
increases, hence smaller reductions in j
1
are required to
o¤set increases of given size in j
2
.)
« Since j
0
is an optimal choice, it attains c (j
0
. r
0
), and therefore lies on
the curve. The gradient of the …xedexpenditure curve at this point
is \
j
c (j
0
. r
0
) = r
0
, by the envelope theorem: since r
0
minimizes
c (j
0
. r
0
) at j
0
, i.e. \
a
c (j
0
. r
0
) = 0, we have for / = 1. 2,
dc (j
0
. r
0
)
dj
¹
=
Jc (j
0
. r
0
)
Jj
¹
+
Jc (j
0
. r
0
)
Jr
¹
Jr
¹
Jj
¹
=
Jc (j
0
. r
0
)
Jj
¹
= r
¹
.
« The gradient vector r
0
of c (j. n
0
) is orthogonal to the tangent of the
level set ¦j s.t. c (j. n
0
) = c (j
0
. n
0
)¦ at j
0
. To see this, think of a vector
39
Figure 4: Scenarios (j
1
÷j
0
) r
0
< 0 (left) and (j
1
÷j
0
) r
0
0 (right)
in the tangent space (i.e. a vector in the direction of the tangent). It
must be a multiple of · = (1. ÷r
0
1
,r
0
2
), since a oneunit increase in j
1
requires j
2
to be reduced by (r
0
1
,r
0
2
) j
1
to keep expenditure constant.
Clearly, r
0
· = 0.
« Since j
1
÷j
0
is the vector "from" j
0
to j
1
, it must lie below the tangent
to the level curve, which is orthogonal to r
0
, when (j
1
÷j
0
) r
0
< 0,
and above when (j
1
÷j
0
) r
0
0. The concavity of the expenditure
function therefore implies that c (j
1
. n
0
) < c (j
0
. n
0
) in the …rst case (so
that there is a welfare improvement), but not necessarily in the second.
As drawn in the right panel of Figure 4, welfare decreases, since j
1
lies
above the level curve at c (j
0
. n
0
), but if the price change (in the same
direction) were su¢ciently large, j
1
could lie beneath the level curve.
Exercise 27 (MWG 3.I.11). Suppose r
1
= r (j
1
. n) is known in addition
to j
0
, j
1
and r
0
. Argue that the agent is worse o¤ at j
1
than at j
0
if
(j
1
÷j
0
) r
1
0, or equivalently if j
0
(r
1
÷r
0
) < 0 (wealth is …xed).
Give graphic intuition for these results. (They can be established via …rst
order approximation of the expenditure function at j
1
, but direct proofs are
enough.)
40
6 ChoiceBased Approach
6.1 Choice Structures
« So far we have assumed that choice behavior is consistent with an un
derlying rational preference ordering. Since preferences are unobserv
able, their existence and properties are not directly testable. In this
lecture, we will build a largely parallel theory of demand that starts
from the properties of observable choices.
« A choice structure (E. C ()) consists of a family E of budget sets and
a choice rule C () that assigns a nonempty subset C (1) _ 1 to every
1 ¸ E.
« In this context, a budget set 1 ¸ E may be thought of as a speci…c
decision problem, of many possible problems, the agent may face. The
problem the agent solves is to choose one or more elements from this
set.
« Since existence of a preference is not assumed, it is not meaningful to
say that objects in C (1) are indi¤erent to each other and preferred to
everything else in 1. The set C (1) simply describes the objects the
agent might be observed to choose when presented with budget set 1,
whatever the reasons.
« One may de…ne a "revealed preference" relation from a choice rule as
follows.
De…nition. %
is the "revealed preference" derived from choice structure
(E. C ()) if r %
¸ ("r is revealed preferred to ¸") if and only if r ¸ C (1)
for some 1 ¸ E such that r. ¸ ¸ 1.
« The relation %
need have none of the rationality properties we assumed
for primitive preferences. For example, r and ¸ are only comparable if
r ¸ C (1) or ¸ ¸ C (1) for some 1 ¸ E such that r. ¸ ¸ 1. If r and ¸
are never chosen, then we have no information about them. Hence %
is not necessarily complete.
41
« There is also no guarantee of transitivity. For example, if C (¦r. ¸¦) =
¦r¦, C (¦¸. .¦) = ¦¸¦, and C (¦r. .¦) = ¦.¦, then r ~
¸ ~
., but
. ~
r.
Exercise 28 (MWG 2.F.4). Laspeyres and Paasche indices measure the
change in consumption between two points in time at …xed prices. Let
j
0
and ¡
0
denote prices and quantities at time 0, and let j
1
and ¡
1
de
note the new prices and quantities at time 1. The Laspeyres index, 1
Q
=
(j
0
r
1
) , (j
0
r
0
), is based on initial prices. The Paasche index, 1
Q
=
(j
1
r
1
) , (j
1
r
0
), uses new prices. Consider also the expenditure change
1
Q
= (j
1
r
1
) , (j
0
r
0
), which allows prices to vary. (If demands refer to
aggregate consumption, this is the percent change in GDP.) Argue:
(a) If 1
Q
< 1, then r
0
is revealed preferred to r
1
.
(b) If 1
Q
1, then r
1
is revealed preferred to r
0
.
(c) If 1
Q
< 1 or 1
Q
1, no revealed preference between r
0
and r
1
can
be established.
« Given a budget set and a preference relation %, one may derive a choice
rule as follows.
De…nition. (E,C
(. %) . ) is the choice struture generated by preference %
if and only if \1 ¸ E,
C
(1. %) = ¦r ¸ 1 s.t. \¸ ¸ 1, r % ¸¦ .
« In principle, the generated choice rule could be empty for some 1 ¸
E. (I.e. there is no mostpreferred element in the budget set.) As
we know, rationality and continuity of % ensure that there exists a
continuous utility representation and a solution to the UMP, hence that
C
(1. %) ,= ?. Whenever we refer to a generated choice structure in
this lecture, we will impicitly assume that it satis…es the de…nition of a
choice structure, i.e. that the generated choices are always nonempty.
42
6.2 Weak Axiom
« In order to have anything substantive to say about observed choices, we
need to impose some minimal consistency. The weak axiom of revealed
preference says that, if r is ever a choice when ¸ is available, then r
must be a choice whenever r is available and ¸ is a choice. In other
words, if we ever observe a preference for r over ¸, then we can never
observe a strict preference for ¸ over r.
De…nition. Choice structure (E. C ()) satis…es the weak axiom (WA) if,
whenever r ¸ C (1) for 1 ¸ E with r. ¸ ¸ 1, and ¸ ¸ C (1
0
) for 1
0
¸ E
with r. ¸ ¸ 1
0
, we also have r ¸ C (1
0
).
Exercise 29 (MWG 2.F.3). The following is partial information about a
consumer’s purchases:
Year 1 Year 2
Quantity Price Quantity Price
Good 1 100 100 120 100
Good 1 100 100 ? 80
.
Give the range of quantities of good 2, consumed in year 2, such that
(a) the choices violate the weak axiom,
(b) the consumption bundle in year 1 is revealed preferred to that in year
2,
(c) the consumption bundle in year 2 is revealed preferred to that in year
1,
(d) neither (a), (b) nor (c) can be concluded based on the data,
(e) given WA holds, good 1 is inferior at some price,
(f) given WA holds, good 2 is inferior at some price.
Exercise 30 (MWG 1.C.2). Argue that WA is equivalent to the following
property. If 1. 1
0
¸ E, with r. ¸ ¸ 1 ¨ 1
0
, then r ¸ C (1) and ¸ ¸ C (1
0
)
imply ¦r. ¸¦ _ C (1) ¨ C (1
0
).
Example. In the nontransitive scenario above, WA is not violated. Since
each pair (¦r. ¸¦, ¦¸. .¦, ¦r. .¦) belongs to only one 1 ¸ E, the axiom is
not tested. It would be a di¤erent matter if we added a fourth budget set
¦r. ¸. .¦ to E. Now any choice rule violates WA. If r ¸ C (¦r. ¸. .¦), then
43
WA stipulates that r is a choice for any subset of ¦r. ¸. .¦ that contains r
and where ¸ or . is a choice. But this was not the case. The same argument
applies to ¸ ¸ C (¦r. ¸. .¦) and . ¸ C (¦r. ¸. .¦), so that C (¦r. ¸. .¦) must
be empty (which is impossible by de…nition).
« Hence, for an arbitrary family of budget sets E, it is not the case
that the revealed preference %
derived from (E. ( ()) which satis…es
WA is necessarily rational. The choice rule needs to be de…ned on a
su¢ciently comprehensive family of budget sets. (This makes it more
restrictive, in the sense that it has to "commit" to a choice in more
decision situations.)
Proposition. Let E include all subsets of A that contain one, two or three
elements. Then, if the choice structure (E. ( ()) satis…es WA, the revealed
preference %
derived from it is rational.
Proof. The relation %
derived from (E. ( ()) is complete: \r. ¸ ¸ A,
we have ¦r. ¸¦ ¸ E. Either r ¸ C (¦r. ¸¦), i.e. r %
¸, or ¸ ¸ C (¦r. ¸¦),
i.e. ¸ %
r, or both. Moreover, %
derived from (E. ( ()) is transitive:
suppose r %
¸ and ¸ %
.. Then there exists 1 such that r. ¸ ¸ 1
and r ¸ C (1), and there exists 1
0
such that ¸. . ¸ 1
0
and ¸ ¸ C (1
0
).
Consider the budget set ¦r. ¸. .¦. If . ¸ C (¦r. ¸. .¦), then WA requires
¸ ¸ C (¦r. ¸. .¦). If ¸ ¸ C (¦r. ¸. .¦), then WA requires r ¸ C (¦r. ¸. .¦).
Hence r ¸ C (¦r. ¸. .¦), whereby r %
..
« Since the revealed preference ordering is fully determined by choices
from twoobject sets, it is in fact unique. (Provided every possible set
of two objects is included in E.)
Exercise 31 (MWG 1.C.3). Let choice structure (E. C ()) satisfy WA,
and de…ne revealed strict preference such that r ~
¸ == ¬1 ¸ E with
r. ¸ ¸ 1, r ¸ C (1) and ¸ , ¸ C (1).
(a) Compare ~
to ~
de…ned such that r ~
¸ == r %
¸ and
not ¸ %
r (where %
is the revealed preference derived from the choice
44
structure). Show that the two de…nitions are equivalent. Does this depend
on WA?
(b) Give an example where ~
is not transitive.
(c) Argue that ~
is transitive if E includes all threeelement subsets of A.
« When the choice rule C () in choice structure (E. C ()) is generated by
a rational preference %, i.e. \1 ¸ E, C (1) = C
(1. %), we say that
% rationalizes C ().
« One can think of generating a choice structure from a rational pref
erence as simulating empirical choice data with a rational preference
model. If the model exactly predicts choices, then the choices are ex
plained (rationalized) by the model (preference).
« If the revealed preference %
derived from choice structure (E. C ()) is
rational, then it rationalizes (E. C ()). This is not quite obvious; we
must verify that %
generates (E. C ()). In fact, that statement is true
only if WA holds for (E. C ()).
Proposition. The revealed preference %
derived from a choice structure
(E. C ()) that satis…es WA generates (E. C ()).
Proof. Our task is to show that \1 ¸ E, C (1) = C
(1. %). Let r ¸
C (1). Since r %
¸, \¸ ¸ 1, we have r ¸ C
(1. %
). Thus C (1) _
C
(1. %
). In the other direction, let r ¸ C
(1. %
), i.e. r %
¸, \¸ ¸ 1.
Then there must exist, for every ¸ ¸ 1, a budget set 1
j
such that r. ¸ ¸
1
j
and r ¸ C (1
j
). Now, some such ¸ ¸ 1 must be chosen from 1, i.e.
¸ ¸ C (1). By WA, this implies r ¸ C (1). Thus C
(1. %
) _ C (1).
Combining the inclusions, C (1) = C
(1. %
).
« Hence, if WAholds for the choice structure, then the revealed preference
derived from it is rational and generates the choice structure. I.e. WA
implies that the choice structure is rationalizable (provided the choice
rule covers all sets of up to three objects).
« It turns out that WA is not only su¢cient (with restrictions on E), but
also necessary for a choice structure to be rationalizable.
45
Proposition. A choice structure (E. C
(. %)) that is generated by a rational
preference % satis…es WA.
Proof. Suppose r ¸ C
(1. %) for 1 ¸ E and ¸ ¸ 1. Since the choice
structure (E. C
(. %)) is generated by %, r % ¸. Let ¸ ¸ C
(1
0
. %) and
r ¸ 1
0
. WA requires that r ¸ C
(1
0
. %). Suppose now that % is rational,
i.e. in particular transitive. Because ¸ ¸ C
(1
0
. %), we have \. ¸ 1
0
, ¸ % .
and by transitivity r % .. Then r ¸ C
(1
0
. %), so WA holds.
« Put together, these results establish that a choice rule, de…ned (at least)
on all sets with up to three elements, re‡ects a rational preference if
and only if it satis…es WA.
« One might be tempted to conclude that the preference and choice
based approaches are basically equivalent, since we could include all
possible budget sets in E. But in the theory of demand, budget sets
have a special form (they satisfy j r _ n), which is restrictive (e.g.
convex). Arbitrary budget sets may not make sense, nor does a solution
to the UMP necessarily exist for them.
« In a meaningful sense, the preferencebased approach (using rational
ity) is less general than the choicebased approach (using the weak
axiom). Rational preference always gives us the weak axiom in a gen
erated choice structure, but choices that satisfy the weak axiom need
not be consistent with rational preference.
Exercise 32 (MWG 1.D.4). If choice structure (E. C ()) is rational
izable, show that it satis…es pathinvariance: \¦1
1
. 1
2
¦ _ E such that
1
1
' 1
2
¸ E and C (1
1
) ' C (1
2
) ¸ E, it is the case that C (1
1
) ' C (1
2
) =
C (C (1
1
) ' C (1
2
)).
Exercise 33 (MWG 1.D.5). On the set of objects A = ¦r. ¸. .¦, de…ne the
family of budget sets E = ¦¦r. ¸¦ . ¦¸. .¦ . ¦r. .¦¦. Think of the choice rule
C () as assigning, to each budget set 1 ¸ E, a probability distribution C (1)
over objects in 1. This stochastic choice rule C () is said to be rationalizable
by preferences if there exists a probability distribution over (strict) preference
relations on A (here there are six such relations) that \1 ¸ E induces C (1).
46
(a) Show that C (¦r. ¸¦) = C (¦¸. .¦) = C (¦r. .¦) = (1,2. 1,2) is ratio
nalizable in this sense.
(b) Show that C (¦r. ¸¦) = C (¦¸. .¦) = C (¦r. .¦) = (1,4. 3,4) is not
rationalizable in this sense.
(c) Find c. c ¸ [0. 1] such that C (¦r. ¸¦) = C (¦¸. .¦) = C (¦r. .¦) =
(c. 1 ÷c) is rationalizable if and only if c ¸ [c. c].
6.3 Relationship with the Law of Demand
« It seems plausible that Walrasian demand could satisfy a version of the
law of demand if we mimic the compensation for price changes that is
implicit in Hicksian demand by adjusting wealth with prices.
« This intuition is correct. In addition to homogeneity of degree zero and
Walras’ law, the weak axiom is the minimal property we need to impose
on (singlevalued) Walrasian demand to make it satisfy a compensated
law of demand.
« In the context of (singlevalued) demand, WA has the following speci…c
form: \(j. n) and \(j
0
. n
0
),
j r (j
0
. n
0
) _ n and r (j. n) ,= r (j
0
. n
0
) == j
0
r (j. n) n
0
.
« I.e. if r (j
0
. n
0
) was available at prices j and wealth n, but r (j. n)
was chosen instead, then r (j. n) must be unavailable at prices j
0
and
wealth n
0
, when r (j
0
. n
0
) is chosen.
« Notice how the singlevaluedness of choices a¤ects WA: since we cannot
require that r ¸ C (1
0
) when r ¸ C (1 ¸ ¸) and ¸ ¸ C (1
0
¸ r), we
stipulate instead that r , ¸ 1
0
. Singlevaluedness implies, when r is
chosen over ¸ in 1, that r is revealed strictly preferred to ¸, and strict
preference is antisymmetric.
Exercise 34 (MWG 2.F.12). Verify that a Walrasian demand function
r (j. n) which is generated by a rational preference satis…es WA.
Exercise 35 (MWG 2.F.14). Argue that a Walrasian demand function
r (j. n) that satis…es WA is homogeneous of degree zero.
47
« Figure 5 illustrates how WA restricts choices given budget sets 1
j,&
and 1
j
0
,&
0 . The top left panel depicts the budget sets; the other panels
show all possible locations for r = C (1
j,&
) and r
0
¸ C (1
j
0
,&
0 ) that
satisfy WA. In the top right panel, r ¸ 1
j,&
¨1
j
0
,&
0 . Since r is available
in both budget sets, the (single) choice of r
0
in 1
j
0
,&
0 reveals it (strictly)
preferred to r. Then r
0
, ¸ 1
j,&
, else not choosing r
0
violates WA. The
same type of argument applies in the other cases.
« Before I give an analytical proof of the equivalence of WA and the
compensated law of demand, a graphic sketch will be helpful. Suppose
we start with bundle r, which lies on the boundary of the budget
set 1
j,&
, in accordance with Walras’ law. Consider a compensated
price change to j
0
, where wealth is adjusted to n
0
such that r remains
a¤ordable. The pivoting of the budget line through r is visible in Figure
6.
« In the left panel, the set of bundles that satis…es WA is dotted. Since
r is a¤ordable in both 1
j,&
and 1
j
0
,&
0 (by the construction of the com
pensation), and was revealed preferred to everything in 1
j,&
, the choice
r
0
in 1
j
0
,&
0 must lie outside 1
j,&
if it is distinct from r. (Otherwise, r
0
would have been chosen prior to the compensated price change.) If we
assume that r
0
also satis…es Walras’ law, then it lies on the boldstriped
portion of the boundary of 1
j
0
,&
0 .
« In the right panel, the set of bundles that satis…es CLD is dotted. Since
j
0
2
< j
2
as drawn, and the price change is compensated, CLD requires
that r
0
2
r
2
. Then r
0
must lie in the triangle above r below the new
budget line. By Walras’ law, r
0
belongs to the boldstriped portion of
the boundary of 1
j
0
,&
0 . This is exactly the same set that satis…es WA.
« Are we done? Actually no. We have only shown that WA is equivalent
to CLD for compensated price changes. But WA is a property that
pertains to arbitrary (possibly uncompensated) price changes. We must
still demonstrate that such price changes will also satisfy WA.
« This point can be made via the contrapositive: if a price change violates
WA, then we can construct a compensated price change that violates
CLD, contrary to the assumption. (Hence no price change can violate
WA.)
48
Figure 5: WA for singlevalued demand
49
Figure 6: Equivalence of WA and CLD for compensated price changes
« In Figure 7, we have situation that is at odds with WA: r and r
0
both
belong to 1
j,&
and 1
j
0
,&
0 (other violations are straightforward to deal
with). We can construct a new budget set 1
j
00
,&
00 that represents a
compensated price change for both r and r
0
(graphically, the boundary
of 1
j
00
,&
00 pivots through both r and r
0
).
« One can see that j
00
2
< j
2
and j
00
2
j
0
2
. Hence CLD requires, for these
compensated changes in demand, that r
00
2
r
2
and r
00
2
< r
0
2
. I.e. r
00
must lie in both dotted triangles, above r and below r
0
. But these are
disjoint sets, so CLD is violated. Since failures of WA (with arbitrary
price changes) lead to contradiction, when CLD holds for compensated
price changes, CLD is in fact su¢cient for WA.
Proposition. For a Walrasian demand function r (j. n) that is homoge
neous of degree zero and satis…es Walras’ law, WA holds if and only if \(j. n)
and \(j
0
. n
0
) such that n
0
= j
0
r (j. n) and r (j. n) ,= r (j
0
. n
0
),
(j
0
÷j) (r (j
0
. n
0
) ÷r (j. n)) < 0.
50
Figure 7: Impossibility of WA failures
Proof. Since r (j
0
. n
0
) satis…es Walras’ law, we have j
0
r (j
0
. n
0
) = n
0
=
j
0
r (j. n), so that the CLD inequality reduces to the equivalent inequality
j (r (j
0
. n
0
) ÷r (j. n)) 0.
(If) WA holds vacuously if r (j. n) = r (j
0
. n
0
) and is satis…ed if j
0
r (j. n)
n
0
or j r (j
0
. n
0
) n. Consider therefore (j. n) and (j
0
. n
0
) where r (j. n) ,=
r (j
0
. n
0
) and n
0
_ j
0
r (j. n). If n
0
= j
0
r (j. n), then the inequality
applies, and j r (j
0
. n
0
) _ n would imply j r (j. n) < n, violating Walras’
law. Hence j r (j
0
. n
0
) n, so that WA is not tested. Similarly for the case
n = j r (j
0
. n
0
).
Thus we can concentrate on j
0
r (j. n) < n
0
and j r (j
0
. n
0
) < n.
This appears to be a violation of WA, but we will show that there exists
a compensated price change for which the CLD cannot hold. Hence the
scenario does not satisfy the assumptions.
Given the strict inequalities, there exists c ¸ (0. 1) such that
(cj + (1 ÷c) j
0
) r (j. n) = (cj + (1 ÷c) j
0
) r (j
0
. n
0
)
(because at c su¢ciently close to 1, the right side is close to j r (j. n) =
n j r (j
0
. n
0
), and at c su¢ciently close to 0, the left side is close to
51
j
0
r (j
0
. n
0
) = n
0
j
0
r (j. n) < n
0
, where the equalities are due to Walras’
law).
Let j
00
= cj+(1 ÷c) j
0
and n
00
= (cj + (1 ÷c) j
0
)r (j
0
. n
0
) = (cj + (1 ÷c) j
0
)
r (j. n). Since
j
00
r (j. n) = n
00
and
j
00
r (j
0
. n
0
) = n
00
.
we have constructed compensated price changes from j to j
00
and from j
0
to
j
00
.
By Walras’ law, j
00
r (j
00
. n
00
) = n
00
, so
j
00
(r (j
00
. n
00
) ÷r (j. n)) = 0
and
j
00
(r (j
00
. n
00
) ÷r (j
0
. n
0
)) = 0.
Because j to j
00
and j
0
to j
00
are compensated price changes, CLD must hold
for both:
j (r (j
00
. n
00
) ÷r (j. n)) 0
and
j
0
(r (j
00
. n
00
) ÷r (j
0
. n
0
)) 0.
From the de…nition of j
00
, we have j = (1,c) j
00
÷((1 ÷c) ,c) j
0
. There
fore, CLD for j to j
00
implies
(j
00
÷(1 ÷c) j
0
) (r (j
00
. n
00
) ÷r (j. n)) 0.
and, since j
00
(r (j
00
. n
00
) ÷r (j. n)) = 0, this means
j
0
r (j. n) j
0
r (j
00
. n
00
) .
Now, since j
0
r (j
0
. n
0
) = n
0
j
0
r (j. n) by assumption, it follows that
j
0
r (j
0
. n
0
) j
0
r (j
00
. n
00
) .
But then the CLD for j
0
to j
00
cannot hold. Hence j
0
r (j. n) < n
0
and
j r (j
0
. n
0
) < n is not possible, and we have shown that WA holds for
all (j. n) and (j
0
. n
0
) if CLD holds for all compensated price changes (and
Walras’ law is in e¤ect).
52
(Only if) Because j
0
r (j. n) = n
0
, the bundle r (j. n) is available at
(j
0
. n
0
). But since r (j
0
. n
0
) is chosen instead at (j
0
. n
0
) (revealing it pre
ferred), WA requires that r (j
0
. n
0
) is unavailable at (j. n), i.e. j r (j
0
. n
0
)
n. From Walras’ law, j r (j. n) = n, which gives the strict inequality
j (r (j
0
. n
0
) ÷r (j. n)) 0.
« Observe that we are not comparing initial demand r (j. n) to the un
compensated demand r (j
0
. n) after the price change to j
0
. Instead,
wealth is adjusted to n
0
= j
0
r (j. n), so that the initial bundle is still
in the budget set at the new prices.
« The remarkable aspect of the result is that nothing more than WA is
needed for the compensated law of (Walrasian) demand (other than
homogeneity of degree zero and Walras’ law, which require only that
preference is locally nonsatiated). In particular, preference does not
have to be rational or continuous.
« Since it is an "if and only if" proposition, WA can be said to be equiv
alent to the compensated law of demand under local nonsatiation.
Exercise 36 (MWG 2.F.5). Let a di¤erentiable Walrasian demand func
tion r (j. n) satisfy homogeneity of degree zero, Walras’ law and the weak
axiom. Suppose r (. ) is also homogeneous of degree one with respect to
wealth n, so that consumption of all goods increases proportionately: \j. n
and \c 0, r (j. cn) = cr (j. n). Show that the law of demand holds also
for uncompensated price changes: \j. j
0
, (j
0
÷j) (r (j
0
. n) ÷r (j. n)) _ 0.
Exercise 37 (MWG 2.F.13). Consider a multivalued Walrasian demand
correspondence r (j. n).
(a) Give the Walrasian version of WA in this case, generalized to choice
sets.
(b) If r (j. n) satis…es WA and Walras’ law, show that r (j. n) also has
the following property: \r ¸ r (j. n) and \r
0
¸ r (j
0
. n
0
),
j r
0
< n == j r n.
53
(c) Moreover, show that if r (j. n) satis…es WA and Walras’ law, then
the following compensated law of demand holds: \(j. n) and \r ¸ r (j. n),
and \(j
0
. n
0
) such that n
0
= j
0
r,
(j
0
÷j) (r
0
÷r) _ 0.
6.4 Strong Axiom
« A choicebased theory that is fully equivalent to the preferencebased
approach with rationality can be based on the strong axiom of revealed
preference.
De…nition. Choice structure (E. C ()) satis…es the strong axiom (SA) if,
for any collection of budget sets
_
1
1
. . . . . 1
.
_
_ E, and r
1
. . . . . r
.
such that
r
a
¸ C (1
a
) and r
a+1
¸ 1
a
for : = 1. . . . . ` ÷ 1, we have r
1
¸ C
_
1
.
_
if
r
1
¸ 1
.
.
« SA is just WA if we consider only collections of ` = 2 budget sets
¦1. 1
0
¦ _ E: then r ¸ C (1) and ¸ ¸ C (1
0
) with ¸ ¸ 1 and r ¸ 1
0
implies r ¸ C (1
0
). I.e. if r
1
is chosen when r
2
is available, then r
1
must be chosen whenever it is available and r
2
is chosen.
« Beyond WA, SA says, if r
1
is chosen when r
2
is available, and r
2
is
chosen when r
3
is available (so that r
1
is indirectly revealed preferred
to r
3
), then r
1
must be chosen whenever it is available and r
3
is chosen.
(And we could make the chain arbitrarily long.) You will recognize the
‡avor of transitivity in SA.
« Transitivity is precisely what WA failed to guarantee in a revealed
preference relation derived from it. SA, however, is su¢cient without
any requirements on the domain of the choice rule. Thus, If the choice
structure (E. C ()) satis…es SA, then the revealed preference %
derived
from it is rational. (The proof is nonelementary.)
« Since SA implies WA, it inherits all properties of WA. Therefore, a
choice structure is generated by a rational preference if it satis…es SA
and only if it satis…es WA.
Exercise 38 (MWG 3.J.1). Show for the consumption set R
2
+
that Wal
rasian demand satis…es WA if and only if it satis…es SA.
54
7 Integrability
7.1 Slutsky and Hicks Compensation
« If choices are generated from rational preferences, then they satisfy the
weak axiom, which is equivalent to the compensated law of demand
(as long as choices are homogeneous of degree zero and satisfy Walras’
law): \(j. n) and \(j
0
. n
0
) such that n
0
= j
0
r (j. n),
(j
0
÷j) (r (j
0
. n
0
) ÷r (j. n)) _ 0.
Hence rational preferences imply the version of the compensated law
of demand we posed in the discussion of the choicebased approach.
« The reverse conclusion is problematic: while the compensated law of
demand implies the weak axiom, the weak axiom is not su¢cient for
the existence of a rational preference that generates the choices.
« In the context of the expenditure minimization problem, we also derived
a compensated law of demand (for continuous and locally nonsatiated
preferences): \j and \j
0
,
(j
0
÷j) (/(j
0
. n) ÷/(j. n)) _ 0.
« Are the two statements of the compensated law of demand equiva
lent? In the choicebased approach, the consumer’s wealth is explic
itly adjusted so that the initial choice r remains a¤ordable at prices
j
0
. Graphically, a price increase (inward pivot of the budget line) is
accompanied by a wealth increase (outward shift of the budget line,
which passes through r), as shown in the left panel of Figure 8. This
type of compensation is called Slutsky compensation.
« In the preferencebased approach, the consumer’s utility is held …xed
and expenditure is allowed to vary, so that the initial utility is still
attained at the optimal choice /(j
0
. n) after the price increase. Graph
ically, a price increase is accompanied by a wealth increase (outward
shift of the budget line, which remains tangent to the indi¤erence curve
at n), as shown in the right panel of Figure 8. This is known as Hicks
compensation.
55
Figure 8: Compensation in the choice and preference approaches
« Visually, it is clear that the two versions of the compensated law of
demand are not the same. However, for in…nitesimal price changes,
they do lead to the same adjustments in choices.
« Our interest in this lecture is in whether the preferencebased com
pensated law of demand implies stronger restrictions on choices that
guarantee rationalizability. I.e. if we observe that choices satisfy it,
can they always be constructed from a rational preference?
« To this end, we will characterize the compensated law of demand in the
choicebased and preferencebased approaches in terms of the substi
tution matrix, which contains price e¤ects. Negative semide…niteness
of the substitution matrix is (almost) equivalent to the choicebased
compensated law of demand and necessary (but not su¢cient) for the
preferencebased compensated law of demand.
« The additional property that the substitution matrix must have if the
preferencebased compensated law of demand holds is symmetry. We
then argue that symmetry is enough to recover a rational preference
from choices, so that the preferencebased compensated law of demand
indeed guarantees rationalizability.
56
7.2 Aside: Dot Product
« Because we will make heavy use of vector notation and transformations
of vector equations in this lecture, I brie‡y review the properties of the
dot product.
« The dot product of vectors r and ¸ is the sum of the products of
corresponding elements (that have the same indices):
r ¸ =
1
¹=1
r
¹
¸
¹
.
« Because multiplication is commutative, so is the dot product: r¸ = ¸r
since
r ¸ =
1
¹=1
r
¹
¸
¹
=
1
¹=1
¸
¹
r
¹
= ¸ r.
« Because multiplication is distributive, so is the dot product: r(¸ +.) =
r ¸ +r ¸ since
r (¸ +.) =
1
¹=1
r
¹
(¸
¹
+.
¹
) =
1
¹=1
r
¹
¸
¹
+
1
¹=1
r
¹
.
¹
= r ¸ +r ..
« The dot product is not associative: (r ¸) . ,= r (¸ .) since, in
general,
(r ¸) . =
1
I=1
_
1
¹=1
r
¹
¸
¹
_
.
I
=
_
1
¹=1
r
¹
¸
¹
_
.
1
+ +
_
1
¹=1
r
¹
¸
¹
_
.
1
,=
_
1
¹=1
¸
¹
.
¹
_
r
1
+ +
_
1
¹=1
¸
¹
.
¹
_
r
1
.
« The dot product of r and ¸ can be written in matrix notation:
r ¸ = r
T
¸.
Commutativity, r
T
¸ = ¸
T
r, and distributivity, r
T
(¸ +.) = r
T
¸+r
T
.,
are of course inherited.
57
« The outer product of vectors r and ¸ is a matrix (as opposed to the
inner product r
T
¸, which is a scalar):
r¸
T
=
_
¸
_
r
1
¸
1
r
1
¸
1
.
.
.
.
.
.
.
.
.
r
1
¸
1
r
1
¸
1
_
¸
_
.
« In matrix notation, there is a version of associativity, which involves a
switch from outer product to inner product:
_
r¸
T
_
. = r
_
¸
T
.
_
since
_
r¸
T
_
. =
_
¸
_
r
1
¸
1
r
1
¸
1
.
.
.
.
.
.
.
.
.
r
1
¸
1
r
1
¸
1
_
¸
_
_
¸
_
.
1
.
.
.
.
1
_
¸
_
=
_
1
¹=1
¸
¹
.
¹
_
r
1
+ +
_
1
¹=1
¸
¹
.
¹
_
r
1
=
_
¸
_
r
1
.
.
.
r
1
_
¸
_
_
_
_
_
¸
1
¸
1
¸
_
¸
_
.
1
.
.
.
.
1
_
¸
_
_
_
_
= r
_
¸
T
.
_
.
Notice that the matrix r¸
T
cannot be expressed in terms of the dot
product between two vectors, so there is no con‡ict with the claim that
the dot product is not associative.
7.3 Substitution Matrix
« The substitution matrix (alternatively known as the Slutsky matrix)
o (j. n) lists, for every two commodities / and /, how much more (or
less) is chosen of commodity / per di¤erential increase in the price of
commodity / at (j. n). The entry in row / and column / is
:
¹I
(j. n) =
dr
¹
(j. n)
dj
I
=
Jr
¹
(j. n)
Jj
I
+
Jr
¹
(j. n)
Jn
J (j r (j
0
. n
0
))
Jj
I
=
Jr
¹
(j. n)
Jj
I
+
Jr
¹
(j. n)
Jn
r
I
(j. n) .
where r (j
0
. n
0
) is the …xed initial demand.
« The …rst term measures the direct e¤ect of the price change on r
¹
(j. n)
and is called the (pure) substitution e¤ect. The second term measures
the e¤ect of the Slutsky compensation on r
¹
(j. n) and is called the
wealth e¤ect.
58
« Even though we denote these choices as Walrasian demands, we do
not necessarily assume here that they arise from the utility maximiza
tion problem. These are observed choices from the budget set 1
j,&
that may or may not be rationalizable by an underlying preference.
(They are Walrasian in the sense that there is no implicit adjustment
in expenditure after a price change, via a …xed utility level. Any such
compensation is explicit.)
« Hence
o (j. n) =
_
¸
_
:
11
(j. n) :
11
(j. n)
.
.
.
.
.
.
.
.
.
:
11
(j. n) :
11
(j. n)
_
¸
_
=
_
¸
_
0a
1
(j,&)
0j
1
+
0a
`
(j,&)
0&
r
1
(j. n)
0a
1
(j,&)
0j
L
+
0a
1
(j,&)
0&
r
1
(j. n)
.
.
.
.
.
.
.
.
.
0a
L
(j,&)
0j
1
+
0a
L
(j,&)
0&
r
1
(j. n)
0a
L
(j,&)
0j
L
+
0a
L
(j,&)
0&
r
1
(j. n)
_
¸
_
= 1
j
r (j. n) +1
&
r (j. n) r (j. n)
T
.
« The following is a characterization of the choicebased compensated
law of demand.
Proposition. If Walrasian demand function r (j. n) is di¤erentiable, ho
mogeneous of degree zero and satis…es Walras’ law and the weak axiom,
then \(j. n) the substitution matrix o (j. n) is negative semide…nite, i.e.
\· ¸ R
1
, ·
T
o (j. n) · _ 0.
Proof. Under the assumptions, the compensated law of demand holds:
\(j. n) and \(j
0
. n
0
) such that n
0
= j
0
r (j. n),
(j
0
÷j) (r (j
0
. n
0
) ÷r (j. n)) = dj dr(j. n) _ 0.
This inequality implies negative semide…niteness of the substitution matrix.
From the total derivative of r, and the fact that, for a compensated price
change,
dn = n
0
÷n = (j
0
÷j) r (j. n) = dj r (j. n) .
59
we have
dr = 1
j
r (j. n) dj +1
&
r (j. n) dn
= 1
j
r (j. n) dj +1
&
r (j. n) (dj r (j. n))
= 1
j
r (j. n) dj +1
&
r (j. n)
_
r (j. n)
T
dj
_
=
_
1
j
r (j. n) +1
&
r (j. n) r (j. n)
T
_
dj.
(using commutativity of the dot product and, after switching to matrix no
tation, associativity and distributivity).
Substituting into the …rst inequality,
dj
_
1
j
r (j. n) +1
&
r (j. n) r (j. n)
T
_
dj _ 0.
Because the magnitude of the price change dj ¸ 1
1
was unrestricted, this
implies dr(j. n) ,dj = o (j. n) is negative semide…nite.
« The converse, that a negative semide…nite substitution matrix implies
the weak axiom, is true provided that o (j. n) is in addition negative
de…nite, i.e. ·
T
o (j. n) · < 0, whenever · is not proportional to j
(· ,= cj for any c).
« As we will see in a moment, the substitution matrix cannot be negative
de…nite for all · ¸ R
1
(i.e. it is in fact never de…nite) since, letting
· = j, we get j
T
o (j. n) j = 0.
« An implication of negative semide…niteness is that all diagonal entries
(the ownprice e¤ects) are nonpositive:
dr
¹
(j. n)
dj
¹
=
Jr
¹
(j. n)
Jj
¹
+
Jr
¹
(j. n)
Jn
r
¹
(j. n) _ 0.
Therefore, a Gi¤en good is necessarily inferior, since Jr
¹
(j. n) ,Jj
¹
0
only if Jr
¹
(j. n) ,Jn < 0.
60
Proposition. If Walrasian demand function r (j. n) is di¤erentiable, homo
geneous of degree zero and satis…es Walras’ law, then \(j. n), o (j. n) j =
j
T
o (j. n) = 0.
Proof. Previously (in Lecture 5), we established
1
j
r (j. n) j +1
&
r (j. n) n = 0.
when r (j. n) is homogeneous of degree zero, and
j
T
1
j
r (j. n) +r (j. n)
T
= 0
as well as j
T
1
&
r (j. n) = 1 when r (j. n) satis…es Walras’ law.
Then
o (j. n) j = 1
j
r (j. n) j +1
&
r (j. n)
_
r (j. n)
T
j
_
= 1
j
r (j. n) j +1
&
r (j. n) n = 0
(where the last equality uses Walras’ law), and furthermore
j
T
o (j. n) = j
T
1
j
r (j. n) +j
T
1
&
r (j. n) r (j. n)
T
= j
T
1
j
r (j. n) +r (j. n)
T
= 0.
« Hence negative semide…niteness is exactly what is implied by the choice
based compensated law of demand (equivalently the weak axiom).
« Since o (j. n) j = 0, zero is an eigenvalue of o (j. n), which means
its determinant [o (j. n)[ = 0. (Recall that ` is an eigenvalue if
o (j. n) j = `j for some j or if [o (j. n) ÷`1[ = 0. The latter im
plies [o (j. n)[ = 0 if ` = 0.) This is useful to know when checking
negative semide…niteness from the signs of the principal minors. We
will see an example later on.
Exercise 39 (MWG 2.F.17). Let the Walrasian demand function r (j. n)
have the form: for / = 1. . . . . 1,
r
I
(j. n) =
n
1
¹=1
j
¹
.
61
Determine whether r (j. n)
(a) is homogeneous of degree zero;
(b) satis…es Walras’ law;
(b) is consistent with the weak axiom;
(c) has a negative semide…nite, symmetric substitution matrix.
Exercise 40 (MWG 2.F.10) Compute the substitution matrix for Wal
rasian demand function r (j. n) where
r
1
(j. n) =
1
j
1
+j
2
+j
3
j
2
j
1
.
r
2
(j. n) =
1
j
1
+j
2
+j
3
j
3
j
2
.
r
3
(j. n) =
1
j
1
+j
2
+j
3
j
1
j
3
.
Demonstrate that the substitution matrix is negative semide…nite, but not
symmetric, at j = (1. 1. 1). Show that it does not satisfy the weak axiom.
Exercise 41 (MWG 2.F.16). Let the Walrasian demand function r (j. n)
be as follows:
r
1
(j. n) =
j
2
j
3
.
r
2
(j. n) = ÷
j
1
j
3
.
r
3
(j. n) =
n
j
3
.
(a) Con…rm that r (j. n) is homogeneous of degree zero and that Walras’
law holds.
(b) Demonstrate that r (j. n) does not satisfy the weak axiom.
(c) Show that \· ¸ R
3
, · o (j. n) · = 0.
7.4 Substitution Matrix with Preference
« If we allow choices to arise from a standard preference, then the du
ality between utility minimization and expenditure minimization lets
62
us express the substitution matrix in terms of Hicksian demand at
n = · (j. n):
J/(j. n)
Jj
=
d/(j. · (j. n))
dj
=
dr(j. n)
dj
= :
¹I
(j. n)
for /. / = 1. . . . . 1.
« In order for a di¤erentiable Hicksian demand function /(j. n) to exist
as a solution to the expenditure minimization problem, n() has to
represent a continuous, locally nonsatiated, strictly convex preference
on A = R
1
+
. I emphasize that the equivalence with a given Walrasian
demand function is only valid if the conditions for the existence of a
Hicksian demand function hold.
« Rewriting the substitution matrix accordingly as
o (j. n) =
_
¸
_
0I
1
(j,&)
0j
1
0I
1
(j,&)
0j
L
.
.
.
.
.
.
.
.
.
0I
L
(j,&)
0j
1
0I
L
(j,&)
0j
L
_
¸
_
= 1
j
/(j. n) .
a further property can be derived, because o (j. n) is now implicitly
con…ned to choices the re‡ect an underlying standard preference.
Proposition. If n() represents a continuous, locally nonsatiated, strictly
convex preference on A = R
1
+
, and the Hicksian demand function /(j. n)
is continuously di¤erentiable at n, then the matrix 1
j
/(j. n) = o (j. n) is
negative semide…nite and symmetric (i.e. d/
¹
(j. n) ,dj
I
= d/
I
(j. n) ,dj
¹
for
/. / = 1. . . . . 1).
Proof. Recall …rst that /(j. n) = \
j
c (j. n) by the envelope theorem, so
that
1
j
/(j. n) = 1
2
j
c (j. n) .
Since c (j. n) is a concave function, we can appeal to the fact that the Hessian
(i.e. the secondderivative matrix) of any concave function is negative semi
de…nite.
I will prove this general result. The Taylor expansion of c (j. n) around
o = 0 in displacement direction · ¸ R
1
is
c (j +o·. n) = c (j. n) +\
j
c (j. n)
T
(o·) +
o
2
2
·
T
1
2
j
c (j +·. n) ·
63
for some ¸ [0. o]. Then
·
T
1
2
j
c (j +·. n) · _ 0.
because the concavity of the expenditure function implies
c (j +o·. n) _ c (j. n) +\
j
c (j. n)
T
(o·) .
This is apparent when, in the de…nition of concavity, i.e. \j, \j
0
and
\c ¸ [0. 1],
c (cj + (1 ÷c) j
0
. n) _ cc (j. n) + (1 ÷c) c (j
0
. n) .
we substitute o· = j
0
÷j for j
0
, so that (after rearrangement)
c (j +o·. n) _ c (j. n) +
c (j + (1 ÷c) o·. n) ÷c (j. n)
1 ÷c
.
and we apply the limit as c ÷ 1 (making the last term a derivative with
respect to j, in direction ·).
Now, if the magnitude o of the displacement in the Taylor expansion is
chosen su¢ciently small, then is very close to zero, and we have
·
T
1
2
j
c (j. n) · _ 0.
where the direction · ¸ R
1
was arbitrary. This establishes that 1
2
j
c (j. n) =
1
j
/(j. n) is negative semide…nite.
Symmetry comes from Clairaut’s theorem, which states that the Hessian
of any function that has continuous second partial derivatives at all points
in the domain is symmetric. Since this is true for c (j. n) by virtue of the
fact that its second derivatives are the …rst derivatives of /(j. n), which were
assumed to be continuous, 1
2
j
c (j. n) = o (j. n) must be symmetric.
« The preferencebased compensated law of demand, which is implicit in
Hicksian demand, therefore requires both negative semide…niteness and
symmetry of the substitution matrix. Symmetry was not an implication
of the choicebased compensated law of demand.
64
« Before we examine this di¤erence in examples, we con…rm that the
substitution matrix will not be negative de…nite.
Proposition. If n() represents a continuous, locally nonsatiated, strictly
convex preference on A = R
1
+
, and /(j. n) is di¤erentiable at n, then
1
j
/(j. n) j = o (j. n) j = 0.
Proof. Under these conditions, Hicksian demand is homogeneous of degree
zero in j, so that /(cj. n) = /(j. n). Di¤erentiating both sides with respect
to c, we have for / = 1. . . . . 1,
1
I=1
J/
¹
(cj. n)
Jcj
I
Jcj
I
Jc
=
1
I=1
J/
¹
(cj. n)
Jj
I
j
I
= 0.
which corresponds to the claim.
Example. Suppose choices have the CobbDouglas form:
r
1
(j. n) =
1
3
n
j
1
. r
2
(j. n) =
1
3
n
j
2
. r
3
(j. n) =
1
3
n
j
3
.
Observe that the choices satisfy homogeneity of degree zero and Walras’ law
(they add up to n). The substitution matrix is easily calculated from partial
derivatives to be
o (j. n) =
1
9
n
_
¸
_
÷2
1
j
2
1
1
j
1
j
2
1
j
1
j
3
1
j
1
j
2
÷2
1
j
2
2
1
j
2
j
3
1
j
1
j
3
1
j
2
j
3
÷2
1
j
2
3
_
¸
_
.
Clearly, this is symmetric. One can check directly, that it is negative semi
de…nite because \· ¸ R
1
,
·
T
o (j. n) · =
1
9
n
_
÷2
·
2
1
j
2
1
+ 2
·
1
·
2
j
1
j
2
+ 2
·
1
·
3
j
1
j
3
÷2
·
2
2
j
2
2
+ 2
·
2
·
3
j
2
j
3
÷2
·
2
3
j
2
3
_
= ÷
1
9
n
_
_
·
1
j
1
÷
·
2
j
2
_
2
+
_
·
1
j
1
÷
·
3
j
3
_
2
+
_
·
2
j
2
÷
·
3
j
3
_
2
_
_ 0
(with equality only if · = cj for any c). Hence CobbDouglas choices sat
isfy the both the choicebased and the preferencebased compensated law of
demand.
65
Example. Consider now the following modi…cation:
r
1
(j. n) =
1
3
n
j
1
+
j
2
j
1
. r
2
(j. n) =
1
3
n
j
2
÷1. r
3
(j. n) =
1
3
n
j
3
.
Neither homogeneity of degree zero nor Walras’ law are violated, e.g.
3
¹=1
j
¹
r
¹
(j. n) =
1
3
n +j
2
+
1
3
n ÷j
2
+
1
3
n = n.
The substitution matrix
o (j. n) =
1
9
n
_
¸
_
÷
1
j
2
1
_
2 + 6
j
2
&
_
1
j
1
j
2
_
1 + 3
j
2
&
_
1
j
1
j
3
_
1 + 3
j
2
&
_
1
j
1
j
2
_
1 + 6
j
2
&
_
÷
1
j
2
2
_
2 + 3
j
2
&
_
1
j
2
j
3
_
1 ÷3
j
2
&
_
1
j
1
j
3
1
j
2
j
3
÷2
1
j
2
3
_
¸
_
is obviously not symmetric.
In this case, it is more convenient to check negative semide…niteness in
directly. One technique is to compute the prinicipal minors, which are de
terminants of : : matrices derived by deleting : ÷: corresponding rows
and columns. A matrix is negative semide…nite if all odd principal minors
(i.e. with : odd) are nonpositive and all even principal minors (i.e. with :
even) are nonnegative.
The …rstorder principal minors are the diagonal entries, which are nega
tive. The secondorder minors have the following form. If row : and column
c are deleted, and i and , are the lowest and highest nondeleted rows, and
/ and  are the lowest and highest nondeleted columns,
`
vc
= :
iI
(j. n) :
)
(j. n)÷:
i
(j. n) :
)I
(j. n) = (÷1)
1(v,c)
n
j
i
j
)
j
I
j

_
1
3
+
j
2
n
_
.
where 1 (:. c) = 1 if : +c is odd, and 1 (:. c) = 0 if : +c is even. Since `
vc
is
negative if and only if : +c is odd, and : = c (i.e. `
vc
is a principal minor)
implies : +c is even, the secondorder principal minors are all positive.
The thirdorder principal minor is the determinant of the matrix, which
we know is always zero for o (j. n). It is worth checking once that, indeed,
[o (j. n)[ = :
11
(j. n) `
11
÷:
12
(j. n) `
12
+:
13
(j. n) `
13
= 0.
66
Hence the matrix is negative semide…nite, which implies that the weak axiom
holds (because all lowerorder principal minors have strict signs, the negative
de…niteness for · ,= cj obtains). These choices do not satisfy the preference
based compensated law of demand, since the substitution matrix fails to be
symmetric.
Exercise 42 (MWG 3.G.14). The following are Walrasian substitution
e¤ects for an agent who has rational preferences and faces prices j = (1. 2. 6):
o (j. n) =
_
_
÷10 ? ?
? ÷4 ?
3 ? ?
_
_
.
Find the missing entries.
Exercise 43 (MWG 2.F.11). Show that, in a twocommodity world, where
r (j. n) is di¤erentiable, homogeneous of degree zero and sats…es Walras’ law,
o (j. n) is always symmetric.
7.5 Integrability
« There is a deeper signi…cance to the observation that the substitution
matrix is symmetric in the derivation from preferencebased demand,
but not in the choicebased approach using the weak axiom. We know
that the weak axiom guarantees that the choice structure is rational
izable (i.e. admits an underlying preference), and is also equivalent to
the compensated law of demand.
« The preferencebased approach is more special in nature, as is evident
from the fact that it implies both the compensated law of demand
and symmetry of the substitution matrix. Hence there is hope that a
demand function that has both properties is rationalizable.
« From a modeling point of view, when we build a theory from a demand
function with certain properties, rather than from preferences, it is
important to know whether the demand function could in fact occur
if the agent were optimizing with respect to some rational preference.
This is known as the integrability problem: are homogeneity of degree
zero, Walras’ law, the compensated law of demand and symmetry of
the substitution matrix su¢cient for rationalizability?
67
« The answer is yes. I.e. choices are rationalizable (given that they sat
isfy homogeneity of degree zero, Walras’ law and the compensated law
of demand) if the substitution matrix is symmetric. (We also presume
that preferences are convex and continuous on in this discussion, so
rationalizability refers here to the existence of a generating preference
that has all the standard properties, not just completeness and transi
tivity. The symmetry of the substitution matrix is also necessary for
rationalizability in terms of such preferences.)
« To understand the connection, we must think about how a preference
would be recovered from choices r (j. n). A preference is fully described
by its upper contour sets
% (r) =
_
c ¸ R
1
+
s.t. c % r
_
=
_
c ¸ R
1
+
s.t. n(c) _ n(r)
_
for some utility function that represents %.
« Equivalently we can express % (r), using the duality between utility
maximization and expenditure minimization, as
%
&(a)
=
_
c ¸ R
1
+
s.t. \j ¸0, c (j. n(r)) _ j c
_
.
This is the set of bundles that are more costly than the cheapest bundle
required to attain n(r), at any given prices. To see why these bundles
form the upper contour set of r, we need insights from abstract duality
theory.
« Duality theory revolves around the idea that any closed, convex set
can be described in terms of its linear approximation by hyperplanes
that separate the set from points not in the set. The sets of interest
to us are the upper contour sets of a continuous and convex preference
relation.
« Fix a utility function n() that represents %, and a level n, and denote
the upper contour set
% (r) =
_
c ¸ 1
1
+
s.t. n(c) _ n(r)
_
by %
&
. A hyperplane that separates the bundle ¸ , ¸%
&
from the set %
&
is parameterized by some vector j and scalar c
&
such that
j ¸ < c
&
_ j r
68
for all r ¸%
&
.
« Given %
&
, the separating hyperplane theorem says that such a hyper
plane (i.e. j and c
&
) exists for every ¸ , ¸%
&
. Each hyperplane separates
R
1
into two halfspaces, one containing the point ¸ and the other con
taining the set %
&
. The intersection of the latter halfspaces is the
original set %
&
, since it excludes all points not in the set.
« For every j, there is an element r
¸%
&
that generates the smallest
value of j r attainable in %
&
. The function that maps j to this value
at r
is called the support function of %
&
:
c
&
(j) = inf ¦j r s.t. r ¸%
&
¦ .
« This function is necessarily concave in j (by the argument we already
made for the concavity of the expenditure function  it increases at
most linearly in j).
« Clearly, we have c
&
(j) _ j r for all r ¸%
&
, at all j. It follows from
the separating hyperplane theorem that there exists some j ¸ 0 for
every ¸ , ¸%
&
such that c
&
(j) j ¸. Hence, no ¸ , ¸%
&
can satisfy
c
&
(j) _ j ¸ for all j ¸0. Then %
&
can be expressed as
%
&
=
_
r ¸ R
1
s.t. \j ¸0. c
&
(j) _ j r
_
.
« The upper contour set at n in the expenditure minimization problem
is convex and closed if preferences are convex and continuous. This
set can be approximated by tangent budget hyperplanes that contain
bundles r
in the upper contour set at which j r is minimized, given
prices. By …nding such a budget hyperplane for every j, we trace
the boundary ot the upper contour set. Knowledge of the minimum
expenditure j r
at every price, i.e. the expenditure function (which
is the support function) would allow us to recover the upper contour
set. See Figure 9.
« Duality theory lets us infer the upper contour set at every utility level
n and therefore recover the preference relation, once we have an expen
diture function. What remains is to obtain the expenditure function
from choices.
69
Figure 9: Upper contour set approximated by c (j. n)
« This is the proper integrability problem, the question whether the sys
tem of partial di¤erential equations \
j
c (j. · (j. n)) = /(j. · (j. n)) =
r (j. n) has a solution. The Frobenius theorem gives necessary and
su¢cient conditions for integrability that are, in particular, satis…ed if
the derivative matrix is symmetric.
« As we have seen, the derivative matrix of \
j
c (j. · (j. n)) is the substi
tution matrix o (j. n) = 1
j
/(j. · (j. n)). Hence symmetry is exactly
what is needed for recovering preferences from choices, i.e. for choices
to be rationalizable.
Exercise 44 (MWG 3.H.6). Derive expenditure and utility function from
the Walrasian demand function r
¹
(j. n) = c
¹
n,j
¹
for / = 1. . . . . 1 with
1
¹=1
c
¹
= 1.
Exercise 45 (MWG 3.H.5). How can expenditure and utility function be
recovered from the indirect utility function?
70
8 Aggregation
8.1 Aggregate Demand Function
« In general, aggregate demand is not a function of aggregate wealth, but
rather a correspondence even if individual demands are functions. At
a given level of aggregate wealth, di¤erent wealth distributions lead to
di¤erent choices, individually and in the aggregate. In this lecture, we
consider the special circumstances under which aggregate demand is a
wellde…ned function of aggregate wealth.
« We also ask whether an aggregate demand function has welfare con
tent. Individual demands are outcomes of utility maximization and
therefore represent the best choices available to an agent. Can we say
that aggregate demand re‡ects the best consumption choices available
to society?
« Finally, we consider whether aggregate demand inherits the weak ax
iom from individual demands, which is an important condition for the
uniqueness of general equilibria.
« Let there be 1 consumers i = 1. . . . . 1 with rational preference relations
%
i
and individual wealths n
i
. Their Walrasian demands are denoted
by r
i
(j. n
i
). Then aggregate demand is:
r (j. n
1
. . . . . n
1
) =
1
i=1
r
i
(j. n
i
) .
Exercise 46 (MWG 4.C.11). Let two agents have identical wealths
n
1
= n
2
= n,2, and let their preferences of over bundles of two goods have
utility representations
n
1
(r
11
. r
21
) = r
11
+ 4
_
r
21
and
n
2
(r
12
. r
22
) = 4
_
r
12
+r
22
.
(a) Derive the individual demand functions and the aggregate demand func
tion.
71
(b) Find the individual Slutsky matrices o
i
(j. n,2) for i = 1. 2 and the
aggregate Slutsky matrix o (j. n). (With two goods, the entire matrix is
determined by one element.) Demonstrate that dj o (j. n) dj < 0 for all
dj ,= 0 that are not proportional to j. Does aggregate demand satisfy the
weak axiom?
« In general, aggregate demand depends on prices and all individual
wealth levels. Can we build a theory where only the aggregate wealth
1
i=1
n
i
= n a¤ects aggregate demand? I.e. when is
1
i=1
r
i
(j. n
i
) =
1
i=1
r
i
(j. n
0
i
) for all (n
1
. . . . . n
1
) and (n
0
1
. . . . . n
0
1
) that distribute the
same aggregate wealth n, so that we can write
r (j. n
1
. . . . . n
1
) =
1
i=1
r
i
(j. n
i
) = r
_
j.
1
i=1
n
i
_
?
« Consider a wealth distribution (n
1
. . . . . n
1
) and some di¤erential change
in wealth (dn
1
. . . . . dn
1
) such that
1
i=1
dn
i
= 0. Since dn is a redis
tribution of wealth that does not a¤ect total wealth, it cannot a¤ect
aggregate consumption of any commodity if r is to be a function of
aggregate wealth only. I.e.
1
i=1
Jr
¹i
(j. n
i
)
Jn
i
dn
i
= 0
for commodity / = 1. . . . . 1.
« Because dn might a¤ect only two individuals, and leave everyone’s
wealth unchanged, it must be that the wealth e¤ect for each commodity
must be the same for any two individuals, i.e. \i. , ¸ 1,
Jr
¹i
(j. n
i
)
Jn
i
=
Jr
¹)
(j. n
)
)
Jn
)
for / = 1. . . . . 1.
« Demand functions have this property at any prices and wealth distrib
ution if and only if preferences admit an indirect utility function of the
Gorman form:
·
i
(j. n
i
) = c
i
(j) +/ (j) n
i
.
72
« For the "if" part, we need Roy’s identity.
Proposition. If n represents a continuous, locally nonsatiated and strictly
convex preference on A = R
1
+
, and the indirect utility function · (. ) is
di¤erentiable at ( j. n) ¸0, then
r ( j. n) = ÷
\
j
· ( j. n)
J· ( j. n) ,Jn
.
Proof. Suppose · ( j. n) = n. By duality of the utility maximization and
expenditure minimization problems, \j, · (j. c (j. n)) = n. Hence
\
j
· (j. c (j. n)) +
J· (j. c (j. n))
Jc (j. n)
\
j
c (j. n) = 0.
Evaluating at j = j, and using \
j
c ( j. n) = /( j. n) = r ( j. n), and replacing
c ( j. n) with n, we have
\
j
· ( j. n) +
J· ( j. n)
Jn
r (j. n) = 0
as claimed.
« Substituting the Gorman form of the indirect utility function into Roy’s
identity, we have
r (j. n) = ÷
\
j
· (j. n)
J· (j. n) ,Jn
= ÷
\
j
c
i
(j)
/ (j)
÷
\
j
/ (j)
/ (j)
n
i
.
Di¤erentiating with respect to wealth,
\
&
i
r (j. n) = ÷
\
j
/ (j)
/ (j)
.
which is independent of n
i
. Since the wealth e¤ect is the same for all
individuals and all wealth levels, aggregate demand depends only on
aggregate wealth.
73
« This argument only provides su¢ciency, but the Gorman form is in
fact necessary for the existence of an aggregate demand function.
Example. Preference % is homothetic if r ~ ¸ implies tr ~ t¸ for all
r. ¸ and all t 0. If n is a utility function that represents homothetic
preference %, then n(r) = n(¸) == n(tr) = n(t¸) for all t 0. It follows
that the utility function is homogeneous of degree one, i.e. n(tr) = tn(r).
Furthermore, the expenditure function is homogeneous of degree one in n:
tc (j. n) = t (j r) = j (tr) = c (j. tn) .
Denote the expenditure required to reach n = 1, given prices, by a new
function c (j. 1) =
~
/ (j). Now one can write the expenditure at an arbitrary
utility n as c (j. n) = nc (j. 1) = n
~
/ (j). Since c (j. n) = n and n = · (j. n)
in the corresponding utility maximization problem, we have
· (j. n) = / (j) n
where / (j) = 1,
~
/ (j), for a homothetic preference. Hence the indirect utility
function has the Gorman form.
We can see directly how the Gorman form relates to linear wealth expan
sion paths in this case. Using Roy’s identity,
r (j. n) = ÷
\
j
/ (j)
/ (j)
n.
Hence, at …xed prices, demand increases linearly (and proportionately for
all commodities) in wealth. Thus \
&
r (j. n) = r (j. n) ,n, and the income
elasticity of demand is, for commodity / = 1. . . . . 1,
¹&
(j. n) = ÷
Jr
¹
(j. n)
Jn
n
r (j. n)
= 1.
This means a …xed share of the budget is spent on each commodity, a property
that you know from demand functions associated with CobbDouglas utility
functions, which belong to the class of homothetic preferences.
Example. Preference is quasilinear in good / if r ~ ¸ implies r + cc
I
~
¸+cc
I
, where c 0 and c
I
is a bundle of one unit of commodity / (and zero
74
of any other commodity). I.e. consuming commodity / does not a¤ect how
the agent values other commodities. As a consequence, r
I
(j. n) must enter
additively into the utility function, e.g. n(r) = r
I
(j. n) + , (r
I
(j. n)).
(The notation r
I
(j. n) refers to quantities of all commodities other than
/.) The function , is nonlinear  if other commodities enter additively, then
it is a condition of optimality that only one of these is consumed.
Recall that utility maximization entails the "tangency"
Jn (r) ,Jr
¹
Jn (r) ,Jr
I
=
j
¹
j
I
for all / (at an interior solution). Since
Jn(r)
Jr
I
= 1.
all marginal utilities are constant with respect to consumption.
For all commodities that enter nonlinearly, such a condition …xes the
quantity at a speci…c level. Only commodity / has a constant marginal
utility at all levels of r
I
(j. n), hence only r
I
(j. n) can adjust to a wealth
change at a solution. It follows that
r
I
(j. n) =
1
j
I
_
n ÷
1
¹6=I=1
j
¹
r
¹
(j)
_
=
n
j
I
÷q (j)
(all remaining wealth is spent on commodity /). Letting c (j) = , ( r
I
(j))÷
q (j) and / (j) = 1,j
I
, the indirect utility function therefore has the Gorman
form:
· (j. n) = c (j) ÷/ (j) n.
The wealth e¤ects of quasilinear demands are constant
Jr
I
(j. n)
Jn
=
1
j
I
.
Jr
¹
(j. n)
Jn
= 0
for all / ,= /. Thus, the wealth expansion path is in this case parallel to the
/axis. With homothetic preference, it is a straight line through the origin
(i.e. the zero bundle).
75
« In some cases, there may be a …xed wealth distribution rule (n
1
(j. n) . . . . . n
1
(j. n))
that depends only on prices and aggregate wealth n. Then aggregate
demand can be written as a function of aggregate wealth,
1
i=1
r
i
(j. n
i
(j. n)) =
1
i=1
~ r
i
(j. n) .
without imposing uniformwealth e¤ects. (Recall that we normally need
them because aggregate wealth could be distributed in many ways. If
we know what the distribution is, then multiplicity is not an issue.)
« If wealth e¤ects are nonuniform, and the distribution of wealth is not
…xed, aggregate demand may depend on certain statistics of the wealth
distribution. One statistic is aggregate wealth (i.e. the mean); we may
also have to observe variance and higherorder moments. Then aggre
gate demand may be a function of prices and distributional statistics
(rather than full information linking each preference to a particular
individual wealth).
8.2 Representative Consumer
« Given that we have an aggregate demand function, we ask now whether
it is rationalizable in the sense that there exists a preference based on
which a hypothetical consumer could make these choices on behalf of
the entire population.
« This is a precondition for the aggregate demand function to have some
welfare content. Without it, one cannot talk about improving on a
given aggregate bundle, e.g. by redistributing wealth.
De…nition. The aggregate demand function r (j. n) admits a positive rep
resentative consumer if there exists a preference % such that r (j. n) is the
Walrasian demand function generated by %. I.e. \(j. n) and \r, r ,= r (j. n)
and j r _ n == r (j. n) ~ r.
« To actually compare aggregate demands, we need to specify how we
would evaluate a particular list of individual outcomes. The rule could
76
be utilitarian (adding up individual utilities) or egalitarian (preferring
less variation between individual utilities), or something else.
De…nition. A BergsonSamuelson social welfare function \ : R
1
÷ R
assigns a value to every utility vector (n
1
. . . . . n
1
) for the 1 agents.
« The optimal (feasible) distribution of wealth (n
1
. . . . . n
1
), given a social
welfare function \ (), is that which attains
max
(&
1
,...,&
I
)
\ (·
1
(j. n
1
) . . . . . ·
1
(j. n
1
)) s.t.
1
i=1
n
i
_ n
= · (j. n) .
« The aggregate indirect utility function · (j. n) (given social welfare
function \ ()) arises as follows. For every price vector j and aggre
gate wealth level n, we record the value of \ () for every possible dis
tribution of n among the 1 individuals, who are assumed to maximize
their utility from consumption within their budget sets 1
j,&
1
. . . . . 1
j,&
I
as determined by the distribution and prices. The highest achievable
value of \ () among all distributions, i.e. the maximum at given prices
and aggregate welath, is the indirect utility at (j. n).
Exercise 47 (MWG 4.D.2). Con…rm that · (j. n), thus constructed has
the usual properties of an indirect utility function (i.e. homogeneous of degree
zero, increasing in n, decreasing in j and quasiconvex).
« Suppose \ () is increasing, concave and di¤erentiable, and the distrib
ution function that solves \ () is such that individual wealth n
i
(j. n)
is di¤erentiable in price and homogeneous of degree one for all i.
« It can be shown that · (j. n) is the maximum of a utility function that
represents the preference of a positive representative consumer. I.e. the
Walrasian demand function derived from · (j. n) via Roy’s identity is
the aggregate of the individual demands underlying (·
1
(j. n
1
) . . . . . ·
1
(j. n
1
)).
77
« Since the aggregate demand function attains maximal utility · (j. n)
at all (j. n), it is chosen by the preference underlying · (j. n) (we could
construct a complete utility function from all values of \ ()). Hence
there exists a preference with respect to which the aggregate demand
at every (j. n) is the utilitymaximizing bundle. This means we have
a positive representative consumer.
« In this particular case, the positive representative consumer is moreover
a normative representative consumer. The normative representative
consumer’s preference chooses an aggregate demand function r (j. n)
that is \(j. n) consistent with individual choices r
1
(j. n
1
) . . . . . r
1
(j. n
1
)
at the wealth distribution (n
1
(j. n) . . . . . n
1
(j. n)) that maximizes \ ().
« Note that a positive representative consumer could choose some other
aggregate demand function, which arises from suboptimal distributions
of wealth (according to \ ()). There may well be a preference that
rationalizes it. The normative requirement is much stronger.
De…nition. The aggregate demand function r (j. n) admits a normative
representative consumer with respect to welfare function \ () if there ex
ists a preference % that generates the welfaremaximizing aggregate demand
function r (j. n) (which re‡ects at every (j. n) the wealth distribution that
maximizes \ ()).
Example. Suppose all agents have homothetic preferences and the social
welfare function is
\ (n
1
. . . . . n
1
) =
1
i=1
c
i
ln n
i
with c
i
0 for all i, and
1
i=1
c
i
= 1. What is the aggregate demand
function of a normative representative consumer? The …rstorder conditions
to maximize \ (), subject to
1
i=1
n
i
= n, are, for i = 1. . . . . 1,
J\ (n
1
. . . . . n
1
)
Jn
i
J·
i
(j. n
i
)
Jn
i
= `.
where ` is the Lagrange multiplier and derivatives are evaluated at the
welfaremaximizing wealth distribution (n
1
. . . . . n
1
).
78
Since
J\ (n
1
. . . . . n
1
)
Jn
i
=
c
i
n
i
=
c
i
·
i
(j. n
i
)
and
J·
i
(j. n
i
)
Jn
i
=
·
i
(j. n
i
)
n
i
(follows from · (j. n) = / (j) n as derived above for homothetic preference),
we have c
i
,n
i
on the left side of the …rstorder conditions. Summing n
i
=
c
i
,` over i,
n =
1
i=1
n
i
=
1
i=1
c
i
`
=
1
`
.
So the optimal wealth distribution is n
i
(j. n) = c
i
n for all i. Then the
welfaremaximizing aggregate demand function is
r (j. n) =
1
i=1
r
i
(j. c
i
n)
(remember, the c
i
are given by the social welfare function).
Exercise 48 (MWG 4.D.8). Suppose for any distribution (n
1
. . . . . n
1
) of
n there is a distribution (n
0
1
. . . . . n
0
1
) of n
0
such that ·
i
(j
0
. n
0
i
) ·
i
(j. n
i
) for
all i. Argue that any normative representative consumer must then prefer
(j
0
. n
0
) to (j. n).
Exercise 49 (MWG 4.D.1). Show that · (j. n) can alternatively derived
by solving the problem
max
a
1
,...,a
I
\ (n
1
(r
1
) . . . . . n
1
(r
1
))
s.t. j
1
i=1
r
i
_ n
for (r
1
(j. n
1
(j. n)) . . . . . r
1
(j. n
1
(j. n))), the individual demands at the op
timal wealth distribution rule (n
1
(j. n) . . . . . n
1
(j. n)). Since this formal
ization, where a planner chooses consumption bundles, is equivalent to the
one where the planner chooses wealth levels (and lets consumers make their
consumption decisions in the market), we have a version of the second welfare
theorem.
79
8.3 Failure of the Weak Axiom
« Suppose aggregate demand is a wellde…ned function r (j. n) of prices
and aggregate wealth.
« It is clear that aggregate demand inherits homogeneity of degree zero
and Walras’ law from individual demand functions.
« The de…nition of the weak axiom in the Walrasian demand setting
extends directly from individual to aggregate demands: it says if j
r (j
0
. n
0
) _ n and r (j. n) ,= r (j
0
. n
0
), then j
0
r (j. n) n. (If
aggregate bundle r (j. n) was chosen over r (j
0
. n
0
) in one situation,
then r (j
0
. n
0
) can only be chosen if r (j. n) is unavailable.)
« In general, the weak axiom does not survive aggregation.
Example. Let wealth n = 10 be distributed equally (n
1
= n
2
= 5) and
consider demands
r
1
(j. n
1
) =
_
0.
5
2
_
. r
2
(j. n
1
) = (3. 1)
at prices (j
1
. j
2
) = (1. 2), as well as
r
1
(j
0
. n
2
) =
_
3
2
. 2
_
. r
2
(j
0
. n
2
) = (2. 1)
at prices (j
0
1
. j
0
2
) = (2. 1). for agent 2. (Here, the subscripts refer to individ
uals, not commodities.)
These demands satisfy the weak axiom, since bundle r
1
(j
0
. n,2) costs
more than n
1
= 5 at prices j, and bundle r
2
(j. n,2) is costs more than
n
2
= 5 at prices j
0
. (We have wellde…ned orderings r
1
(j. n
1
) ~
r
1
(j
0
. n
1
)
and r
2
(j
0
. n
2
) ~
r
2
(j. n,2).)
Aggregate demands are
r (j. n) =
_
3.
7
2
_
. r (j
0
. n) =
_
7
2
. 3
_
.
Since both aggregate bundles cost n = 10 at the prices at which they are
chosen and less than 10 when they are not chosen, they violate the weak
80
Figure 10: Failure of the aggregate weak axiom
axiom, since each is available in budget sets 1
j,&
and 1
j
0
,&
, but r (j. n) ,=
r (j
0
. n). (There is no wellde…ned ordering, since r (j. n) ~
r (j
0
. n) on
1
j,&
, and r (j
0
. n) ~
r (j. n) on 1
j
0
,&
.) As can be seen in Figure 10, the
scaleddown aggregate bundles lie inside both individual budget sets (which
are, in this case, scaleddown versions of the aggregate budget sets).
« Even though the weak axiom fails in general, it does hold when indi
vidual demands are always decreasing in prices (for compensated and
uncompensated price changes). We know from individual consumer
theory that this is not a compelling property. It requires (pure) sub
stitution e¤ects to be su¢ciently large to cancel out any income e¤ects
of inferior goods.
De…nition. Individual demand function r
i
(j. n
i
) satis…es the uncompen
sated law of demand (ULD) if \j. j
0
and \n
i
,
(j
0
÷j) (r
i
(j
0
. n
i
) ÷r
i
(j. n
i
)) < 0
when r
i
(j. n
i
) ,= r
i
(j
0
. n
i
). ULD for the aggregate demand function r
i
(j. n
i
)
is the same property without subscripts.
81
« Unlike the weak axiom, ULD does survive aggregation.
Proposition. If every individual Walrasian demand function r
i
(j. n
i
) sat
is…es ULD, then aggregate demand r (j. n) satis…es ULD.
Proof. If r (j. n) ,= r (j
0
. n), then for some i,
(j
0
÷j) (r
i
(j
0
. n
i
) ÷r
i
(j. n
i
)) < 0
(for all other agents, the inequality is nonpositive). Summing over 1, we
have \j. j
0
and \n,
(j
0
÷j) (r (j
0
. n) ÷r (j. n)) < 0.
« The aggregate version of ULD, in combination the other standard prop
erties, implies the weak axiom. Thus, we can give conditions for on
which aggregate demand satis…es the weak axiom, but these conditions
do not have choicetheoretic foundations.
Proposition. If aggregate demand r (j. n
i
) satis…es homogeneity of degree
zero, Walras’ law and ULD, then it satis…es the weak axiom.
Proof. Given two pricewealth pairs (j. n) and (j
0
. n
0
) with r (j. n) ,=
r (j
0
. n), let j r (j
0
. n
0
) _ n. The weak axiom requires j
0
r (j. n) n
0
, so
that r (j. n) is revealed preferred to r (j
0
. n
0
). Let j
00
= (n,n
0
) j
0
and note
that, because j
00
,n = j
0
,n
0
, homogeneity of degree zero implies r (j
00
. n) =
r (j
0
. n
0
). Thus. j
00
r (j
00
. n) = (n,n
0
) (j
0
r (j
0
. n
0
)) = n (by Walras’ law).
From ULD,
(j
00
÷j) (r (j
00
. n) ÷r (j. n)) < 0.
Since j r (j
00
. n) = j r (j
0
. n
0
) _ n and j r (j. n) = n, it is necessary that
j
00
(r (j
00
. n) ÷r (j. n)) = n ÷j
00
r (j. n) < 0.
i.e. j
00
r (j. n) n. Then (substituting for j
00
), j
0
r (j. n) n
0
.
82
« The compensated law of demand holds if the price derivative matrix
of Hicksian demand, 1
j
/
i
(j. n
i
), which is equal to the substitution
matrix o
i
(j. n
i
) of compensated price e¤ects, is negative semide…nite
(and de…nite when pre and postmultiplied by vectors that are not
proportional to j).
« Similarly, choices satisfy ULD if the price derivative matrix of Wal
rasian demand, 1
j
r
i
(j. n
i
), has the same properties.
Exercise 50 (MWG 4.C.1). Show that r
i
(j. n
i
) satis…es ULD only if
1
j
r
i
(j. n
i
) is negative semide…nite, and conversely, if 1
j
r
i
(j. n
i
) is negative
de…nite (except ·
T
1
j
r
i
(j. n
i
) · = 0 when · = cj for some c), then r
i
(j. n
i
)
satis…es ULD.
« Homothetic preferences are a special case where choices respect ULD
(and therefore the weak axiom).
Example. With a homothetic preference, we saw that \
&
i
r
i
(j. n
i
) =
r
i
(j. n
i
) ,n
i
. Rearranging o
i
(j. n
i
) = 1
j
r
i
(j. n
i
) +\
&
i
r
i
(j. n
i
) r
i
(j. n
i
)
T
gives
1
j
r
i
(j. n
i
) = o
i
(j. n
i
) ÷
1
n
i
r
i
(j. n
i
) r
i
(j. n
i
)
T
.
Since ·
T
o
i
(j. n
i
) · _ 0 (strictly if · is not proportional to j) and ·
T
r
i
(j. n
i
) r
i
(j. n
i
)
T
· =
_
·
T
r
i
(j. n
i
)
_
2
_ 0, we have \· ¸ R
1
, ·
T
1
j
r
i
(j. n
i
) · _ 0 (strictly if · is
not proportional to j). This makes 1
j
r
i
(j. n
i
) negative de…nite for every
individual, so that ULD holds in the aggregate and implies the weak axiom.
Exercise 51 (MWG 4.C.6). Verify that the following claim is true in
the case of a homothetic preference %
i
. If %
i
can be represented by a twice
continuously di¤erentiable, concave utility function n
i
().and \r,
÷
r
i
1
2
n
i
(r
i
) r
i
r
i
\n
i
(r
i
)
< 4.
then r
i
(j. n
i
) satis…es ULD.
83
Exercise 52 (MWG 4.C.7). If every individual has the same consumption
function ~ r (j. n), and individual wealth n is distributed on [0. n] with density
nonincreasing in wealth, show that the aggregate demand function
r (j) =
_
&
0
~ r (j. n) dn
satis…es ULD. On the other hand, demonstrate that there are unimodal distri
butions of wealth (where the density function is singlepeaked, …rst increasing
and then decreasing) for which ~ r (j. n) does not satisfy ULD.
9 Expected Utility
9.1 Lotteries
« In this lecture, we introduce risk into the choice framework.
« The set of ` possible outcomes will be denoted by C (for consequences,
e.g. C = A could be the set of consumption bundles). The decision
maker’s choice induces a probability distribution on C: which of the
consequences will be realized is not certain at the time the choice is
made.
De…nition. A simple lottery is a probability distribution 1 = (j
1
. . . . . j
.
) on
the set of consequences C. I.e. j
a
_ 0 for : = 1. . . . . `, and
.
a=1
j
a
= 1.
« A simple lottery is an element of the (` ÷1)dimensional simplex, i.e.
the set
=
_
j ¸ R
.
+
s.t.
.
a=1
j
a
= 1
_
.
(It is not `dimensional because j
.
is determined by j
1
. . . . . j
.1
and
the requirement that probabilities sum to 1. E.g. if ` = 2, the distri
bution can be represented by a point j ¸ [0. 1] in the onedimensional
line space.)
84
Figure 11: Simplex
« Figure 11 depicts such a simplex for three consequences. For each
consequence, there is a vertex. The simplex is drawn with height 1,
and a lottery 1 = (j
1
. j
2
. j
3
) is mapped to the unique point whose
perpendicular distance from the three edges re‡ects the probabilities
of the consequences.
« Speci…cally, the distance of point 1 from the edge opposite 1, along the
perpendicular line labeled j
1
, is equal to the probability of consequence
1. The distance from the edge opposite 2, which is labeled j
2
, is the
probability of consequence 2, etc. This technique takes advantage of
the fact that the lengths of these perpendicular lines from a point to the
edges always sum to 1. Hence the points in the simplex can represent
probability distributions.
De…nition. A compound lottery is a probability distribution (c
1
. . . . . c
1
) on
a set of 1 simple lotteries 1
1
. . . . . 1
1
, where c
I
_ 0 for / = 1. . . . . 1, and
1
I=1
c
I
= 1.
« It is then an easy matter to reduce a compound lottery on 1
1
. . . . . 1
1
,
where 1
I
=
_
j
I
1
. . . . . j
I
.
_
, to a simple lottery 1 = (j
1
. . . . . j
.
) such that
j
a
=
1
I=1
c
I
j
I
a
85
for : = 1. . . . . `. I.e. the probability of consequence : in the reduced
lottery is the result of adding probabilities j
1
a
. . . . . j
1
a
over simple lot
teries 1
1
. . . . . 1
1
, where each j
I
a
is weighted by the probability c
I
that
1
I
is realized in the compund lottery.
« We can write the reduced lottery as a vector sum:
1 = c
1
1
1
+ +c
1
1
1
¸ .
« Our focus will be entirely on simple lotteries, on the premise that the
decision maker views any compound lottery as equivalent to the simple
lottery it reduces to. Denote the set of all simple lotteries by /.
9.2 Preference over Lotteries
« Analogously to the certain setting, we start by endowing the agent
with a rational preference relation that ranks all elements of /, i.e. all
simple lotteries. Then we add further axioms.
« From two lotteries 1 and 1
0
we can obtain a new lottery c1+(1 ÷c) 1
0
,
where the probability of consequence : in c1+(1 ÷c) 1
0
is a weighted
average of :’s probabilities in 1 and 1
0
, and c ¸ [0. 1] is the weight of
1.
De…nition. Preference % on / is mixturecontinuous if \1. 1
0
¸ / and
\1
00
¸ /, the set of mixtures of 1 and 1
0
that are preferred to 1
00
,
¦c ¸ [0. 1] s.t. c1 + (1 ÷c) 1
0
% 1
00
¦ .
is closed, as is the set of mixtures of 1 and 1
0
to which 1 is preferred,
¦c ¸ [0. 1] s.t. 1
00
% c1 + (1 ÷c) 1
00
¦ .
« Mixture continuity is a weaker property than continuity of %(i.e. upper
and lower contour sets of 1
00
are closed). Whereas continuity implies
that all convergent sequences in % (1
00
) have limits in % (1
00
), mixture
continuity only requires this of a restricted set of sequences (those that
can be constructed by mixing two lotteries and varying the weight c).
The same applies, of course, to  (1
00
).
86
« The ‡avor is, however, much the same. Mixture continuity means that
preference between two lotteries is robust to su¢ciently small changes
in their probabilities (in certain directions): if
~
1 is very similar to 1,
then 1 % 1
0
only if
~
1 % 1
0
.
« One can think of some instances where people may violate continuity for
extreme c. (Because their preferences over consequences are essentially
lexicographic.) E.g. most of us would never commit a violent crime
(lottery 1). We might prefer a completely lawabiding life (lottery
1
0
) to smallscale tax evasion (lottery 1
00
), and prefer smallscale tax
evasion to a mostly lawabiding life with a small chance of committing
a violent crime (mixing 1 and 1
0
).
« Such violations sound plausible, but they are often sensitive to framing.
We have trouble imagining very small probabilities and treat them as
if they are substantial. If you ask, "would you risk your life to watch a
sports game?," you are likely to get a di¤erent answer than if you ask,
"would you drive to your friend’s house to watch the game?"
« Continuity plays the same role here as under certainty: it guarantees
that there exists a utility representation l : / ÷ R such that 1 %
1
0
== l (1) _ l (1
0
). Note that it is conventional to use the capital
letter l to distinguish a utility representation for a preference over
lotteries from a utility representation for a preference over consumption
bundles or consequences.
« There is an obvious relationship between consequence : and the de
generate lottery 1
a
that assigns probability 1 to consequence : (and
zero to all others). The utility value of such a degenerate lottery will
be denoted by n
a
= l (1
a
).
De…nition. Preference % on / satis…es independence if \1. 1
0
,
1 % 1
0
== c1 + (1 ÷c) 1
00
% c1
0
+ (1 ÷c) 1
00
\1
00
¸ / and \c ¸ (0. 1).
« The independence axiom says that preference between 1 and 1
0
should
not be a¤ected by rescaling the probabilities in both lotteries in a given
direction.
87
« A more intuitive rationale can be given by thinking of c as a random
ization. Suppose you prefer 1 to 1
0
and are o¤ered compound lotteries
that result in (1) 1 with probability c and 1
00
with probability 1 ÷ c
and (2) 1
0
with probability c and 1
00
with probability 1 ÷ c. If c is
the probability of some state of the world (e.g. outcome of a toin coss),
then whichever state is realized, you will be at least as happy with the
outcome of compound lottery (1). Hence you should prefer (1).
« The independence axiom is the fundamental di¤erence between the
formalizations of lottery preferences and consumption preferences. It
is primarily responsible for the stronger results we will derive in the
present context.
« It is important to understand that the independence axiom only makes
sense in the absence of complementarities, hence it could not be im
posed on preferences over consumption bundles. While the commodi
ties that make up a bundle are consumed together, the consequences of
lotteries are mutually exclusive. I.e. mixing lotteries does not change
the nature of the consequences in any way, only their probabilities.
9.3 Expected Utility Theorem
« What the independence axiom gives us, in conjunction with mixture
continuity, goes beyond the existence of a continuous utility function
that represents lottery preference. It implies that the utility function
is linear in character. This statement is the expected utility theorem.
We will get to the actual theorem in a few steps.
De…nition. Utility function l : / ÷R has the expected utility form if there
exists (n
1
. . . . . n
.
) ¸ R
.
such that \1 = (j
1
. . . . . j
.
) ¸ /,
l (1) =
.
a=1
j
a
n
a
.
« Autility function with the expected utility formis called a von Neumann
Morgenstern (vNM) utility function.
88
« Notice that the de…nition requires (n
1
. . . . . n
.
) ¸ R
.
to be a utility
function for the restriction of the lottery preference to degenerate lot
teries. For if 1 = 1
a
(the degenerate lottery that assigns probability 1
to consequence :), then l (1
a
) = j
a
n
a
= n
a
if l () has the expected
utility form.
Exercise 53 (MWG 6.B.2). If l () represents a preference % on simple
lotteries / and has the expected utility form, demonstrate that % satis…es
the independence axiom.
Proposition. Utility function l : / ÷ R has the expected utility form if
and only if it is linear:
l
_
1
I=1
c
I
1
I
_
=
1
I=1
c
I
l (1
I
)
for any 1 lotteries 1
1
. . . . . 1
1
¸ / and weights c
1
. . . . . c
1
¸ [0. 1] such that
1
I=1
c
I
= 1.
Proof. (If) Any lottery 1 = (j
1
. . . . . j
.
) can be written in terms of
degenerate lotteries 1
1
. . . . . 1
.
as 1 =
.
a=1
j
a
1
a
. If l () has the linearity
property, then
l (1) = l
_
.
a=1
j
a
1
a
_
=
.
a=1
j
a
l (1
a
) =
.
a=1
j
a
n
a
.
so that l () has the expected utility form.
(Only if) Consider the compound lottery that assigns probabilities (c
1
. . . . . c
1
)
to lotteries 1
1
. . . . . 1
1
, where 1
I
=
_
j
I
1
. . . . . j
I
.
_
for / = 1. . . . . 1. If l () has
the expected utility property, then it assigns to the corresponding reduced
lottery
1
I=1
c
I
1
I
=
_
1
I=1
c
I
j
I
1
. . . . .
1
I=1
c
I
j
I
.
_
the value
l
_
1
I=1
c
I
1
I
_
=
.
a=1
_
1
I=1
c
I
j
I
a
_
n
a
=
1
I=1
c
I
.
a=1
j
I
a
n
a
=
1
I=1
c
I
l (1
I
)
(swapping the order of summation is permitted by distributivity). Hence
l () is linear.
89
Figure 12: Indi¤erence curves with the expected utility form
« Linearity of the utility function implies linear indi¤erence curves for
lotteries in the simplex, as in Figure 12. Suppose 1 ~ 1
0
, so that
l (1) = l (1
0
) = cl (1) + (1 ÷c) l (1
0
). If l () is linear, then
cl (1) + (1 ÷c) l (1
0
) = l (c1 + (1 ÷c) 1
0
), i.e. any convex combi
nation of 1 and 1
0
is indi¤erent to 1 and 1
0
.
« A utility function that has the expected utility form is therefore asso
ciated with linear indi¤erence curves.
« According to the expected utility theorem, the independence axiom
essentially implies the expected utility form (the only other property
needed is continuity). The connection is easy to see graphically, since
nonlinear indi¤erence curves violate the independence axiom.
« Consider the left panel of Figure 13. The curved indi¤erence set con
tains 1 and 1
0
, but not 1
00
= c1 + (1 ÷c) 1
0
. However, since 1 ~ 1
0
,
the independence axiom says c1 + (1 ÷c) 1
0
~ 1
0
+ (1 ÷c) 1
0
, i.e.
1
00
~ 1
0
. Such a contradiction arises whenever an indi¤erence curve is
nonlinear.
« Furthermore, the independence axiom requires the indi¤erence curves
to be parallel (as in Figure 11). The right panel of Figure 13 depicts how
nonparallel indi¤erence curves cause a contradiction. Lotteries 1
1
and
1
0
1
yield utility l
1
, and we construct lotteries 1
2
and 1
0
2
by respectively
90
Figure 13: Independence induces linear and parallel indi¤erence curves
mixing 1
1
and 1
0
1
identically with a lottery 1. I.e. 1
2
= c1
1
+(1 ÷c) 1
and 1
0
2
= c1
0
1
+ (1 ÷c) 1 for some c ¸ (0. 1). Since 1
1
~ 1
0
1
, the
independence axiom says c1
1
+(1 ÷c) 1 ~ c1
0
1
+(1 ÷c) 1, i.e. 1
2
~
1
0
2
. But if the indi¤erence line through 1
2
is not parallel to that for 1
1
and 1
0
1
, then it cannot contain 1
0
2
.
« Here is the expected utility theorem.
Proposition. If rational preference % on / satis…es continuity and inde
pendence, then % admits a utility representation that has the expected utility
form.
Proof. I start with a few observations about % that are intuitive, but
should and can be proven formally from the axioms. There exist lotteries
1 and 1 that are respectively preferred to all and none of the 1 ¸ /. If
c. , ¸ [0. 1], then , c == ,1 + (1 ÷,) 1 ~ c1 + (1 ÷c) 1.
For any 1 ¸ / it is possible to …nd a unique c
1
¸ [0. 1] such that
1 ~ c
1
1 + (1 ÷c
1
) 1. (We veri…ed it in constructing a utility function
for a continuous rational preference. This is where the continuity axiom is
needed.) Now we will show that l (1) = c
1
is a utility function for % and
has the expected utility form.
By the de…nition of c
1
, 1 % 1
0
if and only if
1 ~ c
1
1 + (1 ÷c
1
) 1 % c
1
0 1 + (1 ÷c
1
0 ) 1 ~ 1
0
.
91
i.e. if and only if c
1
_ c
1
0 . Therefore, if l (1) = c
1
for all 1 ¸ /, then
1 % 1
0
== l (1) _ l (1
0
), so that l () represents %.
The expected utility form is equivalent to linearity, i.e. l
_
1
I=1
,
I
1
I
_
=
1
I=1
,
I
l (1
I
) for all 1
1
. . . . . 1
1
and all ,
1
. . . . . ,
1
_ 0 such that
1
I=1
,
I
=
1. Since 1
I
~ c
1
k
1+(1 ÷c
1
k
) 1, induction on the independence axiom gives
1 =
1
I=1
,
I
1
I
~
1
I=1
,
I
_
c
1
k
1 + (1 ÷c
1
k
) 1
_
=
1
I=1
,
I
c
1
k
1 +
_
1 ÷
1
I=1
,
I
c
1
k
_
1.
Thus c
1
=
1
I=1
,
I
c
1
k
, so by the construction of l (),
l
_
1
I=1
,
I
1
I
_
=
1
I=1
,
I
c
1
k
=
1
I=1
,
I
l (1
I
) .
Exercise 54 (MWG 6.B.3). If the set of outcomes C is …nite and rational
preference % on / satis…es independence, demonstrate that there are best
and worst lotteries 1 and 1 in /, such that 1 % 1 % 1 for all 1 ¸ /.
« A utility function that represents a preference over consumption bun
dles is unique up to strictly increasing transformation. Hence it re
‡ects the ordinal nature of the preference: there is no signi…cance in
the magnitude [n(r) ÷n(r
0
)[, only in the fact that n(r) _ n(r
0
) or
n(r) _ n(r/). If [n(r) ÷n(r
0
)[ [n(r) ÷n(r
00
)[, we can …nd an
other utility function that represents the same preference, but where
[~ n(r) ÷ ~ n(r
0
)[ < [~ n(r) ÷ ~ n(r
00
)[.
« In contrast, the next result indicates that a vNM expected utility
function is cardinal in nature, because it is unique only up to pos
itive linear transformation, which preserves the (relative) magnitude
of di¤erences. (If [l (1) ÷l (1
0
)[ [l (1) ÷l (1
00
)[, then \, 0,
[,l (1) ÷,l (1
0
)[ [,l (1) ÷,l (1
00
)[.)
92
Proposition. If utility functions l : / ÷ R and
~
l : / ÷ R represent
preference % on /, and l () has the expected utility form, then
~
l : / ÷R
has the expected utility form if and only if it is a positive linear transformation
of l (), i.e. ¬, 0 and ¬¸ ¸ R such that \1 ¸ /,
~
l (1) = ,l (1) +¸.
Proof. (If) If l (1) has the expected utility form and
~
l (1) = ,l (1)+¸,
then
~
l (1) is linear: since 1 =
1
I=1
c
I
1
I
for some set of lotteries 1
1
. . . . . 1
1
and weights c
1
. . . . . c
1
,
~
l (1) =
~
l
_
1
I=1
c
I
1
I
_
= ,l
_
1
I=1
c
I
1
I
_
+¸
= ,
_
1
I=1
c
I
l (1
I
)
_
+¸ =
1
I=1
c
I
(,l (1
I
) +¸) .
where the last two equalities re‡ect linearity of l () and that
1
I=1
c
I
= 1.
Linearity of
~
l () implies the expected utility form.
(Only if) Suppose l () and
~
l () both have the expected utility form
(thus, both are linear). We will construct constants , 0 and ¸ such that
\1 ¸ /,
~
l (1) = ,l (1) + ¸. Unless all lotteries are indi¤erent (in which
case l () and
~
l () are constant functions, so that the property holds), there
are best and worst lotteries 1 and 1 in / such that 1 ~ 1. (Since l () is
continuous, it has a maximizer and minimizer on the probability simplex /,
which is a compact set.)
For every 1 ¸ /, de…ne
`
1
=
l (1) ÷l (1)
l
_
1
_
÷l (1)
.
Rearrangement gives
l (1) = `
1
l
_
1
_
+ (1 ÷`
1
) l (1)
and, because l () is linear, l (1) = l
_
`
1
1 + (1 ÷`
1
) 1
_
. Thus 1 ~ `
1
1+
(1 ÷`
1
) 1.
Because
~
l () represents % and is linear,
~
l (1) =
~
l
_
`
1
1 + (1 ÷`
1
) 1
_
= `
1
~
l
_
1
_
+ (1 ÷`
1
)
~
l (1) = ,l (1) +¸.
where
, =
~
l
_
1
_
÷
~
l (1)
l
_
1
_
÷l (1)
0. ¸ =
~
l (1) ÷,l (1) .
93
Exercise 55 (MWG 6.B.4). A safety agency is looking for an evacuation
criterion for an area that has a 1% probability of ‡ooding. Four things can
happen: (A) No evacuation is necessary, and none is performed. (B) An un
necessary evacuation is performed. (C) A necessary evacuation is performed.
(D) A necessary evacuation is not performed.
The agency is indi¤erent between sure outcome (B) and a scenario where
(A) occurs with probability j ¸ (0. 1) and (D) occurs with probability 1 ÷j.
The agency is also indi¤erent between sure outcome (C) and a scenario where
(A) occurs with probability ¡ ¸ (0. 1) and (D) occurs with probability 1 ÷¡.
Moreover, it prefers (A) to (D). Suppose the expected utility theorem applies.
(a) Construct an expected utility function for the agency.
(b) Compare the following policy criteria: (1) Evacuate in 90% of ‡ood
ing instances, and evacuate unnecessarily in 10% of the instances where no
‡ooding occurs. (2) Evacuate in 95% of ‡ooding instances, and evacuate un
necessarily in 15% of instances where no ‡ooding occurs. Derive probability
distributions over the four outcomes under these criteria and decide, based
on the expected utility function, which criterion the agency prefers.
(Note that this version of the problem corrects two typos in MWG.)
9.4 Paradoxes
« Expected utility (or, rather, the underlying independence axiom) en
tails some speci…c predictions about choices. These predictions fre
quently fail in some famous experiments that have led to various alter
native axiomatizations in order to explain the observed behavior.
« Consider four lotteries that o¤er the following probabilities over three
prizes:
$2.5 million $0.5 million $0 million
1
1
0 1 0
1
0
1
0.10 0.89 0.01
1
2
0 0.11 0.89
1
0
2
0.10 0 0.90
.
The decision maker is asked to compare 1
1
to 1
0
1
and 1
2
to 1
0
2
.
« If is often observed (roughly half the time) that 1
1
~ 1
0
1
(giving up
the chance to win a greater prize to avoid a small risk of zero), while
94
1
0
2
~ 1
2
(accepting a slightly increased risk of zero for the prospect of
a greater prize).
« These preferences are not necessarily irrational, but they are incon
sistent with expected utility theory. Notice that in both cases 1
0
is
obtainable from 1 by taking 0.11 out of the probability of winning $0.5
million and distributing 0.01 to the probability of $0 and 0.10 to the
probability of $2.5 million. Only the initial probabilities di¤er, but ac
cording to the independence axiom they should not a¤ect preference
for the adjustment.
« Explicitly, if 1
1
~ 1
0
1
, then l (1
1
) = n
0.5
_ 0.10n
2.5
+0.89n
0.5
+0.01n
0
,
which implies, after adding 0.89 (n
0
÷n
0.5
) on both sides, 0.11n
0.5
+
0.89n
0
_ 0.10n
2.5
+ 0.90n
0
, i.e. 1
2
~ 1
0
2
. But 1
0
2
~ 1
2
was observed.
This is known as the Allais paradox.
« Perhaps people worry about regrets they might have if a bad outcome
occurs. (I.e. if 1
0
1
results in zero, it is viewed as a loss of $0.5 million
that the agent would have had with 1
1
, and therefore seems somehow
worse than zero. In contrast, if 1
0
2
results in zero, there was no way to
guarantee a nonzero prize by choosing 1
2
.)
Exercise 56 (MWG 6.B.5). The following property is known as the
"betweenness axiom": \1. 1
0
¸ / and \` ¸ (0. 1), if 1 ~ 1
0
, then `1 +
(1 ÷`) 1
0
~ 1. Suppose there are three possible outcomes.
(a) Show that a preference relation on lotteries satis…es independence only
if it satis…es betweenness.
(b) Depict in the simplex that, if the continuity and betweenness axioms
hold, then indi¤erence curves of a lottery preference must be straight lines.
Conversely, depict that straightline indi¤erence curves imply betweenness.
(c) Argue (from a graphic comparison) that betweenness is weaker than
independence.
(d) Draw an indi¤erence map that satis…es betweenness and produces the
choices from the Allais paradox.
« Consider now the following experiment. The alternatives are lotteries
over outcomes: go to Venice, watch a movie about Venice, stay at home.
95
Even though you prefer the outcomes in this order, you may prefer to
randomize between going to Venice and staying at home, rather than
between going to Venice and watching the movie. This would make
sense if you expect that, in case you cannot go to Venice (but were
hoping to), you will no longer enjoy the movie.
« But it clearly violates the independence axiom. This is known as
Machina’s paradox. It reminds us that preference need not be …xed,
but may depend on realizations of events. If you want to think about
it in terms of regrets, the regret here is not over a choice the decision
maker failed to make, but over the outcome "nature" chose (which is
beyond the agent’s control).
« Perhaps the most in‡uential critique of expected utility theory is based
on the Ellsberg paradox. Suppose there are two urns, 1 and H, that
each contain 100 balls of white or black color. The proportion of colors
is known for 1 (there are 49 white and 51 black balls), but not for H.
The decision maker will win a $1. 000 prize if he can pick a ball of a
speci…ed color (i.e. either white or black) from one of the urns. All he
has to do is choose the urn from which to take the ball.
« Many people will always take the ball from 1 in successive experi
ments where …rst a black ball and then a white ball wins. If a black
ball wins, then 1 induces a lottery 1
1
over outcomes $1. 000 and $0
with probabilities 0.51 and 0.49, whereas H induces a lottery H
1
with
unknown probabilities : and 1 ÷ :. If a white ball wins, 1 induces
1
W
= (0.49. 0.51) and H induces H
1
= (1 ÷:. :).
« The problem is that there is no way to assign a probability : (that
a black ball is chosen from H) that could justify choosing 1 in both
experiments. Since the agent wants to maximize expected winnings,
he can prefer 1
1
= (0.51. 0.49) to H
b
only if : < 0.51. But then
1 ÷: 0.49, so that he must prefer H
W
to 1
W
.
« This failure of expected utility theory is fundamental: behavior appears
to be at times inconsistent with the notion that individuals choose be
tween known (or even estimated) probability distributions over out
comes.
96
« As we will see, expected utility theory can be extended to replace the
objective probabilities in the de…nition of lotteries with subjective (im
plicitly believed) probabilities of events. But this does not address the
Ellsberg paradox, which contradicts the existence of unique probabili
ties altogether.
« An important current research area in decision theory is therefore non
expected utility theory, where agents may e.g. have ambiguous beliefs
(allowing for multiple probability distributions), and choices could re
‡ect optimistic or pessimistic expectations. These situations, where
agents do not use simple probabilistic information or beliefs, are said
to be characterized by (Knightian) uncertainty, rather than risk.
Exercise 57 (MWG 6.F.2). In the setting of the Ellsberg paradox, let
n(0) = 0 and n(1000) = 1 represent the decision maker’s preferences over
sure amounts of money. A probabilistic belief that the color of the Hball
is white can be expressed as : ¸ [0. 1]. Suppose, however, that the decision
maker has a set 1 _ [0. 1] of such beliefs. The available actions are 1 and H
(respectively, picking the ball from urn 1 and H). Denote by \ the choice
situation where the $1000 prize is won if the ball is white and $0 otherwise.
In choice situation 1, a black ball wins $1000 dollars, and the decision maker
gets $0 otherwise.
For each choice situation, let the utility function over actions 1 and
H be as follows. In \, l
W
: ¦1. H¦ ÷ R is such that l
W
(1) = 0.49
and l
W
(H) = min ¦: s.t. : ¸ 1¦. In 1, l
1
: ¦1. H¦ ÷ R is such that
l
W
(1) = 0.51 and l
W
(H) = min ¦1 ÷: s.t. : ¸ 1¦. I.e. l
W
(1) is the
expected utility of $1000 given the (objective) probability implied by the
number of white and black balls in urn 1. But l
W
(H) is the expected
utility of $1000 based on the most pessimistic probability in 1. (This is an
instance of Gilboa and Schmeidler’s theory of nonunique prior beliefs.)
(a) Show that if 1 consists of a single belief, then l
W
and l
1
are derived
from a vNM utility function, and l
W
(1) l
W
(H) == l
1
(1) < l
1
(H).
(b) Find a set 1 such that l
W
(1) l
W
(H) and l
1
(1) l
1
(H).
Exercise 58 (MWG 6.B.6). Sometimes, an agent’s preference over lotter
ies depends on a prior action c ¸ ¹ (e.g. when you have to bring wine to
dinner, you would like to know what kind of meat will be served, so you prefer
97
a degenerate lottery). Such a preference has an induced utility representation
l (1) = max
o2¹
.
a=1
j
a
n
a
(c)
for all 1 = (j
1
. . . . . j
.
) ¸ /, where n
a
(c) is the utility assigned to degenerate
lottery 1
a
if action c ¸ ¹ is taken. Show that l () is convex, but (by
example) not necessarily linear.
9.5 StateSpace Approaches
« It is often possible and useful to impose a bit more structure where
we have so far assumed given probabilities. Let o be a set of states
(decisionrelevant situations that may materialize), where the proba
bility :
c
of state : ¸ o is objectively known (we will relax this to
subjective knowledge later).
« The choice objects are now taken to be random variables r : o ÷C, i.e.
functions that determine which consequence will occur in each state.
(These functions are also called acts.)
« Such random variables are closely related to lotteries via the probability
distribution over states. While a lottery assigns probabilities to conse
quences, a random variable, or act, assigns consequences to events that
are characterized by their probabilities. Both types of choice objects
therefore induce a probability distribution over consequences.
« Preference is de…ned on the space of random variables A and, if it
satis…es continuity and a variant of independence called the surething
principle, can be represented by a function that has a modi…ed expected
utility form.
De…nition. Utility function l : A ÷ R has the extended expected util
ity form if there exists \: ¸ o, a function n
c
: C ÷ R such that \r =
(r
1
. . . . . r
S
) ¸ A,
l (r) =
S
c=1
:
c
n
c
(r
c
) .
98
« Comparing this to the expected utility form, l (1) =
.
a=1
j
a
n
a
, the
probability j
a
of consequence : (that is determined by the choice of
lottery 1) is replaced by state probability :
c
. And utility n
a
of the
lottery that is degenerate in consequence : is replaced by utility n
c
(r
c
)
of consequence r
c
(which the act r associates with state :).
« An extended expected utility function represents preferences on A that
satisfy continuity and the surething principle (provided there are at
least three states). Informally, the surething principle says that pref
erence between r and r
0
should be determined on states in which they
disagree.
« To express the surething principle formally, I introduce some new lan
guage. Denote by r
1
the restriction of random variable r on o to the
event 1 _ o, and by r
1
the restriction of r to the complement of 1.
(I.e. r
1
assigns states : ¸ 1 to consequences.)
« Say that r %
1
r
0
if r
1
r
0
1
% r
0
(i.e. r is preferred to r
0
when r
1
is
replaced by r
0
1
, so that r di¤ers from r
0
only on 1). One can read
r %
1
r
0
as "r is preferred to r
0
on 1."
De…nition. Preference % on A satis…es the surething principle if \1 _ o,
r %
1
r
0
== r %
1
r
0
whenever r
1
= r
1
and r
0
1
= r
0
1
.
« The axiom applies when consequences are identical for states outside
the event 1, so that the choice between r and r
0
, respectively r and
r
0
, matters only if 1 occurs. (Hence the name "sure thing"  if what
happens outside 1 cannot be changed, it should not a¤ect preference.)
So if r % r
0
, and r
1
= r
1
, and r
0
1
= r
0
1
, then it should be the case
that r % r
0
.
« The surething principle takes the place of the independence axiom in
the present setting. It is analogous in that it requires preference to be
independent of events on which the acts do not di¤er. (Or, in alterna
tive formulations, events that have zero probability of occurring.)
« So far, we have taken for granted that there are objective probabilities
for the states. It is, of course, rare to have such information. In fact,
99
nothing forces us to interpret the :
1
. . . . . :
S
in the extended expected
utility function as objective probabilities.
« Suppose preference % on A respects continuity and the surething
principle, so that it has an extended expected utility representation
l (r) =
S
c=1
:
c
n
c
(r
c
). This tell us that, for every state : and act r,
the value :
c
n
c
(r
c
) is uniquely determined up to positive linear trans
formation. If :
c
is not given objectively, then it is arbitrary.
« Is there a compelling way to "disentangle" :
c
from n
c
(r
c
), i.e. …x :
c
so
that we can interpret it as the subjective probability the agent assigns
implicitly to state :? If we are willing to require that preference is
stateuniform (depends only on the consequence chosen in state :, but
ranks consequences the same in every state), then n
c
(r
c
) = n(r
c
) for
all : ¸ o, and :
c
is determined through :
c
n(r
c
) up to scaling. If the
:
1
. . . . . :
S
are to be interpreted as probabilities, then
S
c=1
:
c
= 1, so
that they are completely determined.
« The result that a preference on A that satis…es continuity, the sure
thing principle and state uniformity admits a utility representation of
the form l (r) =
S
c=1
:
c
n(r
c
), with unique probabilities :
1
. . . . . :
S
,
is known as the subjective expected utility theorem, due to Savage. It
implies that expected utility theory does not depend on factual knowl
edge of probabilities, but can be built around personal beliefs that are
probabilistic.
10 Risk
10.1 Money Lotteries
« A lottery over continuous amounts of money r ¸ R can be described
most generally in terms of its cumulative distribution function 1 : R ÷
[0. 1]. (The more direct approach would be to use density functions
, (), but these do not always exists and exclude the case of discrete
outcomes. If , () does exist, then 1 (r) =
_
a
1
, (t) dt.)
« We now take the lottery space / to be the set of distribution func
tions on R. The continuous version of the expected utility theorem
100
guarantees that a continuous preference % on /, that satis…es the in
dependence axiom, can be represented by
l (1) =
_
n(r) d1 (r) .
(Note that , (r) = d1 (r) ,dr, so that d1 (r) = , (r) dr, and thus
_
n(r) d1 (r) =
_
n(r) , (r) dr whenever the density exists.)
« The function n(), which records the values of degenerate lotteries, i.e.
certain amounts of money, is called Bernoulli utility function. Assume
that n() is increasing, continuous and bounded. (If it were unbounded,
small probability events could make a lottery in…nitely desirable  the
St. Petersburg paradox.)
Exercise (MWG 6.C.2). Suppose an individual’s Bernoulli utility function
n() has the quadratic form n(r) = ,r
2
+ ¸r. Show that utility from a
distribution is determined by the mean and variance of the distribution, and
only these moments. (No need to do part b.)
10.2 Risk Attitude
« A riskaverse agent is someone who rejects fair gambles (that have
neither an expected gain nor loss).
De…nition. An agent is riskaverse if she prefers the expected money value
of a lottery to the lottery itself: \1 ¸ /,
n
__
rd1 (r)
_
_ l (1) .
(Strictly riskaverse if this is an equality only if 1 () is degenerate.) The
agent is riskneutral if always indi¤erent between the expected value of a
lottery and the lottery itself, i.e. the above is an equality.
« The criterion for risk aversion implies, when l () has the expected
utility form,
n
__
rd1 (r)
_
_
_
n(r) d1 (r) .
101
This is Jensen’s inequality that characterizes a concave function n().
(If you think of the integral over the probability distribution 1 () as a
weighted sum, the inequality relates to the basic de…nition of a concave
function n(cr + (1 ÷c) r
0
) _ cn(r) + (1 ÷c) n(r
0
).)
« Hence the Bernoulli utility function of a riskaverse agent is concave,
and anyone with a concave Bernoulli function is riskaverse. This fact
has a straightforward explanation: the riskaverse agent’s Bernoulli
utility increases more slowly with gain than it decreases with a loss.
Since the agent has, in utility terms, more to lose than gain from a
lottery that is fair in money terms, she declines the lottery unless the
odds are strictly favorable.
« In Figure 14, the expected utility of a fair lottery, i.e. random vari
able A : ¦r
1
. r
2
¦ ÷ [0. 1] with equal probabilities, is labeled l (A).
The Bernoulli utility of certain amount 1 (A) = (1,2) r
1
+ (1,2) r
2
is labeled n(1 (A)). The individual in the left panel is riskaverse,
and l (A) _ n(1 (A)). (Observe how r
1
and r
2
along the raxis are
equidistant from 1 (A), while the corresponding utility value n(1 (A))
is closer to the utility of the better state, n(r
2
).)
« On the right panel, we have the contrasting case of a riskseeker, whose
Bernoulli utility is convex.
De…nition. The certainty equivalent c (1. n) of lottery 1 () is its money
value to an agent whose preference is represented by Bernoulli utility function
n():
n(c (1. n)) =
_
n(r) d1 (r) .
« It is intuitive that a riskaverse person has a certainty equivalent less
than the expected value of the lottery, i.e. she only values the lottery
the same if it produces on average a gain relative to the certainty equiv
alent. Indeed, concavity of the (increasing) Bernoulli utility function
102
Figure 14: Bernoulli utility functions for riskaverter (left) and riskseeker
(right)
implies:
_
n(r) d1 (r) _ n
__
rd1 (r)
_
== n(c (1. n)) _ n
__
rd1 (r)
_
== c (1. n) _
_
rd1 (r) .
Figure 15 shows the certainty equivalent for riskaverse and riskseeking
agents.
« Beyond these formal characterizations, riskaverse behavior is evident
in the propensity to buy insurance, even when premia are not "actuar
ially fair" (i.e. the expected payout is less than the premium).
Example. A strictly risk averse agent with initial wealth n faces a possible
damage of $1 with probability :. The agent is o¤ered insurance at a fair
premium : per dollarpayout in the event of a loss. If c dollars of insurance
103
Figure 15: Certainty equivalent for riskaverter (left) and riskseeker (right)
(i.e. conditional payout) are purchased at this premium, the agent’s wealth
will be either n ÷ :c (if no damage occurs) or n ÷ :c ÷ 1 + c = n +
(1 ÷:) c ÷ 1 (if there is damage). Expected utility from a choice of c,
which induces a lottery over n+(1 ÷:) c÷1 and n÷:c with probabilities
(:. 1 ÷:), is then
l (c) = :n (n + (1 ÷:) c ÷1) + (1 ÷:) n(n ÷:c) .
The …rstorder condition with respect to c,
n
0
(n + (1 ÷:) c
÷1) = n
0
(n ÷:c
) .
can be solved for n + (1 ÷:) c
÷ 1 = n ÷ :c
, i.e. c
= 1 because n
0
()
is strictly decreasing from strict concavity. Thus, a riskaverse agent insures
fully if the premium is actuarially fair.
Example. Suppose an amount of wealth can be invested in a safe asset,
which yields $1, and a risky asset with earnings distribution 1 (.) such that
expected return
_
.d1 (.) is greater than 1. The portfolio choice problem is
to determine the optimal shares c and , to invest in these assets, such that
c + , = 1. Since the random return, given c and ,, is c + ,., utility is a
104
random variable n(c +,.) = n(1 ÷, +,.). An expected utility maximizer
solves:
max
o2[0,1]
_
n(1 +, (. ÷1)) d1 (.) .
which has …rstorder condition, with respect to ,,
_
(. ÷1) n
0
(1 +,
(. ÷1)) d1 (.) = 0
(at an interior solution). Since the left side is greater than zero at , = 0,
given
_
.d1 (.) 1 and that n() is increasing everywhere, we must have
,
0, whether or not the individual is riskaverse. The general principle
is that an agent will always invest some share of wealth in an actuarially
favorable asset.
Exercise (MWG 6.C.19).
« Risk aversion can be quanti…ed and compared by means of the absolute
and relative coe¢cients of risk aversion.
De…nition. The ArrowPratt coe¢cient of absolute risk aversion is, for a
twice di¤erentiable Bernoulli utility function n() at r,
:
¹
(r. n) = ÷
n
00
(r)
n
0
(r)
.
« This is essentially a measure of the concave curvature of the Bernoulli
utility function (for an increasing concave function, n
0
(r) 0 and
n
00
(r) < 0). Note that we cannot compare the degree of concavity based
on ÷n
00
(r) alone, since this derivative can be scaled by a positive linear
transformation (which yields another utility function that represents
the preference). But such a transformation would also scale n
0
(r), so
it cannot a¤ect :
¹
(r. n).
« For a riskneutral agent, n
0
(r) is constant and n
00
(r) = 0, so :
¹
(r. n) =
0.
105
« The more riskaverse of two agents has a lower certainty equivalent
for any given lottery 1 (). Moreover, 2 is more riskaverse than 1 in
the sense that :
¹
(r. n
2
) _ :
¹
(r. n
1
) if and only if n
2
() is a concave
transformation of n
1
(). I.e. there exists an increasing concave function
· () such that \r, n
2
(r) = · (n
1
(r)).
Exercise (MWG 6.C.6, C.7).
Example. It is possible to recover the preference from the ArrowPratt coef
…cient. Suppose :
¹
(r. n) = c for all r. Integrating n
00
(r) = ÷cn
0
(r) on both
sides, we have n
0
(r) ,n(r) = J ln n(r) ,Jr = ÷c, and integrating once more
on both sides, n(r) = c
oa
, i.e. the utility function is exponential when the
ArrowPratt coe¢cient is constant. I have constructed one particular utility
function, assuming the integration constants are zero, but others are still
exponential (solve the di¤erential equation to see this). Exponential utility
functions therefore constitute the CARA (constant absolute risk aversion)
class.
« The DARA class of Bernoulli utility functions has the plausible prop
erty that wealthier people tend to be less riskaverse.
De…nition. Bernoulli utility function n() exhibits decreasing absolute risk
aversion (DARA) if :
¹
(r. n) is a decreasing function of r, i.e. \r, :
¹
(r. n)
:
¹
(r
0
. n) whenever r < r
0
.
Exercise (MWG 6.C.8).
De…nition. The coe¢cient of relative risk aversion is, for a twice di¤eren
tiable Bernoulli utility function n() at r,
:
1
(r. n) = ÷r
n
00
(r)
n
0
(r)
.
106
« If :
1
(r. n) = r:
¹
(r. n) is decreasing in r, then clearly :
¹
(r. n) must
be decreasing in r. The converse is not true. Therefore, nonincreasing
relative risk reversion is a stronger property than decreasing absolute
risk aversion.
Example. Consider the Bernoulli utility function n(r) = r
1j
, (1 ÷j),
where j ¸ (0. 1). Since :
¹
(r. n) = j,r is decreasing in r, this function is in
the DARA class. But :
1
(r. n) = r:
1
(r. n) = j is constant, so DARA does
not imply DRRA.
Exercise (MWG 6.C.18).
Exercise (MWG 6.C.12).
10.3 Stochastic Dominance
« Up to now, we have compared agents in terms of the risk aversion
exhibited by their utility functions. Now we are interested in compar
ing lotteries and …nding criteria by which they can be ranked, given
properties of preference, such as risk attitude.
« If distribution 1 yields a higher expected utility than lottery G, re
gardless of risk attitude (i.e. the speci…c form of the Bernoulli utility
function), 1 is said to …rstorder stochastically dominate G.
De…nition. Distribution 1 () …rstorder stochastically dominates G(), writ
ten 1 %
1OS1
G, if
_
n(r) d1 (r) _
_
n(r) dG(r)
for any nondecreasing function n : R ÷R.
« If distribution 1 yields a higher expected utility than lottery G for a
riskaverse agent (i.e. a concave Bernoulli utility function), then 1 is
said to secondorder stochastically dominate G.
107
De…nition. Distribution 1 () secondorder stochastically dominates G(),
written 1 %
SOS1
G, if 1 () and G() have the same expectation of r, i.e.
_
rd1 (r) =
_
rdG(r), and
_
n(r) d1 (r) _
_
n(r) dG(r)
for any nondecreasing concave function n : R
+
÷R.
« 1 %
1OS1
G is equivalent to the property: for any r, getting more than
r is more likely under 1 than under G.
« Arelated statement can be made for secondorder dominance. 1 %
SOS1
G is equivalent to the property: for any r, probability mass accumulates
faster toward r under G than under 1. (I.e. G spreads the probability
mass more evenly and gives more weight to the extremes. I will give a
precise characterization in a moment.)
« The next results make use of some integral relationships that may be
found through "integration by parts." The technique is based on the
product rule applied to n(r) 1 (r):
d
dr
n(r) 1 (r) = n
0
(r) 1 (r) +n(r) , (r)
implies
n(r) , (r) =
d
dr
n(r) 1 (r) ÷n
0
(r) 1 (r)
and, integrating both sides,
_
n(r) , (r) dr = c ÷
_
n
0
(r) 1 (r) dr.
where c is a constant (since dn(r) 1 (r) ,dr integrated on (÷·. ·) is
n(·) 1 (·) ÷n(÷·) 1 (÷·), and any distribution function satis…es
1 (÷·) = 0 and 1 (·) = 1).
« Applied to n
0
(r)
_
a
1
1 (t) dt, the product rule gives
d
dr
n
0
(r)
_
a
1
1 (t) dt = n
00
(r)
_
a
1
1 (t) dt + 2n
0
(r) 1 (r)
108
(since the derivative of
_
a
1
1 (t) dt with respect to r is
_
a
1
, (t) dt +
1 (r) = 21 (r) by Leibniz’ rule). Thus
n
0
(r) 1 (r) =
1
2
d
dr
n
0
(r)
_
a
1
1 (t) dr ÷
1
2
n
00
(r)
_
a
1
1 (t) dt
and
_
n
0
(r) 1 (r) dr = , ÷
1
2
_ _
n
00
(r)
_
a
1
1 (t) dt
_
dr.
(where , is a constant, since
1
2
n
0
(·)
_
1
1
1 (t) dt =
1
2
n
0
(·)).
« Hence
_
n(r) d1 (r) = c ÷
_
n
0
(r) 1 (r) dr
= c ÷, +
1
2
_ _
n
00
(r)
_
a
1
1 (t) dt
_
dr.
Proposition. The payo¤ distribution 1 () …rstorder stochastically domi
nates G() if and only if \r, 1 (r) _ G(r).
Proof. (If) Let 1 (r) _ G(r) for all r. From integration by parts,
we have
_
n(r) d1 (r) = c ÷
_
n
0
(r) 1 (r) dr and
_
n(r) dG(r) = c ÷
_
n
0
(r) G(r) dr, so
_
n
0
(r) 1 (r) dr _
_
n
0
(r) G(r) dr ==
_
n(r) d1 (r) _
_
n(r) dG(r) .
If n() is an increasing function, the …rst inequality holds, given that 1 (r) _
G(r) for all r. Hence 1 …rstorder dominates G.
(Only if) Suppose 1 ( r) G( r) for some r. To show that 1 fails to
…rstorder dominate G, we need to …nd a nondecreasing function n() such
that
_
n(r) d1 (r) <
_
n(r) dG(r) at some r. Consider n(r) = 1 for r r
and n(r) = 0 for r _ r, which is nondecreasing. Then
_
n(r) d1 (r) =
_
1
a
d1 (r) = 1 ÷1 ( r)
< 1 ÷G( r) =
_
1
a
dG(r) =
_
n(r) d1 (r) .
109
Exercise (MWG 6.D.1).
Exercise (MWG 6.D.2).
Proposition. The payo¤ distribution 1 () secondorder stochastically dom
inates G() if and only if \r,
_
a
1
G(t) dt _
_
a
1
1 (t) dt.
Proof. (If) Let
_
a
1
1 (t) dt _
_
a
1
G(t) dt for all r. From integrating
by parts, we have
_
n(r) d1 (r) = c ÷ , +
1
2
_
_
n
00
(r)
_
a
1
1 (t) dt
_
dr and
_
n(r) dG(r) = c÷, +
1
2
_
_
n
00
(r)
_
a
1
G(t) dt
_
dr. If n() is concave, then
n
00
(r) < 0, so
_
a
1
1 (t) dt _
_
a
1
G(t) dt ==
_
n(r) d1 (r) _
_
n(r) dG(r) .
Hence 1 secondorder dominates G.
(Only if) Suppose
_
a
1
1 (t) dt
_
a
1
G(t) dt for some r. To show that
1 fails to secondorder dominate G, we need to …nd a concave function
n() such that
_
n(t) d1 (t) <
_
n(t) dG(t) at r. Let n(t) = t, except
in an interval [r. r] containing r where
_
a
0
1
1 (t) dt
_
a
0
1
G(t) dt for all
r
0
¸ [r. r]. On this interval, let n(t) be strictly concave.
From integration by parts,
_
n(t) d1 (t)
_
n(t) dG(t) ==
_ _
n
00
(r)
_
a
1
1 (t) dt
_
dr
_ _
n
00
(r)
_
a
1
G(t) dt
_
dr.
Since n
00
(r) = 0 for all r , ¸ [r. r] and n
00
(r) < 0 for all r ¸ [r. r], where
_
a
a
1 (t) dt
_
a
a
G(t) dt, the left inequality cannot hold, so
_
n(r) d1 (r) <
_
n(r) dG(r). This means G ~
SOS1
1, a contradiction.
110
« 1 %
SOS1
G is also equivalent to the property that G is a "mean
preserving spread" of 1. I.e. if G is obtainable from 1 by replac
ing every certain outcome r with a lottery that yields r + ., where .
is a zeromean random variable, distributed according to H
a
(.) with
_
.H
a
(.) = 0.
Example. Consider two lotteries that reward outcomes of rolling a fair die.
1 gives $1 if a number up to 3 is rolled and $2 if the number is greater
than 3. G pays nothing for a 1 and $5 for a 6, and $1 otherwise. These
lotteries have the same mean, 3,2, and probabilities (1,2. 1,2) over ($1. $2),
respectively (1,6. 2,3. 1,6) over ($0. $1. $5). To obtain G from 1, replace
the $1 and $2 wins in 1 with lotteries that give ($0. $1. $5) respectively with
probabilities (1,3. 7,12. 1,12) and (0. 3,4. 1,4). Observe that the expected
values of these lotteries are $1 and $2. The compound lottery over ($0. $1. $5)
that plays (1,3. 7,12. 1,12) and (0. 3,4. 1,4) with equal probability reduces
to (1,6. 2,3. 1,6) = G. So we have constructed G as a meanpreserving
spread of 1, i.e. 1 %
SOS1
G.
Example. Continuing in the previous scenario, distribution H is called an
"elementary increase in risk" from G if it redistributes all probability mass
to the extreme points in G’s domain (while preserving the mean). I.e. H
is the lottery (7,10. 0. 3,10). This is a meanpreserving spread via lotteries
(1. 0. 0), (4,5. 0. 1,5) and (0. 0. 1) in place of the $0, $1 and $5 wins.
Exercise (MWG 6.D.4).
Exercise (MWG 6.D.3).
11 Pro…t Maximization Problem
11.1 Production Set
« Firms exist in order to transform some goods (inputs) into other goods
(outputs). Which goods are inputs and which are outputs is a matter
of choice for every …rm within the technological constraints. Feasible
production plans are described by the production set 1 , which lists all
possible combinations of input and output quantities. These are simply
bundles, or vectors, in the commodity space.
111
« However, unlike consumption bundles, production vectors necessarily
contain negative entries (every technology requires inputs). The pro
duction set can therefore not be restricted to R
+
. A typical element
is (¸
1
. . . . . ¸
1
) ¸ R
1
, where ¸
¹
< 0 identi…es the /th commodity as an
input for the particular …rm, and ¸
¹
0 means / is an output. The set
of all such feasible vectors is the production set 1 _ R
1
.
« It is commonly assumed that 1 is nonempty and closed (i.e. there are
e¢cient production plans), and that there can be "no free lunch" (no
output without input, ¸ ¸ 1 and ¸ _ 0 == ¸ = 0).
« The following are typical properties that may fail to apply in special
circumstances. If the …rm is able to shut down its operations without
"sunk costs," it has the option of inaction (0 ¸ 1 ). If the …rm can
always use more inputs without reducing output, free disposal applies
(¸ ¸ 1 and ¸
0
_ ¸ == ¸
0
¸ 1 ). If it is impossible to fully recover
inputs from outputs, we have irreversibility (¸ ¸ 1 and ¸ ,= 0 ==
÷¸ , ¸ 1 ).
Exercise (MWG 5.B.5).
« Finally, one often assumes some form of convexity, which can be broken
down into additivity and returnstoscale properties. Additivity says
that two feasible production plans can be combined (¸. ¸
0
¸ 1 ==
¸ + ¸
0
¸ 1 ). (Free entry would imply additivity.) With returns to
scale, we have \¸ ¸ 1 , c¸ ¸ 1 for c ¸ [0. 1] (nonincreasing), c _ 1
(nondecreasing) or c _ 0 (constant). (Note these are all versions of
what is conventionally called "constant returns to scale;" they restrict
the scale parameter to di¤erent intervals.)
Proposition. The production set 1 is additive and exhibits nonincreasing
returns to scale if and only if it is a convex cone, i.e. \¸. ¸
0
¸ 1 and \c. , _
0, c¸ +,¸
0
¸ 1 .
Proof. (If) Take any ¸. ¸
0
¸ 1 and c. , 0. Suppose 1 has the additivity
and nonincreasing returns properties. By / times adding ¸, we have /¸ ¸ 1 .
This can be done for any integer /, so let / max ¦c. ,¦. Because c,/ <
112
Figure 16: Typical convex cone production set (shaded)
1, nonincreasing returns implies (c,/) (/¸) = c¸ ¸ 1 . By the analogous
construction, ,¸
0
¸ 1 . Now additivity gives c¸ + ,¸
0
¸ 1 , so that 1 is a
convex cone.
(Only if) With c = 1 and , = 1, the convex cone is seen to satisfy
additivity. Similarly, with , = 0, it satis…es nonincreasing returns. Hence
1 is a convex cone only if 1 has the additivity and nonincreasing returns
properties.
« If 1 is a convex cone, then 1 is convex (¸. ¸
0
¸ 1 and c ¸ [0. 1] ==
c¸ + (1 ÷c) ¸
0
¸ 1 ). The converse is true only if 0 ¸ 1 (inactivity is
possible), so that nonincreasing returns hold (let ¸
0
= 0), but this is
not su¢cient (for additivity).
« We will typically assume that 1 is convex, which is a weaker prop
erty than convex cone, or that 1 is strictly convex (¸ ,= ¸
0
and c ¸
113
(0. 1) == c¸ + (1 ÷c) ¸
0
is in the interior of 1 ). (A convex cone is
not strictly convex.)
Exercise (MWG 5.B.2).
Exercise (MWG 5.B.3).
Exercise (MWG 5.C.8).
11.2 Transformation Function
« The production set can be expressed in terms of a "transformation
function" 1 : R
1
÷ R that assigns values 1 (¸) _ 0 to ¸ ¸ 1 and
1 (¸) 0 to ¸ , ¸ 1 , with 1 (¸) = 0 if and only if ¸ is a boundary
point of 1 . (Such a function could be de…ned for any set, including the
consumption set, which is however easy enough to express as A = R
+
.)
Example. Suppose each of goods ¡
1
and ¡
2
can be made from 1 and
1. Technologies are CobbDouglas: ¡
1
= /
c
1

1c
1
and ¡
2
= /
o
2

1o
2
, where
/
1
+ /
2
= 1 or 
1
+ 
2
= 1. Given /
1
and 
1
, the maximal output of ¡
2
is
¡
2
= (1 ÷/
1
)
o
(1 ÷
1
)
1o
. The (production possibility) frontier is therefore
described by
1 =
_
¡
1

1c
1
_
1¸c
+
_
¡
2
(1 ÷
1
)
1o
_
1¸o
(rearrange the production functions for /
1
and 1 ÷/
1
, add up). A transfor
mation function is:
1 (¡
1
. ¡
2
. ÷/. ÷) = /
1
+/
2
÷1 =
_
¡
1

1c
1
_
1¸c
+
_
¡
2

1o
2
_
1¸o
÷/
1
÷/
2
.
Notice that 1 (¡
1
. ¡
2
. ÷/. ÷) = 0 if and only if / = /
1
+ /
2
= 1 and  =

1
+ 
2
= 1, and else 1 (¡
1
. ¡
2
. ÷/. ÷) < 0 unless /
1
+ /
2
1 or 
1
+ 
2
1
(which are infeasible).
The transformation function contains all relevant information about pro
duction possibilities. You can recognize outputs and inputs by the fact that
an increase in ¡
1
and ¡
2
increases 1 (), and an increase in a / or  decreases
114
1 (). (When inputs are …xed, an increase in output indicates greater e¢
ciency. When outputs are …xed, an increase in input means lower e¢ciency.)
The production plan ( ¡
1
. ¡
2
. ÷1. ÷1), where 1 ( ¡
1
. ¡
2
. ÷1. ÷1) = 0, is fully
e¢cient.
« Suppose 1 () is di¤erentiable at a boundary point ¸. Then, holding ¸
¹
…xed for / = 3. . . . . 1,
d1 ( ¸) =
J1 ( ¸)
J¸
1
d¸
1
+
J1 ( ¸)
J¸
2
d¸
2
= 0.
and the slope of the transformation function is
d¸
2
d¸
1
= ÷
J1 ( ¸) ,J¸
1
J1 ( ¸) ,J¸
2
.
« If ¸
1
and ¸
2
are outputs, this ratio is called the marginal rate of trans
formation (MRT). It measures the amount by which, at …xed input
levels, one output has to be reduced in order to produce more of the
other. If ¸
1
and ¸
2
are intputs, the ratio is the marginal rate of tech
nical substitution (MRTS). Then it measures the amount by which, at
…xed output levels, one input can be reduced when using more of the
other.
Example. In the CobbDouglas case above, where the transformation func
tion was
1 ( ¡
1
. ¡
2
. ÷1. ÷1) =
_
¡
1

1c
1
_
1¸c
+
_
¡
2
(1 ÷
1
)
1o
_
1¸o
÷1.
we …nd
`11
21
=
d¡
2
d¡
1
= ÷
,
c
( ¡
1
,
1
)
(1c)¸c
( ¡
2
, (1 ÷
1
))
(1o)¸o
= ÷
,
c
(/
1
,
1
)
1c
((1 ÷/
1
) , (1 ÷
1
))
1o
and
`11o
11
=
d1
d1
= ÷
,
1 ÷,
_
1 ÷
1
¡
2
_
1¸o
= ÷
,
1 ÷,
1 ÷
1
1 ÷/
1
.
115
11.3 Pro…t Maximization
« The standard behavioral premise about …rms is that they maximize
pro…ts. This standpoint is not as immediately compelling as maxi
mization with respect to consumer preferences. However, if …rms are
owned by consumers, and …xed shares of pro…t accrue to the owners,
higher pro…ts enlarge the owners’ budget sets and therefore increase
indirect utilities.
« However, this argument relies on the "pricetaking" assumption that
each …rm regards j = (j
1
. . . . . j
1
) ¸0 as independent of its production
plan. Else, an owner may …nd it optimal to manipulate prices by
increasing the production of the goods she likes. We do assume price
taking behavior throughout.
« Pro…t maximization also requires that the technology is certain: else, a
riskaverse owner may favor less risky (but potentially less pro…table)
production plans.
« Finally, if …rms are operated by agents instead of owners, they may
pursue other objectives than pro…t maximization, since the pro…t does
not accrue to them.
Exercise (MWG 5.G.1).
« Pro…t, in the conventional sense, is j ¸ =
1
¹=1
j
¹
¸
¹
(recall that inputs
enter negatively, as costs). From a nonempty and closed production
set, the …rm chooses production plan ¸ at j ¸0 that attains
max
j2Y
j ¸ = max
j2R
L
j ¸
s.t. 1 (¸) _ 0.
« Clearly, the …rm must choose a production vector in the boundary of 1 ,
else it is possible to reduce inputs without reducing outputs, which at
strictly positive prices increases pro…t. Hence the constraint specializes
to 1 (¸) = 0.
116
« A solution to the …rm’s problem is the supply correspondence
¸ (j) = ¦¸ ¸ 1 s.t. j ¸ = : (j)¦ .
« We de…ne the pro…t function as : (j) = max
j2Y
j ¸, i.e. it is the value
function associated with the …rm’s maximization problem (analogous
to the indirect utility function and the expenditure function in demand
theory). It depends only on the parameter j; the optimal choice ¸ (j)
is implicit.
« Suppose the transformation function is di¤erentiable. Then the La
grangean …rstorder conditions for pro…tmaximization are, for / =
1. . . . . 1,
`
J1 (¸
)
J¸
¹
= j
¹
.
or in matrix notation, `\1 (¸
) = j. (Since the gradient of the trans
formation function is proportional to j at a solution ¸
, the pro…t
maximizing direction to adjust ¸, if the technological constraint is re
laxed, is in the direction of prices, such that expensive goods are pro
duced using cheap goods.)
« For any pair of commodities /. , the condition implies
J1 (¸
) ,J¸
I
J1 (¸
) ,J¸

=
j
I
j

.
Example. The …rm from the previous examples uses only inputs 1 and 1,
at prices : and n. Thus, a condition for pro…tmaximization is
÷`11o
11
=
,
1 ÷,
1 ÷
1
1 ÷/
1
=
:
n
.
and another is
÷`11
21
=
,
c
(/
1
,
1
)
1c
((1 ÷/
1
) , (1 ÷
1
))
1o
=
j
1
j
2
.
(Additional …rstorder conditions place implicit restrictions on prices.)
117
Solving jointly, we obtain the optimal input ratios in the production of
each output:
i
1
=
/
1

1
=
_
c
,
j
1
j
2
_
1¸(1c)
_
,
1 ÷,
n
:
_
(1o)¸(1c)
.
i
2
=
1 ÷/
1
1 ÷
1
=
,
1 ÷,
n
:
.
The scale of production is not determined, since the technology exhibits
constant returns. We can express the outputs in terms of the levels of one
input
¡
1
= /
c
1

1c
1
= i
1

1
.
¡
2
= (1 ÷/
1
)
o
(1 ÷
1
)
1o
= i
2
(1 ÷
1
) .
or use the identity 
1
= (i
2
1 ÷1) , (i
2
÷i
1
) to express outputs in terms of
the total use of both inputs:
¡
1
=
i
1
i
2
÷i
1
(i
2
1 ÷1) .
¡
2
=
i
2
i
2
÷i
1
(1 ÷i
1
1) .
Since i
1
increases in j
1
, it makes sense that ¡
1
increases, and ¡
2
decreases,
in i
1
.
Exercise (MWG 5.C.9).
Exercise (MWG 5.C.12).
Proposition. The supply correspondence ¸ () for a production set 1 that is
closed and satis…es free disposal is homogeneous of degree zero, convex if 1
is convex, and singlevalued if 1 is strictly convex. The pro…t function : ()
is homogeneous of degree one and convex.
Proof. Since a price increase from j to /j by / 0 does not a¤ect the pro
duction set and can be factored out of j¸, it does not change the maximizers,
i.e. the supply correspondence ¸ (). Then it is clear that : (j) = j ¸
with
¸
¸ ¸ (j) is homogeneous of degree one.
Suppose 1 is convex. If ¸ ¸ ¸ (j) and ¸
0
¸ ¸ (j), then j ¸ = j ¸
0
(since
all elements of the supply correspondence maximize pro…t at j), so
j ¸ = cj ¸ + (1 ÷c) j ¸
0
= j (c¸ + (1 ÷c) ¸
0
) .
118
i.e. c¸ +(1 ÷c) ¸
0
¸ ¸ (j). Moreover, if 1 is strictly convex, then \¸. ¸
0
¸ 1 ,
where ¸ ,= ¸
0
, and \c ¸ (0. 1), the convex combination c¸ + (1 ÷c) ¸
0
is in
the interior of 1 . Then it is possible to reduce inputs and increase outputs
in 1 , which must lead to strictly higher pro…t than j (c¸ + (1 ÷c) ¸
0
).
Therefore, if ¸ ¸ ¸ (j) and ¸
0
¸ ¸ (j), then c¸ + (1 ÷c) ¸
0
, ¸ ¸ (j), which
contradicts convexity of ¸ (j). Hence ¸ (j) cannot contain distinct ¸ and ¸
0
,
and must be singlevalued.
To see that : () is convex, let c ¸ [0. 1] and j. j
0
¸ 0. For any ¸ ¸
¸ (cj + (1 ÷c) j
0
), by de…nition of : () as a maximum, j ¸ _ : (j) and
j
0
¸ _ : (j
0
). Therefore,
: (cj + (1 ÷c) j
0
) = (cj + (1 ÷c) j
0
) ¸ _ c: (j) + (1 ÷c) : (j
0
) .
which means : () is convex.
11.4 Law of Supply
« The law of supply says that more is produced of an output, less used
of an input, whose price increases. It arises from the convexity of the
pro…t function, which is an intuitive property: a price increase causes
a linear increase in pro…t at …xed supply, so the …rm’s ability to adjust
supply can only further enhance pro…t.
Proposition. If the supply function ¸ () for a production set 1 that is
closed and satis…es free disposal is di¤erentiable at j, then ¸ ( j) = \: ( j)
and 1¸ ( j) = 1
2
: ( j) is a symmetric and positive semide…nite matrix, i.e.
\· ¸ R
1
, ·
T
1¸ ( j) · _ 0 (equality if · = j).
Proof. That ¸ ( j) = \: ( j) follows immediately from the envelope theorem
(¸ ( j) is a maximizer at : ( j) = j ¸ ( j), hence locally constant). Positive
semide…niteness of 1
2
: ( j) is also immediate from the fact that : () is con
vex. Since ¸ () is homogeneous of degree zero, ¸ (c j) = ¸ ( j) for c 0.
Di¤erentiating on the left and on the right with respect to j, we have for
/ = 1. . . . . 1,
1
I=1
J¸
¹
(c j)
Jcj
I
Jcj
I
Jc
=
1
I=1
J¸
¹
(c j)
Jj
I
j
I
= 0.
This is the /th entry in the vector 1¸ ( j) j. Therefore, j
T
1¸ ( j) j = 0.
119
« The statement ¸ ( j) = \: ( j), that the supply function can be recov
ered from the (maximal) proft function, is known as Hotelling’s lemma.
« Positive semide…niteness of 1¸ ( j) is an expression of the law of sup
ply and implies, in particular, that ownprice e¤ects J¸
¹
(j) ,Jj
¹
are
positive. Its nondi¤erential equivalent can be understood directly:
(j ÷j
0
) (¸ ÷¸
0
) = j (¸ ÷¸
0
) +j
0
(¸
0
÷¸) _ 0
for all j, j
0
and ¸ ¸ ¸ (j), ¸
0
¸ ¸ (j
0
). This is true because j ¸ _ j ¸
0
and j
0
¸
0
_ j
0
¸ (by the optimality of the supply correspondence).
12 E¢ciency of Aggregate Supply
12.1 E¢cient Production
« The focus of this lecture are the e¢ciency properties of supply and
aggregate supply. First, we look at a basic notion of e¢ciency that is
de…ned from the properties of the production plan alone. Second, we
discuss the stronger notion of cost minimization, which incorporates
prices. Both are essentially implied by pro…t maximization and survive
aggregation.
« E¢cient production is nonwasteful.
De…nition. Production plan ¸ ¸ 1 is e¢cient if there exists no ¸
0
¸ 1 such
that ¸ ,= ¸ and ¸
0
_ ¸.
« While all e¢cient production plans are boundary points of the produc
tion set 1 , not all boundary points are e¢cient. (Consider a natural
resource that cannot be produced. Any feasible production plan that
does not use the resource is a boundary point, but it may be possible
to reduce waste in other commodities.)
120
Proposition. Production plan ¸ ¸ 1 maximizes pro…t at j ¸ 0 only if it
is e¢cient.
Proof. If ¸ is not e¢cient, i.e. there exists ¸
0
_ ¸ in 1 and ¸
0
,= ¸,
then j ¸
0
j ¸ (given strictly positive prices), so that ¸ cannot be pro…t
maximizing.
« This result becomes the …rst welfare theorem under aggregation. The
partial converse corresponds to the second welfare theorem.
Proposition. If the production set 1 is convex, then there exists for every
e¢cient production plan ¸ ¸ 1 a nonzero price vector j _ 0, given which ¸
is pro…tmaximizing on 1 .
Proof. If ¸ ¸ 1 is e¢cient, then 1 must be disjoint from the set
1
j
=
_
¸
0
¸ R
1
+
s.t. ¸
0
¸¸
_
of production plans that contain more of each output and less of each input.
Since 1
j
is convex (if ¸
0
¸ ¸ and ¸
00
¸ ¸, then any convex combination
strictly exceeds ¸), there exists a hyperplane that separates 1 and 1
j
, i.e.
there is some j ,= 0 such that j ¸
0
_ j ¸
00
for any ¸
0
¸ 1
j
and any ¸
00
¸ 1 .
We need to show that j _ 0 and that ¸ is pro…tmaximizing at prices j.
Suppose j
¹
< 0 for some commodity /. Then one can …nd ¸
0
¸¸ with ¸
0
¹
÷¸
¹
su¢ciently large that j ¸
0
< j ¸. But this was ruled out by the construction
of j.
Suppose there is a feasible production plan ¸
00
¸ 1 such that j ¸
00
j ¸.
Then j ¸
00
j ¸
0
for every ¸
0
in some (su¢ciently close) neighborhood of
¸. Because such a neighborhood contains ¸
0
¸¸, the construction of j such
that j ¸
0
_ j ¸
00
is again violated. It follows that j ¸ _ j ¸
00
for any
¸
00
¸ 1 .
Example. Consider a production plan (¡. ÷.) that is e¢cient on the single
output production set 1 given by concave production function , (), i.e.
, (.) = ¡. From the …rstorder conditions, we can construct prices (¡. n)
at which . is pro…tmaximizing. Represent 1 by the transformation function
1 (¡. ÷.) = ¡ ÷ , (.). Fix the output price j, say at j = 1. Then for every
121
/, divide the …rstorder condition `J1 (¡. ÷.
) ,J.
¹
= `J, (.
) ,J.
¹
= n
¹
by
the …rstorder condition for the optimal output ¡, `J1 (¡. ÷.
) ,J¡ = ` =
j = 1: for / = 1. . . . . 1 ÷1,
n
¹
=
J, (.
)
J.
¹
.
This determines the desired price vector at which the e¢cient production
plan (¡. ÷.) is pro…tmaximizing.
« The de…nition of production e¢ciency makes no reference to prices. A
stronger criterion that selects among e¢cient production plans is cost
minimization.
12.2 Cost Minimization
« I focus on the singleoutput case, where a quantity ¡ is produced using
input vector . ¸ R
11
, and the technological contraint is expressed by
a production function , (.). The production function can be viewed as
a transformation function 1 (¡. ÷.) = ¡ ÷ , (.), since \(¡. ÷.) , ¸ 1 ,
¡ , (.) and \(¡. ÷.) ¸ 1 , ¡ _ , (.) with equality if and only if
(¡. ÷.) is in the boundary of 1 .
« Denote by n ¸ 0 the vector of input prices. The costminimizing
choice of inputs, at a given output, de…nes the cost function:
c (n. ¡) = min
:0
n .
s.t. , (.) _ ¡.
« Notice that the optimal choice of inputs is implicit in the cost function;
its value c (n. ¡) is the lowest cost required to produce ¡ at prices n.
« A solution to the cost minimization problem at di¤erent (n. ¡) is the
(conditional) factor demand correspondence . (n. ¡). (The quali…er
"conditional" refers to the fact that the factor demand depends on
output.)
122
« The factor demands at any solution (¡
. ÷.
) to the pro…t maximization
problem must be costminimizing given ¡
, since pro…t at ¡
, i.e. j¡
÷
n ., is strictly decreasing in the cost n .. Hence pro…t maximization
implies not just e¢ciency, but also that factor demand . solves the cost
minimization problem.
« If the production set is convex (i.e. the production function is concave),
the following …rstorder conditions (together with the production func
tion) characterize .
¸ . (n. ¡). Letting the /th commodity be the
output, for all / = 1. . . . . 1 ÷1,
`
J, (.
)
J.
¹
_ n
¹
with equality if .
¹
0. In matrix notation, `\, (.
) _ n.
Exercise (MWG 5.C.10).
« Since ¡ must be the pro…tmaximizing output, given the cost func
tion, i.e. solve max
q
(j¡ ÷c (n. ¡)), the …rstorder condition for pro…t
maximization, j = Jc (n. ¡) ,J¡, must hold. The multiplier ` in the
costminimization problem is the marginal value of relaxing the tech
nological constraint, i.e. ` = Jc (¡. n) ,J¡. Therefore, ` = j whenever
cost is minimized at the pro…tmaximizing output level.
« The direction of the inequality in the …rstorder condition for cost
minimization is then intuitive: whenever `(J, (.
) ,J.
¹
) n
¹
, the
use of input / should be increased, because / adds value j`1
¹
that
is greater than its price n
¹
. Hence .
¹
0 at an optimum (and
`(J, (.
) ,J.
¹
) = n
¹
) unless `(J, (.
) ,J.
¹
) < n
¹
at .
¹
= 0.
« If .
1
. .
2
,= 0, the …rstorder conditions imply `1
1
,`1
2
= n
1
,n
2
and,
since ÷`1
1
,`1
2
is the slope of , (.) = ¡, and ÷n
1
,n
2
is the slope of
c (n. ¡) = c, tangency of the "isoquant" (graph of , (.) = ¡) and the
"isocost" (graph of c (n. ¡) = c). See Figure 17.
« This picture is reminiscent of the expenditure minimization problem if
you think of ¡ (or the implied pro…t j¡) as the …rm owner’s "utility." In
fact, at given prices and costs, the owner’s wealth and indirect utility
· (j. n) are increasing in ¡, so there is a direct connection.
123
Figure 17: Tangency solution to cost minimization
« This parallel implies that c (n. ¡) inherits properties of the expenditure
function (homogeneous of degree one and concave in input prices n,
and nondecreasing in ¡). Moreover, the factor demand correspondence
has properties of the Hicksian demand correspondence for commodity
bundles (homogeneous of degree zero in n, . (n. ¡) = \
&
c (n. ¡) where
di¤erentiable, 1
&
. (n. ¡) is symmetric and negative semide…nite, i.e.
satis…es the law of demand). (The identity . (n. ¡) = \
&
c (n. ¡) is
called Shepard’s lemma.)
Exercise (MWG 5.C.3). Consider a singleoutput technology 1 , with free
disposal, that is is closed and characterized by production function , ().
Show: if , () is homogeneous of degree one (constant returns to scale), then
c () and . () are homogeneous of degree one in ¡. If , () is concave (non
increasing returns to scale), then c () is a convex function of ¡, so that
marginal cost is nondecreasing in ¡.
« The (minimum) cost function is better behaved than the pro…t function
under nondecreasing returns to scale (when it is optimal to produce
in…nite or zero output). Moreover, the cost function always exists,
whereas the pro…t function only exists if …rms are pricetakers.
124
Exercise (MWG 5.D.4).
Exercise (MWG 5.D.5).
12.3 Aggregate Supply
« Suppose now there are J …rms with production sets 1
1
. . . . . 1
J
(each
nonempty, closed and with free disposal). Denote …rm ,’s pro…t func
tion and supply correspondence by :
)
(j) and ¸
)
(j).
« The aggregate supply correspondence is
¸ (j) =
_
¸ ¸ R
1
s.t. ¬¸
1
. . . . . ¸
J
with ¸
)
¸ ¸
)
(j) and
J
)=1
¸
)
= ¸
_
.
It includes every sum of production plans ¸
1
. . . . . ¸
J
that are individu
ally optimal for the …rms.
« The aggregated supply correspondence admits a representative …rm in
the sense that such a …rm, faced with the aggregated production set
1
1
+ +1
J
, would choose a vector in ¸ (j), i.e. an aggregate of plans
that …rms , = 1. . . . . J would choose in the individual production sets
1
1
. . . . . 1
J
.
« Let ¸
(j) denote the supply correspondence on 1
1
+ +1
J
, and call
the associated pro…t function :
(j).
Proposition. If j ¸0, then any ¸ ¸ ¸
(j) satis…es
¸ =
J
)=1
¸
)
for some ¸ such that ¸
)
¸ ¸
)
(j) for , = 1. . . . . J. Moreover, :
(j) =
J
)=1
:
)
(j).
Proof. It is helpful to establish the second part …rst: :
(j) =
J
)=1
:
)
(j).
Let ¸
)
¸ 1
)
for , = 1. . . . . J. Then
J
)=1
¸
)
¸ 1 = 1
1
+ + 1
J
implies
:
(j) _ j
J
)=1
¸
)
=
J
)=1
(j ¸
)
), i.e. :
(j) _
J
)=1
:
)
(j). On the other
125
hand, if ¸ ¸ 1 , there are ¸
)
¸ 1
)
for , = 1. . . . . J such that
J
)=1
¸
)
= ¸. Then
j ¸ = j
J
)=1
¸
)
=
J
)=1
(j ¸
)
) _
J
)=1
:
)
(j), i.e. :
(j) _
J
)=1
:
)
(j).
For the …rst part, we need ¸
(j) _ ¹ =
_
J
)=1
¸
)
s.t. ¸
)
¸ ¸
)
(j) for , = 1. . . . . J
_
and also ¹ _ ¸
(j). Let ¸ ¸ ¸
(j) and ¸ =
J
)=1
¸
)
for some ¸
)
¸ 1
)
, , =
1. . . . . J. Because ¸ is pro…tmaximizing, j
J
)=1
¸
)
= :
(j) =
J
)=1
:
)
(j).
I.e. j ¸
)
< :
)
(j) for some , is only possible if j ¸
I
:
I
(j) for some
/, which con‡icts with the de…nition of :
I
(j) as a maximum value func
tion. It follows that j ¸
)
= :
)
(j), hence ¸
)
¸ ¸
)
(j), for all ,. I.e.
¸
(j) _ ¹. Now let ¸
)
¸ 1
)
, , = 1. . . . . J. Since j ¸
)
= :
)
(j) for all ,,
j
J
)=1
¸
)
=
J
)=1
(j ¸
)
) = :
(j). Thus
J
)=1
¸
)
¸ ¸
(j), i.e. ¹ _ ¸
(j).
« A key implication of this result is that e¢ciency aggregates: since every
…rm produces a given output at minimal cost, and a hypothetical plan
ner, who has the economy’s combined production possibilities available,
can do not better than to replicate individual choices, aggregate supply
must be costminimizing.
« Recall that the "law of supply" (optimal production plans increase in
commodities whose prices increase) applies under mild conditions (1
is closed and has free disposal).
« If every ¸
)
(j) satis…es the law of supply, then ¸ (j) also does. This is
so because symmetry and positive semide…niteness are properties that
are preserved under matrix addition.
« In fact, the lawof supply inequality for …rm,, (j ÷j
0
)(¸
)
(j) ÷¸
)
(j
0
)) _
0, sums over , = 1. . . . . J to (j ÷j
0
) (¸ (j) ÷¸ (j
0
)) _ 0, its aggregate
version.
13 Partial Competitive Equilibrium
13.1 Competitive Equilibrium
« A private ownership economy is populated by consumers with prefer
ences, endowments and ownership stakes in …rms that have production
126
possibilities. We have discussed the consumer problem of choosing a
bundle r ¸ A from the consumption set A, and the producer problem
of choosing a plan ¸ ¸ 1 from the production set 1 . Aggregate de
mand and supply are functions of prices that we took to be exogenous.
Now we consider how prices are determined through the interaction of
demand and supply.
« Let there by 1 consumers i = 1. . . . . 1, J …rms , = 1. . . . . J, and 1
goods / = 1. . . . . 1. Consumer i’s preference over consumption bundles
r
i
¸ A
i
_ R
1
is represented by the utility function n
i
(). Initially, i
is endowed with a bundle .
i
= (.
¹i
. . . . . .
1i
) ¸ A
i
. The total endowed
quantity of commodity / in the economy is .
¹
=
1
i=1
.
¹i
.
« Firm , implements the production plan ¸
)
¸ 1
)
_ R
1
that maximizes
pro…t. Consumer i holds a share o
i)
in …rm ,, which is a proportional
right to the …rm’s net output (i.e. the consumer provides a fraction
o
i)
of the inputs used by …rm , and receives a fraction o
i)
of the out
puts). Given the production decisions, the total available quantity of
commodity / in the economy is .
¹
+
J
)=1
¸
¹)
.
« An allocation describes, in this context, the consumption bundle each
consumer ends up with and the production activity of each …rm.
De…nition. An allocation (r
1
. . . . . r
1
. ¸
1
. . . . . ¸
J
) is a list of consumption
vectors r
i
¸ A
i
_ R
1
for all consumers i ¸ 1 and production vectors ¸
)
¸
1
)
_ R
1
for all …rms , ¸ J.
« An allocation is said to be feasible, given the economy’s endowments
.
1
. . . . . .
1
of commodities / = 1. . . . . 1, if for each /,
1
i=1
r
¹i
_ .
¹
+
J
)=1
¸
¹)
(i.e. no more than the available quantity of each commodity is allocated
for consumption).
127
De…nition. A competitive equilibrium is an allocation (r
1
. . . . . r
1
. ¸
1
. . . . . ¸
J
)
and price vector j
¸ R
1
such that consumers and …rms optimize, and mar
kets clear:
(i) \i ¸ 1, n
i
(r
) = max
a
i
2A
i
n
i
(r
i
) s.t. j
r
i
_ j
.
i
+
J
)=1
o
i)
_
j
¸
)
_
,
(ii) \, ¸ J, j
¸
)
= max
j
j
2Y
j
¦j
¸
)
¦,
(iii) for / = 1. . . . . 1,
1
i=1
r
¹i
= .
¹
+
J
)=1
¸
¹)
.
« It is implicit in the de…nition of competitive equilibrium that consumers
and …rms treat prices as independent of their choices. However, in the
aggregate these choices determine prices through the market clearing
conditions.
« Notice that a proportional change in all prices, from j to cj with c 0,
has no e¤ect on any of the three aspects of competitive equilibrium,
since demand and supply are homogeneous of degree zero. Therefore,
we can arbitrarily assign a price 1 to some commodity (which is then
called the numeraire) without changing the equilibrium allocation.
« Suppose markets clear under an allocation for all commodities except
/. Then, if consumers exhaust their budgets, the market for / must
clear, too. This is apparent from adding up the budget constraints,
1
i=1
(j r
i
) ÷
1
i=1
(j .
i
) ÷
1
i=1
J
)=1
(o
i)
(j ¸
)
))
=
1
i=1
_
j r
i
÷j .
i
÷j
J
)=1
o
i)
¸
)
_
= j
1
i=1
_
r
i
÷.
i
÷
J
)=1
o
i)
¸
)
_
= 0.
(factoring out the price vector is possible by the distributivity of the
dot product).
« Now if markets for commodities / ,= / clear, i.e.
1
i=1
_
r
¹i
÷.
¹i
÷
J
)=1
o
i)
¸
¹)
_
= 0.
128
then
j
I
1
i=1
_
r
Ii
÷.
Ii
÷
J
)=1
o
i)
¸
I)
_
= 0.
which (provided j
I
0) implies that the market for commodity /
clears.
13.2 Partial Equilibrium
« Often we are interested in the market for a particular good / and would
like to treat the remaining markets as "everything else," absorbing
whatever wealth the consumer does not spend on /. Strictly speaking,
that is a quasilinear scenario, but one may argue that it applies ap
proximately whenever only a small portion of the consumer’s wealth is
spent on /.
« The main prerequisites for such a "partial equilibrium" analysis are
that changes in the price of / do not create wealth e¤ects and that
changes in the quantity demanded of / do not cause price adjustments
in other markets. If they did, then the marginal value of residual wealth
would not be …xed, and it is then not clear what it means to measure
the price of a unit of / in terms of what is given up of other goods. If
money existed, the value of a dollar spent on / would not be constant;
it would depend on how much of / is consumed.
« We will assume a quasilinear environment, but note that it may general
ize a bit beyond to situations where demand for / is just not signi…cant
enough to cause nonnegligible wealth e¤ects and price changes else
where. Clearly, this argument is violated if / has close substitutes or
complements.
« Suppose then that all goods other than /, bundled together, enter con
sumer i’s utility function linearly, and denote the quantity of this com
posite commodity by :
i
("money" left over). It can be treated as the
numeraire, i.e. we normalize its price to 1.
« Let the quantity of good / be r
i
, and i’s quasilinear preferences are
represented by
n
i
(:
i
. r
i
) = :
i
+c
i
(r
i
) .
129
where c
i
() is concave, i.e. c
0
i
(r
i
) 0, c
00
i
(r
i
) < 0 at all r
i
_ 0, and
c
i
(0) = 0 (a normalization). The price of / is j.
« Firms require c
)
(¡
)
) of numeraire commodity to produce ¡
)
_ 0 units of
good /. Let c
)
(¡
)
) be convex and twice di¤erentiable, i.e. c
0
i
(r
i
) 0,
c
00
i
(r
i
) _ 0 at all ¡
)
_ 0. (Note that c (0) = 0 only makes sense in the
long run, when there are no …xed costs.)
« Suppose there is no endowment of /, and an endowment .
ni
0 of the
numeraire.
« Firm , sets ¡
)
so that j
¡
)
attains
max
q
j
0
¦j
¡
)
÷c
)
(¡
)
)¦ .
which has …rstorder conditions j
_ c
0
)
_
¡
)
_
, with equality if ¡
)
0.
« Consumer i solves
max
n
i
2R,a
i
2R
+
¦:
i
+c
i
(r
i
)¦
s.t. :
i
+j
r
i
_ .
ni
+
J
)=1
o
i)
_
j
¡
)
÷c
)
_
¡
)
__
= max
n
i
2R,a
i
2R
+
_
c
i
(r
i
) ÷j
r
i
+.
ni
+
J
)=1
o
i)
_
j
¡
)
÷c
)
_
¡
)
__
_
.
since the budget constraint holds with equality at a solution (so that
we can substitute for :
i
).
« The …rstorder conditions, c
0
i
(r
i
) _ j
(equality if r
i
0), determine
the equilibrium j
together with market clearing,
1
i=1
r
i
=
J
)=1
¡
)
.
Note that none of the equilibrium conditions depend on endowments
or ownership stakes. This is a property of quasilinear preferences.
Exercise (MWG 10.C.2).
Exercise (MWG 10.G.2).
Exercise (MWG 10.C.6).
Exercise (MWG 10.G.5). (Only part a.)
130
13.3 The Long Run
« Let there be in…nitely many …rms that could potentially produce com
modity / with the same cost function. Firms are identical, so we may
denote any individual …rm’s output as ¡ (since it is unique if costs are
strictly convex).
« Given enough time, a …rm can attain zero pro…t by shutting down
completely (c (0) = 0). Assuming free entry and exit, …rms will then
enter as long as positive pro…t can be made, and exit if only negative
pro…t can be made. Therefore, all individual …rms must have zero
pro…t in longrun equilibrium, else (if positive) more would enter or (if
negative) more would exit, so some …rms could not be optimizing.
« This leads to the following special version of competitive equilibrium.
De…nition. A longrun competitive equilibrium, given aggregate demand
function r (j) and cost function c (¡) for potentially active …rms (where
c (0) = 0), is a price j
, a per…rm production level ¡
, and a number of
…rms J
such that …rms optimize, pro…table entry is not possible, and mar
kets clear:
(i) j
¡
÷c (¡
) = max
q0
¦j
¡ ÷c (¡)¦,
(ii) j
¡
÷c (¡
) = 0,
(iii) r (j
) = J
¡
.
« Assume strictly positive demand at a price equal to marginal cost,
r (c
0
(0)) 0, and constant returns to scale: c (¡) = c¡ for some c 0.
Then we must have j
_ c, else every …rm would want to produce
in…nitely by (i). Hence max
q0
(j
÷c) ¡ = 0, so that j
= c. Since
r (c
0
(0)) = r (c) 0, market clearing (iii) requires ¡
0, but ¡
is
indeterminate by (ii), and therefore J
is indeterminate.
« There is no longrun equilibrium with increasing and strictly convex
costs. In this case, if j c
0
(0), …rms make a positive pro…t at some
level of production, so total supply is in…nite. Therefore j _ c
0
(0), but
then supply is zero and demand is positive by assumption, r (c
0
(0)) 0,
i.e. market clearing fails.
131
« To get a determinate longrun number of …rms, there must exists an
e¢cient scale ¡ that minimizes longrun average cost c = c ( ¡) , ¡. If
j c, output is in…nite. If j < c, then no nonzero output level is
pro…table (since c is the minimized average cost). So we must have
j = c, and J = r ( c) , ¡.
Exercise (MWG10.F.2). Longrun and shortrun competitive equilibrium
price
Exercise (MWG 10.F.3). Tax impact in short run and long run
Exercise (MWG 10.F.6). Shortrun and longrun supply function with
…xed factor
14 Welfare Analysis
14.1 Pareto E¢ciency and Surplus
« We now consider the welfare properties of competitive equilibrium.
De…nition. A feasible allocation (r
1
. . . . . r
1
. ¡
1
. . . . . ¡
J
) is Pareto e¢cient
with respect to preferences represented by utility functions n
1
() . . . . . n
1
()
if there exists no other feasible allocation (r
0
1
. . . . . r
0
1
. ¡
0
1
. . . . . ¡
0
J
) such that
n
i
(r
0
i
) _ n
i
(r
i
) for all i = 1. . . . . 1 (and the inequality is strict for someone).
« This is a minimal welfare criterion: there is no distributional fairness
implied (allocating everything to one person is Pareto e¢cient). How
ever, it makes sense to at least require that not everyone can be made
better o¤.
« Consider a partial equilibrium setting where individual utility functions
are quasiconcave, i.e.
n
i
(:. r) = :
i
+c
i
(r
i
)
for i = 1. . . . . 1. As in the previous lecture, :
i
is i’s consumption of a
numeraire commodity, which is a composite of all goods other than /.
Commodity / is consumed by individual i in quantity r
i
and produced
by …rm , in quantity ¡
)
at cost c
)
(¡
)
).
132
« In the quasilinear setting, there is a convenient welfare test. An al
location is Pareto e¢cient if and only if it maximizes the aggregate
surplus
o (r
1
. . . . . r
1
. ¡
1
. . . . . ¡
J
) =
1
i=1
c
i
(r
i
) ÷
J
)=1
c
)
(¡
)
) .
« To understand the connection, consider the utility possibility set:
l =
_
(n
1
. . . . . n
1
) ¸ R
1
s.t. ¬ feasible allocation (r
1
. . . . . r
1
. ¸
1
. . . . . ¸
J
)
with n
i
_ n
i
(r
i
) for i = 1. . . . . 1
_
.
i.e. the set of all attainable "utility allocations" n
1
. . . . . n
1
.
« If individual consumption of good / and production plans are …xed at
( r
1
. . . . . r
1
. ¡
1
. . . . . ¡
J
), then .
n
÷
J
)=1
c
)
( ¡
)
) is available to be spent
on the numeraire good. Since the total utility from consuming / is
1
i=1
c
i
( r
i
), the feasible utility constraint is
1
i=1
n
i
_ .
n
+
1
i=1
c
i
( r
i
) ÷
J
)=1
c
)
( ¡
)
)
= .
n
+o ( r
1
. . . . . r
1
. ¡
1
. . . . . ¡
J
) .
« This means, the boundary of the utility possibility set is linear in the
consumption of the numeraire (with slope ÷1). Altering production or
consumption of good 1 leads to parallel shifts
« Optimal consumption and production levels for good / are those where
the boundary of the utility possibility set is as far out as possible. Then
Paretooptimal allocations can only di¤er in the distribution of the
numeraire among consumers (given the c
i
() are strictly concave and
c
)
() strictly convex, their levels are uniquely determined at optimal
consumption and production choices)
« An increase in aggregate surplus (in the quasilinear setting) expands
the utility possibility set and is therefore welfareimproving with respect
to any other reasonable measure (that respects the Pareto principle).
Hence, in the quasilinear environment, Paretooptimal and surplus
maximizing is equivalent.
133
14.2 E¢ciency of Competitive Equilibrium
« Recall that, in the competitive equilibrium problem, j = c
0
)
(¡
)
) for all
,, and j = c
0
i
(r
i
) for all i at a solution. De…ning the inverse demand
function as r
1
() such that r
1
(r (j)) = j, and the inverse supply
function as ¡
1
() such that ¡
1
(¡ (j)) = j, we have ¡
1
(¡ (j)) = j =
c
0
)
(¡
)
).
« Note that marginal cost and marginal utility, at a solution, are the
same across all consumers and …rms (since they are both equal to the
constant j). Hence we can replace c
0
)
() with the "industry marginal
cost" C
0
(), and r
1
(r (j)) with the price 1 () at which a particular
quantity is demanded. Then c
0
i
( r
i
) = 1 (r) for every i, and c
0
)
( ¡
)
) =
C
0
(¡) for every ,.
« Consider a di¤erential change in aggregate surplus:
do =
1
i=1
c
0
i
(r
i
) dr
i
÷
J
)=1
c
0
)
(¡
)
) d¡
)
= 1 (r)
1
i=1
dr
i
÷C
0
(¡)
J
)=1
d¡
)
= (1 (r) ÷C
0
(r)) dr
(Because 1 (r) and C
0
(r) are constants at a competitive equilibrium,
they do not vary with individual or …rm. Apply feasibility ¡ = r and
1
i=1
dr
i
= dr = d¡ =
J
)=1
d¡
)
in the last equality.)
« Integrating over quantity consumed of /, we get
o (r) = o
0
+
_
a
c=0
(1 (:) ÷C
0
(:)) d:.
which is graphically the area between demand and supply curve below
r, the standard depiction of aggregate surplus. (o
0
equals aggregate
surplus when / is not consumed, i.e. in this case the endowment of the
numeraire.)
« It is clear from the expression that aggregate surplus is maximized at r
such that 1 (r) = C
0
(r), i.e. the competitive equilibrium consumption
level r where price equals marginal cost of /.
134
« Note that aggregate surplus is the area between the inverse demand
function 1 (r) = r
1
(r (j)) = j = c
0
i
(r
i
) and the inverse supply func
tion C
0
(¡) = ¡
1
(¡ (j)) = j = c
0
)
(¡
)
). I.e. the inverse demand function
indicates marginal utility of consumption, and the inverse supply func
tion indicates marginal cost of production. (In a wellde…ned sense,
since these are equalized across individuals in competitive equilibrium.)
« At the competitive equilibrium solution, we therefore have 1 (r) =
C
0
(r) == c
0
i
(r
i
) = c
0
)
(¡
)
). I.e. the marginal utility of another unit
of / equals its opportunity cost in terms of the numeraire. (All this
presumes consumers and …rms are price takers.)
« Thus, competitive equilibrium allocation is Paretooptimal. This is an
instance of the …rst welfare theorem, which says that a competitive
equilibrium allocation is e¢cient.
Exercise (MWG 10.D.3).
Exercise (MWG 10.D.4).
14.3 E¢cient Allocations through the Market Mecha
nism
« Since the competitive equilibrium is surplusmaximizing in the quasi
linear case, it can be thought of as the solution to the problem
max
(a
1
,...,a
I
)0
(q
1
,...,q
J
)0
_
.
n
+
1
i=1
c
i
(r
i
) ÷
J
)=1
c
)
(¡
)
)
_
s.t.
1
i=1
r
i
=
1
i=1
¡
)
.
« Firstorder conditions are: for i = 1. . . . . 1,
c
0
i
(r
i
) _ j
and for , = 1. . . . . J,
÷c
0
)
(¡
)
) _ j.
(With equality if r
i
0, respectively ¡
)
0.) And the feasibility
constraint.
135
« To equate these to …rstorder conditions of the competitive equilibrium
problem, must have j = j. I.e. the shadow price j of the resource
constraint for good / is its price j, i.e. price re‡ects the marginal
social value of the good (also implies, through …rstorder conditions,
c
0
i
(r
i
) = j = c
0
)
(¡
)
), marginal utility from consumption equals mar
ginal cost). Then the problem jointly solves for a competitive equilib
rium and Pareto e¢cient allocation.
« Because the amount of the numeraire commodity .
n
can be allocated
in any conceivable way among individuals without a¤ecting the solution
(but it does, of course, a¤ect individual utility), every possible utility
distribution can be implemented and remains Pareto optimal. This the
second welfare theorem.
« The second welfare theorem says that any Paretoe¢cient allocation is
a competitive equilibrium allocation for some speci…cation of zerosum
transfers, i.e. 1
1
. . . . . 1
1
such that
1
i=1
1
i
= 0, between individual
endowments of the numeraire commodity.
Exercise (MWG 10.C.3).
Exercise (MWG 10.C.4).
15 Externalities
15.1 Externalities
« Up until now, we have assumed that individuals care only about their
own consumption  they are indi¤erent to the consumption choices of
others or production choices of …rms, except insofar as these a¤ect one’s
own budget set. One can, however, think of many examples where
other people’s actions directly impact one’s personal welfare (tra¢c
congestion, noise, pollution, to name a few). These e¤ects are called
externalities.
De…nition. An externality is a welfare e¤ect one agent’s actions have on
another.
136
« To get a sense of the issue, consider initially a bilateral externality
that is imposed by one person on another (as opposed to a multilateral
externality that a¤ects many others). The two concerned individuals
i = 1. 2 are a small part of the economy, i.e. their choices do not a¤ect
prices j ¸ R
1
. Their wealths, given these prices, are n
1
and n
2
.
« Besides having preferences over their consumption of goods r
i
= (r
1i
. . . . . r
1i
),
both individuals also care about an action / ¸ R
+
taken by 1. Thus, 1
imposes a (positive or negative) externality on 2.
« Since we wish to focus on the role of the externality, it is convenient to
de…ne a "derived" preference over levels of /, given that consumption
of the 1 goods is optimized. The associated utility function is
·
i
(j. n
i
. /) = max
a
i
0
n
i
(r
i
. /)
s.t. j r
i
_ n
i
.
« Let preferences be quasilinear with respect to a numeraire (good 1), so
that
n
i
(r
i
. /) = q
i
(r
1i
. /) +r
1i
(where r
1i
= (r
22
. . . . . r
12
. /) denotes consumption of goods 2. . . . . 1).
Evaluating at the optimal demands r
1i
(j. /), which are independent
of wealth, we have
·
i
(j. n
i
. /) = q
i
(r
1i
(j. /) . /) +n
i
÷j
1
r
1i
(j. /)
= c
i
(j. /) +n
i
where c
i
(j. /) = q
i
(r
1i
(j. /) . /) ÷j
1
r
1i
(j. /).
« Since commodity prices j are …xed throughout the discussion, we can
treat c
i
(. ) as a function of / only. Let c
i
() be strictly concave (i.e.
c
00
i
() < 0), and normalize to c
i
(0) = 0.
« We should, however, note that concavity may well be violated in prac
tice. For example, if a …rm is able to shut down operations when a
su¢ciently high negative externality is imposed on it, then its pro…t
function cannot be forever decreasing in /, making it convex some
where. Or if a …rm’s constantreturnstoscale production function is
a¤ected by a positive externality (which acts like another input), then
the …rm e¤ectively operates an increasing returns technology and there
fore has a convex pro…t function.
137
15.2 Ine¢ciency
« Denote by /
the equilibrium level of action /, which must be optimal
for 1, and by /
the (possibly di¤erent) socially optimal level. Re
call that, in the quasilinear setting, the socially optimal allocation is
unambiguously the one that maximizes aggregate surplus
o (/) = c
1
(/) +c
2
(/) +n
1
+n
2
.
(Pareto e¢ciency holds if and only if surplus is maximized).
« The …rstorder condition that the socially optimal externality /
must
satisfy is therefore
c
0
1
(/
) +c
0
2
(/
) _ 0.
with equality if /
0.
« However, 1’s choice of / maximizes 1’s derived utility function, which
has the KuhnTucker …rstorder condition
c
0
1
(/
) _ 0.
with equality if /
0.
« Suppose /
0, so that 1 in fact imposes the externality on 2, and say
that /
0 (otherwise, the externality is obviously ine¢cient). If / is
a negative externality, i.e. c
0
2
() < 0 everywhere, then /
= /
only if
c
0
1
(/
) = ÷c
0
2
(/
) 0.
Since c
0
1
(/) was assumed to be decreasing in / (by concavity), and
c
0
1
(/
) 0 = c
0
1
(/
), we must have /
< /
(too much of the exter
nality is imposed).
« If / is a positive externality, i.e. c
0
2
() 0 everywhere, then /
= /
only if
c
0
1
(/
) = ÷c
0
2
(/
) < 0.
Again, since c
0
1
(/) is decreasing in / and c
0
1
(/
) < c
0
1
(/
), it follows
that /
/
(too little of the externality is imposed).
Exercise (MWG 11.D.1).
138
15.3 Remedies
« One way to eliminate the ine¢ciency is by quota, i.e. by mandating
agent 1 to provide no more or no less than the socially e¢cient level
/
, depending on whether the externality is negative or positive.
« Another option is to tax or subsidize the externality. The optimal tax
(subsidy) has to achieve /
= /
, i.e. c
0
1
(/
) = ÷c
0
2
(/
). Since
the tax subtracts from what agent 1 is able to spend on the numeraire
commodity, it leaves derived utility
·
1
(j. n
1
. /) = c
1
(/) ÷t/ +n
1
.
which is maximized by / such that
c
0
1
(/
) = t
(given that /
0, so that the government wants to induce /
0).
« One can see that /
= /
if
t = ÷c
0
2
(j. /
) .
The interpretation is that 1 is made to bear the cost (receives the
bene…t) bestowed on 2. Such corrective taxes are called Pigouvian
taxes.
Exercise (MWG 11.D.2).
Exercise (MWG 11.B.4).
« Bargaining can also resolve the externality problem. Say, 1 has the
right to choose any level of /, but can sign a contract with 2 to set / to
a speci…c level in return for compensation. If 2 o¤ers a payment of 1
for the implementation of
/, that is su¢cient for 1, then 2 maximizes
derived utility
·
2
_
j. n
2
.
/
_
= c
2
_
/
_
÷1 +n
2
s.t. c
1
_
/
_
+1 +n
1
= c
1
(/
) +n
1
= c
2
_
/
_
÷
_
c
1
(/
) ÷c
1
_
/
__
+n
2
.
after substituting 1 = c
1
(/
) ÷c
1
_
/
_
from the constraint.
139
« Because c
0
1
(/
) = 0 (if /
0), the …rstorder condition is
c
0
2
_
/
_
= ÷c
0
1
_
/
_
.
hence
/ = /
.
« Alternatively, if 2 has the right to choose /, then 2 can let 1 purchase
the right to implement
/ at a price 1. The optimal o¤er maximizes
·
2
_
j. n
2
.
/
_
= c
2
_
/
_
+1 +n
2
s.t. c
1
_
/
_
÷1 +n
1
= c
1
(0) + n
1
= c
2
_
/
_
÷
_
c
1
(0) ÷c
1
_
/
__
+n
2
.
which is identical to the case where 1 owns the right to the externality
(since c
1
(0) = 0) and has the same solution.
« While it does not matter who has the right to externality, it is crucial
that the right is allocated to someone in advance and tradable. This
arguments is known as the "Coase theorem." In the case of …rms, the
transfer of rights may occur through a merger or acquisition, which
creates value by internalizing the externality.
Exercise (MWG 11.B.1).
« Related is the "missing markets" view of externalities. Instead of bar
gaining over the amount of the externality, a market for units of the
externality would also …x the problem. An externality is then simply
a commodity, and ine¢ciency arises when it is not traded (the market
is missing).
« We can generalize to a model where J …rms produce amounts /
1
. . . . . /
J
of the externality and 1 individuals experience amounts /
1
. . . . . /
1
of it.
(To connect with the previous twoperson environment, you can think
of 1 as a …rm and 2 as a person.)
« If …rms must buy the right to impose a unit of the externality on
someone at a price j
I
, then …rm , demands /
)
units that maximize its
pro…t
:
)
(j. /
)
) = ,
)
(/
)
) ÷j
I
/
)
.
140
where ,
)
(/
)
) is ,’s (concave) production function for the numeraire,
which requires /
)
as an input. (This setup re‡ects the interpretation
that the externality is negative. If it is positive, think of ,
)
(/
)
) < 0
as a convex cost function, and of j
I
< 0, i.e. …rms must be paid to
produce the externality. Conclusions are the same.)
« Consumer i supplies /
i
units of the externality, maximizing
·
i
(j. n
i
. /
i
) = c
i
(/
i
) +j
I
/
i
+n
i
.
Firstorder conditions imply
,
0
)
(/
)
) = j
I
= ÷c
0
i
(/
i
)
for …rms that produce, and individuals who consume, a positive amount
of the externality. /
)
0, respectively /
i
0. The market price of the
externality turns out to be the Pigouvian tax.
« Socially optimal provision of the externality maximizes aggregate sur
plus
o (/) =
1
i=1
(c
i
(/
i
) +n
i
) +
J
)=1
,
)
(/
)
) .
(Note that transfers
1
i=1
j
I
/
i
=
J
)=1
j
I
/
)
drop out of the social
objective due to market clearing.)
« Firstorder conditions for
max
I
1
,...,I
I
0
I
1
,...,I
J
0
o (/) s.t.
1
i=1
/
i
=
J
)=1
/
)
give
,
0
)
_
j. /
)
_
= j = ÷c
0
i
(j. /
i
)
for every i and , such that /
i
0 and /
)
0. This implies /
)
= /
)
,
i.e. the market provision of the externality is Paretoe¢cient.
Exercise (MWG 11.D.5). (Assume , (0) = 0.)
141
15.4 Public Goods
« The market solution relied on the depletable nature of the externality:
a …rm could directly sell a unit to a consumer without a¤ecting others.
This is a reasonable assumption for some types of externalities (say,
construction noise next to a single house), but not for others (for ex
ample, air pollution). The latter type of externality is called a public
good (or public bad, as the case may be).
De…nition. A public good is a nonrivalrous (i.e. nondepletable) commod
ity: it can be consumed simultaneously by all agents.
« Suppose …rms , = 1. . . . . J produce quantities /
1
. . . . . /
J
of a public
good at convex cost c
)
(/
)
) in terms of the numeraire. They sell units of
it to individuals at price j
I
. Individuals i = 1. . . . . 1 demand quantities
/
1
. . . . . /
1
, but in fact consume the entire amount of the public good
that …rms supply, i.e.
J
)=1
/
)
=
1
i=1
/
i
= /.
« Then
·
i
(j. n
i
. /
i
) = c
i
(/) ÷j
I
/
i
+n
i
.
and
:
)
(/
)
) = j
I
/
)
÷c
)
(/
)
) .
« The aggregate surplus is
o (/) =
1
i=1
(c
i
(/) +n
i
) ÷c (/) .
where c (/) =
J
)=1
c
)
(/
)
) is the total cost of providing amount / of
the public good (payments
1
i=1
j
I
/
i
=
J
)=1
j
I
/
)
canceled out).
« Surplus is maximized (Pareto e¢ciency attained) only if
1
i=1
c
0
i
(/
) = c
0
(/
)
(provided /
0). This is the Samuelson condition for the optimal
provision of a public good. (It says that the sum of marginal rates
of substitution between a public and private good, in this case the
numeraire, should equal the marginal rate of transformation.)
142
« If the public good is excludable, then government can achieve e¢ciency
by imposing personal prices (j
1
. . . . . j
1
) on individuals for a unit of the
public good: socalled Lindahl prices j
i
= c
0
i
(j. /
) per unit of the
public good consumed, for i = 1. . . . . 1.
« These prices are …rstorder conditions for consumers, hence all individ
uals optimize by demanding /
units. A single …rm (or a consortium
of J …rms) maximizes pro…t : (/) =
_
1
i=1
j
i
_
/ ÷ c (/) by setting
1
i=1
j
i
= c
0
(/
). Hence
1
i=1
c
0
i
(/
) = c
0
(/
), which implies /
= /
.
« The classic (or pure) public good is, however, nonexcludable in ad
dition to being nonrivalrous (i.e. nondepletable). Then individual
i can choose to pay for amount /
i
of the public good, but consume
/ =
1
i=1
/
i
. If the good is sold in the market at price j
I
, the …rst
order conditions for individuals and …rms are
c
0
i
(/
) ÷j
I
_ 0. j
I
÷c
0
)
_
/
)
_
_ 0
with equality if /
i
0, respectively /
)
0. Note that, in a competitive
equilibrium, marginal cost of production is equalized across …rms, so
that c
0
)
_
/
)
_
= c
0
(/
) when /
)
0..
« If at least one unit of the public good is provided, then c
0
i
(/
i
) = j
I
=
c
0
(/
) for some individual i and …rm ,. Given that c
0
i
(/) 0 for all i,
and c
0
)
(/
)
) 0 for all ,, this implies
1
i=1
c
0
i
(/
) c
0
(/
)
For example, if every …rm produced the public good, and every individ
ual consumed it, then
1
i=1
c
0
i
(/
) =
J
)=1
c
0
)
_
/
)
_
= Jc
0
(/
) c
0
(/
).
« Because c
0
i
() is a decreasing function of /, and c
0
() is an increasing
function of /
, it follows that /
< /
, i.e. the public good is under
provided. This is an instance of the freerider problem: individuals do
not pay for the bene…ts they receive from purchases by others, so the
price …rms receive for producing a unit of the public good understates
the social value.
143
« In fact, if there is someone who has a higher marginal utility for the
public good than everyone else, then c
0
1
(/) = j
I
for this individual
entails c
0
i
(/) < j
I
for all others, i.e. no one else pays for a unit of the
good.
« The government can either provide a public good directly or use (mini
mum) quotas or subsidies to induce the e¢cient amount. For example,
a perunit subsidy of the form
:
i
=
I6=i
c
0
I
(j. /
)
would cause consumers to take the total bene…t of their public good
purchases into account.
Exercise (MWG 11.B.5).
Exercise (MWG 11.D.3).
« In general, whenever government must intervene, it faces the funda
mental problem that it has no …rsthand information about the e¤ect
of the public good on individuals. Thus, it may over or underprovide
the costly public good. A central issue is how to design a …nancing
mechanism that elicits the correct information from individuals.
« A solution is a GrovesClarke (or pivotal) mechanism, which speci…es
that every agent pays the "costs" he in‡icts on others. Speci…cally,
the government could ask consumers how their wellbeing is a¤ected
by di¤erent levels of pollution and ask …rms how bene…cial it is for
them to be able to pollute. Based on the responses, the government
implements the level of pollution that is socially optimal, given the
reports. In addition, it pays consumers the bene…ts reported by …rms,
and collects damages from …rms equal to the loss in wellbeing reported
by consumers.
« No one has an incentive to misrepresent their true costs or bene…ts
because one’s own report only a¤ects payments the other side makes
or receives (not one’s own liabilities). It also determines the level of
144
pollution the government sets, but neither …rms nor individuals have
an interest to manipulate it. If …rms report excessive costs, then more
pollution will be allowed, but they will also have to pay higher damages.
If consumers exaggerate their loss in wellbeing, pollution will be more
restricted, but they they receive less compensation.
« Because a GrovesClarke mechanism completely internalizes externali
ties  everyone bears the social cost of their actions (or reaps the social
bene…ts)  it achieves e¢ciency.
16 Monopoly and Product Di¤erentiation
16.1 Monopoly
« For many industries, the pricetaking assumption that is fundamental
to competitive equilibrium is unrealistic. The notion that many small
…rms produce a particular good can be relaxed to varying extents. Most
dramatically, to …rms whose products have no close substitutes (mo
nopolists). Then, to unique products that are, however, imperfectly
substitutable (di¤erentiation). Finally, to perfectly substitutable prod
ucts provided by a limited number of …rms.
« We begin with a monopolist that faces di¤erentiable demand r () and
cost c () for its product (and we also assume that pro…t is quasiconcave,
i.e. …rstorder conditions identify a unique maximum).
« The monopolist sets its price to maximize pro…t
: (j) = jr (j) ÷c (r (j)) .
The …rstorder condition
r (j
) +j
Jr (j
)
Jj
=
Jc (r (j
))
Jr
Jr (j
)
Jj
can be restated in terms of the inverse demand function j (r (j
)) as
Jj (r (j
))
Jr
r (j
) +j
=
Jc (r (j
))
Jr
.
145
i.e. marginal revenue equals marginal cost. This is a general op
timization principle for the …rm, even in competitive environments,
where its marginal impact on price is zero: Jj (r) ,Jr = 0, and thus
j = Jc (r) ,Jr.
« Note on inverting the derivative Jr (j
) ,Jj: it is generally the case
that
Jr (j)
Jj
Jr
1
(r (j))
Jr
= 1.
This can be seen by di¤erentiating the left and right side of r
1
(r (j)) =
j with respect to j (using the chain rule on the left). Then, if we denote
the inverse demand function r
1
(r (j)) by j (r),
Jj (r)
Jr
=
1
Jr (j) ,Jj
.
Alternatively, one can come to the same conclusion by writing pro…t
as j (r) r ÷c (r) and di¤erentiating with respect to r.
Exercise (MWG 12.B.1).
« Rearranging the …rstorder condition to
j
=
Jc (r (j
))
Jr
÷
Jj (r (j
))
Jr
r (j
) .
we see that the monopoly price strictly exceeds marginal cost, provided
Jj (r (j)) ,Jr < 0. The reason is that the monopolist reduces sales from
the socially optimal level where j = Jc (r) ,Jr, in order to increase the
price consumers are willing to pay per unit. (A competitive …rm, on
the other hand, is too small relative to the market to a¤ect the price.)
Exercise (MWG 12.B.6).
146
Figure 18: Surplus sharing under perfect competition (left) and monopoly
(right)
« The restraint in sales causes a "deadweight loss" to society, since the
foregone units cost less to produce than people are willing to pay for
them. From society’s point of view, only the allocation matters, and
the price is irrelevant. However, the price determines how surplus from
trade is divided between the …rm and consumers, so it is not irrelevant
to the monopolist. Figure 18 illustrates.
Example. Let inverse demand be linear, j (r) = c ÷/r, and marginal cost
constant at c. The optimal price and sales quantity satisfy marginal revenue
equals marginal cost:
J
Jr
j (r
) r
= c ÷2/r
= c.
i.e.
r
=
c ÷c
2/
. j
= j (r
) =
c +c
2
.
You can see from the inverse demand function that the …rm would not
produce unless c c, so j
c = j
in the relevant circumstances. Since
147
r (j) = (c ÷j) ,/, the competitive output is r
= r (c) = (c ÷c) ,/ r
.
The deadweight loss is the area of the triange with height j
÷j
= j
÷c
and base length r
÷r
:
1 =
1
2
(r
÷r
) (j
÷j
) =
(c ÷c)
2
4/
0.
Exercise (MWG 12.B.9).
Exercise (MWG 12.B.10).
« If the monopolist were able to (perfectly) pricediscriminate, i.e. charge
every consumer her valuation of each unit sold, then it would be pro…t
maximizing to sell the socially optimal quantity, and no deadweight
loss would occur. This is immediately apparent from the fact that the
monopolist then captures the entire surplus and directly maximizes it.
« There are, however, important practical obstacles to price discrimina
tion, from limited information to the possibility of resale by consumers
who obtain the product at lower prices.
Exercise (MWG 12.B.5).
16.2 Bertrand Price Competition
« The two workhorse oligopoly models are Bertrand and Cournot duopoly.
Both extend straightforwardly to more players. In this lecture, we de
velop Bertrand competition from its pure case, where …rms interact
once with identical products, to repetition and to di¤erentiated prod
ucts.
« Two …rms operate with constant marginal cost c and face total demand
r (j) for their joint output. Each …rm i = 1. 2 faces demand
r
i
(j
i
. j
)
) =
_
_
_
0 if j
i
j
)
1
2
r (j
i
) if j
i
= j
)
r (j
i
) if j
i
< j
)
individually.
148
« The …rms simultaneously set prices. The solution of interest is a Nash
equilibrium. I will not give a formal de…nition of strategy here: in
words, it speci…es an action for every possible information state the
player may …nd himself in. For our purposes, think of a player’s strategy
as a direct action or, if there is a sequence of moves, as an action
conditional on the prior moves. Let o
i
be the set of such actions or
conditional actions available to player i.
De…nition. A Nash equilibrium is a strategy pro…le : = (:
1
. :
2
) such that,
for i = 1. 2, :
i
¸ o
i
and
:
i
(:
i
. :
)
) _ :
i
(:
0
i
. :
)
)
for all :
0
i
¸ o
i
.
« A Nash equilibrium speci…es the strategy chosen by each player. It is
stable in the sense that, given ,’s strategy, i cannot gain by chang
ing hers, and vice versa. Since the payo¤s are in this context pro…ts
:
i
(j
i
) = (j
i
÷c) r
i
(j
i
. j
)
), a Nash equilibrium is a price for each …rm
that is pro…tmaximizing, given the other …rm’s price.
« Pure Bertrand duopoly has a unique Nash equilibrium (j
1
. j
2
) in which
j
1
= j
2
= c. Given j
)
= c, j
i
< 0 gains all sales, but makes a negative
pro…t per sale, whereas j
i
0 loses all sales. Hence neither improves
i’s payo¤. This means j
i
= c for i = 1. 2 is a Nash equilibrium.
« Suppose there existed a distinct other Nash equilibrium (j
0
1
. j
0
2
). If
j
0
i
< j
0
)
, then i can strictly increase pro…t by raising the price (regard
less of whether j
0
i
_ c or j
0
i
< c). Hence none of the prices can be
larger than the other; we must have j
0
i
= j
0
)
. Now if j
0
i
< c, then
increasing the price would strictly increase i’s pro…t. On the other
hand if j
0
i
c, decreasing the price would gain all sales and strictly
increase i’s pro…t. It follows that only j
0
i
= j
0
)
= c quali…es as a Nash
equilibrium. (Hence the Nash equilibrium is unique.)
« The …nding that interaction of even just two …rms reverts to marginal
cost pricing is striking, but typically not observed in reality. Next, we
consider two departures from pure Bertrand competition that restore
higher prices and positive pro…ts.
149
Exercise (MWG 12.C.1).
Exercise (MWG 12.C.4). (Only part a.)
16.3 Repetition
« One factor the pure Bertrand model ignores is that most …rms know
that they will face the joint pricesetting problem repeatedly over time.
This circumstance greatly expands the …rms’ options. They can now
condition their pricing on the rival’s past actions and react aggressively
to price cuts.
« Above marginal cost pricing may become viable because …rms maxi
mize intertemporal, rather than immediate pro…t and will consider the
e¤ect of setting a low price today on the rival’s behavior tomorrow.
Intertemporal pro…ts are
1
t=1
o
t1
:
it
.
where :
it
is i’s pro…t in period t and o ¸ (0. 1) is a discount factor. The
discount factor may be interpreted in various ways, e.g. as the constant
probability (at each point in time) that the …rms will compete again in
the following period.
Exercise (MWG 12.D.1).
« Note that, if :
it
is a constant value :
i
, then the in…nite sum can be
reduced to
1
t=1
o
t1
:
i
=
1
1 ÷o
:
i
.
since
(1 ÷o)
1
t=1
o
t1
:
i
=
1
t=1
o
t1
:
i
÷
1
t=1
o
t
:
i
=
1
t=1
o
t1
:
i
÷
1
t=2
o
t1
:
i
= :
i
.
150
« Cooperation on high prices (for example, the monopoly price j
n
) might
be sustainable in Nash equilibrium if both …rms play the trigger strat
egy
j
i
(t) =
_
j
n
if t = 1 or j
)
(t) = j
n
for t = 1. . . . . t ÷1
c otherwise
for i = 1. 2. I.e. at any given time, …rm i sets a high price j
n
only if ,
has set j
n
at all previous times. Otherwise, i "punishes" , by pricing
at marginal cost, so that , can make no more than zero pro…t.
« Such a strategy is called Nash reversion, because it switches from the
best to the worst Nash equilibrium in the punishment phase. It is also
a "grim" strategy in that it never forgives the rival for setting a low
price.
« In theory, if monopoly prices can be maintained forever in equilibrium,
punishment will never actually occur  all that matters is the threat.
This is why an unforgiving trigger strategy such as the one above can
be optimal: the threat is never tested.
« More pragmatic trigger strategies may maintain low prices for a su¢
ciently long time to deter the rival in the future. In case the punish
ment is ever triggered, say because the rival makes a mistake, marginal
cost pricing hurts the …rm in the present, but can be understood as
an investment in reputation that permits pro…table cooperation in the
future.
« If o _ 1,2 (i.e. there is not too much discounting of future income),
then j
i
(t) for i = 1. 2 is a Nash equilibrium of an in…nitely repeated
Bertrand game. The reasoning is inductive. In the …rst period, the
strategies prescribe (j
1
(1) . j
2
(1)) = (j
n
. j
n
). Nowsuppose (j
1
(t ÷1) . j
2
(t ÷1)) =
(j
n
. j
n
) and j
1
(t) = j
n
. The best alternative for 2 to j
2
(t) = j
n
is to slightly undercut and earn a pro…t of (j
n
÷c) ¡
n
, instead of
(j
n
÷c) (¡
n
,2).
« Undercutting triggers j
1
(t) = c for t = t + 1. . . . . · (so that the
maximal pro…t 2 can receive in future periods is zero), so it is optimal
for 2 in period t if and only if
(j
n
÷c) ¡
n
1
1 ÷o
(j
n
÷c)
¡
n
2
.
151
i.e. if the pro…t from capturing all sales today exceeds the discounted
value of future pro…t streams when sharing the sales.
« Because the inequality holds only when o < 1,2, it is optimal for 2
to set j
2
(t) = j
n
if o _ 1,2. Since the …rms share the market at
t = 1, and continue to do so whenever they have shared it previously,
induction implies that they share the market in all periods. Hence
(j
1
(1) . j
2
(1)) = (j
n
. j
n
) is a Nash equilibrium of the in…nitely re
peated game.
« Since o < 1,2 implies that the optimal coordination on (j
1
(t) . j
2
(t)) =
(j
n
. j
n
) cannot be sustained, it is intuitively clear that no other kind
of coordination can work either. I.e. either j
1
(t) = c or j
2
(t) = c at
all times t.
« However, o _ 1,2 admits many other cooperative equilibria besides
the pro…tmaximizing one. This is an instance of the Folk Theorem
from game theory, which says that equilibria leading to any payo¤s that
exceed the best each player can achieve independently (the minmax, i.e.
the best attainable in the worstcase scenario) exist for some su¢ciently
large o.
Exercise (MWG 12.D.4).
« Here, the minmax pro…ts are zero, since every player can always guar
antee zero by setting price to marginal cost, but no more than that.
Therefore, any combination of pro…ts exceeding zero is possible (to see
this, just replace j
n
above with any other prices, possibly asymmetric,
and consider o very close to 1, so that threats are arbitrarily damaging
in the long run). Which equilibrium occurs often depends on history
and focal points.
« Collusion is often not maintained as smoothly as suggested here. In
practice, demands and costs ‡uctuate, and it may not be obvious to
other parties whether a …rm is ceasing to cooperate, or simply reacting
to environmental changes. To prevent …rms from taking advantage of
this ambiguity, it becomes necessary to punish deviations, even if it is
152
not clear why they happened, and this leads to intermittent price wars,
between periods of high prices.
Exercise (MWG 12.D.5).
16.4 Product Di¤erentation
« When products are di¤erentiated, …rmi’s demand is not perfectly price
elastic. It may instead be a continuously decreasing function r
i
(j
i
. j
)
)
of …rm i’s price, given the price j
)
of the competitor.
« If …rms face constant marginal cost c 0, they choose their prices to
maximize
:
i
(j
i
) = (j
i
÷c) r
)
(j
i
. j
)
) .
Typically, …rms retain some positive demand when pricing above mar
ginal cost, from consumers who value their unique products. Hence
pricing above marginal cost will be the pro…tmaximizing strategy, even
if it means getting undercut.
« There are two main approaches to product di¤erentation: in one, the
Hotelling model, a consumer demands only one of the products; in
the other, the DixitStiglitz model, products are used together, there
is an explicit preference for variety. Monopolistic competition is the
special case of the DixitStiglitz model where variety is in…nite (there
is a continuum of di¤erentiated products with small market shares, but
some market power). I focus on the Hotelling model.
« Consider duopolists that serve a continuum of consumers whose de
mands arise in the following manner. The …rms are associated with
points 0 and 1 in a spectrum [0. 1] of possible products. Every indi
vidual occupies at a point . ¸ [0. 1] that re‡ects her most preferred
product. In purchasing either 0 or 1, the individual incurs a "travel
cost" .t or (1 ÷.) t, which indicates lost satisfaction from consuming
a nonideal product.
« Someone who prefers . is therefore willing to pay at most
1
0
(.) = · ÷.t
153
for product 0, and at most
1
1
(.) = · ÷(1 ÷.) t
for product 1 (· is the good’s undiscounted value).
« Individual demand is for either one or zero units, hence no one consumes
both products, and some might consume neither. If the …rms charge
j
0
and j
1
for a unit of their products, then the person who prefers .
will consume product 0 if 1
0
(.) _ j
0
, i.e.
. _
· ÷j
0
t
and 1
0
(.) ÷j
0
_ 1
1
(.) ÷j
1
, i.e.
. _
1
2
+
j
1
÷j
0
2t
= ^ ..
On the other hand, this person will consumer product 1 if
. _ 1 ÷
· ÷j
1
t
and the second inequality is reversed.
« The point ^ ., where 1
0
(.) ÷ j
0
= 1
1
(.) ÷ j, belongs to the "marginal
consumer," who is just indi¤erent between the products.
« Suppose . is uniformly distributed on [0. 1], so that every possible prod
uct is preferred by the same mass of consumers. Then the total demand
for product 0 is the length of the interval [0. .
0
], where
.
0
= min
_
· ÷j
0
t
. ^ .
_
.
and total demand for product 1 is the length of the interval [.
1
. 1], i.e.
1 ÷.
1
, where
.
1
= max
_
1 ÷
· ÷j
1
t
. ^ .
_
.
(Assuming, of course, that prices are such that .
0
. .
1
¸ (0. 1). When
ever this is violated, one …rm makes no sales, which cannot be pro…t
maximizing unless the other …rm prices below marginal cost.)
154
« For simplicity, let · be su¢ciently large that the market is "covered"
in equilibrium: every consumer purchases one of the products. Then
the marginal consumer determines the …rm’s demand functions
r
0
(j
0
. j
1
) = ^ .. r
1
(j
0
. j
1
) = 1 ÷ ^ ..
« Pro…ts
:
0
(j
0
. j
1
) = (j
0
÷c) ^ .. :
1
(j
0
. j
1
) = (j
1
÷c) (1 ÷ ^ .)
are maximized when the …rstorder conditions with respect to j
0
and
j
1
,
j
0
(j
1
) =
j
1
+c +t
2
.
j
1
(j
0
) =
j
0
+c +t
2
.
are met.
« These are the best responses to the other …rm’s price. A joint solution,
where prices are mutual best responses, is a Nash equilibrium:
j
0
(j
1
) = c +t = j
1
(j
0
) .
I.e. product di¤erentiation allows the …rms to price above marginal
cost in Bertrand equilibrium, and increasingly so the more consumers
discount for di¤erences from their most preferred products. In the
limiting case, where consumers do not care about such di¤erences and
thus t = 0, prices equal marginal cost, and pro…ts are zero.
« This simple case can be extended to multiple …rms, arbitrary consumer
taste distributions, individual demands other than zeroone, asymmet
ric costs, as well as endogenous …rm locations (strategic product posi
tioning). Above marginal cost pricing is not merely owed to the fact
that few …rms compete in the market. In fact, entry may not lower
prices because it tends to attract consumers on the margin (who have
the lowest willingness to pay for the product). Once the incumbent
…rm loses these pricesensitive consumers, it may actually increase its
price further.
Exercise (MWG 12.C.17). Linear city with di¤erent costs
Exercise (MWG 12.C.16). Circular city with quadratic cost
155
17 Capacity Constraints
17.1 CapacityConstrained Pricing
« An assumption that was implicit in Bertrand oligopoly, and turns out
to be crucial for its marginal cost pricing equilibrium, is that …rms
can serve arbitrarily large demands (i.e. capacity is free or can be
increased instantly). Then, no …rm can a¤ord to be undercut, since all
sales would be lost to competitors. These circumstances lead to very
aggressive pricing.
« With limited production capacities, the lowerpriced of two identical
products is sold to early buyers, and the higherpriced alternative may
still be demanded by latecomers. This tends to soften pricing, because
a …rm can make a pro…t despite being undercut. Capacity constraints
lead to the Cournot model of oligopoly, which can be viewed as quantity,
rather than price, competition in the sense that the …rm’s fundamental
choices are their production capacities.
« For simplicity, consider again a duopoly in which …rms i = 1. 2 simul
taneously choose capacities ¡
1
and ¡
2
(at a positive perunit cost), and
subsequently engage in price competition. At the latter point, each
…rm is able to sell up to its capacity at a constant marginal cost c _ 0.
Both capacities are known to both …rms at the time when prices are
set.
« We assume that the products are identical, and demand for the total
output is a continuous, strictly decreasing and concave function r (j).
Denote the inverse demand by j (). Which of the two prices e¤ectively
constrains demand depends on the capacities. If j
i
< j
)
and ¡
i
_ r (j
i
),
then …rm i serves everyone, so j
i
determines how much is demanded. If
¡
i
< r (j
i
), then …rm , serves those who cannot buy from …rm i, which
sells at capacity, so j
)
determines how much is demanded.
« We must be speci…c about who buys from i at the lower price when
demand exceeds i’s capacity. This is determined through a rationing
rule. We will impose that consumers with the highest willingness to
156
pay are at the head of the queue. Then demand is, for i = 1. 2,
r
i
(j
1
. j
2
) =
_
_
_
min ¦¡
i
. r(j
i
)¦ if j
i
< j
)
min
_
¡
i
. max
_
r (j
i
) ÷¡
)
.
1
2
r (j
i
)
__
if j
i
= j
)
min ¦¡
i
. max ¦r (j
i
) ÷¡
)
. 0¦¦ if j
i
j
)
.
« This rationing rule where "highest valuations served …rst" is known in
industrial organization as the "e¢cient rationing rule." (An alternative
is, for example, the proportional rationing rule, where anyone is equally
likely to be at the head of the queue, so that …rm ,’s customers have
the same expected willingness to pay as …rm i’s. This would increase
,’s demand, and decrease i’s demand, at any given prices.)
« We begin by analyzing the pricing game for some given capacities ¡
1
0
and ¡
2
0. (If one …rm does not invest in capacity, we simply have a
monopoly.) Denoting by / (¡
)
) the optimal quantity …rm i would sell
if it were not capacityconstrained (and …rm , sold ¡
)
), let ¡
1
_ / (¡
2
)
and ¡
2
_ / (¡
1
). (Hence, the capacity constraints bind.)
« Since capacity is costly to build, both …rms must sell a nonzero quantity
in equilibrium: else they could avoid losses by not investing in capacity.
« This means prices must be equal. Otherwise, if j
i
< j
)
, then …rm
i would sell at capacity (given that consumers …rst go to the lower
priced seller, and …rm , still manages to sell something). But then
i could slightly increase its price and still sell at capacity, giving it a
strictly higher pro…t.
« Similarly, it is not possible that j
1
= j
2
< j (¡
1
+¡
2
), in which case
market demand would exceed both capacities, so that each …rm could
pro…tably increase its price. On the other hand, if j
1
= j
2
j (¡
1
+¡
2
),
then market demand does not cover both capacities. At least one …rm
has spare capacity and will want to slightly lower its price (i.e. undercut
like a pure Bertrand competitor) as long as j
1
= j
2
c (otherwise, if
j
1
= j
2
_ c, it is not optimal to invest in capacity).
« The only alternative is j
1
= j
2
= j (¡
1
+¡
2
). It remains to be shown
that this is a Nash equilibrium given ¡
1
and ¡
2
. Suppose therefore
that j
)
= j (¡
1
+¡
2
). Neither of i’s deviations from j
i
= j (¡
1
+¡
2
) is
pro…table. Namely, j
i
< j (¡
1
+¡
2
) simply lowers the price, but cannot
157
increase sales, since …rm i is already at capacity. And j
i
j (¡
1
+¡
2
)
is undesirable by the assumption / (¡
)
) _ ¡
i
, which implies that …rm i’s
pro…t increases in sales when , is at, and i is below, capacity. Hence i
should lower price while j
i
_ j (¡
1
+¡
2
).
« We conclude that j
1
= j
2
= j (¡
1
+¡
2
) is the unique Nash equilibrium
of the pricing game with capacities ¡
i
¸ (0. / (¡
)
)]. (Incidentally, there
is no equilibrium if ¡
i
/ (¡
)
) and j (¡
1
+¡
2
) c, since i then sells
below capacity, but at symmetric prices it would always be pro…table
to slightly undercut.)
« This insight motivates the Cournot model of oligopoly, which takes as
given that each …rm will produce up to capacity and set the market
clearing price j (¡
1
+¡
2
). The Cournot model focuses on the capacity
or quantitysetting game preceding price formation.
17.2 Cournot Quantity Competition
« Let the inverse demand j () be a decreasing, di¤erentiable function of
¡
i
+¡
)
with j (0) c. Duopolist i = 1. 2 solves
max
q
i
0
(j (¡
i
+¡
)
) ÷c) ¡
i
.
which has …rstorder condition
j
0
(¡
i
+¡
)
) ¡
i
+j (¡
i
+¡
)
) ÷c _ 0
(equality if ¡
i
0).
« Suppose ¡
i
= 0. Then j (¡
i
+¡
)
) = j (¡
)
) _ c, and ¡
)
= 0 contradicts
j (0) c. On the other hand, ¡
)
0 would imply j
0
_
¡
)
_
¡
i
+j
_
¡
)
_
= c
at ¡
i
= ¡
i
, i.e. j
_
¡
)
_
< c (since j
0
() < 0 everywhere). Again, this
contradicts j (0) c. Under the assumptions, we must have ¡
i
0 for
i = 1. 2, so that the …rstorder conditions apply with equality.
« Then they are functions
¡
i
= ÷
j (¡
i
+¡
)
) ÷c
j
0
(¡
i
+¡
)
)
that implicitly determine the best quantity choice for …rm i, given …rm
,’s quantity choice.
158
« A Nash equilibrium
_
¡
i
. ¡
)
_
jointly solves these bestresponse functions,
thus:
j
_
¡
i
+¡
)
_
= c ÷j
0
_
¡
i
+¡
)
_ ¡
i
+¡
)
2
= c ÷j
0
_
¡
i
+¡
)
_
¡
i
.
Proposition. Given identical, constant marginal costs c 0, and in
verse demand for the industry’s combined output such that j (0) 0 and
j
0
(¡
1
+¡
2
) < 0 whenever ¡
1
+¡
2
_ 0, the Cournot equilibrium price satis…es
c < j (¡
1
+¡
2
) < j (¡
n
) .
where ¡
n
is the optimal monopoly output.
Proof. From the Nash equilibrium condition above, it is immediate that
j (¡
1
+¡
2
) c. So we show j (¡
1
+¡
2
) < j (¡
n
). Since j
0
() < 0, this will
be the case if ¡
1
+ ¡
2
¡
n
. If ¡
1
+ ¡
2
< ¡
n
, then …rm i can increase its
production to ¡
i
= ¡
n
÷¡
)
, in which case the industry supplies the monopoly
quantity, at the monopoly price and monopoly pro…t. Because the increase
in ¡
i
lowers the price, while ¡
)
remains …xed, this must strictly reduce ,’s
pro…t. But combined pro…t cannot decrease, since the monopoly pro…t is
an upper bound on industry pro…t. Hence i’s pro…t strictly increases. This
violates Nash equilibrium, hence we must have ¡
1
+¡
2
_ ¡
n
.
If ¡
1
+ ¡
2
= ¡
n
, then the industry acts like a monopolist, and the …rst
order condition for Cournot equilibrium should be identical to the …rstorder
condition for a monopoly equilibrium at ¡
1
+ ¡
2
= ¡
n
, namely j (¡) = c ÷
j
0
(¡) ¡, where ¡ = ¡
1
+¡
2
. You can check that it is not so. Thus ¡
1
+¡
2
¡
n
.
« Note that ¡
1
= / (¡
2
) and ¡
2
= / (¡
1
) in Cournot equilibrium: the
quantity choices can be interpreted as binding capacity choices that lead
to j (¡
1
+¡
2
) as a price equilibrium in a Bertrand game with capacity
constraints.
« Unlike pure Bertrand oligopoly, Cournot oligopoly does not lead to
competitive (marginal cost) pricing. The reason is that, given capacity
159
constraints, neither …rm can capture the entire market by undercutting.
Since each …rm’s demand is less than perfectly elastic, it can make a
positive pro…t by pricing above marginal cost.
« Yet the …rms are unable to maximize industry pro…t, i.e. set the
monopoly price. In determining its capacity, each …rm only takes into
account how an increase in sales, and therefore a lower marketclearing
price, a¤ects its own revenue. It ignores the e¤ect on the rival’s revenue.
Hence the …rms do not respond to the full bene…t (to the industry) of
keeping sales low and price high as a monopolist would; they sell "too
much."
« This is easy to see from the Nash equilibrium condition: Cournot
duopolists markup on marginal cost by ÷j
0
_
¡
i
+¡
)
_
¡
i
, i.e. the ef
fect of the price reduction on own revenue. By contrast, a monopolist
marks up by ÷j
0
(¡
) ¡
, with ¡
= ¡
i
+ ¡
)
, i.e. the e¤ect on industry
revenue.
Example. If inverse demand is linear, j (¡
1
+¡
2
) = c ÷ / (¡
1
+¡
2
) (where
c c and / 0), then pro…ts are (c ÷/¡
i
÷/¡
)
÷c) ¡
i
, resulting in …rstorder
conditions (bestresponse functions)
¡
1
(¡
2
) =
c ÷/¡
2
÷c
2/
and
¡
2
(¡
1
) =
c ÷/¡
1
÷c
2/
.
(In this context, bestresponse functions are often referred to as reaction
functions.) Solving jointly, we obtain the Nash equilibrium
¡
1
(¡
2
) =
c ÷c
3/
= ¡
2
(¡
1
) .
with marketclearing price
j (¡
1
+¡
2
) =
1
3
c +
2
3
c.
Since c c, this means j (¡
1
+¡
2
) c. The monopoly quantity is …rm i’s
best response to ¡
)
= 0. Hence ¡
n
= ¡
i
(0) = (c ÷c) , (2/) and j (¡
n
) =
c,2 +c,2 j (¡
1
+¡
2
).
160
Exercise (MWG 12.C.9).
Exercise (MWG 12.C.20).
Exercise (MWG 12.D.3).
17.3 Competitive Limit
« In the generalization to J …rms producing nonzero quantities with iden
tical marginal costs c, …rstorder conditions imply that the outputs are
equal, and adding up over , = 1. . . . . J, i.e.
J¡
= ÷J
j (J¡
) ÷c
j
0
(J¡
)
.
can be arranged for the equilibrium price
j (J¡
) = c ÷j
0
(J¡
) ¡
.
« Intuitively, as J ÷·, each …rm’s production level becomes very small,
so that j (J¡
) ÷ c. (I.e. competitive pricing is approached in the
limit.) If the number of …rms in the industry is determined by free entry
in response to pro…t opportunities, the competitive limit is obtained as
demand increases.
Exercise (MWG 12.C.7).
« Consider parameterized demand functions r
c
(j) = cr (j), where r (j)
is some aggregate demand function, and c 0 is the "market size."
Let j
c
be the equilibrium price, and Q
c
the equilibrium total output
(after all pro…table entry has occurred), when the market size is c.
« Assume that each …rm has the same minimum average cost c at some
level of output ¡ 0 (we do not impose constant marginal cost here).
Then Q
c
+ ¡ _ cr ( c) because otherwise, if demand exceeded supply
(plus an additional quantity ¡) at j = c, the marketclearing price after
a further entry (at the minimum average cost level ¡) would be greater
than c, and entry would be pro…table.
161
« This imposes a lower bound on the equilibrium supply Q
c
: it cannot
fall short by more than ¡ of the quantity that would be supplied if
j = c. It follows that the equilibrium price j
c
is bounded above: it
cannot exceed c by so much that the post entry price, after another
…rm enters and supplies ¡, is still greater than c.
« Let j
c
() be the inverse supply function at market size c, and de…ne
the di¤erence between this upper bound on j
c
and c (which is a lower
bound such that no …rm exits) as
´j
c
= j
c
(cr ( c) ÷ ¡) ÷j
c
(cr ( c)) .
« In order to see what happens to the equilibrium price interval as market
size increases from c = 1 toward in…nity, rewrite ´j
c
in terms of j (),
the inverse supply function at c = 1. Since ¡
c
= r
c
= cr = c¡, we
have j (¡) = j
c
(¡
c
,c), i.e.
´j
c
= j
_
cr ( c) ÷ ¡
c
_
÷j (r ( c)) .
which goes to zero as c ÷·.
« This means the equilibrium price converges to j
c
(r
c
( c)) = c, the long
run competitive price, as market size gets large. The industry output
is then also competitive (and …rms are small relative to the market,
since they produce at the …xed minimum average cost level ¡). With
constant marginal cost c, c = c.
Exercise (MWG 12.F.3).
Exercise (MWG 12.F.2).
Exercise (MWG 12.F.4).
18 Precommitment and Entry
18.1 Precommitment
« In some industries, …rms act sequentially, rather than simultaneously
as assumed so far. In fact, it may super…cially appear that sequential
162
moves are always more realistic, but this view misses the point of the
distinction between the game forms. It is a matter of information,
rather than timing.
« In a sequential game, strategy sets di¤er in that some players are able
to condition their actions on what other players are observed to do.
This is not necessarily a bene…cial power to have: …rst movers can
potentially exploit it by manipulating the incentives for late movers.
« The e¤ect of an observed action by …rm i on the behavior of another
…rm , is captured by ,’s best response function. If , bestresponds
to a change in i’s strategy with a change in the same direction, i.e.
d/
)
(:
i
) ,d:
i
0, then :
i
and :
)
= /
)
(:
i
) are strategic complements. If
,’s best response is in the opposite direction, i.e. d/
)
(:
i
) ,d:
i
< 0, then
we have strategic substitutes.
« For instance, in pure Bertrand competition, a price cut by one …rm
(above marginal cost) is answered by a price cut by the other …rm, so
these are strategic complements. In a Cournot setting, a sales increase
by one …rm often creates incentives for the other …rm to reduce its
sales, so these are strategic substitutes.
Example. Suppose costs depend on an investment / that …rm i can make
(for example, in process innovation), i.e. c
0
(/) < 0. Let the investment stage
be followed by Cournot competition with linear inverse demand j (¡
i
+¡
)
) =
c ÷ / (¡
i
+¡
)
) (where c c _ 0 and / 0) and constant marginal costs
(namely, c (/) for i, and c for ,). With …rmspeci…c marginal costs, the
bestresponse functions are
¡
i
(¡
)
. /) =
c ÷c (/)
2/
+
1
2
¡
)
.
¡
)
(¡
i
(/)) =
c ÷c
2/
+
1
2
¡
i
(/) .
leading to equilibrium
¡
i
_
¡
)
. /
_
=
c ÷2c (/) +c
3/
.
¡
)
(¡
i
(/)) =
c ÷2c +c (/)
3/
.
163
The bestresponse level of output for i is increasing in the costreduction: for
any ¡
)
, ¡
0
i
(/) 0. Di¤erentiating ,’s best response with respect to / gives
J¡
)
(¡
i
(/))
J/
= ÷
1
2
¡
0
i
(/) < 0,
so output levels are strategic substitutes, and i’s investment in cost reduc
tion leads to less aggressive behavior from ,, which reinforces the inherent
advantage of lower cost.
Exercise (MWG 12.G.1).
« A Cournot duopolist may take advantage of the strategic substitutes
relationship by precommitting to high sales, so that the competitor will
concede a large share of the market.
« The sequential version of Cournot duopoly is called Stackelberg duopoly.
One …rm (the "leader") sets its quantity …rst, and the choice is observed
by the other …rm (the "follower"). The follower’s optimal reaction is
anticipated by the leader. Given the leader’s choice ¡
1
, the follower
maximizes
:
2
(¡
1
. ¡
2
) = (j (¡
1
+¡
2
) ÷c) ¡
2
.
The resulting bestresponse function for the follower is
¡
2
(¡
1
) = ÷
j (¡
1
+¡
2
) ÷c
j
0
(¡
1
+¡
2
)
.
This is identical to a Cournot …rm’s best response function.
« The leader, however, maximizes against the follower’s best response
function (rather than a …xed value):
:
1
(¡
1
. ¡
2
(¡
1
)) = (j (¡
1
+¡
2
(¡
1
)) ÷c) ¡
1
.
This leads to the leader’s best response
¡
1
(¡
2
) = ÷
j (¡
1
+¡
2
(¡
1
)) ÷c
j
0
(¡
1
+¡
2
(¡
1
))
1
1 +¡
0
2
(¡
1
)
=
¡
2
(¡
1
)
1 +¡
0
2
(¡
1
)
.
If ¡
1
and ¡
2
are strategic substitutes, i.e. ¡
0
2
(¡
1
) < 0, then ¡
1
(¡
2
)
¡
2
(¡
1
). If ¡
1
and ¡
2
are strategic complements, i.e. ¡
0
2
(¡
1
) 0, then
¡
1
(¡
2
) < ¡
2
(¡
1
).
164
« Since ¡
2
(¡
1
) is the Cournot best response function, it is clear that total
output increases in the case of strategic substitutes, and total output
decreases in the case of strategic complements. If inverse demand j ()
is decreasing in combined sales, then the industry price decreases with
strategic substitutes and increases with strategic complements relative
to Cournot competition.
Example. With linear inverse demand, j (¡
1
+¡
2
) = c ÷ / (¡
1
+¡
2
), the
follower’s best response in the Stackelberg model is
¡
2
(¡
1
) =
c ÷c
2/
÷
¡
1
2
.
Since 1 +¡
0
2
(¡
1
) = 1,2,
¡
1
(¡
2
) = 2¡
2
(¡
1
) =
c ÷c
/
÷¡
1
=
c ÷c
2/
.
whereas
¡
2
(¡
1
) =
c ÷c
4/
.
This is the strategic substitutes case (¡
0
2
(¡
1
) = ÷1,2); the leader sells a
larger quantity, knowing that the follower will practice restraint in order
avoid too much price deterioration. Compared to the simultaneous Cournot
outcome ¡
1
(¡
2
) = ¡
2
(¡
1
) = (c ÷c) , (3/), the overall quantity sold increases
from ¡
1
+¡
2
= (2,3) (c ÷c) ,/ to ¡
1
+¡
2
= (3,4) (c ÷c) ,/ under Stackelberg
competition. As a result, price decreases from j (¡
1
+¡
2
) = c,3 + (2,3) c to
j (¡
1
+¡
2
) = c,4 +(3,4) c (keeping in mind that c c). You can check that
the leader’s pro…t is larger, and the follower’s pro…t smaller, than in Cournot
competition.
« In the Stackelberg model, both …rms produce a nonzero quantity, as
long as price exceeds marginal cost. (There is always at least a small
strictly positive pro…t to be made.) When production requires a …xed
initial outlay (e.g. in product development or a plant), the leader might
deter the follower from incurring the startup cost by committing to an
aggressive response (high sales at a low price). This intention can be
signaled credibly (in the sense that it becomes the best response) by
investing in high capacity and low marginal cost.
165
« The leader might then be able to operate as a monopolist, possibly
under constraints to keep the price lowenough to continually discourage
entry.
18.2 Entry Equilibrium
« In the remainder of the lecture, we return to a symmetric environment
and consider how much entry will occur in equilibrium with a …xed
entry cost 1 0. There is an in…nite number of potential entrants
who decide, at stage 1 of the game, whether to invest 1 or not, and
the entrants compete at stage 2 as oligopolists.
« Assume that, for any number J of entrants, there is in stage 2 a unique,
symmetric Nash equilibrium, yielding pro…t :
J
for each entrant (this
excludes the entry cost 1). The equilibrium number of entrants J
is such that no …rm wants to either enter or exit, given the prevailing
pro…t :
J
:
:
J
_ 1 and :
J
+1
< 1.
(We assume …rms enter if indi¤erent, i.e. if :
J
+1
= 1.)
« If :
J
is decreasing in J, and :
J
÷ 0 as J ÷ ·, then the equilibrium
J
is unique.
Example. Suppose stage 2 is a pure Bertrand game, where …rms have con
stant marginal costs c, and inverse demand is linear, j (¡) = c ÷ /¡ with
c c _ 0 and / 0. We know that :
J
= 0 when J 1. If the monopoly
pro…t :
n
exceeds entry cost 1, a single …rm will enter (i.e. J
= 1)
and set the monopoly price. Since :
n
(¡) = (j (¡) ÷c) ¡ is maximized by
¡
= (c ÷c) , (2/), the optimal pro…t is :
n
(¡
) = (c ÷c)
2
, (4/). The entry
criterion :
n
(¡
) _ 1 is therefore equivalent to 1 _ (c ÷c)
2
, (4/).
Exercise (MWG 12.E.1).
« In an alternative setup, entry cost may be incurred only if the …rm
makes nonzero sales. Thus, a …rm can observe how many other …rms
enter, and bear the …xed cost only if it can make a nonnegative pro…t.
This approach restores something close to marginal cost pricing in the
166
Bertrand entry game, since …rms enter if and only if there is positive
pro…t to be made.by undercutting the prevailing price.
« Amonopolist could in this case not set a price above j
= (1 +cr (j)) ,r(j),
which approaches c when industry demand r (j) is large. Else, another
…rm could enter without prepaying the entry cost 1. In other words,
the entrant would not have to worry that it may be undercut in the
ensuing Bertrand competition and be left with a loss of 1. It can sim
ply respond to the incumbent’s current price, since it can exit freely
if the incumbent’s behavior changes. Industries where such "hit and
run" entry is possible are called contestable markets.
Example. Under the same cost and demand conditions, consider a Cournot
game in stage 2. Each entrant maximizes
:
J
= (j (J¡
J
) ÷c) ¡
J
= (c ÷(J ÷1) / ¡
J
÷/¡
J
÷c) ¡
J
.
so that
¡
J
=
c ÷c
2/
÷
J ÷1
2
¡
J
=
c ÷c
/
1
J
+ 1
.
since all …rms symmetrically set ¡
J
= ¡
J
. Thus
:
J
=
1
/
_
c ÷c
J
+ 1
_
2
is strictly decreasing in J
, and :
J
÷0 as J
÷0.
Since :
~
J
= 1 at
~
J =
c ÷c
_
/1
÷1.
the equilibrium number of entrants J
is the greatest integer below
~
J. As
1 ÷0 or / ÷0, J
÷· and the industry price
j (J
¡
J
) = c ÷J
/¡
J
= c ÷
J
J
+ 1
(c ÷c)
=
1
J
+ 1
c +
J
J
+ 1
c
approaches marginal cost.
167
18.3 Socially Optimal Entry
« Entry in an industry provides, on the one hand, valued goods to con
sumers and, on the other, duplicates entry cost for …rms. Given that
…rms choose their production levels to maximize pro…t at the postentry
stage, given the market structure, there exists an e¢cient number of
entrants from a social perspective that optimally resolves this tradeo¤.
« This number J
maximizes consumption bene…ts (as measured by the
willingness to pay for each unit) less production cost (which includes
each …rm’s entry cost 1):
o (J) =
_
Jq
J
0
j (:) d: ÷J (c (¡
J
) +1) .
Example. Returning to the pure Bertrand example, the socially optimal
number of entrants can be no more than two, since price equals marginal
cost if two …rms compete (hence surplus is maximized, and it cannot be
socially desirable to incur further entry costs). Since a single …rm enters in
equilibrium, the number is either socially optimal, or one fewer than socially
optimal.
Example. In the Cournot example that was already described,
o (J) = J
_
(c ÷c) ¡
J
÷
1
2
J/¡
2
J
÷1
_
=
_
J
J + 1
÷
1
2
_
J
J + 1
_
2
_
(c ÷c)
2
/
÷J1.
The welfaremaximizing number of …rms J
satis…es
o
0
(J
) =
1
(J
+ 1)
3
(c ÷c)
2
/
÷1 = 0.
so
(J
+ 1)
3
=
_
c ÷c
_
/1
_
2
.
168
Recall that
~
J = (c ÷c) ,
_
/1 ÷ 1 was the equilibrium number of entrants.
It exceeds the socially optimal number, since J
+ 1 =
_
~
J + 1
_
2¸3
<
~
J + 1.
Exercise (MWG 12.E.3).
« Entry may be ine¢cient with Bertrand as well as with Cournot com
petition: in the …rst case, we have (possibly) too little, in the second
generally too much, entry. There are two kinds of failures.
« No …rm may …nd it pro…table to enter an industry as a monopolist
because the maximal pro…t the …rm can make with perunit pricing
does not recover the cost of entry. However, from a social perspective
it may be desirable that the monopolist operate, since the consumer
surplus may o¤set losses in pro…t. Of course, the potential monopolist
ignores these bene…ts to consumers, unless it can extract them through
price discrimination or is o¤ered a subsidy.
« The same logic could apply to an additional potential entrant in an
industry with some number of incumbents  it may be socially valu
able, but not privately optimal, for the entry to occur. This reasoning
suggests that underentry by one …rm (relative to J
) is possible.
« More typically, there is overentry because …rms consider only whether
their pro…ts can cover the cost of entry, but ignore the reduction in
incumbents’ pro…ts. Because the incumbents sell fewer units after an
additional entry, their contribution to surplus declines, while the com
bined entry cost they incurred remains …xed. From a social point of
view, the net increase in total output (and consumption bene…ts) may
not justify the duplication of entry costs.
« But for the …rm, entry may still be pro…table because its sales exceed
the increase in total output (the di¤erence being output reductions
by incumbents) and could therefore cover the entry cost. In short,
overentry is possible due to the "businessstealing" e¤ect.
« These insights can be stated as a general result: under standard condi
tions, equilibrium entry may well exceed the socially optimal level, but
may fall short by at most one …rm.
169
Proposition. If marginal …rm pro…ts are positive for any number of entrants
J (i.e. j (J¡
J
) ÷ c
0
(¡
J
) _ 0), and more entry increases industry output
( J J
0
== J¡
J
_ J
0
¡
J
) but decreases …rm output and pro…t ( J J
0
==
¡
J
_ ¡
J
0 ), and if furthermore j
0
() < 0 and c
00
() _ 0, then the equilibrium
number of entrants J
_ J
÷ 1, where J
is the socially optimal number
of entrants.
Proof. Since the claim is obviously true when J
= 1, let J
1. By
de…nition, \ (J
) _ \ (J
÷1), hence:
_
J
q
J
(J
1)q
J
1
j (:) d: ÷J
(c (¡
J
) ÷c (¡
J
1
)) ÷c (¡
J
1
) _ 1.
Under the assumptions, J
¡
J
(J
÷1) ¡
J
1
, and price is a decreasing
function of industry supply, so j ((J
÷1) ¡
J
1
) is the maximum in the
price interval from j (J
¡
J
) to j ((J
÷1) ¡
J
1
). Therefore,
_
J
q
J
(J
1)q
J
1
j (:) d: _ j ((J
÷1) ¡
J
1
) (J
¡
J
÷(J
÷1) ¡
J
1
) .
(Think of the fact that the area under the curve j (¡) between ¡
0
= (J
÷1) ¡
J
1
and
¡
1
= J
¡
J
is contained in the rectangle formed by ¡
1
÷¡
0
and the maximal
height of the function j (¡).)
Then, by rearranging the …rst inequality,
J
(j ((J
÷1) ¡
J
1
) (¡
J
1
÷¡
J
) +c (¡
J
) ÷c (¡
J
1
))
_ j ((J
÷1) ¡
J
1
) ¡
J
1
÷c (¡
J
1
) ÷1.
Now, the convexity of the cost function (c
00
() _ 0) implies
c (¡
J
) ÷c (¡
J
1
) =
_
q
J
q
J
1
c
0
(:) d: _ c
0
(¡
J
1
) (¡
J
÷¡
J
1
)
since c
0
(¡
J
1
) is the minimum in the cost interval between c
0
(¡
J
1
) and
c
0
(¡
J
). Thus,
J
(j ((J
÷1) ¡
J
1
) ÷c
0
(¡
J
1
)) (¡
J
1
÷¡
J
)
_ j ((J
÷1) ¡
J
1
) ¡
J
1
÷c (¡
J
1
) ÷1.
Since ¡
J
1
¡
J
and j ((J
÷1) ¡
J
1
) ÷ c
0
(¡
J
1
) _ 0 by assump
tion, the right side is positive, i.e.
j ((J
÷1) ¡
J
1
) ¡
J
1
÷c (¡
J
1
) = :
J
1
_ 1.
170
Then …rms must enter at least until their number is J
÷ 1, because :
J
is
decreasing in J:
:
J
÷:
J1
= j (J¡
J
) ¡
J
÷j ((J ÷1) ¡
J1
) ¡
J1
÷(c (¡
J
) ÷c (¡
J1
))
_ j (J¡
J
) ¡
J
÷j ((J ÷1) ¡
J1
) ¡
J1
÷c
0
(¡
J1
) (¡
J
÷¡
J1
)
= (j (J¡
J
) ÷c
0
(¡
J1
)) ¡
J
÷(j ((J ÷1) ¡
J1
) ÷c
0
(¡
J1
)) ¡
J1
_ 0.
since ¡
J
_ ¡
J1
and j (J¡
J
) _ j ((J ÷1) ¡
J1
) from J¡
J
_ (J ÷1) ¡
J1
and
j
0
() < 0.
Exercise (MWG 12.E.2).
Exercise (MWG 12.E.4).
171
Contents
1 Preference 1.1 Consumption Set . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Rational Preference . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Utility Functions . . . . . . . . . . . . . . . . . . . . . . . . . 3 3 5 7
2 Utility 8 2.1 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 Quasiconcavity . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3 Demand I: Utility Maximization Problem 15 3.1 Budgets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.2 The Utility Maximization Problem (UMP) . . . . . . . . . . . 16 3.3 Indirect Utility Function . . . . . . . . . . . . . . . . . . . . . 21 4 Demand II: Expenditure Minimization 4.1 EMP and Hicksian Demand . . . . . . 4.2 Expenditure Function . . . . . . . . . . 4.3 Duality . . . . . . . . . . . . . . . . . . 5 Comparative Statics 5.1 Wealth E¤ects . . . . 5.2 Price E¤ects . . . . . 5.3 Law of Demand . . . 5.4 Elasticity . . . . . . 5.5 Money Metric . . . . 5.6 Welfare Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Problem 23 . . . . . . . . . . . . . 23 . . . . . . . . . . . . . 26 . . . . . . . . . . . . . 28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 30 30 33 34 35 36 41 41 43 47 54
6 ChoiceBased Approach 6.1 Choice Structures . . . . . . . . . . . . 6.2 Weak Axiom . . . . . . . . . . . . . . . 6.3 Relationship with the Law of Demand 6.4 Strong Axiom . . . . . . . . . . . . . .
7 Integrability 55 7.1 Slutsky and Hicks Compensation . . . . . . . . . . . . . . . . 55 7.2 Aside: Dot Product . . . . . . . . . . . . . . . . . . . . . . . . 57 7.3 Substitution Matrix . . . . . . . . . . . . . . . . . . . . . . . . 58 1
7.4 Substitution Matrix with Preference . . . . . . . . . . . . . . . 62 7.5 Integrability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 8 Aggregation 71 8.1 Aggregate Demand Function . . . . . . . . . . . . . . . . . . . 71 8.2 Representative Consumer . . . . . . . . . . . . . . . . . . . . . 76 8.3 Failure of the Weak Axiom . . . . . . . . . . . . . . . . . . . . 80 9 Expected Utility 9.1 Lotteries . . . . . . . . . . 9.2 Preference over Lotteries . 9.3 Expected Utility Theorem 9.4 Paradoxes . . . . . . . . . 9.5 StateSpace Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 84 86 88 94 98
10 Risk 10.1 Money Lotteries . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Risk Attitude . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Stochastic Dominance . . . . . . . . . . . . . . . . . . . . . 11 Pro…t Maximization Problem 11.1 Production Set . . . . . . . 11.2 Transformation Function . . 11.3 Pro…t Maximization . . . . 11.4 Law of Supply . . . . . . . . 12 E¢ ciency of Aggregate 12.1 E¢ cient Production . 12.2 Cost Minimization . 12.3 Aggregate Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
100 . 100 . 101 . 107 111 . 111 . 114 . 116 . 119
Supply 120 . . . . . . . . . . . . . . . . . . . . . . . 120 . . . . . . . . . . . . . . . . . . . . . . . 122 . . . . . . . . . . . . . . . . . . . . . . . 125
13 Partial Competitive Equilibrium 126 13.1 Competitive Equilibrium . . . . . . . . . . . . . . . . . . . . . 126 13.2 Partial Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . 129 13.3 The Long Run . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 14 Welfare Analysis 132 14.1 Pareto E¢ ciency and Surplus . . . . . . . . . . . . . . . . . . 132 14.2 E¢ ciency of Competitive Equilibrium . . . . . . . . . . . . . . 134 2
. . . . . . . . . . . .1 Externalities . . . . . . . . .1 CapacityConstrained Pricing . . . . . . . . . . . . . . 153
16 Monopoly and Product Di¤erentiation 16. .1 Monopoly . 142 145 . . . . . We will also look at connections with the "revealed preference" approach that makes assumptions directly about choices. . . . . . . . . . . . . . .4 Product Di¤erentation . . . . . . . . . . . . . 156 17. . . . 15. . . . . . . . . . . . . . 16. . . A consumption bundle is a list
3
. . . . .3 Competitive Limit . 136 . 161 18 Precommitment and Entry 162 18.14. . . . . 162 18. . Let there be L di¤erent commodities. . . 150 . . . . . . . . . . . . . . . . 168
1
1. . . . . . . . . . .1
Preference
Consumption Set
In the …rst part of these lectures. . . 16. . . 158 17. . . . . . 135 15 Externalities 15. 16.
17 Capacity Constraints 156 17. . . . . . . . . . . . . . . . . . . .3 E¢ cient Allocations through the Market Mechanism . . . .2 Ine¢ ciency . . . . . . . . 148 . . . . 139 . . . . . . 138 . 145 . . 166 18. . . . . . . . . . . . . . . . . . . . . . . . Most of our attention will be on the primitive approach that starts with "rational" individual preferences and derives optimal choices for given prices and endowments. . . . . . . . . . . . . 15. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Socially Optimal Entry . . . . . . . . . . . . .1 Precommitment . . . . . . . . . . . .2 Entry Equilibrium . . . we consider the consumer decision problem in a market economy. . . . . . . .3 Remedies . .4 Public Goods . . . . . . . . . . . . 136 . . . . . . 15. . . . . . . . . . . . . . . . . . . . . . . . where goods are o¤ered at posted prices. . . . . . . . . . . .3 Repetition . . . .2 Bertrand Price Competition . .2 Cournot Quantity Competition . .
Example. not all vectors) can be feasibly consumed. xL in the commodity space RL . xL of quantities of each commodity. we cannot consume in two places simultaneously. Suppose the commodities are time (in minutes) spent watching television. the description of a commodity may have to include detailed context information. Note that RL includes consumption bundles that list negative quantities of some goods. This assumption rules out indivisible commodities. 7 x=4 . But one cannot watch television 4
.e. Strictly speaking. shelter) to survive. and time (in minutes) spent in a rollercoaster car on "The Beast" (Kings Island near Cincinnati). 5 . and time is limited. Any two items that could sell at di¤erent prices should be modeled as distinct commodities. we have to consume enough of certain things (food. It is possible to watch no television and ride "The Beast" once (call this bundle x). such as "a diet coke can from a vending machine at Grand Central Station in the summer of 2008. represented by a vector 2 3 x1 6 . if x and y are feasible. or we may wish to impose restrictions in a particular model. or the consumption set. a ride which takes approximately …ve minutes. x2 . It may be the case that not all such objects (i. These can be interpreted as debts or giving some of one’ s endowment to others. and it is also possible to watch television for one hour and not ride "The Beast" (call this bundle y). That is. is convex. Physical constraints include: we need time to consume. : : : . 1) is also feasible.x1 ." The de…nition of the commodity space tells us what kind of object a consumption bundle is. then any mixture z = x + (1 ) y with 2 (0. A restriction we will impose for modeling purposes is that the set of feasible consumption bundles.
x` 0 for ` = 1. we can think of X more generally as a choice set.2
Rational Preference
The …rst step in analyzing individual choices from the consumption set is to de…ne preference over its elements. y) 2%. We interpret (x. even though many commodities are actually indivisible.t. Or they may be something else one can express a preference about. a % xg fb 2 X s..t. and propositions for the idealized world are also informative about reality.
1.. Any economic model is an idealized representation of reality. Most of the time. they preserve the spirit of the problem.. where the elements may in particular be consumption bundles. x 2 RL s. : : : . for example sports teams. Convexity of the consumption set is a convenience restriction that should not alter the thrust of our results. For the moment. the consumption set for these two commodities is not convex. Since this bundle (call it z) is a mixture of x and y.for half an hour and ride "The Beast" for twoandahalf minutes (unless it is a oneperiod model . which is commonly denoted x % y. as "x is at least as good as y. the assumptions limit the applicability of the model to a more speci…c problem.(x) fa 2 X s. L . A preference % is a binary relation on X: a subset of X X.t. x % bg :
The intersection of the upper and lower contour sets is the indi¤erence set at x: (x) fb 2 X s." The preference % associates with every x 2 X its betterthan (or upper contour) set and worsethan (or lower contour) set: % (x) . We will make many assumptions that are in some sense too strong to ever hold exactly.). we will assume that the consumption set X contains all bundles with only nonnegative quantities: X = RL + which is convex. In some cases.t. In others. x bg : 5
.
then y for z.2).B. especially when the choice objects are complex and unfamiliar. y 2 X. Suppose the agent initially has x. However. This scenario could be accommodated by statedependent preference. 1. Moreover. But it is never the case that the agent could not say whether he prefers x or y (or …nds them indi¤erent). Preference % is rational i¤ (i) % is complete: 8x. or else they are indi¤erent. If these trades are repeated often enough. people often violate transitivity unwittingly. y. then 8x.B. y. Transitivity is most closely related to the usual notion of rationality: it requires the agent to rank alternatives consistently and predictably. e. a transitivity violator is exposed to "Dutch book" trades. In practice. 6
. x % y % z =) x % z.g. x y % z % x. one could o¤er z for x. (ii) relations and are transitive. Completeness says that any two elements of X can be compared: one is always preferred. De…nition. most would revise a stated preference when one points out to them that it is intransitive. preferences are assumed to be rational in the following sense. but he has made a payment. (ii) % is transitive: 8x. Since the preference is cyclical. they will bankrupt the agent. Exercise 1 (MWG 1. Which leaves the agent with the initial bundle x. for example because he is seeking more information. z 2 X. where the state is de…ned by what the agent learns. z 2 X (i) x x.From % derive the strict preference and indi¤erence relations x x y i¤ x % y and not y % x y i¤ x % y and also y % x:
and
:
In almost all economic theory. and the agent should be willing to pay some nonzero amount to exchange y for x. (iii) x y % z =) x z.1. x % y or y % x (or both). then x for y (at a price). Show: if % is rational.
we have u (x) u (y) u (z). x % y () u (x) u (y) :
If u represents %. then any strictly increasing function f : R ! R can be composed with u to give a new function v : X ! R (where v (x) = f (u (x)) for all x 2 X). "if % is not rational. Exercise 2 (MWG 1. with the property that v (x) v (x0 ) () 0 u (x) u (x ). So existence of a utility function implies that % is rational. then there exists a utility function u : X ! R that represents %.3
Utility Functions
To apply tools from calculus in analyzing choices. utility functions are nonunique. Proposition. Suppose a utility function u : X ! R represents preference % on X.5).B. Then % is complete: for any x. It never exists for a preference that is not rational. y 2 X. it is useful to represent the preference relation by a function. The function u : X ! R is a utility function that represents preference relation % if 8x. y 2 X. The contrapositive. On the other hand. then there does not exist a utility function.
De…nition. hence u (x) u (z) and then x % z. Thus v also represents %. Only a rational preference relation can be represented by a utility function. we have u (x) u (y) or u (y) u (x)." follows. Proof. Clearly.
7
. a utility function may fail to exist. so x % y or y % x. Show: if X is …nite and % is a rational preference relation on X.1. And % is transitive: when x % y % z.
n=1 n=1 8
.(x). so that % (x) is closed. When is % represented by a utility function u that is twice di¤erentiable and has a (unique) maximum? Conditions for di¤erentiability are fairly di¢ cult to establish. Informally. a continuous preference is equivalently de…ned by the closedness of all lower and upper contour sets. De…nition. so we focus on the related. Informally. Preference relation % is continuous if x % y whenever x and y are the limits of sequences fxn g1 and fy n g1 such that xn % y n for all n=1 n=1 n. under limits: whenever x is very close to y in the domain.
De…nition.2
2. too. n=1 Then continuity requires y % x. di¤erentiability requires the absence of kinks in the utility function. In addition to continuity. In particular. . Thus. the lower contour set. a continuous preference ranks x and y the same as it ranks objects that are very similar to x and y. f (x) is very close to f (y) in the image. In the context of functions. By the analogous argument. The converse is true. the upper contour set. x is preferred to y if objects that are very similar to x are preferred to y. we discuss which properties of preferences ensure the existence of the type of utility function to which the standard optimization techniques apply. but weaker. The function u : X ! R is continuous if f (x) is the limit of sequence ff (xn )g1 whenever x is the limit of sequence fxn g1 X. is also closed. suppose fy n g1 converges to y. continuity preserves closeness. and y n % x for all n. rather than preference. is also in % (x). Hence the limit of every convergent sequence in % (x).1
Utility
Continuity
In this lecture. property of continuity.
Monotonicity implies: for every + 0 such that x % e and < 1 such consumption bundle x. 1) 2 RL be a vector of 1s.The proof of existence of a continuous utility function is simpli…ed by (but actually valid without) imposing a property called monotonicity on preferences. A (x) and A+ (x) are closed. Let e = (1. sequences of bundles that n=1 have equal quantities of all commodities). Proof. Assume that preference % is continuous (and monotone) on RL . Given x.e.
De…nition. It rules out bads. which is a connected set. : : : . since . but this is not a real problem. there exists that e % x (just let = 0 and large enough so that all quantities in e are greater than the corresponding quantities in x). Proposition. Any continuous rational preference relation can be represented by a continuous utility function. Monotonicity welcomes more of anything. y =) x y. desirable only up to a point (you may like some ice cream. since bads can be relabeled as goods (e. e % xg and A (x) f 2 R+ s. but not tons of it delivered to your home). Moreover. L.(i. x e
. and together cover R+ . To sum. A (x) and A+ (x) are nonempty. however. Its de…nition presupposes that the consumption set is X = RL .t. x % eg are nonempty.t. Many things are. Preference relation % is monotone if x
Monotonicity is a fairly strong assumption. + We will construct utility values and show that the resulting function represents % and is continuous. : : : . closed. so that preference must be preserved under limits of sequences f n eg1 .(x) and % (x) are closed. we have 2 A (x) or 2 A+ (x) for any 2 R+ by completeness.g. + 9 2 R+ s. This means that the sets A+ (x) f 2 R+ s. it is inconsistent with limited wants. + The notation x y stands for x` > y` for ` = 1. Hence A (x) \ A+ (x) = is nonempty for all x 2 RL . so R+ A (x) [ A+ (x). replace "waste" with "waste disposal"). perhaps a lot.t.
n 1 Therefore fu (x )gn=1 must have a convergent subsequence. Continuity of u requires: for any sequence fxn g1 with limit x. With lexicographic preferences. etc.g. ] where = 0 and is the highest quantity of any commodity in fxn g1 n=n(") . which con‡ icts with ^ e u (x) e. if u (x) u (y). we have. Example. and also ^ e u (x) e.g. there exists an ordering of commodities.C. Then by monotonicity. and u (x). is preferred to y such that n=1 (1) (2) y(1) = 0 and y(2) = 1.2). But x u (x) e. 0 0 00 since e x for all > and x e for all < . Then xm(n) u xm(n) e ^ e. so x % y by monotonicity. Analogously. suppose there is a 1 n=1 convergent subsequence u xm(n) n=1 (m is an increasing function) that converges to 6= u (x). We may therefore de…ne a function u : X ! R+ by u (x) = 2 A (x) \ A+ (x) for all x 2 X. every convergent subsequence has limit u (x). the sen=1 quence fu (xn )g1 converges to u (x). every bundle in the sequence 1 fxn g1 . u xm(n) > ^ for all Because u xm(n) ! n > N . 10
. > u (x). where xn = n and xn = 0 for all n.) These preferences violate continuity: e. then % is continuous. because % is continuous and % (x) therefore closed.Monotonicity implies further that there is only one such that x e. for some N . which implies x % ^ e. Hence fu (xn )g1 n=n(") lies in a compact set: namely. < u (x) leads to a contradicton. in the interval [ . Prove the converse: if a continuous utility function represents %. (As in alphabetic entries in a telephone book or dictionary. then e u (x) e by monotonicity. To see this. Exercise 3 (MWG 3. so u (x) u (y) : Conversely. which is therefore the limit of the sequence fu (xn )g1 . n=1 n=1 there exists for any " a number n (") such that kxn xk < " for n n ("). Note that. 0) is worse than y. Furthermore. then e e. If e. Lexicographic preferences are not continuous and do not admit a utility representation. Suppose x % y and x = e and y = e. because fxn g1 converges. It remains to be shown that u represents % and is continuous. x(1) = y(1) and x(2) > y(2) =) x % y. but the limit x = (0. where ^ = 1 ( + u (x)) lies between 2 > ^ . : : : . so that x(1) > y(1) =) x % y (where (1) refers to the highestpriority commodity).
e. Exercise 4 (MWG 3. > u (x) > u (y 0 ) > 0 ). Whatever values u (x) and u (y) > u (x) the utility function assigns to x and y. But the utility values come from the (in some sense smaller) set R. z 2% (x) =) 8 2 [0.2
Quasiconcavity
Besides continuity (di¤erentiability). One can understand the argument intuitively by noting that. Preference relation.4). y(2) = 2. y + (1 ) z x. with lexicographic preferences. % is convex if y % x and z % x imply 8 2 [0. Lexicographic preference implies > 0 =) 0 r ( ) > r ( 0 ) (since x with x(1) = is preferred to y 0 with y(1) = 0 . every point of RL . we can …nd a rational number r ( ) 2 (u (x) .Suppose u is a utility function representing lexicographic preference %.e.e.% is convex if 8x 2 X the upper contour set % (x) is convex: y. i. the indi¤erence sets are singletons.
2. we would like to know whether the utility function has a maximum. This property (quasiconcavity) relates to the convexity of preference. u (y)). i.
11
. Hence the function r maps onetoone from R (an uncountable set) to Q (a countable set). 1). y + (1 x.C. provided y 6= z. which allows us to apply calculus in solving the utility maximization problem. Let x(1) = y(1) = and x(2) = 1. 1] y + (1 ) z 2% (x) :
I. so that a di¤erent utility value has to be assigned for every bundle. Find a preference relation that is not continuous but has a utility function that represents it. 1].
)z %
Preference relation % is strictly convex if y % x and z % x imply 8 2 (0.
De…nition. a contradiction.
1) u ( x + (1 provided y 6= z. By the same token.1). u (y)g :
Utility function u is strictly quasiconcave if 8x. I am willing to trade aggressively (give up a lot) for a more balanced bundle. y 2 X and 8 2 (0. The graph of a convex function always lies above a line segment between two of its points. Quasiconcavity is weaker than concavity. a convex function satis…es: 8x. u (y)g .C. then I …nd a mixture of the two (a more balanced bundle) more appealing.Convex preference can be interpreted as a desire for variety: if I like two bundles equally. Utility function u is quasiconcave if 8x 2 X the upper contour set % (x) fa 2 X s. 1) u ( x + (1 ) y) u (x) + (1 ) u (y) min fu (x) . u (a) u (x)g is convex. which decreases as I acquire more. u (y)g : (Why are the de…nitions equivalent? Suppose x % y. Show that lexicographic preference is rational. y 2 X and 8 2 [0. Exercise 5 (MWG 3. suppose x. hence u ( x + (1 ) y) u (y). monotone and strictly convex. y 2 X and 8 2 (0. I dislike extremes. y 2% (z). and note that u ( x + (1 ) y) min fu (x) . then convexity of % (y) implies x + (1 ) y 2% (y). u (y)g u (z) implies x + (1 ) y 2% (z). This results in a "diminishing marginal rate of substitution": a high relative valuation for commodities of which I have little. Conversely.
. The graph of a quasiconcave function only has to lie above the lower of the two points.) Of course. De…nition. If I have a lot of one commodity and little of the others.t. 1] u ( x + (1 ) y) min fu (x) . which is easily apparent from an equivalent de…nition: u is quasiconcave if 8x. 12 ) y) > min fu (x) . Figure 1 illustrates the concave and quasiconcave cases in panels (a) and (b).
and x + (1 ) y % y whenever x % y. Consider a weakly convex preference relation. If a utility function represents a (strict) convex preference relation. x + (1 )y % x whenever y % x. I demonstrate that it implies quasiconcavity (the strict case is analogous). the global maximum may not be unique if it is part of a plateau. To show: u ( x + (1 or u ( x + (1 ) y) u (y) for all x. resulting in a "gap" (nonconvexity) in the upper contour set. Proof. y 2 X. If u represents 13 ) y) u (x)
. If there existed a greater maximum or a minimum. With convex preference. 1).Figure 1: Concave and quasiconcave functions
Quasiconcavity is su¢ cient to guarantee that a local extremum is a global maximum. then the utility function would have a trough somewhere and increase on both sides of it. with 2 (0. Proposition. However. then it is (strictly) quasiconcave.
C. the CES utility function represents in the limit the same preferences as the CobbDouglas utility function u (x) = x1 1 x2 2 . : : : . y 2 X. x2 g. 0. y 2 X. and u ( x + (1 u (x) u (y).
14
. x + e1 x.5). the graphs of the indi¤erence. 8 > 0. Preference relation.%. the …rst inequality) if u (y) (i.) Exercise 8 (MWG 3. (b) Let e1 = (1.e.
) y) ) y)
u (x) (i. Show: a continuous preference is homothetic i¤ it admits a utility function that is homogeneous of degree 1.sets are linear. Exercise 7 (MWG 3. : : : . xL ).Preference relation % is homothetic if 8x. 0) denote the bundle that contains 1 unit of commodity 1 and none of the other commodities. then % is locally nonsatiated.1).6). (a). (x + e1 ) (y + e1 ).e. the second inequality) if
Exercise 6 (MWG 3. Show: a continuous preference is quasilinear with respect to commodity 1 if it admits a utility function of the form u (x) = x1 + (x2 . the CES utility function has in the limit the same indi¤erence sets as the Leontief utility function u (x) = min fx1 . y x =) y x. (No need to show a converse. (c) As ! 1.C.% is quasilinear with respect to good 1 if 8x. and x y =) 8 0. Consider preferences in a twocommodity world represented by the CES (constant elasticity of substitution) utility function u (x) = ( 1 x1 + 2 x2 )1= : Show: (a) When = 1. and x y =) 8 2 R. (b) As ! 0. it must satisfy u ( x + (1 u (y) u (x). Show: if % is monotone. Utility function u is homogeneous of degree 1 if 8 > 0 u ( x) = u (x). x y.B.
then 8 2 [0. and h hours of leisure.1
Demand I: Utility Maximization Problem
Budgets
A consumer is constrained in her choices by a limited budget. what is the Walrasian budget set? 15
. For an individual who consumes amount x of a commodity priced at p. p x + w :
Notice that a decline in any price enlarges the budget set: more bundles satisfy the constraint. 1] we have p x00 = p x + p (1 ) x0 w. the budget set is bounded by an (L hyperplane. de…ned by p x = w. More generally. 7 p = 4 . the budget set is the area under the budget line p1 x1 + p2 x2 = w. is called the budget set.3
3. . In a twocommodity world. and commodities sell at market prices 2 3 p1 6 . : : : . L (you can at most spend all your wealth on one commodity) Exercise 9 (MWG 2. 1)dimensional
The budget set is convex: if p x w and p x0 w.2). for given prices and wealth. De…nition. It is also compact: closed and bounded. The Walrasian budget set is the set of all bundles x 2 RL + a¤ordable with wealth w at prices p: Bp. If she has wealth w.w = x 2 RL s. since x` w=p` for ` = 1. 5 2 RL . ++ pL then her chosen bundle x 2 RL must satisfy + p x = p1 x1 + + pL xL w:
The set of bundles that meet the budget contraint. when the hourly wage is 1.D.t.
The map from prices and wealth to the set of utilitymaximizing consumption bundles in the associated budget set is called the Walrasian demand correspondence. : : : . The consumption set is X = RL . from the previous lecture.t. L.3.e. g (x) p x w 0. In the UMP. and locally nonsatiated . : : : .2
The Utility Maximization Problem (UMP)
Let the consumer preferences be rational." which we will get to later on). and the h` (x` ) x` 0. the solution can be stated in terms of KuhnTucker conditions: for ` = 1. the consumer picks the mostpreferred bundle in her Walrasian budget set. @x`
where
0 and the
`
0 are Lagrange multipliers (shadow prices). This set depends on the parameters of the budget constraint: commodity prices and individual wealth.which implies. L are the nonnegativity constraints. the interior …rstorder condition holds). that there is a continuous and quasiconcave utility function representing it.
is the budget constraint. nonempty set has a maximum. all of which give maximal utility. If the relevant constraint is nonbinding. which gives utility max u (x) s. a continuous realvalued function on a compact. + The "utility maximization problem" is one of two equivalent ways to frame the consumer choice problem (the other is the "expenditure minimization problem. 16
. Therefore the UMP has a solution as long as preferences are continuous. A solution to the UMP is in principle a set of consumption bundles in the budget set. respectively ` = 0 (i.
` = 1. = 0. p x w:
x 0
By the Weierstrass theorem. continuous. @g (x) X @u (x) = + @x` @x` `=1
L `
@h` (x) . In general.
the slopes have to be equal. Consider preferences represented by the CobbDouglas utility function u (x1 . x2 ) = ln u (x1 . then so 2 must u (x1 . x2 ) = ln x1 + (1 ~ ) ln x2
17
. Example. the slope of the indi¤erence curve (solve du = M U1 dx1 + M U2 dx2 = 0 for dx2 =dx1 ).
@u(x) @xL
3
7 5:
In a twocommodity world.e. where the budget constraint binds and both goods are consumed. if u represents preferences. and solving for dx2 =dx1 . Tangency is necessary for an interior optimum. The right side is the slope of the budget line (found by totally di¤erentiating the budget line constraint. x2 ) = x1 x1 . the budget constraint is needed to …x the solution. i. @u (x) =@x2 p2 which is incidentally the tangency condition familiar from diagrams in intermediate micro books. i. p1 dx1 + p2 dx2 = 0. these conditions imply p1 @u (x) =@x1 = . See Figure 2.e.The conditions reduce to @u (x) = p` @x` or equivalently @u (x) @x` p` p . . To be tangent.). but there are in…nitely many points satisfying it for di¤erent levels of wealth. Therefore. Note that. The left side is the marginal rate of substitution. More concisely we can write this as 5u (x) (with equalities where x` > 0) in terms of the gradient vector 2 6 4
@u(x) @x1
5u (x)
. .
with equality if x` > 0.
x2 . since the logarithm is an increasing transformation. Thus. the constrained choice is the solution to the Lagrangean problem max L (x1 .Figure 2: Optimal choice at the point of tangency
represent them. x2 ) is strictly increasing in x1 and x2 .
which satis…es …rstorder conditions @L (x1 . ) 1 = p2 = 0 @x2 x2 @L (x1 . x2 . ) = p1 = 0 @x1 x1 @L (x1 . ) = w p1 x1 p2 x2 = 0: @ The …rst two reduce to 1 x2 p1 = .x2
ln x1 + (1
) ln x2 + (w
p1 x1
p2 x2 ) . x1 p2 18
. ) =
x1 . Because u (x1 . the budget constraint ~ p1 x1 + p2 x2 w
must bind. x2 . x2 .
and if u is strictly quasiconcave. p x = w. Local nonsatiation (which also has the e¤ect of ruling out thick indifference sets) is a more plausible property than monotonicity. actually su¢ ces. w) is convex. If u is in addition quasiconcave. w). 19
. w). Walras’law is satis…ed if the consumer’ choice exhausts the budget. ) w : p2
w and x2 = (1 p1
If preference is monotonic. we have w hence x1 = p1 x1 = p2 x2 . w) satis…es Walras’ law if 8p 0. local nonsatiation. then x (p. w) = k x (p. w) the Walrasian demand correspondence is homogeneous of degree zero and satis…es Walras’law.% is locally nonsatiated if 8x 2 X and 8" > 0.8w > 0 and 8x 2 x (p. De…nition. Preference relation. then x (p.Thus p2 x2 = (1 ) p1 x1 : From the budget constraint (the third …rstorder condition). De…nition. But a weaker property. since it allows for limited wants. since all commodities are valuable. the budget will always be exhausted. The Walrasian demand correspondence x (p. If u is a continuous utility function that represents locally nonsatiated preference relation % on X = RL . + 8x 2 x (p. 8w > 0. Proposition. s The demand correspondence is homogeneous of degree k if x ( p. 9y 2 X such that ky xk " and y x. then 8p 0. w) is a singleton.
w) : Exercise 10 (MWG 3. Informally it says that solutions in one constraint set should still be solutions at a very similar constraint set.Proof. then 8p 0. then u (x00 ) = u ( x + (1 ) x0 ) > min (u (x) . so that x0 is in the budget set. x (p. x0 2 x (p. But then x is not a utilitymaximizing choice.t. If u is strictly quasiconcave and x (p. If we pick a su¢ cienty small neighborhood. w) are equally preferred. which contradicts x. then Hence x (p.D. w) =% (x) = fa 2 X s. Verify that the above proposition holds for the Walrasian demand function with CobbDouglas utility. If p x < w for x 2 x (p. because it is convex) is strictly preferred to both x and x0 . w). then p x0 < w. If u is quasiconcave. If u is a continuous utility function that represents locally nonsatiated preference relation % on X = RL . w) inherits the convexity of % (x). Since all elements of x (p. n=1 Proposition. The Walrasian demand correspondence is upper hemicontinuous if 8 (p. x 2 x (p. this implies x00 (which is in the budget set. then by nonsatiation there exists in every neighborhood of x an x0 such that x0 x. then its upper contour sets are convex. w) whenever (p. a construction. u (x0 )). 20
. u (x0 )) : x (p. u (a) u (x)g :
But since u (x) = u (x0 ) = min (u (x) . w). w) is also the upper contour set of any of its members: if x 2 x (p. 8w > 0 the + Walrasian demand correspondence is upper hemicontinuous.1). The following is a natural generalization of the continuity concept for pointvalued functions to setvalued correspondences. De…nition. w) and x are limits of sequences f(pn . Homogeneity of degree zero follows from the fact that the budget constraint is unchanged when prices and wealth are scaled by : p x w () p x w:
Walras’law re‡ ects nonsatiation. wn ) for all n. w) has two distinct elements x and x0 . wn )g1 n=1 and fxn g1 such that xn 2 x (pn . w).
the result implies that. which implies that the utilitymaximizing choice lies in the boundary of the budget set. and continuous in p and w. i. By continuity of u. it must be the case for all su¢ ciently large n that y 2 Bpn . Since (pn . and we have xn 2 x (pn .t. 8w > 0. strictly increasing in w and nonincreasing in p` for ` = 1. Since wealth increases and price decreases enlarge the budget set. so that utility associated with it cannot decrease.3
Indirect Utility Function
The value of the utility function at a solution to the UMP is the highest it can attain on the budget set. then 8p + indirect utility function is homogeneous of degree zero. The indirect utility function is homogeneous of degree zero because the Walrasian demand correspondence is: since the scaling of prices and wealth does not a¤ect the set of utilitymaximizing choices. there exists also y 2 Bp.Proof.e. w) s. w) = v (p. a contradiction. But as fxn g1 n=1 converges to x. v (p.wn .w such that u (y) > u (x).e.
Of course. v ( p. : : : . but x 2 x (p.wn . the indirect utility function must strictly increase in wealth. this argument leads to u (x) u (y). it cannot a¤ect the utility derived from them. 21
. if the Walrasian demand correspondence is in fact a function. the nonsatiated preference relation % on X = RL . w). w).
3. L. then this function is continuous. w) n=1 n=1 and x.w such that u (x0 ) > u (x). w) vg is convex. wn ) converges to (p. Because of local nonsatiation. Suppose sequences f(pn . w). If u is a continuous utility function that represents locally 0. quasiconvex. wn ) for all n. since xn is in the choice set for Bpn . wn )g1 and fxn g1 converge to (p. The map from prices and wealth to the highest attainable utility value is called the independent utility function. the best available choice can only improve. Then there = exists x0 in Bp. This implies u (xn ) u (y). for a particular set of prices and wealth. Proof.
Proposition. i. The indirect utility function v is quasiconvex if 8v the lower contour set f(p.
v (p. and because v was arbitrary. (c) Verify that the general properties of Walrasian demand and indirect utility functions hold in this case. w) v and v (p0 . Exercise 11 (MWG 3.w0 . i. (b) What are the …rstorder conditions for the UMP? Derive Walrasian demand and the indirect utility function. In the second case. so u (x) v (p.w .To establish quasiconvexity.6). the maximal utility attainable in the budget set de…ned by prices p00 = p + (1 ) p0 and wealth w00 = w + (1 For any x in this budget set.2). w). x 2 Bp. let preferences be represented by the utility function u (x) = (x1 b1 ) (x2 b2 ) (x3 b3 ) :
(a) Why is there no loss of generality from imposing + + = 1? Assume this for the remaining parts. Verify that the above proposition holds for the inverse utility function with the log transformation of CobbDouglas utility. Exercise 12 (MWG 3. In a threecommodity setting. In the more restrictive case that u is strictly quasiconcave. so that it is also continuous. ) w0 :
so p x w or p0 x w0 must be true.D. w00 ). w) is continuous. suppose v (p. all bundles in Bp00 . w0 ). the lower contour set at v. w) s. w0 ) v and considerv (p00 . x (p. and u (x (p. x 2 Bp0 .w00 belong to the lower contour set at v.
22
. Because x was arbitrary. w)) is a composition of continuous functions. so u (x) v (p0 . w) vg. v is quasiconvex. In the …rst case.e. x 2 f(p. p x + (1 ) p0 x w + (1 ) w0 . Thus u (x) v.D.t.
u (x)
We assume p 0 and u > u (0) throughout (so that at least one commodity must be consumed. + In parallel to the UMP. @x`
where 0 and the ` 0 are Lagrange multipliers (shadow prices). and none in in…nite quantity).e.1
Demand II: Expenditure Minimization Problem
EMP and Hicksian Demand
In the UMP ("utility maximization problem"). we assume that u represents a continuous. a consumer picks the consumption bundle for which expenditure is min p x s. If the utility constraint is nonbinding at x. locally nonsatiated preference on RL . In the EMP. then ` = 0. + The set of consumption bundles that are solutions to the EMP at prices p and required utility u is denoted as h (p. u (x)
x 0 x 0
u u:
= max ( p x) s. a set in which all bundles yield at least a certain utility). If the `th nonnegativity constraint is nonbinding at x. i. Argue that a solution to the EMP exists if p 0 and u (x) u for some x 2 RL .3). we apply the KuhnTucker conditions: at a solution x . L. : : : . an optimal choice maximizes utility on a …xed budget set.e.t. This approach is called the EMP ("expenditure minimization problem"). we have for ` = 1. if singlevalued). u) RL and called the + Hicksian demand correspondence (or function. x` > 0. and is di¤erentiable. 23
. Analogously. u (x) > u. we have = 0.4
4. we can de…ne optimal choice as minimizing expenditure on a …xed upper contour set (i. Furthermore.e. Exercise 13 (MWG 3.t.E. i. @ (u u (x )) X @ (p x ) = + @x` @x` `=1
L `
@ ( x` ) .
since expen2 diture p x is strictly decreasing in x1 and x2 . The constrained choice therefore satis…es the "tangency" condition. (after rearranging and relabeling the multiplier ~ @u (x ) @x` with equality if x` > 0 (so that
`
1= )
~ p`
= 0). Example. p` = @u (x ) @x`
`.
u
. u (x) = u (0) if x1 or x2 is zero.
i.) Moreover. From the utility constraint. we have x 1 x 21 hence x2 = x2 x1 24 = u.
This is exactly analogous to the …rstorder conditions in the UMP.Resolving the derivatives. which specializes to x2 p1 = : 1 x1 p2 given the CobbDouglas marginal utilities. the solution x is again characterized by the tangency condition: @u (x ) =@x1 p1 = : @u (x ) =@x2 p2 The only di¤erence is that the remaining constraint that …xes x is now not the budget constraint. where the utility constraint binds (so that ~ < 1) and both goods are consumed. (If it did not bind. you could slightly lower consumption of one or both of the goods and attain a lower expenditure within the utility constraint. Reconsider preferences represented by the CobbDouglas utility function u (x1 . so they must be strictly positive for x to attain u > u (0). The utility constraint has to bind. x2 ) = x1 x1 . but the utility constraint u (x ) = u.e. In a twocommodity world.
) No excess utility re‡ ects continuity of the utility function. u (x) = u. Proposition. a contradiction. then the Hicksian demand correspondence h (p. u) : (It is only the expenditure that changes. u). u). u) is on RL . Proof. But the consumption bundle x may change so as to make the increase in expenditure as small as possible. Homogeneity of degree zero follows from the fact that the upper contour set at u (the constraint set). and therefore any solution to EMP. expenditure is implicitly adjusted as needed in order to keep utility constant. u) such that u (x) > u. u) () x 2 h (p. then upper contour sets are convex. which is inherited from preferences. Let x. is una¤ected by scaling prices: x 2 h ( p. not the bundle that minimizes it the scaling leaves relative prices unaltered. then we could construct a scaleddown bundle x0 = x with 2 (0. x0 2 h (p. and singlevalued if preference is strictly convex. Hicksian demand is also called compensated demand. The Hicksian demand correspondence is said to have no excess utility if 8x 2 h (p. If there were a solution x 2 h ( p. so x00 = x + (1 25 ) x0
. 1) that satis…es p x0 < p x (since x0 x) and u (x0 ) u (for close enough to 1. Hicksian demand has properties that correspond to those of Walrasian demand. locally nonsatiated preferences 0. is convex if preference is convex. so that p x = p x0 . In this sense. has no excess utility. If preference is convex (utility quasiconcave). But this means x0 . because if prices increase. and p + homogeneous of degree zero in p. by continuity). can be a solution to EMP at u.and x2 = 1 p1 p2 u and x1 = 1 p2 p1
1
u:
Notice that x1 and x2 are functions of prices and the required utility (not of wealth) in the EMP. The wealth that is needed to attain u is allowed to vary. If u represents continuous. and not x.
i. then x00 x (and x00 x0 since h (p. Suppose e (p. u) were not strictly increasing in u.e. so that the expenditure function is homogeneous of degree one in p. so x0 . and p + of degree one in p. If preference is strictly convex (utility strictly quasiconcave). then the expenditure function e (p.e. u). The expenditure function is homogeneous of degree one in p if 8 > 0. 1) such that x00 x.
Proposition. p x00 = p x + p (1 ) x0 = p x + (1 ) p x0 = p x:
It follows that x00 also minimizes expenditure and belongs to h (p. Since the constraint set is not a¤ected ` 26
. 1) that satis…es u (x) > u0 and p x < p x00 p x0 (since x x00 ). x 2 h (p. u).) Continuity implies there is a bundle x = x00 for some 2 (0. u) were strictly decreasing in p` for some `.
4. then e (p00 . = a contradiction. concave in p. then there would exist solutions 0 x and x00 . u) satis…es no excess utility. u) must be a singleton. u) and called the (minimum) expenditure. u) cannot exist. Since scaling up the price does not change the constraint set (i. Proof. u) < e (p0 . Moreover. respectively at u0 and u00 > u0 .e. x00 6= 0. Hence e ( p. so that x0 x). we have p x00 < p x. strictly increasing in u and nondecreasing in p` for ` = 1.attains utility u. I. u). u) = e (p.e. u) is homogeneous on RL . u). (We maintain u > u (0). Hence two such elements x. If u represents a continuous. Continuity implies there exists 2 (0. u). If e (p.2
Expenditure Function
The value of p x at a solution x to the EMP is denoted as e (p. L. it does not a¤ect the solution x . u ( x00 ) > u (x). u) = p x = e (p. But since x00 x00 and p x00 = p x. which implies that x does not minimize expenditure on the constraint set. i. the upper contour set at u). locally nonsatiated preference 0. and continuous in p and u. x0 of h (p. e ( p. such that p x0 p x00 > 0. But then x0 cannot be a solution to EMP. : : : . if we compare expenditure at two price vectors p0 and p00 di¤ering only in that p00 p0` . and h (p.
E.E. u) = p00 x00 = p x00 + (1 e (p.
27
. Exercise 16 (MWG 3.by the price di¤erence. Hence the expenditure function is concave in p. u) increases less than linearly in p. derive the Hicksian demand function and the expenditure function. The consumer can always attain u at an expenditure no greater than e (p0 . u) + (1 ) p0 x00 ) e (p0 . For the CES (constant elasticity of substitution) utility function u (x) = (x1 + x2 )1= . show that the Hicksian demand functions for the remaining goods are invariant to u. but may be able adjust x to x0 to attain u at reduced expenditure. u).
since x00 is available at prices p and p0 (so expenditure with prices p and p0 at the minimizing bundles x and x0 cannot be larger then at x00 ). the same x is a solution with both p and p0 . then e (p00 . Con…rm that the general properties of the Hicksian demand function and the expenditure function hold with CobbDouglas preferences. Exercise 14 (MWG 3. u). and verify their general properties in this case.6).E.7). e (p0 . if p is increased to p0 and x kept …xed. u). so e (p00 . a contradiction. If x00 solves EMP at u with prices p00 = p + (1 ) p0 for 2 [0. Since this is true for all p.2). expenditure (to maintain utility level u) increases linearly to e (p0 . If preferences are quasilinear with respect to the …rst good. u) = p00 x p0 x = e (p0 . We do not prove continuity. thus is concave. Exercise 15 (MWG 3. and …nd the form of the expenditure function.
Concavity follows from the fact that. u) . 1].
I. and (equivalently) when utility u in the EMP is …xed at the level of maximized (or indirect) utility u (x ) v (p. w) = h (p.e. w) in the UMP. xw = xh if w = p xh . since p x
h
= p1 = = p1
1 1
p2 p1
1
1
u + p2 + 1
1 !
p1 p2 p1 p1 2
u u
p2 1
1
u
e (p. w)
in the UMP. w)) :
Example.3
Duality
The connection between Walrasian and Hicksian demand is that UMP and EMP have the same solutions when wealth w in the UMP is …xed at the level of minimized expenditure p x e (p.
28
. Compare the Walrasian and Hicksian demands for CobbDouglas preferences: w w and xw = (1 ) xw = 2 1 p1 p2 and 1 1 p2 p1 xh = u and xh = u: 1 2 1 p1 p2 Equating either the x1 s or x2 s gives w= Since u (xw ) = w p1 (1 ) w p2
1
p1 1
p2
1
u:
=
1 p1 p2
1
w
v (p. v (p. e (p. On the other hand. u) in the EMP.4. u) = x (p. u) .
in the EMP. h (p. u)) and x (p. we see that xw = xh if u = u (xw ).
The UMP and EMP are generally equivalent in the following sense. Proposition. If u represents continuous, locally nonsatiated preferences 0, then: on RL , and p + (i) if x solves the UMP at wealth w > 0, then x solves the EMP at u = u (x ), and e (p; u) = w; (ii) if x solves the EMP at utility u > u (0), then x solves the UMP at w = e (p; u), and u (x ) = u. Proof. (i) Suppose x is a solution to UMP at w, but not to EMP at u = u (x ). Let instead x0 be a solution to EMP at u = u (x ), so that p x0 < p x and u (x0 ) u. Local nonsatiation implies that there exists x00 such that u (x00 ) > u (x0 ) su¢ ciently close to x0 for p x00 < p x to still hold. But then x00 2 Bp;w and u (x00 ) > u (x ), so that x was not a solution to UMP. By contradiction, x solves EMP, and therefore e (p; u) = p x = w. (ii) Conversely, suppose x is a solution in EMP at u > u (0), but not in UMP at w = p x . Let instead x0 be a solution to UMP at w, so that u (x0 ) > u (x ) and p x0 p x . By continuity, we can scale x0 down to x00 = x0 , with 2 (0; 1) su¢ ciently close to 1, such that u (x00 ) > u (x ) still holds. (Note that u > u (0) implies x 6= 0 and x0 6= 0, so p x > 0.) Since x00 x , we have p x00 < p x , and x cannot be a solution to EMP. By contradiction, x solves UMP, and therefore u (x ) = u. Exercise 17 (MWG 3.E.9). Using the equivalence of the UMP and EMP, show that the general properties of the indirect utility function (homogeneous of degree zero, strictly increasing in w, nonincreasing in prices, quasiconvex, continuous in p and w) imply the general properties of the expenditure function (homogeneous of degree one in p, strictly increasing in u, nondecreasing in prices, concave in p, continuous in p and u), and vice versa. Exercise 18 (MWG 3.E.10). Using the equivalence of the UMP and EMP and the properties of indirect utility and expenditure functions, show that properties of the Walrasian demand function for continuous, locally nonsatiated preferences (homogeneous of degree zero, Walras’law, convex / singlevalued if preferences are convex / strictly convex) imply properties of the Hicksian demand function (homogeneous of degree zero in p, no excess utility, convex / singlevalued if preferences are convex / strictly convex), 29
and vice versa. (Note there is a typo in the book  you are to show that Proposition 3.D2 implies Proposition 3.E3, not 3.E4). The relationship between UMP and EMP is just a special case of a far more general theory of duality. In this connection, duality means that a constrained maximization problem can be expressed as a constrained minimization problem, swapping objective and constraint.
5
5.1
Comparative Statics
Wealth E¤ects
In this lecture, we examine the behavior of the demand function as prices and wealth change. We assume for now that demand is singlevalued (i.e. preferences are strictly convex). The wealth e¤ect for the `th good is @x` (p; w) =@w. If the wealth e¤ect is positive, i.e. @x` (p; w) =@w 0, we say that the good is normal. If the wealth e¤ect is negative, i.e. @x (p; w) =@w < 0, we say that the good is inferior. Goods are inferior when there are higherquality, costlier substitutes for the agent to switch to as wealth increases. (E.g. supermarket bread vs. fresh bread from a bakery.) The total wealth e¤ect is given by the derivative vector with respect to w: 3 2 6 Dw x (p; w) = 4
@x1 (p;w) @w
. . .
@xL (p;w) @w
7 L 52R :
5.2
Price E¤ects
The price e¤ect for the `th good is @x` (p; w) =@p` . Typically, we expect the price e¤ect to be negative. A good for which the price e¤ect is positive, so that the agent consumes more of it after its price increases, is called a Gi¤en good. 30
A Gi¤en good is similar to an inferior good  actually, it is just a very inferior good. As price increases, the agent’ budget set shrinks, and if s the wealth e¤ect (having more of an inferior good when you are poorer) is su¢ ciently powerful, we have a Gi¤en good. The total price e¤ect is given by the derivative matrix with respect to p: 2 @x1 (p;w) @x1 (p;w) 3 We consider now some relationships between price and wealth e¤ects. Recall that Walrasian demand is homogeneous of degree zero. An impliciation is that changes to the consumption bundle in response to simultaneous (and proportionate) increases in prices and wealth cancel. This merely re‡ ects the invariance of the budget constraint. Proposition. If Walrasian demand x (p; w) is homogeneous of degree zero, then 8p and 8w, Dp x (p; w) p + Dw x (p; w) w = 0: Proof. By zerohomogeneity, 8 > 0, x ( p; w) i.e. 3 x1 ( p; w) 6 7 . . 4 5 . xL ( p; w) 2 2 x (p; w) = 0; 3 2 3 x1 (p; w) 0 6 7 4 5 . . 4 5= 0 . 0 xL (p; w) 6 Dp x (p; w) = 4
@p1
. . .
...
@pL
. . .
@xL (p;w) @p1
@xL (p;w) @pL
7 5:
This is true, in particular, for , we have for ` = 1; : : : ; L,
= 1. Totally di¤erentiating with respect to
L X @x` ( p; w) @x` ( p; w) pk + w = 0; @pk @w k=1
which corresponds to the claim in matrix notation when
= 1.
31
@p`
which corresponds to the second claim in matrix notation. : : : . : : : . If Walrasian demand x (p. w) = 1: Proof. Similarly. Proposition. p x (p. has to be of the same magnitude. we have for ` = 1. and therefore the decrease in required wealth from reduced consumption. w)T + pT Dp x (p. p Dp x (p. w) p` = 1. we have for ` = 1. w) = w. x (p. an increase in available wealth has to be matched by the increase in required wealth from greater consumption. …rst with respect to prices. @w `=1
32
. w). Di¤erentiating Walras’ law with respect to wealth. w) +
L X k=1
pk
@xk (p. notice that a price increase for all commodities would increase the required wealth in proportion x (p. L x` (p.
which corresponds to the …rst claim in matrix notation. w) = 0T and pT Dw x (p. Di¤erentiating Walras’law. To understand how the following results relate to this intuition. and that an increase in wealth (while prices remain …xed) brings with it a corresponding increase in consumption. w) = 0. w).Walras’ law implies that an increase in prices (while wealth remains …xed) must be accompanied by an o¤setting decrease in consumption. p Dw x (p. w) satis…es Walras’ law. then 8p and 8w. L L X @x (p. w).
Interpret. x3 (p. Exceptions are the Gi¤en goods. w). show that "`w (p.e. w) is homogeneous of degree one with respect to w. This statement is.E. L. If x (p. In the Hicksian de…nition of demand. If x (p.3). w) p = Interpret. w) = p1 + p2 + p3 p2 p1 w .e.E. Exercise 20 (MWG 2. w) = p1 + p2 + p3 p3 w:
verify the three derivative conditions for a demand function x (p. x (p.E.Exercise 19 (MWG 2. : : : ." i. x ( p. w) = x (p. 8 > 0. so it does not "impoverish. w) = 1 for ` = 1.6). Hence the two de…nitions of demand di¤er in comparative statics." and the Gi¤en e¤ect is absent. Hicksian demand is said to obey the "compensated law of demand.3
Law of Demand
The "law of demand" refers to the intuitive property that a price increase should reduce demand for a commodity. (singlevalued) demand for any commodity diminishes if its price increases. i. They arise because a price increase e¤ectively makes the agent poorer and may lead to an increase in the consumption of relatively cheap commodities. Exercise 21 (MWG 2. and satis…es Walras’ law.e. w). however.
5. and satis…es Walras’law. w) = x (p. w) is homogeneous of degree zero. w) = w p2 p1 + p2 + p3 p1 p3 w x2 (p. w) that is homogeneous of degree zero and satis…es Walras’law. 8 > 0. i. show that p Dp x (p. not generally valid for Walrasian demand. In the case of the demand function x1 (p.4). 33
. a price increase is accompanied by an increase in expenditure.
u)) 0:
Proof. "`k (p. p00 . @x` (p." If ` 6= k. w) =@p` depends in this case on whether petrol is sold by gallon or by liter (about a quarter gallon). u) p00 h (p0 . and h (p0 . a $1 increase in the price of petrol would cause a greater response in demand than a U1 increase (which translates into about one cent). w)
34
. @x` (p. Moreover. u) .Proposition.
since h (p00 . u) h (p0 . "`k (p. w) w : @w x` (p. (p00 p0 ) (h (p00 . and the Hicksian demand correspondence h (p.4
Elasticity
The partial derivatives are not unitfree measures. Subtracting p0 h (p0 . If u represents continuous. u) on the left of the …rst inequality preserves signs and gives the result. u) minimizes expenditure when prices are p0 . w) =@p` would depend on the currency in which we quote prices: clearly. locally nonsatiated preferences on RL ." The wealth elastisticity of demand x` (p. u) p0 h (p00 .
5. w) @pk x` (p. Simply observe that p00 h (p00 . w) is called "crossprice elasticity. then 8p0 . w) If ` = k. u) on the right and p0 h (p00 . w) @x` (p. w) with respect to the kth good is: @x` (p. u) is singlevalued for + all p 0. An alternative way to describe price and wealth e¤ects is in terms of elasticities. The price elastisticity of demand x` (p.
De…nition. w) pk : "`k (p. w) is called "ownprice elasticity.g. w) is: "`w (p. u) p0 h (p0 . E. which are unitfree. u) minimizes expenditure when prices are p00 .
the percent change in x` (p. locally s nonsatiated and the expenditure and indirect utility functions di¤erentiable. w) with respect to the kth good can be expressed as "`k (p. Let the consumer’ preference relation be rational. First.E. Exercise 22 (MWG 2. w) 0. then she is better o¤ after the s price change if and only if v (p1 . we consider the welfare implications of price changes. w) = 0:
5. we can restate the relationship between the responses in demand for commodity ` to changes in prices and wealth (as implied by homogeneity) as:
L X k=1
"`k (p. w). If consumer’ preferences are known. we are also interested in conditions under which a price is welfareimproving that are based on minimal information (such as the initial price and demand and the new price). when the change is very small): e. Suppose wealth is …xed at w > 0. w)) . @ ln pk
and derive a corresponding expression for "`w (p.g. and the price changes from p0 to p1 .8).
35
. w) + "`w (p. Demonstrate that the price elasticity of demand x` (p. Since data availability is normally a signi…cant constraint. w) v (p0 . In terms of elasticities. w) = @ ln (x` (p.5
Money Metric
In the remainder. continuous.Elasticities can be interpreted as the approximate relative percent change in two variables (this relationship holds exactly only in the limit. w) in response to a small percent change in p` . we derive a quantitative measure of welfare change in terms of the expenditure function.
Recall that the expenditure function e (p. u1
since e (p0 . w) is called the money metric (indirect utility function). at given prices.w0 . v (p0 . so is the money metric. but those prices are …xed in the money metric. (Expenditure increases in its price arguments. I. w)) 0 (for any indirect utility function). w)) 0. v (p0 .
e (p. w) and u1 in welfare by e p0 . w). u0 = e p0 .) Clearly. if a bundle that yields the maximal utility attainable at prices p1 is more expensive than a bundle that yields the maximal utility attainable at prices p0 .6
Welfare Comparisons
Let u0 v (p0 . u0 ) = e (p0 . the expenditure function is itself an indirect utility function when u is evaluated at u = v (p. v (p1 .e. we have. w0 ) . w)) = e (p1 . v (p. w0 )) is a meaningful quantitative measure of welfare change. Hence it is unique up to the choice of p. w)) e (p. as all utility representas tions select the same consumption bundle at given prices and wealth. It is the change in wealth needed to buy the optimal bundle when the budget set changes from Bp. v (p0 .
36
. the agent is better o¤ at prices p1 than at price p0 if and only if e (p. Note that because indirect utility is decreasing in prices. u) is strictly increasing in u. The expenditure function evaluated at u = v (p. for any …xed price p e (p. Hence. Then we can de…ne the change e p1 . we can express the indirect utility at p1 and p0 in terms of the expenditure required to attain it. i. w)) e (p. w).
5. w)) = w = e (p1 . w)
and e (p. u1 ) through the equivalence of UMP and EMP. u1 . It is independent of the utility representation we choose for the agent’ preference.Due to the duality of the UMP and EMP.e. v (p. v (p1 .w to Bp0 . w0 )) () v (p. v (p0 . u1 v (p1 . v (p0 .
e p0 .
5). however. w) for any prices p0 and p1 . 1 1 so that u1 > u0 . these tell us the change in wealth that is required to maintain utility at u1 and u0 after a price change. I. giving the new price vector p1 p0 . Exercise 24 (MWG 3. Demonstrate that CV (p0 . Show that CV (p0 .I. this means more of it is demanded at optimum. w e p0 . at all wealth levels w. w e p0 . : : : . u1 e p1 . u0 e p1 . p1 . w) if good ` is inferior. less of it is demanded.) Suppose the price of good ` falls (other price remained …xed). Suppose u (x) is quasilinear with respect to the …rst good (and p = 1 is …xed).3).
This is what the agent has to spend at the new prices p1 in order to maintain the original utility u0 that was available at the old prices p0 . p1 . and CV p0 . u0 = e p0 . If good 1 is normal. w) > EV (p0 .This is the additional wealth the agent would have needed at the old prices p0 in order to attain the level of utility u1 that is available under the new prices p1 . (Respectively. p1 . u1
is called the "equivalent variation" (EV). then these measures would not coincide. Exercise 23 (MWG 3. Hence the chosen bundles and their associated expenditures would depend on how u is …xed. For 37
. w) = EV (p0 . then u1 is associated with greater wealth at given prices. The following is also a plausible way to de…ne the change in welfare from the money metric: e p1 . u1 e p1 . Let there be a population of consumers indexed by i = 1. Welfare change as measured by EV p0 . with utility functions ui (x) and individual wealths wi .6). p1 .I. If p0 > p1 . u0 . If good 1 is inferior. u0
is called the "compensating variation" (CV). u0 e p1 . p1 . Exercise 25 (MWG 3. Consider a price change in good 1 only.I. that if there were wealth e¤ects. Note. p1 .
By Walras’law. p0 x0 = w.e. and by local nonsatiation there exists a bundle x 2 Bp1 . w) than under (p0 . w0 ) to (p1 . See Figure 3 for illustrations of the consumer surplus from good 1 before and after a price cut. u) dp`
p0 `
0
L XZ `=1
p1 `
h` (p. u) @p` (by the envelope theorem. since e (p. w) if (p1 p0 ) x0 < 0. u) dp`
0
h` (p. and we wish to evaluate the impact of a change in prices to p1 ? Proposition. wi ) for all i. Extend this test of welfare improvement to changes in prices and wealth from (p0 . u) = p h (p. @e (p.12). then the agent is strictly better o¤ under (p1 .P any change in prices from p0 to p1 such that i CVi (p0 . L. w) > 0. : : : . such as the initial price p0 and choice x0 x (p0 . so (p1 p0 ) x0 < 0 implies p1 x0 < w.u
1
= =
L XZ `=1
L XZ `=1
p0 `
h` (p. u)) to derive e p . i=1 i wi and vi (p . w). the two de…nitions of welfare change are equivalent. there exist wealth P 0 P 0 1 0 levels fwi gI such that i wi vi (p0 .)
38
.w that is strictly preferred to x0 . for ` = 1. where it is now not necessarily the case that w1 = w0 .I. u) = h` (p. wi ) Absent wealth e¤ects. Proof. u) is locally invariant to changes in its minimizer h (p. I. w1 ). If preference is locally nonsatiated. show that it is possible to compensate everyone for lost utility. What if only partial information is available about demand. p1 . Exercise 26 (MWG 3. (No need to do this for equivalent and compensating variation. u) dp` :
p1 `
This is the change in consumer surplus as a result of the price change.u
0
e p . But then x0 is in the interior of the budget set at p1 . We can use the fact that.
it attains e (p0 . x0 ). The gradient of the …xedexpenditure curve at this point is rp e (p0 . because the expenditure function is concave in each price. e (p. To appreciate the di¤erence. x0 ) @e (p0 . think of a vector 39
. x0 ) @x` @e (p0 . by the envelope theorem: since x0 minimizes e (p0 .) Since p0 is an optimal choice.e. x0 ) de (p0 . x0 ) = + = = x` : dp` @p` @x` @p` @p` The gradient vector x0 of e (p. x0 ) at p0 . x0 ) is drawn as a convex curve. u0 ) = e (p0 . u0 )g at p0 . rx e (p0 . x0 ) = x0 . expenditure increases at a diminishing rate as p2 increases. and therefore lies on the curve. @e (p0 . x0 ) = 0. keeping p1 …xed. To see this. hence smaller reductions in p1 are required to o¤set increases of given size in p2 . where the two scenarios are depicted in price space. 2. (I.t. look at Figure 4. we have for ` = 1.Figure 3: Change in consumer surplus
The converse is not always true: (p1 p0 ) x0 > 0 does not imply that the agent is worse o¤ after the price change.e. u0 ) is orthogonal to the tangent of the level set fp s. The set of prices that keep expenditure constant at e (p0 . i.
It must be a multiple of v (1. x0 v = 0.e.I. 2 1 Clearly. p1 could lie beneath the level curve. but direct proofs are enough. when (p1 p0 ) x0 < 0.11). p1 and x0 . which is orthogonal to x0 .) 40
. Suppose x1 = x (p1 . but not necessarily in the second. As drawn in the right panel of Figure 4. since p1 lies above the level curve at e (p0 . u0 ) < e (p0 . u0 ) in the …rst case (so that there is a welfare improvement). and above when (p1 p0 ) x0 > 0. w) is known in addition to p0 . but if the price change (in the same direction) were su¢ ciently large. since a oneunit increase in p1 2 1 requires p2 to be reduced by (x0 =x0 ) p1 to keep expenditure constant. Give graphic intuition for these results. u0 ). (They can be established via …rstorder approximation of the expenditure function at p1 . Since p1 p0 is the vector "from" p0 to p1 . a vector in the direction of the tangent). Exercise 27 (MWG 3. Argue that the agent is worse o¤ at p1 than at p0 if (p1 p0 ) x1 > 0. it must lie below the tangent to the level curve. The concavity of the expenditure function therefore implies that e (p1 . or equivalently if p0 (x1 x0 ) < 0 (wealth is …xed).Figure 4: Scenarios (p1
p0 ) x0 < 0 (left) and (p1
p0 ) x0 > 0 (right)
in the tangent space (i. x0 =x0 ). welfare decreases.
The problem the agent solves is to choose one or more elements from this set.
De…nition. y 2 B. whatever the reasons. then we have no information about them. A choice structure (B. x and y are only comparable if x 2 C (B) or y 2 C (B) for some B 2 B such that x. their existence and properties are not directly testable.1
ChoiceBased Approach
Choice Structures
So far we have assumed that choice behavior is consistent with an underlying rational preference ordering. If x and y are never chosen. the agent may face. Since preferences are unobservable. y 2 B.
41
. In this lecture. In this context. C ( )) if x % y ("x is revealed preferred to y") if and only if x 2 C (B) for some B 2 B such that x. C ( )) consists of a family B of budget sets and a choice rule C ( ) that assigns a nonempty subset C (B) B to every B 2 B. % is the "revealed preference" derived from choice structure (B. a budget set B 2 B may be thought of as a speci…c decision problem. The relation % need have none of the rationality properties we assumed for primitive preferences. The set C (B) simply describes the objects the agent might be observed to choose when presented with budget set B. Hence % is not necessarily complete. of many possible problems. we will build a largely parallel theory of demand that starts from the properties of observable choices. Since existence of a preference is not assumed. it is not meaningful to say that objects in C (B) are indi¤erent to each other and preferred to everything else in B. For example. One may de…ne a "revealed preference" relation from a choice rule as follows.6
6.
zg) = fyg. De…nition. but z x. x % yg :
In principle. (B. (I. yg) = fxg. For example. Given a budget set and a preference relation %. we will impicitly assume that it satis…es the de…nition of a choice structure.
42
. C (B. one may derive a choice rule as follows.4). ) is the choice struture generated by preference % if and only if 8B 2 B. The Paasche index.) Argue: (a) If LQ < 1.e. then x0 is revealed preferred to x1 . (If demands refer to aggregate consumption. this is the percent change in GDP.) As we know. which allows prices to vary. Let p0 and q 0 denote prices and quantities at time 0. then x y z. (c) If EQ < 1 or EQ > 1. %) = fx 2 B s. the generated choice rule could be empty for some B 2 B. and C (fx.There is also no guarantee of transitivity. there is no mostpreferred element in the budget set. that the generated choices are always nonempty. LQ (p0 x1 ) = (p0 x0 ). then x1 is revealed preferred to x0 . no revealed preference between x0 and x1 can be established. The Laspeyres index. and let p1 and q 1 denote the new prices and quantities at time 1.t. i. (b) If PQ > 1. %) . is based on initial prices. zg) = fzg. hence that C (B. PQ (p1 x1 ) = (p1 x0 ). C (fy. uses new prices.F. 8y 2 B.e. Laspeyres and Paasche indices measure the change in consumption between two points in time at …xed prices. if C (fx.C ( . Whenever we refer to a generated choice structure in this lecture. rationality and continuity of % ensure that there exists a continuous utility representation and a solution to the UMP. Consider also the expenditure change EQ (p1 x1 ) = (p0 x0 ). Exercise 28 (MWG 2. %) 6= ?.
then we can never observe a strict preference for y over x. In other words. Since each pair (fx. It would be a di¤erent matter if we added a fourth budget set fx. fy.2
Weak Axiom
In order to have anything substantive to say about observed choices.3). the axiom is not tested. Choice structure (B. (b) the consumption bundle in year 1 is revealed preferred to that in year 2. yg. then x 2 C (B) and y 2 C (B 0 ) imply fx. y 2 B 0 . zg) belongs to only one B 2 B. and y 2 C (B 0 ) for B 0 2 B with x. whenever x 2 C (B) for B 2 B with x. Exercise 30 (MWG 1. (b) nor (c) can be concluded based on the data. y. we also have x 2 C (B 0 ). zg to B. such that (a) the choices violate the weak axiom. yg C (B) \ C (B 0 ). y. Now any choice rule violates WA. good 1 is inferior at some price. good 2 is inferior at some price. WA is not violated. if we ever observe a preference for x over y. y 2 B. In the nontransitive scenario above.F. The weak axiom of revealed preference says that. then x must be a choice whenever x is available and y is a choice. B 0 2 B. The following is partial information about a consumer’ purchases: s Year 1 Quantity Price Good 1 100 100 Good 1 100 100 Year 2 Quantity Price : 120 100 ? 80
Give the range of quantities of good 2. then 43
. (c) the consumption bundle in year 2 is revealed preferred to that in year 1. zg). Argue that WA is equivalent to the following property. we need to impose some minimal consistency.C.2).
De…nition. (d) neither (a). with x. If x 2 C (fx. consumed in year 2.6. (f) given WA holds. if x is ever a choice when y is available. zg. fx. (e) given WA holds. Exercise 29 (MWG 2. C ( )) satis…es the weak axiom (WA) if. If B. Example. y 2 B \ B 0 .
zg. y.3). y. z 2 B 0 and y 2 C (B 0 ). y.e. Consider the budget set fx. x 2 C (B) and y 2 C (B). yg). it is in fact unique. The same argument applies to y 2 C (fx. C ( )) satisfy WA.) Exercise 31 (MWG 1. Then there exists B such that x. Proof. Let B include all subsets of X that contain one. if the choice structure (B. or y 2 C (fx. (Provided every possible set of two objects is included in B. zg). C ( )) satis…es WA. C ( )) which satis…es WA is necessarily rational. (This makes it more restrictive. y. for an arbitrary family of budget sets B. Hence. whereby x % z.WA stipulates that x is a choice for any subset of fx. zg) must be empty (which is impossible by de…nition).e. in the sense that it has to "commit" to a choice in more decision situations. it is not the case that the revealed preference % derived from (B. If y 2 C (fx. The relation % derived from (B. But this was not the case. zg). Hence x 2 C (fx. % derived from (B. and there exists B 0 such that y. i. Either x 2 C (fx. y % x. y 2 B. yg 2 B. so that C (fx. zg) and z 2 C (fx.
Since the revealed preference ordering is fully determined by choices from twoobject sets. Then. y. i. Moreover. The choice rule needs to be de…ned on a su¢ ciently comprehensive family of budget sets. y. zg). y 2 X. zg). x % y. If z 2 C (fx. two or three elements. C ( )) is transitive: suppose x % y and y % z. zg). = (a) Compare to de…ned such that x y () x % y and not y % x (where % is the revealed preference derived from the choice 44
. zg that contains x and where y or z is a choice. C ( )) is complete: 8x. y. then WA requires x 2 C (fx. y. yg).) Proposition.C. y 2 B and x 2 C (B). we have fx. the revealed preference % derived from it is rational. and de…ne revealed strict preference such that x y () 9B 2 B with x. y. then WA requires y 2 C (fx. or both. y. Let choice structure (B. zg).
this implies x 2 C (B).e. It turns out that WA is not only su¢ cient (with restrictions on B). WA implies that the choice structure is rationalizable (provided the choice rule covers all sets of up to three objects).structure). 8B 2 B. we have x 2 C (B. In fact. y 2 C (B). By WA. C ( )). C ( )). % ). % ). The revealed preference % derived from a choice structure (B. If the model exactly predicts choices. but also necessary for a choice structure to be rationalizable. Let x 2 C (B). then the choices are explained (rationalized) by the model (preference). some such y 2 B must be chosen from B.e. Does this depend on WA? (b) Give an example where is not transitive.
45
. for every y 2 B. I. we must verify that % generates (B. Then there must exist. i. if WA holds for the choice structure. If the revealed preference % derived from choice structure (B. % ). C (B) = C (B. % ) C (B). C ( )) that satis…es WA generates (B. 8y 2 B. %). In the other direction. When the choice rule C ( ) in choice structure (B. Proof. C (B) = C (B. 8y 2 B. C ( )). a budget set By such that x. Thus C (B. This is not quite obvious. Proposition. y 2 By and x 2 C (By ). let x 2 C (B. C ( )) is generated by a rational preference %. C ( )). Combining the inclusions. %).e. Thus C (B) C (B. i. C ( )) is rational. One can think of generating a choice structure from a rational preference as simulating empirical choice data with a rational preference model. then it rationalizes (B. then the revealed preference derived from it is rational and generates the choice structure.e. that statement is true only if WA holds for (B. we say that % rationalizes C ( ). i. Since x % y. % ). Show that the two de…nitions are equivalent.
Hence. Now. x % y. C (B) = C (B. (c) Argue that is transitive if B includes all threeelement subsets of X. Our task is to show that 8B 2 B.
4). Since the choice structure (B. Let y 2 C (B 0 . %)) that is generated by a rational preference % satis…es WA. On the set of objects X = fx. to each budget set B 2 B.Proposition. zg. show that it satis…es pathinvariance: 8 fB1 . but choices that satisfy the weak axiom need not be consistent with rational preference. Rational preference always gives us the weak axiom in a generated choice structure. Suppose x 2 C (B. C ( )) is rationalizable. since we could include all possible budget sets in B. Suppose now that % is rational. re‡ ects a rational preference if and only if it satis…es WA. Proof.e. the preferencebased approach (using rationality) is less general than the choicebased approach (using the weak axiom).D. Think of the choice rule C ( ) as assigning. it is the case that C (B1 ) [ C (B2 ) = C (C (B1 ) [ C (B2 )). de…ned (at least) on all sets with up to three elements. i. budget sets have a special form (they satisfy p x w). A choice structure (B. y. %). One might be tempted to conclude that the preference. fy. zg . If choice structure (B. C ( . zgg.D. yg . so WA holds. in particular transitive. In a meaningful sense. %) and x 2 B 0 . which is restrictive (e.5). %)) is generated by %. Because y 2 C (B 0 . %) for B 2 B and y 2 B. a probability distribution C (B) over objects in B. convex). we have 8z 2 B 0 . de…ne the family of budget sets B = ffx. x % y. %). C ( .g. Exercise 32 (MWG 1. B2 g B such that B1 [ B2 2 B and C (B1 ) [ C (B2 ) 2 B. nor does a solution to the UMP necessarily exist for them. This stochastic choice rule C ( ) is said to be rationalizable by preferences if there exists a probability distribution over (strict) preference relations on X (here there are six such relations) that 8B 2 B induces C (B). Then x 2 C (B 0 . WA requires that x 2 C (B 0 . y % z and by transitivity x % z. %). these results establish that a choice rule. fx. Arbitrary budget sets may not make sense. But in the theory of demand.
Put together. Exercise 33 (MWG 1.and choicebased approaches are basically equivalent. 46
.
6. zg) = C (fx. then x (p. (b) Show that C (fx. we stipulate instead that x 2 B 0 . 2 [0. w) which is generated by a rational preference satis…es WA. (c) Find . zg) = ( . zg) = C (fx. Singlevaluedness implies. Verify that a Walrasian demand function x (p.3
Relationship with the Law of Demand
It seems plausible that Walrasian demand could satisfy a version of the law of demand if we mimic the compensation for price changes that is implicit in Hicksian demand by adjusting wealth with prices. w) that satis…es WA is homogeneous of degree zero.14). w) must be unavailable at prices p0 and wealth w0 .F. Argue that a Walrasian demand function x (p. w0 ).(a) Show that C (fx. 1=2) is rationalizable in this sense. but x (p. 3=4) is not rationalizable in this sense. yg) = C (fy. w0 ) w and x (p. WA has the following speci…c form: 8 (p. zg) = (1=4. w) was chosen instead.1 ) is rationalizable if and only if 2 [ . the weak axiom is the minimal property we need to impose on (singlevalued) Walrasian demand to make it satisfy a compensated law of demand. zg) = C (fx.e. p x (p0 . Notice how the singlevaluedness of choices a¤ects WA: since we cannot require that x 2 C (B 0 ) when x 2 C (B 3 y) and y 2 C (B 0 3 x). In the context of (singlevalued) demand. Exercise 35 (MWG 2. This intuition is correct. w0 ) =) p0 x (p. 1] such that C (fx. In addition to homogeneity of degree zero and Walras’law.12). when x is = chosen over y in B. w) > w0 :
I. w) and 8 (p0 . Exercise 34 (MWG 2.F. ]. that x is revealed strictly preferred to y. w) 6= x (p0 . when x (p0 . if x (p0 . yg) = C (fy. yg) = C (fy. zg) = (1=2. w0 ) is chosen. w0 ) was available at prices p and wealth w. and strict preference is antisymmetric.
47
.
Since x is available in both budget sets. contrary to the assumption. Since p02 < p2 as drawn. Are we done? Actually no. Since x is a¤ordable in both Bp.w0 (by the construction of the compensation). Then x0 must lie in the triangle above x below the new budget line. (Hence no price change can violate WA. in accordance with Walras’ law.w0 ) that satisfy WA. But WA is a property that pertains to arbitrary (possibly uncompensated) price changes. x 2 Bp.w \Bp0 .w . By Walras’law. This point can be made via the contrapositive: if a price change violates WA. In the top right panel. else not choosing x0 violates WA. where wealth is adjusted to w0 such that x remains a¤ordable. the set of bundles that satis…es WA is dotted.w . The top left panel depicts the budget sets.w0 reveals it (strictly) preferred to x. (Otherwise. Then x0 2 Bp.w ) and x0 2 C (Bp0 . CLD requires that x02 > x2 .w0 . the (single) choice of x0 in Bp0 .w0 . We have only shown that WA is equivalent to CLD for compensated price changes. which lies on the boundary of the budget set Bp. This is exactly the same set that satis…es WA. x0 belongs to the boldstriped portion of the boundary of Bp0 . and the price change is compensated. the other panels show all possible locations for x = C (Bp. The pivoting of the budget line through x is visible in Figure 6. the choice x0 in Bp0 .w and Bp0 . Before I give an analytical proof of the equivalence of WA and the compensated law of demand. and was revealed preferred to everything in Bp. The = same type of argument applies in the other cases. then it lies on the boldstriped portion of the boundary of Bp0 .w . a graphic sketch will be helpful. the set of bundles that satis…es CLD is dotted.w0 .) If we assume that x0 also satis…es Walras’law. Consider a compensated price change to p0 . In the right panel. Suppose we start with bundle x.) 48
.w if it is distinct from x.w0 must lie outside Bp. then we can construct a compensated price change that violates CLD. x0 would have been chosen prior to the compensated price change.w and Bp0 . In the left panel.w0 . We must still demonstrate that such price changes will also satisfy WA.Figure 5 illustrates how WA restricts choices given budget sets Bp.
Figure 5: WA for singlevalued demand
49
.
w) that is homogeneous of degree zero and satis…es Walras’law. so CLD is violated.e. WA holds if and only if 8 (p. x00 2 2 must lie in both dotted triangles. we have situation that is at odds with WA: x and x0 both belong to Bp. I.w and Bp0 . that x00 > x2 and x00 < x02 .w0 (other violations are straightforward to deal with). We can construct a new budget set Bp00 . when CLD holds for compensated price changes. Proposition. w) 6= x (p0 . (p0 p) (x (p0 . above x and below x0 . w0 ) x (p. w) and 8 (p0 . w0 ). for these 2 2 compensated changes in demand. For a Walrasian demand function x (p. Hence CLD requires. But these are disjoint sets. the boundary of Bp00 . CLD is in fact su¢ cient for WA. w) and x (p.w00 that represents a compensated price change for both x and x0 (graphically.w00 pivots through both x and x0 ). One can see that p00 < p2 and p00 > p02 . Since failures of WA (with arbitrary price changes) lead to contradiction.Figure 6: Equivalence of WA and CLD for compensated price changes
In Figure 7. w)) < 0:
50
. w0 ) such that w0 = p0 x (p.
Consider therefore (p. and p x (p . so that the CLD inequality reduces to the equivalent inequality
0
p (x (p0 . w0 ) w would imply p x (p. w0 ) satis…es Walras’law. w) < w. If w0 = p0 x (p. w). Given the strict inequalities. Similarly for the case w = p x (p0 . but we will show that there exists a compensated price change for which the CLD cannot hold. w0 ) and is satis…ed if p0 x (p.Figure 7: Impossibility of WA failures
Proof. the right side is close to p x (p. and at su¢ ciently close to 0. w) = x (p0 . Hence p x (p0 . This appears to be a violation of WA. w) = w > p x (p0 . w) and (p0 . w0 )
x (p. there exists 2 (0. w0 ) = w0 = p x (p. w). Since x (p0 . w0 ) where x (p. w0 ) > w. w)) > 0:
(If) WA holds vacuously if x (p. we have p0 x (p0 . w0 ) < w. 1) such that ( p + (1 ) p0 ) x (p. the left side is close to 51
. w0 ) > w. w0 ) and w0 p0 x (p. w) > w0 or p x (p0 . w). then the inequality 0 applies. Hence the scenario does not satisfy the assumptions. w0 ). w) 6= x (p0 . w) = ( p + (1 ) p0 ) x (p0 . Thus we can concentrate on p0 x (p. w0 )
(because at su¢ ciently close to 1. w0 ). so that WA is not tested. w) < w0 and p x (p0 . violating Walras’ law.
w) by assumption. p00 x (p00 .p0 x (p0 .
and. w0 ) = ( p + (1 ) p0 ) x (p. w00 ) x (p0 . w). since p00 (x (p00 . w0 )) > 0: ((1 ) = ) p0 . w0 ) = w0 > p0 x (p. w)) = 0. w)) > 0 and p0 (x (p00 . w00 ) x (p0 . w00 ) x (p. since p0 x (p0 . this means
p0 x (p. w00 ) = w00 . Hence p0 x (p. w) < w0 and p x (p0 . w00 ) : Now. where the equalities are due to Walras’ law). and we have shown that WA holds for all (p. w0 ) = w00 . By Walras’law. so p00 (x (p00 . w00 ) and p00 (x (p00 . w) = w00 and p00 x (p0 . we have constructed compensated price changes from p to p00 and from p0 to p00 . it follows that p0 x (p0 . w) < w0 . ThereFrom the de…nition of p00 . w0 ) < w is not possible. w0 ) = w0 > p0 x (p. CLD must hold for both: p (x (p00 . w00 ) x (p. w00 )
x (p. we have p = (1= ) p00 fore. w0 ) if CLD holds for all compensated price changes (and Walras’law is in e¤ect). 52
. Let p00 p+(1 ) p0 and w00 ( p + (1 ) p0 ) x (p0 . CLD for p to p00 implies (p00 (1 ) p0 ) (x (p00 . w0 ) > p0 x (p00 . w) and (p0 . w0 )) = 0: Because p to p00 and p0 to p00 are compensated price changes. w00 ) : But then the CLD for p0 to p00 cannot hold. w) > p0 x (p00 . w)) = 0
x (p. Since p00 x (p. w)) > 0.
w). w0 ). WA requires that x (p0 . w) = w. wealth is adjusted to w0 = p0 x (p. Exercise 37 (MWG 2. w) and 8x0 2 x (p0 . Let a di¤erentiable Walrasian demand function x (p. ) is also homogeneous of degree one with respect to wealth w. Suppose x ( . Instead. which require only that preference is locally nonsatiated). p x (p. p0 . x (p. But since x (p0 . which gives the strict inequality
0
p (x (p0 . Exercise 36 (MWG 2. w). w). The remarkable aspect of the result is that nothing more than WA is needed for the compensated law of (Walrasian) demand (other than homogeneity of degree zero and Walras’ law.13). show that x (p. w) after the price change to p0 . generalized to choice sets. w)) 0. (p0 p) (x (p0 .(Only if) Because p0 x (p. w) satisfy homogeneity of degree zero. w0 ) (revealing it preferred).5).F. w and 8 > 0. w) x (p. w0 ) is unavailable at (p. w0 )
x (p. w)) > 0:
Observe that we are not comparing initial demand x (p. w) = x (p. Since it is an "if and only if" proposition. p x0 < w =) p x > w: 53
. Consider a multivalued Walrasian demand correspondence x (p. p x (p0 . Walras’ law and the weak axiom.e. w0 ). preference does not have to be rational or continuous. the bundle x (p. In particular. w) satis…es WA and Walras’law. (b) If x (p. w0 ) > w. w) to the uncompensated demand x (p0 .F. w0 ) is chosen instead at (p0 . w) is available at (p . w) also has the following property: 8x 2 x (p. w) = w0 . so that consumption of all goods increases proportionately: 8p. (a) Give the Walrasian version of WA in this case. w). so that the initial bundle is still in the budget set at the new prices. Show that the law of demand holds also for uncompensated price changes: 8p. From Walras’law. i. WA can be said to be equivalent to the compensated law of demand under local nonsatiation.
a choice structure is generated by a rational preference if it satis…es SA and only if it satis…es WA. Thus. I. C ( )) satis…es the strong axiom (SA) if. w). Choice structure (B.
De…nition.(c) Moreover. if x1 is chosen when x2 is available. and x2 is chosen when x3 is available (so that x1 is indirectly revealed preferred to x3 ). then the revealed preference % derived from it is rational. w0 ) such that w0 = p0 x. : : : .J. show that if x (p. B N B. w) satis…es WA and Walras’ law.) You will recognize the ‡ avor of transitivity in SA. Therefore. : : : . w) and 8x 2 x (p. Beyond WA. Show for the consumption set R2 that Wal+ rasian demand satis…es WA if and only if it satis…es SA. N 1. Exercise 38 (MWG 3. SA. then x1 must be chosen whenever it is available and x2 is chosen. (And we could make the chain arbitrarily long. and x1 . then the following compensated law of demand holds: 8 (p. (p0 p) (x0 x) 0:
6. however. for any collection of budget sets B 1 . xN such that xn 2 C (B n ) and xn+1 2 B n for n = 1. SA is just WA if we consider only collections of N = 2 budget sets fB. 54
.4
Strong Axiom
A choicebased theory that is fully equivalent to the preferencebased approach with rationality can be based on the strong axiom of revealed preference. Transitivity is precisely what WA failed to guarantee in a revealed preference relation derived from it. it inherits all properties of WA. (The proof is nonelementary. and 8 (p0 .e. C ( )) satis…es SA. if x1 is chosen when x2 is available. we have x1 2 C B N if x1 2 B N . If the choice structure (B. is su¢ cient without any requirements on the domain of the choice rule. SA says.1).) Since SA implies WA. then x1 must be chosen whenever it is available and x3 is chosen. B 0 g B: then x 2 C (B) and y 2 C (B 0 ) with y 2 B and x 2 B 0 implies x 2 C (B 0 ). : : : .
Graphically. as shown in the left panel of Figure 8. the weak axiom is not su¢ cient for the existence of a rational preference that generates the choices. 55
. then they satisfy the weak axiom. w0 ) x (p. so that the initial utility is still attained at the optimal choice h (p0 . which passes through x). a price increase (inward pivot of the budget line) is accompanied by a wealth increase (outward shift of the budget line. as shown in the right panel of Figure 8. w) and 8 (p0 . (p0 p) (h (p0 . In the context of the expenditure minimization problem. u) h (p. w0 ) such that w0 = p0 x (p. This is known as Hicks compensation. w). u)) 0:
Are the two statements of the compensated law of demand equivalent? In the choicebased approach. The reverse conclusion is problematic: while the compensated law of demand implies the weak axiom.7
7. a price increase is accompanied by a wealth increase (outward shift of the budget line. we also derived a compensated law of demand (for continuous and locally nonsatiated preferences): 8p and 8p0 . In the preferencebased approach. which is equivalent to the compensated law of demand (as long as choices are homogeneous of degree zero and satisfy Walras’ law): 8 (p. the consumer’ wealth is explics itly adjusted so that the initial choice x remains a¤ordable at prices p0 . u) after the price increase. This type of compensation is called Slutsky compensation. Graphically.1
Integrability
Slutsky and Hicks Compensation
If choices are generated from rational preferences. which remains tangent to the indi¤erence curve at u). w)) 0:
Hence rational preferences imply the version of the compensated law of demand we posed in the discussion of the choicebased approach. the consumer’ utility is held …xed s and expenditure is allowed to vary. (p0 p) (x (p0 .
which contains price e¤ects. it is clear that the two versions of the compensated law of demand are not the same.e. so that the preferencebased compensated law of demand indeed guarantees rationalizability. can they always be constructed from a rational preference? To this end. The additional property that the substitution matrix must have if the preferencebased compensated law of demand holds is symmetry. We then argue that symmetry is enough to recover a rational preference from choices.Figure 8: Compensation in the choice. Our interest in this lecture is in whether the preferencebased compensated law of demand implies stronger restrictions on choices that guarantee rationalizability. for in…nitesimal price changes. 56
. However. if we observe that choices satisfy it.and preference approaches
Visually. Negative semide…niteness of the substitution matrix is (almost) equivalent to the choicebased compensated law of demand and necessary (but not su¢ cient) for the preferencebased compensated law of demand. I. they do lead to the same adjustments in choices. we will characterize the compensated law of demand in the choicebased and preferencebased approaches in terms of the substitution matrix.
so is the dot product: x (y + z) = x y + x y since x (y + z) =
L X `=1
x` (y` + z` ) =
L X `=1
x` y` +
L X `=1
x` z` = x y + x z:
The dot product is not associative: (x y) z 6= x (y z) since. xT y = y T x. ! ! ! L L K L X X X X x` y` z1 + + x` y` zK x` y` zk = (x y) z = 6=
k=1 `=1 L X `=1
y` z`
!
x1 +
+
`=1 L X `=1
y` z`
!
`=1
xK :
The dot product of x and y can be written in matrix notation: x y = xT y: Commutativity. xT (y + z) = xT y +xT z. The dot product of vectors x and y is the sum of the products of corresponding elements (that have the same indices): x y
L X `=1
x` y` :
Because multiplication is commutative. so is the dot product: x y = y x since L L X X x y= x` y` = y` x` = y x:
`=1 `=1
Because multiplication is distributive. in general. 57
. I brie‡ review the properties of the y dot product.7. and distributivity. are of course inherited.2
Aside: Dot Product
Because we will make heavy use of vector notation and transformations of vector equations in this lecture.
w). 7 = . . w) @x` (p. 5 @ y1 . .
`=1 xL y1 xL yL zL 30 2 31 x1 z1 6 . w0 )) dx` (p. 5 xy T z = 4 .
7. w) @ (p x (p0 . for every two commodities ` and k. there is a version of associativity. how much more (or less) is chosen of commodity ` per di¤erential increase in the price of commodity k at (p.. xL zL `=1
2
Notice that the matrix xy T cannot be expressed in terms of the dot product between two vectors. w) and is called the (pure) substitution e¤ect. x L y1 xL yL
In matrix notation. The entry in row ` and column k is s`k (p. w) @x` (p. w) . . . .. 7: .The outer product of vectors x and y is a matrix (as opposed to the inner product xT y. so there is no con‡ with the claim that ict the dot product is not associative. . w) = + dpk @pk @w @pk @x` (p. . . w0 ) is the …xed initial demand. 58
. . 54 . 5A = x y T z : = 4 .. . w) = + xk (p. w) lists. which involves a switch from outer product to inner product: xy T z = x y T z since 2 32 3 ! ! x1 y1 x1 yL z1 L L X X 6 . w) @x` (p. y` z` x1 + y` z` xL + . 76 . w) and is called the wealth e¤ect. which is a scalar): 2 3 x1 y1 x1 yL 6 . 7C yL 4 . 5 xy T 4 . The second term measures the e¤ect of the Slutsky compensation on x` (p. The …rst term measures the direct e¤ect of the price change on x` (p. 7B 6 . @pk @w
where x (p0 .3
Substitution Matrix
The substitution matrix (alternatively known as the Slutsky matrix) S (p.
From the total derivative of x. . w) @p1 6 . S (p. .
59
. If Walrasian demand function x (p. w) 2 @x1 (p. dw w0 w = (p0 p) x (p. v T S (p. w) and 8 (p0 . w)) dp dx (p. w) s1L (p. .e. . Under the assumptions. w) (p. w) sLL (p. (They are Walrasian in the sense that there is no implicit adjustment in expenditure after a price change. 8v 2 RL . @xL (p.w) xL @w
3 7 5
= Dp x (p. Proof. = 4 . w) @p1
T
2
@x1 (p. w) is negative semide…nite. i. .. .w) @pL @xL (p. w) x (p. . . w) 6 7 . w) = 4 5 . . Proposition.w) + @w x1 (p. the compensated law of demand holds: 8 (p. . w) 0:
This inequality implies negative semide…niteness of the substitution matrix. w0 ) such that w0 = p0 x (p.w) @pL
+ +
@x1 (p.) Hence 3 s11 (p. w) + Dw x (p.. w) v 0. and the fact that. These are observed choices from the budget set Bp. w0 ) x (p. via a …xed utility level.w) @x` (p. then 8 (p.Even though we denote these choices as Walrasian demands. . (p0 p) (x (p0 . w) is di¤erentiable. Any such compensation is explicit. w) . we do not necessarily assume here that they arise from the utility maximization problem.w that may or may not be rationalizable by an underlying preference. w)
. w) = dp x (p. w) the substitution matrix S (p. w) :
The following is a characterization of the choicebased compensated law of demand.w) @xL (p. homogeneous of degree zero and satis…es Walras’ law and the weak axiom. sL1 (p.
@xL (p.w) + @w x1 (p. w). for a compensated price change.w) xL @w
(p.
w) + Dw x (p. An implication of negative semide…niteness is that all diagonal entries (the ownprice e¤ects) are nonpositive: @x` (p. w) = + x` (p. w) dp` @p` @w 0:
Therefore. we get pT S (p. this implies dx (p. is true provided that S (p. As we will see in a moment. w)) = Dp x (p. w)T dp = Dp x (p. w)T dp:
(using commutativity of the dot product and. letting v = p. that a negative semide…nite substitution matrix implies the weak axiom.
60
. w) dx` (p. a Gi¤en good is necessarily inferior. w) (dp x (p. w) v < 0. w) =dp = S (p. w) dw = Dp x (p. the substitution matrix cannot be negative de…nite for all v 2 RL (i. v T S (p. w) x (p. w) dp + Dw x (p. whenever v is not proportional to p (v 6= p for any ). w) =@w < 0.we have dx = Dp x (p. w) x (p. after switching to matrix notation. it is in fact never de…nite) since. w) p = 0. w) is in addition negative de…nite. w)T dp 0:
Because the magnitude of the price change dp 2 RL was unrestricted. w) dp + Dw x (p.e.e. i. w) @x` (p. w) is negative semide…nite. Substituting into the …rst inequality. associativity and distributivity). dp Dp x (p.
The converse. w) x (p. w) =@p` > 0 only if @x` (p. since @x` (p. w) + Dw x (p. w) dp + Dw x (p.
w) is homogeneous of degree zero. homogeneous of degree zero and satis…es Walras’ law. w) p = 0. w) = 0. and furthermore pT S (p.) This is useful to know when checking negative semide…niteness from the signs of the principal minors. w) satis…es Walras’law. Exercise 39 (MWG 2. w xk (p. w) + x (p. w)T = pT Dp x (p. w) p + Dw x (p. w) = PL 61 p` :
`=1
. Then S (p.Proposition. w)T = 0:
Hence negative semide…niteness is exactly what is implied by the choicebased compensated law of demand (equivalently the weak axiom). zero is an eigenvalue of S (p. The latter implies jS (p. and pT Dp x (p. w) = pT Dp x (p. w) p = p for some p or if jS (p. w) x (p. We will see an example later on. w) + pT Dw x (p. w) p = Dp x (p. w). Proof. which means its determinant jS (p. w). w)T = 0 as well as pT Dw x (p. If Walrasian demand function x (p. Since S (p. w) w = 0 (where the last equality uses Walras’law). w)T p = Dp x (p. w) p = pT S (p. (Recall that is an eigenvalue if S (p. then 8 (p.F. w) p + Dw x (p. w) is di¤erentiable. Previously (in Lecture 5). L.17). : : : . w) have the form: for k = 1. w) = 1 when x (p. w) Ij = 0. w)j = 0. w) p + Dw x (p. w)j = 0 if = 0. w) x (p. w) + x (p. Let the Walrasian demand function x (p. w) w = 0: when x (p. we established Dp x (p. S (p.
16). Let the Walrasian demand function x (p. 1. (b) Demonstrate that x (p. x2 (p. then the duality between utility minimization and expenditure minimization lets
62
. w) = Demonstrate that the substitution matrix is negative semide…nite. w) = p1 + p2 + p3 p2 1 p1 x3 (p. p3 w : x3 (p. w) = : p1 + p2 + p3 p3 x1 (p. at p = (1.F.
7. w) = (a) Con…rm that x (p. w) = p3 x1 (p. p3 p1 x2 (p. (b) satis…es Walras’law. w) where p2 1 . p1 + p2 + p3 p1 1 p3 . w) = . w) (a) is homogeneous of degree zero. (b) is consistent with the weak axiom. Exercise 40 (MWG 2. (c) has a negative semide…nite. Exercise 41 (MWG 2.F.Determine whether x (p. (c) Show that 8v 2 R3 . but not symmetric. v S (p. symmetric substitution matrix. w) be as follows: p2 . Show that it does not satisfy the weak axiom.4
Substitution Matrix with Preference
If we allow choices to arise from a standard preference. w) does not satisfy the weak axiom. w) is homogeneous of degree zero and that Walras’ law holds.10) Compute the substitution matrix for Walrasian demand function x (p. 1). w) v = 0.
u) around = 0 in displacement direction v 2 RL is
2
7 5 = Dp h (p. u) v
. w)
.
. u) by the envelope theorem. w): @h (p. so that 2 Dp h (p. w) is negative semide…nite and symmetric (i. the secondderivative matrix) of any concave function is negative semide…nite. v (p. L). locally nonsatiated. . Rewriting the substitution matrix accordingly as 2 @h1 (p. The Taylor expansion of e (p.u) @p1
@hL (p.
@hL (p. u) =dp` for `.u) @h1 (p.
@pL
. I will prove this general result.u) 3 6 S (p. u) is a concave function. w) = 4
@p1
s`k (p. In order for a di¤erentiable Hicksian demand function h (p. u) =dpk = dhk (p. k = 1. w) = = @p dp dp for `. : : : . Proof. L.e.us express the substitution matrix in terms of Hicksian demand at u = v (p. we can appeal to the fact that the Hessian (i. u) + is continuously di¤erentiable at u. strictly convex preference on X = RL . then the matrix Dp h (p. u) to exist as a solution to the expenditure minimization problem. w) is now implicitly con…ned to choices the re‡ an underlying standard preference. locally nonsatiated.e. dh` (p. u) : Since e (p. u)T ( v) + 63
2
2 v T Dp e (p + "v. Recall …rst that h (p. u) . u) = e (p. I emphasize that the equivalence with a given Walrasian + demand function is only valid if the conditions for the existence of a Hicksian demand function hold.. k = 1. . u) dh (p. .u) @pL
a further property can be derived. strictly convex preference on X = RL . u) + rp e (p. ect Proposition. . : : : .
. u) = rp e (p. u ( ) has to represent a continuous. u) = Dp e (p.
e (p + v. u) = S (p. because S (p. w)) dx (p. and the Hicksian demand function h (p. If u ( ) represents a continuous.
e. This establishes that Dp e (p.
64
. therefore requires both negative semide…niteness and symmetry of the substitution matrix. u) = Dp h (p. 8p. Symmetry comes from Clairaut’ theorem. Symmetry was not an implication of the choicebased compensated law of demand. u) + e (p + (1 ) v. u) 1 e (p. 8p0 and 8 2 [0. u) . u) v
0. if the magnitude of the displacement in the Taylor expansion is chosen su¢ ciently small. u) by virtue of the fact that its second derivatives are the …rst derivatives of h (p.
The preferencebased compensated law of demand. u) e (p. u). which states that the Hessian s of any function that has continuous second partial derivatives at all points in the domain is symmetric. u) is negative semide…nite. in direction v). u) = S (p. Since this is true for e (p. Then
2 v T Dp e (p + "v. Now.
p for p0 . u) v
0. then " is very close to zero. u) + rp e (p. 1]. which is implicit in Hicksian demand. so that (after rearrangement) e (p. u) e (p.for some " 2 [0. i.
e (p + v. and we have
2 v T Dp e (p. u) . u)T ( v) :
This is apparent when. in the de…nition of concavity.
2 where the direction v 2 RL was arbitrary. w) must be symmetric. which were 2 assumed to be continuous. u)
and we apply the limit as ! 1 (making the last term a derivative with respect to p. e ( p + (1 we substitute v p0 ) p0 . u) + (1 ) e (p0 . Dp e (p.
because the concavity of the expenditure function implies e (p + v. ].
Proof. locally nonsatiated. w) = w 4 p1 p2 5 2 9 1 1 1 2 p2 p1 p3 p2 p3
3
Clearly. : : : . w) = : 3 p1 3 p2 3 p3
Observe that the choices satisfy homogeneity of degree zero and Walras’law (they add up to w). we con…rm that the substitution matrix will not be negative de…nite. u) @ pk X @h` ( p. Di¤erentiating both sides with respect to . u) p = S (p.
L L X @h` ( p. @ pk @ @pk k=1 k=1
which corresponds to the claim. Under these conditions. Suppose choices have the CobbDouglas form: x1 (p. w) v = = 1 w 9 1 w 9 2
2 v1 v1 v2 v1 v3 +2 +2 2 p1 p1 p2 p1 p3
2 v3 p3
2 v2 v2 v3 +2 2 p2 p2 p3 2
2 v3 p3
2 v3 p2 3
v1 p1
v2 p2
2
+
v1 p1
+
v2 p2
2
!
0
(with equality only if v = p for any ). w) p = 0. The substitution matrix is easily calculated from partial derivatives to be 2 3 1 2 p12 p11p2 p1 p3 1 6 1 1 2 p12 p21p3 7 : S (p. w) = 1w 1w 1w . u) is di¤erentiable at u. v T S (p. Proposition. Hicksian demand is homogeneous of degree zero in p. u) = pk = 0. this is symmetric. L. and h (p. Example. we have for ` = 1. u) = h (p. that it is negative semide…nite because 8v 2 RL . so that h ( p. One can check directly.Before we examine this di¤erence in examples. x2 (p. x3 (p. 65
. then + Dp h (p. w) = . u). Hence CobbDouglas choices satisfy the both the choicebased and the preferencebased compensated law of demand. If u ( ) represents a continuous. strictly convex preference on X = RL .
w) sjl (p. w) = p2 1w 1w + . If row r and column c are deleted. w) = 3 p1 p1 3 p2 1. which are negative. with m odd) are nonpositive and all even principal minors (i. w) = w + p2 + w 3 3 `=1
1 p2 + w = w: 3
The substitution matrix 2
1 p2 1 1 p1 p2
is obviously not symmetric. jS (p. c) = 0 if r + c is even.c) w pi pj pk pl 1 p2 + 3 w . It is worth checking once that. One technique is to compute the prinicipal minors. Since Mrc is negative if and only if r + c is odd. which we know is always zero for S (p. Mrc = sik (p.e. w)j = s11 (p.g. w) = w 4 9
2 + 6 p2 w p2 1+6w
1 p1 p3
1 p1 p2 1 p2 2
1 + 3 p2 w p2 2+3w
1 p2 p3
1 p1 p3 1 p2 p3
1 + 3 p2 w p2 1 3w 2 p12
3
3 7 5
where I (r. and k and l are the lowest and highest nondeleted columns. it is more convenient to check negative semide…niteness indirectly. A matrix is negative semide…nite if all odd principal minors (i. w) M12 + s13 (p. w). w) sil (p.e. x3 (p. The secondorder minors have the following form. which are determinants of n n matrices derived by deleting n m corresponding rows and columns.
3 X
1 1 p` x` (p. w) = ( 1)I(r. indeed. with m even) are nonnegative. In this case. x2 (p. w) M11 s12 (p.e. The …rstorder principal minors are the diagonal entries. w) M13 = 0: 66
. and I (r. w) sjk (p.
1 6 S (p. w) = 1w : 3 p3
Neither homogeneity of degree zero nor Walras’law are violated. The thirdorder principal minor is the determinant of the matrix. e. and i and j are the lowest and highest nondeleted rows. c) = 1 if r + c is odd. the secondorder principal minors are all positive. Consider now the following modi…cation: x1 (p. Mrc is a principal minor) implies r + c is even.Example. and r = c (i.
it is important to know whether the demand function could in fact occur if the agent were optimizing with respect to some rational preference. 6): 2 3 10 ? ? 4 ? 5: S (p. The preferencebased approach is more special in nature. w) is di¤erentiable. Exercise 42 (MWG 3. rather than from preferences. 2. admits an underlying preference). homogeneous of degree zero and sats…es Walras’law.
Exercise 43 (MWG 2. since the substitution matrix fails to be symmetric. where x (p. but not in the choicebased approach using the weak axiom.e.5
Integrability
There is a deeper signi…cance to the observation that the substitution matrix is symmetric in the derivation from preferencebased demand. and is also equivalent to the compensated law of demand. when we build a theory from a demand function with certain properties.G. The following are Walrasian substitution e¤ects for an agent who has rational preferences and faces prices p = (1. w) = 4 ? 3 ? ?
Find the missing entries. Show that. the compensated law of demand and symmetry of the substitution matrix su¢ cient for rationalizability? 67
.11).F. which implies that the weak axiom holds (because all lowerorder principal minors have strict signs.14). Hence there is hope that a demand function that has both properties is rationalizable. S (p. We know that the weak axiom guarantees that the choice structure is rationalizable (i. These choices do not satisfy the preferencebased compensated law of demand. as is evident from the fact that it implies both the compensated law of demand and symmetry of the substitution matrix. the negative de…niteness for v 6= p obtains). in a twocommodity world. This is known as the integrability problem: are homogeneity of degree zero.
7.Hence the matrix is negative semide…nite. w) is always symmetric. Walras’ law. From a modeling point of view.
choices are rationalizable (given that they satisfy homogeneity of degree zero. A preference is fully described by its upper contour sets % (x) = = a 2 RL s. at any given prices. e (p. as %u(x) = a 2 RL s.
Equivalently we can express % (x).t. we need insights from abstract duality theory. Walras’law and the compensated law of demand) if the substitution matrix is symmetric. 8p + 0.e. so rationalizability refers here to the existence of a generating preference that has all the standard properties. To see why these bundles form the upper contour set of x. A hyperplane that separates the bundle y 2%u from the set %u = is parameterized by some vector p and scalar eu such that p y < eu 68 p x
L % (x) = a 2 R+ s. using the duality between utility maximization and expenditure minimization. and a level u. The sets of interest to us are the upper contour sets of a continuous and convex preference relation. The symmetry of the substitution matrix is also necessary for rationalizability in terms of such preferences. not just completeness and transitivity.t. u (a) + a 2 RL s. (We also presume that preferences are convex and continuous on in this discussion. I. we must think about how a preference would be recovered from choices x (p.t. u (x)) p a :
This is the set of bundles that are more costly than the cheapest bundle required to attain u (x).The answer is yes. Duality theory revolves around the idea that any closed. a % x + u (x)
for some utility function that represents %. convex set can be described in terms of its linear approximation by hyperplanes that separate the set from points not in the set. w). u (a)
u (x)
. and denote the upper contour set by %u .t.) To understand the connection. Fix a utility function u ( ) that represents %.
for all x 2%u . Given %u , the separating hyperplane theorem says that such a hyperplane (i.e. p and eu ) exists for every y 2%u . Each hyperplane separates = RL into two halfspaces, one containing the point y and the other containing the set %u . The intersection of the latter halfspaces is the original set %u , since it excludes all points not in the set. For every p, there is an element x 2%u that generates the smallest value of p x attainable in %u . The function that maps p to this value at x is called the support function of %u : eu (p) inf fp x s.t. x 2%u g :
This function is necessarily concave in p (by the argument we already made for the concavity of the expenditure function  it increases at most linearly in p). Clearly, we have eu (p) p x for all x 2%u , at all p. It follows from the separating hyperplane theorem that there exists some p 0 for every y 2%u such that eu (p) > p y. Hence, no y 2%u can satisfy = = eu (p) p y for all p 0. Then %u can be expressed as %u = x 2 RL s.t. 8p 0; eu (p) p x :
The upper contour set at u in the expenditure minimization problem is convex and closed if preferences are convex and continuous. This set can be approximated by tangent budget hyperplanes that contain bundles x in the upper contour set at which p x is minimized, given prices. By …nding such a budget hyperplane for every p, we trace the boundary ot the upper contour set. Knowledge of the minimum expenditure p x at every price, i.e. the expenditure function (which is the support function) would allow us to recover the upper contour set. See Figure 9. Duality theory lets us infer the upper contour set at every utility level u and therefore recover the preference relation, once we have an expenditure function. What remains is to obtain the expenditure function from choices. 69
Figure 9: Upper contour set approximated by e (p; u)
This is the proper integrability problem, the question whether the system of partial di¤erential equations rp e (p; v (p; w)) = h (p; v (p; w)) = x (p; w) has a solution. The Frobenius theorem gives necessary and su¢ cient conditions for integrability that are, in particular, satis…ed if the derivative matrix is symmetric. As we have seen, the derivative matrix of rp e (p; v (p; w)) is the substitution matrix S (p; w) = Dp h (p; v (p; w)). Hence symmetry is exactly what is needed for recovering preferences from choices, i.e. for choices to be rationalizable. Exercise 44 (MWG 3.H.6). Derive expenditure and utility function from the Walrasian demand function x` (p; w) = ` w=p` for ` = 1; : : : ; L with PL `=1 ` = 1. Exercise 45 (MWG 3.H.5). How can expenditure and utility function be recovered from the indirect utility function?
70
8
8.1
Aggregation
Aggregate Demand Function
In general, aggregate demand is not a function of aggregate wealth, but rather a correspondence even if individual demands are functions. At a given level of aggregate wealth, di¤erent wealth distributions lead to di¤erent choices, individually and in the aggregate. In this lecture, we consider the special circumstances under which aggregate demand is a wellde…ned function of aggregate wealth. We also ask whether an aggregate demand function has welfare content. Individual demands are outcomes of utility maximization and therefore represent the best choices available to an agent. Can we say that aggregate demand re‡ ects the best consumption choices available to society? Finally, we consider whether aggregate demand inherits the weak axiom from individual demands, which is an important condition for the uniqueness of general equilibria. Let there be I consumers i = 1; : : : ; I with rational preference relations %i and individual wealths wi . Their Walrasian demands are denoted by xi (p; wi ). Then aggregate demand is: x (p; w1 ; : : : ; wI ) =
I X i=1
xi (p; wi ) :
Exercise 46 (MWG 4.C.11). Let two agents have identical wealths w1 = w2 = w=2, and let their preferences of over bundles of two goods have utility representations p u1 (x11 ; x21 ) = x11 + 4 x21 and p u2 (x12 ; x22 ) = 4 x12 + x22 :
(a) Derive the individual demand functions and the aggregate demand function. 71
e. it cannot a¤ect aggregate consumption of any commodity if x is to be a function of aggregate wealth only. @x`i (p. : : : . wI ) and (w1 . j 2 I. Since dw is a redisi=1 tribution of wealth that does not a¤ect total wealth.e. L. wi ) = i=1 wi i=1 PI 0 0 0 xi (p.(b) Find the individual Slutsky matrices Si (p. : P. i. (With two goods.
I X @x`i (p. so that we can write ! I I X X x (p. : : : . wi ) i=1
@wi
dwi = 0
for commodity ` = 1. Demand functions have this property at any prices and wealth distribution if and only if preferences admit an indirect utility function of the Gorman form: vi (p. : : : . Because dw might a¤ect only two individuals. w=2) for i = 1. wI ) that distribute the i=1 same aggregate wealth w. Does aggregate demand satisfy the weak axiom? In general. w). I. Can we build a theory where only the aggregate wealth PI P w a¤ects aggregate demand? I. wI ) = xi (p. wi ) for all (w1 . w1 .e. : : : . 8i. wi ) @x`j (p.) Demonstrate that dp S (p. when is I xi (p. : : : . wj ) = @wi @wj for ` = 1. the entire matrix is determined by one element. L. dwI ) such that I dwi = 0. : : : . wi ) = ai (p) + b (p) wi : 72
. aggregate demand depends on prices and all individual wealth levels. 2 and the aggregate Slutsky matrix S (p. and leave everyone’ s wealth unchanged. wi ?
i=1 i=1
Consider a wealth distribution (w1 . w) dp < 0 for all dp 6= 0 that are not proportional to p. it must be that the wealth e¤ect for each commodity must be the same for any two individuals. wI ) and some di¤erential change :: in wealth (dw1 . wi ) = x p.
w) = rp v (p. and replacing e (p. ) is + di¤erentiable at (p. locally nonsatiated and strictly convex preference on X = RL . Since the wealth e¤ect is the same for all individuals and all wealth levels. then x (p. Hence rp v (p. e (p. w) : @v (p. u) = x (p. and using rp e (p. w) x (p. 8p. we have x (p. rwi x (p. u)) rp e (p. w) = @v (p. and the indirect utility function v ( . e (p. w) + as claimed. If u represents a continuous. u)
Evaluating at p = p. w) = u. u) with w. u) = 0: @e (p. By duality of the utility maximization and expenditure minimization problems. w) =@w rp ai (p) b (p) rp b (p) wi : b (p)
Di¤erentiating with respect to wealth. @v (p. u)) = u. w). e (p. u) = h (p. b (p)
which is independent of wi .For the "if" part. v (p. we have rp v (p. u)) + @v (p. w) 0. s Proposition. w) = rp b (p) . w) =@w
Proof. w) = 0 @w
Substituting the Gorman form of the indirect utility function into Roy’ s identity. 73
. Suppose v (p. we need Roy’ identity. aggregate demand depends only on aggregate wealth. w) = rp v (p.
s x (p. we have v (p. Example. by a new function e (p. w) = b (p) w where b (p) 1=~ (p). for a homothetic preference. "`w (p. w) = rp b (p) w: b (p)
Hence. w) = x (p. a property that you know from demand functions associated with CobbDouglas utility functions. tu) : Denote the expenditure required to reach u = 1. but the Gorman form is in fact necessary for the existence of an aggregate demand function. w) w = 1: @w x (p. u) = ue (p. u) = w and u = v (p. Since e (p. i. 1) ~ (p). u) = t (p x) = p (tx) = e (p. : : : . w) b in the corresponding utility maximization problem. demand increases linearly (and proportionately for all commodities) in wealth. w) =w. and the income elasticity of demand is. Preference is quasilinear in good k if x y implies x + ek y + ek . Using Roy’ identity. then u (x) = u (y) () u (tx) = u (ty) for all t > 0. It follows that the utility function is homogeneous of degree one. Furthermore. Preference % is homothetic if x y implies tx ty for all x. L. the expenditure function is homogeneous of degree one in u: te (p. Now one can write the expenditure at an arbitrary b utility u as e (p.e. We can see directly how the Gorman form relates to linear wealth expansion paths in this case. w)
This means a …xed share of the budget is spent on each commodity. Example. which belong to the class of homothetic preferences. w) = @x` (p. at …xed prices. where > 0 and ek is a bundle of one unit of commodity k (and zero 74
. y and all t > 0. 1) = u~ (p). for commodity ` = 1. given prices. Thus rw x (p.This argument only provides su¢ ciency. If u is a utility function that represents homothetic preference %. u (tx) = tu (x). Hence the indirect utility b function has the Gorman form.
if other commodities enter additively. w) = a (p) b (p) w:
The wealth e¤ects of quasilinear demands are constant 1 @x` (p. @xk all marginal utilities are constant with respect to consumption. the wealth expansion path is in this case parallel to the kaxis.e. w) refers to quantities of all commodities other than k. w). such a condition …xes the quantity at a speci…c level. then it is a condition of optimality that only one of these is consumed. Recall that utility maximization entails the "tangency" p` @u (x) =@x` = @u (x) =@xk pk for all ` (at an interior solution). For all commodities that enter nonlinearly.e. the indirect utility function therefore has the Gorman form: v (p. Thus. I. As a consequence. it is a straight line through the origin (i. w) @xk (p. w) + f (x k (p. hence only xk (p. e. Letting a (p) f (x k (p)) g (p) and b (p) = 1=pk . With homothetic preference. w) = .g. w) = pk pk `6=k=1 (all remaining wealth is spent on commodity k). w) can adjust to a wealth change at a solution. (The notation x k (p. u (x) = xk (p.of any other commodity). It follows that ! L X w 1 p` x` (p) = w g (p) xk (p. Only commodity k has a constant marginal utility at all levels of xk (p. xk (p. w) must enter additively into the utility function. w)). =0 @w pk @w for all ` 6= k. Since @u (x) = 1.) The function f is nonlinear . consuming commodity k does not a¤ect how the agent values other commodities.
75
. the zero bundle).
there may be a …xed wealth distribution rule (w1 (p. Without it. If we know what the distribution is.e.g. w) . by redistributing wealth.) If wealth e¤ects are nonuniform. e.
De…nition. ~
without imposing uniform wealth e¤ects. w)) that depends only on prices and aggregate wealth w. I. we need to specify how we would evaluate a particular list of individual outcomes. : : : . x 6= x (p. the mean). One statistic is aggregate wealth (i. one cannot talk about improving on a given aggregate bundle.e. Then aggregate demand can be written as a function of aggregate wealth. w) .
8. (Recall that we normally need them because aggregate wealth could be distributed in many ways. we ask now whether it is rationalizable in the sense that there exists a preference based on which a hypothetical consumer could make these choices on behalf of the entire population. 8 (p. w) is the Walrasian demand function generated by %. wi (p. aggregate demand may depend on certain statistics of the wealth distribution. w) admits a positive representative consumer if there exists a preference % such that x (p. wI (p. Then aggregate demand may be a function of prices and distributional statistics (rather than full information linking each preference to a particular individual wealth).
I X i=1
xi (p. w)) =
I X i=1
xi (p. then multiplicity is not an issue.2
Representative Consumer
Given that we have an aggregate demand function. To actually compare aggregate demands. w) x. we may also have to observe variance and higherorder moments. The rule could 76
. w) and p x w =) x (p. This is a precondition for the aggregate demand function to have some welfare content. and the distribution of wealth is not …xed. The aggregate demand function x (p. w) and 8x.In some cases.
Exercise 47 (MWG 4. De…nition. w) via Roy’ identity is s the aggregate of the individual demands underlying (v1 (p. Bp. w). wI )) s. wI ). The optimal (feasible) distribution of wealth (w1 . and the distribution function that solves W ( ) is such that individual wealth wi (p. w). the maximum at given prices and aggregate welath. thus constructed has the usual properties of an indirect utility function (i.
I X i=1
(w1 . or something else. It can be shown that v (p. we record the value of W ( ) for every possible distribution of w among the I individuals. Con…rm that v (p. : : : . w) (given social welfare function W ( )) arises as follows. w) :
The aggregate indirect utility function v (p. vI (p.be utilitarian (adding up individual utilities) or egalitarian (preferring less variation between individual utilities).t. w) is the maximum of a utility function that represents the preference of a positive representative consumer.e. vI (p. For every price vector p and aggregate wealth level w. is the indirect utility at (p. concave and di¤erentiable.2).wI as determined by the distribution and prices. who are assumed to maximize their utility from consumption within their budget sets Bp. The highest achievable value of W ( ) among all distributions. 77
. the Walrasian demand function derived from v (p.e.w1 . w1 ) . wI )). i. : : : .:::. A BergsonSamuelson social welfare function W : RI ! R assigns a value to every utility vector (u1 . increasing in w. decreasing in p and quasiconvex). is that which attains max W (v1 (p. : : : .wI )
wi
w
v (p. I. given a social welfare function W ( ).e.D. w) is di¤erentiable in price and homogeneous of degree one for all i. : : : . homogeneous of degree zero. w1 ) . : : : . uI ) for the I agents. Suppose W ( ) is increasing.
wi ) = . subject to I wi = w. wI ) at the wealth distribution (w1 (p. : : : . w) the wealth distribution that maximizes W ( )). Example. xI (p. There may well be a preference that rationalizes it.Since the aggregate demand function attains maximal utility v (p. w) at all (p. w)) that maximizes W ( ). : : : . Hence there exists a preference with respect to which the aggregate demand at every (p. : : : . for i = 1. uI ) = i ln ui
i=1
with i > 0 for all i. The aggregate demand function x (p. L. w) s that is 8 (p. w) . w). 78
PI
. What is the aggregate demand function of a normative representative consumer? The …rstorder conditions P to maximize W ( ). w) is the utilitymaximizing bundle. The normative representative consumer’ preference chooses an aggregate demand function x (p. i=1 @W (u1 . : : : . In this particular case. : : : . The normative requirement is much stronger. which arises from suboptimal distributions of wealth (according to W ( )). are. w) (we could construct a complete utility function from all values of W ( )). : : : . it is chosen by the preference underlying v (p. and i=1 i = 1. w) consistent with individual choices x1 (p. w) (which re‡ects at every (p. w1 ) . Suppose all agents have homothetic preferences and the social welfare function is I X W (u1 . This means we have a positive representative consumer. wI (p. uI ) @vi (p. w) admits a normative representative consumer with respect to welfare function W ( ) if there exists a preference % that generates the welfaremaximizing aggregate demand function x (p. De…nition. the positive representative consumer is moreover a normative representative consumer. wI ). @ui @wi where is the Lagrange multiplier and derivatives are evaluated at the welfaremaximizing wealth distribution (w1 . Note that a positive representative consumer could choose some other aggregate demand function.
wi ) for all i. : : : . w)).Since
@W (u1 .
Exercise 48 (MWG 4. uI ) i i = = @ui ui vi (p. the individual demands at the optimal wealth distribution rule (w1 (p. w) can alternatively derived by solving the problem
x1 . w) = b (p) w as derived above for homothetic preference). w))).1). wI (p. we have i =wi on the left side of the …rstorder conditions. wi )
@vi (p. where a planner chooses consumption bundles.xI
max
W (u1 (x1 ) . wi ) = @wi wi (follows from v (p.:::. Argue that any normative representative consumer must then prefer (p0 .
i w)
are given by the social welfare function). Exercise 49 (MWG 4. uI (xI )) s. : : : . w0 ) to (p. p
I X i=1
xi
w
for (x1 (p. wi ) vi (p. we have a version of the second welfare theorem. wI (p. Summing wi = i = over i. 79
. w1 (p.D.8).t. Suppose for any distribution (w1 . the
i I X i=1
iw
for all i. Then the
xi (p. Since this formalization. w) = welfaremaximizing aggregate demand function is x (p. Show that v (p.D. : : : . xI (p. wi ) > vi (p. w). : : : . PI I X 1 i w= wi = i=1 = :
i=1
and
So the optimal wealth distribution is wi (p. w) = (remember. wI ) of w0 such that vi (p0 . w) . w)) . is equivalent to the one where the planner chooses wealth levels (and lets consumers make their consumption decisions in the market). : : : . wI ) of 0 0 0 w there is a distribution (w1 . : : : .
w1 ) 0 and x2 (p . w1 ) = 0. x2 (p0 . w2 ) = (2. not commodities. (We have wellde…ned orderings x1 (p. x2 (p. then x (p0 . since bundle x1 (p0 . w2 ) = 3 . w0 ) can only be chosen if x (p. p2 ) = (1.3 : 2
Since both aggregate bundles cost w = 10 at the prices at which they are chosen and less than 10 when they are not chosen. w0 ) in one situation.8. w) = 3. It is clear that aggregate demand inherits homogeneity of degree zero and Walras’law from individual demand functions.) These demands satisfy the weak axiom.) In general. p02 ) = (2. w) is unavailable. w0 ) w and x (p. w2 ) x2 (p. 2). The de…nition of the weak axiom in the Walrasian demand setting extends directly from individual to aggregate demands: it says if p x (p0 . for agent 2. w) 6= x (p0 . they violate the weak 80
. (If aggregate bundle x (p. as well as x1 (p0 . 1)
at prices (p1 . w0 ). w=2) is costs more than w2 = 5 at prices p0 . w) = 7 . w1 ) = (3. w) of prices and aggregate wealth. the weak axiom does not survive aggregation. then p0 x (p. and bundle x2 (p.) Aggregate demands are x (p. 2 . w=2) costs more than w1 = 5 at prices p. x (p0 .3
Failure of the Weak Axiom
Suppose aggregate demand is a wellde…ned function x (p. w1 ) x1 (p0 . 1) 2
at prices (p01 . 1). 7 2 . the subscripts refer to individuals. w) > w.
Example. w=2). 5 2 . w) was chosen over x (p0 . (Here. Let wealth w = 10 be distributed equally (w1 = w2 = 5) and consider demands x1 (p.
Figure 10: Failure of the aggregate weak axiom
axiom, since each is available in budget sets Bp;w and Bp0 ;w , but x (p; w) 6= x (p0 ; w). (There is no wellde…ned ordering, since x (p; w) x (p0 ; w) on 0 Bp;w , and x (p ; w) x (p; w) on Bp0 ;w .) As can be seen in Figure 10, the scaleddown aggregate bundles lie inside both individual budget sets (which are, in this case, scaleddown versions of the aggregate budget sets). Even though the weak axiom fails in general, it does hold when individual demands are always decreasing in prices (for compensated and uncompensated price changes). We know from individual consumer theory that this is not a compelling property. It requires (pure) substitution e¤ects to be su¢ ciently large to cancel out any income e¤ects of inferior goods. De…nition. Individual demand function xi (p; wi ) satis…es the uncompensated law of demand (ULD) if 8p; p0 and 8wi , (p0 p) (xi (p0 ; wi ) xi (p; wi )) < 0 when xi (p; wi ) 6= xi (p0 ; wi ). ULD for the aggregate demand function xi (p; wi ) is the same property without subscripts.
81
Unlike the weak axiom, ULD does survive aggregation. Proposition. If every individual Walrasian demand function xi (p; wi ) satis…es ULD, then aggregate demand x (p; w) satis…es ULD. Proof. If x (p; w) 6= x (p0 ; w), then for some i, (p0 p) (xi (p0 ; wi ) xi (p; wi )) < 0
(for all other agents, the inequality is nonpositive). Summing over I, we have 8p; p0 and 8w, (p0 p) (x (p0 ; w) x (p; w)) < 0:
The aggregate version of ULD, in combination the other standard properties, implies the weak axiom. Thus, we can give conditions for on which aggregate demand satis…es the weak axiom, but these conditions do not have choicetheoretic foundations. Proposition. If aggregate demand x (p; wi ) satis…es homogeneity of degree zero, Walras’law and ULD, then it satis…es the weak axiom. Proof. Given two pricewealth pairs (p; w) and (p0 ; w0 ) with x (p; w) 6= x (p0 ; w), let p x (p0 ; w0 ) w. The weak axiom requires p0 x (p; w) > w0 , so that x (p; w) is revealed preferred to x (p0 ; w0 ). Let p00 (w=w0 ) p0 and note that, because p00 =w = p0 =w0 , homogeneity of degree zero implies x (p00 ; w) = x (p0 ; w0 ). Thus. p00 x (p00 ; w) = (w=w0 ) (p0 x (p0 ; w0 )) = w (by Walras’law). From ULD, (p00 p) (x (p00 ; w) x (p; w)) < 0: Since p x (p00 ; w) = p x (p0 ; w0 ) p00 (x (p00 ; w) w and p x (p; w) = w, it is necessary that p00 x (p; w) < 0;
x (p; w)) = w
i.e. p00 x (p; w) > w. Then (substituting for p00 ), p0 x (p; w) > w0 .
82
The compensated law of demand holds if the price derivative matrix of Hicksian demand, Dp hi (p; ui ), which is equal to the substitution matrix Si (p; wi ) of compensated price e¤ects, is negative semide…nite (and de…nite when pre and postmultiplied by vectors that are not proportional to p). Similarly, choices satisfy ULD if the price derivative matrix of Walrasian demand, Dp xi (p; wi ), has the same properties. Exercise 50 (MWG 4.C.1). Show that xi (p; wi ) satis…es ULD only if Dp xi (p; wi ) is negative semide…nite, and conversely, if Dp xi (p; wi ) is negative de…nite (except v T Dp xi (p; wi ) v = 0 when v = p for some ), then xi (p; wi ) satis…es ULD. Homothetic preferences are a special case where choices respect ULD (and therefore the weak axiom). Example. With a homothetic preference, we saw that rwi xi (p; wi ) = xi (p; wi ) =wi . Rearranging Si (p; wi ) = Dp xi (p; wi ) + rwi xi (p; wi ) xi (p; wi )T gives 1 xi (p; wi ) xi (p; wi )T : Dp xi (p; wi ) = Si (p; wi ) wi Since v T Si (p; wi ) v 0 (strictly if v is not proportional to p) and v T xi (p; wi ) xi (p; wi )T v = 2 v T xi (p; wi ) 0, we have 8v 2 RL , v T Dp xi (p; wi ) v 0 (strictly if v is not proportional to p). This makes Dp xi (p; wi ) negative de…nite for every individual, so that ULD holds in the aggregate and implies the weak axiom. Exercise 51 (MWG 4.C.6). Verify that the following claim is true in the case of a homothetic preference %i . If %i can be represented by a twice continuously di¤erentiable, concave utility function ui ( ).and 8x, xi D2 ui (xi ) xi < 4; xi rui (xi ) then xi (p; wi ) satis…es ULD.
83
w) does not satisfy ULD. w] with density ~ nonincreasing in wealth.e. pN 1 and the requirement that probabilities sum to 1. if N = 2. pn 0 for n = 1.C.e. ~
9
9.
n=1
(It is not N dimensional because pN is determined by p1 . E. the distribution can be represented by a point p 2 [0.7).
De…nition. : : : .)
84
. C = X could be the set of consumption bundles). show that the aggregate demand function Z w x (p. N . On the other hand.Exercise 52 (MWG 4. 1] in the onedimensional line space. i.g. pN ) on P the set of consequences C. and N pn = 1. the set ( ) N X N pn = 1 : = p 2 R+ s. we introduce risk into the choice framework. The set of N possible outcomes will be denoted by C (for consequences. : : : . w). If every individual has the same consumption function x (p. and individual wealth w is distributed on [0. e.g.t. A simple lottery is a probability distribution L = (p1 . …rst increasing and then decreasing) for which x (p. w) dw ~ x (p) =
0
satis…es ULD. : : : . n=1 A simple lottery is an element of the (N 1)dimensional simplex. demonstrate that there are unimodal distributions of wealth (where the density function is singlepeaked.1
Expected Utility
Lotteries
In this lecture. I. The decision maker’ choice induces a probability distribution on C: which of the s consequences will be realized is not certain at the time the choice is made.
which is labeled p2 . p3 ) is mapped to the unique point whose perpendicular distance from the three edges re‡ ects the probabilities of the consequences. to a simple lottery L = (p1 . and a lottery L = (p1 . : : : . etc. The simplex is drawn with height 1. K ) on a set PK of K simple lotteries L1 . there is a vertex. is the probability of consequence 2. De…nition. A compound lottery is a probability distribution ( 1 . : : : .Figure 11: Simplex
Figure 11 depicts such a simplex for three consequences. It is then an easy matter to reduce a compound lottery on L1 . is equal to the probability of consequence 1. This technique takes advantage of the fact that the lengths of these perpendicular lines from a point to the edges always sum to 1. For each consequence. Speci…cally. Hence the points in the simplex can represent probability distributions. LK . pk . K. LK . p2 . and k=1 k = 1. the distance of point L from the edge opposite 1. : : : . pN ) such that 1 N pn = 85
K X k=1 k k pn
. where Lk = pk . The distance from the edge opposite 2. : : : . along the perpendicular line labeled p1 . : : : . : : : . where k 0 for k = 1.
: : : . 86
. and 2 [0.for n = 1. L00 % L + (1 ) L00 g :
Mixture continuity is a weaker property than continuity of % (i.2
Preference over Lotteries
Analogously to the certain setting.
De…nition. i.t. From two lotteries L and L0 we can obtain a new lottery L+(1 ) L0 . the probability of consequence n in the reduced lottery is the result of adding probabilities p1 . 1] s. upper and lower contour sets of L00 are closed). 1] s. as is the set of mixtures of L and L0 to which L is preferred.(L00 ).e. where the probability of consequence n in L + (1 ) L0 is a weighted 0 average of n’ probabilities in L and L . is closed. on the premise that the decision maker views any compound lottery as equivalent to the simple lottery it reduces to. the set of mixtures of L and L0 that are preferred to L00 . : : : . Then we add further axioms. of course. L0 2 L and 8L00 2 L. We can write the reduced lottery as a vector sum: L=
1 L1
+
+
K LK
2
:
Our focus will be entirely on simple lotteries. Whereas continuity implies that all convergent sequences in % (L00 ) have limits in % (L00 ). LK . to . where each pk is weighted by the probability k that n Lk is realized in the compund lottery. 1] is the weight of s L. N . pK over simple lotn n teries L1 .t.e. I.
9. all simple lotteries. f 2 [0. Denote the set of all simple lotteries by L. mixture continuity only requires this of a restricted set of sequences (those that can be constructed by mixing two lotteries and varying the weight ). Preference % on L is mixturecontinuous if 8L. we start by endowing the agent with a rational preference relation that ranks all elements of L. : : : . L + (1 ) L0 % L00 g .e. The same applies. f 2 [0.
L % L0 () L + (1 ) L00 % L0 + (1 ) L00 8L00 2 L and 8 2 (0. 87
. and prefer smallscale tax evasion to a mostly lawabiding life with a small chance of committing a violent crime (mixing L and L0 ). L0 . ~ then L % L0 only if L % L0 . De…nition. "would you risk your life to watch a sports game?. Preference % on L satis…es independence if 8L. If you ask." you are likely to get a di¤erent answer than if you ask.) E. One can think of some instances where people may violate continuity for extreme . We might prefer a completely lawabiding life (lottery L0 ) to smallscale tax evasion (lottery L00 ). most of us would never commit a violent crime (lottery L). however. "would you drive to your friend’ house to watch the game?" s Continuity plays the same role here as under certainty: it guarantees that there exists a utility representation U : L ! R such that L % L0 () U (L) U (L0 ). The independence axiom says that preference between L and L0 should not be a¤ected by rescaling the probabilities in both lotteries in a given direction. There is an obvious relationship between consequence n and the degenerate lottery Ln that assigns probability 1 to consequence n (and zero to all others). Note that it is conventional to use the capital letter U to distinguish a utility representation for a preference over lotteries from a utility representation for a preference over consumption bundles or consequences. 1). We have trouble imagining very small probabilities and treat them as if they are substantial.The ‡ avor is. Mixture continuity means that preference between two lotteries is robust to su¢ ciently small changes ~ in their probabilities (in certain directions): if L is very similar to L. The utility value of such a degenerate lottery will be denoted by un U (Ln ). (Because their preferences over consequences are essentially lexicographic.g. but they are often sensitive to framing. much the same. Such violations sound plausible.
This statement is the expected utility theorem. then whichever state is realized. mixing lotteries does not change the nature of the consequences in any way. only their probabilities. pN ) 2 L.A more intuitive rationale can be given by thinking of as a randomization. It implies that the utility function is linear in character.g.3
Expected Utility Theorem
What the independence axiom gives us. outcome of a toin coss). in conjunction with mixture continuity. Suppose you prefer L to L0 and are o¤ered compound lotteries that result in (1) L with probability and L00 with probability 1 and (2) L0 with probability and L00 with probability 1 . It is primarily responsible for the stronger results we will derive in the present context. 88
. hence it could not be imposed on preferences over consumption bundles. It is important to understand that the independence axiom only makes sense in the absence of complementarities. We will get to the actual theorem in a few steps. While the commodities that make up a bundle are consumed together.
9. : : : .
De…nition. I.e. U (L) =
N X n=1
pn un :
A utility function with the expected utility form is called a von NeumannMorgenstern (vNM) utility function. The independence axiom is the fundamental di¤erence between the formalizations of lottery preferences and consumption preferences. If is the probability of some state of the world (e. the consequences of lotteries are mutually exclusive. : : : . goes beyond the existence of a continuous utility function that represents lottery preference. Utility function U : L ! R has the expected utility form if there exists (u1 . Hence you should prefer (1). uN ) 2 RN such that 8L = (p1 . you will be at least as happy with the outcome of compound lottery (1).
(If) Any lottery L = (p1 . 1] such that PK k=1 k = 1. If U ( ) represents a preference % on simple lotteries L and has the expected utility form. : : : . : : : . k=1 k pN U
K X k=1 k Lk
K)
!
=
N K X X n=1 k=1
k k pn
!
un =
K X k=1
k
N X n=1
pk un n
=
K X k=1
kU
(Lk )
(swapping the order of summation is permitted by distributivity). then it assigns to the corresponding reduced P PK PK k k lottery K the value k=1 k Lk = k=1 k p1 . : : : . (Only if) Consider the compound lottery that assigns probabilities ( 1 . : : : .2). : : : . Proof. then U (Ln ) = pn un = un if U ( ) has the expected utility form. If U ( ) has the linearity n=1 property. where Lk = pk . K. : : : . Exercise 53 (MWG 6. Proposition. then ! N N N X X X pn un . For if L = Ln (the degenerate lottery that assigns probability 1 to consequence n). LK 2 L and weights 1 . Hence U ( ) is linear. If U ( ) has 1 N the expected utility property. : : : . LN as L = N pn Ln . pN ) can be written in terms of P degenerate lotteries L1 . pk for k = 1. Utility function U : L ! R has the expected utility form if and only if it is linear: ! K K X X U = k U (Lk ) k Lk
k=1 k=1
for any K lotteries L1 .
89
. demonstrate that % satis…es the independence axiom. K 2 [0. to lotteries L1 . LK . : : : .B. pn U (Ln ) = U (L) = U p n Ln =
n=1 n=1 n=1
so that U ( ) has the expected utility form. : : : . uN ) 2 RN to be a utility function for the restriction of the lottery preference to degenerate lotteries. : : : .Notice that the de…nition requires (u1 .
Suppose L L0 . i. Consider the left panel of Figure 13. A utility function that has the expected utility form is therefore associated with linear indi¤erence curves.e. any convex combination of L and L0 is indi¤erent to L and L0 . The right panel of Figure 13 depicts how nonparallel indi¤erence curves cause a contradiction. the independence axiom essentially implies the expected utility form (the only other property needed is continuity).Figure 12: Indi¤erence curves with the expected utility form
Linearity of the utility function implies linear indi¤erence curves for lotteries in the simplex. the independence axiom says L + (1 ) L0 L0 + (1 ) L0 . According to the expected utility theorem. as in Figure 12.e. but not L00 = L + (1 ) L0 . and we construct lotteries L2 and L02 by respectively 90
. The connection is easy to see graphically. since nonlinear indi¤erence curves violate the independence axiom. If U ( ) is linear. Lotteries L1 and L01 yield utility U1 . so that 0 0 U (L) = U (L ) = U (L) + (1 ) U (L ). Such a contradiction arises whenever an indi¤erence curve is nonlinear. The curved indi¤erence set contains L and L0 . the independence axiom requires the indi¤erence curves to be parallel (as in Figure 11). since L L0 . L00 L0 . However. Furthermore. i. then U (L) + (1 ) U (L0 ) = U ( L + (1 ) L0 ).
then % admits a utility representation that has the expected utility form. But if the indi¤erence line through L2 is not parallel to that for L1 and L01 . then it cannot contain L02 . then > () L + (1 )L L + (1 ) L. For any L 2 L it is possible to …nd a unique L 2 [0. I. Since L1 L1 . 1). There exist lotteries L and L that are respectively preferred to all and none of the L 2 L. This is where the continuity axiom is needed. Here is the expected utility theorem. (We veri…ed it in constructing a utility function for a continuous rational preference. the and L2 = L1 + (1 ) L. 1] such that L L L + (1 L ) L. If . By the de…nition of L . I start with a few observations about % that are intuitive. L2 independence axiom says L1 + (1 )L L01 + (1 L02 .
.e. L2 = L1 +(1 0 0 0 ) L for some 2 (0. but should and can be proven formally from the axioms. 2 [0.Figure 13: Independence induces linear and parallel indi¤erence curves
)L mixing L1 and L01 identically with a lottery L. If rational preference % on L satis…es continuity and independence. i.e. L % L0 if and only if L
LL
+ (1
L) L
% 91
L0 L
+ (1
L0 ) L
L0 .) Now we will show that U (L) = L is a utility function for % and has the expected utility form. Proof. Proposition. 1].
(If jU (L) U (L0 )j > jU (L) U (L00 )j. : : : . A utility function that represents a preference over consumption bundles is unique up to strictly increasing transformation. Therefore. ! K K K X X X Lk = = k k Lk k U (Lk ) :
k=1 k=1 k=1
Exercise 54 (MWG 6. j U (L) U (L0 )j > j U (L) U (L00 )j. if and only if L L0 . K k=1 k = 1. because it is unique only up to positive linear transformation. i.
k=1 K X k=1
k
Lk L
+
k
Lk
!
L:
L
k=1
k
U
so by the construction of U ( ). so that U ( ) represents %. demonstrate that there are best and worst lotteries L and L in L. induction on the independence axiom gives Lk L+(1 L
K X k=1 k Lk K X k Lk L
+ (1 1
Lk ) L K X k=1
= Thus = PK
Lk .e. such that L % L % L for all L 2 L.e. Since Lk Lk ) L. u ~ u ~ In contrast. LK and all 1 .i.B. Hence it re‡ ects the ordinal nature of the preference: there is no signi…cance in the magnitude ju (x) u (x0 )j. which preserves the (relative) magnitude of di¤erences. the next result indicates that a vNM expected utility function is cardinal in nature.) 92
. If the set of outcomes C is …nite and rational preference % on L satis…es independence. If ju (x) u (x0 )j > ju (x) u (x00 )j. U k=1 k Lk = PK P 0 such that K k=1 k U (Lk ) for all L1 . PK The expected utility form is equivalent to linearity. : : : .3). but where j~ (x) u (x0 )j < j~ (x) u (x00 )j. then 8 > 0. we can …nd another utility function that represents the same preference. only in the fact that u (x) u (x0 ) or u (x) u (x0). if U (L) = L for all L 2 L. then 0 0 L % L () U (L) U (L ).
Thus L
LL +
+ (1 ~ U L U L
L) L
=
LU
~ L + (1 ~ U (L)
L) U
~ (L) = U (L) + . both are linear). which is a compact set. because U ( ) is linear. there are best and worst lotteries L and L in L such that L L.
U (L) U L
U (L) : U (L)
Rearrangement gives U (L) =
LU
L + (1
L) U
(L)
L) L
and. LK and weights 1 . so that the property holds). Linearity of U ~ (Only if) Suppose U ( ) and U ( ) both have the expected utility form (thus. K . ~ Because U ( ) represents % and is linear.) For every L 2 L. ~ Proof. (If) If U (L) has the P expected utility form and U (L) = U (L)+ . i. 93
U (L) :
. ! ! K K X X ~ ~ U (L) = U = U + k Lk k Lk =
k=1 K X k=1 kU
(Lk )
!
k=1
+
=
P where the last two equalities re‡ linearity of U ( ) and that K ect k=1 k = 1. it has a maximizer and minimizer on the probability simplex L. We will construct constants > 0 and such that ~ 8L 2 L.~ Proposition.e. and U ( ) has the expected utility form. : : : . (Since U ( ) is continuous. : : : . U (L) = U (L) + .
~ U (L) U (L)
> 0. then U : L ! R has the expected utility form if and only if it is a positive linear transformation ~ of U ( ). ~ ~ U (L) = U where
LL
. U (L) = U (L) + . de…ne
L
K X k=1
k
( U (Lk ) + ) . ~ ( ) implies the expected utility form. If utility functions U : L ! R and U : L ! R represent ~ preference % on L. K ~ then U (L) is linear: since L = k=1 k Lk for some set of lotteries L1 . 9 > 0 and 9 2 R such that 8L 2 L. Unless all lotteries are indi¤erent (in which ~ case U ( ) and U ( ) are constant functions. U (L) = U L L + (1 (1 L ) L.
it prefers (A) to (D). rather. (b) Compare the following policy criteria: (1) Evacuate in 90% of ‡ ooding instances.)
9. The agency is also indi¤erent between sure outcome (C) and a scenario where (A) occurs with probability q 2 (0.4). Derive probability distributions over the four outcomes under these criteria and decide. (a) Construct an expected utility function for the agency. while 94
. which criterion the agency prefers. and evacuate unnecessarily in 15% of instances where no ‡ ooding occurs. (2) Evacuate in 95% of ‡ ooding instances. The agency is indi¤erent between sure outcome (B) and a scenario where (A) occurs with probability p 2 (0. (Note that this version of the problem corrects two typos in MWG. Moreover. 1) and (D) occurs with probability 1 q. Suppose the expected utility theorem applies. (C) A necessary evacuation is performed. (D) A necessary evacuation is not performed. and none is performed. Consider four lotteries that o¤er the following probabilities over three prizes: $2:5 million $0:5 million $0 million L1 0 1 0 L01 0:10 0:89 0:01 : L2 0 0:11 0:89 L02 0:10 0 0:90 The decision maker is asked to compare L1 to L01 and L2 to L02 .4
Paradoxes
Expected utility (or. If is often observed (roughly half the time) that L1 L01 (giving up the chance to win a greater prize to avoid a small risk of zero). based on the expected utility function. A safety agency is looking for an evacuation criterion for an area that has a 1% probability of ‡ ooding. These predictions frequently fail in some famous experiments that have led to various alternative axiomatizations in order to explain the observed behavior. the underlying independence axiom) entails some speci…c predictions about choices. Four things can happen: (A) No evacuation is necessary.B. and evacuate unnecessarily in 10% of the instances where no ‡ ooding occurs. 1) and (D) occurs with probability 1 p.Exercise 55 (MWG 6. (B) An unnecessary evacuation is performed.
But L02 L2 was observed. if the continuity and betweenness axioms hold. (I. then U (L1 ) = u0:5 0:10u2:5 + 0:89u0:5 + 0:01u0 . Only the initial probabilities di¤er. (c) Argue (from a graphic comparison) that betweenness is weaker than independence.) Exercise 56 (MWG 6. Perhaps people worry about regrets they might have if a bad outcome occurs.B. stay at home. Explicitly. 0:11u0:5 + 0:89u0 0:10u2:5 + 0:90u0 .5). The alternatives are lotteries over outcomes: go to Venice. and therefore seems somehow worse than zero. Notice that in both cases L0 is obtainable from L by taking 0:11 out of the probability of winning $0:5 million and distributing 0:01 to the probability of $0 and 0:10 to the probability of $2:5 million.e. which implies. if L L0 . if L02 results in zero. (b) Depict in the simplex that. there was no way to guarantee a nonzero prize by choosing L2 . then L + (1 ) L0 L. These preferences are not necessarily irrational. i. Consider now the following experiment. if L01 results in zero. depict that straightline indi¤erence curves imply betweenness. Suppose there are three possible outcomes. 1). (d) Draw an indi¤erence map that satis…es betweenness and produces the choices from the Allais paradox. (a) Show that a preference relation on lotteries satis…es independence only if it satis…es betweenness. then indi¤erence curves of a lottery preference must be straight lines. watch a movie about Venice. but they are inconsistent with expected utility theory. it is viewed as a loss of $0:5 million that the agent would have had with L1 . L2 L02 .e. This is known as the Allais paradox. after adding 0:89 (u0 u0:5 ) on both sides.L02 L2 (accepting a slightly increased risk of zero for the prospect of a greater prize). The following property is known as the "betweenness axiom": 8L. L0 2 L and 8 2 (0. Conversely. if L1 L01 . but according to the independence axiom they should not a¤ect preference for the adjustment. 95
. In contrast.
All he has to do is choose the urn from which to take the ball. Since the agent wants to maximize expected winnings. but over the outcome "nature" chose (which is beyond the agent’ control). then R induces a lottery RB over outcomes $1. rather than between going to Venice and watching the movie. The decision maker will win a $1. s Perhaps the most in‡ uential critique of expected utility theory is based on the Ellsberg paradox. R and H. But then 1 > 0:49. Many people will always take the ball from R in successive experiments where …rst a black ball and then a white ball wins. The proportion of colors is known for R (there are 49 white and 51 black balls). This is known as Machina’ paradox. that each contain 100 balls of white or black color. either white or black) from one of the urns. ). the regret here is not over a choice the decision maker failed to make. It reminds us that preference need not be …xed. If a black ball wins. 000 and $0 with probabilities 0:51 and 0:49. If you want to think about it in terms of regrets. so that he must prefer HW to RW . in case you cannot go to Venice (but were hoping to).Even though you prefer the outcomes in this order.
96
. But it clearly violates the independence axiom. The problem is that there is no way to assign a probability (that a black ball is chosen from H) that could justify choosing R in both experiments. s but may depend on realizations of events. This would make sense if you expect that.e. but not for H. you will no longer enjoy the movie. Suppose there are two urns. he can prefer RB = (0:51. whereas H induces a lottery HB with unknown probabilities and 1 . 0:49) to Hb only if < 0:51. you may prefer to randomize between going to Venice and staying at home. 000 prize if he can pick a ball of a speci…ed color (i. R induces RW = (0:49. 0:51) and H induces HB = (1 . This failure of expected utility theory is fundamental: behavior appears to be at times inconsistent with the notion that individuals choose between known (or even estimated) probability distributions over outcomes. If a white ball wins.
a black ball wins $1000 dollars. then UW and UB are derived from a vNM utility function. however. so you prefer 97
.g.B.F.) s (a) Show that if P consists of a single belief. Suppose.e. These situations. rather than risk. 1] of such beliefs. UW (R) is the expected utility of $1000 given the (objective) probability implied by the number of white and black balls in urn R. In W .t. have ambiguous beliefs (allowing for multiple probability distributions). In the setting of the Ellsberg paradox. Hg ! R is such that UW (R) = 0:51 and UW (H) = min f1 s. expected utility theory can be extended to replace the objective probabilities in the de…nition of lotteries with subjective (implicitly believed) probabilities of events. picking the ball from urn R and H).g. that the decision maker has a set P [0. An important current research area in decision theory is therefore nonexpected utility theory. Hg ! R is such that UW (R) = 0:49 and UW (H) = min f s. 2 P g. But UW (H) is the expected utility of $1000 based on the most pessimistic probability in P . which contradicts the existence of unique probabilities altogether. you would like to know what kind of meat will be served.2). Sometimes. let the utility function over actions R and H be as follows. For each choice situation. when you have to bring wine to dinner. 1]. and UW (R) > UW (H) () UB (R) < UB (H). an agent’ preference over lotters ies depends on a prior action a 2 A (e.6).As we will see. 2 P g. Exercise 57 (MWG 6. But this does not address the Ellsberg paradox. Denote by W the choice situation where the $1000 prize is won if the ball is white and $0 otherwise. In B. and choices could re‡ optimistic or pessimistic expectations. In choice situation B. and the decision maker gets $0 otherwise. UW : fR. (This is an instance of Gilboa and Schmeidler’ theory of nonunique prior beliefs. are said to be characterized by (Knightian) uncertainty. where ect agents do not use simple probabilistic information or beliefs. Exercise 58 (MWG 6. I. A probabilistic belief that the color of the Hball is white can be expressed as 2 [0. let u (0) = 0 and u (1000) = 1 represent the decision maker’ preferences over s sure amounts of money.t. The available actions are R and H (respectively. UB : fR. where agents may e. (b) Find a set P such that UW (R) > UW (H) and UB (R) > UB (H).
: : : . pN ) 2 L.) Such random variables are closely related to lotteries via the probability distribution over states. The choice objects are now taken to be random variables x : S ! C. a function us : C ! R such that 8x = (x1 . Utility function U : X ! R has the extended expected utility form if there exists 8s 2 S. : : : . or act. where the probability s of state s 2 S is objectively known (we will relax this to subjective knowledge later). Such a preference has an induced utility representation U (L) = max
a2A N X n=1
pn un (a)
for all L = (p1 . can be represented by a function that has a modi…ed expected utility form. assigns consequences to events that are characterized by their probabilities. Both types of choice objects therefore induce a probability distribution over consequences. Show that U ( ) is convex.a degenerate lottery).
9. (These functions are also called acts. a random variable. Preference is de…ned on the space of random variables X and. if it satis…es continuity and a variant of independence called the surething principle.5
StateSpace Approaches
It is often possible and useful to impose a bit more structure where we have so far assumed given probabilities. While a lottery assigns probabilities to consequences. functions that determine which consequence will occur in each state. but (by example) not necessarily linear.e. xS ) 2 X . Let S be a set of states (decisionrelevant situations that may materialize). S X U (x) = s us (xs ) :
s=1
98
. where un (a) is the utility assigned to degenerate lottery Ln if action a 2 A is taken.
De…nition. i.
x is preferred to x0 when x E is replaced by x0 E . U (L) = N pn un .e. Denote by xE the restriction of random variable x on S to the event E S.
The axiom applies when consequences are identical for states outside the event E.P Comparing this to the expected utility form.) Say that x %E x0 if xE x0 E % x0 (i. And utility un of the lottery that is degenerate in consequence n is replaced by utility us (xs ) of consequence xs (which the act x associates with state s). and xE = xE . rare to have such information. (I. and by x E the restriction of x to the complement of E. so that the choice between x and x0 . (Or. the surething principle says that preference between x and x0 should be determined on states in which they disagree.) So if x % x0 .if what happens outside E cannot be changed. then it should be the case that x % x0 . it should not a¤ect preference. It is. (Hence the name "sure thing" . I introduce some new language. in alternative formulations. One can read x %E x0 as "x is preferred to x0 on E. S.) So far. xE assigns states s 2 E to consequences. we have taken for granted that there are objective probabilities for the states. To express the surething principle formally. The surething principle takes the place of the independence axiom in the present setting. of course." De…nition. and x0E = x0E . An extended expected utility function represents preferences on X that satisfy continuity and the surething principle (provided there are at least three states). respectively x and x0 .e. the n=1 probability pn of consequence n (that is determined by the choice of lottery L) is replaced by state probability s . Preference % on X satis…es the surething principle if 8E x %E x0 () x %E x0 whenever xE = xE and x0E = x0E . It is analogous in that it requires preference to be independent of events on which the acts do not di¤er. events that have zero probability of occurring. matters only if E occurs. so that x di¤ers from x0 only on E). 99
. In fact. Informally.
due to Savage. (The more direct approach would be to use density functions f ( ). so that they are completely determined. It implies that expected utility theory does not depend on factual knowledge of probabilities. for every state s and act x. utility function as objective probabilities. so that it has an extended expected utility representation P U (x) = S s=1 s us (xs ). with unique probabilities 1 . : : : . then F (x) = 1 f (t) dt. but these do not always exists and exclude the case of discrete Rx outcomes. is known as the subjective expected utility theorem. then it is arbitrary.e. If the PS 1 .nothing forces us to interpret the 1 .)
We now take the lottery space L to be the set of distribution functions on R. …x s so that we can interpret it as the subjective probability the agent assigns implicitly to state s? If we are willing to require that preference is stateuniform (depends only on the consequence chosen in state s. If f ( ) does exist. i.
10
10. S are to be interpreted as probabilities. : : : . then s=1 s = 1. S . If s is not given objectively. but ranks consequences the same in every state). Is there a compelling way to "disentangle" s from us (xs ). : : : . the surething principle and state uniformity admits a utility representation of P the form U (x) = S s=1 s u (xs ). This tell us that.
S
in the extended expected
Suppose preference % on X respects continuity and the surething principle. The result that a preference on X that satis…es continuity. 1]. the value s us (xs ) is uniquely determined up to positive linear transformation.1
Risk
Money Lotteries
A lottery over continuous amounts of money x 2 R can be described most generally in terms of its cumulative distribution function F : R ! [0. but can be built around personal beliefs that are probabilistic. The continuous version of the expected utility theorem 100
. then us (xs ) u (xs ) for all s 2 S. and s is determined through s u (xs ) up to scaling.
The criterion for risk aversion implies. and only these moments. An agent is riskaverse if she prefers the expected money value of a lottery to the lottery itself: 8F 2 L.e. continuous and bounded. and thus R u (x) dF (x) = u (x) f (x) dx whenever the density exists.)
The function u ( ). i. Suppose an individual’ Bernoulli utility function s 2 u ( ) has the quadratic form u (x) = x + x. that satis…es the independence axiom.2
Risk Attitude
A riskaverse agent is someone who rejects fair gambles (that have neither an expected gain nor loss). which records the values of degenerate lotteries. Show that utility from a distribution is determined by the mean and variance of the distribution. Z Z u xdF (x) u (x) dF (x) : 101
.C. Assume that u ( ) is increasing. is called Bernoulli utility function.) Exercise (MWG 6. (No need to do part b. certain amounts of money.e. so that dF (x) = f (x) dx. i. (If it were unbounded.guarantees that a continuous preference % on L. Petersburg paradox. small probability events could make a lottery in…nitely desirable . (Note that f (x)R dF (x) =dx.)
10.
De…nition. the above is an equality.the St. can be represented by Z U (F ) = u (x) dF (x) .) The agent is riskneutral if always indi¤erent between the expected value of a lottery and the lottery itself. when U ( ) has the expected utility form. Z u xdF (x) U (F ) : (Strictly riskaverse if this is an equality only if F ( ) is degenerate.2).
u)) = u (x) dF (x) :
It is intuitive that a riskaverse person has a certainty equivalent less than the expected value of the lottery. The individual in the left panel is riskaverse. she declines the lottery unless the odds are strictly favorable. while the corresponding utility value u (E (X)) is closer to the utility of the better state. and U (X) u (E (X)).) Hence the Bernoulli utility function of a riskaverse agent is concave. i. x2 g ! [0. (Observe how x1 and x2 along the xaxis are equidistant from E (X). the expected utility of a fair lottery. Since the agent has. u) of lottery F ( ) is its money value to an agent whose preference is represented by Bernoulli utility function u ( ): Z u (c (F. in utility terms. we have the contrasting case of a riskseeker. concavity of the (increasing) Bernoulli utility function
102
. whose Bernoulli utility is convex.e. more to lose than gain from a lottery that is fair in money terms.) On the right panel. the inequality relates to the basic de…nition of a concave function u ( x + (1 ) x0 ) u (x) + (1 ) u (x0 ). u (x2 ). is labeled U (X). i. and anyone with a concave Bernoulli function is riskaverse. In Figure 14. random variable X : fx1 . she only values the lottery the same if it produces on average a gain relative to the certainty equivalent. De…nition.This is Jensen’ inequality that characterizes a concave function u ( ). s (If you think of the integral over the probability distribution F ( ) as a weighted sum.e. This fact has a straightforward explanation: the riskaverse agent’ Bernoulli s utility increases more slowly with gain than it decreases with a loss. The Bernoulli utility of certain amount E (X) = (1=2) x1 + (1=2) x2 is labeled u (E (X)). Indeed. The certainty equivalent c (F. 1] with equal probabilities.
Beyond these formal characterizations.e. Example. u)) u Z c (F. riskaverse behavior is evident in the propensity to buy insurance. The agent is o¤ered insurance at a fair premium per dollarpayout in the event of a loss. even when premia are not "actuarially fair" (i. u) xdF (x) :
Z
Figure 15 shows the certainty equivalent for riskaverse and riskseeking agents.Figure 14: Bernoulli utility functions for riskaverter (left) and riskseeker (right)
implies: Z u (x) dF (x)
u
Z
xdF (x)
() ()
xdF (x) u (c (F. If dollars of insurance 103
. A strictly risk averse agent with initial wealth w faces a possible damage of $D with probability . the expected payout is less than the premium).
Expected utility from a choice of . a riskaverse agent insures fully if the premium is actuarially fair. The portfolio choice problem is to determine the optimal shares and to invest in these assets. R a risky asset with earnings distribution F (z) such that and expected return zdF (z) is greater than 1. such that + = 1. i. is + z. Example. u0 (w + (1 ) D) = u0 (w ).e. Suppose an amount of wealth can be invested in a safe asset. given and .
can be solved for w + (1 ) D=w . which induces a lottery over w + (1 ) D and w with probabilities ( . = D because u0 ( ) is strictly decreasing from strict concavity.Figure 15: Certainty equivalent for riskaverter (left) and riskseeker (right)
(i.1 ). which yields $1. is then U ( ) = u (w + (1 ) D) + (1 ) u (w ):
The …rstorder condition with respect to . Since the random return.e. conditional payout) are purchased at this premium. utility is a 104
. the agent’ wealth s will be either w (if no damage occurs) or w D+ = w+ (1 ) D (if there is damage). Thus.
For a riskneutral agent. u) = u00 (x) : u0 (x)
This is essentially a measure of the concave curvature of the Bernoulli utility function (for an increasing concave function. But such a transformation would also scale u0 (x). The ArrowPratt coe¢ cient of absolute risk aversion is. for a twice di¤erentiable Bernoulli utility function u ( ) at x. rA (x. we must have > 0. so it cannot a¤ect rA (x. Since the left side is greater than zero at = 0.C. An expected utility maximizer solves: Z max u (1 + (z 1)) dF (z) . so rA (x. Exercise (MWG 6. with respect to .
2[0. whether or not the individual is riskaverse.random variable u ( + z) = u (1 + z). Note that we cannot compare the degree of concavity based on u00 (x) alone. u0 (x) is constant and u00 (x) = 0. The general principle is that an agent will always invest some share of wealth in an actuarially favorable asset. De…nition. u) = 0. Risk aversion can be quanti…ed and compared by means of the absolute and relative coe¢ cients of risk aversion. 105
. given zdF (z) > 1 and that u ( ) is increasing everywhere. Z (z 1) u0 (1 + (z 1)) dF (z) = 0
(at anRinterior solution).19). u0 (x) > 0 and u00 (x) < 0).1]
which has …rstorder condition. since this derivative can be scaled by a positive linear transformation (which yields another utility function that represents the preference). u).
C. Bernoulli utility function u ( ) exhibits decreasing absolute risk aversion (DARA) if rA (x. I have constructed one particular utility function.e. for a twice di¤erentiable Bernoulli utility function u ( ) at x. i. u) whenever x < x0 .e. De…nition. Exercise (MWG 6.C. the utility function is exponential when the ArrowPratt coe¢ cient is constant. Moreover.8). Suppose rA (x. u) is a decreasing function of x. 8x. assuming the integration constants are zero. u1 ) if and only if u2 ( ) is a concave transformation of u1 ( ). but others are still exponential (solve the di¤erential equation to see this). Exercise (MWG 6. u) = x u00 (x) : u0 (x)
106
. u) > rA (x0 .The more riskaverse of two agents has a lower certainty equivalent for any given lottery F ( ). u2 ) rA (x. we have u0 (x) =u (x) = @ ln u (x) =@x = a.e. 2 is more riskaverse than 1 in the sense that rA (x. I. Integrating u00 (x) = au0 (x) on both sides. u2 (x) = (u1 (x)). and integrating once more on both sides. there exists an increasing concave function ( ) such that 8x. i.6. u (x) = e ax . rR (x. It is possible to recover the preference from the ArrowPratt coef…cient. rA (x.C. De…nition.7). Example. The coe¢ cient of relative risk aversion is. u) = a for all x. Exponential utility functions therefore constitute the CARA (constant absolute risk aversion) class. The DARA class of Bernoulli utility functions has the plausible property that wealthier people tend to be less riskaverse.
e. written F %F OSD G. Now we are interested in comparing lotteries and …nding criteria by which they can be ranked. then clearly rA (x. so DARA does not imply DRRA. where 2 (0. u) = xrA (x. regardless of risk attitude (i. Therefore.
107
. Distribution F ( ) …rstorder stochastically dominates G ( ). u) = is constant. Consider the Bernoulli utility function u (x) = x1 = (1 ). But rR (x. 1).
De…nition. u) = =x is decreasing in x. The converse is not true.C. Since rA (x. the speci…c form of the Bernoulli utility function). then F is said to secondorder stochastically dominate G.e.18).12).3
Stochastic Dominance
Up to now.C. If distribution F yields a higher expected utility than lottery G for a riskaverse agent (i. we have compared agents in terms of the risk aversion exhibited by their utility functions. if Z Z u (x) dF (x) u (x) dG (x) for any nondecreasing function u : R ! R. Example. a concave Bernoulli utility function).
10. this function is in the DARA class. such as risk attitude. nonincreasing relative risk reversion is a stronger property than decreasing absolute risk aversion. If distribution F yields a higher expected utility than lottery G. u) must be decreasing in x.If rR (x. Exercise (MWG 6. given properties of preference. F is said to …rstorder stochastically dominate G. u) = xrR (x. u) is decreasing in x. Exercise (MWG 6.
if F ( ) and G ( ) have the same expectation of x. i." The technique is based on the product rule applied to u (x) F (x): d u (x) F (x) = u0 (x) F (x) + u (x) f (x) dx implies u (x) f (x) = d u (x) F (x) dx Z u0 (x) F (x)
and.De…nition. and Z Z u (x) dF (x) u (x) dG (x) for any nondecreasing concave function u : R+ ! R. A related statement can be made for secondorder dominance. and any distribution function satis…es F ( 1) = 0 and F (1) = 1). I will give a precise characterization in a moment.e. Distribution F ( ) secondorder stochastically dominates G ( ). F %F OSD G is equivalent to the property: for any x. Rx Applied to u0 (x) 1 F (t) dt. G spreads the probability mass more evenly and gives more weight to the extremes. R R xdF (x) = xdG (x). F %SOSD G is equivalent to the property: for any x.e. (I. the product rule gives Z x Z x d 0 00 u (x) F (t) dt = u (x) F (t) dt + 2u0 (x) F (x) dx 1 1 108
. Z u (x) f (x) dx =
u0 (x) F (x) dx.
where is a constant (since du (x) F (x) =dx integrated on ( 1. integrating both sides. written F %SOSD G.) The next results make use of some integral relationships that may be found through "integration by parts. probability mass accumulates faster toward x under G than under F . getting more than x is more likely under F than under G. 1) is u (1) F (1) u ( 1) F ( 1).
F (x) G (x). Proof. From Rintegration by parts. which is nondecreasing. Thus Z x Z x 1 d 0 1 00 0 u (x) F (x) = u (x) F (t) dx u (x) F (t) dt 2 dx 2 1 1 and Z u (x) F (x) dx =
0
1 2
(where Hence Z
1 is a constant. we need to …nd a nondecreasing function u ( ) such R R that u (x) dF (x) < u (x) dG (x) at some x. since 2 u0 (1)
Z
u (x) R1
1
00
F (t) dt = 1 u0 (1)). (x) we have u (x) dF (x) = u0 (x) F (x) dx and u (x) dG (x) = R 0 u (x) G (x) dx. (Only if) Suppose F (x) > G (x) for some x.Rx Rx (since the derivative of 1 F (t) dt with respect to x is 1 f (t) dt + F (x) = 2F (x) by Leibniz’rule). To show that F fails to …rstorder dominate G. Hence F …rstorder dominates G. Consider u (x) = 1 for x > x and u (x) = 0 for x x. The payo¤ distribution F ( ) …rstorder stochastically dominates G ( ) if and only if 8x. Then Z Z 1 u (x) dF (x) = dF (x) = 1 F (x) x Z 1 Z < 1 G (x) = dG (x) = u (x) dF (x) :
x
109
.R (If) Let F (x) GR for all x. so Z Z Z Z 0 0 u (x) F (x) dx u (x) G (x) dx () u (x) dF (x) u (x) dG (x) : If u ( ) is an increasing function. given that F (x) G (x) for all x. 2
Z
x
F (t) dt dx:
1
u (x) dF (x) = =
Z
u0 (x) F (x) dx Z Z 1 00 + u (x) 2
x
F (t) dt dx:
1
Proposition. the …rst inequality holds.
x]. (If) Let 1 F (t) dt G (t) dt for all x. a contradiction. Let u (t) = t.2). Z x Z x G (t) dt F (t) dt:
1 1
Rx Rx Proof.1). R x x (Only if) Suppose 1 F (t) dt > 1 G (t) dt for some x. On this interval. Z Z Z Z x Z Z 00 00 u (t) dF (t) > u (t) dG (t) () u (x) F (t) dt dx > u (x)
1
x
G (t) dt dx:
1
Since u00 (x) = 0 for all x 2 [x. then 2 u00 (x) < 0. Exercise (MWG 6.
110
. we have u (x) dF (x) = +2 u00 (x) 1 F (t) dt dx and R R Rx u (x) dG (x) = +1 u00 (x) 1 G (t) dt dx.D. where = Rx Rx R F (t) dt > x G (t) dt. The payo¤ distribution F ( ) secondorder stochastically dominates G ( ) if and only if 8x. If u ( ) is concave. we need to …nd a concave function R R u ( ) such that u (t) dF (t) < u (t) dG (t) at x.D. From integration by parts. From integrating 1 R R Rx 1 by parts. so u (x) dF (x) < Rx u (x) dG (x). so Z x Z x Z Z F (t) dt G (t) dt () u (x) dF (x) u (x) dG (x) :
1 1
Hence F secondorder R dominates G.Exercise (MWG 6. x] containing x where 1 F (t) dt > 1 G (t) dt for all x0 2 [x. x]. Proposition. the left inequality cannot hold. x] and u00 (x) < 0 for all x 2 [x. let u (t) be strictly concave. This means G SOSD F . To show that F fails to secondorder dominate G. except R x0 R x0 in an interval [x.
1=4) with equal probability reduces to (1=6. 1=12) and (0. $1. 111
. $5) respectively with probabilities (1=3. 1=5) and (0. To obtain G from F . 1=6) = G. and $1 otherwise. Which goods are inputs and which are outputs is a matter of choice for every …rm within the technological constraints. 1=2) over ($1. 1=6) over ($0. or vectors. Feasible production plans are described by the production set Y . 0. respectively (1=6. in the commodity space. $1 and $5 wins. 1) in place of the $0. $1.e. $5) that plays (1=3.1
Pro…t Maximization Problem
Production Set
Firms exist in order to transform some goods (inputs) into other goods (outputs).3). I.e. $5). (4=5. $2). Observe that the expected values of these lotteries are $1 and $2. Example. These are simply bundles. 3=10). Consider two lotteries that reward outcomes of rolling a fair die. 7=12. F %SOSD G. 0). 0. Exercise (MWG 6. H s is the lottery (7=10. These lotteries have the same mean. 7=12. Example. where z is a zeromean random variable. F gives $1 if a number up to 3 is rolled and $2 if the number is greater than 3. Continuing in the previous scenario. and probabilities (1=2.F %SOSD G is also equivalent to the property that G is a "meanpreserving spread" of F .e. $1. 2=3. 3=4. This is a meanpreserving spread via lotteries (1. which lists all possible combinations of input and output quantities. 3=4. 1=4). 0.D. The compound lottery over ($0. i. replace the $1 and $2 wins in F with lotteries that give ($0. distribution H is called an "elementary increase in risk" from G if it redistributes all probability mass to the extreme points in G’ domain (while preserving the mean). G pays nothing for a 1 and $5 for a 6.4). 1=12) and (0.
11
11. I. 2=3. So we have constructed G as a meanpreserving spread of F . Exercise (MWG 6.D. if G is obtainable from F by replacing every certain outcome x with a lottery that yields x + z. 3=2. distributed according to Hx (z) with R zHx (z) = 0. 0.
Finally. (Free entry would imply additivity.) Proposition. (Note these are all versions of what is conventionally called "constant returns to scale. This can be done for any integer k. y 0 2 Y and . 0. and that there can be "no free lunch" (no output without input.5). > 0." they restrict the scale parameter to di¤erent intervals. 1 (nondecreasing) or 0 (constant). Additivity says that two feasible production plans can be combined (y.) With returns to scale. where y` < 0 identi…es the `th commodity as an input for the particular …rm. one often assumes some form of convexity. y 2 Y for 2 [0. The production set can therefore not be restricted to R+ . yL ) 2 RL ." it has the option of inaction (0 2 Y ). we have ky 2 Y . there are e¢ cient production plans). y 0 2 Y =) y + y 0 2 Y ). which can be broken down into additivity and returnstoscale properties. we have 8y 2 Y . By k times adding y. production vectors necessarily contain negative entries (every technology requires inputs). Suppose Y has the additivity and nonincreasing returns properties. If it is impossible to fully recover inputs from outputs. If the …rm can always use more inputs without reducing output. : : : . we have irreversibility (y 2 Y and y 6= 0 =) y 2 Y ). 1] (nonincreasing).e. If the …rm is able to shut down its operations without "sunk costs. Because =k < 112
. i. (If) Take any y. = Exercise (MWG 5. Proof.B. 8y.e. The production set Y is additive and exhibits nonincreasing returns to scale if and only if it is a convex cone. y 0 2 Y and 8 . so let k > max f .However. and y` > 0 means ` is an output. unlike consumption bundles. A typical element is (y1 . y + y 0 2 Y . g. It is commonly assumed that Y is nonempty and closed (i. free disposal applies (y 2 Y and y 0 y =) y 0 2 Y ). The following are typical properties that may fail to apply in special circumstances. y 2 Y and y 0 =) y = 0). The set of all such feasible vectors is the production set Y RL .
but this is not su¢ cient (for additivity). By the analogous construction. Hence Y is a convex cone only if Y has the additivity and nonincreasing returns properties. or that Y is strictly convex (y 6= y 0 and 2 113
. so that nonincreasing returns hold (let y 0 = 0). with = 0.
If Y is a convex cone.Figure 16: Typical convex cone production set (shaded)
1. which is a weaker property than convex cone. nonincreasing returns implies ( =k) (ky) = y 2 Y . 1] =) y + (1 ) y 0 2 Y ). y 0 2 Y . y 0 2 Y and 2 [0. (Only if) With = 1 and = 1. then Y is convex (y. it satis…es nonincreasing returns. The converse is true only if 0 2 Y (inactivity is possible). Now additivity gives y + y 0 2 Y . Similarly. We will typically assume that Y is convex. the convex cone is seen to satisfy additivity. so that Y is a convex cone.
add up). q2 . k. Exercise (MWG 5. the maximal output of q2 is q2 = (K k1 ) (L l1 )1 . l) < 0 unless k1 + k2 > K or l1 + l2 > L (which are infeasible). Exercise (MWG 5.B. You can recognize outputs and inputs by the fact that an increase in q1 and q2 increases F ( ). Suppose each of goods q1 and q2 can be made from K and 1 1 L. l) = k1 + k2 K= q1
1 l1 1=
k1 . with F (y) = 0 if and only if y is a boundary = point of Y . Given k1 and l1 . and an increase in a k or l decreases 114
.
) y 0 is in the interior of Y ).B.2).8). 1) =) y + (1 not strictly convex. The transformation function contains all relevant information about production possibilities. and else F (q1 . k. (Such a function could be de…ned for any set. k. Technologies are CobbDouglas: q1 = k1 l1 and q2 = k2 l2 .2
Transformation Function
The production set can be expressed in terms of a "transformation function" F : RL ! R that assigns values F (y) 0 to y 2 Y and F (y) > 0 to y 2 Y . (A convex cone is
11.) Exercise (MWG 5. q2 .)
Example. where k1 + k2 = K or l1 + l2 = L.(0. including the consumption set.3). which is however easy enough to express as X = R+ . q2 .C. A transforq2
1 l2 1=
+
k1
k2 :
Notice that F (q1 . The (production possibility) frontier is therefore described by !1= 1= q2 q1 + K= 1 l1 (L l1 )1 (rearrange the production functions for k1 and K mation function is: F (q1 . l) = 0 if and only if k = k1 + k2 = K and l = l1 + l2 = L.
When outputs are …xed. dF (y) = @F (y) @F (y) dy1 + dy2 = 0. L. this ratio is called the marginal rate of transformation (MRT). K. an increase in input means lower e¢ ciency. K. is fully e¢ cient. Example. Suppose F ( ) is di¤erentiable at a boundary point y. Then. where F (q1 . K. at …xed input levels. q2 . (When inputs are …xed.F ( ). If y1 and y2 are intputs. It measures the amount by which. F (q1 . L) = 0. q2 . q2 . where the transformation function was !1= 1= q1 q2 K. holding y` …xed for ` = 3. L) = 1 + l1 (L l1 )1 we …nd M RT21 and M RT SLK dL = dK L 1 115 q2 l1
1=
dq2 = dq1
(q1 =l1 )(1 (q2 = (L
)= )=
l1 ))(1
=
(k1 =l1 )1 ((K k1 ) = (L l1 ))1
=
1
L K
l1 : k1
. one input can be reduced when using more of the other. Then it measures the amount by which. L). In the CobbDouglas case above. one output has to be reduced in order to produce more of the other. : : : . an increase in output indicates greater e¢ ciency. at …xed output levels. @y1 @y2
and the slope of the transformation function is dy2 = dy1 @F (y) =@y1 : @F (y) =@y2
If y1 and y2 are outputs. the ratio is the marginal rate of technical substitution (MRTS).) The production plan (q1 .
Exercise (MWG 5. Finally. they may pursue other objectives than pro…t maximization.11. However. However. Else. F (y)
0:
Clearly. and …xed shares of pro…t accrue to the owners. This standpoint is not as immediately compelling as maximization with respect to consumer preferences. From a nonempty and closed production set. as costs). We do assume pricetaking behavior throughout. a riskaverse owner may favor less risky (but potentially less pro…table) production plans. the …rm must choose a production vector in the boundary of Y . which at strictly positive prices increases pro…t. if …rms are operated by agents instead of owners. higher pro…ts enlarge the owners’ budget sets and therefore increase indirect utilities.t. else it is possible to reduce inputs without reducing outputs. if …rms are owned by consumers. is p y = L p` y` (recall that inputs `=1 enter negatively. Pro…t maximization also requires that the technology is certain: else. an owner may …nd it optimal to manipulate prices by increasing the production of the goods she likes.3
Pro…t Maximization
The standard behavioral premise about …rms is that they maximize pro…ts. since the pro…t does not accrue to them. P Pro…t.G. the …rm chooses production plan y at p 0 that attains max p y = max p y
y2Y y2RL
s.1).
116
. pL ) 0 as independent of its production plan. Hence the constraint specializes to F (y) = 0. : : : . in the conventional sense. this argument relies on the "pricetaking" assumption that each …rm regards p = (p1 .
the optimal choice y (p) is implicit. a condition for pro…tmaximization is M RT SLK = and another is M RT21 = (k1 =l1 )1 ((K k1 ) = (L l1 ))
1
1
L K
l1 r = . it is the value function associated with the …rm’ maximization problem (analogous s to the indirect utility function and the expenditure function in demand theory).A solution to the …rm’ problem is the supply correspondence s y (p) = fy 2 Y s. @F (y ) = p` . Suppose the transformation function is di¤erentiable. Then the Lagrangean …rstorder conditions for pro…tmaximization are. such that expensive goods are produced using cheap goods. l. The …rm from the previous examples uses only inputs K and L. Thus. p y = (p)g :
We de…ne the pro…t function as (p) maxy2Y p y.t. k1 w
=
p1 : p2
(Additional …rstorder conditions place implicit restrictions on prices. @y` or in matrix notation. if the technological constraint is relaxed. : : : . It depends only on the parameter p. for ` = 1.) For any pair of commodities k.e. the condition implies pk @F (y ) =@yk = : @F (y ) =@yl pl
Example. is in the direction of prices.) 117
. L. rF (y ) = p. at prices r and w. (Since the gradient of the transformation function is proportional to p at a solution y . the pro…tmaximizing direction to adjust y. i.
1
increases in p1 . The supply correspondence y ( ) for a production set Y that is closed and satis…es free disposal is homogeneous of degree zero. We can express the outputs in terms of the levels of one input
1 q1 = k1 l1 = 1 l1 . it does not change the maximizers. it makes sense that q1 increases. Exercise (MWG 5. and singlevalued if Y is strictly convex. i.
. convex if Y is convex.C. Since a price increase from p to kp by k > 0 does not a¤ect the production set and can be factored out of p y. Proposition.C.
The scale of production is not determined. the supply correspondence y ( ). Proof.e. we obtain the optimal input ratios in the production of each output:
1
2
k1 p1 = l1 p2 K k1 = L l1 1
1=(1
)
1 w : r
w r
(1
)=(1
)
.Solving jointly. so p y = p y + (1 ) p y 0 = p ( y + (1 118 ) y0) .12).
or use the identity l1 = ( 2 L the total use of both inputs: q1 =
K) = (
1 2
2
to express outputs in terms of K) .
Exercise (MWG 5.
1 L) :
( 2L
1
q2 =
2
2 1
(K
Since in 1 . then p y = p y 0 (since all elements of the supply correspondence maximize pro…t at p). q2 = (K k1 ) (L l1 )1
=
1)
2
(L
l1 ) . The pro…t function ( ) is homogeneous of degree one and convex. Suppose Y is convex. If y 2 y (p) and y 0 2 y (p). Then it is clear that (p) = p y with y 2 y (p) is homogeneous of degree one. and q2 decreases.9). since the technology exhibits constant returns.
let 2 [0.i. ) p0 ) y (p) + (1 ) (p0 ) . p0 0.e. It arises from the convexity of the pro…t function. Since y ( ) is homogeneous of degree zero. p y (p) and p0 y (p0 ). then y (p) = r (p) and Dy (p) = D2 (p) is a symmetric and positive semide…nite matrix. L. so the …rm’ ability to adjust s supply can only further enhance pro…t. which = contradicts convexity of y (p). y 0 2 Y . which is an intuitive property: a price increase causes a linear increase in pro…t at …xed supply. Therefore. To see that ( ) is convex. 119
Proposition. then 8y. and must be singlevalued. 0 where y 6= y . Moreover. 1). we have for ` = 1. hence locally constant).
This is the `th entry in the vector Dy (p) p. if y 2 y (p) and y 2 y (p). pT Dy (p) p = 0. the convex combination y + (1 ) y 0 is in the interior of Y . i. ( p + (1 which means ) p0 ) = ( p + (1 ( ) is convex. Positive semide…niteness of D2 (p) is also immediate from the fact that ( ) is convex. 8v 2 RL . Hence y (p) cannot contain distinct y and y 0 . if Y is strictly convex. : : : . y + (1 ) y 0 2 y (p). then y + (1 ) y 2 y (p).
11. Then it is possible to reduce inputs and increase outputs in Y . 1] and p.4
Law of Supply
The law of supply says that more is produced of an output. Di¤erentiating on the left and on the right with respect to p. less used of an input. v T Dy (p) v 0 (equality if v = p). For any y 2 0 y ( p + (1 ) p ). That y (p) = r (p) follows immediately from the envelope theorem (y (p) is a maximizer at (p) = p y (p). 0 0 Therefore. by de…nition of ( ) as a maximum. whose price increases. which must lead to strictly higher pro…t than p ( y + (1 ) y 0 ).e. L L X @y` ( p) @ pk X @y` ( p) = pk = 0: @ pk @ @pk k=1 k=1
. Proof. y ( p) = y (p) for > 0. Therefore. If the supply function y ( ) for a production set Y that is closed and satis…es free disposal is di¤erentiable at p. and 8 2 (0.
While all e¢ cient production plans are boundary points of the production set Y . (Consider a natural resource that cannot be produced. Its nondi¤erential equivalent can be understood directly: (p p0 ) (y y 0 ) = p (y y 0 ) + p0 (y 0 y) 0
for all p. First. E¢ cient production is nonwasteful. This is true because p y p y 0 and p0 y 0 p0 y (by the optimality of the supply correspondence). but it may be possible to reduce waste in other commodities. Any feasible production plan that does not use the resource is a boundary point. we look at a basic notion of e¢ ciency that is de…ned from the properties of the production plan alone. Production plan y 2 Y is e¢ cient if there exists no y 0 2 Y such that y 6= y and y 0 y. that ownprice e¤ects @y` (p) =@p` are positive.
12
12. p0 and y 2 y (p). that the supply function can be recovered from the (maximal) proft function. in particular. s Positive semide…niteness of Dy (p) is an expression of the law of supply and implies. y 0 2 y (p0 ). Second.
De…nition. we discuss the stronger notion of cost minimization.)
120
. Both are essentially implied by pro…t maximization and survive aggregation. is known as Hotelling’ lemma.1
E¢ ciency of Aggregate Supply
E¢ cient Production
The focus of this lecture are the e¢ ciency properties of supply and aggregate supply.The statement y (p) = r (p). which incorporates prices. not all boundary points are e¢ cient.
We need to show that p 0 and that y is pro…tmaximizing at prices p. Consider a production plan (q. The partial converse corresponds to the second welfare theorem. Then one can …nd y 0 y with y` y` 0 su¢ ciently large that p y < p y. there exists y 0 y in Y and y 0 6= y. Example. there is some p 6= 0 such that p y 0 p y 00 for any y 0 2 Py and any y 00 2 Y . Then p y 00 > p y 0 for every y 0 in some (su¢ ciently close) neighborhood of y. f (z) = q. Proof.t. then Y must be disjoint from the set Py y 0 2 RL s. If y 2 Y is e¢ cient. Proof. z) = q f (z). we can construct prices (q. the construction of p such that p y 0 p y 00 is again violated. Suppose there is a feasible production plan y 00 2 Y such that p y 00 > p y.e.
This result becomes the …rst welfare theorem under aggregation. Since Py is convex (if y 0 y and y 00 y. 0 then p y > p y (given strictly positive prices). given which y is pro…tmaximizing on Y . It follows that p y p y 00 for any y 00 2 Y . i. i. so that y cannot be pro…tmaximizing. But this was ruled out by the construction of p. then there exists for every e¢ cient production plan y 2 Y a nonzero price vector p 0.e. y 0 + y
of production plans that contain more of each output and less of each input. Proposition. Production plan y 2 Y maximizes pro…t at p 0 only if it is e¢ cient. Because such a neighborhood contains y 0 y. then any convex combination strictly exceeds y). w) at which z is pro…tmaximizing. Represent Y by the transformation function F (q. If the production set Y is convex.Proposition. Then for every 121
. From the …rstorder conditions. 0 Suppose p` < 0 for some commodity `. z) that is e¢ cient on the singleoutput production set Y given by concave production function f ( ). there exists a hyperplane that separates Y and Py . Fix the output price p. say at p = 1. i.e. If y is not e¢ cient.
de…nes the cost function: c (w. q) is the lowest cost required to produce q at prices w. z) 2 Y .2
Cost Minimization
I focus on the singleoutput case. divide the …rstorder condition @F (q. q). f (z) q:
Notice that the optimal choice of inputs is implicit in the cost function. w` = @f (z ) : @z`
This determines the desired price vector at which the e¢ cient production plan (q. q f (z) with equality if and only if (q. z ) =@q = = p = 1: for ` = 1. = q > f (z) and 8 (q. z) 2 Y . where a quantity q is produced using input vector z 2 RL 1 . : : : . The production function can be viewed as a transformation function F (q. L 1. z ) =@z` = @f (z ) =@z` = w` by the …rstorder condition for the optimal output q.
12. and the technological contraint is expressed by a production function f (z).`. z) is in the boundary of Y .)
122
. The costminimizing choice of inputs. (The quali…er "conditional" refers to the fact that the factor demand depends on output. since 8 (q. The de…nition of production e¢ ciency makes no reference to prices. A solution to the cost minimization problem at di¤erent (w. its value c (w. z) is pro…tmaximizing. Denote by w 0 the vector of input prices. at a given output. q) min
z 0
w z s.t. A stronger criterion that selects among e¢ cient production plans is cost minimization. @F (q. z) = q f (z). q) is the (conditional) factor demand correspondence z (w.
given the cost function. the following …rstorder conditions (together with the production function) characterize z 2 z (w. = @c (q.e. i.
Since q must be the pro…tmaximizing output.The factor demands at any solution (q . the production function is concave). Hence z` > 0 at an optimum (and (@f (z ) =@z` ) = w` ) unless (@f (z ) =@z` ) < w` at z` = 0. and w1 =w2 is the slope of c (w. : : : ." In s fact. q)). the owner’ wealth and indirect utility s v (p. the use of input ` should be increased. If the production set is convex (i.e. 123
. p q w z. z2 6= 0. q) =@q.10).C. for all ` = 1. since M P1 =M P2 is the slope of f (z) = q.e. q). Hence pro…t maximization implies not just e¢ ciency. must hold. the …rstorder conditions imply M P1 =M P2 = w1 =w2 and.
with equality if z` > 0. p = @c (w. This picture is reminiscent of the expenditure minimization problem if you think of q (or the implied pro…t pq) as the …rm owner’ "utility. but also that factor demand z solves the cost minimization problem. The multiplier in the costminimization problem is the marginal value of relaxing the technological constraint. i. so there is a direct connection. If z1 . solve maxq (pq c (w. since pro…t at q . q) = c. q) = c). @f (z ) @z` w` w. at given prices and costs. L 1. z ) to the pro…t maximization problem must be costminimizing given q . In matrix notation. rf (z ) Exercise (MWG 5. Letting the `th commodity be the output. tangency of the "isoquant" (graph of f (z) = q) and the "isocost" (graph of c (w. i. is strictly decreasing in the cost w z. because ` adds value pM P` that is greater than its price w` . The direction of the inequality in the …rstorder condition for costminimization is then intuitive: whenever (@f (z ) =@z` ) > w` . w) are increasing in q. = p whenever cost is minimized at the pro…tmaximizing output level. the …rstorder condition for pro…t maximization.e. See Figure 17. Therefore. w) =@q.
the factor demand correspondence has properties of the Hicksian demand correspondence for commodity bundles (homogeneous of degree zero in w. q) is symmetric and negative semide…nite. q) inherits properties of the expenditure function (homogeneous of degree one and concave in input prices w. z (w. The (minimum) cost function is better behaved than the pro…t function under nondecreasing returns to scale (when it is optimal to produce in…nite or zero output). (The identity z (w. the cost function always exists. q) where di¤erentiable. Dw z (w. Consider a singleoutput technology Y . Moreover. whereas the pro…t function only exists if …rms are pricetakers. Show: if f ( ) is homogeneous of degree one (constant returns to scale). q) = rw c (w. q) is called Shepard’ lemma. i. 124
.Figure 17: Tangency solution to cost minimization
This parallel implies that c (w.3). then c ( ) and z ( ) are homogeneous of degree one in q. satis…es the law of demand). q) = rw c (w. that is is closed and characterized by production function f ( ). Moreover. so that marginal cost is nondecreasing in q.C. with free disposal.e. and nondecreasing in q). If f ( ) is concave (nonincreasing returns to scale). then c ( ) is a convex function of q.) s Exercise (MWG 5.
Exercise (MWG 5. The aggregate supply correspondence is ( y (p) )
y 2 RL s. J. would choose a vector in y (p). Denote …rm j’ pro…t funcs tion and supply correspondence by j (p) and yj (p). If p 0. Y J . Moreover.3
Aggregate Supply
Suppose now there are J …rms with production sets Y1 .e. j=1 j (p). : : : . YJ (each nonempty.D. 9y1 . i. and call
yj
for some y such that yj 2 yj (p) for j = 1.D.e. Proposition. It is helpful to establish the second part …rst: (p) = J j=1 j (p). (p) = PJ j=1 j (p). then any y 2 y (p) satis…es y=
J X j=1
+ YJ . J would choose in the individual production sets Y1 . PJ Let yj 2 Yj for j = 1. Then j=1 yj 2 Y Y1 + + YJ implies PJ PJ PJ (p) p (p) j=1 yj = j=1 (p yj ). i. : : : .4). : : : . P Proof. Exercise (MWG 5. : : : . On the other 125
. J.
12. yJ that are individually optimal for the …rms. The aggregated supply correspondence admits a representative …rm in the sense that such a …rm. : : : . : : : . an aggregate of plans that …rms j = 1.t. faced with the aggregated production set Y1 + + YJ .5). closed and with free disposal). : : : . Let y (p) denote the supply correspondence on Y1 + the associated pro…t function (p). yJ with yj 2 yj (p) and
J X j=1
yj = y
:
It includes every sum of production plans y1 .
I. yj 2 yj (p) for j = 1. Since p yj = j (p) for all j. o PJ For the …rst part. J. sums over j = 1. j = 1.e. PJ PJ P (p). J j=1 PJ and also A y (p). Because y is pro…tmaximizing. i.P hand. J. Let y 2 y (p) and y = j=1 yj for some yj 2 Yj .t. hence yj 2 yj (p). A y (p). (p p0 ) (yj (p) yj (p0 )) 0. I.1
Partial Competitive Equilibrium
Competitive Equilibrium
A private ownership economy is populated by consumers with preferences. It follows that p yj = j (p). J such that J yj = y. p (p) = J j=1 yj = j=1 j (p). aggregate supply must be costminimizing.
13
13. Now let yj 2 Yj . : : : . the law of supply inequality for …rm j. This is so because symmetry and positive semide…niteness are properties that are preserved under matrix addition. j = PJ P 1. p yj < j (p) for some j is only possible if p yk > k (p) for some k. for all j. and a hypothetical planner. : : : . then y (p) also does. we need y (p) A yj s. If every yj (p) satis…es the law of supply. y (p) A. Then j=1 PJ P PJ PJ p y=p yj = J (p yj ) (p). J to (p p0 ) (y (p) y (p0 )) 0. endowments and ownership stakes in …rms that have production 126
. In fact. which con‡ icts with the de…nition of k (p) as a maximum value function. i. : : : . : : : . (p) j j=1 j=1 j=1 n j=1 j (p). Recall that the "law of supply" (optimal production plans increase in commodities whose prices increase) applies under mild conditions (Y is closed and has free disposal). p j=1 yj = j=1 (p yj ) = j=1 A key implication of this result is that e¢ ciency aggregates: since every …rm produces a given output at minimal cost. who has the economy’ combined production possibilities available. : : : . Thus J yj 2 y (p). its aggregate version. if y 2 Y . there are yj 2 Yj for j = 1.e.e.e. s can do not better than to replicate individual choices.
: : : . xI . the consumer provides a fraction s of the inputs used by …rm j and receives a fraction ij of the outij puts). and the producer problem of choosing a plan y 2 Y from the production set Y . : : : . I quantity of commodity ` in the economy is ! ` i=1 ! `i . L. Let there by I consumers i = 1. yJ ) is a list of consumption vectors xi 2 Xi RL for all consumers i 2 I and production vectors yj 2 L Yj R for all …rms j 2 J. if for each `. : : : . : : : . j=1
An allocation describes.
De…nition. Consumer i’ preference over consumption bundles s xi 2 Xi RL is represented by the utility function ui ( ). : : : . L. i is endowed with a bundle ! i = (! `i . Given the production decisions.
I X i=1
x`i
!` +
J X j=1
y`j
(i. : : : . J …rms j = 1: : : : . J.e. given the economy’ endowments s ! 1 . Now we consider how prices are determined through the interaction of demand and supply. I.e. the total available quantity of P commodity ` in the economy is ! ` + J y`j . the consumption bundle each consumer ends up with and the production activity of each …rm. Consumer i holds a share ij in …rm j.
127
. ! Li ) 2 XPThe total endowed i. no more than the available quantity of each commodity is allocated for consumption). Aggregate demand and supply are functions of prices that we took to be exogenous. We have discussed the consumer problem of choosing a bundle x 2 X from the consumption set X. Initially.possibilities. An allocation is said to be feasible. Firm j implements the production plan yj 2 Yj RL that maximizes pro…t. which is a proportional right to the …rm’ net output (i. An allocation (x1 . in this context. y1 . : : : . and L goods ` = 1. ! L of commodities ` = 1.
P P (iii) for ` = 1. L. yJ ) and price vector p 2 RL such that consumers and …rms optimize. However. ij y`j
i=1 j=1
128
.
Notice that a proportional change in all prices.De…nition. from p to p with > 0. A competitive equilibrium is an allocation (x1 . since demand and supply are homogeneous of degree zero. we can arbitrarily assign a price 1 to some commodity (which is then called the numeraire) without changing the equilibrium allocation. in the aggregate these choices determine prices through the market clearing conditions.
I X i=1
(p xi )
=
I X i=1
I X i=1
(p ! i )
p xi
p !i
p
J X j=1
= p = 0. ui (x ) = maxxi 2Xi ui (xi ) s. Suppose markets clear under an allocation for all commodities except k. Therefore. : : : . p yj = maxyj 2Yj fp yj g. i=1 j=1 It is implicit in the de…nition of competitive equilibrium that consumers and …rms treat prices as independent of their choices. : : : . if consumers exhaust their budgets. y1 . I x`i = ! ` + J y`j .t. too.
I X i=1
J X j=1 ij yj
I J XX i=1 j=1 ij yj
(
ij
(p yj ))
xi
!i
!
!
(factoring out the price vector is possible by the distributivity of the dot product). Then. and markets clear: P yj . Now if markets for commodities ` 6= k clear. ! I J X X x`i ! `i = 0. i. This is apparent from adding up the budget constraints. p xi p ! i + J j=1 ij p (ii) 8j 2 J.e. : : : . the market for k must clear. (i) 8i 2 I. has no e¤ect on any of the three aspects of competitive equilibrium. xI .
xi ) = mi + i (xi ) . Clearly. Strictly speaking. enter consumer i’ utility function linearly. We will assume a quasilinear environment. It can be treated as the numeraire. and it is then not clear what it means to measure the price of a unit of ` in terms of what is given up of other goods. The main prerequisites for such a "partial equilibrium" analysis are that changes in the price of ` do not create wealth e¤ects and that changes in the quantity demanded of ` do not cause price adjustments in other markets.then pk
I X i=1
xki
! ki
J X j=1
ij ykj
!
= 0. we normalize its price to 1. bundled together. it would depend on how much of ` is consumed.
which (provided pk > 0) implies that the market for commodity k clears. then the marginal value of residual wealth would not be …xed. and i’ quasilinear preferences are s represented by ui (mi . the value of a dollar spent on ` would not be constant. and denote the quantity of this coms posite commodity by mi ("money" left over).
13. Suppose then that all goods other than `." absorbing whatever wealth the consumer does not spend on `. 129
. If they did. Let the quantity of good ` be xi . but note that it may generalize a bit beyond to situations where demand for ` is just not signi…cant enough to cause nonnegligible wealth e¤ects and price changes elsewhere. If money existed. but one may argue that it applies approximately whenever only a small portion of the consumer’ wealth is s spent on `. i. that is a quasilinear scenario.e.2
Partial Equilibrium
Often we are interested in the market for a particular good ` and would like to treat the remaining markets as "everything else. this argument is violated if ` has close substitutes or complements.
Firm j sets qj so that p qj attains max fp qj
qj 0
cj (qj )g . (Only part a.
which has …rstorder conditions p Consumer i solves
mi 2R. c0j qj . Exercise (MWG 10.G.C.e. 00 (xi ) < 0 at all xi i i (0) = 0 (a normalization).t. 0i (xi ) > 0.C.
=
mi 2R. I xi = J qj . mi + p xi (
! mi +
p qj
J X j=1
c j qj p qj c j qj ) . when there are no …xed costs.xi 2R+
max
fmi +
i
(xi )g
J X j=1 ij
s.G. Exercise (MWG 10.e. 0i (xi ) p (equality ifPi > 0). i. Let cj (qj ) be convex and twice di¤erentiable. The price of ` is p. Exercise (MWG 10.xi 2R+
max
i
(xi )
p xi + ! mi +
ij
since the budget constraint holds with equality at a solution (so that we can substitute for mi ). with equality if qj > 0. determine x P the equilibrium p together with market clearing.2). and an endowment ! mi > 0 of the numeraire. (Note that c (0) = 0 only makes sense in the i long run.5).2). and
Firms require cj (qj ) of numeraire commodity to produce qj 0 units of good `. c00 (xi ) 0 at all qj 0.where i ( ) is concave.6).) 130
.) Suppose there is no endowment of `. i=1 j=1 Note that none of the equilibrium conditions depend on endowments or ownership stakes.
0. c0i (xi ) > 0. This is a property of quasilinear preferences. The …rstorder conditions. Exercise (MWG 10. i.
x (c0 (0)) > 0. i.
131
. and constant returns to scale: c (q) = cq for some c > 0. but then supply is zero and demand is positive by assumption. market clearing fails.3
The Long Run
Let there be in…nitely many …rms that could potentially produce commodity ` with the same cost function. so that p = c. and markets clear: (i) p q c (q ) = maxq 0 fp q c (q)g. …rms will then enter as long as positive pro…t can be made. In this case. pro…table entry is not possible. so some …rms could not be optimizing. a …rm can attain zero pro…t by shutting down completely (c (0) = 0).e. given aggregate demand function x (p) and cost function c (q) for potentially active …rms (where c (0) = 0). else (if positive) more would enter or (if negative) more would exit. is a price p . all individual …rms must have zero pro…t in longrun equilibrium. x (c0 (0)) > 0. Hence maxq 0 (p c) q = 0. …rms make a positive pro…t at some level of production. if p > c0 (0). Since 0 x (c (0)) = x (c) > 0. Then we must have p c. so we may denote any individual …rm’ output as q (since it is unique if costs are s strictly convex). Therefore p c0 (0). a per…rm production level q . and a number of …rms J such that …rms optimize. Firms are identical. Therefore. and exit if only negative pro…t can be made. Given enough time. so total supply is in…nite. (ii) p q c (q ) = 0. Assume strictly positive demand at a price equal to marginal cost. (iii) x (p ) = J q . Assuming free entry and exit. market clearing (iii) requires q > 0. This leads to the following special version of competitive equilibrium. A longrun competitive equilibrium. and therefore J is indeterminate. else every …rm would want to produce in…nitely by (i). but q is indeterminate by (ii).13. There is no longrun equilibrium with increasing and strictly convex costs.
De…nition.
qJ ) such that 0 ui (xi ) ui (xi ) for all i = 1. I.F. Exercise (MWG 10. Shortrun and longrun supply function with …xed factor
14
14. This is a minimal welfare criterion: there is no distributional fairness implied (allocating everything to one person is Pareto e¢ cient). q1 . and J = x (c) =q. So we must have p = c. 132
. x) = mi +
i
(xi )
for i = 1.F. : : : . If p < c. I (and the inequality is strict for someone).6). : : : . xI .1
Welfare Analysis
Pareto E¢ ciency and Surplus
We now consider the welfare properties of competitive equilibrium. : : : . then no nonzero output level is pro…table (since c is the minimized average cost). qJ ) is Pareto e¢ cient with respect to preferences represented by utility functions u1 ( ) . ui (m.
De…nition. Tax impact in short run and long run Exercise (MWG 10. : : : . Consider a partial equilibrium setting where individual utility functions are quasiconcave. it makes sense to at least require that not everyone can be made better o¤.e. i. As in the previous lecture. q1 . x0I . : : : . However. : : : . A feasible allocation (x1 . uI ( ) 0 0 if there exists no other feasible allocation (x01 .3). If p > c.To get a determinate longrun number of …rms. Commodity ` is consumed by individual i in quantity xi and produced by …rm j in quantity qj at cost cj (qj ). Longrun and shortrun competitive equilibrium price Exercise (MWG 10. output is in…nite. which is a composite of all goods other than `.2). mi is i’ consumption of a s numeraire commodity. there must exists an e¢ cient scale q that minimizes longrun average cost c = c (q) =q.F. : : : .
: : : . in the quasilinear environment. qJ ). xI . xI . : : : . q1 . q1 .
i. yJ ) with ui ui (xi ) for i = 1. : : : . : : : . : : : . q1 . qJ ) :
I X i=1
i
(xi )
J X j=1
cj (qj )
This means.e. : : : . their levels are uniquely determined at optimal consumption and production choices) An increase in aggregate surplus (in the quasilinear setting) expands the utility possibility set and is therefore welfareimproving with respect to any other reasonable measure (that respects the Pareto principle). Altering production or consumption of good 1 leads to parallel shifts Optimal consumption and production levels for good ` are those where the boundary of the utility possibility set is as far out as possible. y1 . consider the utility possibility set: U= (u1 . Paretooptimal and surplusmaximizing is equivalent. Then Paretooptimal allocations can only di¤er in the distribution of the numeraire among consumers (given the i ( ) are strictly concave and cj ( ) strictly convex. : : : . : : : . xI . the set of all attainable "utility allocations" u1 . An allocation is Pareto e¢ cient if and only if it maximizes the aggregate surplus S (x1 . then ! m j=1 cj (qj ) is available to be spent on the numeraire good. : : : . If individual consumption of good Pand production plans are …xed at ` J (x1 .In the quasilinear setting. xI . the feasible utility constraint is
I X i=1
ui
!m +
= ! m + S (x1 . 133
. uI . : : : . Hence. : : : . there is a convenient welfare test. uI ) 2 RI s. the boundary of the utility possibility set is linear in the consumption of the numeraire (with slope 1). qJ ) =
I X i=1 i
(xi )
J X j=1
cj (qj ) :
To understand the connection. 9 feasible allocation (x1 . I . Since the total utility from consuming ` is PI i=1 i (xi ).t.
and c0j (qj ) = C 0 (q) for every j. De…ning the inverse demand function as x 1 ( ) such that x 1 (x (p)) = p. we get Z x S (x) = S0 + (P (s) C 0 (s)) ds. (S0 equals aggregate surplus when ` is not consumed.) i=1 dxi = dx = dq = Integrating over quantity consumed of `. are the same across all consumers and …rms (since they are both equal to the constant p).) It is clear from the expression that aggregate surplus is maximized at x such that P (x) = C 0 (x). they do not vary with P individual or …rm. and x 1 (x (p)) with the price P ( ) at which a particular quantity is demanded. Hence we can replace c0j ( ) with the "industry marginal cost" C 0 ( ). Then 0i (xi ) = P (x) for every i. and p = 0i (xi ) for all i at a solution.14. Apply feasibility q = x and PI J j=1 dqj in the last equality.2
E¢ ciency of Competitive Equilibrium
Recall that.
s=0
which is graphically the area between demand and supply curve below x. and the inverse supply function as q 1 ( ) such that q 1 (q (p)) = p. Consider a di¤erential change in aggregate surplus: dS =
I X i=1 0 i
(xi ) dxi
= P (x)
I X i=1
J X j=1 0
c0j (qj ) dqj
J X j=1
dxi
C (q)
dqj = (P (x)
C 0 (x)) dx
(Because P (x) and C 0 (x) are constants at a competitive equilibrium. the competitive equilibrium consumption level x where price equals marginal cost of `. Note that marginal cost and marginal utility. at a solution. we have q 1 (q (p)) = p = c0j (qj ).e. in this case the endowment of the numeraire. 134
. the standard depiction of aggregate surplus. i. i.e. p = c0j (qj ) for all j. in the competitive equilibrium problem.
e. respectively qj > 0.:::. I. which says that a competitive equilibrium allocation is e¢ cient. J. 135
. the marginal utility of another unit of ` equals its opportunity cost in terms of the numeraire. I.xI ) 0 (q1 .:::. since these are equalized across individuals in competitive equilibrium. Exercise (MWG 10.
I X i=1
xi =
I X i=1
qj :
Firstorder conditions are: for i = 1.4). we therefore have P (x) = C 0 (x) () 0i (xi ) = c0j (qj ). (All this presumes consumers and …rms are price takers.
14. I.) At the competitive equilibrium solution. it can be thought of as the solution to the problem ( ) I J X X max !m + cj (qj ) i (xi )
(x1 .D. c0j (qj ) (With equality if xi > 0.
0 i
(xi ) :
and for j = 1. This is an instance of the …rst welfare theorem.D.Note that aggregate surplus is the area between the inverse demand function P (x) = x 1 (x (p)) = p = 0i (xi ) and the inverse supply function C 0 (q) = q 1 (q (p)) = p = c0j (qj ). : : : .3
E¢ cient Allocations through the Market Mechanism
Since the competitive equilibrium is surplusmaximizing in the quasilinear case.) And the feasibility constraint. Exercise (MWG 10.) Thus.t. (In a wellde…ned sense.3).qJ ) 0 i=1 j=1
s. competitive equilibrium allocation is Paretooptimal.e. : : : . the inverse demand function indicates marginal utility of consumption. and the inverse supply function indicates marginal cost of production.
3). An externality is a welfare e¤ect one agent’ actions have on s another.4). 136
. I. think of many examples where other people’ actions directly impact one’ personal welfare (tra¢ c s s congestion. through …rstorder conditions. : : : .e. to name a few). 0 = c0j (qj ). These e¤ects are called externalities. the shadow price of the resource constraint for good ` is its price p. Then the problem jointly solves for a competitive equilibrium and Pareto e¢ cient allocation.1
Externalities
Externalities
Up until now. This the second welfare theorem. The second welfare theorem says that any Paretoe¢ cient allocation is a competitive equilibrium allocation P some speci…cation of zerosum for transfers. every possible utility distribution can be implemented and remains Pareto optimal. marginal utility from consumption equals mari (xi ) = ginal cost). i. we have assumed that individuals care only about their own consumption . One can. Because the amount of the numeraire commodity ! m can be allocated in any conceivable way among individuals without a¤ecting the solution (but it does. however. a¤ect individual utility).they are indi¤erent to the consumption choices of others or production choices of …rms.
15
15.C. between individual i=1 endowments of the numeraire commodity. Exercise (MWG 10. i.C.e. must have = p.To equate these to …rstorder conditions of the competitive equilibrium problem. except insofar as these a¤ect one’ s own budget set. T1 .
De…nition. price re‡ ects the marginal social value of the good (also implies. pollution. TI such that I Ti = 0. noise.e. of course. Exercise (MWG 10.
00 i ( ) < 0). h) = gi (x 1i (p. Since we wish to focus on the role of the externality. then the …rm e¤ectively operates an increasing returns technology and therefore has a convex pro…t function. 2 are a small part of the economy. h). then its pro…t function cannot be forever decreasing in h. The two concerned individuals i = 1. 1 imposes a (positive or negative) externality on 2. Or if a …rm’ constantreturnstoscale production function is s a¤ected by a positive externality (which acts like another input). their choices do not a¤ect prices p 2 RL . xLi ). note that concavity may well be violated in practice.To get a sense of the issue. Their wealths. h) = gi (x 1i . h)
(p. The associated utility function is vi (p. we have vi (p.e. h) denotes consumption of goods 2. wi . L). h) + wi where
i
p
1
x
1i
(p. h) = max ui (xi . making it convex somewhere. so that ui (xi . h) + wi i (p. h)
p
1
x
1i
(p. we can treat i ( . p xi
wi :
Let preferences be quasilinear with respect to a numeraire (good 1). h) . h) + x1i (where x 1i = (x22 . i. We should. are w1 and w2 . For example. wi . h) = gi (x
1i
(p. ) as a function of h only. Let i ( ) be strictly concave (i. h). Besides having preferences over their consumption of goods xi = (x1i . given these prices. xL2 . both individuals also care about an action h 2 R+ taken by 1. which are independent of wealth. h) .
Since commodity prices p are …xed throughout the discussion. however. and normalize to i (0) = 0. : : : . 137
. given that consumption of the L goods is optimized. : : : . consider initially a bilateral externality that is imposed by one person on another (as opposed to a multilateral externality that a¤ects many others). Evaluating at the optimal demands x 1i (p.t. Thus. h)
xi 0
s. it is convenient to de…ne a "derived" preference over levels of h. : : : .e. if a …rm is able to shut down operations when a su¢ ciently high negative externality is imposed on it.
2
Ine¢ ciency
Denote by h the equilibrium level of action h. Exercise (MWG 11. only if 0 1 (h ) =
0 1 0 2
( ) > 0 everywhere.1). 02 ( ) < 0 everywhere.
(Pareto e¢ ciency holds if and only if surplus is maximized).
with equality if h > 0. since (h) is decreasing in h and 01 (h ) < 01 (h ). The …rstorder condition that the socially optimal externality h satisfy is therefore 0 0 0. If h is a positive externality. If h is a negative externality. then h = h
0 2
(h ) < 0:
Again. in the quasilinear setting. must
However. Recall that. Suppose h > 0.e. i. so that 1 in fact imposes the externality on 2. then h = h only if
0 1
(h ) =
0 2
(h ) > 0:
Since 01 (h) was assumed to be decreasing in h (by concavity). and 0 0 < h (too much of the exter1 (h ) > 0 = 1 (h ). the socially optimal allocation is unambiguously the one that maximizes aggregate surplus S (h) =
1
(h) +
2
(h) + w1 + w2 . the externality is obviously ine¢ cient). which must be optimal for 1. we must have h nality is imposed). 138
. 1’ choice of h maximizes 1’ derived utility function. i. which s s has the KuhnTucker …rstorder condition
0 1
(h )
0. 1 (h ) + 2 (h ) with equality if h > 0. it follows that h > h (too little of the externality is imposed). and say that h > 0 (otherwise. and by h the (possibly di¤erent) socially optimal level.D.15.e.
D. w1 . Exercise (MWG 11.e. i. Such corrective taxes are called Pigouvian taxes.t. 1 has the right to choose any level of h. depending on whether the externality is negative or positive. h = = after substituting T =
1 2
h
1
s.3
Remedies
One way to eliminate the ine¢ ciency is by quota.
139
. 01 (h ) = 2 (h ). but can sign a contract with 2 to set h to a speci…c level in return for compensation.e.15. Another option is to tax or subsidize the externality. If 2 o¤ers a payment of T for the implementation of h. Since the tax subtracts from what agent 1 is able to spend on the numeraire commodity. it leaves derived utility v1 (p. 1 (h ) 1 h
1
(h )
h from the constraint. i. so that the government wants to induce h > 0). if t=
0 2
One can see that h = h
(p.2). then 2 maximizes derived utility v2 p. Say. The optimal tax 0 (subsidy) has to achieve h = h .
(h ) = t
(given that h
> 0. w2 . that is su¢ cient for 1. by mandating agent 1 to provide no more or no less than the socially e¢ cient level h .B. Bargaining can also resolve the externality problem.4). h) = which is maximized by h such that
0 1 1
(h)
th + w1 .
2
h
T + w2 h + T + w1 = 1 (h ) + w1 + w2 . Exercise (MWG 11. h ) :
The interpretation is that 1 is made to bear the cost (receives the bene…t) bestowed on 2.
the …rstorder condition is
0 2
h =
0 1
h . While it does not matter who has the right to externality.t.1). Alternatively. 1 (0) 1 h
2
=
which is identical to the case where 1 owns the right to the externality (since 1 (0) = 0) and has the same solution.
hence h = h . 1 h T + w1 = 1 (0) + w1 = 2 h + w2 . An externality is then simply a commodity. h h + T + w2 s." In the case of …rms. if 2 has the right to choose h.Because
0 1
(h ) = 0 (if h > 0).) If …rms must buy the right to impose a unit of the externality on someone at a price ph . j (p. which creates value by internalizing the externality. Related is the "missing markets" view of externalities. The optimal o¤er maximizes v2 p. then 2 can let 1 purchase the right to implement h at a price T . then …rm j demands hj units that maximize its pro…t ph hj . it is crucial that the right is allocated to someone in advance and tradable. : : : . Exercise (MWG 11. the transfer of rights may occur through a merger or acquisition. This arguments is known as the "Coase theorem. you can think of 1 as a …rm and 2 as a person.B. (To connect with the previous twoperson environment. hj ) = fj (hj ) 140
. hI of it. Instead of bargaining over the amount of the externality. We can generalize to a model where J …rms produce amounts h1 . a market for units of the externality would also …x the problem. : : : . hJ of the externality and I individuals experience amounts h1 . and ine¢ ciency arises when it is not traded (the market is missing). w2 .
(This setup re‡ ects the interpretation that the externality is negative.:::.hJ 0
I X i=1
hi =
J X j=1
hj
give fj0 p. respectively hi > 0. and of ph < 0. hj > 0.
i=1 j=1
(Note that transfers i=1 ph hi = objective due to market clearing.D.)
141
. s which requires hj as an input. a positive amount of the externality. (Assume f (0) = 0. This implies hj = hj .where fj (hj ) is j’ (concave) production function for the numeraire. wi . i.e. hi ) = Firstorder conditions imply fj0 (hj ) = ph =
0 i i
(hi ) + ph hi + wi :
(hi )
for …rms that produce.:::. Exercise (MWG 11. Socially optimal provision of the externality maximizes aggregate surplus I J X X S (h) = ( i (hi ) + wi ) + fj (hj ) . think of fj (hj ) < 0 as a convex cost function.) Firstorder conditions for max S (h) s. i. the market provision of the externality is Paretoe¢ cient.hI 0 h1 . The market price of the externality turns out to be the Pigouvian tax.e. …rms must be paid to produce the externality.) Consumer i supplies hi units of the externality. hj = =
0 i
(p. Conclusions are the same.
PI
PJ
j=1
ph hj drop out of the social
h1 . If it is positive.5). maximizing vi (p. and individuals who consume. hi )
for every i and j such that hi > 0 and hj > 0.t.
: : : . This is the Samuelson condition for the optimal provision of a public good. This is a reasonable assumption for some types of externalities (say.4
Public Goods
The market solution relied on the depletable nature of the externality: a …rm could directly sell a unit to a consumer without a¤ecting others. hJ of a public good at convex cost cj (hj ) in terms of the numeraire. as the case may be). i. (It says that the sum of marginal rates of substitution between a public and private good. Individuals i = 1. hI . The latter type of externality is called a public good (or public bad. They sell units of it to individuals at price ph . I demand quantities h1 . : : : . i=1 j=1 vi (p. air pollution). A public good is a nonrivalrous (i. : : : . hi ) = and
j i
Then
(h)
ph hi + wi . in this case the numeraire.15. but in fact P consume the entire amount of the public good P that …rms supply. Suppose …rms j = 1. J produce quantities h1 . J hj = I hi h.) 142
. i=1 j=1 (h ) = c0 (h )
(provided h > 0).
Surplus is maximized (Pareto e¢ ciency attained) only if
I X i=1 0 i
where c (h) = j=1 cj (hj ) is the total cost of providing amount h of P P the public good (payments I ph hi = J ph hj canceled out).
De…nition. should equal the marginal rate of transformation. : : : . cj (hj ) :
(hj ) = ph hj
The aggregate surplus is S (h) = PJ
I X i=1
(
i
(h) + wi )
c (h) .e. nondepletable) commodity: it can be consumed simultaneously by all agents. construction noise next to a single house).e. but not for others (for example. wi .
it follows that h < h . then 0i (hi ) = ph = c0 (h ) for some individual i and …rm j. then government can achieve e¢ ciency by imposing personal prices (p1 . for i = 1. pI ) on individuals for a unit of the public good: socalled Lindahl prices pi = 0i (p.
The classic (or pure) public good is. h ) per unit of the public good consumed. in a competitive equilibrium. however. and c0j (hj ) > 0 for all j. : : : . the …rsti=1 order conditions for individuals and …rms are
0 i
(h )
ph
0. Note that. This is an instance of the freerider problem: individuals do not pay for the bene…ts they receive from purchases by others. These prices are …rstorder conditions for consumers. Then individual i canP choose to pay for amount hi of the public good.
Because 0i ( ) is a decreasing function of h. which implies h = h . then I i (h ) = i=1 j=1 cj hj = Jc (h ) > c (h ). so that c0j hj = c0 (h ) when hj > 0. : : : . if every …rm produced the public good. the public good is underprovided. Given that 0i (h) > 0 for all i. marginal cost of production is equalized across …rms.e. Hence i=1 i (h ) = c (h ). If at least one unit of the public good is provided.. nonexcludable in addition to being nonrivalrous (i. A single …rm (or a consortium PI of J …rms) maximizes pro…t (h) = c (h) by setting i=1 pi h PI PI 0 0 0 i=1 pi = c (h ). ph
c0j hj
0
with equality if hi > 0. and c0 ( ) is an increasing function of h . 143
. this implies
I X i=1 0 i
(h ) > c0 (h )
For example. I. nondepletable). but consume h = I hi . so the price …rms receive for producing a unit of the public good understates the social value. If the good is sold in the market at price ph . respectively hj > 0.e. i.If the public good is excludable. hence all individuals optimize by demanding h units. and every individP PJ 0 0 0 0 ual consumed it.
Speci…cally. For example.e. A solution is a GrovesClarke (or pivotal) mechanism. the government could ask consumers how their wellbeing is a¤ected by di¤erent levels of pollution and ask …rms how bene…cial it is for them to be able to pollute. given the reports.or underprovide the costly public good. In general. if there is someone who has a higher marginal utility for the public good than everyone else. it faces the fundamental problem that it has no …rsthand information about the e¤ect of the public good on individuals. no one else pays for a unit of the good. Thus.B. It also determines the level of s 144
. A central issue is how to design a …nancing mechanism that elicits the correct information from individuals. The government can either provide a public good directly or use (minimum) quotas or subsidies to induce the e¢ cient amount. whenever government must intervene. the government implements the level of pollution that is socially optimal. then 01 (h) = ph for this individual entails 0i (h) < ph for all others.In fact. h )
k6=i
would cause consumers to take the total bene…t of their public good purchases into account. Exercise (MWG 11.5). In addition.D. which speci…es that every agent pays the "costs" he in‡ icts on others. Exercise (MWG 11. it pays consumers the bene…ts reported by …rms. a perunit subsidy of the form X 0 si = k (p. Based on the responses. No one has an incentive to misrepresent their true costs or bene…ts because one’ own report only a¤ects payments the other side makes s or receives (not one’ own liabilities). i.3). and collects damages from …rms equal to the loss in wellbeing reported by consumers. it may over.
We begin with a monopolist that faces di¤erentiable demand x ( ) and cost c ( ) for its product (and we also assume that pro…t is quasiconcave. pollution will be more restricted. but they will also have to pay higher damages. to perfectly substitutable products provided by a limited number of …rms.1
Monopoly and Product Di¤erentiation
Monopoly
For many industries. to …rms whose products have no close substitutes (monopolists). i.
16
16. If consumers exaggerate their loss in wellbeing. but neither …rms nor individuals have an interest to manipulate it. If …rms report excessive costs. but they they receive less compensation. The notion that many small …rms produce a particular good can be relaxed to varying extents. Finally.pollution the government sets. Then. to unique products that are. however. Because a GrovesClarke mechanism completely internalizes externalities . then more pollution will be allowed. …rstorder conditions identify a unique maximum). Most dramatically. the pricetaking assumption that is fundamental to competitive equilibrium is unrealistic.everyone bears the social cost of their actions (or reaps the social bene…ts) . The monopolist sets its price to maximize pro…t (p) = px (p) The …rstorder condition x (p ) + p @c (x (p )) @x (p ) @x (p ) = @p @x @p c (x (p)) :
can be restated in terms of the inverse demand function p (x (p )) as @p (x (p )) @c (x (p )) x (p ) + p = . imperfectly substitutable (di¤erentiation). @x @x 145
.e.it achieves e¢ ciency.
1).B.e. The reason is that the monopolist reduces sales from the socially optimal level where p = @c (x) =@x. Then. @p (x) 1 = : @x @x (p) =@p Alternatively. Exercise (MWG 12. This is a general optimization principle for the …rm.B. (A competitive …rm. Note on inverting the derivative @x (p ) =@p: it is generally the case that @x (p) @x 1 (x (p)) = 1: @p @x This can be seen by di¤erentiating the left and right side of x 1 (x (p)) = p with respect to p (using the chain rule on the left).) Exercise (MWG 12. one can come to the same conclusion by writing pro…t as p (x) x c (x) and di¤erentiating with respect to x. marginal revenue equals marginal cost. even in competitive environments. on the other hand. where its marginal impact on price is zero: @p (x) =@x = 0.i. @x
we see that the monopoly price strictly exceeds marginal cost. Rearranging the …rstorder condition to p = @c (x (p )) @x @p (x (p )) x (p ) .
146
. is too small relative to the market to a¤ect the price. and thus p = @c (x) =@x. if we denote the inverse demand function x 1 (x (p)) by p (x). in order to increase the price consumers are willing to pay per unit.6). provided @p (x (p)) =@x < 0.
and s the price is irrelevant.
a+c : 2b 2 You can see from the inverse demand function that the …rm would not produce unless a > c. p = p (x ) = 147
. Example. p (x) = a bx. Figure 18 illustrates. so it is not irrelevant to the monopolist. so p > c = p in the relevant circumstances.Figure 18: Surplus sharing under perfect competition (left) and monopoly (right)
The restraint in sales causes a "deadweight loss" to society. From society’ point of view. since the foregone units cost less to produce than people are willing to pay for them. the price determines how surplus from trade is divided between the …rm and consumers. Since . and marginal cost constant at c. only the allocation matters.e. x = a c 2bx = c. The optimal price and sales quantity satisfy marginal revenue equals marginal cost: @ p (x ) x = a @x i. Let inverse demand be linear. However.
If the monopolist were able to (perfectly) pricediscriminate. important practical obstacles to price discrimination.e. however.
.9). This is immediately apparent from the fact that the monopolist then captures the entire surplus and directly maximizes it. Exercise (MWG 12. i.B. Exercise (MWG 12. Each …rm i = 1. The deadweight loss is the area of the triange with height p p =p c and base length x x: 1 L = (x 2 Exercise (MWG 12. and no deadweight loss would occur. we develop Bertrand competition from its pure case. charge every consumer her valuation of each unit sold. the competitive output is x = x (c) = (a c) =b > x . from limited information to the possibility of resale by consumers who obtain the product at lower prices. Both extend straightforwardly to more players. There are. where …rms interact once with identical products. In this lecture.10). then it would be pro…tmaximizing to sell the socially optimal quantity.2
Bertrand Price Competition
The two workhorse oligopoly models are Bertrand and Cournot duopoly.B. 2 faces demand 8 0 if pi > pj < 1 x (pi ) if pi = pj xi (pi .B. x ) (p p )= (a 4b c)2 > 0:
16. pj ) = : 2 x (pi ) if pi < pj 148
individually. to repetition and to di¤erentiated products.5). Two …rms operate with constant marginal cost c and face total demand x (p) for their joint output.x (p) = (a p) =b.
i cannot gain by changs ing hers. p2 ) in which p1 = p2 = c. Since the payo¤s are in this context pro…ts c) xi (pi . The solution of interest is a Nash equilibrium. but typically not observed in reality.
149
. A Nash equilibrium speci…es the strategy chosen by each player. Given pj = c.The …rms simultaneously set prices. sj )
for all s0i 2 Si . pj ). If p0i < p0j . s2 ) such that. as an action conditional on the prior moves. given the other …rm’ price. Hence neither improves i’ payo¤. then increasing the price would strictly increase i’ pro…t. given j’ strategy. It follows that only p0i = p0j = c quali…es as a Nash s equilibrium. Let Si be the set of such actions or conditional actions available to player i. for i = 1. For our purposes. It is stable in the sense that. a Nash equilibrium is a price for each …rm i (pi ) = (pi that is pro…tmaximizing. 2.) The …nding that interaction of even just two …rms reverts to marginal cost pricing is striking. Next. De…nition. Now if p0i < c. and vice versa. A Nash equilibrium is a strategy pro…le s = (s1 . (Hence the Nash equilibrium is unique. This means pi = c for i = 1. On the other s 0 hand if pi > c. think of a player’ strategy s as a direct action or. si 2 Si and
i
(si . whereas pi > 0 loses all sales. Hence none of the prices can be larger than the other. p02 ). I will not give a formal de…nition of strategy here: in words. pi < 0 gains all sales. but makes a negative pro…t per sale. s Suppose there existed a distinct other Nash equilibrium (p01 . s Pure Bertrand duopoly has a unique Nash equilibrium (p1 . we consider two departures from pure Bertrand competition that restore higher prices and positive pro…ts. sj )
i
(s0i . 2 is a Nash equilibrium. if there is a sequence of moves. then i can strictly increase pro…t by raising the price (regardless of whether p0i c or p0i < c). it speci…es an action for every possible information state the player may …nd himself in. we must have p0i = p0j . decreasing the price would gain all sales and strictly increase i’ pro…t.
The s discount factor may be interpreted in various ways. (Only part a. They can now condition their pricing on the rival’ past actions and react aggressively s to price cuts. This circumstance greatly expands the …rms’options.)
16.
Exercise (MWG 12.4).3
Repetition
One factor the pure Bertrand model ignores is that most …rms know that they will face the joint pricesetting problem repeatedly over time.1).D. as the constant probability (at each point in time) that the …rms will compete again in the following period. Above marginal cost pricing may become viable because …rms maximize intertemporal. 1) is a discount factor.Exercise (MWG 12. Note that.C.g. Exercise (MWG 12. rather than immediate pro…t and will consider the e¤ect of setting a low price today on the rival’ behavior tomorrow.1).C. s Intertemporal pro…ts are 1 X t 1 it .
=
1 1
since (1 )
t 1
i
= =
1 X t=1 1 X t=1
t 1
i
1 X t=1 1 X t=2
t
i
t 1
i
t 1
i
=
i:
150
. e.
it
then the in…nite sum can be
i. if reduced to is a constant value
1 X t=1 1 X t=1 t 1 i i.
t=1
where it is i’ pro…t in period t and 2 (0.
because it switches from the best to the worst Nash equilibrium in the punishment phase.e. so it is optimal for 2 in period t if and only if (pm c) q m > 1 1 (pm c) qm . then pi (t) for i = 1.Cooperation on high prices (for example. p2 (1)) = (pm . say because the rival makes a mistake. the strategies prescribe (p1 (1) . More pragmatic trigger strategies may maintain low prices for a su¢ ciently long time to deter the rival in the future. It is also a "grim" strategy in that it never forgives the rival for setting a low price. the monopoly price pm ) might be sustainable in Nash equilibrium if both …rms play the trigger strategy pi (t) = pm if t = 1 or pj ( ) = pm for c otherwise = 1. p2 (t (pm . at any given time. In theory. In the …rst period. Otherwise. Undercutting triggers p1 ( ) = c for = t + 1. This is why an unforgiving trigger strategy such as the one above can be optimal: the threat is never tested. so that j can make no more than zero pro…t. if monopoly prices can be maintained forever in equilibrium. I. there is not too much discounting of future income). i "punishes" j by pricing at marginal cost. Such a strategy is called Nash reversion. If 1=2 (i. The reasoning is inductive.e. instead of (pm c) (q m =2). marginal cost pricing hurts the …rm in the present. but can be understood as an investment in reputation that permits pro…table cooperation in the future. t 1
for i = 1. Now suppose (p1 (t 1) . 2 is a Nash equilibrium of an in…nitely repeated Bertrand game. 2. pm ). : : : . In case the punishment is ever triggered. punishment will never actually occur . 2
1)) =
151
. 1 (so that the maximal pro…t 2 can receive in future periods is zero). : : : . The best alternative for 2 to p2 (t) = pm is to slightly undercut and earn a pro…t of (pm c) q m .all that matters is the threat. …rm i sets a high price pm only if j has set pm at all previous times. pm ) and p1 (t) = pm .
it is optimal for 2 to set p2 (t) = pm if 1=2. pm ) is a Nash equilibrium of the in…nitely repeated game. it is intuitively clear that no other kind of coordination can work either. either p1 (t) = c or p2 (t) = c at all times t. so that threats are arbitrarily damaging in the long run). which says that equilibria leading to any payo¤s that exceed the best each player can achieve independently (the minmax.e. Collusion is often not maintained as smoothly as suggested here.e. Here. possibly asymmetric. i. 1=2 admits many other cooperative equilibria besides the pro…tmaximizing one. p2 (1)) = (pm . even if it is
152
. and it may not be obvious to other parties whether a …rm is ceasing to cooperate. p2 (t)) = (pm . pm ) cannot be sustained. the minmax pro…ts are zero. I. Therefore. just replace pm above with any other prices. Which equilibrium occurs often depends on history and focal points. Because the inequality holds only when < 1=2. Exercise (MWG 12. and consider very close to 1. if the pro…t from capturing all sales today exceeds the discounted value of future pro…t streams when sharing the sales. This is an instance of the Folk Theorem from game theory. the best attainable in the worstcase scenario) exist for some su¢ ciently large .i. it becomes necessary to punish deviations.4). Hence (p1 (1) . demands and costs ‡ uctuate. Since the …rms share the market at = 1. However. and continue to do so whenever they have shared it previously. In practice. any combination of pro…ts exceeding zero is possible (to see this.e. but no more than that. or simply reacting to environmental changes.D. Since < 1=2 implies that the optimal coordination on (p1 (t) . since every player can always guarantee zero by setting price to marginal cost. To prevent …rms from taking advantage of this ambiguity. induction implies that they share the market in all periods.
but some market power). In purchasing either 0 or 1. the individual incurs a "travel cost" zt or (1 z) t. they choose their prices to maximize c) xj (pi . There are two main approaches to product di¤erentation: in one. in the other. which indicates lost satisfaction from consuming a nonideal product.4
Product Di¤erentation
When products are di¤erentiated. Every individual occupies at a point z 2 [0. …rm i’ demand is not perfectly prices elastic. there is an explicit preference for variety. Someone who prefers z is therefore willing to pay at most P0 (z) = v 153 zt
. pj ) of …rm i’ price. …rms retain some positive demand when pricing above marginal cost. between periods of high prices. It may instead be a continuously decreasing function xi (pi . a consumer demands only one of the products. the DixitStiglitz model. products are used together. 1] of possible products.D. s If …rms face constant marginal cost c > 0.5). Monopolistic competition is the special case of the DixitStiglitz model where variety is in…nite (there is a continuum of di¤erentiated products with small market shares. Hence pricing above marginal cost will be the pro…tmaximizing strategy. from consumers who value their unique products.not clear why they happened. Consider duopolists that serve a continuum of consumers whose demands arise in the following manner. 1] that re‡ ects her most preferred product.
16. I focus on the Hotelling model. the Hotelling model. pj ) : i (pi ) = (pi Typically. The …rms are associated with points 0 and 1 in a spectrum [0. and this leads to intermittent price wars. given the price pj of the competitor. Exercise (MWG 12. even if it means getting undercut.
z and P0 (z) p0 P1 (z) z p1 . If the …rms charge p0 and p1 for a unit of their products. i. Suppose z is uniformly distributed on [0. where z0 min v t p0 . this person will consumer product 1 if z 1
and the second inequality is reversed. z0 ]. then the person who prefers z will consume product 0 if P0 (z) p0 . one …rm makes no sales.e. s Individual demand is for either one or zero units.z : ^ t (Assuming. belongs to the "marginal ^ consumer. so that every possible product is preferred by the same mass of consumers. Whenever this is violated. The point z .e. ^
and total demand for product 1 is the length of the interval [z1 . and at most P1 (z) = v (1 z) t
for product 1 (v is the good’ undiscounted value). 1]. 1 p1 p0 + 2 2t v t p1 z: ^ v t p0
On the other hand. 1]. i. where P0 (z) p0 = P1 (z) p. that prices are such that z0 . i. of course.e.) 154
. which cannot be pro…tmaximizing unless the other …rm prices below marginal cost. z1 2 (0." who is just indi¤erent between the products.z . 1). hence no one consumes both products. where v p1 z1 max 1 . 1 z1 . Then the total demand for product 0 is the length of the interval [0.for product 0. and some might consume neither.
p1 ) = z . Above marginal cost pricing is not merely owed to the fact that few …rms compete in the market. let v be su¢ ciently large that the market is "covered" in equilibrium: every consumer purchases one of the products.For simplicity. it may actually increase its price further. individual demands other than zeroone. prices equal marginal cost. and increasingly so the more consumers discount for di¤erences from their most preferred products. ^
1
(p0 . p1 + c + t .e.C. Circular city with quadratic cost 155
. arbitrary consumer taste distributions. is a Nash equilibrium: p0 (p1 ) = c + t = p1 (p0 ) : I. as well as endogenous …rm locations (strategic product positioning).17). Then the marginal consumer determines the …rm’ demand functions s x0 (p0 . Once the incumbent …rm loses these pricesensitive consumers.16). and pro…ts are zero. p0 (p1 ) = 2 p0 + c + t p1 (p0 ) = . s where prices are mutual best responses. entry may not lower prices because it tends to attract consumers on the margin (who have the lowest willingness to pay for the product). Linear city with di¤erent costs Exercise (MWG 12. p1 ) = (p1
c) (1
z) ^
are maximized when the …rstorder conditions with respect to p0 and p1 . 2 are met. product di¤erentiation allows the …rms to price above marginal cost in Bertrand equilibrium. In fact. These are the best responses to the other …rm’ price. p1 ) = 1 ^ Pro…ts
0
z: ^
(p0 .C. p1 ) = (p0
c) z . A joint solution. Exercise (MWG 12. x1 (p0 . In the limiting case. asymmetric costs. This simple case can be extended to multiple …rms. where consumers do not care about such di¤erences and thus t = 0.
and subsequently engage in price competition. We will impose that consumers with the highest willingness to
156
. strictly decreasing and concave function x (p).e. This tends to soften pricing. At the latter point. so pi determines how much is demanded. rather than price.1
Capacity Constraints
CapacityConstrained Pricing
An assumption that was implicit in Bertrand oligopoly. and the higherpriced alternative may still be demanded by latecomers. so pj determines how much is demanded. which sells at capacity. Which of the two prices e¤ectively constrains demand depends on the capacities. This is determined through a rationing s rule. because a …rm can make a pro…t despite being undercut. and demand for the total output is a continuous. since all sales would be lost to competitors. competition in the sense that the …rm’ fundamental s choices are their production capacities. Capacity constraints lead to the Cournot model of oligopoly. For simplicity. Then. no …rm can a¤ord to be undercut. which can be viewed as quantity. then …rm i serves everyone. 2 simultaneously choose capacities q1 and q2 (at a positive perunit cost).17
17. With limited production capacities. the lowerpriced of two identical products is sold to early buyers. If qi < x (pi ). We must be speci…c about who buys from i at the lower price when demand exceeds i’ capacity. These circumstances lead to very aggressive pricing. consider again a duopoly in which …rms i = 1. If pi < pj and qi x (pi ). and turns out to be crucial for its marginal cost pricing equilibrium. Both capacities are known to both …rms at the time when prices are set. Denote the inverse demand by p ( ). capacity is free or can be increased instantly). We assume that the products are identical. is that …rms can serve arbitrarily large demands (i. then …rm j serves those who cannot buy from …rm i. each …rm is able to sell up to its capacity at a constant marginal cost c 0.
we simply have a monopoly. p2 ) = : min fqi . pi < p (q1 + q2 ) simply lowers the price. Neither of i’ deviations from pi = p (q1 + q2 ) is s pro…table. At least one …rm has spare capacity and will want to slightly lower its price (i. 0gg if pi > pj
. in which case market demand would exceed both capacities.This rationing rule where "highest valuations served …rst" is known in industrial organization as the "e¢ cient rationing rule. (Hence.) Denoting by b (qj ) the optimal quantity …rm i would sell if it were not capacityconstrained (and …rm j sold qj ). 2 x (pi ) if pi = pj : xi (p1 . (If one …rm does not invest in capacity. undercut like a pure Bertrand competitor) as long as p1 = p2 > c (otherwise.e. the proportional rationing rule.) s s We begin by analyzing the pricing game for some given capacities q1 > 0 and q2 > 0. Suppose therefore that pj = p (q1 + q2 ). then market demand does not cover both capacities. j’ demand. both …rms must sell a nonzero quantity in equilibrium: else they could avoid losses by not investing in capacity. But then i could slightly increase its price and still sell at capacity. so that …rm j’ customers have s the same expected willingness to pay as …rm i’ This would increase s. giving it a strictly higher pro…t. 2. where anyone is equally likely to be at the head of the queue. Similarly. let q1 b (q2 ) and q2 b (q1 ). and decrease i’ demand. This means prices must be equal." (An alternative is. max x (pi ) qj . for i = 1. Namely. for example. It remains to be shown that this is a Nash equilibrium given q1 and q2 . but cannot 157
pay are at the head of the queue. if p1 = p2 > p (q1 + q2 ). if pi < pj . Then demand is. x (pi )g 1 min qi . Otherwise. at any given prices. if p1 = p2 c. The only alternative is p1 = p2 = p (q1 + q2 ). the capacity constraints bind. it is not possible that p1 = p2 < p (q1 + q2 ). it is not optimal to invest in capacity). On the other hand.) Since capacity is costly to build. 8 if pi < pj < min fqi . max fx (pi ) qj . then …rm i would sell at capacity (given that consumers …rst go to the lowerpriced seller. so that each …rm could pro…tably increase its price. and …rm j still manages to sell something).
and qj = 0 contradicts p (0) > c.increase sales. which takes as given that each …rm will produce up to capacity and set the marketclearing price p (q1 + q2 ). Then they are functions qi = p (qi + qj ) c p0 (qi + qj ) c 0
that implicitly determine the best quantity choice for …rm i.e. Then p (qi + qj ) = p (qj ) c. And pi > p (q1 + q2 ) is undesirable by the assumption b (qj ) qi . Duopolist i = 1. i. Under the assumptions. 2 solves max (p (qi + qj )
qi 0
c) qi . Suppose qi = 0. p qj < c (since p0 ( ) < 0 everywhere). since i then sells below capacity. there is no equilibrium if qi > b (qj ) and p (q1 + q2 ) > c. qj > 0 would imply p0 qj qi + p qj = c at qi = qi . We conclude that p1 = p2 = p (q1 + q2 ) is the unique Nash equilibrium of the pricing game with capacities qi 2 (0. (Incidentally. which implies that …rm i’ s pro…t increases in sales when j is at. we must have qi > 0 for i = 1. so that the …rstorder conditions apply with equality.) This insight motivates the Cournot model of oligopoly. s 158
. and i is below. The Cournot model focuses on the capacityor quantitysetting game preceding price formation. Again. but at symmetric prices it would always be pro…table to slightly undercut. given …rm j’ quantity choice. capacity. On the other hand.
which has …rstorder condition p0 (qi + qj ) qi + p (qi + qj ) (equality if qi > 0). di¤erentiable function of qi + qj with p (0) > c. b (qj )]. 2.2
Cournot Quantity Competition
Let the inverse demand p ( ) be a decreasing. since …rm i is already at capacity.
17. Hence i should lower price while pi p (q1 + q2 ). this contradicts p (0) > c.
then …rm i can increase its production to qi = q m qj . Thus q1 + q2 > q m . namely p (q) = c p0 (q) q. Cournot oligopoly does not lead to competitive (marginal cost) pricing. But combined pro…t cannot decrease. where q m is the optimal monopoly output. the Cournot equilibrium price satis…es c < p (q1 + q2 ) < p (q m ) . From the Nash equilibrium condition above. at the monopoly price and monopoly pro…t. Because the increase s in qi lowers the price. where q = q1 + q2 . this will be the case if q1 + q2 > q m . since the monopoly pro…t is an upper bound on industry pro…t. Since p0 ( ) < 0. Proof. So we show p (q1 + q2 ) < p (q m ). Hence i’ pro…t strictly increases. You can check that it is not so. Given identical. This s violates Nash equilibrium. The reason is that. constant marginal costs c > 0. and the …rstorder condition for Cournot equilibrium should be identical to the …rstorder condition for a monopoly equilibrium at q1 + q2 = q m . this must strictly reduce j’ pro…t. given capacity 159
. thus: p qi + qj = c = c p 0 qi + qj p 0 qi + qj qi + q j 2 qi :
Proposition. then the industry acts like a monopolist. If q1 + q2 < q m . and inverse demand for the industry’ combined output such that p (0) > 0 and s 0 p (q1 + q2 ) < 0 whenever q1 + q2 0. Unlike pure Bertrand oligopoly. hence we must have q1 + q2 q m . in which case the industry supplies the monopoly quantity. it is immediate that p (q1 + q2 ) > c. qj jointly solves these bestresponse functions. If q1 + q2 = q m .
Note that q1 = b (q2 ) and q2 = b (q1 ) in Cournot equilibrium: the quantity choices can be interpreted as binding capacity choices that lead to p (q1 + q2 ) as a price equilibrium in a Bertrand game with capacity constraints.A Nash equilibrium qi . while qj remains …xed.
they sell "too much. each …rm only takes into account how an increase in sales. set the monopoly price. we obtain the Nash equilibrium q2 (q1 ) = q1 (q2 ) = a 3b c = q2 (q1 ) . i. a monopolist marks up by p0 (q ) q . a¤ects its own revenue. 160
. then pro…ts are (a bqi bqj c) qi . neither …rm can capture the entire market by undercutting. the e¤ect on industry revenue.constraints. Since each …rm’ demand is less than perfectly elastic. By contrast. In determining its capacity. It ignores the e¤ect on the rival’ revenue. i. and therefore a lower marketclearing price.e. i. this means p (q1 + q2 ) > c. with q = qi + qj .
a
with marketclearing price 1 2 p (q1 + q2 ) = a + c: 3 3 Since a > c. If inverse demand is linear. resulting in …rstorder conditions (bestresponse functions) q1 (q2 ) = and a bq2 2b c
bq1 c : 2b (In this context. Example. The monopoly quantity is …rm i’ s m m best response to qj = 0. Yet the …rms are unable to maximize industry pro…t.e. the effect of the price reduction on own revenue. Hence q = qi (0) = (a c) = (2b) and p (q ) = a=2 + c=2 > p (q1 + q2 ). bestresponse functions are often referred to as reaction functions. s Hence the …rms do not respond to the full bene…t (to the industry) of keeping sales low and price high as a monopolist would." This is easy to see from the Nash equilibrium condition: Cournot duopolists markup on marginal cost by p0 qi + qj qi . p (q1 + q2 ) = a b (q1 + q2 ) (where a > c and b > 0). it can make a s positive pro…t by pricing above marginal cost.e.) Solving jointly.
161
. p0 (Jq )
can be arranged for the equilibrium price p (Jq ) = c p0 (Jq ) q :
Intuitively.C. Exercise (MWG 12. J.D. the marketclearing price after a further entry (at the minimum average cost level q) would be greater than c.3). s so that p (Jq ) ! c.9).3
Competitive Limit
In the generalization to J …rms producing nonzero quantities with identical marginal costs c.e. if demand exceeded supply (plus an additional quantity q) at p = c. i.) If the number of …rms in the industry is determined by free entry in response to pro…t opportunities. as J ! 1. Jq = J p (Jq ) c . and entry would be pro…table. and Q the equilibrium total output (after all pro…table entry has occurred). competitive pricing is approached in the limit. each …rm’ production level becomes very small. Then Q + q x (c) because otherwise.e." Let p be the equilibrium price. …rstorder conditions imply that the outputs are equal.C. Exercise (MWG 12. the competitive limit is obtained as demand increases.7).C.
17.Exercise (MWG 12. Assume that each …rm has the same minimum average cost c at some level of output q > 0 (we do not impose constant marginal cost here). where x (p) is some aggregate demand function. : : : . and > 0 is the "market size. (I.20). when the market size is . and adding up over j = 1. Consider parameterized demand functions x (p) = x (p). Exercise (MWG 12.
This imposes a lower bound on the equilibrium supply Q : it cannot fall short by more than q of the quantity that would be supplied if p = c. It follows that the equilibrium price p is bounded above: it cannot exceed c by so much that the post entry price, after another …rm enters and supplies q, is still greater than c. Let p ( ) be the inverse supply function at market size , and de…ne the di¤erence between this upper bound on p and c (which is a lower bound such that no …rm exits) as 4p p ( x (c) q) p ( x (c)) :
In order to see what happens to the equilibrium price interval as market size increases from = 1 toward in…nity, rewrite 4p in terms of p ( ), the inverse supply function at = 1. Since q = x = x = q, we have p (q) = p (q = ), i.e. 4p = p which goes to zero as ! 1. x (c) q p (x (c)) ;
This means the equilibrium price converges to p (x (c)) = c, the longrun competitive price, as market size gets large. The industry output is then also competitive (and …rms are small relative to the market, since they produce at the …xed minimum average cost level q). With constant marginal cost c, c = c. Exercise (MWG 12.F.3). Exercise (MWG 12.F.2). Exercise (MWG 12.F.4).
18
18.1
Precommitment and Entry
Precommitment
In some industries, …rms act sequentially, rather than simultaneously as assumed so far. In fact, it may super…cially appear that sequential 162
moves are always more realistic, but this view misses the point of the distinction between the game forms. It is a matter of information, rather than timing. In a sequential game, strategy sets di¤er in that some players are able to condition their actions on what other players are observed to do. This is not necessarily a bene…cial power to have: …rst movers can potentially exploit it by manipulating the incentives for late movers. The e¤ect of an observed action by …rm i on the behavior of another …rm j is captured by j’ best response function. If j bestresponds s to a change in i’ strategy with a change in the same direction, i.e. s dbj (si ) =dsi > 0, then si and sj = bj (si ) are strategic complements. If j’ best response is in the opposite direction, i.e. dbj (si ) =dsi < 0, then s we have strategic substitutes. For instance, in pure Bertrand competition, a price cut by one …rm (above marginal cost) is answered by a price cut by the other …rm, so these are strategic complements. In a Cournot setting, a sales increase by one …rm often creates incentives for the other …rm to reduce its sales, so these are strategic substitutes. Example. Suppose costs depend on an investment k that …rm i can make (for example, in process innovation), i.e. c0 (k) < 0. Let the investment stage be followed by Cournot competition with linear inverse demand p (qi + qj ) = a b (qi + qj ) (where a > c 0 and b > 0) and constant marginal costs (namely, c (k) for i, and c for j). With …rmspeci…c marginal costs, the bestresponse functions are c (k) 1 + qj ; 2b 2 a c 1 qj (qi (k)) = + qi (k) ; 2b 2 qi (qj ; k) = a leading to equilibrium q i qj ; k = a a 163 2c (k) + c ; 3b 2c + c (k) : 3b
qj (qi (k)) =
The bestresponse level of output for i is increasing in the costreduction: for any qj , qi 0 (k) > 0. Di¤erentiating j’ best response with respect to k gives s @qj (qi (k)) = @k 1 0 q (k) < 0, 2 i
so output levels are strategic substitutes, and i’ investment in cost reducs tion leads to less aggressive behavior from j, which reinforces the inherent advantage of lower cost. Exercise (MWG 12.G.1). A Cournot duopolist may take advantage of the strategic substitutes relationship by precommitting to high sales, so that the competitor will concede a large share of the market. The sequential version of Cournot duopoly is called Stackelberg duopoly. One …rm (the "leader") sets its quantity …rst, and the choice is observed by the other …rm (the "follower"). The follower’ optimal reaction is s anticipated by the leader. Given the leader’ choice q1 , the follower s maximizes c) q2 : 2 (q1 ; q2 ) = (p (q1 + q2 ) The resulting bestresponse function for the follower is q2 (q1 ) = p (q1 + q2 ) c : p0 (q1 + q2 )
This is identical to a Cournot …rm’ best response function. s The leader, however, maximizes against the follower’ best response s function (rather than a …xed value):
1
(q1 ; q2 (q1 )) = (p (q1 + q2 (q1 ))
c) q1 :
This leads to the leader’ best response s q1 (q2 ) = p (q1 + q2 (q1 )) c 1 q2 (q1 ) = : 0 (q + q (q )) 1 + q 0 (q ) p 1 1 + q2 0 (q1 ) 2 1 2 1
If q1 and q2 are strategic substitutes, i.e. q2 0 (q1 ) < 0, then q1 (q2 ) > q2 (q1 ). If q1 and q2 are strategic complements, i.e. q2 0 (q1 ) > 0, then q1 (q2 ) < q2 (q1 ). 164
then the industry price decreases with strategic substitutes and increases with strategic complements relative to Cournot competition.Since q2 (q1 ) is the Cournot best response function. the leader sells a larger quantity. knowing that the follower will practice restraint in order avoid too much price deterioration.g. the overall quantity sold increases from q1 + q2 = (2=3) (a c) =b to q1 + q2 = (3=4) (a c) =b under Stackelberg competition. 165
. both …rms produce a nonzero quantity. You can check that the leader’ pro…t is larger. Compared to the simultaneous Cournot outcome q1 (q2 ) = q2 (q1 ) = (a c) = (3b). as long as price exceeds marginal cost. (There is always at least a small strictly positive pro…t to be made. q1 (q2 ) = 2q2 (q1 ) = whereas q2 (q1 ) = a b a c c q1 = a 2b c . it is clear that total output increases in the case of strategic substitutes. This intention can be signaled credibly (in the sense that it becomes the best response) by investing in high capacity and low marginal cost. With linear inverse demand. a 2b c q1 : 2 b (q1 + q2 ). in product development or a plant). the leader might deter the follower from incurring the startup cost by committing to an aggressive response (high sales at a low price). Example. price decreases from p (q1 + q2 ) = a=3 + (2=3) c to p (q1 + q2 ) = a=4 + (3=4) c (keeping in mind that a > c). the
: 4b This is the strategic substitutes case (q2 0 (q1 ) = 1=2). If inverse demand p ( ) is decreasing in combined sales. than in Cournot s s competition.) When production requires a …xed initial outlay (e. In the Stackelberg model. p (q1 + q2 ) = a follower’ best response in the Stackelberg model is s q2 (q1 ) = Since 1 + q2 0 (q1 ) = 1=2. As a result. and the follower’ pro…t smaller. and total output decreases in the case of strategic complements.
There is an in…nite number of potential entrants who decide. The entry criterion m (q ) K is therefore equivalent to K (a c)2 = (4b). J = 1) and set the monopoly price.e. In an alternative setup.2
Entry Equilibrium
In the remainder of the lecture. for any number J of entrants. given the prevailing pro…t J : K and J +1 < K: J (We assume …rms enter if indi¤erent. entry cost may be incurred only if the …rm makes nonzero sales. and bear the …xed cost only if it can make a nonnegative pro…t. Since m (q) = (p (q) c) q is maximized by q = (a c) = (2b). Suppose stage 2 is a pure Bertrand game.The leader might then be able to operate as a monopolist. at stage 1 of the game. and inverse demand is linear. whether to invest K or not. yielding pro…t J for each entrant (this excludes the entry cost K). we return to a symmetric environment and consider how much entry will occur in equilibrium with a …xed entry cost K > 0. If the monopoly pro…t m exceeds entry cost K. p (q) = a bq with a > c 0 and b > 0.E. Assume that. the optimal pro…t is m (q ) = (a c)2 = (4b). and the entrants compete at stage 2 as oligopolists.)
! 0 as J ! 1. a single …rm will enter (i. where …rms have constant marginal costs c. This approach restores something close to marginal cost pricing in the 166
. We know that J = 0 when J > 1. symmetric Nash equilibrium. and J is unique. then the equilibrium
Example.1). Exercise (MWG 12. The equilibrium number of entrants J is such that no …rm wants to either enter or exit.e. a …rm can observe how many other …rms enter.
18. if If J is decreasing in J. there is in stage 2 a unique.
J J +1
= K. possibly under constraints to keep the price low enough to continually discourage entry. i. Thus.
Industries where such "hit and s run" entry is possible are called contestable markets. In other words. which approaches c when industry demand x (p) is large. A monopolist could in this case not set a price above p = (K + cx (p)) =x (p). Each entrant maximizes
J
= (p (JqJ )
c) qJ = (a a 2b a b c c
(J J
1) bqJ 1 2
bqJ
c) qJ .
so that qJ = = qJ
1 . since it can exit freely s if the incumbent’ behavior changes. Since J = K at ~ ~ a c 1.Bertrand entry game. As K ! 0 or b ! 0. Thus
J
=
1 b
a c J +1
is strictly decreasing in J . 167 J bqJ = a J (a J +1 c) 1 J a+ c J +1 J +1
. Else. the entrant would not have to worry that it may be undercut in the ensuing Bertrand competition and be left with a loss of K. J ! 1 and the industry price p (J qJ ) = a = approaches marginal cost. Under the same cost and demand conditions. It can simply respond to the incumbent’ current price. another …rm could enter without prepaying the entry cost K. consider a Cournot game in stage 2. Example. J +1
2
since all …rms symmetrically set qJ = qJ . and J ! 0 as J ! 0. J=p bK ~ the equilibrium number of entrants J is the greatest integer below J.by undercutting the prevailing price. since …rms enter if and only if there is positive pro…t to be made.
since price equals marginal cost if two …rms compete (hence surplus is maximized.18. Given that …rms choose their production levels to maximize pro…t at the postentry stage. Example. the number is either socially optimal. Since a single …rm enters in equilibrium. S (J) = J (a = J J +1 c) qJ 1 2 1 2 JbqJ 2 J J +1
2
K ! (a b c)2 JK:
The welfaremaximizing number of …rms J
0
satis…es K = 0.3
Socially Optimal Entry
Entry in an industry provides. This number J maximizes consumption bene…ts (as measured by the willingness to pay for each unit) less production cost (which includes each …rm’ entry cost K): s S (J) = Z
JqJ
p (s) ds
J (c (qJ ) + K) :
0
Example. duplicates entry cost for …rms. on the other. on the one hand. there exists an e¢ cient number of entrants from a social perspective that optimally resolves this tradeo¤. In the Cournot example that was already described. or one fewer than socially optimal. Returning to the pure Bertrand example. the socially optimal number of entrants can be no more than two. and it cannot be socially desirable to incur further entry costs). valued goods to consumers and.
2
1 (a c)2 S (J ) = b (J + 1)3 so (J + 1) =
3
a c p bK
:
168
. given the market structure.
unless it can extract them through price discrimination or is o¤ered a subsidy. More typically. The same logic could apply to an additional potential entrant in an industry with some number of incumbents . equilibrium entry may well exceed the socially optimal level. there is overentry because …rms consider only whether their pro…ts can cover the cost of entry.it may be socially valuable. overentry is possible due to the "businessstealing" e¤ect. but ignore the reduction in incumbents’pro…ts. we have (possibly) too little. No …rm may …nd it pro…table to enter an industry as a monopolist because the maximal pro…t the …rm can make with perunit pricing does not recover the cost of entry. In short. However. But for the …rm. since J + 1 = J + 1 < J + 1. while the combined entry cost they incurred remains …xed. Exercise (MWG 12. since the consumer surplus may o¤set losses in pro…t.3). There are two kinds of failures. Of course. the net increase in total output (and consumption bene…ts) may not justify the duplication of entry costs. From a social point of view.E. Entry may be ine¢ cient with Bertrand as well as with Cournot competition: in the …rst case.1 was the equilibrium number of entrants.
~ Recall that J = (a
p c) = bK
169
. but may fall short by at most one …rm. the potential monopolist ignores these bene…ts to consumers. from a social perspective it may be desirable that the monopolist operate. These insights can be stated as a general result: under standard conditions. This reasoning suggests that underentry by one …rm (relative to J ) is possible. entry may still be pro…table because its sales exceed the increase in total output (the di¤erence being output reductions by incumbents) and could therefore cover the entry cost. entry. in the second generally too much. but not privately optimal. for the entry to occur. their contribution to surplus declines. Because the incumbents sell fewer units after an additional entry. 2=3 ~ ~ It exceeds the socially optimal number.
where J is the socially optimal number of entrants. Therefore. Thus. i. so p ((J 1) qJ 1 ) is the maximum in the price interval from p (J qJ ) to p ((J 1) qJ 1 ).e. J qJ > (J 1) qJ 1 .e. and if furthermore p0 ( ) < 0 and c00 ( ) 0. If marginal …rm pro…ts are positive for any number of entrants J (i. let J > 1. J (p ((J 1) qJ 1 ) p ((J 1) qJ 1 ) qJ Since qJ 1 > qJ and p ((J tion. Proof. p ((J 1) qJ
1 ) qJ 1
0
and
1
c0 (qJ c (qJ
1)
1 )) (qJ 1)
1
qJ ) 0 by assumpK:
K:
1)
1) qJ c (qJ
c0 (qJ
1) J
1
170
. and more entry increases industry output ( J > J 0 =) JqJ J 0 qJ ) but decreases …rm output and pro…t ( J > J 0 =) qJ qJ 0 ). and price is a decreasing function of industry supply. By de…nition. the convexity of the cost function (c00 ( ) 0) implies Z qJ c (qJ ) c (qJ 1 ) = c0 (s) ds c0 (qJ 1 ) (qJ
qJ
1
1
and
c (qJ
1 ))
qJ
1) 1)
since c (qJ 1 ) is the minimum in the cost interval between c0 (qJ c0 (qJ ). W (J ) W (J 1). then the equilibrium number of entrants J J 1. hence: Z J qJ p (s) ds J (c (qJ ) c (qJ 1 )) c (qJ 1 ) K:
(J 1)qJ
1
Under the assumptions.) Then. by rearranging the …rst inequality.Proposition. Z J qJ p (s) ds p ((J 1) qJ 1 ) (J qJ (J 1) qJ 1 ) :
(J 1)qJ
1
(Think of the fact that the area under the curve p (q) between q0 (J 1) qJ q1 J qJ is contained in the rectangle formed by q1 q0 and the maximal height of the function p (q). the right side is positive. p (JqJ ) c0 (qJ ) 0). Since the claim is obviously true when J = 1. J (p ((J 1) qJ 1 ) (qJ 1 qJ ) + c (qJ ) p ((J 1) qJ 1 ) qJ 1 c (qJ 1 ) K: Now.
and p (JqJ )
p ((J
1) qJ
1)
from JqJ
and
Exercise (MWG 12.4).2).
since qJ qJ 0 p ( ) < 0. because
J
is
= p (JqJ ) qJ p ((J 1) qJ 1 ) qJ p (JqJ ) qJ p ((J 1) qJ 1 ) qJ = (p (JqJ ) c0 (qJ 1 )) qJ (p ((J
1
1 1
(c (qJ ) c (qJ 1 )) c0 (qJ 1 ) (qJ qJ 1 ) 1) qJ 1 ) c0 (qJ 1 )) qJ (J 1) qJ
1
1
0.
171
.Then …rms must enter at least until their number is J decreasing in J:
J J 1
1.E. Exercise (MWG 12.E.