BL MSc.

Micro 20 1
Lecture 1: Consumer Theory
The Axioms of Consumer Preference
The Utility Function
The Utility Maximisation Problem
The Expenditure Minimisation Problem
The Slutsky Equation and the Law of Demand
Utility Maximisation with Endowments
An Application: The Life-Cycle Model
BL MSc. Micro 20 2
1.Introduction
Consumer theory explains how a rational decision-maker chooses to allocate her
income between different goods and services, taking as fixed the prices for these
goods and services.
Basic consumer theory can easily be extended to study how a decision-maker makes
her labour supply and savings decisions.
The consumer: a rational decision-maker who has a fixed income at her disposal, and
who values goods and services.
[Note that in practice, expenditure decisions are usually made by groups of people
living together (households), although there are increasing numbers of one-person
households.]
BL MSc. Micro 20 3
2. Consumption Bundles and the Consumption Set
The consumer values n different commodities i = 1,..n
x
i
- the amount of commodity i purchased and consumed by the consumer
x
i
is a non-negative real number (commodities infinitely divisible).
x = (x
1
, x
2
, ..x
n
) a vector of consumption levels, the consumption bundle.
We will denote specific bundles by letters (x, y, z, etc). In the case of two goods:
x
1
y
x
2
y
1
y
2
BL MSc. Micro 20 4
The consumption set X is the set of all possible consumption bundles.
Without any further restrictions, X is simply the set of all non-negative consumption
bundles:
However, in practice, certain minimum levels of consumption of goods may be needed
e.g. to keep the consumer alive and so the consumption set is modified to take these
constraints into account. [Exercise: draw the consumption set in the case of n = 2, if
the consumer needs at least amount y
1
of good 1, and y
2
of good 2]
x
1
x
2
X
BL MSc. Micro 20 5
3. Consumer Preference Relations
Modern theory only assumes that consumers can rank pairs of consumption bundles,
and that these rankings satisfy certain assumptions.
It can then be shown that given these assumptions (axioms) on rankings, consumers
behave as if they maximise utility.
x R y - the consumer judges bundle x to be at least as good as bundle y.
R is called the weak preference relation.
Note that R does not in itself tell us whether the consumer strictly prefers x to y or
whether the consumer is indifferent between x and y.
However, once we know R, we can easily calculate which of these is the case.
BL MSc. Micro 20 6
The strict preference relation P on X is defined as follows:
x P y if and only if x R y and not y R x .
The indifference relation I on X is defined as follows:
x I y if and only if x R y and y R x .
In consumer theory, we make some assumptions about the weak preference relation R.
These assumptions are called axioms, but are really hypotheses about consumer
behaviour that are in principle, testable.
BL MSc. Micro 20 7
Axiom 1: Completeness. For all xand yin X, either x R yor y R x.
This says that the consumer can rank any pair of consumption bundles. The
completeness axiom obviously implies that either x P yor y P xor y I x
Axiom 2: Transitivity. For any three bundles x ,y, z in X, if x R yand y R z then x R
z.
This assumes that a consumers choices be consistent.
Note that transitivity of R implies transitivity of P and transitivity of I (Q1, problem set 1).
Without the transitivity axiom, it would be possible that a consumer would prefer xto y,
and y to z, and finally z to x. So, she may not be able to identify any feasible
alternative as best, and so could not make a rational choice.
BL MSc. Micro 20 8
The next axiom is the continuity axiom.
Consider a sequence of consumption bundles {x
n
}= (x
1
, x
2
, x
3
, x
n
) which
converges to a limit bundle y.
Also, let z be some other consumption bundle.
Axiom 3: Continuity. If {x
n
} converges to y, and x
n
R z, all n = 1,2,.then y R z also.
In words: if all consumption bundles close to y (in n-dimensional space) are preferred
to z, bundle y is preferred to bundle z also.
BL MSc. Micro 20 9
Example where Axiom 3 is not satisfied. (lexicographic preferences)
Assume n = 2, and that good 1 is beer, and good 2 is cheese.
Given any two bundles x, y, the consumer strictly prefers x as long as it contains more
beer than y If x, y have the same amount of beer, the consumer strictly prefers the
bundle that has more cheese. [So, the consumer ranks beer as infinitely better than
cheese].
Now consider the sequence {x
n
}, where x
n
= (1+1/n, 0), and the bundle z = (1,1).
Note that:
- x
n
converges to y = (1,0) as n
- The consumer strictly prefers any x
n
to z as x
n
has more beer (i.e. x
n
P z, all n)
- But, the consumer strictly prefers z to y as z has more cheese
So, axiom 3 is not satisfied as x
n
P z, all n = 1,2,.but z P y.
BL MSc. Micro 20 10
Axiom 4: Strict Monotonicity. Assume that bundle x contains at least as much of all
goods as bundle y (i.e.
i i
y x >
, all i = 1,..n) and strictly more of at least one good as y
(i.e.
i i
y x >
, some i = 1,..n). Then, x P y.
This says that consumers are always greedy; no matter how much they have of any
commodity, they always want more.
4. Representation of Preferences by a Utility Function
Under assumptions 1-4, the preference relation R can be represented, or described, by
a utility function u mapping consumption bundles to real numbers.
Formally, a utility function is a function u: X 9 where u(x) is the number assigned to
consumption bundle x. (9 = the set of real numbers)
The utility function u: X 9 represents consumer preferences if: u(x) > u(y) if and
only if x R y.
BL MSc. Micro 20 11
Note:
1. u(x) is a function of n variables i.e. u(x) = u(x
1
,x
2
x
n
) where x
i
is the level of
consumption of the ith good.
2. By definition, if u(.) represents a particular consumers preferences, then so does any
strictly increasing function of u e.g. u
2
, lnu, u+1, etc. (Why?)
So, if a preference relation is represented by any utility function, always represented by
a whole family of utility functions that are strictly increasing transformations of each
other.
Theorem. Suppose that the consumer preference relation is complete, transitive,
continuous, and strictly monotonic (i.e. satisfies Axioms 1-4). Then there exists a
continuous utility function which represents those preferences. Moreover, u(x) =
u(x
1
,x
2
x
n
) is strictly increasing in each of its arguments.
BL MSc. Micro 20 12
45
o
x
1
x
2
x
u(x)
u(x)
z
Proof
It can be proved that for any xeX, there exists a unique utility number u(x) see
Appendix to notes.
So, as x varies, we can think of u(x) as a function mapping all x in X to real numbers.
We call this function the utility function.
Step 1: Construction of a Utility Number
Consider some bundle x = (x
1
, x
2
) in X.
Then, given A1-A4, we can find a bundle
z = (u(x),u(x)) i.e. with equal amounts of
goods 1 and 2 which is indifferent to x, i.e.
(u(x),u(x)) I x
u(x) - the utility number of x.
BL MSc. Micro 20 13
Step 2: Proof that the Candidate Utility Function Represents Preferences
Consider two bundles x and y. W e need to show that x R y if and only if u(x) u(y).
But by definition:
(u(x),u(x)) I x, (u(y),u(y)) I y (*)
But then given (*),
x R y (u(x),u(x)) I x and xR y and y I (u(y),u(y))
(u(x),u(x)) R (u(y),u(y)) (by A2, transitivity)
u(x) u(y) (by A4, strict monotonicity)
BL MSc. Micro 20 14
5. Indifference Curves
For any reference bundle y, define the indifference set to be the set of consumption
bundles ranked as indifferent to y:
I(y) = {x|u(x)=u( y)}
From the strict monotonicity axiom A4, we can see that the indifference sets must be
thin and downward-sloping, as shown below.
x
1
x
2
I(y)
y
BL MSc. Micro 20 15
Due to this shape, indifference sets are usually called indifference curves, and we use
this terminology from now on.
6. Convexity of Preferences
The final axiom is the convexity axiom. It is not required for the existence of a utility
function. Rather, it is needed to make sure that the demand functions generated by
utility maximisation are well-behaved i.e. single-valued and continuous.
Let x = (x
1
, x
2
, ..x
n
) , y = (y
1
, y
2
, ..y
n
). Then a convex combination of x and y is
x+(1-)y = (x
1
+(1-)y
1
, x
2
+(1-)y
2
,. (x
n
+(1-)y
n
), for some 0 s s 1.
For example, if x = (1,2), and y = (2, 1), then a convex combination is:
x+(1-)y = (1+(1-)2, 2+(1-)1) for some 0 s s 1.
BL MSc. Micro 20 16
Axiom 5: Strict Convexity. Given consumption bundles x, y, z with x R z, y R z, and x
y, then x+(1-)y P z for all 0 < <1.
Implication: A5 implies indifference curves convex to the origin.
x
2
x
1
1
2
x
y
x+(1-)y for
some 0 s s 1
1
2
Geometrically, a convex
combination of two
consumption bundles lies
on a straight line between
the two bundles.
BL MSc. Micro 20 17
x
1
x
2
x
y
x+(1-)y
In this example, x,y,z lie
on the same IC, so x R
z, y R z.
So, by A5, we have
x+(1-)y P z, all 0 <
<1
So, all points on the
straight line between x
and y are preferred to z
So, z must lie below the
straight line connecting
x and y by A5
z
BL MSc. Micro 20 18
Theorem. The utility function is strictly quasi-concave if and only if preferences are
strictly convex i.e. A5 holds.
Recall strict quasi-concavity of u is:
For any x,y and 0 < <1,
)} ( ), ( min{ ) ) 1 ( ( y u x u y x u > +
Proof of if
A5: Given consumption bundles x, y, z with x R z, y R z, and x y, then x+(1-)y P z
for all 0 < <1
Assume w.l.o.g. that x R y, and also that z=y. Then A5 implies then x+(1-)y P y for
all 0 < <1, or
)} ( ), ( min{ ) ( ) ) 1 ( ( y u x u y u y x u = > +
Proof of only if left as an exercise.
BL MSc. Micro 20 19
7. Utility Maximisation
The consumer is assumed to choose the consumption bundle that is the most
preferred one in her consumption set, given also the financial constraints that she
faces. We assume A1-A5 throughout.
The Budget Set
p
i
price of one unit of good i
m total income of the consumer
We assume
0 , 0 > > m p
i
Expenditure on goods is

=
= + +
n
i
i i n n
x p x p x p x p
1
2 2 1 1
...
The budget constraint : expenditure be no more than income i.e.
m x p
n
i
i i
s

=1
m and p
i
are assumed
exogenous to the
consumer, and are strictly
positive real numbers.
BL MSc. Micro 20 20
The budget set B - the set of consumption bundles that are physically feasible and
satisfy the budget constraint.
Generally, the budget set is defined as: B = {x |x
i
> 0, i = 1,..n, Ep
i
x
i
s m }. Note that it
is closed and bounded.
x
1
x
2
Budget line, with
slope: p
1
/p
2
m/p
1
m/p
2
B
BL MSc. Micro 20 21
The Consumers Utility Maximisation Problem
(UMP) To choose (x
1
,x
2
, ..x
n
) to maximise utility u(x
1
,x
2
, ..x
n
) subject to the constraint
that (x
1
,x
2
, ..x
n
) lies in the budget set B.
Note:
- As B is compact (closed and bounded) and u(x
1
,x
2
, ..x
n
) is continuous, this problem
has at least one solution (Extreme value theorem, maths lecture notes).
- From Axiom 4, the budget constraint will hold with equality at any solution:
m x p
n
i
i i
=

=1
.
- From Axiom 5, strict convexity, the solution to this problem is unique.
We assume from now on that u(x
1
,..x
n
) is differentiable i.e. no kinks in indifference
curves.
BL MSc. Micro 20 22
Then, as u(.) is strictly quasi-concave, by the pre-sessional maths notes, the solution to
the UMP is fully described by the Kuhn-Tucker conditions (Lagrangean Approach).
There then two possibilities:
A. Interior Solution: Indifference curve tangent to the budget line, strictly positive amounts of both
goods consumed.
B. Corner solution: Indifference curve at a corner of the budget set, only one good consumed.
We assume an interior solution in what follows.
BL MSc. Micro 20 23
Solving the UMP
Then, the UMP can be stated as:
(UMP) choose (x
1
,x
2
, ..x
n
) to maximise u(x
1
,x
2
, ..x
n
) subject to
m x p
n
i
i i
=

=1
.
To solve, write the Lagrangian
] [ ) ,... , (
2 2 1 1 2 1 n n n
x p x p x p m x x x u L + =
The FOC for a solution to UMP are:
n i p
x
u
x
L
i
i i
,.. 1 , 0 = =
c
c
=
c
c

, (1)
0
2 2 1 1
= =
c
c
n n
x p x p x p m
L

(2)
Equations (1) and (2) are
n+1 equations that can be
solved for the n+1
unknowns x
1
,x
2
, ..x
n
and .
BL MSc. Micro 20 24
Example: Cobb-Douglas utility
Assume u =
n
n
x x x
o o o
) ....( ) ( ) (
2 1
2 1 , o
i
> 0.
We take the log of u before solving UMP (this is legitimate) to get

=
i
i i
x u ln ln o
Then, Lagrangean is:

+ =
i
i i
i
i i
x p m x L ) ( ln o
So, the n+1 equations are:
n i p
x x
L
i
i
i
i
,.. 1 , 0 = = =
c
c

o
(1)
0
2 2 1 1
= =
c
c
n n
x p x p x p m
L

(2)
Unknowns are x
1
,x
2
,..x
n
,
, and parameters are
m, p
1
,p
2
, ..p
n
, o
1
,... o
n
BL MSc. Micro 20 25
Step I: Find formula for in terms of parameters
(1)
n i x p
i i i
,.. 1 , = = o
Now sum over i=1,..n :
n i x p
n
i
i i
n
i
i
,.. 1 ,
1 1
= =

= =
o

m
n
i
i
o =

=1
(using (2))

m m
n
i
i
|
o
= =

=1
(3)
NB: This
method will also
work in more
complex
problems.
BL MSc. Micro 20 26
Step 2: Find formulae for x
1
, ..x
n
in terms of parameters, using (1), (2), (3):
n i
p
m
p
x
i
i
i
i
i
,.. 1 , = = =
|
o

o
Special case n = 2, (used below):
2
2
1
1 2 1
) 1 (
, , 1 ,
p
m
x
p
m
x
o o
o o o o

= = = =
BL MSc. Micro 20 27
Marshallian Demands
The solutions x
1
,..x
n
to UMP can be written as functions of parameters p
1
,..p
n
, m:
x
i
=x
i
(p
1
,..p
n
, m) (Marshallian demand functions)
Key Properties of Marshallian Demands:
1. Continuity. x
i
(p
1
,..p
n
, m) is continuous in (p
1
,..p
n
, m).
2. Homogeneity. If all prices and incomes double (or change by some common
multiple) then as the budget set is unchanged, Marshallian demands do not change.
Mathematically, x
i
(p
1
,..p
n
, m) is homogenous of degree zero in (p
1
,..p
n
, m).
3. Under certain technical conditions (J-R, Theorem 1.5, assumed satisfied) , x
i
(p
1
,..p
n
,
m) is differentiable in (p
1
,..p
n
, m)
4. Budget balance.

=
=
n
i
i i
m x p
1
BL MSc. Micro 20 28
Indirect Utility
The indirect utility function measures the maximum utility that the consumer can
achieve by choice of a consumption bundle, given prices p
1
,..p
n
and income m.
v(p
1
,..p
n
,m) u(x
1
(p
1
,..p
n
,m), x
2
(p
1
,..p
n
,m),. x
n
(p
1
,..p
n
,m))
Example: two goods (n = 2), and Cobb-Douglas utility u =
o o 1
2 1
) ( ) ( x x
.
In this special case, o o o o = = 1 ,
2 1
, so the Marshallian demands are
1
2 1 1
) , , (
p
m
m p p x
o
=
,
2
2 1 2
) 1 (
) , , (
p
m
m p p x
o
=
Substituting back into the utility function:
1
2 1
1
1
2 1
2 1
) ( ) ( ) 1 (
) 1 (
) , , (

=
|
|
.
|

\
|
|
|
.
|

\
|
=
o o o o
o o
o o
o o
p p m
p
m
p
m
m p p v
BL MSc. Micro 20 29
Properties of the Indirect Utility Function
1. Continuous in (p
1
,..p
n
, m)
2. Strictly increasing in m, and decreasing in p
i
.
3. v(p
1
,..p
n
, m) is homogenous of degree zero in (p
1
,..p
n
, m). If all prices and incomes
double (or change by some common multiple) then as the budget set is unchanged,
indirect utility is unchanged.
4. If
) , ,.. (
1
m p p v
n
is differentiable, Roys identity:
) , ,.. (
/
/
1
m p p x
m v
p v
n i
i
=
c c
c c

. This says that

the Marshallian demand function can be recovered from the indirect utility function.
BL MSc. Micro 20 30
Proof of Roys Identity (example of the envelope theorem)
First, note that we can write:
v(p
1
,..p
n
,m) u(x
1
(p
1
,p
2
,m), x
2
(p
1
,p
2
,m))
u(x
1
(p
1
,p
2
,m), x
2
(p
1
,p
2
,m)) + ( p
1
,p
2
,m )[m p
1
x
1
(p
1
,p
2
,m) p
2
x
2
(p
1
,p
2
,m)]
So, differentiating both sides with respect to m:
| |
2 2 1 1
2
2
1
1
2
2
1
1
1 1 x p x p
m m
x
p
m
x
p
m
x
x
u
m
x
x
u
m
v

c
c
+
(

c
c

c
c
+
c
c
c
c
+
c
c
c
c
=
c
c

=
+
c
c
|
|
.
|

\
|

c
c
+
c
c
|
|
.
|

\
|

c
c
=
m
x
p
x
u
m
x
p
x
u
2
2
2
1
1
1
By a similar argument,
i
i
x
p
v
=
c
c
. Combining
i
i
x
p
v
=
c
c
,
=
c
c
m
v
gives Roys identity.
In the last line, we
have used the fact
that the terms in the
brackets are zero
when the consumer
is choosing
consumption
optimally.
BL MSc. Micro 20 31
8 The Dual Approach to Consumer and Producer Behaviour
The Expenditure Minimisation Problem
An alternative way of thinking about consumer behaviour is to observe that the
consumer can be thought of as taking a utility level (rather than income, m) as given. In
this case, a rational consumer will minimise the level of expenditure on goods needed
to achieve this utility level.
x
1
x
2
e/p
2
e/p
1
Iso-expenditure line combinations of
bundles of goods that can be bought given
expenditure e
All iso-expenditure lines have slope p
1
/p
2
BL MSc. Micro 20 32
The expenditure minimisation problem (EMP) of the consumer is then to find the
minimum level of expenditure required to obtain a target level of utility u.
x
1
x
2
I
I
e
1
e
2
e
3
I I fixed indifference curve
The consumer can achieve the target utility level
at least cost by spending e
2
(e
1
is too much, and e
3
not enough).
The resulting consumption levels are known as
Hicksian demands (h
1
,h
2
)
h
1
h
2
BL MSc. Micro 20 33
Formally, the EMP is:
Choose (x
1
,x
2
, ..x
n
) to minimise e =
n n
x p x p x p ..
2 2 1 1
+ +
subject to
u x x u
n
> ) ,. (
1
where u - the target level of utility.
Note:
- The constraint set is closed (but not bounded) and u(x
1
,x
2
, ..x
n
) is continuous. The
extreme value theorem does not apply but a different argument can be made that
EMP has at least one solution (J-R, p35).
- From Axiom 4, the utility constraint will hold with equality at any solution:
u x x u
n
= ) ,. (
1
.
- From Axiom 5, strict convexity, the solution to this problem is unique.
BL MSc. Micro 20 34
Take the Lagrangean approach: Lagrangean is
)] ,... , ( [
2 1 2 2 1 1 n n n
x x x u u x p x p x p L + + + =
The FOC for a solution to UMP are:
n i
x
u
p
x
L
i
i
i
,.. 1 , 0 = =
c
c
=
c
c

, (1)
0 ) ,.. (
1
= =
c
c
n
x x u u
L

(2)
The solution to this problem gives the Hicksian demands
) , ,.. (
1
u p p h
n i , which can then
be plugged back into the expenditure formula to get the expenditure function:
Equations (1) and (2) are
n+1 equations that can
be solved for the n+1
unknowns x
1
,x
2
, ..x
n
and
.
BL MSc. Micro 20 35

=
=
n
i
n i i n
u p p h p u p p e
1
1 1
) , ,.. ( ) , ,.. (
Two ways of finding the expenditure function:
1. Direct/brute force method. Lagrangean method above to find
) , ,.. (
1
u p p h
n i and then
plug back into expenditure formula to get the expenditure function.
2. Indirect/inversion method. Suppose that at given prices p
1
,..p
n
, u is the maximum
utility that is attainable with income m i.e.
u = v(p
1
,..p
n
, m) (1)
Then, the minimum expenditure required to attain u must be m i.e.
m = e(p
1
,..p
n
, u) (2)
We can obtain the
expenditure function by
solving (1) for m as a
function of p
1
,..p
n
, u. This is
often convenient as (i) the
UMP is often easier to solve
than the EMP; (ii) it avoids
the need to solve both
problems.
BL MSc. Micro 20 36
Example
Assume n= 2 and Cobb-Douglas preferences. Then we know that
1
2 1
1
2 1
) ( ) ( ) 1 ( ) , , (

=
o o o o
o o p p m m p p v
So, equation (1) becomes
u = m p p
1
2 1
1
) ( ) ( ) 1 (

o o o o
o o
So, solving this equation for m, we get
u p p m
o o o o
o o

=
1
2 1
) 1 (
) ( ) ( ) 1 (
On the right-hand side, we have the formula for the expenditure function, as required
i.e.
u p p u p p e
o o o o
o o

=
1
2 1
) 1 (
2 1
) ( ) ( ) 1 ( ) , , (
BL MSc. Micro 20 37
Properties of the Expenditure Function
1. Continuous in (p
1
,..p
n
, u)
2. Strictly increasing in u, and increasing in p
i
.
3. e(p
1
,..p
n
, u) is homogenous of degree one in (p
1
,..p
n
). E.g. if all prices double then
twice as much as expenditure is needed to achieve u.
4. It is concave in (p
1
,..p
n
) i.e.
) , ' ( ) 1 ( ) , ( ) , ' ) 1 ( ( u p e u p e u p p e + > +
.
5. If
) , ,.. (
1
u p p e
n
is differentiable, Shepards Lemma :
) , ,.. (
1
u p p h
p
e
n i
i
=
c
c
. This says that
the Hicksian demand function can be recovered from the indirect utility function.
Properties 4 and 5 can be used to derive several interesting properties of the Hicksian
demands.
BL MSc. Micro 20 38
Shepards Lemma: An Example
Recall in the two-good Cobb-Douglas case,
u p p u p p e
o o
k

=
1
2 1 2 1
) ( ) ( ) , , (
where
) 1 (
) 1 (
o o
o o k

=
Then by Shepards Lemma,
) , , ( ) (
2 1 1
1
1
2
1
u p p h u
p
p
p
e
= =
c
c
o
ok
,
) , , ( ) ( ) 1 (
2 1 2
2
1
2
u p p h u
p
p
p
e
= =
c
c
o
k o
BL MSc. Micro 20 39
Concavity of the Expenditure Function in Prices: Proof
Notation. Let p,x be quantity and price vectors, denote inner products as

=
=
n
i
i i
x p x p
1
.
.
Suppose:
1
x
achieves u at minimum cost when prices are
1
p
2
x
achieves u at minimum cost when prices are
2
p
*
x
achieves u at minimum cost when prices are
, ) 1 (
2 1 *
p p p + =
1 0 s s
Then by definition:
BL MSc. Micro 20 40
* 1 1 1
x p x p s
,
* 2 2 2
x p x p s
Multiply first inequality by

, second by
1
, and add:
* * * 2 1 * 2 * 1 2 2 1 1
) ) 1 ( ( ) 1 ( ) 1 ( x p x p p x p x p x p x p = + = + s +
But by definition:
1 1 1
) , ( x p u p e =
,
2 2 2
) , ( x p u p e =
,
* * *
) , ( x p u p e =
) , ) 1 ( ( ) , ( ) , ( ) 1 ( ) , (
2 1 * 2 1
u p p e u p e u p e u p e + = s +
As required. QED.
BL MSc. Micro 20 41
Proof of Shepards Lemma
To prove Shepards Lemma, we apply the envelope theorem. First, note that in the
case of two goods, we can write:
e(p
1
,p
2
,u) p
1
h
1
(p
1
,p
2
,u)+p
2
h
2
(p
1
,p
2
,u))
p
1
h
1
(p
1
,p
2
,u)+p
2
h
2
(p
1
,p
2
,u) + (p
1
,p
2
,u)[u u(h
1
(p
1
,p
2
,u), h
2
(p
1
,p
2
,u)) ]
So, differentiating both sides with respect to (say) p
1
:
| | ) , (
2 1
1 1
2
2 1
1
1 1
2
2
1
1
1 1
1
h h u u
p p
h
x
u
p
h
x
u
p
h
p
p
h
p h
p
e

c
c
+
(

c
c
c
c

c
c
c
c
+
c
c
+
c
c
+ =
c
c

1
1
2
2
2
1
1
1
1 1
h
p
h
x
u
p
p
h
x
u
p h
=
c
c
|
|
.
|

\
|
c
c
+
c
c
|
|
.
|

\
|
c
c
+ =
In the last line, we have
used the fact that the
terms in the brackets
are zero when the
consumer is choosing
consumption optimally
in the EMP.
BL MSc. Micro 20 42
Interpreting Hicksian Demands
Hicksian demands measure the change in demand due to
changes in prices, given that the consumer is
compensated by changes to her income m so that her
utility is unchanged i.e. she stays on the same indifference
curve.
x
1
x
1
p
1
h
1
(p
1
, p
2
, u)
A
B
A
B
AA old budget line with slope
2 1
/ p p
A

BB new budget line with slope

2 1
/ p p
B

Note
A B
p p
1 1
>
A
p
1
B
p
1
x
2
BL MSc. Micro 20 43
Properties of Hicksian Demands
As Hicksian demands are related to the expenditure function via Shepards Lemma, the
properties of the expenditure function imply certain properties of the Hicksian demands.
1. h
i
(p
1
,..p
n
, u) is homogenous of degree zero in (p
1
,..p
n
).i.e. if all prices double (or
change by some common multiple) then h
i
does not change.
Reason: relative prices do not change i.e. the slope of the budget line is unchanged, so
the same consumption bundle is required to achieve the target level of utility at
minimum cost.
[Note: this result follows mathematically from Property 1 of the expenditure function: as
e is HOD1, all its partial derivatives are HODO].
BL MSc. Micro 20 44
2. Symmetry of Hicksian demands.
i
n j
j
n i
p
u p p h
p
u p p h
c
c
=
c
c
) , ,... (
) , ,... (
1
1
i.e. the marginal effect of
the price of good j on the Hicksian demand for good i is equal to the marginal effect
of the price of good i on the Hicksian demand for good j
Reason: from Shepards Lemma,
j i
n
j
n i
p p
u p p e
p
e p p h
c c
c

c
c ) , ,. ( ) , ,.. (
1
2
1
,
i j
n
i
n j
p p
u p p e
p
e p p h
c c
c

c
c
) , ,. (
) , ,.. (
1
2
1
But from Youngs Theorem ,
i j
n
j i
n
p p
u p p e
p p
u p p e
c c
c
=
c c
c ) , ,. ( ) , ,. (
1
2
1
2
.
From now on, we use the short-hand
j
n i
ij
p
e p p h
h
c
c
=
) , ,.. (
1
for the substitution effects.
BL MSc. Micro 20 45
3. The matrix of substitution effects S is negative semi-definite (NSD).
To interpret this condition, note that when n = 2, the matrix of substitution effects is
|
|
.
|

\
|
=
22 21
12 11
h h
h h
S
.
From the principal minor test for negative semi-definiteness, NSD of S requires:
(i) h
11
, h
22
s 0 (i.e. own-price substitution effects negative)
(ii) h
11
h
22
h
12
h
21
= h
11
h
22
- (h
12
)
2
> 0 (own-price substitution effects dominate cross-
price substitution effects)
Reason: as e(p
1
,..p
n
,u) is concave in p
1
,..p
n
, the n x n matrix of second derivatives of e
with respect to prices (the Hessian matrix of e) is negative semi-definite. But as shown
above, the second derivative e
ij
of e is equal to the substitution effect h
ij
.
BL MSc. Micro 20 46
Hicksian Substitutes and Complements
h
ij
< 0: goods i,j Hicksian complements, h
ij
>0: goods i,j Hicksian substitutes
Theorem. For any good i, h
ij
>0 for some ji; that is, every good is a Hicksian substitute
with at least one other good
Implies if n=2, goods 1,2 are Hicksian substitutes.
Proof. By property 3 of the expenditure function,
) , ,.. (
1
u p p e
n is HOD1 in prices. So
as
) , ,.. (
1
u p p h
n i is a derivative of
) , ,.. (
1
u p p e
n , by Shepards Lemma,
) , ,.. (
1
u p p h
n i is HODO in prices. But then using Eulers homogenous function
theorem, we get:
0 .. ..
2 2 1 1
= + + +
n in i ii i i
p h p h p h p h
But then, as h
ii
<0, and p
i
>0, all i, it must be that h
ij
>0, some i. QED.
BL MSc. Micro 20 47
The Slutsky Equation
x
2
x
1
A
0
B
OA old budget line with slope
2 1
/ p p
A

OB new budget line with slope

2 1
/ p p
B

CC- hypothetical budget line resulting from

increase in price from
A
p
1
to
B
p
1
with a utility-
compensating rise in m
C
C
Substitution
effect on good 1
Income effect on
good 1
BL MSc. Micro 2013 48
Using duality theory, we can derive a mathematical expression for this
decomposition.
By definition:
) , ,.. ( ) , ,.. (
1 1
m p p x u p p h
n i n i

, where
) , ,.. (
1
u p p e m
n

By substitution:
)) , ,.. ( , ,.. ( ) , ,.. (
1 1 1
u p p e p p x u p p h
n n i n i

As this is an identity, we can differentiate both sides with respect to p
j
to get:
j
i
j
i
j
i
p
e
m
x
p
x
p
h
c
c
c
c
+
c
c

c
c
(S1)
But now: j
j
h
p
e
=
c
c
by Shepards Lemma, and h
j
= x
j
by definition. So:
j
i
j
i
j
i
x
m
x
p
x
p
h
c
c
+
c
c

c
c
(S2)
BL MSc. Micro 2013 49
Rearranging, we get the Slutsky equation:
In the special case of j=i, we have:
income on substituti
x
m
x
p
h
p
x
j
i
j
i
j
i
c
c

c
c
=
c
c
( ) ( ?)
i i i
i
i i
x h x
x
p p m
substitution income
c c c
=
c c c

General implication: the law of
demand i.e. the demand curve
slopes down if the good is normal
BL MSc. Micro 2013 50
Utility maximization with Endowments
So far, we have treated income m as exogenous and independent of prices
But now suppose that

=
=
n
i
i i
e p m
1
, where e
1
,..e
n
are non-negative endowments
of goods (applications: labour supply, savings).
Marshallian demand for good i is
) .... , ,.. ( ) , ,.. (
2 2 1 1 1 1 n n n i n i
e p e p e p p p x m p p x + + =
Note that: (i) i
x
is HODO in p
1
,..p
n
, and m: (ii) m is HOD1 in p
1
, ..p
n.
So, i
x
is HODO in p
1
,..p
n
BL MSc. Micro 2013 51
Excess demand = Marshallian demand endowment:
i n i n i
e p p p x p p p z = ) ,.. , ( ) ,... , (
2 1 2 1 , i=1,..n
Example
) 1 , 5 . 0 ( = e ) , (
2 1
p p p =

2 1
5 . 0 p p m + =
,
o o
=
1
2 1
) ( ) ( x x u
So,
1
2 1
1
1
) 5 . 0 (
p
p p
p
m
x
+
= =
o o
and
2
2 1
2
2
) 5 . 0 )( 1 ( ) 1 (
p
p p
p
m
x
+
=

=
o o
So,
5 . 0
) 5 . 0 (
1
2 1
1 1 1

+
= =
p
p p
e x z
o
and
1
) 5 . 0 )( 1 (
2
2 1
2 2 2

+
= =
p
p p
e x z
o
///
General properties of excess demands:
1. HOD0 in prices (as
h
i
x
is HOD0 in p
1
,..p
n
)
2. Uniquely defined and continuous in prices (from axiom A5, strict convexity).
BL MSc. Micro 2013 52
The Slutsky Equation with Endowments
With endowments, Slutsky equation is modified:
+ +
i
m
n n n i n i
e e p e p e p p p x p p z ) .... , ,.. ( ) ,.. (
2 2 1 1 1 1

j
i
j
i
j j
i
j
i
j
i
Slutsky
j
i
j
i
j
i
j
i
j
i
z
m
x
p
h
e x
m
x
p
h
e
m
x
x
m
x
p
h
e
m
x
p
x
p
z
c
c

c
c
=
c
c

c
c
=
c
c
+
c
c

c
c
=
c
c
+
c
c
=
c
c
) (

So, now, if i=j
(?) ) ( income on substituti
z
m
x
p
h
p
z
i
i
i
i
i
i

c
c

c
c
=
c
c
Now, income effect depends on excess demand z
i
If z
i
<0, and good i is normal,
0 >
c
c

j
i
z
m
x
So, then, if the income effect is large enough,
0 >
c
c
i
i
p
z
General implication: the law of demand need no longer
hold when excess demands are negative, and income effect
is large
BL MSc. Micro 2013 53
Applications:
1. Labour supply. Endowment of leisure greater than consumption of leisure,
so excess demand for leisure is negative. So, the income effect of an
increase in the wage is to increase the demand for leisure. Thus, if the
income effect dominates the substitution effect, the labour demand curve
can be upward-sloping.
Consumption, C
Substitution effect: l*l
Income effect: ll*
Leisure
hours, l
24
l
l** l*
BL MSc. Micro 2013 54
Savings. A two-period consumption-savings model, with income (i.e. endowment
of the first period consumption good) only in the first period. Then, the
endowment of the first-period consumption good is greater than actual first-
period consumption (positive saving), so excess demand for first-period
consumption is negative. So, the income effect of an increase in relative price of
first-period consumption is to increase the demand for first-period consumption
i.e. reduce savings. But, the price of first-period consumption is 1+r. Thus, if the
income effect dominates the substitution effect, an increase in r can reduce
savings.
Substitution effect: c*c
Income effect: cc*
First-period
consumption
y
c
c** c*
Second-
period
consumption
y(1+r)
BL MSc. Micro 2013 55
The Life-Cycle Consumption-Savings Model
A household has income y
t
in periods t=1,..T
In each period, it must divide this income between consumption, c
t
, and
savings, s
t
.
Per period budget constraint: t t t t
y Rs s c + = +
1 , R=1+r, t=1,..T
Budget constraints are then
c +s = Rs
0
+ y
c +s = Rs +y
.
.
c
T
= Rs
T-1
+y
T
(wlog, we set s
T
=0)
Eliminate s ..,s
T-1
by substitution to get present-value budget constraint
BL MSc. Micro 2013 56
m y
R
y
R
y Rs c
R
c
R
c
T T T T
+ + + = + +
1 2 1 0 1 2 1
1
....
1 1
.....
1
So, prices of goods c
1
,c
2
,..c
T
are
1 2 1
1
,.....
1
, 1

= = =
T T
R
p
R
p p
Preferences over goods c
1
,c
2
,..c
T
are described by
) ,.... (
1 T
c c u
Much focus on special case of additive separability and exponential discounting
i.e.
) ( .... ) ( ) ( ) ,.... (
2 2 1 1 1 T T T
c u c u c u c c u + + =
(additive separability)
) ( .... ) ( ) (
1
2 1 T
T
c u c u c u

+ + = o o
(exponential discounting)
BL MSc. Micro 2013 57
Additive separability rules out habit formation i.e. past consumption levels do not
affect MRS between current and future consumption
Exponential discounting captures in a simple way preference for current over
future consumption (but strong assumption).
We can solve using Lagrangean technique
(

+ + + =

T T T T
T
c
R
c
R
c m c u c u c u L
1 2 1
1
2 1
1
.....
1
) ( .... ) ( ) ( o o
FOC are:
1
1
) ( '

=
t
t
t
R
c u

o
, t=1,..T
By eliminating

, FOC imply the consumption Euler equation:
BL MSc. Micro 2013 58
= =
+

) ( ' ) ( ) ( ' ) ( ) ( ' ) (
1
1 1
t
t
t
t
t
t
c u R c u R c u R o o o
CEE c u R c u
t t
) ( ' ) ( '
1 +
= o
The CEE implies that consumption is increasing/constant/decreasing over time
as R is greater than 1/equal to 1/less than 1.
Also, only present value of income matters for consumption.
Example
Assume log utility i.e.
c c u ln ) ( =
. Then possible to show that 1
1
.. 1
) (

+ +
=
T
t
t
m R
c
o o
o
(see
Q4, Problem Set 2). In special case 1 = o R , an increase in $1 in any period
leads to increase of $1/T in consumption in each period.
BL MSc. Micro 2013 59
Appendix: Proof of the Existence of a Unique Utility Number
Define A= {t0| (t,t) R x }, B= {t0| x R (t,t) }
By monotonicity (A4), if t eA, then t>t eA also. By monotonicity (A4), if t eB,
then t<t eB also. So, A, B are intervals.
By continuity (A3), A and B are both closed sets . (Why?)
So, A must be a closed interval of the form
) , [ t
and B must be a closed interval
of the form
] , 0 [ t
.
Now for any t0, completeness of R, (A1), implies that either (t,t) R x, or x R (t,t).
This means that for any t0, either t eA or t eB, or both. So,
t t s
, which implies
that AB=
] , [ t t
and this thus non-empty.
Finally, choose t* e AB. By the definition of A,B, we know (t*,t*) R x and x R
(t*, t*) i.e. (t*,t*) I x.
BL MSc. Micro 2013 60
So, t* is the candidate utility number i.e. t*=u(x) and consequently a utility
number u(x) exists for all x.
To establish the uniqueness of the Utility Number, the proof is by contradiction.
Suppose to the contrary that there are two candidate utility numbers u(x), v(x)
for some bundle x. Then
(u(x),u(x)) I x, (v(x),v(x)) I x
Then, as the indifference relation I is transitive (A2), we must have
(u(x),u(x)) I (v(x),v(x))
Suppose that u(x) v(x) e.g. u(x) > v(x). But then by strict monotonicity (A4) ,
(u(x),u(x)) P (v(x),v(x)), a contradiction. So, it must be that u(x)=v(x) i.e. the
utility number is unique. QED.
BL MSc. Micro 2013 61
Overview of Consumer Theory
Utility maximisation Expenditure Minimisation
Choose x
1
x
n
to
maximise u(x
1
,..x
n
)
subject to

=
=
n
i
i i
m x p
1
Marshallian demands
x
i
(p
1
,.p
n
, m)
Solve
Indirect utility function
v(p
1
,.p
n
, m)
Roys
identity
Substitute
into u
Choose x
1
x
n
to
minimise

=
=
n
i
i i
x p e
1
subject to u(x
1
,..x
n
) = u
subject to
Solve
Hicksian demands
h
i
(p
1
,.p
n
, u)
Shepards
Lemma
Substitute
into e
Expenditure function
e(p
1
,.p
n
, u)
inversion