Microeconomics I. Antonio Zabalza.

University of Valencia 1

Lesson 6: Extensions and applications of

consumer theory

6.1 The approach of revealed preference

The basic result of consumer theory (discussed in

previous lesson) is that the own substitution effect is
non-positive.

This proposition can also be derived using another

approach, revealed preference, which proceeds
without any reference to indifference curves or
utility functions. Originator: Paul Samuelson, 1938.

x
Microeconomics I. Antonio Zabalza. University of Valencia 2

Suppose that at prices ( px , p y ) and income m,

bundle A is chosen. Suppose we have another bundle
B that can also be afforded by the consumer (that is,
lies on the bl or inside the bl). Can we say anything
about the preferences of this consumer respect these
two bundles? Answer: yes. If at these prices and
income, despite that both bundles were affordable,
A was chosen, it must be because the consumer
prefers A to B.

In mathematical terms
( x, y ) is the bundle chosen (A) when prices and
income are ( px , p y , m ) . ( x′, y′ ) is the bundle B,
which is affordable at these prices and income, but is
not chosen.
Bundle B is affordable: px x′ + p y y ′ ≤ m
Bundle A is chosen : px x + p y y = m

Substituting the second equation into the first,

px x + p yy ≥ p xx′ + p y y′

If this inequality holds, then we say that bundle A is

revealed preferred to bundle B.

In words: Bundle A is revealed preferred to bundle B

if bundle A is chosen when B could have been
chosen.
Microeconomics I. Antonio Zabalza. University of Valencia 3

The concept of revealed preference is not empty. It

excludes the possibility of some types of behaviour.
Supose bundles A and B in the following graph.

(p’, m)

(p, m)
B

If at ( p, m ) A is chosen, at ( p′, m ) B could not be

chosen. Why? Because if at the second set of prices
B was chosen, we would be saying that both A is
revealed preferred to B, and B is revealed preferred
to A, which is a contradiction.

The concept of revealed preference allows us to state

the following result:
Microeconomics I. Antonio Zabalza. University of Valencia 4

Weak Axiom of Revealed Preference (WARP)

If px x + p yy ≥ p xx′ + p y y′ ;
Then p′x x ′ + p′y y ′ < p′x x + p′y y .

That is, if at ( px , p y ) A is revealed preferred (rp) to

B, at ( p′x , p′y ) , if B is chosen, it must be because at
these latter prices A is not affordable.

A new look to the substitution effect

Suppose that at ( px , p y , m ) the bundle ( x, y ) is

chosen. Now suppose that prices change and we
compensate the consumer so that he can still afford
his old bundle; that is the budget data are now
( p′x , p y′ , m ′). What will the new equilibrium be?
y

R (x,y)

(x’,y’)
S

T x
Microeconomics I. Antonio Zabalza. University of Valencia 5

Can it be to the left of the initial equilibrium

(segment RS)? No. Bundle (x,y) was revealed
preferred to all bundles on that segment. Now, being
still bundle (x,y) affordable, no other bundle in that
segment can be chosen.

Can it be to the right of the initial equilibrium

(segment ST)? Yes. Bundle (x,y) was not revealed
preferred to the bundles on this segment , so they in
principle could now be revealed preferred to the
initial bundle and be chosen.

If this is so, it must mean that as a result of a change

in prices and income as the one depicted, a
compensated change in prices, the new bundle will
be such that necessarily more of the good that has
become relatively cheaper will be bought, and less
of the good that has become relatively more
expensive will be bought. This is the negative
substitution effect.

In maths

Call the new bundle (x’,y’). If the consumer has been

compensated so that he can afford the old bundle, it
must be that
p′x x′ + p′y y ′ = px′ x + p′y y (1)
If the new bundle is chosen, it must be because at
the new prices is revealed preferred to the old
Microeconomics I. Antonio Zabalza. University of Valencia 6

bundle, and that at the old prices if it was not chosen

it was because it was not affordable. So,
p x x′ + p y y ′ > p x x + p y y (2)
Subtract (1) from (2)
p
px x′ + p y y ′ − p′x x′ − p′y y ′ > px x + p y y − px′ x − p′y y
( px − p ′x ) ( x′ − x ) + ( p y − p ′y ) ( y ′ − y ) > 0
Suppose that the price of good y remains constant,
so that the second term in the sum is zero. Then
( px − p ′x ) ( x′ − x ) > 0
In the figure the first parenthesis is positive (the
price of x goes down). For the whole expression to
hold, the second parenthesis also has to be positive;
That is, the quantity of x must go up. This is the
substitution effect, without the need of indifference
curves or utility functions.

If px ↓ and the consumer is compensated so that

he is able to just afford his old bundle, then x ↑ .

6.2 Index numbers

One of the most immediate use of revealed

preference analysis is for the derivation of indices of
welfare change. If prices and income change,
normally the consumption bundle also changes. Can
Microeconomics I. Antonio Zabalza. University of Valencia 7

we say whether welfare has increased or decreased

as a result of this change? This is the sort of question
that the theory of index numbers tries to answer.

Definitions:
Base period: (p ,p )
b
x
b
y (x , y )
b b

Final period: (p ,p )
t
x
t
y (x , y )
t t

Quantity indices
Their objective is to measure how the consumption
of a basket of goods has evolved over time. They
measure the change in real consumption, holding
prices constant.

Goods are heterogeneous and therefore cannot be

compared directly. We must convert bundles of
goods into the value of these bundles. That is, to
weight each good by some price. In general the
structure that these indices (we will call them I q )
take is the following one:

wx x t + wy y t
Iq =
wx x b + w y y b

The important thing to notice is that the weights

remain constant in both periods; all that varies are
the quantities consumed. In general we will interpret
these indices in the following manner:
Microeconomics I. Antonio Zabalza. University of Valencia 8

If I q > 1, real consumption has increased.

If I q < 1, real consumption has decreased.

Choice of weights
If weights are base prices: Laspeyres quantity index,
Lq .
If weights are final prices: Paasche quantity index,
Pq .

pbx x t + p by y t
Lq =
pxb x b + p yb y b
pxt x t + pty y t
Pq =
ptx x b + pty yb

Inferences about changes in welfare

Say, Pq > 1 ⇒ pxt x t + pty y t > ptx xb + pty yb

Clearly, from the principle of revealed preference

(prp) we can conclude that
( xt , yt ) is rp to ( xb , yb ) .The consumer (or the
country) to which Pq refers, is better off at t than at
b.
Microeconomics I. Antonio Zabalza. University of Valencia 9

Pq < 1 ⇒ ptx xt + pty y t < p tx x b + pty yb

Here the prp cannot be applied because the bundle

( xb , yb ) is not affordable at the final year prices.
Lq < 1 ⇒ pxb x t + pby y t < pbx x b + pby y b

or pbx x b + pby y b > pxb xt + pyb y t

From the prp we conclude that

( xb , yb ) is rp to ( xt , yt ) . Therefore welfare has
decreased between b and t. The consumer (country)
is worse off.

Lq > 1 ⇒ pbx x b + pby y b < pxb xt + pyb y t

Here the prp cannot be applied since the final

bundle is not affordable at the base prices.

Price indices
The same can be applied to the measurement of the
evolution of prices. This may be more familiar to
you (recall IPC). Now what varies are the prices and
what stays put (the weights) are the quantities. As
Microeconomics I. Antonio Zabalza. University of Valencia 10

before there are two types of indices depending on

whether se use base or final year prices as weights.

ptx xb + pty yb
Lp =
pbx x b + pby y b
ptx x t + p ty y t
Pp =
pxb x t + pby y t

In the case of price indices, prices in the numerator

differ from tose in the denominator. Consequently,
the prp cannot be applied (see that for the prp to be
applicable, the set of prices at which you compare
two bundles has to be the same).

However, if you have additional information on how

expenditure has evolved over time (that is, on how
income has evolved), then something can be said.

Define a new index M:

ptx xt + pty y t
M=
pxb x b + p yb y b

See that M is just the ratio between expenditure

(income) in the final year and expenditure (income)
in the base year. With this information, there are two
cases in which we can say something definite:
Microeconomics I. Antonio Zabalza. University of Valencia 11

If Pp > M ⇒ Welfare has gone down

If Lp < M ⇒ Welfare has gone up

Proof that if Pp > M then welfare has gone down.

pxt x t + pyt y t pxt xt + pty y t

>
pbx xt + pby y t pxb xb + pby y b
(p xt t
x + pty y t )( pbx xb + pyb y b ) > ( pxb x t + p yb y t )( pxt x t + pyt y t )
pbx x b + p by y b > pbx x t + pby y t
∴ ( x , y ) is rp to ( x , y )
b b t t

Exercise: Proof that if L p < M then welfare has gone

up. Explain why in the other two cases,
Pp < M and L p > M , revealed preference cannot
help you to say whether welfare improves or
worsens.
Microeconomics I. Antonio Zabalza. University of Valencia 12

6.3 Budget constraints with fixed endowments

Suppose that instead of having income, the

consumer has a given endowment ( x , y ) of the two
goods. To fix ideas, think of a farmer or of a country
that produces oil. The budget constraint then is:

px x + p y y = p x x + p y y

Solving for y we have the formula for the new

budget line:

 px  px
y=y+ x− x
 
py  py

The slope is, as before, the relative price. The
vertical intercept however is a bit more complicated:
it depends on the endowments and on the price ratio.
Another important property of this budget line is that
always goes through point E: the endowment bundle
( x , y ) . (You should check this property by finding
the value of y when x = x ).

This property means that a change in the price ratio

will rotate the budget line around this point E. If x
becomes relatively cheaper, the bl will rotate
anticlockwise. If it becomes relatively more
expensive, the rotation will be clockwise.
Microeconomics I. Antonio Zabalza. University of Valencia 13

px
↑
py
 px 
 y + x
 p y 

E
y
px
↓
py

x  py  x
x + y
 px 
Of course, if the endowments change, with prices
constant, the bl will shift in a parallel fashion, as
with a conventional income effect. (Check that you
understand that).

Equilibrium

The equilibrium of the consumer need not be at

point E. If, for instance, the equilibrium is at the
right of point E, the consumer will be buying good x
(we will say he is a net buyer of x) and selling good
y (we will say that he is a net seller of y). And vice
versa if the equilibrium is at the left of point E. How
can the consumer buy x? With the resources he
obtains selling y. By selling y he gets resources by
the amount of py ( y − y ) . And with this resources,
Microeconomics I. Antonio Zabalza. University of Valencia 14

he spends px ( x − x ) on x. Will the former resources

be enough to pay for the latter expenditure? Or, in
other words, is this buying and selling consistent
with the budget constraint? Yes. See that the revenue
he obtains by selling y has to be equal to the
expenditure on x.
py ( y − y ) = px ( x − x )
But this is precisely the budget constraint, as you
can check yourself with a bit of manipulation of this
equation.
Response to price changes
Suppose the consumer is a net buyer of x,
( x − x ) > 0 , and that px ↓ (good x is normal). Will
he consume more or less x?
y

q’’

q
q’

x
x x’ x’’
 Microeconomics I. Antonio Zabalza. University of Valencia 15

See that in this case both the substitution effect

(from q to q’) and the income effect (from q’ to q’’)
work in the same direction and, therefore, the
consumer consumes more x.

Suppose now that for the same price change as

above, the consumer instead is a net seller of x,
( x − x ) < 0 . Will he consume also more x?
y

q
q’

q’’

x x x

Here we see that the substitution and the income

effect work in opposite directions. Depending on the
size of the income effect, the final equilibrium q’’
could be to the right of the initial equilibrium (more
Microeconomics I. Antonio Zabalza. University of Valencia 16

x consumed) or to the left (less x consumed). So in

this case it is not possible to say anything definite.

Why do we find these different results depending on

whether the consumer is a net buyer or a net seller of
good x? It must have to do with the income effect,
because the substitution effect works in the same
direction in both cases. Remember that the price of x
goes down. If you are a net buyer of x, this price fall
means a positive income effect (you are richer
because of this price fall). However, if you are a net
seller of x, this price falls makes you poorer, and
therefore the income effect in this case is negative.
This could never happen in the previous lesson in
which income is not in terms of physical
endowments but in terms of money.
Microeconomics I. Antonio Zabalza. University of Valencia 17

Switch from net buyer to net seller and vice versa

Even without knowing anything on preferences we

can say quite a few things about the behaviour of the
consumer concerning his desired position as seller or
buyer by using the principle of revealed preference
(prp).

Net buyer of x ( x − x ) > 0; px ↓

C
E

O B D

By the prp the new equilibrium must be on segment

ED. The bundle q is revealed preferred (rp) to all the
set of opportunities OAB; so, since CE and q are still
affordable at the new prices, CE will not be chosen.
If the price of a good goes down, a net buyer of this
good will remain a net buyer.
Microeconomics I. Antonio Zabalza. University of Valencia 18

Net buyer of x ( x − x ) > 0; px ↑

A
E

O D B

The prp cannot say anything on the new equilibrium.

The bundle q is rp to all the set of opportunities
OAB. But at the new prices q is no longer
affordable, so we cannot say what bundles in new bl
will be or not preferable to q (remember that to be
able to make a comparison both bundles have to be
affordable).

Exercise
Equipped with the above reasoning show (with the
corresponding graphs) that for a net seller:
a) If the price of x goes down the prp cannot help
you to tell whether or not the consumer will
remain a net seller or not.
Microeconomics I. Antonio Zabalza. University of Valencia 19

b) If the price of x goes up, a net seller of x will

remain a net seller of x.

6.4 Labour supply

A particular application of a budget constraint with

endowments is the analysis of labour supply.

We measure labour supply by H (hours of work per

period of time; say, per day). Clearly there is a
maximum of hours per day that we could work,
namely 24. This is our endowment of time. The
other good is C, consumption; the more we work,
the more we can consume.

Considering H and C would be like considering a

“bad” and a “good”. To make the analysis more
similar to the previous exposition of consumer
theory and work with two “goods”, instead of
considering “hours of work”, we will consider
“hours of leisure” and call it L. The endowment of
leisure is the maximum amount of leisure that we
could conceivably have (24). Call this endowment
L . Then the hours that an individual works is this
maximum amount minus the actual leisure that he
consumes,
H = L−L
Microeconomics I. Antonio Zabalza. University of Valencia 20

If out of the 24 hours possible, I decide to consume

16 hours of leisure (8 to sleep, 3 to eat and 5 to
enjoy myself), it means that I work 8 hours (=24-
16). The utility function is then defined as

U = U (C , L )

wL w
P slope =
P

wH
C=
P

L L L

To build the budget constraint consider the

following: If I work H hours (equivalent to having L
hours of leisure; L = L − H ) and the hourly wage is
w, then my income will be wH. With income wH, if
the price of C, consumption, is P, I will be able to
spend on consumption
Microeconomics I. Antonio Zabalza. University of Valencia 21

wH=CP (1)

which means that the real units of consumption will

be

wH
C=
P
In the extreme, suppose I work all day H = L , then
L=0 and

wL
C=
P
This gives the vertical intercept of the budget
constraint.

So, the slope of the budget constraint is the vertical

intercept divided by the horizontal intercept. That is,

wL P w
Slope = =
L P

The slope of the budget constraint is the real wage.

By now it should be easy to derive the formula for

the budget line. From (1) we have
Microeconomics I. Antonio Zabalza. University of Valencia 22

w ( L − L ) = CP
CP + wL = wL

This is the budget line: the “goods” are C and L and

the endowment of leisure is L . Solving for C we
have

wL w
C= − L
P P

A straight line with vertical intercept wL P and

slope − w P .

Response to an increase in the wage

An increase in w will rotate clockwise the bl around

the horizontal intercept (check you understand this).
The substitution effect is as always: the higher the
wage, the less leisure is consumed (the more hours
of work are supplied). This is the movement from A
to B. But see that there is also a positive income
effect: we are all net sellers of time (of leisure); if w,
the price of time goes up, we are richer. Where is the
final equilibrium? Here we have a problem, because
in this case the two effects work in opposite
directions. If we suppose that leisure is a normal
good (as we should), then the final equilibrium, D,
can be to the right of A or to the left of A. Normally
it will be to the left of A (we will work more, when
Microeconomics I. Antonio Zabalza. University of Valencia 23

w goes up). But this is not an unambiguous

conclusion. We might feel so rich as a result of the
increase in w that we may decide to work less (in
that case the final equilibrium would be to the right
of A). Empirically, this tends to happen for people
who already work a lot and therefore have, even
before the increase of the wage, quite a lot of
income.

w1L
P

w0 L D
P
B

L L

In the graph we have depicted the more frequent

case: an increase in w results in an increase in hours
of work. But see that there is no reason why we
could not draw an indifference map such that point
D was at the right of point A.
Microeconomics I. Antonio Zabalza. University of Valencia 24

Exercise to do in the practical class

Suppose that in addition to an endowment of leisure

of L you also have a given endowment of non-
labour income M . That is, you also have some rent
(money) per period of time that does not depend on
the hours you work. What is the budget constraint in
the (C,L) space?

6.5 Consumption over time

So far we have analysed the consumption of goods

in the present moment of time. Now, in the last
application of consumer theory, we analyse the
consumption of one good in different periods of
time.

To be more precise. Suppose we have a composite

good, consumption, which we denote as above by C.
We want to see what will be the amount of
consumption now, C1 , and the amount of
consumption in some future period of time
(tomorrow, next week, next year), C2 . That is, the
consumer’s utility function is

U = U ( C1 , C2 )
Microeconomics I. Antonio Zabalza. University of Valencia 25

Suppose, to simplify things, that prices are 1 in both

periods. At the end of this section we will relax this
assumption. Suppose also that the consumer has a
source of income in each period. Income in period 1
is m1 and income in period 2 is m2 . m1 and m2 are
the endowments the consumer has.

The last assumption we make is that the consumer

can lend and borrow money at a rate of interest of r.

Budget constraint

Maximum amount of consumption in period 2

(vertical intercept).

C2 ( max ) = m2 + m1 (1 + r )

m2 is the money the consumer has in period 2. If in

period 1 has saved some money and has lent it, in
period 2 he will have m1 (1 + r ) .

Maximum amount of consumption in period 1

(horizontal intercept).

m2
C1 (max) = m1 +
(1 + r )

Here the difficult thing to understand is the second

term. This term is the amount of money I could
Microeconomics I. Antonio Zabalza. University of Valencia 26

borrow now on the expectation that in period 2 my

income will be m2 . Why? If today I borrow x, next
period I will have to return x(1+r). To pay this, next
period I will have m2 . So, the x I can borrow now, so
that next period I will be able to repay with my
income then, is:

x(1 + r ) = m2
m2
x=
(1 + r )

C2

m2 + m1(1 + r) slope = −(1 + r )

C2 A

E
m2
B

C1 m1 m2 C1
m1 +
(1 + r )
Microeconomics I. Antonio Zabalza. University of Valencia 27

The slope, as usual is (minus) the vertical intercept

divided by the horizontal intercept. Check that this
gives –(1+r).

The rate of interest, r, is the relative price in this

case: is the price of present consumption relative to
future consumption. If r goes up, present
consumption becomes relatively more expensive and
the bl rotates clockwise around E. If r goes down,
present consumption becomes relatively cheaper, so
the bl rotates anticlockwise around E. Remember
that the endowment is always available: whatever
the change in prices, E is always affordable (in this
and in all other applications).

Formula for the budget line

To derive the formula for the bl, suppose the

equilibrium is at point A. The consumer saves
(m1 − C1 ) . This means that the future consumption he
will be able to have is

C2 = m2 + (m1 − C1 )(1 + r )
Or,

C2 = [ m2 + m1 (1 + r ) ] − (1 + r )C1
Microeconomics I. Antonio Zabalza. University of Valencia 28

A straight line with vertical intercept [ m2 + m1 (1 + r ) ]

and slope –(1+r).

Exercise to do at the practical class

Suppose the consumer’s equilibrium is at point B
(that is, he is a borrower). Derive the formula for the
budget line. It should, of course, be the same as the
one derived for a saver.
Another (important) way of expressing the budget
line

If in the previous formula of the bl you multiply

both sides of the equation by (1+r) and rearrange the
terms you have

C2 m2
C1 + = m1 +
(1 + r ) (1 + r )

This form for the budget line says that the present
value of consumption must be equal to the present
value of income. Why is the expression on the left
of the equality sign the present value of
consumption, and why the expression on the right is
the present value of income?
Microeconomics I. Antonio Zabalza. University of Valencia 29

Equilibrium

The equilibrium position is found as usual: point of

tangency between bl and ic. The problem here is that
the ic is not between two different goods, but
between the same good at two different time periods.
So, we must first interpret the meaning of the slope
of this ic.

C2

∆C2

∆C1

45º

C1

If at the intersection with the 45º line the slope of the

ic was –1, it would mean that I value equally present
and future consumption. This in fact is not like this.
Normally we give more value to present than to
future consumption. Why? Think “What would I
prefer €1,000 now or €1,500 in five years time?”
(Difficult choice? Well, think that in five years time
I may be dead).
Microeconomics I. Antonio Zabalza. University of Valencia 30

Normally slope should be larger than 1 in absolute

terms, meaning that people value more present than
future consumption.

∆C2  ∆C2   ∆C2 − ∆C1 

slope = − = − 1 + − 1 = − 1 + 
∆C1  ∆ C1   ∆ C1 
The term ( ∆C2 − ∆C1 ) ∆C1 is the extra
compensation I need now to give up the
consumption of present income. Call it θ , the
subjective rate of time preference. Then, the slope of
the ic is

slope = − (1 + θ )

Recalling that the slope of the bl is –(1+r), the

equilibrium position, that is the tangency between
the ic and the bl, is a point where the market rate of
interest equals the subjective rate of time preference.
r =θ
Microeconomics I. Antonio Zabalza. University of Valencia 31

C2

Equilibrium point; r = θ

C1

Changes in equilibrium

We suppose both C1 and C2 are normal goods.

Again, it is an application of revealed preference in
the context of changing endowments.

A. First suppose the consumer is a lender

( m1 − C1 ) > 0 .
A.1 Increase in the rate of interest.

In that case, by the prp, the consumer will continue

to be a lender. Do you agree with that?
Microeconomics I. Antonio Zabalza. University of Valencia 32

Concerning the effect on consumption in the two

periods, the result is given by the following table:

Effects C1 C2

Substitution - +
Income + +

Total ? +

C2

C
q’’

q’

C1
D B
Microeconomics I. Antonio Zabalza. University of Valencia 33

A.2 Fall in interest rate

In this case, the prp cannot help us to predict the

final segment of the budget constraint where the
consumer will be. So, in this case, we must conclude
that the consumer could switch and become a
borrower.

Regarding the qualitative change in the consumption

levels of present and future consumption, the
relevant table is:

Effects C1 C2

Substitution + -
Income - -

Total ? -

C2
Microeconomics I. Antonio Zabalza. University of Valencia 34

A
q

C q’

E
q’’

C1
B D

Exercise: Complete the analysis when the consumer

is a borrower and for the cases in which the rate of
interest: a) increases; and b) decreases. Show in
particular that in the first case the consumer could
switch to being a lender, and that in the second case
we can assert definitely that the consumer will
remain a borrower.
Microeconomics I. Antonio Zabalza. University of Valencia 35

Inflation

So far we have considered that the price of

consumption in the two periods is the same: This in
index terms is equivalent to saying that the price in
both periods is equal to 1: p1 = 1 and p2 = 1. In terms
of the budget constraint, if we want to consider these
prices explicitly, the equation is

p 2C2 pm
p1C1 + = p1m1 + 2 2
1+ r 1+ r

Suppose now that there is inflation and that p2 > p1 .

(If, say, p1 = 1 and p2 = 1.10 , it would mean that
between period 1 and period 2, prices have increased
10%). Then the budget constraint is:

C2 m
C1 + p2 = m1 + p2 2
1+ r 1+ r
 1+ r  1+ r
C2 =  m2 + m1  − C1
 p2  p2

That is, the slope of the budget line, the relative

price of present consumption changes from (1 + r ) to
(1 + r ) p2 . Inflation, therefore, it is like if the
discounting factor had changed.
Microeconomics I. Antonio Zabalza. University of Valencia 36

Let us relate this new discounting factor more

explicitly to inflation. Call π the rate of inflation, so
that

1+ r 1+ r
=
p2 1 + π

In our example, π = 0.10 , or a 10% rate of inflation.

Then, let us call (1+ρ) this new discounting factor.

1+ r
=1+ ρ
1+π

ρ is the real rate of interest. Approximately is equal

to the nominal rate of interest minus the rate of
inflation.

ρ ; r −π
So, if the nominal rate of interest is 15% and the rate
of inflation is 10%, the real rate of interest is 5%.
So, when there is inflation, the budget constraint is

C2 =  m2 + (1 + ρ ) m1  − (1 + ρ ) C2

The discounting factor is then defined with the real

rate of interest.