Economics 217

Chapter 5: “Income and Substitution Effects”

Nadine Yamout
American University of Beirut

Spring 2022-2023

1
Class Outline

1. Demand Functions
2. Changes in Income
3. Changes in a Good’s Price
4. The Individual’s Demand Curve
5. Compensated (Hicksian) Demand Curves and Functions
6. Mathematical Development of Response to Price Changes
7. Class Summary

Readings: Nicholson and Snyder, Chapter 5.1-5.6

2
Demand Functions
Demand Functions

• It is possible to solve the necessary conditions of a utility maximum

for the optimal levels of x1 , x2 , . . . , xn as functions of all prices and
income.
• This can be expressed as n demand functions of the form:

x1∗ = x1 (p1 , p2 , . . . , pn , I )
x2∗ = x2 (p1 , p2 , . . . , pn , I )
..
.
xn∗ = xn (p1 , p2 , . . . , pn , I )
• If there are only two goods, x and y, notation can be simplified as:

x∗ = x(px , py , I )
∗
y = y (px , py , I )
• The notation stresses that prices and income are “exogenous” to this
process; that is, these are parameters over which the individual has
no control at this stage of the analysis. 3
Homogeneity

• If we were to double all prices and income, then the optimal quantities
demanded would remain unchanged.
• Doubling all prices and income changes only the units by which we
count, not the “real” quantity of goods demanded.
• Doubling px , py , and I does not affect the budget constraint since
px x + py y = I is the same as 2px x + 2py y = 2I . Hence x ∗ , y ∗ will
still be same combination chosen.
• We can write this result as saying that, for any good xi,

xi∗ = xi (p1 , p2 , . . . , pn , I ) = xi (tp1 , tp2 , . . . , tpn , tI )

• Individual demand functions are homogeneous of degree zero in all
prices and income.
• This result shows that (in theory) individuals’ demands will not be
affected by a pure inflation during which all prices and incomes
increase proportionally.
4
Homogeneity: Examples

• With a Cobb-Douglas utility function

utility = U(x, y ) = x 0.3 y 0.7

the demand functions are:
0.3I 0.7I
x∗ = y∗ =
px py
Note that a doubling of both prices and income would leave x ∗ and
y ∗ unaffected.

• With a CES utility function

utility = U(x, y ) = x 0.5 + y 0.5

the demand functions are:
1 I 1 I
x∗ = · y∗ = ·
1 + px /py px 1 + py /px py
Note again that a doubling of both prices and income would leave x ∗
and y ∗ unaffected. 5
Changes in Income
Changes in Income

• As a person’s purchasing power increases, it is natural to expect that

the quantity of each good purchased will also increase.

• An increase in income will cause the budget constraint out in a parallel

fashion.

• Since px /py does not change, the MRS will stay constant as the
worker moves to higher levels of satisfaction.

6
Increase in Income and Normal Goods

• As expenditures increase from I1 to I2 to I3 , the quantity of x

demanded increases from x1 to x2 to x3 . Also, the quantity of y
increases from y1 to y2 to y3 .
• Both x and y increase as income increases, i.e. ∂x/∂I > 0 and
∂y /∂I > 0 .
• Goods that have this property are called normal goods.

7
Increase in Income and Inferior Goods

• For increases in income in the range shown, less of z is chosen.

• z decreases as income increases, i.e. ∂z/∂I < 0 .
• Goods that have this property are called inferior goods.

8
Changes in a Good’s Price
Changes in a Good’s Price

• A change in the price of a good involves changing not only one of the
intercepts of the budget constraint but also its slope.

• Consequently, moving to the new utility-maximizing choice entails not

only moving to another indifference curve but also changing the MRS.

• When a price changes, two analytically different effects come into

play:

1. substitution effect: even if the individual were to stay on the same

indifference curve, consumption patterns would be allocated so as to
equate the MRS to the new price ratio.

2. income effect: arises because a price change necessarily changes an

individual’s “real” income and therefore he must move to a new
indifference curve

9
Graphical Analysis of Decrease in Price

10
Graphical Analysis of a Decrease in Price

• This individual is initially maximizing utility by consuming the

combination x ∗ , y ∗ .
• If the price of x decreases to px2 , the new budget constraint is given
by the equation I = px2 x + py y .
• The new position of maximum utility is at x ∗∗ , y ∗∗ , where the new
budget line is tangent to the indifference curve U2 .
• The movement to this new point can be viewed as being composed
of two effects. First, the change in the slope of the budget constraint
would have motivated a move to point B, even if choices had
been confined to those on the original indifference curve U1 . This
movement is a graphic demonstration of the substitution effect.
• The additional move from B to the optimal point x ∗∗ , y ∗∗ is
analytically identical to changes in income. Because the price of x
has decreased, this person has a greater real income and can afford
a utility level U2 that is greater than that which could previously be
11
attained. This is a income effect.
Graphical Analysis of an Increase in Price

12
Graphical Analysis of an Increase in Price

• This individual is initially maximizing utility by consuming the

combination x ∗ , y ∗ .
• If the price of x increases to px2 , the new budget constraint is given
by the equation I = px2 x + py y .
• The new position of maximum utility is at x ∗∗ , y ∗∗ , where the new
budget line is tangent to the indifference curve U1 .
• The movement to this new point can be viewed as being composed
of two effects. First, even if this person could stay on the initial
indifference curve U2 , there would still be an incentive to substitute
y for x and move along U2 to point B. This movement is a graphic
demonstration of the substitution effect.
• However, because purchasing power has been reduced by the increase
in the price of x, he or she must move to a lower level of utility. This
movement is again called the income effect.
13
Price Changes for Normal Goods

• Notice that in the examples we considered, both the income and

substitution effect work in the same direction and cause the quantity
of x demanded to be reduced in response to an increase in its price.

• If a good is normal, substitution and income effects reinforce one

another.

• When price falls, both effects lead to a rise in quantity demanded.

• When price rises, both effects lead to a drop in quantity demanded.

14
Price Changes for Inferior Goods

• If a good is inferior, substitution and income effects move in opposite

directions
• The combined effect is indeterminate:
• when price rises, the substitution effect leads to a drop in quantity
demanded, but the income effect is opposite
• when price falls, the substitution effect leads to a rise in quantity
demanded, but the income effect is opposite
• If the income effect of a price change is strong enough, the change
in price and the resulting change in the quantity demanded could
actually move in the same direction.
• Legend has it that the English economist Robert Giffen observed this
paradox in nineteenth-century Ireland: When the price of potatoes
rose, people reportedly consumed more of them.
• Giffen’s paradox: The possibility of an increase in the quantity
demanded in response to an increase in the price of a good.
15
The Individual’s Demand Curve
The Individual’s Demand Curve

• An individual’s demand for x depends on preferences, all prices, and

income:
x ∗ = x(px , py , I )
• It may be convenient to graph the individual’s demand for x assuming
that income and the price of y (py ) are held constant:

x ∗ = x(px , p¯y , I¯)

where the bars over py and I indicate that these determinants of

demand are being held constant.

• Individual demand curve: shows the relationship between the price

of a good and the quantity of that good purchased by an individual,
assuming that all other determinants of demand are held constant.

16
Construction of an Individual’s Demand Curve

17
Shifts in the Demand Curve

• Three factors were held constant in deriving this demand curve:

1. income
2. prices of other goods (py )
3. the individual’s preferences
• If any of these were to change, the entire demand curve might shift
to a new position.
• If I were to increase, the curve would shift outward (provided that
∂x/∂I > 0). More x would be demanded at each price.
• If another price (say, py ) were to change, then the curve would shift
inward or outward, depending precisely on how x and y are related.
• The demand curve would shift if the individual’s preferences for good
x were to change.
• It is important to keep clearly in mind the difference between a
movement along a given demand curve caused by a change in px
and a shift in the entire curve caused by a change in income, in one
18
of the other prices, or in preferences.
Compensated (Hicksian)
Demand Curves and Functions
Compensated Demand Curves

• The actual level of utility varies along the demand curve. The reason
this happens is that the demand curve is drawn on the assumption that
nominal income and other prices are held constant; hence a decline in
px makes this person better off by increasing his or her real purchasing
power.

• Although this is the most common way to impose the ceteris paribus
assumption indeveloping a demand curve, it is not the only way.

• An alternative approach holds real income (or utility) constant while

examining reactions to changes in px .

• Here, the effects of the price change are “compensated” so as to

constrain the individual to remain on the same indifference curve and
so reactions to price changes include only substitution effects.

19
Construction of a Compensated Demand Curve

20
Compensated Demand Curves

• A compensated demand curve shows the relationship between the

price of a good and the quantity purchased on the assumption that
other prices and utility are held constant.
• Therefore, the curve (which is sometimes termed a Hicksian demand
curve after the British economist John Hicks) illustrates only
substitution effects.
• Mathematically, the curve is a two-dimensional representation of the
compensated demand function:
x c = c x (px , py , U)
• Notice that the only difference between the compensated demand
function and the uncompensated demand functions is whether utility
or income enters the functions. Hence the major difference between
compensated and uncompensated demand curves is whether utility or
income is held constant in constructing the curves.
21
Compensated and Uncompensated Demand Curves

22
Compensated and Uncompensated Demand Curves

• At p ′′ x the curves intersect because at that price the individual’s

income is just sufficient to attain utility level Ū. Hence x ′′ is demanded
under either demand concept.
• For prices below p ′′ x , however, the individual suffers a compensating
reduction in income on the curve x c that prevents an increase in
utility arising from the lower price. Assuming x is a normal good,
it follows that less x is demanded at p ′′′ x along x c than along the
uncompensated curve x.
• Alternatively, for a price above p ′′ x , income compensation is positive
because the individual needs some help to remain on the same utility.
Again, assuming x is a normal good, at px′ more x is demanded along
x c than along x.
• For a normal good the compensated demand curve is somewhat less
responsive to price changes than is the uncompensated curve. This
is because the latter reflects both substitution and income effects,
whereas the compensated curve reflects only substitution effects. 23
Shepard’s Lemma

• Consider the dual expenditure minimization problem. The Lagrangian

expression for this problem:
L = px x + py y + λ U(x, y ) − Ū
 

• The solution to this problem yields the expenditure function

E (px , py , U). Because this is a value function, the envelope
theorem applies. This means that we can interpret derivatives of
the expenditure function by differentiating the original Lagrangian
expression that produced it.
• Differentiation with respect to the price of good x, for example, yields:
∂E (px , py , U) ∂L
= = x c (px , py , U)
∂px ∂px

• So, the compensated demand function for a good can be found from
the expenditure function by differentiation with respect to that good’s 24
Shepard’s Lemma

• One of the many insights that can be derived from Shephard’s lemma
concerns the slope of the compensated demand curve. Since the
expenditure function must be concave in prices then:

∂x c (px , py , U) ∂ [∂E (px , py , U)/∂px ] ∂ 2 E (px , py , U)

= = <0
∂px ∂px ∂px2

• Hence the compensated demand curve must have a negative slope.

• The ambiguity that arises when substitution and income effects work
in opposite directions for Marshallian demand curves does not arise
in the case of compensated demand curves because they involve only
substitution effects.

25
Example: Compensated Demand Functions

• Consider the utility function given by:

utility = U(x, y ) = x 0.5 y 0.5

• The Marshallian demand functions are:
I I
x= y=
2px 2py
• The indirect utility function and expenditure function are:
I
V (I , px , py ) = E (px , py , U) = 2px0.5 py0.5 U
2px0.5 py0.5
• We can now use Shephard’s lemma to calculate the compensated
demand functions as:
∂E (px , py , U)
x c (px , py , U) = = px−0.5 py0.5 U
∂px
∂E (px , py , U)
y c (px , py , U) = = px0.5 py−0.5 U
∂py
26
Mathematical Development of
Response to Price Changes
Direct Approach

• Our goal is to use the utility-maximization model to learn something

about how the demand for good x changes when px changes; that is,
we wish to calculate ∂x/∂px .
• We could approach this problem using comparative static methods
by differentiating the three first-order conditions for a maximum with
respect to px . This would yield three new equations containing the
partial derivative we seek. These could then be solved using matrix
algebra and Cramer’s rule.
• Unfortunately, obtaining this solution is cumbersome and the steps
required yield little in the way of economic insights.
• Hence we will instead adopt an indirect approach that relies on the
concept of duality. In the end, both approaches yield the same
conclusion, but the indirect approach is much richer in terms of the
economics it contains.
27
Indirect Approach

• We will assume there are only two goods (x and y ) and focus on the
compensated demand function, x c (px , py , U), and its relationship to
the ordinary demand function, x(px , py , I ).
• By definition we know that the quantity demanded is identical for the
compensated and uncompensated demand functions when income is
exactly what is needed to attain the required utility level:

x c (px , py , U) = x [px py , E (px , py , U)]

• We can now differentiate the compensated demand function with

respect to px :
∂x c ∂x ∂x ∂E
= + ·
∂px ∂px ∂E ∂px
• Rearranging terms yields:
∂x ∂x c ∂x ∂E
= − ·
∂px ∂px ∂E ∂px

28
Substitution and Income Effects

∂x ∂x c ∂x ∂E
= − ·
∂px ∂px ∂E ∂px

• The first term is the slope of the compensated demand curve. But
that slope represents movement along a single indifference curve; it
is, in fact, what we called the substitution effect earlier.

• The second term reflects the way in which changes in px affect the
demand for x through changes in purchasing power. Therefore, this
term reflects the income effect. The negative sign reflects the inverse
relationship between changes in prices and changes in purchasing
power.

29
The Slutsky Equation

• The substitution effect can be written as:

∂x c

∂x 
substitution effect = =
∂px ∂px U= constant

• The income effect can be written as:

∂x ∂E ∂x ∂E ∂x
income effect = − · =− · = −x c
∂E ∂px ∂I ∂px ∂I

• The Slutsky equation is then:


∂x ∂x  ∂x
= substitution effect + income effect = −x
∂px ∂px U= constant ∂I
where we have made use of the fact that x(px , py , I ) = x c (px , py , V )
at the utility-maximizing point.
30
The Slutsky Equation


∂x ∂x  ∂x
= −x
∂px ∂px U= constant ∂I

• The first term is the substitution effect. It is always negative as

long as MRS is diminishing since we showed that the slope of the
compensated demand curve must be negative.

• The second term is the income effect and its sign depends on the sign
of ∂x/∂I :
• if x is a normal good, then ∂x/∂I > 0 and the entire income effect is
negative
• if x is an inferior good, then ∂x/∂I < 0 and the entire income effect
is positive

31
Example: Slutsky Decomposition

• For the Cobb-Douglas Utility function U(x, y ) = x 0.5 y 0.5 , the

Marshallian and Hicksian demand functions are:
0.5I
x(px , py , I ) = and x c (px , py , U) = px−0.5 py0.5 U
px

• The total effect of a price change on Marshallian demand can be found

by differentiating the Marshallian demand function:

∂x(px , py , I ) −0.5I
=
∂px px2

• We wish to show that this is the sum of the two effects that Slutsky
identified.

32
Example: Slutsky Decomposition

• The substitution effect is found by differentiating the compensated

demand function:
∂x c (px , py , U)
substitution effect = = −0.5px−1.5 py0.5 U
∂px

• Now in place of U, we can use the indirect utility function

V (px , py , I ) = 0.5Ipx−0.5 py−0.5 :
−0.25I
substitution effect = −0.5px−1.5 py0.5 V =
px2
• The income effect can be computed as:
 
∂x 0.5I 0.5 −0.25I
income effect = − =− · =
∂I px px px2

• The sum of the substitution and income effects is exactly ∂x/∂px :

−0.25I −0.25I −0.5I ∂x
substitution effect + income effect = + = =
px2 px2 px2 ∂px33
Class Summary
Class Summary

• Proportional changes in all prices and income do not shift the

individual’s budget constraint and therefore do not change the
quantities of goods chosen. In formal terms, demand functions are
homogeneous of degree 0 in all prices and income.
• For normal goods, an increase in purchasing power causes more to
be chosen. In the case of inferior goods, however, an increase in
purchasing power causes less to be purchased.
• A change in the price of a good causes substitution and income effects
that, for a normal good move in the same direction. For inferior goods,
however, substitution and income effects work in opposite directions,
and no unambiguous prediction is possible.
• Compensated (or Hicksian) demand functions show how quantities
demanded are functions of all prices and utility. The compensated
demand function for a good can be generated by partially
differentiating the expenditure function with respect to that good’s
price (Shephard’s lemma). 34