Lecture Note 6 – Demand Functions: Income Effects,

Substitution Effects, and Labor Supply

David Autor, MIT Department of Economics

1
1 Demand Functions
Now, let’s use the Indirect Utility function and the Expenditure function to get Demand
functions. Up to now, we have been solving for:

• Utility as a function of prices and budget

• Expenditure as a function of prices and utility

Implicitly we have already found demand schedules—a demand schedule is immediately im-
plied by an individual utility function. For any utility function, we can solve for the quantity
demanded of each good as a function of its price, holding the price of all other goods constant
and holding either income or utility constant.

2 Two types of demand curves: Marshallian (uncompen-

sated) and Hicksian (compensated)
Alfred Marshall was the first economist to draw supply and demand curves. The ‘Marshallian
cross’ is the staple tool of blackboard economics. Marshallian demand curves correspond to
conventional market-level or consumer-level demand curves. They answer the question:

• Holding income and all other prices constant, how does the quantity of good X de-
manded change with px ? We notate this demand function as dx (px , py , I). Marshallian
demand curves implicitly combine income and substitution effects. The income effect
is the change in consumption that arises if the consumer’s income falls but if prices
stay the same. The substitution effect is the change in consumption that arises if the
prices change but the agent is given enough income to maintain the same utility they
had at the initial prices. They are net demands that sum over these two conceptually
distinct behavioral responses to price changes.

2
• One can also conceive of a demand curve that is composed solely of substitution ef-
fects. This is called Hicksian demand (after the economist J. R. Hicks) and it answers
the question: Holding consumer utility constant, how does the quantity of good X de-
manded change with px . We notate this demand function as hx (px , py , U ). The presence
of U as a parameter in the Hicksian demand function indicates that this function holds
consumer utility constant–on the same indifference curve–as prices change. Hicksian
demand is also called compensated demand. This name follows from the fact that to
keep the consumer on the same indifference curve as prices vary, one would have to
adjust the consumer’s income, i.e., compensate them. For the analogous reason, the
Marshallian demand is called uncompensated demand.

• A demand curve for x as a function of px

3
 • So a demand function is a set of tangency points between indifference curves and budget
set holding I and py (all other prices) constant.

Mathematics of uncompensated (‘Marshallian’) demand—Holding

income constant
• In our previous example where:

U (x, y) = x.5 y .5

we derived:

I
x(px , py , I) = 0.5
px
I
y(px , py , I) = 0.5
py

4
 • In general we will write these demand functions (for individuals) as:

x∗1 = d1 (p1 , p2 , ..., pn , I)

x∗2 = d2 (p1 , p2 , ..., pn , I)
...
x∗n = dn (p1 , p2 , ..., pn , I)

• We call this “Marshallian” demand after Alfred Marshall (who first drew demand
curves). You are also welcome to call it uncompensated demand.

2.1 Compensated (‘Hicksian’) demand—Holding utility constant

• This is called “Hicksian” or compensated demand after John Hicks. This demand
function takes utility as an argument, not income. This turns out to be an important
distinction.

5
Mathematics of compensated (‘Hicksian’) demand—Holding utility
constant
• We can write this mathematically using the dual problem to utility maximization, which
is expenditure minimization. Using the utility function from earlier examples, we had
previously derived that:
 .5
py
x(px , py , U ) = Up
px
 .5
px
y(px , py , U ) = Up
py

• These are Hicksian (compensated) demand function because they take prices and utility
as arguments rather than prices and income.

• We will denoted the Hicksian demand functions (of an individual) as:

x∗1,c = h1 (p1 , p2 , ..., pn , U )

x∗2,c = h2 (p1 , p2 , ..., pn , U )
...
x∗n,c = hn (p1 , p2 , ..., pn , U )

3 The effect of price changes on Marshallian (uncompen-

sated) demand
• A simple change in the consumer’s budget (i.e., an increase or decrease or I) involves
a parallel shift of the feasible consumption set inward or outward from the origin.
 The
economics of this are straightforward. Since this shift preserves the price ratio ppxy , it
 
typically has no effect on the consumer’s marginal rate of substitution (MRS), UUxy ,
unless the chosen bundle is either initially or ultimately at a corner solution.

• By contrast, a rise in the price of one good holding constant both income and the prices
of other goods has economically more complex effects:

1. It shifts the budget set inward toward the origin for the good whose price has
risen. In other words, the consumer is now effectively poorer. This component is
the ‘income effect.’

6
 2. It changes the slope of the budget set so that the consumer faces a different set of
market trade-offs. This component is the ‘price effect.’

• Although both shifts occur simultaneously, they are conceptually distinct and have
potentially different implications for consumer behavior.

Impact of an Increase in Px on the Budget Set

3.1 Substitution effect

• In a two good economy, what happens to consumption of X1 if

p1
↑
p2

while utility is held constant? Formally, we are asking for the sign of this derivative:
 
∂X1
Sign |U =U0 .
∂p1

∂X1
• Provided that the axiom of diminishing MRS applies, we’ll have |
∂p1 U =U0
< 0.

• In words, holding utility constant, the substitution effect is always negative.

7
3.2 Income effect
• Consider now the effect of a change in income on in consumption of good X. The
income effect is defined as ∂X/∂I, the change in the consumer’s quantity demanded of
X for a rise in income I.

• The income effect can be either negative or positive, depending on the specific good
and the consumer’s income level.

• If ∂X/∂I > 0 at given prices and income, good X is said to be a “normal” good. If
∂X/∂I < 0 at given prices and income, good X is said to be an “inferior” good. Inferior
goods can be further subdivided into two groups, as we’ll see below.

• What is the impact of an inward shift in the budget set in a 2-good economy (X1 , X2 )?

1. Total consumption? [Falls]

2. Utility? [Falls]
3. Consumption of X1 ? [Answer depends on normal, inferior]
4. Consumption of X2 ? [Answer depends on normal, inferior]

Comparison of Compensated (hx ) and Uncompensated (dx ) Demand Curves for a Normal
Good

8
Comparison of Compensated (hx ) and Uncompensated (dx ) Demand Curves for an Inferior
Good

4 Income, substitution, and labor supply

Although we typically think of demand functions as describing goods consumption, the same
reasoning applies to labor supply functions. And seeing this same logic through the labor
supply lens will deepen your understanding of the material. [Don’t worry: we’ll return to
demand for goods in the next lecture when we study Giffen goods.]
Consider the figure below, which depicts the Production Possibility Frontier of a subsis-
tence household—that is, a household that works only for itself. For simplicity, let’s assume
there’s only one household member, Zara. She can choose any bundle on the frontier of this
budget set (or in the interior—but she would not, by non-satiation). If Zara works 24 hours
per day, she maximizes consumption (point CC). If she sleeps or relaxes 24/7 (chillaxes as
they say in Singapore) her consumption is zero (point LL). The curvature of the PPF comes
from diminishing returns to home production (technically, this is a diminishing marginal rate
of technical substitution between production and leisure/sleep). Presumably the first hours
of home production in the day are highly productive (cooking, using the loo). Marginal

9
productivity falls as Zara does all of the essential stuff and becomes increasingly exhausted.1

Production Possibility Frontier of a Subsistence Household

Let’s say Zara chooses point A on this PPF, meaning that A corresponds to the point
where the PPF is tangent to the highest attainable indifference curve. Zara devotes time
LL − L0 to household production and consumes goods C0 . She spends hours L0 − 0 on leisure
and sleep (the complement of the time spent on household production).
1
This graphical depiction of the household time’s allocation trade-offs among sleep/leisure, home produc-
tion, market production, and consumption, was introduced in a 1977 Journal of Political Economy paper by
Reuben Gronau, “Leisure, Home Production, and Work – the Theory of the Allocation of Time Revisited.”
It’s a remarkably compact graphical representation of a sophisticated problem.

10
 Consumption-Leisure Bundle of Subsistence Household
Now let’s introduce a market wage, Wm . This will change the equilibrium in a complex
and interesting way. The wage Wm enters the figure as a slope that measures the additional
consumption attainable per hour of labor supply. Wm connects to the PPF at the point
of tangency between the wage slope and the PPF’s frontier. It thus expands the PPF by
allowing Time-Consumption bundles that were previously infeasible (beyond the frontier).
How does the market wage affect home production, labor supply, and consumption?
A first thing to notice is that it will never make sense for Zara to spend more than
LL−H1 hours on home production. Why? Because to the left of H1 , she gets more additional
consumption per hour by supplying labor to the market at wage Wm than she does through
home production. This is because the new wage-augmented PPF is steeper to the left of H1
than is the subsistence PPF.
How much does Zara work? Zara’s new utility maximizing bundle is at point B. Her
utility has risen to U1 , her home production has fallen to LL − H1 and her market work
has risen from zero to H1 − L1 . Market work has risen by more than total work because
home production has fallen. Zara’s hours of sleep and leisure have declined (L1 − 0 is shorter

11
than L0 − 0). Consumption has risen to C1 , with the length 0 − Ch corresponding to home-
produced goods and the length C1 −Ch corresponding to market-produced goods. (The figure
implicitly assumes that home- and market-produced goods are perfect substitutes; that is,
one unit of either is exactly as good as one unit of the other. If this were not the case, the
figure would need another dimension). Thus, Z is consuming more, working more, chillaxing
less, and happier for it (seen in her higher indifference curve).

Adding a Market Wage

Let’s now consider the impact the Earned Income Tax Credit. For simplicity, assume
initially that the EITC is exclusively an hourly wage subsidy that does not have the complex
phase-in and phase-out properties of the actual EITC; it just raises the hourly wage from
Wm to Wm + WE as depicted below.

12
 Adding an EITC Subsidy to the Market Wage
What effect does this have on home production, market work, consumption, and well-
being? You can see that home production unambiguously falls (from LL − H1 to LL −
H2 ). This is immediate because the point of tangency between the wage locus and the
PPF has moved southeast (down and to the right). Similarly, utility must rise: as long as
Zara is participating in the subsidy (that is, her chosen location on the PPF is along the
length containing Wm + WE ), she must be better off; otherwise she would not participate.
Graphically, utility has risen from point B on indifference curve U1 to point C on indifference
curve U2 . Notice also that consumption has risen to C2 , which exceeds consumption at the
previous location B. Although consumption of home produced goods fell, consumption of
market produced goods rose by more, leading to a net increase in consumption.
The most interesting observation here is how the wage subsidy affects labor supply. In-
tuitively, you might think that Zara’s labor supply would necessarily increase since her wage
has risen—and in the picture, it does. But the rise is modest (from H1 − L1 to H2 − L2 ).
This modest increase reflects the net impact of two countervailing effects. Holding income
constant, the rise in the wage would certainly cause Zara to work more; this is a pure substi-

13
tution effect. You can visualize this by rotating the ray Wm through point B until it parallels
the ray Wm + WE . The indifference curve tangent to that rotated ray would clearly lie north-
west of U1 , meaning more hours of labor supply, more consumption, and higher utility. To
reiterate, this notional change would correspond to the pure substitution effect of a higher
wage rate offered at the initial income level.
But that’s not precisely what’s happening here. If Zara made no adjustment in her hours
at all, her income/consumption would still rise when the EITC was implemented. Prior to the
EITC policy, she was working hours H1 − L1 in the figure above. When the EITC is enacted,
her income rises by the amount (H1 − L1 ) × WE with no change in labor supply. Thus, the
wage subsidy has two components from Zara’s perspective: (1) a transfer of income equal
to (H1 − L1 ) × WE ; and (2) an increase of WE in the hourly wage for each additional hour
worked. All else equal, this income transfer is likely to reduce labor supply because leisure is,
for most people, a normal good—the wealthier you get, the more of it you consume (holding
the wage constant). Because of these countervailing substitution and income effects, a wage
subsidy may either increase or reduce labor supply among those already working. Notice,
however, that for a person who is not initially working, the wage subsidy has no income effect,
but it does generate a potential substitution effect. A wage subsidy will therefore normally
cause some non-workers to enter the labor market. This observation is quite important for
the Eissa-Liebman paper, which we’ll be discussing next.
The next figure depicts a case where the wage subsidy is substantially larger, reflected in
a steeper slope for Wm + WEE as compared to Wm + WE in the figure above. Relative to
the case with no EITC subsidy (wage is Wm ), home production falls from the LL − H1 to
LL − H3 while market labor supply shifts from H1 − L1 to H3 − L3 . As drawn, the inward
shift from L1 to L3 is slightly larger than the outward shift from H1 to H3 , meaning that
market labor supply has fallen. Thus, the income effect of this large wage subsidy dominates
the substitution effect for Zara. The subsidy raises her welfare, increases her consumption,
increases her leisure and reduces her time spent on both home production and market work.
She’s definitely more chillaxed than she used to be.

14
 Adding a Generous EITC Subsidy to the Market Wage
The green line in the final figure depicts the complex budget set created by the EITC.
The first segment of the EITC budget set, starting from the PPF, is a subsidy to the hourly
wage. This is referred to as the phase-in. (Though in fact, the marginal hourly subsidy is
the same at each hour of work along this segment—so there’s really no phase-in, it’s all-at-
once). The second segment, which parallels the original wage Wm , is the plateau in which
the EITC neither increases nor decreases the marginal wage. Although there is no marginal
wage subsidy in this range, the worker receives a subsidy on hours worked during the phase-
in range; thus inframarginal but not marginal hours are subsidized. The final segment is
the phase-out range; here the marginal wage is below WM while the average wage is above
Wm . This means that each additional hour of work is taxed, but the average subsidy is still
positive because of the subsidy earned on the hours in the first segment (i.e. during the
phase in). At the point where the phase-out rejoins Wm , the EITC is fully phased out. An
individual who works at or above this level receives no subsidy at the margin or on average.
(The logic is that if a worker’s earnings are sufficiently high, he or she is not entitled to a
wage subsidy.)

15
 Summing up: in the phase-in range, the EITC has both income and substitution effects
(except for individuals who were not initially working—for them, there is no income effect).
We generally expect the EITC to increase labor supply in this region, though it’s possible
for the income effect to dominate, lowering labor supply. In the plateau range, the EITC
has only an income effect; it raises total earnings but does not raise earnings at the margin,
so we expect it to (modestly) decrease labor supply in this range. In the phase-out range,
where the EITC reduces the marginal wage below Wm , both the income and substitution
effects work in the direction of lowering labor supply. We thus expect the EITC to reduce
labor supply in this range.
How does Zara respond to the EITC? Her hours of home production fall from LL − H1
to LL − H4 while her hours of market work rise from H1 − L1 to H4 − L4 (both shift
right, but home production shifts rightward by more). The EITC permits Zara to choose
modestly higher consumption, modestly higher labor market hours, modestly higher leisure,
and modestly lower household production. She is, of course, better off.

Adding the Full EITC Budget Set to the Market Wage

As this figure underscores, the complexity of the EITC budget set implies that many

16
things can happen. Because the EITC reduces the marginal wage during the phase-out
range, it provides an incentive to reduce labor supply; many people will shift labor supply
until they are at the kink where the plateau meets the phase-out region. Others whose initial
market labor supply was quite low or zero will face a strong substitution effect but have little
or no income effect, which should increase labor supply. For a person like Zara whose hours
were already in the plateau range, there is an income effect but no substitution effect. Total
hours worked (home + market) are likely to decline due to the income effect. But market
hours may rise (as in this figure) because home production hours fall by more than total labor
supply (meaning market hours have risen). In all cases, individuals receiving the EITC are
made better off: whether they increase or decrease total work hours, market hours, and/or
leisure, those who receive the subsidy must necessarily obtain a bundle of consumption, work,
and leisure that is preferred to the initial starting point.

5 Bringing the theory to the data: the Tax Reform Act

of 1986 (TRA86)
The 1996 paper by Eissa and Liebman in the Quarterly Journal of Economics provides
a credible test of whether the EITC affects labor supply in practice. TRA86 yielded a
substantial expansion in the generosity of the EITC. As Eissa/Liebman note, at every level
of earnings, the EITC amount after the expansion was at least as large as it was before. We
would therefore expect it to increase the fraction of potentially eligible workers participating
in the labor force (due to the substitution effect). For eligible workers already participating in
the labor force, however, we would generally expect the EITC expansion to modestly reduce
hours (due to the income effect). We’ll discuss the results of the Eissa/Liebman study in
class.

17
MIT OpenCourseWare
https://ocw.mit.edu

14.03 / 14.003 Microeconomic Theory and Public Policy

Fall 2016

For information about citing these materials or our Terms of Use, visit: https://ocw.mit.edu/terms.