C H A P T E R

19
Distortionary Taxes and Subsidies

This chapter, the second in a series of three that focuses on implications of price
distortions on the first welfare theorem, revisits the topics of distortionary taxes
and subsidies, topics that had previously been treated solely within the consumer
model and, to some extent (in end-of-chapter exercises) in the producer model. In
these previous treatments, we simply assumed that a tax raised the price on con-
sumers or the marginal costs of producers, or a subsidy lowered these. In the utility
maximization framework, we were able to demonstrate that the deadweight losses
from such price distortions could be measured on marginal willingness to pay (or
compensated) curves — or, in the special case of quasilinear tastes, they could me
measured on (uncompensated) demand curves. We return to these issues now be-
cause we now have the tools to illustrate how such taxes and subsidies affect con-
sumer and producer prices in equilibrium — and how the work we previously did
fits into this equilibrium framework. Building on our work with price elasticities
in Chapter 18, we can furthermore differentiate these impacts based on relative
price elasticities of demand and supply. And we can show how the pictures that
you probably saw in previous economics courses apply for the case of quasilinear
tastes but also how such pictures can be deeply misleading when the quasilinearity
assumption is not appropriate.

Chapter Highlights
The main points of the chapter are:

1. The statutory incidence of taxes and subsidies bears no relation to the eco-
nomic incidence of taxes and subsidies, with the latter arising from the price
elasticities of demand and supply. The more price inelastic part of the mar-
ket will end up bearing the greater burden of a tax because the tax can more
easily be passed onto that side of the market regardless of how the tax law is
written.

2. The size of tax revenues one can expect from a given tax rate is similarly de-
pendent on the price elasticities of supply and demand, with revenues higher
1
 Distortionary Taxes and Subsidies 2

the more price inelastic at least one side of the market is. For most taxes, we
can derive a Laffer curve that illustrates tax revenues as initially increasing
with the tax rate but eventually decreasing. As a rough rule of thumb, dead-
weight losses increase by t 2 for any t -fold increase in tax rates — leading to
the conventional wisdom that it is more efficient to tax a large base at a low
rate rather than a small base at a high rate.

3. When underlying tastes are quasilinear, deadweight losses can be measured

as areas beneath the demand and above the supply curves — with dead-
weight losses on the consumer side arising from substitution effects as pre-
viously discussed in earlier chapters.

4. When underlying tastes are not quasilinear, measuring deadweight losses

may be masked by offsetting income and substitution effects on the curve
that arises from utility maximization. This is particularly relevant for wage
and capital taxes as underlying tastes by workers and savers almost certainly
give rise to such competing effects. In such cases, one cannot use the de-
mand and supply picture as the sole guide for estimating deadweight losses
but must instead isolate the deadweight loss from substitution effects using
compensated curves.

5. Land taxes represent an unusual real world tax in that such taxes, when ap-
propriately designed, give rise to no deadweight losses and no substitution
effects.

19A Solutions to Within-Chapter-Exercises for

Part A

Exercise 19A.1

Will the increase in price from the tax be larger or smaller in the long run? (Hint:
How is the price elasticity of supply in the long run usually related to the price elas-
ticity of supply in the short run?)
Answer: The increase in price will be larger in the long run since supply curves
are more price-elastic in the long run.

Exercise 19A.2

Using a pencil, redraw the graphs in panel (a) and (b) but this time label clearly
which price buyers end up paying and sellers end up receiving, taking into account
that sellers have to pay the tax in panel (a) and buyers have to pay the tax in panel
(b). Then erase the shifted curves in your two graphs. Do the two graphs now look
identical to each other and to the graph in panel (c)? (The answer should be yes.)
3 19A. Solutions to Within-Chapter-Exercises for Part A

Answer: In panel (a), p ′ = p d and p ′ − t = p s (where p d is the price paid by

consumers and p s is the price received by firms). In panel (b), p ′′ + t = p d and
p ′′ = p s . (Note that the graph is not drawn here.)

Exercise 19A.3

Illustrate how the equilibrium changes when the subsidy is paid to sellers (thus
reducing their MC ). Compare this to how the equilibrium changes when the sub-
sidy is paid to buyers (thus shifting the demand curve). Can you see how both of
these types of subsidies will result in an economic outcome summarized in Graph
19.2?

Answer: This is done in panels (a) and (b) of Exercise Graph 19A.3. In panel (a),
the subsidy to producers shifts the supply curve down by the amount of the per-
unit subsidy s — taking us from the initial equilibrium A to the new equilibrium
B with equilibrium price p ′ . In panel (b), the subsidy to consumers shifts up the
demand curve by s, taking us from the initial equilibrium A to the new equilibrium
C with equilibrium price p ′′ . Although the equilibrium prices are different in the
two pictures, the end result is equivalent: In panel (a), p ′ represents the amount
that consumers pay to producers — but producers still get the per-unit subsidy s
on top of it. Thus, the consumer price is p d = p ′ and the supplier price is p s =
p ′ + s = p ′′ . In panel (b), the equilibrium price p ′′ is the price that consumers pay to
producers — but they then receive a per-unit subsidy s, which implies that the true
price consumers pay is p d = p ′′ − s = p ′ while the price suppliers get is p s = p ′′ .

Exercise Graph 19A.3 : Price Subsidies for Producers and Consumers

 Distortionary Taxes and Subsidies 4

Exercise 19A.4

During the 2008 presidential campaign in the United States, oil prices increased
sharply. Some candidates advocated a “tax holiday” on gasoline taxes to help con-
sumers. Others argued that this would have little effect on gasoline prices in the
short run. Assuming each side was honest, how must they have disagreed on their
estimates of underlying price elasticities?
Answer: If you think the tax holiday will help consumers, you must think that
the demand curve is inelastic relative to the supply curve — and thus the lower tax
is primarily passed onto consumers. If you think the tax holiday will not help con-
sumers very much, then you must think that the supply curve is inelastic relative
to the demand curve — with the benefits of lower tax rates captured primarily by
suppliers.

Exercise 19A.5

In graphs with demand and supply curves similar to those in Graph 19.3, illus-
trate the economic impact on buyers and sellers of subsidies. How does the benefit
of a subsidy relate to relative price elasticities?
Answer: This is illustrated in Exercise Graph 19A.5 where the benefit of the sub-
sidy disproportionately accrues to consumers in panel (a) where demand is rela-
tively inelastic and to producers in panel (b) where supply is relatively inelastic.
Thus, just as tax burdens are passed onto the side of the market that behaves rela-
tively inelastically, so subsidy benefits are passed onto the side of the market that
behaves relatively inelastically.

Exercise Graph 19A.5 : Price Subsidies and Elasticities

5 19A. Solutions to Within-Chapter-Exercises for Part A

Exercise 19A.6

Does the impact of subsidies on market output also rise with the price-responsiveness
of buyers and sellers?
Answer: Yes. This is easily seen by simply drawing the green price differential in
Graph 19.4 of the text to the other side of the equilibrium price where the subsidy
price differential is determined.

Exercise 19A.7

Suppose the government has already imposed the taxes graphed in Graph 19.4
and is now considering raising this tax. Can you see in these graphs under what
circumstances this would result in a decrease in overall tax revenues?
Answer: It would result in a decline in tax revenues in panel (a) where the quan-
tity response is large for even a small increase in the green tax differential.

Exercise 19A.8

Suppose the tax on fuel-efficient cars is low and the tax on gas-guzzling cars is
high. Is it likely that our partial equilibrium estimate of a tax on gasoline will cause
us to over- or underestimate the full impact on government revenues?
Answer: The high tax on gas-guzzling cars will cause a decrease in demand in
that market and an increase in demand in the low-taxed fuel-efficient car market.
Thus, we are likely to over-estimate the tax revenues in the high-tax market and
under-estimate the tax revenues in the low tax market, with the former likely to
outweigh the latter as consumers avoid paying taxes by changing behavior.

Exercise 19A.9

Illustrate, using an analogous set of steps we just used as we worked our way
through Graph 19.6, how wage subsidies are inefficient even when workers are
completely unresponsive to changes in wages. (Hint: If you get stuck, read the next
section and come back.)
Answer: This is done in Exercise Graph 19A.9. In panel (a), we illustrate the shal-
lower (pre-subsidy) wage w ∗ and the steeper (post-subsidy) wage (x ∗ +s). After the
subsidy, the worker chooses A. This implies that the government pays this worker
the vertical distance S. The government could instead have given the worker a lump
sum subsidy L and made him just as well off. The difference between L and S is the
deadweight loss — and it arises solely because of the substitution effect from A to
B. In panel (b) of the graph, the analogous points to A, B and C are plotted on
the labor supply and compensated labor supply curves. The worker’s surplus at the
subsidized wage (w ∗ + s) is area (a + b). The total payment S made by the govern-
ment is the difference between the two wages times the amount the worker works
at A — giving us S = a + c. In the absence of the wage subsidy (and with just the
lump sum subsidy), the worker’s surplus is b. But the worker is indifferent between
 Distortionary Taxes and Subsidies 6

A and B — which means that the lump sum subsidy must have been equal to L = a.
Since we concluded that the price subsidy costs (a + c), this leaves us with dead-
weight loss DW L = c.

Exercise Graph 19A.9 : Wage Subsidies and Deadweight Loss

Exercise 19A.10

How large does deadweight loss get if the tax rate rises to 3t ? What if it rises to
4t ?
Answer: $9,000 and $16,000. (You can see this by adding up the relevant squares
in Graph 19.8a, and of course this follows exactly the rule that a k-fold increase in
tax rates causes a k 2 -fold increase in the deadweight loss.

Exercise 19A.11

Illustrate the relationship between subsidy rates, the deadweight loss from a
subsidy and the cost of the subsidy using the same initial graph of supply and de-
mand as in Graph 19.8a in graphs analogous to panels (b) and (c) of Graph 19.8.
Answer: This is done in Exercise Graph 19A.11. Panel (a) is the same as the
panel in the text, except that the subsidy shows up to the right of the equilibrium
price. The deadweight loss graph in panel (b) is exactly identical to that in the text
— because the deadweight loss triangles that accumulate are the same as those in
the text except that they appear on the other side of the initial equilibrium. For
instance, if s leads to a deadweight loss of 1, 000, then the deadweight loss triangle
to the right of the equilibrium in panel (a) for the small subsidy s has area 1,000.
When the subsidy rate rises to 2s, we are adding the equivalent of 3 further such
deadweight loss triangles (with each of the two squares in panel (a) equal to 1,000
7 19A. Solutions to Within-Chapter-Exercises for Part A

and the two triangles that remain jointly equal to 1,000). Thus, the new deadweight
loss is the initial 1,000 plus 3,000 — or a total of 4,000. The major difference in the
graphs comes in panel (c) which differs from panel (c) in the text. The reason for
this is that tax revenues will eventually fall to zero as tax rates are increased, but
subsidy costs for the government never fall as the government subsidizes more —
demand for the subsidy just keeps on going up. Thus, the subsidy cost increases at
an increasing rate as the subsidy rate s rises.

Exercise Graph 19A.11 : Subsidies, DWL and Costs

Exercise 19A.12

What would be the economic impact of a 100% tax on land rents (levied on
owners)?
Answer: A 100% tax on land rents would reduce land prices to zero — thus in ef-
fect transferring (in a lump sum way) the wealth of land owners to the government.
The reason that the transfer happens in a lump sum way is that there is nothing
land owners can do to avoid paying any of this tax. They either continue to rent the
land out and hand the rental revenues to the government, or they sell the land. But
if they sell it, no one is willing to pay more than zero for the land. Either way, the
landowners will pay the land rent tax.
 Distortionary Taxes and Subsidies 8

19B Solutions to Within-Chapter-Exercises for

Part B

Exercise 19B.1

Demonstrate that, whenever εd is less in absolute value than εs , consumers will

bear more than half the incidence of the tax, and whenever the reverse is true, they
will bear less than half of the incidence of the tax.
Answer: If |εd | < εs , this implies

|εd − εs | > |2εd | ( 19B.1.i)

(because εd is negative while εs is positive). We can then write

εd εd 1
< = . ( [Link])
εd − εs 2εd 2
Given the expression for d p s /d t we derived in the text, we then simply have to
multiply this by negative 1 (which reverses the inequality) to get

d ps εd 1
=− > − or d p s > −0.5d t . ( [Link])
dt εd − εs 2
We know that the only difference between p d and p s is t — which implies d p d =
d p s + d t . Substituting our result from ( [Link]), we can then write this as

d p d = d p s + d t > −0.5d t + d t = 0.5d t ( [Link])

or simply d p d > 0.5d t ; i.e. the increase in the consumer’s price is more than half
the per unit tax and consumers therefore bear more than half the burden of the tax.
To prove the reverse, the same steps can be used, with each inequality pointing in
the opposite directions.

Exercise 19B.2

Can you show that d p d /d t = εs /(εs − εd )? (Hint: Note that equation (19.2) im-
plies d p d /d t = (d p s /d t ) + 1.)
Answer: Dividing the expression d p d = d p s + d t by d t , we get the expression
given in the hint. Substituting d p s /d t = −(εd /(εd − εs )), we then get

d pd d ps εd
= +1 = − +1 ( 19B.2.i)
dt dt εd − εs
εd εs − εd εs
= + = . ( [Link])
εs − εd εs − εd εs − εd

Note that this again has an intuitive interpretation. It says that if price elastic-
ity of supply is zero (i.e. supply is perfectly inelastic), consumers experience no
9 19B. Solutions to Within-Chapter-Exercises for Part B

price change as a result of a tax (because the entire burden of the tax rests with
producers). If, on the other hand, price elasticity of demand is zero, d p d /d t = 1 or
d p d = d t and the entire burden of the tax is on consumers.

Exercise 19B.3

What is the price elasticity of demand for x? What is the cross-price elasticity of
demand for y?
Answer: The price elasticity of demand is

d x p −α p
= 2 α = −1. ( 19B.3.i)
dp x p p

The cross-price elasticity of demand for y is

dy p
= 0. ( [Link])
dp y

Exercise 19B.4

What is the price elasticity of supply?

Answer: The price elasticity of supply is

d xs p p
=β = 1. ( 19B.4)
d p xs βp

Exercise 19B.5

Can you verify that our answer for ∆P S is correct by simply calculating the area
of the rectangle d and the triangle (e) in Graph 19.10?
Answer: The area of the rectangle d is (10 − 6.18)(61.8) ≈ 236. The area of the
triangle e is (10−6.18)(100−61.8)/2 ≈ 73. Adding these together, we get a change in
producer surplus equal to 309.

Exercise 19B.6

Can you derive this expenditure function more directly through an expenditure
minimization problem?
Answer: The expenditure minimization problem is

min p x + y subject to u = αln x + y. ( 19B.6.i)

x,y

Setting up the Lagrangian and solving in the usual way, we get compensated
demand functions
α α
x= and y = u − αln . ( [Link])
p p
 Distortionary Taxes and Subsidies 10

Plugging these in to the objective function p x + y, we then get the expenditure

function

α α α
E (p, u) = p + u − αln = u + α − αln . ( [Link])
p p p

Exercise 19B.7

Can you verify that the expenditure necessary to reach the after tax utility at the
pre-tax price is always less than (or equal to) I ?
Answer: From what we derived in the text, we know that

p∗
µ ¶
E (p ∗, u t ) = αln + I. ( 19B.7)
pd
Whether this is greater or less than I then depends on whether ln(p ∗ /p d ) is
greater or less than 0. For any positive tax t , we know that p d ≥ p ∗ — i.e. taxes will
never lower the price consumers have to pay. Thus, (p ∗ /p d ) ≤ 1 — which implies
ln(p ∗ /p d ) ≤ 0 (since the natural log of 1 is zero and the natural log of anything
between zero and 1 is negative). Thus, E (p ∗, u t ) ≤ I .

Exercise 19B.8

What has to be true for E (p ∗, u t ) = I to hold?

Answer: The only way E (p ∗ , u t ) = I is if ln(p ∗ /p d ) = 0 — which is the case if
and only if p ∗ = p d (since the natural log of 1 is zero). Thus, the only way that the
income necessary to reach the after-tax utility level at pre-tax prices is equal to I is if
consumer price is unchanged as a result of the tax — which in turn can only happen
if supply is perfectly inelastic. This should make intuitive sense: If consumer price
is unaffected by that tax, then consumers are unaffected by the tax and their after-
tax utility is the same as their before-tax utility. And the only way you can reach the
before tax utility at pre-tax prices is for you to have I .

Exercise 19B.9

Can you show that in general, before substituting in specific pre- and post-tax
prices, equation (19.13) (which we derived using integral calculus) and equation
(19.18) (which we derived using the expenditure function) yield identical results?
Answer: The first equation tells us that ∆C S = α(ln p d − ln p ∗ ). This can be re-
written as ∆C S = −α(ln p ∗ − ln p d ). Using the property of logarithm that (ln a −
ln b) = ln(a/b), this further implies that

p∗
∆C S = −αln ( 19B.9)
pd
which is what we derived using the expenditure function.
11 19B. Solutions to Within-Chapter-Exercises for Part B

Exercise 19B.10

Does the Laffer curve in this example have a peak? Why or why not?
Answer: No, in this case the Laffer curve does not have a peak but rather ap-
proaches 1000 as t gets large. This is because we have chosen a demand curve that
has elasticity of −1 — which implies that total consumer spending on x remains
the same as the price of x increases. For instance, with α = 1000, the consumer
buys x = 1000/1 = 1000 when p d = 1 — for total spending of $1000. As t goes up,
the consumer reduces the quantity purchased such that p d x remains constant at
$1000. This further implies that, as t increases, a larger and larger fraction of the
consumer’s spending on x goes toward paying the tax. As t gets large, tax revenue
then converges to the amount the consumer spends on x — which is $1,000.

Exercise 19B.11

Verify that this labor supply function has zero wage elasticity of supply.
Answer: Since dl s /d w = 0, the wage elasticity of supply (dl s /d w)(w/l s ) = 0.

Exercise 19B.12

What is the wage elasticity of labor demand?

Answer: With labor demand function l d (w) = 25, 000, 000/w 2 , the wage elastic-
ity of demand for labor is

dl d w −50, 000, 000 w

= = −2. ( 19B.12)
d w ld w3 25,000,000
w2

Exercise 19B.13

Verify this.
Answer: The Lagrangian for the minimization problem is

L = wℓ + c + λ u t − c α ℓ(1−α) .
¡ ¢
( 19B.13.i)

The two first order conditions then imply that ℓ = (1 − α)c/(αw). Plugging this
into the constraint u t = c α ℓ(1−α) and solving for c, we get the compensated con-
sumption demand and plugging that back into ℓ = (1 − α)c/(αw), we get the com-
pensated leisure demand as given in the text. Plugging these into the objective
function wℓ + c, we then get the expenditure function
 Distortionary Taxes and Subsidies 12

µ ¶α ³ αw ´(1−α)
1−α
E (w, u t ) = w ut + ut
αw 1−α
·µ ¶α ³
α ´(1−α)
¸
1−α
= w (1−α) u t +
α 1−α
α (1−α)
+ α(1−α) αα
· ¸
(1 − α) (1 − α)
= w (1−α) u t
αα (1 − α)(1−α)
· ¸
(1−α) 1−α+α
=w ut
αα (1 − α)(1−α)
w (1−α) u t
= . ( [Link])
αα (1 − α)(1−α)

Exercise 19B.14

Can you find in a graph such as panel (b) of Graph 19.6 the various numbers
calculated above?
Answer: Referring to Graph 19.6 (in the text), we calculated that T = 400, L =
423.93 and DW L = 23.93. The utility associated with A and B in panel (b) is u A =
193.

Exercise 19B.15

Verify these.
Answer: For any per-unit tax t , we know that p d = p s + t . We can then write the
demand function as xd = (A −(p s +t ))/α and the supply function as x s = (p s −B)/β.
Setting these equal to each other and solving for p s , we get

βA + αB − βt
ps = . ( 19B.15.i)
α+β

From this, we can get the consumer after-tax price

βA + αB − βt βA + αB + αt
pd = ps + t = +t = . ( [Link])
α+β α+β
Plugging p s into either xd = (A − (p s + t ))/α or x s = (p s − B)/β, we get

A −B −t
xt = . ( [Link])
α+β

Exercise 19B.16

Verify the expression for deadweight loss. (Hint: There are two ways of doing
this: You can either take the appropriate integrals of the supply and demand func-
tions evaluated over the appropriate ranges of prices, or you can add rectangles and
triangles in a graph.)
13 19B. Solutions to Within-Chapter-Exercises for Part B

Answer: In this answer, we refer to Exercise Graph 19B.16, with the initial part
based on panel (a). Note that the vertical intercepts of the demand and supply
curves in panel (a) are A and B respectively. When t = 0, output is x ∗ = (A −B)/(α+
β) and price is p ∗ = (βA + αB)/(α + β).

Exercise Graph 19B.16 : Taxes and Deadweight Loss

Thus, initial consumer surplus is area (a + b + c) which is

(A − p ∗ )x ∗ 1 α(A − B)2
· ¸µ ¶
βA + αB A −B
C S∗ = = A− = ( 19B.16.i)
2 2 α+β α+β 2(α + β)2

and, similarly, initial producer surplus is (d + e + f ) which is

(p ∗ − B)x ∗ β(A − B)2

P S∗ = = . ( [Link])
2 2(α + β)2
This gives us total initial surplus of (a + b + c + d + e = F ) or simply

(α + β)(A − B)2 (A − B)2

T S∗ = C S∗ + P S∗ = = . ( [Link])
2(α + β)2 2(α + β)
Consumer surplus under a tax t , on the other hand, is just area (a) or

(A − p d )x t 1 βA + αB + αt A −B −t α(A − B − t )2
· ¸µ ¶
C St = = A− = .
2 2 α+β α+β 2(α + β)2
( [Link])
and producer surplus is area ( f ) or

β(A − B − t )2
· ¸µ ¶
(p s − B)x t 1 βA + αB − βt A −B −t
PSt = = −B = . ( 19B.16.v)
2 2 α+β α+β 2(α + β)2
 Distortionary Taxes and Subsidies 14

Finally, tax revenue is area (b + d) which is

t (A − B − t )
T R t = t xt = . ( [Link])
α+β
Total surplus under the per-unit tax t is then (a + b + d + f ) — or

(A − B − t )(A − B + t )
T St = C St + P St + T Rt = . ( [Link])
2(α + β)
The deadweight loss from the tax is then T S ∗ − T S t — which is equal to (c + e)
— or

(A − B)2 (A − B − t )(A − B + t ) (A − B)2 (A − B)2 − t 2 t2

DW L(t ) = − = − = .
2(α + β) 2(α + β) 2(α + β) 2(α + β) 2(α + β)
( [Link])
We could instead use integrals to calculate areas under demand and supply
functions which are the inverses of demand and supply curves. These are graphed
in panel (b) of Exercise Graph 19B.16. The deadweight loss — area (e + c) — is then
equal to the change in consumer surplus (b+c) plus the change in producer surplus
(d + e) minus tax revenue (d + b) where

(2A − p ∗ )p ∗
Zp d Zp d
A−p (2A − p d )p d
µ ¶ µ ¶
∆C S = xd (p)d p = dp = −
p∗ p∗ α 2α 2α
( [Link])
Zp ∗ Zp ∗ µ ∗
p −B (p − 2B)p ∗ (p s − 2B)p s
¶ µ ¶
∆P S = x s (p)d p = dp = − ( 19B.16.x)
ps ps β 2β 2β
t (A − B − t )
T R = t xt = . ( [Link])
α+β

Adding the first two and subtracting the last, substituting our expressions for
p ∗ , p s and p d and simplifying, we then get the same expression for deadweight
loss.

Exercise 19B.17

For t = 0.5, verify that the marginal product columns of the table report the
correct results.
Answer: When t = 0.5, k 1 = 200 and k 2 = 800. The marginal product of capital
in housing is
α 100 50
MP K 1 (k 1 ) = 1/2 = 1/2 = 1/2 . ( 19B.17.i)
2k 1 2k 1 k1
When k 1 = 200 is plugged in, we therefore get MP K 1 ≈ 3.54 — which means that
the after tax return on capital in housing is (1 − t )MP K 1 = 0.5(3.54) = 1.77. In the
non-housing sector, marginal product is
15 19B. Solutions to Within-Chapter-Exercises for Part B

β 100 50
MP K 2 (k 2 ) = = = . ( [Link])
2k 21/2 2k 21/2 k 21/2
When k 2 = 800, we then get MP K 2 ≈ 1.77. Thus, in equilibrium the after tax rate
of return in housing is equal to the rate of return in the untaxed sector.

Exercise 19B.18

If capital is “sector-specific” and cannot move from one use to another, would
you still expect the housing tax to be shifted? Explain.
Answer: No. If capital cannot be shifted, then the mechanism through which
the general equilibrium effect arises is no longer operative.

Exercise 19B.19

Why is the relative size of the housing sector relevant for determining how much
owners of capital in other sectors are affected by a tax on housing capital?
Answer: If the housing sector is small, then even a large tax on housing capital
will result in a shift of capital that is small relative to the non-housing sector — and
thus will have relatively little impact on the non-housing sector. Of course such a
tax would raise relatively little in revenue if the taxed sector is small.
 Distortionary Taxes and Subsidies 16

19C Solutions to Odd Numbered

End-of-Chapter Exercises
Exercise 19.1

In our discussion of economic versus statutory incidence, the text has focused
primarily on the incidence of taxes. This exercise explores analogous issues related to
the incidence of benefits from subsidies.

A: Consider a price subsidy for x in a partial equilibrium model of demand and

supply in the market for x.
(a) Explain why it does not matter whether the government gives the per-unit
subsidy s to consumers or producers.
Answer: This is illustrated in panel (a) of Exercise Graph 19.1 where the
pre-subsidy demand curve D intersects the pre-subsidy supply curve S at
price p ∗ and output level x ∗ . If the subsidy is given to consumers, de-
mand curves shift up by s to D ′ — causing the equilibrium to shift to
A with price p A and output level x s . If the subsidy is given to produc-
ers, supply shifts down by s to S ′ — causing the equilibrium to shift to
B with price p B and output level x s . The output level in equilibrium A
and B is therefore the same — and independent of whether the subsidy
was given to consumers or producers. The equilibrium price differs —
but the equilibrium impact on consumers and producers is identical. In
equilibrium A, consumers pay p A to producers but then receive s per unit
back from the government — implying that the suppliers’ price is p s = p A
and consumer prices are p d = p A − s = p B . In equilibrium B, consumers
pay p B to producers — and producers now receive a per unit subsidy s
for each good that is sold. Thus, the prices received by producers are
p s = p B + s = p A and prices paid by consumers are p d = p B . Whether
the subsidy is given to producers or consumers, it is therefore the case
that p s = p A , p d = p B and output is x s .
(b) Consider the case where the slopes of demand and supply curves are roughly
equal in absolute value at the no-subsidy equilibrium. What does this im-
ply for the way in which the benefits of the subsidy are divided between
consumers and producers?
Answer: This is what we just illustrated in panel (a) of Exercise Graph 19.1
where the price received by producers rises from p ∗ to p A and the price
paid by consumers falls from p ∗ to p B . When the slopes of the demand
and supply curves are roughly equal in absolute value at the initial equi-
librium, the distance between p A and p ∗ will be roughly equal to the dis-
tance between p ∗ and p B — i.e. consumers will benefit by about as much
as producers will from the subsidy.
(c) How does your answer change if the demand curve is steeper than the sup-
ply curve at the no-subsidy equilibrium?
17 19C. Solutions to Odd Numbered End-of-Chapter Exercises

Exercise Graph 19.1 : The Economic Incidence of Subsidies

Answer: This is illustrated in panel (b) of Exercise Graph 19.1 where the
consumer price p d drops substantially more than the producer price p s
rises from the initial p ∗ . Thus, when consumers are relatively less respon-
sive to price changes, they will obtain the bulk of the benefit from the
subsidy.
(d) How does your answer change if the demand curve is shallower than the
supply curve at the no-subsidy equilibrium?
Answer: This is illustrated in panel (c) of Exercise Graph 19.1. Now the
producer price p s rises more than the consumer price p d falls from the
initial p ∗ . Thus, when consumers are relatively more responsive to price
changes than producers, the bulk of the benefit from the subsidy accrues
to producers.
(e) Can you state your general conclusion — using the language of price elas-
ticities — on how much consumers will benefit relative to producers when
price subsidies are introduced. How is this similar to our conclusions on
tax incidence?
Answer: The general conclusion is that the benefits of subsidies are shifted
disproportionately to the side of the market that is less sensitive to price
— i.e. the side of the market that is more price inelastic. This is similar to
what we concluded about the burden of a tax — which is also shifted to
the side of the market that is more price inelastic.
(f) Do any of your answers depend on whether the tastes for x are quasilinear?
Answer: No — our prediction of market outcomes in terms of prices and
quantities are based on uncompensated curves that include income and
substitution effects — because if we want to know what actually happens,
we have to look at actual or uncompensated curves. It is only when we try
to assess changes in surplus — i.e. welfare changes — that we need to use
compensated (or marginal willingness to pay) curves. Had we drawn into
 Distortionary Taxes and Subsidies 18

our graphs consumer surplus areas along the demand curves, we would
therefore have had to assume that the uncompensated demand curves in
the graphs are also equal to marginal willingness to pay curves — which
would only be true if tastes are quasilinear in x.
B: In Section 19B.1, we derived the impact of a marginal per-unit tax on the
price received by producers — i.e. d p s /d t .
(a) Repeat the analysis for the case of a per-unit subsidy and derive d p s /d s
where s is the per-unit subsidy.
Answer: Consider the general case where demand is given by xd (p), sup-
ply is given by x s (p) and the no-tax equilibrium has price p ∗ and quantity
x ∗ . Now suppose a small subsidy s (to be received by consumers for each
unit of x that is purchased) is introduced. This implies that the price p d
effectively paid by buyers is s lower than the price p s at which the good is
purchased from suppliers; i.e. p d = p s − s. Taking the differential of this,
we get
d p d = d p s − d s; (19.1.i)

i.e. the change in the consumer price p d is equal to the change in the pro-
ducer price p s minus the change in s. In the new equilibrium, demand
has to equal supply, with each evaluated at the relevant price; i.e.

xd (p d ) = x s (p s ). ([Link])

Taking the differential of this, we can write

d xd d xs
d pd = d ps ([Link])
d pd d ps
and substituting equation (19.1.i) into equation ([Link]), this becomes

d xd d xs
(d p s − d s) = d ps . ([Link])
d pd d ps
Rearranging terms in this equation, we can write it as
µ ¶
d xd d x s d xd
− d ps = d s. (19.1.v)
d pd d ps d pd
Before the subsidy is introduced, the equilibrium was at the intersection
of supply and demand at p ∗ and x ∗ — a point on both the supply and
demand curve. Multiplying equation (19.1.v) by p ∗ /x ∗ , it becomes

d xd p ∗ d x s p ∗ d xd p ∗
µ ¶
− d p s = d s, ([Link])
d pd x∗ d ps x∗ d pd x∗
which you should notice contains several price elasticity terms (evaluated
at the no-tax equilibrium). Rewriting the equation in terms of these price
elasticities, it becomes
19 19C. Solutions to Odd Numbered End-of-Chapter Exercises

(εd − εs )d p s = εd d s ([Link])

where εd is the price elasticity of demand and εs is the price elasticity of

supply. Re-arranging terms, we can also then write this as

d ps εd
= . ([Link])
ds εd − εs

(b) What is d p d /d s?
Answer: Equation (19.1.i) implies that

d pd d ps
= − 1. ([Link])
ds ds
Substituting equation ([Link]) into this equation, we then get
µ ¶
d pd εd εd εd − εs εs
= −1 = − = (19.1.x)
ds εd − εs εd − εs εd − εs εd − εs

(c) What do your results in (a) and (b) tell you about the economic incidence of
a per-unit subsidy when the price elasticity of demand is zero? What about
when the price elasticity of supply is zero?
Answer: When εd = 0, we get

d pd d ps
= −1 and = 0; ([Link])
ds ds
i.e. d p d = −d s and d p s = 0. Thus, the entire benefit of the subsidy goes to
consumers whose price drops by the amount of the subsidy. When εs = 0,
we get

d pd d ps
= 0 and = 1; ([Link])
ds ds
i.e. d p d = 0 and d p s = d s. Thus, the entire benefit of the subsidy goes to
producers whose price increases by the amount of the subsidy.
(d) What does your analysis suggest about the economic incidence of the sub-
sidy when the price elasticities of demand and supply are equal (in abso-
lute value) at the no-subsidy equilibrium?
Answer: In the case where |εd | = εs , we get

d pd εs 1 d ps εd 1
= = − and = = . ([Link])
ds εd − εs 2 ds εd − εs 2

Thus, consumer prices fall by the same amount as producer prices in-
crease — and the benefits of the subsidy are shared equally.
 Distortionary Taxes and Subsidies 20

(e) More generally, can you show which side of the market gets the greater ben-
efit when the absolute value of the price elasticity of demand is less than the
price elasticity of supply?
Answer: When |εd | < εs , |εd − εs | > 2|εd |. This implies

|εd | |εd | 1
< = ([Link])
|εd − εs | 2|εd | 2

which implies

d ps εd |εd | 1
= = < . ([Link])
ds εd − εs |εd − εs | 2

Thus, producers will receive less than half the benefit of the subsidy when
demand is more price inelastic than supply — which is exactly our con-
clusion in part A(e). Put differently, consumers receive a disproportion-
ately larger share of the benefit of the subsidy when they are more price
inelastic than producers. (We could similarly derive d p d /d s < −0.5, which
says the same thing.)

Exercise 19.3

In the text, we discussed deadweight losses that arise from wage taxes even when
labor supply is perfectly inelastic. We now consider wage subsidies.

A: Suppose that the current market wage is w ∗ and that labor supply for all
workers is perfectly inelastic. Then the government agrees to pay employers a
per-hour wage subsidy of $s for every worker hour they employ.
(a) Will employers get any benefit from this subsidy? Will employees?
Answer: Employers will get no benefit from the subsidy because the entire
economic incidence will fall on employees whose labor supply is perfectly
inelastic. Thus, wages paid by employers will remain at w ∗ while wages
received by employees will be (w ∗ + s).
(b) In a consumer diagram with leisure ℓ on the horizontal and consumption
c on the vertical axes, illustrate the impact of the subsidy on worker budget
constraints.
Answer: This is illustrated in panel (a) of Exercise Graph 19.3.
(c) Choose a bundle A that is optimal before the subsidy goes into effect. Locate
the bundle that is optimal after the subsidy.
Answer: This is also illustrated in panel (a) of the Graph — where C lies
vertically above A because the (uncompensated) labor response to the
wage subsidy is perfectly inelastic.
(d) Illustrate the size of the subsidy payment S as a vertical distance in the
graph.
Answer: This is again illustrated in panel (a) of Exercise Graph 19.3 where
S is the vertical distance between A and C .
 21 19C. Solutions to Odd Numbered End-of-Chapter Exercises

Exercise Graph 19.3 : Dead Weight Losses from Wage Subsidies with Inelastic Labor Supply

(e) Illustrate how much P we could have paid the worker in a lump sum way
(without distorting wages) to make him just as well off as he is under the
wage subsidy. Then locate the deadweight loss of the wage subsidy as a
vertical distance in your graph.
Answer: This is also illustrated in panel (a) of Exercise Graph 19.3 where
P is the difference between the parallel budgets. The higher of these is
the budget necessary to get the worker to his post-subsidy utility level
uC at the pre-subsidy wage w ∗ . The deadweight loss is then simply the
difference between S and P .
(f) On a separate graph, illustrate the inelastic labor supply curve as well as
the before and after-subsidy points on that curve. Then illustrate the ap-
propriate compensated labor supply curve on which to measure the dead-
weight loss. Explain where this deadweight loss lies in your graph.
Answer: This is done in panel (b) of Exercise Graph 19.3. The subsidy
causes workers to move from A to C on their inelastic supply curve. The
compensated labor supply curve that corresponds to the after-subsidy
utility level uC must then pass through C — and as long as there is at least
some substitutability between leisure and consumption, it must slope
up (because it only incorporates substitution effects). When measuring
worker surplus on this compensated labor supply curve, we find surplus
of (a + c) under the wage subsidy but only surplus of (c) under the lump
sum subsidy that eliminates the wage subsidy. Since the worker is equally
happy at B and C , the lump sum subsidy must therefore be equal to (a);
i.e. P = a. The actual wage subsidy paid, however, is s times l ∗ — which is
equal to area (a + b); i.e. S = a + b. Thus, the deadweight loss is DW L = b.
(g) True or False: As long as leisure and consumption are at least somewhat
substitutable, compensated labor supply curves always slope up and wage
 Distortionary Taxes and Subsidies 22

subsidies that increase worker wages create deadweight losses.

Answer: This is true. Substitution effects tell us that consumption of
leisure decreases when leisure becomes more expensive — which is equiv-
alent to saying that labor supplied increases as w increases. Compen-
sated labor supply curves only incorporate substitution effects — and
thus, so long as there is any substitutability between leisure and con-
sumption, more labor will be supplied as w goes up. Put differently, com-
pensated labor supply curves must slope up — and it is that upward slope
that gives rise to the deadweight loss from wage subsidies.

B: Suppose that, as in our treatment of wage taxes, tastes over consumption c

and leisure ℓ can be represented by the utility function u(c, ℓ) = c α ℓ(1−α) and
that all workers have leisure endowment of L (and no other source of income).
Suppose further that, again as in the text, the equilibrium wage in the absence
of distortions is w ∗ = 25.
(a) If the government offers a $11 per hour wage subsidy for employers, how
does this affect the wage costs for employers and the wages received by em-
ployees?
Answer: Since labor supply is perfectly inelastic for these tastes (as shown
in the text), the entire benefit of the subsidy accrues to workers. Thus,
wages for workers increase to $36 per hour while wages paid by employers
remain unchanged at $25 per hour.
(b) Assume henceforth that α = 0.5. What is the utility level u s attained by
workers under the subsidy (as a function of leisure endowment L)?
Answer: From the utility maximization problem, we get that consump-
tion demand is c = 36(0.5)L = 18L while leisure demand is 0.5L. (This is
derived in the text — we simply plugged in the after-subsidy wage of $36
and α = 0.5.) Plugging these back into the utility function, we get

u s = (18L)0.5 (0.5L)0.5 = 3L. (19.3.i)

(c) What’s the least (in terms of leisure endowment L) we would need to give
each worker in a lump sum way to get them to agree to give up the wage
subsidy program?
Answer: In the text, we derived the compensated leisure demand and
consumption demand equations and, from these, the expenditure func-
tion

w (1−α) u
E (w, u) = . ([Link])
αα (1 − α)(1−α)
The expenditure necessary to get the worker to the utility level u s at the
pre-subsidy wage w = 25 is E (25, 3L). Plugging in u = 3L, w = 25 and
α = 0.5, we get E = 30L. Thus, the worker would have to have 30L in order
to be just as happy without the subsidy as he is with the subsidy when the
value of his leisure endowment is 25L. We would therefore have to give
23 19C. Solutions to Odd Numbered End-of-Chapter Exercises

the worker 5L in a lump sum way to make him as well off at a wage of $25
per hour as he is under the subsidized wage of $36 per hour.
(d) What is the per worker deadweight loss (in terms of leisure endowment L)
of the subsidy?
Answer: The deadweight loss is then the difference between what we ac-
tually have to pay in a wage subsidy and what we could have paid in a
lump sum way without making workers worse off. We just concluded
that a lump sum payment of 5L would be just as good for the worker as
the wage subsidy. Under the wage subsidy, the government has to pay $11
per hour worked — and workers always work 0.5L hours (taking the rest of
their endowment as leisure). Thus, the wage subsidy costs 11(0.5)L = 5.5L
— implying a deadweight loss of DW L = 0.5L per worker.
(e) Use the compensated labor supply curve to verify your answer.
Answer: In the text, we derived the compensated labor demand curve as
µ ¶α
1−α
l sc (w, u) = L − u. ([Link])
αw
Plugging in α = 0.5 and the after-subsidy utility level u − 3L, we get
µ ¶
3L
l sc (w) = L − . ([Link])
w 0.5
The equivalent lump sum payment is the area under this function be-
tween the before and after-subsidy wage; i.e.
Z36 · µ ¶¸
3L
Lump Sum Payment = L− d w = 5L. (19.3.v)
25 w 0.5
Subtracting this from the actual subsidy cost of 11(0.5L) = 5.5L, we get a
deadweight loss of DW L = 0.5L as before.
Exercise 19.5

(This exercise builds on exercise 19.4 which you should do before proceeding.)
Through the income tax code, governments typically tax most interest income — but,
through a variety of retirement programs, they often subsidize at least some types of
interest income.
A: Suppose all capital is supplied by individuals that earn income now but don’t
expect to earn income in some future period — and therefore save some of their
current income. Suppose further that these individuals do not change their cur-
rent consumption (and thus the amount they put into savings) as interest rates
change.
(a) What is the economic incidence of a government subsidy of interest in-
come? What is the economic incidence of a tax on interest income?
Answer: The economic incidence of taxes and subsidies always falls more
heavily on the side of the market that behaves more inelastically. In this
 Distortionary Taxes and Subsidies 24

case, savers are perfectly inelastic — which implies they will enjoy the
full benefit of interest subsidies and pay the full cost of interest taxes. (If
you have trouble seeing this, draw a graph with the savings supply curve
almost perfectly inelastic — you should see that the incidence of taxes
and subsidies then falls almost entirely on savers.)
(b) In the text, we illustrated the deadweight loss from a subsidy on inter-
est income when savings behavior is unaffected by changes in the interest
rate. Now consider a tax on interest income. In a consumer diagram with
current consumption c 1 on the horizontal and future consumption c 2 on
the vertical axis, illustrate the deadweight loss from such a tax for a saver
whose (uncompensated) savings supply is perfectly inelastic.
Answer: This is done in panel (a) of Exercise Graph 19.5(1).

Exercise Graph 19.5(1) : Dead Weight Loss from Taxing Interest Income

The after tax optimal bundle (on the shallower budget) is A — giving util-
ity u A . At that bundle, the saver pays the distance T in taxes. But he would
have been willing to pay up to L in order to keep the distortionary tax from
being implemented — with the difference between L and T constituting
the deadweight loss from the tax on interest income.
(c) What does the size of the deadweight loss depend on? Under what special
tastes does it disappear?
Answer: The size of the deadweight loss depends on the distance between
A and B — which in turn depends on the degree of substitutability be-
tween consumption now and in the future. The more complementary
consumption is across time, the shorter this distance — and the less the
deadweight loss. If consumption across the two periods is perfectly com-
plementary, B and A would be the same point — and the deadweight
loss would disappear because the substitution effect that gives rise to
the deadweight loss would be gone. However, the uncompensated la-
bor supply curve would then not be perfectly inelastic. Instead, for the
25 19C. Solutions to Odd Numbered End-of-Chapter Exercises

deadweight loss to disappear and the labor supply curve to be perfectly

inelastic, there would have to be a kink point at A that is sufficiently large
to keep a substitution effect from appearing while still allowing point C
to lie directly above A in panel (a) of the graph.

(d) On a separate graph, illustrate the inelastic savings (or capital) supply
curve. Then illustrate the compensated savings supply curve that allows
you to measure the deadweight loss from the tax on interest income. Ex-
plain where in the graph this deadweight loss lies.
Answer: This is illustrated in panel (b) of Exercise Graph 19.5(1). The
deadweight loss will be measured on the compensated supply curve that
corresponds to the after-tax utility level u A (and therefore passes through
A.) Saver surplus under the interest rate (r −t ) is just area b — while saver
surplus under the interest rate r is (a + b + c). But the saver is equally
happy at A and B because at B he had to pay a lump sum tax that pushes
him to the indifference curve u A at the pre-tax interest rate. This implies
that the lump sum tax he is willing to pay is (a + c). But the tax revenue
that is collected under the distortionary tax is just area a. This leaves us
with deadweight loss c.

(e) What happens to the compensated savings supply curve as consumption

becomes more complementary across time — and what happens to the
deadweight loss as a result?
Answer: As consumption becomes more complementary across time, the
distance between A and B in panel (a) of the graph decreases — which
also implies the distance between C and B decreases in panel (b) of the
graph Thus, the compensated savings supply curve becomes more in-
elastic as consumption becomes more complementary across time — be-
cause the substitution effect becomes smaller. This causes the deadweight
loss area c in the graph to shrink. If consumption is perfectly complemen-
tary across time, C and B will lie on top of one another in panel (b) of the
graph — with the compensated savings supply curve perfectly inelastic.
In that case, the deadweight loss area disappears.

(f) Is the special case when there is no deadweight loss from taxing interest in-
come compatible with a perfectly inelastic uncompensated savings supply
curve?
Answer: Yes and no. We would need consumption across time to be per-
fectly complementary — but that would imply that there is only a wealth
effect and no substitution effect. And this would imply that savings falls
with an increase in the interest rate — i.e. C would lie to the right of A
in panel (a) of the graph. This further implies a downward sloping sav-
ings supply curve, not a perfectly inelastic curve. But you could still have
a kink at A that is not a right angle and that still allows C to lie directly
above A (as discussed in the answer to (c)).
 Distortionary Taxes and Subsidies 26

B: Suppose everyone’s tastes and economic circumstances are the same as those
described in part B of exercise 19.4 — with α = 0.5 and I = 100, 000.1
(a) Suppose further that there are 10,000,000 consumers like this — and they
are the only source of capital in the economy. How much capital is supplied
regardless of the interest rate?
Answer: The (inelastic) supply of capital is given by

K s = 10, 000, 000k s = 10, 000, 000(1−α)I = 10, 000, 000(0.5)(100, 000) = 500, 000, 000, 000
(19.5.i)
— i.e. $500 trillion.
(b) Suppose next that demand for capital is given by K d = 25, 000, 000, 000/r .
What is the equilibrium real interest rate r ∗ in the absence of any price
distortions?
Answer: Setting K d equal to the inelastic capital supply of $500,000,000,000,000
and solving for r , we get r ∗ = 0.05.
(c) Suppose that, for any dollar of interest earned, the government provides
the person who earned the interest a 50 cent subsidy. What will be the new
(subsidy-inclusive) interest rate earned by savers, and what will be the in-
terest rate paid by borrowers? What if the government instead taxed 50%
of interest income?
Answer: Since the supply of savings is perfectly inelastic, savers will re-
ceive the full benefit of the subsidy and bear the full burden of the tax.
Thus, under the subsidy, the interest rate earned by savers would be 1.5(r ∗ ) =
1.5 ∗ 0.05 = 0.075. Under the tax, the interest rate earned by savers would
fall to 0.5(r ∗ ) = 0.5(0.05) = 0.025. In both cases, borrowers would still pay
interest r ∗ = 0.05.
(d) Consider the subsidy introduced in (c). How much utility V will each saver
attain under this subsidy?
Answer: In exercise 19.4 we derived the indirect utility function as

V (r, I ) = αα (1 − α)(1−α) (1 + r )(1−α) I . ([Link])

Substituting in α = 0.5, I = 100, 000 and the post-subsidy interest rate for
savers (r = 0.075), we get V ≈ 51, 841.1. (You can of course also simply
use the demand functions for current and future consumption to get c 1 =
50, 000 and c 2 = 53, 750 and plug these into the utility function to get the
same answer.)
(e) How much current income would each saver have to have in order to ob-
tain the same utility V at the pre-subsidy interest rate r ∗ ? In terms of fu-
ture dollars, how much would it therefore cost the government to make
1 Among other functions, you should have derived uncompensated and compensated savings func-
tion as
1 + r (1−α)
" µ ¶ #
k s (r, I ) = (1 − α)I and k sc = 1 − α I. (19.5)
1+r
27 19C. Solutions to Odd Numbered End-of-Chapter Exercises

each saver as well off in a lump sum way as it does using the interest rate
subsidy?
Answer: In part (e) of exercise 19.4, we derived the expenditure function
as

u
E (r, u) = . ([Link])
αα (1 − α)(1−α) (1 + r )(1−α)
We now simply need to plug α = 0.5, r = 0.5 and u = 51, 841.1 to get E ≈
101, 183.47. Thus, in current dollars, each saver’s income would have to
increase by $1,183.47 — which is equal to (1 + 0.05)1, 183.47 ≈ $1, 242.65.
(f) How much interest will the government have to pay to each saver (in the
future) under the subsidy? Use this and your previous answer to conclude
the amount of deadweight loss per saver in terms of future dollars. Given
the number of savers in the economy, what is the overall deadweight loss?
Answer: Since savers will always save $50,000, they will earn (1+0.075)50, 000 =
$3, 750 in interest income under the subsidy — which is $1,250 more than
they would have earned in the absence of the subsidy. Thus, the subsidy
costs the government $1,250 per saver (in terms of future dollars). Sub-
tracting the lump sum payment that would have made savers equally well
off, we get a deadweight loss of

DW L = 1, 250 − 1, 242.65 = $7.35 ([Link])

per saver. Given that there are 10 million savers, the overall deadweight
loss is therefore about $73.5 million in terms next period.
(g) Derive the compensated savings function (as a function of r ) given the
post-subsidy utility level V .
Answer: In exercise 19.4 we derived the compensated savings function as
" µ ¶(1−α) #
1+r
k sc (r, r ) = 1−α I (19.5.v)
1+r

where r is the interest rate that determines the compensated utility level.
If we want the compensated savings function at the post-subsidy utility,
we therefore set r to 0.075. Substituting in α = 0.5 and I = 100, 000, we
then get

51841.1
k sc (r ) ≈ 100, 000 − . ([Link])
(1 + r )0.5
You can of course also derive this by substituting the compensated de-
mand for current consumption evaluated at the post-subsidy utility level
from I to get the same answer.
(h) Use your answer to (g) to derive the aggregate compensated capital supply
function — and then find the area that corresponds to the deadweight loss.
Compare this to your answer in part (f ).
 Distortionary Taxes and Subsidies 28

Answer: The aggregate compensated capital supply function is simply k sc

multiplied by the number of savers — i.e.
µ ¶
51841.1
K sc (r ) = 10, 000, 000 100, 000 − ([Link])
(1 + r )0.5
In panel (a) of Exercise Graph 19.5(2), the inverse of this — which we have
called the compensated capital supply curve — is depicted.

Exercise Graph 19.5(2) : Compensated Capital Supply Curve and Function

This is analogous to Graph 19.7 in the text where we concluded that area c
in the graph is the deadweight loss. Panel (b) of the graph inverts this back
to plot the compensated capital supply function K sc (r ) — with areas from
panel (a) again labeled. We can now see that the deadweight loss area
c is simply the rectangle (a + c) minus the integral between r = 0.05 and
r = 0.075. The rectangle (a+c) is simply the total amount of capital K sup-
plied under the subsidy multiplied by 0.025 — where K is simply 50,000
times the 10 million savers. Thus, (a+c)=500,000,000,000(0.025)=12,500,000,000.
The deadweight loss area c is therefore

Z 0.075 µ ¶
51841.1
DW L = 12, 500, 000, 000 − 10, 000, 000 100, 000 − dr
0.05 (1 + r )0.5
= 12, 500, 000, 000 − 10, 000, 000 100, 000r − 2(51841.1)(1 + r )0.5 |0.075
£¡ ¢ ¤
0.05
= 73, 531, 136 ≈ $73.5 million. ([Link])

This is precisely what we calculated in (f).

(i) Repeat parts (d) through (h) for the case of the tax on interest income de-
scribed in part (c).
Answer: To get the after-tax utility level, we would substitute r = 0.025
rather than r = 0.075 into our expression for V to get V ≈ 50, 621.14. To
29 19C. Solutions to Odd Numbered End-of-Chapter Exercises

get the same utility level under the pre-tax interest rate r ∗ = 0.05, we
would not substitute r = 0.025 (rather than r = 0.075) into our expres-
sion for the expenditure function — as well as our new value for V . This
gives us E ≈ $98, 802.35 — i.e. each saver would be willing to give up
100, 000 − 98, 802.35 = $1, 197.65 in current dollars or 1, 197.65(1 + 0.05) ≈
1, 257.53 in future dollars. The government would actually receive only
$1,250 in tax revenue (in terms of future dollars) — implying that individ-
uals would be willing to give up $7.53 to avoid the tax. For 10 million
individuals, that amounts to a deadweight loss of approximately $75.3
million. To calculate the same deadweight loss using the compensated
capital supply function, we can again use k sc (r, r ) defined in part (g) —
but now r = 0.025 giving us the equation

50621.14
k sc (r ) = 100, 000 − . ([Link])
(1 + r )0.5

Integrating this between r = 0.025 and r = 0.05, we get the lump sum
payment an individual would be willing to make to not incur the tax —
which is again $1,257.53. Or, aggregating the function across 10,000,000
individuals, integrating in the same way and then subtracting the total
tax payments received, we get approximately $75.3 million in deadweight
loss.
(j) You have calculated deadweight losses for interest rates that are reason-
able for 1-year time horizons. If we consider distortions in people’s deci-
sions over longer time horizons (such as when they plan for retirement), a
more reasonable time frame might be 25 years. With annual market inter-
est rates of 0.05 in the absence of distortions, can you use your compensated
savings function (given in the footnote to the problem) to estimate again
what the deadweight losses from a subsidy that raises the effective rate of
return by 50% and from a tax that lowers it by 50% would be?
Answer: If the undistorted interest rate is 0.05 annually, then a dollar in-
vested now will result in (1+0.05)2 5 ≈ $3.39 — giving us an effective inter-
est rate of 2.39 (or 239 percent) over the 25 years. If a subsidy raised that
by 50%, it would raise it to approximately 3.585. Using our compensated
savings function, we would then get

" µ ¶(1−α) #
1+r
k sc (r ) = 1−α I
1+r
1 + 3.585 0.5
· µ ¶ ¸
= 100, 000 − 50, 000
1+r
107, 063
= 100, 000 − . (19.5.x)
(1 + r )0.5

Using the same steps as in (h), we then get the deadweight loss
 Distortionary Taxes and Subsidies 30

Z 3.585 µ ¶
107, 063
DW L = 50, 000(3.585 − 2.39)(10, 000, 000) − 10, 000, 000 100, 000 −
2.39 (1 + r )0.5
= 45, 019, 422, 909 ≈ $45 billion ([Link])

or about $4,500 per individual. You can also verify this by using the ex-
penditure function as we did in the earlier parts of this exercise. For a tax
that cuts the effective return from 2.39 to 1.195, on the other hand, we
would use the compensated savings function

" µ ¶(1−α) #
1+r
k sc (r ) = 1−α I
1+r
1 + 1.195 0.5
· µ ¶ ¸
= 100, 000 − 50, 000
1+r
74, 077.66
= 100, 000 − . ([Link])
(1 + r )0.5

Subtracting the actual tax revenue from the appropriate integral (that gives
us the lump sum payment individuals would be willing to make to avoid
the tax), we get

· Z 2.39 µ ¶¸
74, 077.66
DW L = 10, 000, 000 100, 000 − − 1.195(50, 000)(10, 000, 000)
1.195 (1 + r )0.5
= 64, 671, 207, 082 ≈ $62.7 billion. ([Link])

Exercise 19.7

Business and Policy Application: Land Use Policies: In most Western democra-
cies, it is settled law that governments cannot simply confiscate land for public pur-
poses. Such confiscation is labeled a “taking” — and, even when the government
has compelling reasons to “take” someone’s property for public use, it must compen-
sate the landowner. But, while it is clear that a “taking” has occurred when the gov-
ernment confiscates private land without compensation, constitutional lawyers dis-
agree on how close the government has to come to literally confiscating private land
before the action constitutes an unconstitutional “taking”.
A: Any restriction that alters the way land would otherwise be used reduces the
annual rental value of that land and, from the owner’s perspective, can therefore
be treated as a tax on rental value.
(a) Explain why the above statement is correct.
Answer: There are two parts to the statement: First, restrictions on land
use cause rental values to decline, and second, that this is equivalent to
a tax on land rents from the owner’s perspective. Prior to any restrictions
on the land, the user of the land (whether this is a renter or the owner who
31 19C. Solutions to Odd Numbered End-of-Chapter Exercises

can be though of as renting the land from himself) employs it in the op-
timal way. If the restrictions placed on the land still permit the land to be
used in this way, the rental value of the land is unaffected. For instance,
if the optimal use of the land is to have an office building on it and a re-
striction is passed that prohibits users to plant corn, nothing has really
changed. But if the restriction impacts the use of the land, then by defini-
tion we are no longer using it the way we would have in the absence of the
restriction — i.e. we are no longer employing it in the optimal way from
the individual’s perspective. Thus, the user gets less out of the land which
lessens his demand for the land — i.e. the rental value declines. From the
owner’s perspective, it does not really matter what causes the rental value
to decline — whatever it is, this reduces the value of the land. Since taxing
land rents lowers land value, we can then set the tax just "right" so as to
achieve a reduction in land value — including the reduction that occurs
as a result of a land use regulation.
(b) Suppose a land use regulation is equivalent (from the owner’s perspective)
to a tax of t % on land rents to be statutorily paid by landowners (where
0 < t < 1). How does it affect the market value of the land?
Answer: Under such a tax, the owners would still collect the same amount
of land rent as before but would then have to give up t % of it. This lowers
the land rents that owners can keep to (1−t )% of what they kept originally
— implying that land value (which is just the present discounted value of
all future land rents) falls by t %.
(c) I am about to buy an acre of land from you in order to build on it. Right
before we agree on a price, the government imposes a new zoning regula-
tion that limits what I can do on the land. Who is definitively made worse
off by this?
Answer: You are definitively made worse off — because your land value
drops immediately by the net present value of the decrease in land rents
caused by the regulation.
(d) Suppose you own 1000 acres of land that is currently zoned for residential
development. Then suppose the government determines that your land is
home to a rare species of salamander — and that it is in the public interest
for no economic activity to take place on this land in order to protect this
endangered species. From your perspective, what approximate tax rate on
land rents that you collect is this regulation equivalent to? Do you think
this is a “taking”?
Answer: Since you are effectively prohibited from using (or renting) the
land, this is equivalent to you paying a 100% tax on land rents. Were the
government to actually impose such a tax, it would have essentially con-
fiscated your land because all rents from it would go to the government
instead of to you. But from your perspective, the regulation to protect the
salamanders is no different — you again lose all the value from your land.
The only difference is that the government has actually not taken posses-
 Distortionary Taxes and Subsidies 32

sion of the value of the land the way it would under the 100% land rent
tax. But from the owner’s perspective, it sure looks like a taking.
(e) Suppose that, instead of prohibiting all economic activity on your 1000
acres, the government reduces your ability to build residential housing on
it to a single house. How does your answer change? What if it restricts
housing development to 500 acres? Do you think this would be a “taking”?
Answer: If you are now only allowed to build a single house, some small
fraction of the original value of the land is retained — and the regula-
tion therefore is not a complete “taking” from your perspective. But it’s
pretty close to a complete taking. If the government restricts housing de-
velopment to only 500 of the 1000 acres, this is approximately equivalent
to the government taxing your land rents at 50% — or confiscating half
your land. Courts would almost certainly not call this a “taking” — but
from the owner’s perspective, it’s just like the government just took half
the land.
B: Suppose that people gain utility from housing services h and other consump-
tion x, with tastes described by the utility function u(x, h) = ln x + ln h. Con-
sumption is denominated in dollars (with price therefore normalized to 1). Hous-
ing services, on the other hand, are derived from the production process h =
k 0.5 L α where k stands for units of capital and L for acres of land. Suppose
0 < α < 1. Let the rental rate of capital be denoted by r , and assume each person
has income of 1000.
(a) Write down the utility maximization problem and solve for the demand
function for land assuming a rental rate R for land.
Answer: Substituting the housing production function into the utility func-
tion, we can write the utility function as

u(x, k, L) = ln x + ln k 0.5 L α = ln x + 0.5ln k + αln L.

¡ ¢
(19.7.i)

The utility maximization problem can then be written as

max ln x + 0.5ln k + αln L subject to 1000 = x + r k + RL. ([Link])

x,k,L

We can then write the Lagrangian function as

L = ln x + 0.5ln k + αln L + λ(1000 − x − r k − RL) . ([Link])

The first order conditions are then

1 0.5 α
= λ; = λr and = λR. ([Link])
x k L
We can then use the first and third of these to write x in terms of L(i.e.
x = RL/α) and we can use the second and third of these to write k in terms
of L (i.e. k = 0.5RL/(αr )). Substituting these into the budget constraint,
we can then solve for the demand function for land
33 19C. Solutions to Odd Numbered End-of-Chapter Exercises

1000α
L= . (19.7.v)
(1.5 + α)R

(b) Suppose your city consists of 100,000 individuals like this — and there are
25,000 acres of land available. What is the equilibrium rental rate per acre
of land (as a function of α)?
Answer: Setting demand equal to supply, we get the equation
µ ¶
1000α
100, 000 = 25, 000 ([Link])
(1.5 + α)R

that solves for

4000α
R∗ = . ([Link])
(1.5 + α)

(c) Using your answers above, derive the amount of land each person will con-
sume.
Answer: Substituting equation ([Link]) into (19.7.v), we get

1000α 1000α(1.5 + α) 1
L= ³ ´= = ([Link])
4000α
(1.5 + α) (1.5+α) (1.5 + α)(4000α) 4

— i.e. everyone consumes a quarter of an acre in equilibrium.

(d) Suppose the government imposes zoning regulations that reduce the coef-
ficient α in the production function from 0.5 to 0.25. What happens to the
equilibrium rental value of land?
Answer: This implies that the equilibrium rental rate for an acre of land
falls from

4000(0.5)
= 1000 ([Link])
(1.5 + 0.5)

to

4000(0.25)
≈ 571.43. (19.7.x)
(1.5 + 0.25)

(e) Suppose that what you have calculated so far is the monthly rental value
of land. What happens to the total value of an acre of land as a result of
these zoning regulations assuming that people use a monthly interest rate
of 0.5% to discount the future?
Answer: This implies that land values fall from 1000/0.005 = $200, 000 to
571.43/0.005 ≈ $114, 286.
 Distortionary Taxes and Subsidies 34

(f) Suppose that, instead of lowering α from 0.5 to 0.25 through regulation,
the government imposes a tax t on the market rental value of land and
statutorily requires renters to pay. Thus, if the market land rental rate is R
per acre, those using the land must pay t R on top of the rent R for every
acre they use. Set up the renters’ utility maximization problem, derive the
demand for land and aggregate it over all 100,000 individuals. Then derive
the equilibrium land rent per acre as a function of t (assuming α = 0.5).
Answer: The only thing that changes in the utility maximization problem
is the rental price in the budget constraint — which must now include the
tax. Thus, the utility maximization problem becomes

max ln x + 0.5ln k + 0.5ln L subject to 1000 = x + r k + (1+ t )RL. ([Link])

x,k,L

The first order conditions are then

1 0.5 0.5
= λ; = λr and = λ(1 + t )R. ([Link])
x k L
Solving these as before, we get the demand function for land

250
L= . ([Link])
(1 + t )R
Aggregating across 100,000 consumers and setting equal to supply, we get
the equation
µ ¶
250
100, 000 = 25, 000. ([Link])
(1 + t )R
Solving for R, we get the equilibrium rental rate

1000
R∗ = . ([Link])
(1 + t )

(g) Does the amount of land consumed by each household change?

Answer: Plugging the equilibrium land rental rate into equation ([Link]),
we get L = 1/4 as before. So — no, the amount of land consumed by each
household does not change — because the tax inclusive rental rate for
land remains unchanged. (Before, the land rental rate was 1000 prior to
the imposition of zoning; the tax inclusive rental rate now is (1 + t )R ∗ =
1000.)
(h) Suppose you own land that you rent out. What level of t makes you indif-
ferent between the zoning regulation that drove α from 0.5 to 0.25 and the
land rent tax that does not change α?
Answer: We calculated before that the rental rate on an acre of land falls
to 571.43 under the zoning regulation. You would therefore be indifferent
35 19C. Solutions to Odd Numbered End-of-Chapter Exercises

between this and a land rent tax if the rental rate under the land rent tax
also fell to 571.43; i.e. if

1000
= 571.43. ([Link])
(1 + t )
This solves to t = 0.75.
(i) Suppose the government statutorily collected the land rent tax from the
owner instead of from the renter. What would the tax rate then have to be
set at to make the land owner indifferent between the zoning regulation
and the tax?
Answer: In this case, the consumer does not statutorily face a tax — and
so the utility maximization problem is the same as in (a) leading to the
same equilibrium rental rate R ∗ = 1000 as calculated in (b). Landown-
ers would therefore collect $1,000 per acre but would then have to pay
t (1000) to the government. In order for this to leave them with an after-
tax rent of $571.43 (as under the zoning regulation), the tax rate would
have to be set at t ≈ 0.4286.

Exercise 19.9

Policy Application: Rent Control: Is it a Tax or a Subsidy?: In exercise 18.11 we

analyzed the impact of rent control policies that impose a price ceiling in the hous-
ing rental market. The stated intent of such policies is often to make housing more
affordable. Before answering this question, you may wish to review your answers to
exercise 18.11.

A: Begin by illustrating the impact of the rent control price ceiling on the price
received by landlords and the eventual equilibrium price paid by renters.
(a) Why is it not an equilibrium for the price ceiling to be the rent actually paid
by renters?
Answer: It is not an equilibrium because, at the rent controlled price,
more people demand apartments than are supplied — which implies some
non-price rationing mechanism must allocate the scarce apartments. This
mechanism will raise the real cost of apartments to consumers until de-
mand is once again equal to supply.
(b) If you wanted to implement a tax or subsidy policy that achieves the same
outcome as the rent control policy, what policy would you propose?
Answer: This is illustrated in panel (a) of Exercise Graph 19.9(1).
The price ceiling results in a price of p s received by suppliers and a price
p c paid by consumers (once all costs of the non-price rationing mecha-
nism have been taken into account). We can achieve the same prices for
consumers and producers (as well as the same reduction in housing units
provided) by simply imposing a per-unit tax in the amount t = (p c − p s ).
(c) Could you credibly argue that the alternative policy you proposed in (b)
was designed to make housing more affordable?
 Distortionary Taxes and Subsidies 36

Exercise Graph 19.9(1) : Rent Control, Taxes and Subsidies

Answer: Since prices for consumers are raised from p ∗ to p c , it would not
be possible to argue this (since at least some of the tax will be passed to
consumers).
(d) If you did actually want to make housing more affordable (rather than try-
ing to replicate the impact of rent control policies), would you choose a
subsidy or a tax?
Answer: You would want to use a subsidy — which would result in p c <
p ∗ < p s as well as an increase in housing units.
(e) Illustrate your proposal from (d) — and show what would happen to the
rental price received by landlords and the rents paid by renters. What hap-
pens to the number of housing units available for rent under your new pol-
icy?
Answer: This is illustrated in panel (b) of Exercise Graph 19.9(1) where a
subsidy of s raises producer prices to p s and lowers consumer prices to
p c — while resulting in an increase in output. The subsidy s can in fact be
set so as to insure that housing will be exactly as affordable as advocates
of rent control wish when they set a price ceiling. (Under rent control, of
course, this would not be successful.)
(f) True or False: Policies that make housing more affordable must invariably
increase the equilibrium quantity of housing — and rent control policies
fail because they reduce the equilibrium quantity of housing while subsi-
dies succeed for the opposite reason.
Answer: This is true as illustrated in the previous parts and in the Graph.
(g) True or False: Although rental subsidies succeed at the goal of making
housing more affordable (while rent control policies fail to do so), we can-
not in general say that deadweight loss is greater or less under one policy
rather than the other.
37 19C. Solutions to Odd Numbered End-of-Chapter Exercises

Answer: This is also true. In panel (a) of Exercise Graph 19.9(1), the dead-
weight loss from rent control is at least (a + b) and may be as much as
(a + b + c + d) depending on the rationing mechanism used. In panel (b)
of the Graph, the deadweight loss from the subsidy is equal to area (e).
Depending on the relative elasticities of supply and demand, (e) may or
may not be smaller than (a +b). Of course the more the rationing mecha-
nism under rent control causes deadweight loss to be larger than (a + b),
the more likely it is that the subsidy will definitively be more efficient.
Still, the example illustrates that we may care more about the policy goal
of creating more affordable housing than the precise size of deadweight
loss — and if that is the case, then the subsidy is clearly the better policy
as it actually achieves the policy goal rather than doing the opposite.
B: Suppose, again as in exercise 18.11, that the aggregate monthly demand curve
is p = 10000 − 0.01x while the supply curve is p = 1000 + 0.002x. For simplicity,
suppose again that there are no income effects.
(a) Calculate the equilibrium number of apartments x ∗ and the equilibrium
monthly rent p ∗ in the absence of any price distortions.
Answer: Re-writing the demand and supply curves as demand and supply
functions (by writing them as functions of p), setting them equal to one
another and solving, we get p ∗ = 2, 500. Plugging this back into either the
demand or supply function, we get x ∗ = 750, 000.
(b) In exercise 18.11, you were asked to consider the impact of a $1,500 price
ceiling. What housing tax or subsidy would result in the same economic
impact?
Answer: Imposing a price ceiling of 1,500 will result in a reduction of x
supplied. We can solve for the quantity by substituting 1,500 for p into
the supply curve and solving for x to get x = 250, 000. Thus, suppliers will
supply 250,000 housing units at the price ceiling of 1500. Consumers will
then have to compete for the limited number of apartments — exerting
effort that adds to their cost of renting these units. In equilibrium, the
effort cost needs to be sufficient so that demand is equal to supply —
which means the overall price (including effort cost) must be

p c = 10000 − 0.01(250, 000) = 7, 500. (19.9.i)

The rent control policy therefore results in a producer price of p s = 1, 500

and a consumer price of p c = 7, 500 — a result we could equally well get
by simply imposing a per unit tax of $6,000.
(c) Suppose that you wanted to use tax/subsidy policies to actually reduce rents
to $1,500 — the stated goal of the rent control policy. What policy would
you implement?
Answer: You would want to implement a subsidy program. In order to get
the consumer price to be 1500, we will need sufficient numbers of apart-
ments so that every consumer who wants to buy one at that price can get
one. Thus, plugging 1,500 into the demand curve and solving for x, we
 Distortionary Taxes and Subsidies 38

get that we need a total of 850,000 apartments. In order for producers to

supply that many apartments, they have to be able to charge

p s = 1000 + 0.002(850, 000) = 2, 700. ([Link])

Thus, we need to provide a per unit subsidy of $1,200 in order to cre-

ate sufficient incentives for the market to provide housing at a consumer
price of $1,500 (and a seller’s price of $2,700).
(d) Consider the policies you derived in (b) and (c). Under which policy is the
deadweight loss greater?
Answer: Exercise Graph 19.9(2) illustrates the supply and demand curves
for this problem as well as the various numbers we have calculated. The
deadweight loss from the tax is the triangle (a + b + c) to the left of the
equilibrium while the deadweight loss from the subsidy is the triangle
(d + e) to the right of the equilibrium. (The deadweight loss under rent
control is at least the size of the deadweight loss from the tax but possibly
more depending on the rationing mechanism). It is easily seen in the
picture that the deadweight loss from the tax (and from rent control) is
larger than the deadweight loss from the subsidy. (You can easily verify
that the tax deadweight loss triangle is $1,750,000,000 while the subsidy
deadweight loss triangle is $60,000,000.)

Exercise Graph 19.9(2) : Rent Control Taxes versus Housing Subsidies

Exercise 19.11

Policy Application: Mortgage Interest Deductibility, Land Values and the Equi-
librium Rate of Return on Capital: In the text, we suggested that the property tax
can be thought of in part as a tax on land and in part as a tax on capital invested
in housing. In the U.S., property taxes are typically levied by local governments —
while the major piece of federal housing policy is contained in the federal income
39 19C. Solutions to Odd Numbered End-of-Chapter Exercises

tax code which allows individuals to deduct (from income) the interest they pay on
home mortgages prior to calculating the amount of taxes owed.
A: Whereas we can think of the property tax as a tax on both land and housing
structures, we can think of the homeownership subsidy in the federal tax code as
a subsidy on land and housing structures.
(a) If your marginal federal income tax rate is 25% and you are financing 100%
of your home value, how much of your housing consumption is being sub-
sidized through the tax code? What if you are only financing 50% of the
value of your home?
Answer: In the first scenario, about 25% of your housing consumption is
subsidized by the government through the tax code (because almost all of
your mortgage payments are tax deductible since most of your payment
is made up of interest that is deductible). If you are only financing 50%,
only about 12.5% of your housing consumption is subsidized.
(b) Suppose homeowners are similar to one another in terms of their marginal
tax rate and how much of their home they are financing, and suppose that
this implies a subsidy of s for every dollar of housing/land consumption.
How would you predict the value of suburban residential land (assumed
to be in fixed supply) is different as a result of this than it would have been
in the absence of this policy?
Answer: Exercise Graph 19.11 illustrates the market for residential land.

Exercise Graph 19.11 : Tax Code Subsidy for Residential Land

In the graph, land is initially denominated in units such that the original
equilibrium price (at A) is $1. The effective subsidy s per dollar then raises
the rental price of residential land to (1+ s) — with the entire incidence of
the subsidy going to land owners — i.e. land owners now collect (1+s) per
unit of land that they originally collected just $1 on. We would therefore
predict that residential land value increases by the present discounted
value of all future streams of s for each unit of land.
 Distortionary Taxes and Subsidies 40

(c) When s was first introduced, who benefitted from the implicit land sub-
sidy: current homeowners or future homeowners?
Answer: Current homeowners (i.e. those who owned homes when the
subsidy was introduced) benefitted — because all increases in rents from
land (due to the subsidy) are immediately incorporated into the land price.
If you owned land, your asset therefore became more valuable. If you did
not own land when the subsidy was introduced, then you would not ben-
efit from the subsidy because you will pay for the benefit of the subsidy
streams when you buy land at the value that already incorporates all those
future streams of subsidies.
(d) Now consider s as a subsidy on housing capital. Do you think houses are
larger or smaller as a result of the federal income tax code?
Answer: This subsidy on housing capital reduces the price of building on
land — and therefore will cause an increase in housing consumption (so
long as housing is not a Giffen good). We would therefore tend to think
that houses are larger as a result of the federal income tax code.
(e) Suppose that the overall amount of capital in the economy is fixed and
that capital is mobile across sectors. Thus, any given unit of capital can
be invested in housing or alternatively in some other non-housing sector
where it earns some rate of return. If the overall amount of capital in the
economy is fixed, what happens to the fraction of capital invested in the
housing sector?
Answer: If houses are larger as a result of the tax subsidy (as we just con-
cluded), this means that more capital is invested in housing than would
otherwise be the case. Thus, the fraction of capital invested in the hous-
ing sector increases.
(f) What would you predict will happen to the rate of return on capital in the
non-housing sector? Explain.
Answer: As capital shifts from the non-housing sector to the housing sec-
tor, the marginal product of capital must increase in the non-housing sec-
tor — which implies the rate of return on capital increases in the non-
housing sector. Thus, the benefit of the subsidy on housing capital (as
opposed to the subsidy on land) accrues to all forms of capital, not just
housing capital.
(g) True or False: Even though only housing capital is statutorily subsidized,
the economic incidence of this subsidy falls equally on all forms of capital
(so long as capital is mobile between sectors).
Answer: This is true as just explained in the previous part.

B: Suppose we model owners of capital as a “representative investor” who chooses

to allocate K units of capital between the housing sector and other sectors of the
economy. With k 1 representing capital invested in housing and k 2 representing
capital invested in other sectors, suppose f 1 (k 1 ) = αk 10.5 and f 2 (k 2 ) = βk 20.5 are
the production functions of the two sectors.
41 19C. Solutions to Odd Numbered End-of-Chapter Exercises

(a) In the absence of any policy distortions, calculate the fraction of total cap-
ital (K ) that is invested in the housing sector.
Answer: Our representative investor then wants to maximize her total re-
turn by optimally choosing the allocation of her capital K across the two
sectors. Put differently, she wants to solve the maximization problem

max f 1 (k 1 ) + f 2 (k 2 ) subject to k 1 + k 2 = K . (19.11.i)

k 1 ,k 2

The solution to this problem is

α2 K β2 K
k 1∗ = and k ∗
2 = . ([Link])
α2 + β2 α2 + β2

(Note that at this solution, the marginal product of capital is the same in
the two sectors). The fraction of total capital invested in housing is then
(α2 /(α2 + β2 ).
(b) What changes as a result of the federal income tax code’s implicit housing
subsidy s.
Answer: The problem then becomes

max (1 + s) f 1 (k 1 ) + f 2 (k 2 ) subject to k 1 + k 2 = K . ([Link])

k 1 ,k 2

The solution to this problem is

(1 + s)2 α2 K β2 K
k 1∗ = and k ∗
2 = . ([Link])
(1 + s)2 α2 + β2 (1 + s)2 α2 + β2

Thus, the fraction of capital invested in the non-housing sector falls while
the fraction invested in the housing sector increases.
(c) What happens to the marginal product of capital in the non-housing sec-
tor?
Answer: The marginal product of capital in the non-housing sector is
given by

∂ f 2 (k 2 )
MP K 2 = = 0.5βk −0.5 . (19.11.v)
∂k 2

Plugging k 2 from equation ([Link]) into this, we get

¶−0.5
β2 K (1 + s)2 α2 + β2
µ µ ¶
MP K 2 = 0.5β = 0.5 . ([Link])
(1 + s)2 α2 + β2 K

Thus, as s increases, the marginal product of non-housing capital also

increases.
 Distortionary Taxes and Subsidies 42

(d) What happens to the equilibrium rate of return on capital?

Answer: In equilibrium, the marginal product of capital in the non-housing
sector increases (as just demonstrated) until it is equal to the subsidy-
inclusive marginal product of capital in the housing sector — thus, the
equilibrium marginal return on capital increases across all sectors.
(e) True or False: The general equilibrium subsidy incidence of the implicit
subsidy of housing capital falls equally on all forms of capital.
Answer: This is true — capital moves between the sectors until the rate of
return is equalized.

Conclusion: Potentially Helpful Reminders

1. Remember that the actual impact of a tax or subsidy — how much price
changes for buyers and sellers, and how much the quantity transacted in the
market changes — is determined solely by the uncompensated demand and
supply curves. Thus, when thinking about tax incidence and output changes,
we do not need to worry about substitution and income (or wealth) effects
because the combined effect is what matters. The price elasticities of the
curves then become the key fact to focus on.

2. It is when we start to evaluate the welfare changes from taxes and subsidies
— the changes in surplus and deadweight loss — that isolating substitution
effects may become important. Being told that the underlying good is quasi-
linear, however, is your “get-out-of-jail-free” card: In that case, there are no
income (or wealth) effects to worry about, and the regular (uncompensated)
demand and supply curves are all we need.

3. The assumption of quasilinearity is in fact the key (and usually unspoken)

assumption in Principles of Economics courses where supply and demand
graphs are typically drawn with surplus and deadweight loss read directly off
those curves.

4. It is only if there is a good reason to believe that the underlying tastes of con-
sumers, workers or savers are not quasilinear that we have to deviate from the
uncompensated demand and supply graphs to estimate welfare changes —
i.e. changes in surplus and deadweight loss. To be more precise, in consumer
goods markets, we have to worry about compensated demand curves; and
in labor and capital markets, we have to worry about compensated supply
curves.

5. This becomes particularly important when there is reason to believe that

substitution and income (or wealth) effects are offsetting and thus mask each
other in the uncompensated graph. In such cases, it becomes possible that
the uncompensated graph shows little or no deadweight loss when in reality
the deadweight loss is masked by these offsetting effects.
43 19C. Solutions to Odd Numbered End-of-Chapter Exercises

6. Finally, a quick graphing hint (that will carry through the next chapters): Even
though demand and supply curves are shifting as a result of taxes and sub-
sidies, it is much easier to find ways of identifying incidence and welfare
changes without shifting the curves. In the case of taxes, this simply involves
drawing the “tax wedge” to the left of the no-tax equilibrium; and in the case
of subsidies, it involves drawing the “subsidy wedge” to the right of the no-tax
equilibrium.