IE54500 – Problem Set 1

Dr. David Johnson

Fall 2020

1. Affordability
We commonly hear people say, “I can’t afford to purchase 𝑥,” where 𝑥 is some quantity of a good or
service.

a) In the diagram below, which assumes that prices and income are exogenous, explain what
someone might mean when they say that the quantity 𝑥 𝑎 is “unaffordable.”

In this diagram, the region below the intersections of the budget constraint with the x and y axes
represent the feasible set of quantities of x and y that the individual can purchase with income M
and prices 𝑝𝑥 and 𝑝𝑦 . So the quantity 𝑥 𝑎 is literally unaffordable.

b) In the same diagram, explain what someone might mean if they say that 𝑥 𝑏 is “unaffordable.”
Frequently, people will refer to some other quantity of x, say 𝑥 𝑏 , that lies on the budget set, as
being unaffordable. What they mean here is that they are unwilling to trade any more of their y,
given current prices and income, to obtain more x. So 𝑥 𝑏 is in fact affordable, but it is not
consistent with their current preferences. What the individual likely means here is that that they
would like to have more x, but not if it entails having to sacrifice some y to obtain it.

c) How would your answers to the previous questions change if income is earned in the labor
market rather than given exogenously?
Most individuals earn their income in the labor market, wH, where H is hours of work. If we treat
the wage, w, as exogenous (assuming individuals have some set of labor market abilities that
they cannot affect through additional human capital investments), their exogenous income is
wT, where T, for example, is 24 hours in a day. So, when an individual says they cannot afford
some quantity of x that is outside the current budget set, what they most likely mean is that they
are unwilling to trade away more leisure to obtain more x.
2. Concave indifference curves
Suppose indifference curves between two goods are concave to the origin, rather than convex. (As an
example from Lecture 3, you might imagine that 𝑈(𝑥, 𝑦) = 𝑥 2 + 𝑦 2 .) How would that consumer allocate
income between the two goods in order to maximize utility?

The consumer would allocate all of their income to one of the goods. To see this, imagine a standard
linear budget constraint that intersects the indifference curve at some point not on the x or y intercept.
Because the indifference curve is concave, the individual could remain on the same indifference curve
while spending less money. This argument holds for any point that is not an intercept, so that is where
they would locate.

3. Perfect substitutes
Consider a consumer whose utility equals the sum of their caffeine consumption. The consumer can
obtain caffeine from either coffee or tea. Coffee yields 4 units of caffeine, while tea yields 3 units of
caffeine.

a) Draw this consumer’s indifference curve.

The utility function is 𝑈(𝑡, 𝑐 ) = 𝑈 0 = 4𝑡 + 3𝑐, so the indifference curve is a straight line with
slope -4/3 or -3/4 depending on which product is placed on each axis.

b) What is the consumer’s optimal choice of tea and coffee when the relative price of tea and
coffee, 𝑝𝑡 /𝑝𝑐 , is (i) 4/3, (ii) 3/4, or (iii) 2/4?
Since coffee and tea are perfect substitutes in this case, the consumer will allocate all of their
income to whichever good is the least expensive, in terms of the smallest implied unit cost of
caffeine. When coffee and tea are equally expensive in terms of cost per unit of caffeine, the
consumer’s choice cannot be determined (in this case, the indifference curve coincides with the
budget constraint).

4. Optimal fencing
a) Prove that a rectangle of fixed perimeter encloses the largest possible area when it is a square.
The optimization problem here is to maximize the area of a rectangle with width w and length l,
𝐴 = 𝑤𝑙, subject to the constraint 2𝑤 + 2𝑙 = 𝑃 for some fixed perimeter P. First-order conditions
for ℒ = 𝑤𝑙 + 𝜆(𝑃 − 2𝑙 − 2𝑤 ) imply

𝑙 − 2𝜆 = 0
𝑤 − 2𝜆 = 0
𝑃 − 2𝑙 − 2𝑤 = 0

The first two FOCs imply that 𝑙 ∗ = 𝑤 ∗ .

b) Derive an expression for the Lagrange multiplier and interpret its meaning.

From the FOCs above, 𝜆∗ = 𝑙/2. The Lagrange multiplier corresponds to the marginal increase in
area associated with an increase in perimeter P.
 c) Use the expression for the Lagrange multiplier to evaluate how the area of the fenced-in area
will change if the perimeter of the fence increases from 500 to 501 feet. Assume that the fence
designer behaves optimally and wants to maximize the area enclosed.

If the original perimeter is 500 ft, that implies that one side of the fence is 125 ft. Therefore, the
area of the fence would increase by approximately 𝜆∗ ∙ Δ𝑙 = 125/2 ∙ 1 = 62.5 ft.

5. Evaluating [quasi-]concavity
a) Show that 𝑓(𝑥, 𝑧) = 𝑥 1⁄2 𝑧 1⁄4 is both concave and quasi-concave. Assume 𝑥 > 0 and 𝑧 > 0.
To evaluate this, we need to form the Hessian:

3 1 1 3
𝑓 𝑓𝑥𝑧 −1/4𝑥 −2 𝑧 4 −1/8𝑥 −2 𝑧 −4
𝐇 = [ 𝑥𝑥 ]=[ 1 3 1 7]
𝑓𝑧𝑥 𝑓𝑧𝑧
−1/8𝑥 −2 𝑧 −4 −3/16𝑥 2 𝑧 −4

𝑓(𝑥, 𝑧) is concave if the matrix 𝐇 is negative definite, which requires |𝐇𝟏 | < 0 and |𝐇𝟐 | > 0,
where |𝐇𝟏 | and |𝐇𝟐 | are the first and second principal minors of 𝐇:

3 1
|𝐇𝟏𝒙 | = 𝑓𝑥𝑥 = −1/4𝑥 −2 𝑧 4 < 0
1 7
|𝐇𝟏𝒛 | = 𝑓𝑧𝑧 = −3/16𝑥 2 𝑧 −4 < 0
3
|𝐇𝟐 | = 𝑓𝑥𝑥 𝑓𝑧𝑧 − 𝑓𝑧𝑥 𝑓𝑥𝑧 = 132𝑥 −1 𝑧 2 > 0

Thus, 𝑓 (𝑥, 𝑧) = 𝑥 1⁄2 𝑧 1⁄4 is concave. The function is also quasi-concave because every concave
function is quasi-concave.

b) Show that 𝑓(𝑥, 𝑧) = 𝑥 5⁄2 𝑧 5⁄4 is quasi-concave but not concave. Assume 𝑥 > 0 and 𝑧 > 0.
The same approach will show that 𝑓(𝑥, 𝑧) = 𝑥 5⁄2 𝑧 5⁄4 is not concave. To determine whether it is
quasi-concave, we must first form the bordered Hessian:

0 𝑓𝑥 𝑓𝑧
𝐇 = [𝑓𝑥 𝑓𝑥𝑥 𝑓𝑥𝑧 ]
𝑓𝑧 𝑓𝑧𝑥 𝑓𝑧𝑧

𝐼𝑓 |𝐇𝟏 | < 0 and |𝐇𝟐 | > 0, where

0 𝑓𝑥
𝐇𝟏 = [ ]
𝑓𝑧 𝑓𝑥𝑥

𝑎𝑛𝑑 𝐇𝟐 = 𝐇. Evaluating these determinants, you will see that 𝑓 (𝑥, 𝑧) is in fact quasi-concave.
You might also note that this function is an increasing monotonic transformation of the function
from part a). Recall that an increasing monotonic transformation of a concave function preserves
quasi-concavity, but it does not necessarily preserve concavity.
 c) Show that the second-order conditions for obtaining a maximum will always be satisfied in the
case of a consumer who maximizes a strictly quasi-concave utility function subject to a linear
budget constraint. You can assume that utility is only a function of two goods.

Note that this solution assumes two goods but easily generalizes to n variables.

A consumer seeks to maximize utility by choosing quantities of x and y, subject to a linear budget
constraint 𝑀 = 𝑝𝑥 𝑥 + 𝑝𝑦 𝑦:

max ℒ = 𝑈(𝑥, 𝑦) + 𝜆(𝑀 − 𝑝𝑥 𝑥 − 𝑝𝑦 𝑦)

𝑥,𝑦,𝜆

𝑇ℎ𝑒 𝑓𝑜𝑙𝑙𝑜𝑤𝑖𝑛𝑔 𝐹𝑂𝐶𝑠 ℎ𝑜𝑙𝑑 𝑎𝑡 𝑡ℎ𝑒 𝑜𝑝𝑡𝑖𝑚𝑎𝑙 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛:

ℒ𝑥 = 𝑈𝑥 − 𝜆𝑝𝑥 = 0
ℒ𝑦 = 𝑈𝑦 − 𝜆𝑝𝑦 = 0
ℒ𝜆 = 𝑀 − 𝑝𝑥 𝑥 − 𝑝𝑦 𝑦 = 0

These FOCs achieve a maximum if the following SOCs hold: |𝐇𝟏 | < 0 and |𝐇𝟐 | > 0, where

0 −𝑝𝑥 −𝑝𝑦
𝐇 = [−𝑝𝑥 𝑈𝑥𝑥 𝑈𝑥𝑦 ]
−𝑝𝑦 𝑈𝑦𝑥 𝑈𝑦𝑦

𝐵𝑢𝑡 𝑖𝑓 𝑈(𝑥, 𝑦) is quasi-concave, then it must be true that |𝐁𝟏 | < 0 and |𝐁𝟐 | > 0 where

0 𝑈𝑥 𝑈𝑦
𝐁 = [𝑈𝑥 𝑈𝑥𝑥 𝑈𝑥𝑦 ]
𝑈𝑦 𝑈𝑦𝑥 𝑈𝑦𝑦

𝐹𝑟𝑜𝑚 𝑡ℎ𝑒 𝐹𝑂𝐶𝑠, 𝑤𝑒 𝑘𝑛𝑜𝑤 𝑡ℎ𝑎𝑡 𝑈𝑥 = 𝜆𝑝𝑥 and 𝑈𝑦 = 𝜆𝑝𝑦 . Substituting those values into 𝐇, we
have that |𝐇𝟏 | = λ2 |𝐁𝟏 | < 𝟎 and |𝐇𝟐 | = λ2 |𝐁𝟐 | > 0. Therefore, the optimal value of 𝑈(𝑥, 𝑦) is
guaranteed to be a maximum for a quasi-concave utility function with a linear constraint.

6. Ordinal utility
𝛽 𝛽2 𝛽3
Suppose a consumer’s utility function is 𝑈 = 𝑥1 1 𝑥2 𝑥3 , where ∑3𝑖=1 𝛽𝑖 = 1. Assume the consumer
faces prices 𝑝1 , 𝑝2 , and 𝑝3 and has exogenous income 𝑀.

a) Derive the consumer’s Marshallian demand functions.

We need to maximize the utility subject to the budget constraint. Using a monotonic
transformation, this is equivalent to solving the following problem:

max ℒ = 𝛽1 ln 𝑥1 + 𝛽2 ln 𝑥2 + 𝛽3 ln 𝑥3 + 𝜆(𝑀 − 𝑝𝑥1 𝑥1 − 𝑝𝑥2 𝑥2 − 𝑝𝑥3 𝑥3 )

𝑥1 ,𝑥2 ,𝑥3 ,𝜆
 𝛽𝑖 𝑀
Solving the system of equations formed by the FOCs, we get 𝑥𝑖∗ = for 𝑖 = 1,2,3.
𝑝𝑖
3𝛽 3𝛽2 3𝛽3
b) Now suppose the consumer’s utility function is 𝑈 = 𝑥1 1 𝑥2 𝑥3 , where ∑3𝑖=1 𝛽𝑖 = 1. Derive
the consumer’s Marshallian demand functions and compare your result to part a). Why should
you have expected this result?

This new utility function is an increasing monotonic transformation of the original function.
Therefore, they will yield the same set of demand functions. You can verify this by resolving the
utility maximization problem with the new function.

7. Engel and Cournot aggregations

Assume there exists a set of Marshallian demand functions, 𝑔(𝑝, 𝑀). The budget constraint implies that
∑𝑖 𝑝𝑖 𝑔𝑖 (𝑝, 𝑀) = 𝑀.

a) Show that the sum of the changes in a consumer’s expenditures given a small change in income,
𝑀, equals one.

Let 𝐸 = ∑𝑖 𝑝𝑖 𝑔𝑖 (𝑝, 𝑀) = 𝑀, and take the derivative of E with respect to M:

𝜕𝐸 𝜕𝑔𝑖 𝜕𝑀
= ∑ 𝑝𝑖 = =1
𝜕𝑀 𝜕𝑀 𝜕𝑀
𝑖

This result tells us that the change in expenditure on all goods due to a change in income must
equal the change in income.

b) Show that the sum of the changes in expenditures given a small change in the price of good 1,
𝑝1 , equals zero.

Take the derivative of E with respect to 𝑝1 :

𝜕𝐸 𝜕𝑔1 𝜕𝑝1 𝜕𝑔𝑖 𝜕𝑀

= [𝑝1 + 𝑔1 ( )] + ∑ 𝑝1 = =0
𝜕𝑝1 𝜕𝑝1 𝜕𝑝1 𝜕𝑝1 𝜕𝑝1
𝑖≠1

This says that if the price of one good changes, the change in expenditures over all goods must
sum to zero. This makes sense because the income has not changed. The first term in brackets is
the change in expenditure on good 1, and the second term (the sum) is the change in
expenditures on all other goods.

8. Fuel Efficiency
Individuals gain utility from vehicle miles traveled, 𝑣, vehicle safety, 𝑆, and some other good, 𝑥. Assume
further that automobile safety is a decreasing function of fuel efficiency (miles per gallon, 𝑀𝑃𝐺). This is
plausible as, ceteris paribus, making a car safer often requires the designer to add extra weight, reducing
the fuel efficiency. Assume that consumers can choose the level of fuel efficiency they want for their car.
The price of fuel is 𝑝𝑔 and the price of 𝑥 is 𝑝𝑥 . For the purposes of this problem, assume that there are
no costs other than fuel costs associated with owning and operating the vehicle.

a) Set up the individual’s utility maximization problem and derive the system of equations that
describes the individual’s optimal choice of 𝑣, 𝑀𝑃𝐺, and 𝑥.

𝑝𝑔
max ℒ = 𝑈(𝑣, 𝑆(𝑀𝑃𝐺 ), 𝑥 ) + 𝜆 (𝑀 − 𝑣 − 𝑝𝑥 𝑥)
𝑣,𝑀𝑃𝐺,𝑥,𝜆 𝑀𝑃𝐺

The price of miles driven is the price of gas divided by the MPG, or, the real price of driving an
additional mile. Taking FOCs, we have

𝑝𝑔
ℒ𝑣 = 𝑈𝑣 − 𝜆 ∙ =0
𝑀𝑃𝐺
𝑝𝑔 𝑣
ℒ𝑀𝑃𝐺 = 𝑈𝑆 𝑆𝑀𝑃𝐺 + 𝜆 ∙ ∙ =0
𝑀𝑃𝐺 𝑀𝑃𝐺
ℒ𝑥 = 𝑈𝑥 − 𝜆𝑝𝑥 = 0
𝑝𝑔
ℒ𝜆 = 𝑀 − 𝑣 − 𝑝𝑥 𝑥 = 0
𝑀𝑃𝐺

b) Using this system of equations, explain in words the tradeoff individuals make when they
choose vehicles with relatively high fuel efficiency.

Manipulating the first two equations above yields

𝑣
𝑈 = −𝑈𝑆 𝑆𝑀𝑃𝐺
𝑀𝑃𝐺 𝑣

Now note that vehicle miles travelled is the product of gallons of gas and miles per gallon,
v=g∙MPG. Making this substitution, we get

𝑔𝑀𝑃𝐺 𝜕𝑈
( ) = −𝑈𝑆 𝑆𝑀𝑃𝐺
𝑀𝑃𝐺 𝜕𝑔𝑀𝑃𝐺

Holding the gallons of gas constant, this reduces to

𝑈𝑀𝑃𝐺 = −𝑈𝑆 𝑆𝑀𝑃𝐺

This tells us that individuals choose fuel efficiency so as to equate the marginal utility of
increasing fuel efficiency with the marginal disutility of the decline in automobile safety that
arises because of that increase in fuel efficiency. Thus, individuals trade off the benefit of being
able to drive more miles on a gallon of gas with its costs in terms of decreased automobile
safety.

c) Explain how you would derive a formal expression for how the demand for fuel efficiency
responds to an increase in the price of gas. Can you put a sign on this price effect
unambiguously? Explain your answer.
 (i) Totally differentiate the FOCs with respect to all endogenous variables (𝑥, 𝑣, 𝑐, 𝑀𝑃𝐺, 𝑎𝑛𝑑 𝜆)
and exogenous variables (𝑀, 𝑝𝑔 , 𝑝𝑥 ). This translates the FOCs into a system of equations that
describes how a change in the exogenous variables relates to a change in the endogenous
variables. (ii) Express this new system of equations in matrix form. (iii) Use Cramer’s Rule to solve
for 𝑑𝑀𝑃𝐺. (iv) Holding other exogenous variables constant (e.g., setting dM and 𝑑𝑝𝑥 to zero),
derive an expression for 𝑑𝑀𝑃𝐺 ∗ /𝑑𝑝𝑔 .

The sign of this price effect cannot be determined without making further assumptions.
Assuming all goods (x, v, and S) are normal, when the price of gas increases, consumers can
adjust in two different ways. Holding MPG constant, they can substitute away from v toward x.
Alternatively, they can increase MPG (therefore diminishing the effect of the increase in 𝑝𝑔 on
the real price of v), but at the cost of decreasing S. Whether consumers will reduce v, increase
MPG (and so decrease S), or do both, cannot be determined a priori.

d) Now suppose that the U.S. government decides to regulate fuel efficiency standards in an effort
to reduce CO2 emissions. Use your answer above to explain why mandated increases in fuel
efficiency might not cause CO2 emissions to fall. You can assume that aggregate CO2 emissions
are a linear function of fuel consumption.

An increase in fuel efficiency lowers the price of vehicle miles travelled and so causes individuals
to drive more than they otherwise would. How this increase in vehicle miles travelled affects
aggregate CO2 emissions depends on the price elasticity of demand for gasoline. A doubling of
MPG lowers the effective price of vehicle miles by half. If the price elasticity of demand for miles
travelled is unitary, then a decrease in the price of vehicle miles travelled by half will lead to a
doubling of demand for vehicle miles travelled. The demand for gasoline would then remain
constant, and CO2 emissions would be unchanged. If the price elasticity of demand is inelastic,
then a decrease in the price by half would lead to a less than doubling of the demand; the
demand for gasoline, and thus CO2 emissions, would fall. If the price elasticity of demand is
elastic, then a decrease in the price would lead to a more than doubling of demand, increasing
emissions.

9. Natural Disaster Preparedness

Hurricane Laura has reminded cities around the country that they should explore ways to increase the
level of disaster preparedness among their residents. This question asks you to analyze two policy
mechanisms commonly used by governments to incentivize disaster preparedness.

Assume residents of Los Angeles receive utility from individual disaster preparedness, 𝑑, and other
goods, 𝑥. The price of one unit of disaster preparedness is 𝑝 and the price of other goods is normalized
to 1. Also assume that residents receive exogenous income 𝑀.

a) Write down the resident’s optimization problem, assuming a Cobb-Douglas utility function.

Applying a log transformation,

 max ℒ = 𝛽 ln 𝑑 + (1 − 𝛽 ) ln 𝑥 + 𝜆(𝑀 − 𝑝𝑑 − 𝑥 )
𝑑,𝑥,𝜆

b) Solve for the optimal choices of 𝑑 and 𝑥.

Taking FOCs and solving the system of equations leads to

𝛽𝑀
𝑑 ∗ (𝑝, 𝑀) =
𝑝
𝑥 ∗ (𝑝, 𝑀) = (1 − 𝛽 )𝑀

c) Show formally how the demand for disaster preparedness responds to income and its own
price.

Take the derivative of d* with respect to its own price, p:

𝜕𝑑 ∗ 𝛽𝑀
=− 2 <0
𝜕𝑝 𝑝

d) Suppose there are two types of residents, those with high incomes, 𝑀𝐻 , and those with low
incomes, 𝑀𝐿 . Use your answer above to explain whether high-income or low-income residents
will be more prepared for natural disasters (i.e., which group will purchase a greater quantity of
disaster preparedness?).

Given the Cobb-Douglas utility function, disaster preparedness is a normal good:

𝜕𝑑 ∗ 𝛽
= >0
𝜕𝑀 𝑝

Therefore, higher income residents will spend more on disaster preparedness than lower
income residents.

e) Now suppose the city of Los Angeles decides to subsidize residential spending on disaster
preparedness. (Go back to assuming only one type of resident with income 𝑀.) The subsidy rate
(the proportion of the good’s cost paid by the government) is 𝑠, where 0 < 𝑠 < 1. How does the
subsidy change the resident’s budget constraint? How does the subsidy change the optimal
choice of 𝑑? Does the subsidy also change the optimal choice of 𝑥?

The subsidy has the effect of lowering the price of disaster preparedness from p to (1-s)p. So, the
budget constraint becomes 𝑀 = (1 − 𝑠)𝑝𝑑 + 𝑥. Since disaster preparedness increases as its
price declines, the quantity of preparedness will increase with the subsidy.

The demand for x is not a function of the price of disaster preparedness (because of the property
of a Cobb-Douglas utility function), so the quantity of x demanded will not change.
f) Graphically, show the resident’s budget constraint for units of disaster preparedness and other
goods, before and after the introduction of the subsidy. Label your graph (i.e., note the slope of
each budget line, the endpoints, and axes). Note the optimal choice before the subsidy as 𝑑 𝑂
and the optimal choice with the subsidy as 𝑑 𝑆 . Discuss the income and substitution effects
associated with the introduction of the subsidy.

The effect of the subsidy is to shift the budget constraint out along the d axis, since more d can
be purchased when a subsidy is offered than when it is not offered. To find the income and
substitution effects, ask, “How would the demand for d change if the consumer faced the new
set of prices (1-s)p but was constrained to remain at the same level of utility associated with the
old price p?” To find this quantity, find the tangency of the new budget constraint to the old
indifference curve.

The income effect is found by shifting the budget constraint associated with the new prices from
its tangency with the old indifference curve to the new curve. This parallel shift holds prices
constant but allows income to change.

g) Now suppose the city replaces the subsidy program with a new program that gives the resident
a lump sum transfer, 𝑇, which must be spent on disaster preparedness (assume perfect
enforcement). Write down the resident’s optimization problem under the transfer program (you
don’t have to solve the optimization problem).

The resident now faces an additional budget constraint, which is that they must spend the
transfer T on disaster preparedness; in other words, 𝑝𝑑 ≥ 𝑇.

max 𝛽 ln 𝑑 + (1 − 𝛽 ) ln 𝑥
𝑑,𝑥
𝑀 + 𝑇 = 𝑝𝑑 + 𝑥
𝑝𝑑 ≥ 𝑇

h) Show graphically the resident’s budget constraint under the new transfer program, under the
old subsidy program, and under no program at all. Assume each resident’s transfer equals the
amount the city gave the resident under the old subsidy program. Again, label your graph, and
indicate the optimal choice under the new transfer program as 𝑑 𝑇 . Does the resident purchase
more or fewer units of 𝑑 compared to the subsidy program? Discuss the substitution and
income effects created by switching to the transfer program from the subsidy program.|

If the amount of the transfer is based on the amount of subsidy that consumers received under
the subsidy program, it must be the case that the budget constraint under the transfer program
passes through the optimal solution with the subsidy in place. In other words, the transfer
program would allow the resident to purchase the same amount of d as under the subsidy
program. However, the consumer will not choose to do so, because the transfer income will also
create more demand for x. Therefore, the transfer program does not increase the demand for
disaster preparedness as much as the subsidy program does.
i) In part d), we examined the difference in disaster preparedness between residents with
different income levels. If the city especially cares about raising the level of disaster
preparedness among its low-income residents, use your answer above to discuss whether the
city should or should not base the lump-sum transfer on the amount of the subsidy paid to each
resident under the old program.

Basing the transfer on the value of the subsidy will result in a lower transfer of funds to the poor
than to the rich. If the city wants to close the gaps between the rich and poor in disaster
preparedness, it should base the transfer on some metric that is not directly tied to pre-transfer
demand for d.