ECON-SHU 3 MICROECONOMICS

Practice Questions

Xiaoye Liao Eric Set Danyan Zha

Updated April 2021

1
Contents
Contents
1 Consumer Theory and Demand 3

2 Producer Theory and Competitive Supply 32

3 Competitive Markets, Market failure, and Economic Policies 57

4 Monopoly and Pricing Strategies 76

5 Choice under Uncertainty 95

6 Introductory Game Theory and Oligopoly 113

7 Information Economics and Mechanism Design 145

2
 1

Consumer Theory and Demand

1 Consumer Theory and Demand

3
Exercise 1. Define 𝑈 = 3 + 2𝑋 − 𝑌 + 5𝑍. Given this utility function, answer the following questions with
(𝑋, 𝑌 , 𝑍) = (5, 3, 4).

(a) Find the utility level;

(b) Find the value (a real number) of the partial derivative of 𝑈 with respect to 𝑋;
(c) Find the value of the partial derivative of 𝑈 with respect to 𝑌 .

Exercise 2. Define 𝑈 = 4𝑋 7𝑌 8 . Given this utility function, answer the following questions with 𝑋 = 2
and 𝑌 = 1.

(a) Compute the value of the marginal utility of 𝑋;

(b) Find the value of the marginal rate of substitution (i.e., 𝑀𝑈𝑋 /𝑀𝑈𝑌 ).

Exercise 3. Suppose there are three options, 𝐴, 𝐵, and 𝐶. A preference relation % over the options is
given as:
𝐴 % 𝐵, 𝐴 % 𝐶, 𝐵 % 𝐶, 𝐶 % 𝐴, 𝐴 % 𝐴.

(a) Is the preference relation complete? Justify your answer.

(b) Is the preference relation transitive? Justify your answer.

Exercise 4. Assume that one’s utility function over consumption 𝐶 and leisure 𝐿 is

𝑈 (𝐶, 𝐿) = ln 5 + 0.6 ln 𝐶 + 0.4 ln 𝐿.

When 𝐶 = 10 and 𝐿 = 5,

(a) what is the value of the marginal utility of consumption?

(b) what is the marginal rate of substitution (i.e., 𝑀𝑈𝐶 /𝑀𝑈 𝐿 )?

Exercise 5. Assume that Jim’s utility function over leisure 𝐿 and consumption 𝐶 is 𝑈 = 𝐶 𝐿. Let 𝐻
stand for hours spent working. Then we have 𝐿 = 24 − 𝐻, that is, leisure is time not working. When
(𝐶, 𝐻) = (16, 8),

(a) using the chain-rule, calculate the partial derivative of 𝑈 with respect to 𝐻 at the current level of
consumption and hours worked;
(b) find MRS𝐶 𝐿 (i.e., 𝑀𝑈𝐶 /𝑀𝑈 𝐿 ).
√
Exercise 6. Define the utility function 𝑈 = 𝑋 + 𝑌 . Let (𝑋, 𝑌 ) = (8, 16). Find the value of the marginal
rate of substitution MRS𝑋𝑌 (i.e., 𝑀𝑈𝑋 /𝑀𝑈𝑌 ).
This is an example of a quasi-linear utility function. X is usually money (the idea is “everything else you
could buy if not Y”), and Y is the main thing you are interested in describing (imagine using statistics to

4
estimate the parameters). Baked into the function is the comparison of “how much money is one more unit
of Y worth of me.” So, while utility is not really about intensity and interpersonal comparison, it’s very
common to directly include money as a comparison good when the context allows for it (when people
would be okay with non-diminishing marginal utility for money).
Exercise 7. Suppose 𝑈 = 𝑋 0.4𝑌 0.8 . As we have seen in the slides, with a simple budget, these Cobb-
Douglas preferences yields a demand function for 𝑋 of

0.4 𝐼
𝑋= · ,
0.4 + 0.8 𝑃 𝑋

where 𝐼 is the income and 𝑃 𝑋 is the price of 𝑋. For concreteness, suppose 𝑃 𝑋 = 6 and 𝐼 = 100. I asserted
that with Cobb-Douglas preferences and the simple budget problem, the proportion of the budget spent on
goods 𝑋 and 𝑌 are unaffected by either prices or the income. Using this property, find the price elasticity
of demand for consumers whose preferences are Cobb-Douglas. Make sure your answer is a negative
number.

Exercise 8. We now take a standard approach to define, based on a preference relation %, the indifference
relation ∼ and the strict preference relation . In particular, we define 𝐴 ∼ 𝐵 if 𝐴 % 𝐵 and 𝐵 % 𝐴, and
define 𝐴  𝐵 if 𝐴 % 𝐵 is true but 𝐵 % 𝐴 is false.
(a) Define the relation ≺ (i.e., strictly worse than), that is, give a definition for 𝐴 ≺ 𝐵 based on %, using
the way introduced above.
(b) Using the notation and definitions developed above, along with the axioms (for preference relations),
prove each of the following:

(i) If 𝐴 ∼ 𝐵 and 𝐵 ∼ 𝐶, then 𝐴 ∼ 𝐶, that is, the indifference relation is transitive;

(ii) If 𝐴  𝐵 and 𝐵  𝐶, then 𝐴  𝐶, that is, the strict preference relation is also transitive;
(iii) If 𝐴 ∼ 𝐵 and 𝐵  𝐶, then 𝐴  𝐶.

Exercise 9. Ann’s preferences over goods 𝑋 and 𝑌 are represented by the utility function

𝑈 (𝑥, 𝑦) = 6𝑥 2 𝑦 2 .

Her total budget 𝐼 = 360, and the prices are 𝑝 𝑥 = 1 and 𝑝 𝑦 = 2. Find Ann’s optimal consumption bundle.
Exercise 10. Kevin’s utility on good 𝑋 and good 𝑌 is given by

𝑈 (𝑥, 𝑦) = 3𝑥 + 4𝑦.

His total budget 𝐼 = 240. The price of good 𝑋 is 1 and that of good 𝑌 is 8. Let (𝑥 ∗ , 𝑦 ∗ ) denote Kevin’s
optimal consumption bundle. Find (𝑥 ∗ , 𝑦 ∗ ).

5
Exercise 11. Assume that one’s preferences are represented by the following Leontief utility function

𝑈 (𝑥, 𝑦) = min{𝑎𝑥, 𝑦},

where 𝑎 > 0, 𝑥 is the quantity of good 𝑋, and 𝑦 is the quantity of good 𝑌 .

(a) If 𝑎 = 5, how many units of 𝑋 are needed in order to compensate for the loss of 10 units of good 𝑌
at the consumption bundle (𝑥, 𝑦) = (0, 17) (so that the utility level is unchanged)?
(b) Suppose that 𝑎 = 3 and the budget constraint is 6𝑥 + 𝑦 = 36, 𝑥, 𝑦 > 0. Find the utility-maximizing
quantity of good 𝑌 .

Exercise 12. Bob’s utility over good 𝑋 and good 𝑌 is captured by the function

𝑈 (𝑥, 𝑦) = 𝑎 ln 𝑥 + 6 ln 𝑦 + 5,

where 𝑎 > 0. The prices are 𝑝 𝑥 = 2 and 𝑝 𝑦 = 3, respectively. Bob’s total budget 𝐼 > 0.

(a) If the optimal consumption bundle is (31, 31), find the values for 𝑎 and 𝐼, respectively;
(b) If 𝑎 = 2 and 𝐼 = 120, find the optimal quantity of good 𝑋.

Exercise 13. Donald’s preferences over good 𝑋 and good 𝑌 are represented by the utility function

𝑈 (𝑥, 𝑦) = 𝑥 2 (2𝑦 + 1).

His total income 𝐼 = 160. The price of 𝑋 is 10 and that of 𝑌 is 10. Find the optimal consumption bundle.
Exercise 14. The axioms of consumer preferences are given below for reference. Let % be the preference
relation under consideration.

• Completeness: For all consumption bundles 𝐴 and 𝐵, either 𝐴 % 𝐵 or 𝐵 % 𝐴.

• Transitivity: For all consumption bundles 𝐴, 𝐵, and 𝐶, if 𝐴 % 𝐵 and 𝐵 % 𝐶, then 𝐴 % 𝐶 must be
true.
• Strict Monotonicity: More is better, that is, if 𝐴 and 𝐵 are real vectors such that there is more of
everything in 𝐴 than 𝐵, then 𝐴  𝐵.
• Convexity: If 𝐴 % 𝐵 then 𝑡 𝐴 + (1 − 𝑡)𝐵 % 𝐵, where 0 6 𝑡 6 1.

Use the axioms to answer the following questions.

(a) Argue that if % is complete, then for each consumption bundle 𝐴 we must have 𝐴 % 𝐴.
(b) Let % satisfy all of the four axioms above. Show that if 𝐴 % 𝐶 and 𝐵 % 𝐶, then 𝑡 𝐴 + (1 − 𝑡)𝐵 % 𝐶
for all 0 6 𝑡 6 1.
(c) Let 𝐴 = (3, 4), 𝐵 = (6, 4), and 𝐶 = (9, 4). Why would 𝐶  𝐴  𝐵 specifically violate one of the
axioms above? Justify your answer carefully.

6
 (d) Refer to the diagram below. Suppose that bundle 𝐴 is at the intersection of two distinct indifference
curves 𝐼1 and 𝐼2 , which capture a preference relation % that is complete. Argue that this will lead to
a violation of some of the four axioms.

Good 𝑌

𝐴
𝐼1
𝐼2

𝑂 Good 𝑋

Exercise 15. Find the price elasticity of each of the following demand functions at the given price.
4
(a) 𝑄( 𝑝) = at 𝑝 = 7;
𝑝4
(b) ln 𝑄( 𝑝) = 162 − 6 ln 𝑝 at 𝑝 = 13;

(c) 𝑄( 𝑝) = 100 − 2𝑝 2 at 𝑝 = 5.

Exercise 16. Tom has a monthly budget of $80 to spend on bubble tea and burritos. Burritos cost $8 each
and bubble tea costs $4 each. To celebrate the anniversary of the bubble tea shop, the owner decides to
offer a loyalty program that allows a consumer to buy bubble tea at $2 each once she has purchased 6 cups
of bubble tea in that month.

(a) What is the maximum quantity of bubble tea Tom can afford this month?
(b) Put bubble tea on the horizontal axis and burritos on the vertical axis. What are the coordinates of
the kinky point of Tom’s budget line?

Exercise 17. A family has a yearly income of $80, 000 to spend on the kids’ education and all other goods
and services. Put dollars spent on education on the horizontal axis and that on all other goods and services
on the vertical axis.

(a) Find the horizontal and vertical intercepts of the family’s budget line.
(b) If the family has to pay an income tax at the rate 10%, find the horizontal and vertical intercepts of
the family’s budget line in this case.

7
 (c) If the family has to pay a 10% income tax but receives a school voucher of $8, 000 that can only
be used to cover education expenditure, find the horizontal and vertical intercepts of the family’s
budget line in this case.

Exercise 18. Sara has utility for protein bars (good 𝑋) and vitamin water (good 𝑌 ) summarized by the
Cobb-Douglas function 𝑈 (𝑋, 𝑌 ) = 𝑋𝑌 . Her income is 80, the price of protein bars 𝑝 𝑋 = 1, and the price
of vitamin water 𝑝𝑌 = 1.

(a) Find the optimal consumption bundle for Sara. What is her utility level at the optimal consumption
bundle?
(b) Give the equation for Sara’s income-consumption curve.

Now let us analyze the substitution effect and income effect when the price of protein bars changes.
Suppose that the price of protein bars increases to 4.

(c) At the new prices, find Sara’s optimal consumption bundle;

(d) Take as given that the substitution effect of the price change on vitamin water is +40 (i.e., Sara
will consume 40 units more vitamin water if there is only substitution effect due to the increase
of the price of protein bars), find the corresponding income effect on vitamin water as well as the
substitution and income effects on protein bars, respectively.

Exercise 19. Andrew’s preferences over good 𝑋 and good 𝑌 are represented by the utility function

𝑈 (𝑥, 𝑦) = 9𝑥 𝛼 𝑦 𝛽 + 8,

where 𝑥 (𝑦, respectively) represents the quantity of good 𝑋 (𝑌 , respectively). The prices are 𝑝 𝑥 = 4 and
𝑝 𝑦 = 8, which are constant throughout. Andrew’s income 𝐼 > 0 and he (optimally) spends one third of his
income on good 𝑌 .

(a) If 𝛽 = 5, find the value for 𝛼.

(b) When Andrew (optimally) spends 300 on good 𝑌 , how many units of good 𝑋 does he consume?
(c) For Andrew’s consumption of good 𝑋 to increase by 8 units, what is the corresponding increase in
his income?
(d) What is the cross-price elasticity of Andrew’s demand for good 𝑌 when 𝐼 = 93? (Your answer
should be a number.)

8
Exercise 20. Assume that Bernie’s utility over good 𝑋 and good 𝑌 is given by

𝑈 (𝑥, 𝑦) = 3𝑥 + 5𝑦 + 6.

His income 𝐼 = 120 (which is constant throughout) and the initial prices are 𝑝 𝑥 = 4 and 𝑝 𝑦 = 6.

(a) What is Bernie’s optimal consumption bundle under the initial prices?
(b) One day the price of 𝑌 drops to 5. What are the substitution and income effects of this price change
on 𝑌 , respectively?
(c) Later the price of 𝑋 drops to 2 (the price of 𝑌 is still 5). The situation in (b) being the reference
point, find the substitution and income effects of this price change on 𝑋, respectively.

Exercise 21. There are 100 consumers on the market for good 𝑍, each having the individual demand

𝑞 𝑑 = 2.2 − 0.02𝑝 + 0.005𝑄 𝑑 ,

where 𝑝 is the price of good 𝑍 and 𝑄 𝑑 is the quantity demanded by the whole market (i.e., the market
demand). The supply of 𝑍 is given by 𝑄 𝑠 = 200 + 6𝑝.

(a) What is the slope of the market demand curve (i.e., d𝑄 𝑑 /d𝑝)? What is the maximum price for this
market?
(b) Find the market equilibrium price 𝑝 ∗ and quantity 𝑄 ∗ .
(c) What is the (aggregate) consumer surplus at the market equilibrium?

Exercise 22. Eric really likes milk tea, but he is indifferent between brands: Milk tea from TPlus (good 𝑇)
is just as good as milk tea from Kingyo (good 𝐾) to him. A cup of milk tea is sold for $3 in TPlus and $2
in Kingyo.

(a) If 𝐾 is on the vertical axis, what is the slope of Eric’s indifference curve when he consumes 8 cups
of 𝑇 and 9 cups of 𝐾?

For the following questions, Eric’s weekly milk tea budget is $60.

(b) How many cups of milk tea could Eric buy if he spent his entire budget on TPlus?
(c) Find the optimal quantity of 𝐾 for Eric per week.
(d) In a week, Eric would like to try the delivery service of milk tea. The delivery fee for TPlus is $1 per
cup and for Kingyo is $3 per cup. Assume that Eric’s total milk tea budget, including the delivery
fee, is still $60. What is the optimal quantity of 𝑇 for Eric in that week?

9
Exercise 23. Kate is a big fan of sports. Her total income per month is $100. When the price is $4 per
hour, she spends 10 hours at the gym a month.

(a) The gym now provides an alternative plan. Instead of paying $4 per hour, one can pay a membership
fee of $30 per month to get 25% off for the usage fee (i.e., only paying $3 per hour). Which plan will
Kate choose? Explain using a graph.
(b) The gym later cancels the original plan (i.e., $4 per hour and no membership fee) and only provides
a plan that charges a membership fee of $30 per month and $1 per hour for that month. After the
price change, Kate chooses to spend 20 hours at the gym per month. Is Kate better off or worse off?
Explain. Use a graph if necessary.

Exercise 24. Suppose the demand curve of good 𝑍 is 𝑄 𝑑 = 𝑎 − 𝑏 𝑝, and its supply curve is 𝑄 𝑠 = 2𝑝 − 𝑇,
where 𝑝 is the price of good 𝑍. Here 𝑇 is a measure of technology in the industry, and a lower 𝑇 means
more advanced technology. The current technology 𝑇 = 4.
For (a)-(d), assume that 𝑎 = 60 and 𝑏 = 6.

(a) What is the price of good 𝑍 when the price elasticity of demand equals −9?
(b) Compute the market equilibrium price 𝑝 ∗ and quantity 𝑄 ∗ . What is the consumer surplus at the
market equilibrium?
(c) Is the demand at equilibrium elastic, inelastic, or unit elastic? Explain.
(d) For the equilibrium price to decrease by 0.3, what level must the industry technology reach?

Now consider the following market structure: The market demand 𝑄 𝑑 is obtained by aggregating 100
individual demands, each described by

𝑞 𝑑 = 0.3 − 0.04𝑝 + 0.005𝑄 𝑑 .

(e) With the market structure, find the slope of the demand curve (i.e., d𝑄 𝑑 /d𝑝).
(f) The individual demand exhibits network externalities on the market of good 𝑍. According to the
individual demand, are the network externalities positive or negative? Give an example of good 𝑍
that has this property and justify your answer briefly.

Exercise 25. In a two-good society (the two goods are 𝑋 and 𝑌 . You may interpret 𝑋 as food and 𝑌 as
non-food goods and services), a researcher is conducting a case study on a consumer’s preferences and
purchasing behavior. For the sake of convenience, the researcher uses % (i.e., weakly preferred to) to
denote the consumer’s preference relation.

(a) Initially, the researcher hypothesizes that % is represented by a linear utility function

𝑈 (𝑥, 𝑦) = 3𝑥 + 12𝑦.

10
 Given this hypothesis, to compensate the consumer for a loss of one unit of good 𝑌 , how many units
of good 𝑋 is needed?
(b) For some reason the researcher abandons his hypothesis in (a) and hypothesizes instead that % is
represented by a Cobb-Douglas utility function

𝑈 (𝑥, 𝑦) = 𝐴𝑥𝑦 4 ( 𝐴 > 0),

based on which he predicts that when the consumer’s income 𝐼 = 240, 𝑝 𝑋 = 3, and 𝑝𝑌 = 3, the
consumer will choose the consumption bundle (𝑥 ∗ , 𝑦 ∗ ). Find (𝑥 ∗ , 𝑦 ∗ ).
(c) Later, the researcher realizes that there is something wrong with his prediction in (b) after seeing the
following datum about the consumer’s purchasing behavior:

Quantity of 𝑋 Quantity of 𝑌 𝑝𝑋 𝑝𝑌 𝐼
20 15 3 4 120

If the researcher still maintains that % is represented by a Cobb-Douglas utility function, he must
update his prediction in (b). Find the updated prediction for (𝑥 ∗ , 𝑦 ∗ ).

Now more data about the consumer’s purchasing behavior are available:

Quantity of 𝑋 Quantity of 𝑌 𝑝𝑋 𝑝𝑌 𝐼
75 48 8 12.5 1200
75 75 8 8 1200
60 60 8 8 960

Maintaining that % is Cobb-Douglas, the researcher finds the data sufficient to reveal the effect of 𝑝𝑌 ’s
dropping from 12.5 to 8.

(d) Fill in the blanks: According to the data, the total effect of the price change on good 𝑌 is ,
the income effect on good 𝑌 is 15 (i.e., 75 − 60), and the substitution effect on good 𝑌 is .
(e) According to the conclusion in (d), is 𝑌 normal/inferior/Giffen to the consumer? Briefly explain.

***********

Exercise 26. Three rational agents have different preferences over goods 𝑋 and 𝑌 :

Agent 1: 𝑈1 = 2(𝑋 − 1)𝑌 2 + 100

Agent 2: 𝑈2 = 3𝑋 2 (𝑌 − 1) − 100
Agent 3: 𝑈3 = 2 ln 𝑋 + 2 ln 𝑌

All have the same budget of 100. The prices are 𝑝 𝑋 = 𝑝𝑌 = 10.

11
 (a) What is the value of Agent 2’s marginal rate of substitution (i.e., 𝑀𝑈𝑋 /𝑀𝑈𝑌 ) at the utility-
maximizing bundle? (Your answer should be a number.)
(b) What is Agent 1’s utility-maximizing level of good 𝑌 ?
(c) What is Agent 3’s utility-maximizing level of good 𝑋?

Exercise 27. Jing’s preferences satisfy all four of our basic assumptions. She has strictly positive income
and 𝑃 𝑎 , 𝑃𝑏 > 0 (𝑃 𝑎 , 𝑃𝑏 are the prices of goods 𝑎 and 𝑏, respectively). Currently, she consumes at a
point such that 𝑀𝑈𝑎 /𝑃 𝑎 > 𝑀𝑈𝑏 /𝑃𝑏 . Can we conclude that Jing is maximizing her utility? If she is
maximizing her utility, explain why this is unusual. If she is not maximizing her utility, explain what
she should do to improve. If you cannot determine the answer from the information above, explain. You
should not refer to utility functions in your answer (i.e. don’t say “she should maximize her utility”).
Exercise 28. We are interested in modeling fairness in a simple task. A person (called the Dictator)
decides how to split money between himself and a random stranger. Let 𝑋 and 𝑌 denote the money given
to the Dictator and the random stranger, respectively. The proposed utility function of the Dictator is

𝑈 = 𝑋 + 𝑎𝑌 𝑏 ,

where 𝑎 > 0 and 0 < 𝑏 < 1.

(a) What is the marginal utility of 𝑋?

(b) Does this function exhibit diminishing marginal rate of substitution? Explain.
(c) Suppose the Dictator is asked to divide 1 dollar between himself and the stranger. How much do
they allocate to the stranger when (𝑎, 𝑏) = (1, 0.5)?
(d) Suppose that the Dictator has 1000 dollars to split and maintain that (𝑎, 𝑏) = (1, 0.5). How much
does the Dictator allocate to the stranger? Does this make sense to you? If not, would changing 𝑎
and/or 𝑏 help? Why/Why not?

Exercise 29. The San Francisco Chronicle reported that the toll on the Golden Gate Bridge was raised
from 2 to 3 dollars. Following the toll increase, traffic fell by 5%.

(a) Based on this information, calculate the (point) price elasticity of demand. You need to have the
correct sign.
(b) Is the demand elastic, inelastic, or unit elastic? Explain.
(c) Did toll revenue increase or decrease as a result of the toll increase?

Exercise 30. You believe that Lisa’s utility over good 𝑋 (protein bars) and good 𝑌 (cups of yogurt) takes
one of the following three possible forms:

𝑈 (𝑥, 𝑦) = min{𝑎𝑥, 2𝑦}

12
 𝑈 (𝑥, 𝑦) = 𝑎𝑥 + 𝑦,
𝑈 (𝑥, 𝑦) = 𝑥 𝑎 𝑦,

where 𝑎 > 0. Although she sometimes buys more or less, you discover that Lisa always buys 3 protein
bars for every 2 cups of yogurt, even when there are discounts.
(a) Find Lisa’s utility function (include the specific value of 𝑎).
(b) Suppose 𝑈 (𝑥, 𝑦) = min{3𝑥, 𝑦}. Derive Lisa’s demand function for yogurt (i.e., 𝑦 as a function of
prices 𝑝 𝑥 , 𝑝 𝑦 , and income 𝐼).

Exercise 31. Consider the standard consumer problem. For sufficiently high levels of income, preferences
can be captured by the utility function

𝑈 = (𝑥 − 𝑚) 𝑎 (𝑦 − 𝑛) 𝑏 ,

where 𝑥, 𝑦 are the quantities of goods 𝑋 and 𝑌 , respectively, and 𝑚, 𝑛, 𝑎, 𝑏 are strictly positive constants.
(a) Suppose 𝑎 = 0.3, 𝑏 = 0.2, 𝑝 𝑥 = 2 (the price of 𝑋), 𝑝 𝑦 = 3 (the price of 𝑌 ), and income is sufficiently
high. What is the utility-maximizing level of good 𝑌 as a function of 𝑚, 𝑛, and the utility-maximizing
level of good 𝑋? Your answer should be an explicit function.
(b) Again, if income is sufficiently high and 𝑎, 𝑏, 𝑚, and 𝑛 are all strictly positive constants, is the
cross-price elasticity of demand for 𝑋 with respect to the price of 𝑌 greater than, less than, or equal
to zero? Is it something else? Explain.
(c) Why would this not be a good utility function to use when income is not “sufficiently high”? Hint.
Imagine that 𝑚 = 𝑛 = 10 and 𝑎 = 𝑏 = 2.
(d) Give an economic interpretation to this by-now-familiar utility function.

Exercise 32. Two societies, East and West, are separated from each other and do not normally trade.
Coincidently, each society produces two identical and divisible goods, 𝑋 and 𝑌 .
• In East, the prices are 𝑝 𝐸
𝑋 = 5 and 𝑝𝑌 = 10 (in the local currency, say, RMB);
𝐸

• In West, the prices are 𝑝 𝑊

𝑋 = 100 and 𝑝𝑌 = 50 (in the local currency, say, USD).
𝑊

Although both societies exhibit diverse preferences and sometimes extreme income inequality, every
member of each society consumes at least some of both goods. Assume that everyone in either East or
West is rational and has preferences consistent with our axioms.
(a) Find the marginal rate of substitution (i.e., 𝑀𝑈𝑋 /𝑀𝑈𝑌 ) for a citizen of East and that for a citizen of
West at their optimal consumption bundle, respectively.
Quite randomly, a citizen of East (Sun Wukong) meets a citizen of West (Wonder Woman), both while
exploring. Also, quite conveniently, they are both carrying their optimal basket. Using the universal
language, someone proposes that they trade. Assume that trade is only motivated by their own consumption
preferences.

13
(b) There will be trade. Who trades what for what? Justify your answer.
(c) At the end of trading, both Sun Wukong and Wonder Woman have some positive levels of 𝑋 and 𝑌
in their baskets. Compare Wonder Woman’s utility before and after trading. Justify your answer.
(d) Wukong concludes trading with Wonder Woman and returns to East with a magical leap. He meets
his friend Bajie immediately after returning, still with their baskets of goods. Again, assuming
Wukong and Bajie trade (if any) only for consumption. Is there an opportunity for trade between
Wukong and Bajie to happen? Explain.

14
Chapter 1 Answer Key

1. (a) Plugging the data in, one gets 𝑈 (5, 3, 4) = 3 + 2 × 5 − 3 + 5 × 4 = 30.

(b) 𝜕𝑈/𝜕 𝑋 = 2, which does not depend on the values of 𝑋, 𝑌 , and 𝑍.
(c) 𝜕𝑈/𝜕𝑌 = −1, which does not depend on the values of 𝑋, 𝑌 , and 𝑍.

2. (a) By definition, 𝑀𝑈𝑋 = 𝜕𝑈/𝜕 𝑋 = 4(7𝑋 7−1 )𝑌 8 = 28𝑋 6𝑌 8 . So when (𝑋, 𝑌 ) = (2, 1),

𝑀𝑈𝑋 = 28 × 26 × 18 = 1792.

(b) To figure out the marginal rate of substitution, we still need to pin down 𝑀𝑈𝑌 . By definition,
𝑀𝑈𝑌 = 𝜕𝑈/𝜕𝑌 = 4𝑋 7 (8𝑌 8−1 ) = 32𝑋 7𝑌 7 , and so when (𝑋, 𝑌 ) = (2, 1),

𝑀𝑈𝑌 = 32 × 27 × 17 = 4096.

Therefore, the marginal rate of substitution

𝑀𝑈𝑋 1792 7
MRS𝑋𝑌 = = = = 0.4375.
𝑀𝑈𝑌 4096 16

3. (a) The preference relation is incomplete since we are not able to compare, for example, 𝐵 with
itself, since we do not have 𝐵 % 𝐵.
(b) The preference relation is NOT transitive, since we have both 𝐵 % 𝐶 and 𝐶 % 𝐴, but 𝐵 % 𝐴
is not present. (Recall that we need to check if 𝑋 % 𝑍 holds when both 𝑋 % 𝑌 and 𝑌 % 𝑍 are
present for all possible tuple (𝑋, 𝑌 , 𝑍).)

4. (a) By definition, the marginal utility of consumption

𝜕𝑈 0.6 3
𝑀𝑈𝐶 = = = .
𝜕𝐶 𝐶 5𝐶
Thus, when (𝐶, 𝐿) = (10, 5), 𝑀𝑈𝐶 = 0.6/10 = 0.06.
(b) The marginal utility of leisure

𝜕𝑈 0.4 2
𝑀𝑈 𝐿 = = = ,
𝜕𝐿 𝐿 5𝐿
and so, using our result in (a), the marginal rate of substitution

𝑀𝑈𝐶 0.6/𝐶 3𝐿
MRS𝐶 𝐿 = = = .
𝑀𝑈 𝐿 0.4/𝐿 2𝐶

15
 Thus, when 𝐶 = 10 and 𝐿 = 5, we have

3×5
MRS𝐶 𝐿 = = 0.75.
2 × 10

5. (a) Employing the chain-rule, one has (viewing 𝐿 as a function of 𝐻, i.e., 𝐿 = 24 − 𝐻)

𝜕𝑈 𝜕𝑈 d𝐿
= · = −𝐶.
𝜕𝐻 𝜕𝐿 d𝐻
|{z} |{z}
=𝐶 =−1

Thus, when (𝐶, 𝐻) = (16, 8), we have 𝜕𝑈/𝜕𝐻 = −16.

(b) Since 𝑀𝑈𝐶 = 𝐿, it follows that

𝑀𝑈𝐶 𝐿
MRS𝐶 𝐿 = = .
𝑀𝑈 𝐿 𝐶

Therefore, given (𝐶, 𝐻) = (16, 8), we deduce that 𝐿 = 24 − 8 = 16, and so MRS𝐶 𝐿 = 16/16 = 1.

6. Straightforwardly,

1 1 1
𝑀𝑈𝑋 = 𝜕𝑈/𝜕 𝑋 = 1, 𝑀𝑈𝑌 = 𝜕𝑈/𝜕𝑌 = 𝑌 − 2 = √ .
2 2 𝑌
So we have √
𝑀𝑈𝑋 1
MRS𝑋𝑌 = = √ = 2 𝑌.
𝑀𝑈𝑌 1/(2 𝑌 )
√
When (𝑋, 𝑌 ) = (8, 16), MRS𝑋𝑌 = 2 16 = 8.

7. Let the utility function be a general Cobb-Douglas function 𝑈 = 𝑋 𝛼𝑌 𝛽 , where 𝛼, 𝛽 > 0. In a simple
budget problem, consumer must exhaust the budget, and so 𝑃 𝑋 · 𝑋 ∗ + 𝑃𝑌 · 𝑌 ∗ = 𝐼 (𝑋 ∗ and 𝑌 ∗ are the
demands for good 𝑋 and 𝑌 , respectively). According to the “fixed-proportion property",

𝛼 𝐼 𝛼𝐼
𝑋∗ = · = ,
𝛼 + 𝛽 𝑃𝑋 (𝛼 + 𝛽)𝑃 𝑋

𝛽 𝐼 𝛽𝐼
𝑌∗ = · = .
𝛼 + 𝛽 𝑃𝑌 (𝛼 + 𝛽)𝑃𝑌
Therefore, the price elasticities for 𝑋 and 𝑌 are

𝜕 𝑋 ∗ 𝑃𝑋 𝛼𝐼 𝑃𝑋
𝜀𝑋 = · ∗ =− 2
· = −1,
𝜕𝑃 𝑋 𝑋 (𝛼 + 𝛽)𝑃 𝑋 𝛼𝐼/[(𝛼 + 𝛽)𝑃 𝑋 ]
𝜕𝑌 ∗ 𝑃𝑌 𝛽𝐼 𝑃𝑌
𝜀𝑌 = · ∗ =− 2
· = −1,
𝜕𝑃𝑌 𝑌 (𝛼 + 𝛽)𝑃𝑌 𝛽𝐼/[(𝛼 + 𝛽)𝑃𝑌 ]

which do not rely on the values of 𝑋 and 𝑌 . This is an example of iso-elastic demand, that is, the

16
 (price) elasticity is constant. In particular, since the elasticity is constantly equal to −1, we also call
it unit elastic.

8. (a) We can define the relation ≺ as: 𝐴 ≺ 𝐵 if 𝐵 % 𝐴 is true but 𝐴 % 𝐵 is false.

(b) Proof.For (i), note that 𝐴 ∼ 𝐵 and 𝐵 ∼ 𝐶 imply that (1) 𝐴 % 𝐵 and 𝐵 % 𝐶, and (2) 𝐵 % 𝐴 and
𝐶 % 𝐵. Since % is transitive, we learn from (1) that 𝐴 % 𝐶 and from (2) that 𝐶 % 𝐴. Then, by
definition, 𝐴 ∼ 𝐶.
For (ii), note that 𝐴  𝐵 and 𝐵  𝐶 yield that (1) 𝐴 % 𝐵 and 𝐵 % 𝐶, but (2) neither 𝐵 % 𝐴 nor
𝐶 % 𝐵 holds. By (1), 𝐴 % 𝐶, and it remains to show that 𝐶 % 𝐴 is false. By contradiction, if
𝐶 % 𝐴 held, then we would have, using 𝐴 % 𝐵 and transitivity, that 𝐶 % 𝐵, which contradicts
(2). Thus, the desired result follows.
For (iii), the condition tells that (1) 𝐴 % 𝐵, 𝐵 % 𝐶, and 𝐵 % 𝐴, while (2) 𝐶 % 𝐵 is false. By (1)
and transitivity we obtain 𝐴 % 𝐶. To exclude 𝐶 % 𝐴, we again prove by contradiction. Suppose
that 𝐶 % 𝐴. Then by (1) (and transitivity) we have 𝐶 % 𝐵, which contradicts (2). 

9. Standard approach: The optimal consumption bundle (𝑥 ∗ , 𝑦 ∗ ) is characterized by the following

system
∗ ∗
 𝑝𝑥 · 𝑥 + 𝑝𝑦 · 𝑦 = 𝐼




𝑀𝑈 𝑥 𝑝𝑥

 =
( 𝑥,𝑦)=( 𝑥 ∗ ,𝑦 ∗ )
 𝑀𝑈 𝑝𝑦
 𝑦

Since 𝑀𝑈 𝑥 = 12𝑥𝑦 2 and 𝑀𝑈 𝑦 = 12𝑥 2 𝑦, the system can be further reduced to

∗ ∗
 𝑥 + 2𝑦 = 360




𝑦∗ 1
 ∗ =


𝑥 2
Solving which we obtain 𝑥 ∗ = 180 and 𝑦 ∗ = 90.
Shortcut: Since the utility function is Cobb-Douglas, we know that the portion of budget that should
be optimally allocated to each good equals its “weight”, that is, its power divided by the aggregate
power. Here, the power (or “importance”) of 𝑋 is 2 and so is that of 𝑌 . Thus, 2/(2 + 2) = 1/2 of
Ann’s total budget should be spent on 𝑋 and the other half should be spent on 𝑌 . Thus, we have
𝑝 𝑥 · 𝑥 ∗ = 360/2 and 𝑝 𝑦 · 𝑦 ∗ = 360/2, which yields that 𝑥 ∗ = 180 and 𝑦 ∗ = 90.

10. Approach 1: Note that we can use the budget constraint 𝑥 + 8𝑦 = 240 to eliminate 𝑥 in the objective
function (i.e., the utility function) and so it becomes 3(240 − 8𝑦) + 4𝑦, which can be simplified to
720 − 20𝑦. Thus, the consumer’s problem is reformulated as

max 720 − 20𝑦 s.t. 0 6 𝑦 6 30,

𝑦

where 30 is the maximum units of good 𝑌 Kevin can afford (i.e., 𝐼/𝑝 𝑦 ). Since the objective function
is decreasing in 𝑦, its maximum is achieved when the value of 𝑦 is minimized. Thus, 𝑦 ∗ = 0. So

17
 Kevin should optimally spend all of his budget on good 𝑋, and so 𝑝 𝑥 · 𝑥 ∗ = 𝐼, from which we solve
𝑥 ∗ = 240.
Approach 2: Consider the marginal utility obtained by spending one unit of the budget on good 𝑋
and that on good 𝑌 . Clearly, one unit of budget can buy 1/𝑝 𝑥 = 1 unit of 𝑋 and so the increased
utility equals 4 × 1 = 3; however, one unit of budget can by 1/𝑝 𝑦 = 0.5 unit of 𝑌 and so the increased
utility equals 4 × 0.5 = 2. Thus, 𝑋 constantly yields higher utility per dollar than 𝑌 , and so in
optimum Kevin will only consume 𝑋. That is, 𝑥 ∗ = 240 and 𝑦 ∗ = 0.

11. (a) At (0, 17), it is easy to see that the utility level is 𝑈 (0, 17) = min{5 × 0, 17} = 0. If the
quantity of 𝑌 decreases by 10, the corresponding consumption bundle becomes (0, 7). Since
𝑈 (0, 7) = min{5 × 0, 7} = 0, we see that the utility level is the same before and after the change,
and so no additional 𝑋 is needed to maintain the original utility level.
(b) Since the utility function is Leontief, the following system characterizes the optimal consumption
bundle (𝑥 ∗ , 𝑦 ∗ ): (
3𝑥 ∗ = 𝑦 ∗
6𝑥 ∗ + 𝑦 ∗ = 36
Solving this system, we obtain 𝑥 ∗ = 4, 𝑦 ∗ = 12.

12. (a) Since Bob always optimally exhausts his budget, we have 𝑝 𝑥 · 𝑥 ∗ + 𝑝 𝑦 · 𝑦 ∗ = 𝐼, that is,

𝐼 = 2 × 31 + 3 × 31 = 155.

Since the utility function is logarithmic, we know that in optimum a fixed proportion of the
budget should be spent on each good, and the proportion equals its “weight”. So we have

𝑎 𝑝 𝑥 · 𝑥 ∗ 2 × 31 2
= = = ,
6 𝑝 𝑦 · 𝑦 ∗ 3 × 31 3

from which we obtain 𝑎 = 4.

(b) Here we use the “fixed-proportion property”, that is,

2
𝑝 𝑥 · 𝑥∗ = ·𝐼

 (
2𝑥 ∗ = 30



 2+6
⇒
 6 3𝑦 ∗ = 90
 𝑝 𝑦 · 𝑦∗ =

 ·𝐼
 2+6
from which we solve 𝑥 ∗ = 15, 𝑦 ∗ = 30.

13. Since 𝑀𝑈 𝑥 = 2𝑥(2𝑦 + 1) and 𝑀𝑈 𝑦 = 2𝑥 2 , the optimal consumption bundle (𝑥 ∗ , 𝑦 ∗ ) is characterized

18
 by the system
∗ ∗
 10𝑥 + 10𝑦 = 160




2𝑥 ∗ (2𝑦 ∗ + 1) 10

 =
2(𝑥 ∗ ) 2 10


The system then yields that 𝑥 ∗ = 11 and 𝑦 ∗ = 5.

14. (a) Proof.Completeness implies that for each pair of consumption bundles (𝑋, 𝑌 ), either 𝑋 % 𝑌 or
𝑌 % 𝑋 (or both) is true. In particular, let 𝑋 = 𝑌 = 𝐴. Then both 𝑋 % 𝑌 and 𝑌 % 𝑋 are equivalent
to 𝐴 % 𝐴. Since at least one of the two holds, 𝐴 % 𝐴 must hold. 
(b) Proof.By completeness, we have either 𝐴 % 𝐵 or 𝐵 % 𝐴 (or both). Without loss of generality,
assume that 𝐴 % 𝐵 (the case with 𝐵 % 𝐴 can be shown in a similar way). Then by convexity
one deduces that 𝑡 𝐴 + (1 − 𝑡)𝐵 % 𝐵 for all 𝑡 ∈ [0, 1]. Since 𝐵 % 𝐶 by assumption, we conclude,
using transitivity, that 𝑡 𝐴 + (1 − 𝑡)𝐵 % 𝐶 for all 0 6 𝑡 6 1. 
(c) It violates the axiom of convexity. Notice that 𝐶  𝐴 implies that 𝐶 % 𝐴 (by definition), and so
by convexity we have 0.5𝐶 + 0.5𝐴 % 𝐴. Notice that 0.5𝐶 + 0.5𝐴 = (6, 4) = 𝐵, and so 𝐵 % 𝐴,
which then contradicts the fact that 𝐴  𝐵.
(d) Here strong monotonicity and transitivity cannot both be true. To see this, notice that we can
find a consumption bundles 𝐵 on 𝐼1 and 𝐶 on 𝐼2 such that 𝐶 has more of everything than 𝐵, as
illustrated in Figure 1 below.

Good 𝑌

𝐶
𝐵
𝐴
𝐼1
𝐼2

𝑂 Good 𝑋

Figure 1: Two crossing indifference curves

Since 𝐶 and 𝐴 are on the same indifference curve, it follows that 𝐶 ∼ 𝐴. The same argument
also implies that 𝐵 ∼ 𝐴. Then by transitivity, 𝐵 ∼ 𝐶, which violates strong monotonicity.

15. (a) By definition, the price elasticity

16 𝑝 5
 
d𝑄( 𝑝) 𝑝 1 𝑝
𝜀= · = 4 −4 · 5 · = − · = −4.
d𝑝 𝑄( 𝑝) 𝑝 4/𝑝 4 𝑝5 4

Notice that the demand curve is iso-elastic and so the elasticity does not rely on the value of 𝑝.

19
 (b) Differentiate both sides with respect to 𝑝 (using the chain-rule), we obtain (view ln 𝑄( 𝑝) as a
composition of ln 𝑥 and 𝑄( 𝑝))
1 d𝑄 𝑑 6
· =− .
𝑄( 𝑝) d𝑝 𝑝
Multiplying 𝑝 on both sides, it follows that

d𝑄 𝑑 𝑝
𝜀= · = −6.
d𝑝 𝑄( 𝑝)

This is also an example of iso-elastic demand curve.

(c) Straightforwardly, one has

d𝑄( 𝑝) 𝑝 𝑝 −4𝑝 2
𝜀= · = −4𝑝 · = .
d𝑝 𝑄(𝑑) 100 − 2𝑝 2 100 − 2𝑝 2

Plugging 𝑝 = 5 to the expression above, 𝜀 = (−4 × 52 )/(100 − 2 × 52 ) = −2.

16. (a) The maximum units of bubble tea Tom can afford is the units he can buy if all monthly budget is
spent on it. Here it is obvious that Tom can afford more than 9 cups of bubble tea, and so in
0 = 2)
addition to the 9 cups, he can buy at most (at the lower price 𝑝 𝑋

0 𝐼 − 6 × 4 80 − 24
𝑋max = 0 = = 28.
𝑝𝑋 2

As a result, the maximum quantity of bubble tea he can afford is

0
𝑋max = 𝑋max + 9 = 37.

(b) The slope of the budget line equals −𝑝 𝑋 /𝑝𝑌 , which may change if the prices change. A kinky
point on the budget line is a point at which the slope of budget line changes. As a result, the
kinky point is reached when 𝑋 = 6 since the price of each additional unit of good 𝑋 will drop to
$2 after that. The vertical coordinate of the kinky point equals (80 − 6 × 4)/8 = 7. Thus, the
kinky point is (6, 7).

17. (a) The horizontal intercept of the budget line is the maximum quantity of dollars the family can
spend on education, which is clearly the total yearly income $80, 000. Clearly, the vertical
intercept is also $80, 000.
(b) Now the disposable income (i.e., income net the tax) becomes $80, 000(1 − 10%) = $72, 000.
Thus, the two intercepts become $72, 000.
(c) Since the school voucher can only be used for education, now the maximum dollars the family
can spend on education, and thus the horizontal axis, become $72, 000 + $8, 000 = $80, 000. The
maximum dollars the family can spend on all other goods and services (the vertical intercept) is
still $72, 000.

20
18. (a) Since Sara’s utility is Cobb-Douglas, the “fixed-proportion property” holds. Since the power of
either 𝑋 or 𝑌 is equal to 1, it follows that the “weight” of good 𝑋 and that of good 𝑌 are all equal
to 1/2. Let (𝑋 ∗ , 𝑌 ∗ ) be the optimal consumption bundle. Then we have
(
𝑝 𝑋 · 𝑋 ∗ = 𝐼/2
𝑝𝑌 · 𝑌 ∗ = 𝐼/2

Plugging in the data, we can solve 𝑋 ∗ = 𝑌 ∗ = 40. Sara’s utility level at the optimal consumption
bundle is
𝑈 (𝑋 ∗ , 𝑌 ∗ ) = 40 × 40 = 1600.

(b) We obtain from the system in (a) that 𝑝 𝑋 · 𝑋 ∗ = 𝑝 𝑦 · 𝑌 ∗ (notice that now the income 𝐼 is viewed
as a variable). Thus, the income-consumption curve is (using the condition that 𝑝 𝑋 = 𝑝𝑌 = 1)

𝑌 ∗ = 𝑋∗.

(c) Now let 𝑝 𝑋 = 4 in the system above, it is easy to get that the new optimal consumption bundle
(𝑋 ∗∗ , 𝑌 ∗∗ ) = (10, 40).
(d) By (a) and (b) we know that the total effect of the price change on 𝑌 (vitamin water) is 40 − 40 = 0.
Since the total effect equals the sum of the substitution effect and the income effect while the
substitution effect is +40 (as given), the corresponding income effect on vitamin water equals
0 − 40 = −40.

Let (𝑋 𝑐 , 𝑌 𝑐 ) be the optimal consumption bundle provided that there is only substitution effect.
Then 𝑌 𝑐 = 𝑌 ∗ + 40 = 80. Since (𝑋 ∗ , 𝑌 ∗ ) and (𝑋 𝑐 , 𝑌 𝑐 ) must yield to Sara the same level of
utility, we have
1600 = 𝑈 (𝑋 ∗ , 𝑌 ∗ ) = 𝑈 (𝑋 𝑐 , 𝑌 𝑐 ) = 𝑋 𝑐𝑌 𝑐 = 80𝑋 𝑐 ,

from which we deduce that 𝑋 𝑐 = 20. Therefore, the substitution effect on good 𝑋 (protein bars)
equals 𝑋 𝑐 −𝑋 ∗ = 20−40 = −20, and the income effect on good 𝑋 equals 𝑋 ∗∗ −𝑋 𝑐 = 10−20 = −10.

19. (a) Since the utility function is Cobb-Douglas, according to the “fixed-proportion property”, the
optimal proportion of budget that should be spent on good 𝑌 equals its weight, that is,

𝛽 1
= .
𝛼+𝛽 3

Since 𝛽 = 5, we solve from the equation above that 𝛼 = 10.

(b) Since Andrew always spends one third of the total budget on 𝑌 , he always spends two thirds of
his total budget on 𝑋. In other words, the budget spent on 𝑋 is always twice as much as that on 𝑌 .
So when he spends 300 on 𝑌 , he spends 600 on 𝑋. Thus, he consumes 600/4 = 150 units of 𝑋.
(c) Denote the change in Andrew’s income by Δ𝐼. Then using the “fixed-proportion property”, we

21
 have (𝑥 ∗ being the optimal quantity for good 𝑋 at the income level 𝐼)

2
(𝐼 + Δ𝐼) = 𝑥 ∗ + Δ𝑥.
3
Since 𝑥 ∗ = 2𝐼/3 and Δ𝑥 = 8, we can solve that Δ𝐼 = 3Δ𝑥/2 = 12.
(d) Since a fixed proportion of budget is spent on good 𝑌 regardless of the prices, we know that the
demand for good 𝑌 does not reply on 𝑝 𝑥 , and so d𝑦 ∗ /d𝑝 𝑥 = 0 (𝑦 ∗ is the demand for good 𝑌 ).
Thus, the cross-price elasticity

d𝑦 ∗ 𝑝 𝑥 𝑝𝑥
𝜀𝑦𝑥 = · ∗ = 0 · ∗ = 0.
d𝑝 𝑥 𝑦 𝑦

20. (a) Since the utility function is linear, it suffices to analyze the “marginal utility per dollar” for each
good and conclude that it will be optimal to spend all budget on the one with the highest marginal
utility per dollar. Clearly, the marginal utility per dollar for good 𝑋 equals (1/4) × 3 = 3/4, and
that for good 𝑌 equals (1/6) × 5 = 5/6. Since 5/6 > 3/4, it is optimal to consume only good 𝑌 .
Thus, the optimal consumption bundle is (𝑥 ∗ , 𝑦 ∗ ) = (0, 20).
(b) A price drop of good 𝑌 generates an even higher “marginal utility per dollar” for 𝑌 , and so under
the new prices Bernie will still solely consume good 𝑌 . Therefore, the optimal consumption
bundle now is (𝑥 ∗∗ , 𝑦 ∗∗ ) = (0, 24). To identify the substitution and income effects, we solve the
expenditure minimization problem under the new prices

min 4𝑥 + 5𝑦 s.t. 3𝑥 + 5𝑦 + 6 = 106, 𝑥, 𝑦 > 0

𝑥,𝑦

Notice that the constraint materializes the requirement that the utility level must be constant
(i.e., 𝑈 (𝑥, 𝑦) = 𝑈 (𝑥 ∗ , 𝑦 ∗ ) = 106) if we isolate the substitution effect. Let us denote the solution
to the expenditure minimization problem by (𝑥 𝑐 , 𝑦 𝑐 ). To solve it, we plug the constraint to the
objective function to eliminate 𝑦 (i.e., replace 𝑦 with (106 − 6 − 3𝑥)/5 in 4𝑥 + 5𝑦) to reformulate
the problem as
100
min 100 + 𝑥 s.t. 0 6 𝑥 6 ,
𝑥 3
where 100/3 is the maximum value 𝑥 can achieve given that 3𝑥 + 5𝑦 + 6 = 106 and 𝑦 > 0.
Clearly, the new objective function is increasing in 𝑥, and so to minimize it 𝑥 should be set as
low as possible. The lowest possible value for 𝑥 is 0 (i.e., not to consume good 𝑋), so we have
𝑥 𝑐 = 0, which implies that 𝑦 𝑐 = (106 − 6 − 3𝑥 𝑐 )/5 = 20. Therefore, the substitution effect on 𝑌
equals 𝑦 𝑐 − 𝑦 ∗ = 20 − 20 = 0, and the income effect on 𝑌 equals 𝑦 ∗∗ − 𝑦 𝑐 = 24 − 20 = 4.
(c) Now the marginal utility per dollar of good 𝑋 equals (1/2) × 3 = 3/2 while that of good 𝑌 is
(1/5) × 5 = 1. Thus, it is now optimal to only consume good 𝑋. Thus, the optimal consumption
bundle (𝑥 0, 𝑦 0) = (120/2, 0) = (60, 0).
To pin down the substitution and income effects on 𝑋, we again consider the following expenditure

22
 minimization problem under the new prices (recall that 𝑈 (𝑥 ∗∗ , 𝑦 ∗∗ ) = 126)

min 2𝑥 + 5𝑦 s.t. 3𝑥 + 5𝑦 + 6 = 126,

𝑥,𝑦

which can be reduced to

min 120 − 𝑥 s.t. 0 6 𝑥 6 40.
𝑥

Denote the solution to this problem by (𝑥 00, 𝑦 00). Since the new objective function is decreasing in
𝑥, to minimize it one need to set 𝑥 as high as possible. Thus, 𝑥 00 = 40 and 𝑦 00 = (126−6−3𝑥 00)/5 =
0. As a result, the substitution effect on 𝑋 equals 𝑥 00 − 𝑥 ∗∗ = 40 − 0 = 40, and the income effect
on 𝑋 equals 𝑥 0 − 𝑥 00 = 60 − 40 = 20.

21. (a) We need to first identify the market demand. Since the market demand is the sum of 100
(identical) individual demands, we have 𝑞 𝑑 = 𝑄 𝑑 /100. Thus,

𝑄𝑑
= 2.2 − 0.02𝑝 + 0.005𝑄 𝑑 ,
100
from which we obtain the market demand curve

𝑄 𝑑 = 440 − 4𝑝.

As a result, the slope of the market demand curve

d𝑄 𝑑
= −4.
d𝑝

The maximum price for the market, denoted by 𝑝,

¯ is the price at which the quantity of demand
equals 0. Thus,
0 = 440 − 4 𝑝,
¯

which yields that 𝑝¯ = 110.

(b) The market equilibrium price 𝑝 ∗ and 𝑄 ∗ solve the system

𝑄 𝑠 = 200 + 6𝑝 ∗






𝑄 𝑑 = 440 − 4𝑝 ∗

 𝑄 𝑠 = 𝑄 𝑑 = 𝑄∗



Solving the system, one obtains ( 𝑝 ∗ , 𝑄 ∗ ) = (24, 344).
(c) At the market equilibrium the aggregate consumer surplus, by definition, equals

1 1
( 𝑝¯ − 𝑝 ∗ )𝑄 ∗ = (110 − 24) × 344 = 14792.
2 2

22. (a) Since Eric is indifferent between brands, one cup of milk tea from TPlus is equally good as that

23
 from Kingyo. In other words, to compensate a loss of one 𝑇, one 𝐾 is just enough. That says,
the marginal rate of substitution is always 1, and so the slope of Eric’s indifference curve is −1
at each consumption bundle.
(b) This equals (the total budget) / (the price of 𝑇), i.e., 60/3 = 20.
(c) Since 𝐾 and 𝑇 are equally good but 𝑇 is more expensive, it is clear that Eric will choose to
buy the cheaper one (i.e., 𝐾) only. Thus, the optimal quantity of 𝐾 per week is 60/2 = 30.
Alternatively, you can formulate Eric’s problem as

max 𝑇 + 𝐾 s.t. 3𝑇 + 2𝐾 = 60,

𝑇 ,𝐾

which is equivalent to (using the budget constraint to eliminate 𝑇)

𝐾
max 20 + , s.t. 0 6 𝐾 6 30.
𝐾 3
Since the objective function is increasing in 𝐾, the maximum is achieved at the right extreme 30.
(d) We can think of the price of milk tea as the sum of its real price and the delivery fee (as the
latter is charged per cup). Thus, the price of 𝑇 becomes $3 + $1 = $4 and that of 𝐾 becomes
$2 + $3 = $5. Since 𝑇 is cheaper than 𝐾 now, it is optimal to buy 𝑇 only. Thus, the optimal
quantity for 𝑇 is 60/4 = 15 in that week.

23. (a) Kate will choose the original plan (i.e., $4/ hour without a membership fee). The reason is that
the opportunity set with the original plan fully contains that with the alternative plan (one can
easily check that the two intercepts of the original budget line are 100 and 25, while those of the
alternative are 70 and 23.3), and so whatever consumption bundle feasible with the alternative
plan is also feasible with the original one. As a result, she should find the original plan better.
(b) Kate will be better off. Notice that she can still afford the original consumption bundle (i.e.,
spending 10 hours in the gym and $60 on other goods/services), because it costs 30 + 10 × 1 + 60 =
100 under the new plan. Since Kate has chosen an alternative consumption bundle while her
original consumption is feasible, she must have found the new one better.

24. (a) By definition, the price elasticity of demand

d𝑄 𝑑 𝑝 −𝑏 𝑝
𝜀𝑑 = · = .
d𝑝 𝑄 𝑑 𝑎 − 𝑏 𝑝

Thus, when 𝜀 𝑑 = −9, the corresponding price 𝑝 must satisfy

−6𝑝
−9 = ,
60 − 6𝑝

and so 𝑝 = 9.

24
(b) The market equilibrium price 𝑝 ∗ equates the quantity of demand and the quantity of supply,
that is, 60 − 6𝑝 ∗ = 2𝑝 ∗ − 4, from which we get 𝑝 ∗ = 8, and so the equilibrium quantity
𝑄 ∗ = 60 − 6𝑝 ∗ = 12. The consumer surplus at equilibrium is the area below the demand curve
and above the market equilibrium price (i.e., the gray area in Figure 2 below). Then it is easy to
have
1 1
Consumer Surplus = (10 − 𝑝 ∗ )𝑄 ∗ = (10 − 8) × 12 = 12.
2 2

𝑝
𝑄 𝑠 = 2𝑝 − 4
10
𝑝∗
𝑄 𝑑 = 60 − 6𝑝

0 𝑄∗ 𝑄

Figure 2: Consumer Surplus

(c) Using the formula for elasticity in (a), we have

−6𝑝 ∗ −6 × 8
𝜀𝑑 𝑝= 𝑝 ∗
= ∗
= = −4.
60 − 6𝑝 60 − 6 × 8

Since the absolute value of the elasticity is greater than 1. the demand at equilibrium is elastic.
(d) If the equilibrium price decreases by 0.3, it will become 8 − 0.3 = 7.7. For 7.7 to be the
equilibrium price, we must have 60 − 6 × 7.7 = 2 × 7.7 − 𝑇 0 (i.e., demand equals supply), where
𝑇 0 is the required level of technology. Therefore, we have 𝑇 0 = 1.6, i.e., the industry technology
must reach 1.6.
(e) The market demand equals the sum of 100 (identical) individual demands, that is, 𝑞 𝑑 = 𝑄 𝑑 /100.
Thus, replacing 𝑞 𝑑 by 𝑄 𝑑 /100 in the individual demand function, we have

𝑄𝑑
= 0.3 − 0.04𝑝 + 0.005𝑄 𝑑 ,
100
which yields the market demand curve 𝑄 𝑑 = 60 − 8𝑝. As a result, the slope of the market
demand curve
d𝑄 𝑑
= −8.
d𝑝
(f) The network externalities are positive since the individual demand is increasing in the market
demand (because d𝑞 𝑑 /d𝑄 𝑑 = 0.005 > 0). Examples exhibiting this property include telephone,
software, etc., the common feature being that the good/service is more valuable to an individual

25
 if it is consumed by more people. For example, telephone service was not quite useful to the
public when very few people were using it. But it becomes a must-have when most people are
using it since it provides a convenient way to reach out to most people.

25. (a) Let Δ𝑥 be the additional amount of 𝑋 that are needed. Then we have by definition 3(𝑥 + Δ𝑥) +
12(𝑦 − 1) = 3𝑥 + 12𝑦, which implies that 3Δ𝑥 − 12 = 0 and so Δ𝑥 = 4.
(b) Since the utility function is Cobb-Douglass, the optimal decision follows the fixed-proportion
rule. It is easy to see that the weight for 𝑋 is 1/(1 + 4) = 1/5 and the weight for 𝑌 is 1 − 1/5 = 4/5.
Therefore, the consumer will optimally spend 1/5 of his income on 𝑋 and 4/5 of that on 𝑌 , that
is,
𝐼 4𝐼
3𝑥 ∗ = 𝑝 𝑋 · 𝑥 ∗ = = 48, 3𝑦 ∗ = 𝑝𝑌 · 𝑦 ∗ = = 192,
5 5
from which we can solve 𝑥 ∗ = 16, 𝑦 ∗ = 64.
(c) The datum reveals that the consumer has spent one half of his income on 𝑋 and the other half on
𝑌 (as the expenditure on 𝑋 is 20 × 3 = 60 and that on 𝑌 is 15 × 4 = 60). Thus, the weight for
𝑋 should be 1/2 rather than 1/5 and that for 𝑌 should be 1/2 rather than 4/5, given that % is
represented by a Cobb-Douglass utility function. Therefore, when 𝐼 = 240 and 𝑝 𝑋 = 𝑝𝑌 = 3,
we have
𝐼 𝐼
3𝑥 ∗ = 𝑝 𝑋 · 𝑥 ∗ = = 120, 3𝑦 ∗ = 𝑝𝑌 · 𝑦 ∗ = = 120,
2 2
and so the updated prediction for (𝑥 ∗ , 𝑦 ∗ ) is (40, 40).
(d) The total effect equals the change in quantity of consumption before and after the price change,
holding the income constant. Then it is clear that the first row represents the situation before the
price change and the second row represents the situation after the price change. Therefore, the
total effect (𝑇 𝐸) on good 𝑌 is 75 − 48 = 27. Since total effect equals the sum of substitution
effect (𝑆𝐸) and income effect (𝐼 𝐸) (i.e., 𝑇 𝐸 = 𝑆𝐸 + 𝐼 𝐸), we have 𝑆𝐸 = 𝑇 𝐸 − 𝐼 𝐸. Since 𝐼 𝐸 = 15,
𝑆𝐸 = 𝑇 𝐸 − 𝐼 𝐸 = 27 − 15 = 12.

Caveat. In principle, we must make sure that the third row represents the situation after the
price change but the consumer gets compensation in income. Although this is not directly given,
we can back this out from the data, provided that % is represented by a Cobb-Douglas utility
function: It is clear from the data that the consumer always spends one half of his income
on 𝑋 and the other half on 𝑌 so the weight for each good is 1/2 and so the utility function is
𝑈 (𝑥, 𝑦) = 𝐴𝑥𝑦 for some 𝐴 > 0, and so it is easy to see that the consumer achieves the same level
of utility in the situation of the first row and that of the third row (as 75 × 48 = 60 × 60).
(e) Since the income effect (on good 𝑌 ) is positive (15) as 𝑝𝑌 drops, it is normal to the consumer by
definition.

26. (a) Since 𝑈2 is differentiable, in optimum the marginal rate of substitution must equal the corre-
sponding price ratio, that is,
𝑀𝑈𝑋 𝑝𝑋
= ,
𝑀𝑈𝑌 (𝑋 ∗ ,𝑌 ∗ ) 𝑝𝑌

26
 where (𝑋 ∗ , 𝑌 ∗ ) represents the utility-maximizing consumption bundle. Since 𝑝 𝑋 = 𝑝𝑌 = 10,
the value of Agent 2’s marginal rate of substitution at the utility-maximizing bundle equals
10/10 = 1. Note that it is not a good idea to first find (𝑋 ∗ , 𝑌 ∗ ) and then plug it back to the
expression of 𝑀𝑈𝑋 /𝑀𝑈𝑌 since the work would be unnecessarily more.
(b) Since 𝑀𝑈𝑋 = 2𝑌 2 and 𝑀𝑈𝑌 = 4(𝑋 − 1)𝑌 , the optimal consumption bundle (𝑋 ∗ , 𝑌 ∗ ) is
characterized by the system

10𝑋 ∗ + 10𝑌 ∗ = 100


 (
𝑋 ∗ + 𝑌 ∗ = 10




⇒ ,
 2(𝑌 ∗ ) 2 𝑌 ∗ = 2(𝑋 ∗ − 1)
 4(𝑋 ∗ − 1)𝑌 ∗ = 1



from which we solve 𝑌 ∗ = 6.
(c) Denote by 𝑋 ∗ Agent 3’s utility-maximizing level of good 𝑋. Since the utility function is
logarithmic, the “fixed-proportion rule” holds, and so

2
10𝑋 ∗ = 𝑝 𝑋 · 𝑋 ∗ = × 100 = 50,
2+2
which yields 𝑋 ∗ = 5.

27. We cannot conclude that Jing is maximizing her utility, although this is possible. In particular, if
Jing does have maximized her utility, then the point must be corner, that is, good 𝑏 is not consumed.
This is possible if 𝑎 and 𝑏 are perfect substitutes (but this is not the only possibility) and the marginal
utility per dollar of good 𝑎 is higher than that of good 𝑏. If, however, it turns out that she has not yet
maximized her utility, then since the current marginal utility per dollar of good 𝑎 is higher than that
of good 𝑏, Jing should increase the purchase of good 𝑎 and reduce that of good 𝑏 to improve.

28. (a) By definition, 𝑀𝑈𝑋 = 𝜕𝑈/𝜕 𝑋 = 1, which is constant for all 𝑋 and 𝑌 .
(b) Notice that we are discussing the property of diminishing marginal rate along a fixed indifference
curve. Since 𝑈 is increasing in both 𝑋 and 𝑌 , we know that each indifference curve is downward
sloping, and so 𝑋 and 𝑌 are changing in the opposite direction when we move along an
indifference curve.
By definition, the marginal rate of substitution equals the ratio between the marginal utility of 𝑋
and that of 𝑌 , namely,
𝑀𝑈𝑋 1 1 1−𝑏
MRS𝑋𝑌 = = 𝑏−1
= 𝑌
𝑀𝑈𝑌 𝑎𝑏𝑌 𝑎𝑏
As 𝑋 increases, 𝑌 must decrease, and so the marginal rate of substitution decreases, because this
expression is increasing in 𝑌 .
In general, diminishing marginal rate of substitution is equivalent to convex indifference curve.
You may use a software to graph a typical indifference curve for our utility function here and
you should see that it is convex.

27
(c) The Dictator’s problem is
max 𝑋 + 𝑎𝑌 𝑏 s.t. 𝑋 + 𝑌 = 1,
𝑋 ,𝑌

which is similar to a consumer’s problem (although it is not in the setup) wherein the income
equals 1, the prices of 𝑋 and 𝑌 are all equal to 1, 𝑀𝑈𝑋 = 1, and 𝑀𝑈𝑌 = 𝑌 1−𝑏 /(𝑎𝑏). To
characterize the optimal solution (𝑋 ∗ , 𝑌 ∗ ), we resort to the system

∗ ∗
 𝑋 +𝑌 = 1




𝑎𝑏
 ∗ 1−𝑏 = 1


 (𝑌 )

Plugging in the data (𝑎, 𝑏) = (1, 0.5), we solve (𝑋 ∗ , 𝑌 ∗ ) = (0.75, 0.25). Thus, the Dictator will
allocate 75 cents to himself and 25 cents to the stranger.
(d) As we can see from the second equation of the system in (c), the value of 𝑌 ∗ does not rely on the
total amount of money for splitting. Thus, when having 1000 dollars to split, the Dictator will
still give the stranger 25 cents. This clearly fails to satisfactorily capture the sense of “fairness”.
Here, changing 𝑎 and/or 𝑏 could affect 𝑌 ∗ , but the allocation will eventually become “unfair” as
the total amount of money increases. In its most general form,
1
𝑌 ∗ = (𝑎𝑏) 1−𝑏 ,

and it is easy to see that 𝑌 ∗ will increase if you increase either 𝑎 or 𝑏. Intuitively, when 𝑎 > 0
and 0 < 𝑏 < 1, increasing 𝑎 or 𝑏 will increase the marginal utility of allocating money to the
stranger, and so will induce the Dictator to give more money to the latter. But since 𝑌 ∗ does not
rely on the total money for splitting, the allocation will become “unfair” when the Dictator has
sufficiently high money to split.
Caveat. The method we have employed above is actually not universally valid. Indeed, it is only
justified when the Dictator has “enough” money to split. To see this, imagine what will happen
if the total amount of money for splitting is less than 25 cents. In that situation, the system will
give us a negative value for 𝑋 ∗ , which is absurd. In fact, now the system does not work, and the
optimal allocation is corner: The Dictator will behave altruistically and give all the money to
the stranger. The justification for this claim is straightforward: When the initial money to split,
denoted by 𝑀, is less than 25 cents, one can show that the marginal utility per dollar of good 𝑋
is always lower than that of good 𝑌 . Concretely, we can eliminate 𝑋 in the objective function
using the feasibility constraint and reduce the problem to

max (1 − 𝑌 ) + 𝑌 0.5 , s.t. 0 6 𝑌 6 𝑀.

𝑌 | {z }
≡ 𝑓 (𝑌 )

28
 Since for all 𝑌 ∈ (0, 0.25]
1
𝑓 0 (𝑌 ) = √ − 1 > 0,
2 𝑌
it is clear that the maximum is achieved at the right bound, that is, 𝑌 ∗ = 𝑀.

29. (a) Using the definition of point price elasticity, we have

Δ𝑄 𝑑 /𝑄 𝑑 −5%
𝜀= = = −0.1.
Δ𝑝/𝑝 (3 − 2)/2

(b) By our result in (a), since |𝜀| < 1, the demand is inelastic.
(c) Since the demand is inelastic, the toll revenue increased as as a result of the toll increase.

30. (a) Since there is a fixed ratio (i.e., 3/2) between the quantity of protein bars and yogurt, the
underlying utility function must be Leontief, that is, 𝑈 (𝑥, 𝑦) = min{𝑎𝑥, 2𝑦}. To determine
the value of 𝑎, notice that in optimum one always has 𝑎𝑥 ∗ = 2𝑦 ∗ , (𝑥 ∗ , 𝑦 ∗ ) being the optimal
consumption bundle, from which we deduce 𝑥 ∗ /𝑦 ∗ = 2/𝑎. Thus, 3/2 = 2/𝑎, from which we
learn that 𝑎 = 4/3. Therefore, Lisa’s utility function is
 
4
𝑈 (𝑥, 𝑦) = min 𝑥, 𝑦 .
3

(b) The optimal consumption bundle (𝑥 ∗ , 𝑦 ∗ ) is characterized by the system

(
𝑝 𝑥 · 𝑥∗ + 𝑝 𝑦 · 𝑦∗ = 𝐼
3𝑥 ∗ = 𝑦 ∗

which yields that

3𝐼
𝑦∗ = ,
𝑝 𝑥 + 3𝑝 𝑦
which is the demand for yogurt.

31. (a) The optimal consumption bundle (𝑥 ∗ , 𝑦 ∗ ) satisfies the condition

𝑀𝑈 𝑥 𝑝𝑥
= .
𝑀𝑈 𝑦 ( 𝑥 ∗ ,𝑦 ∗ ) 𝑝𝑦

Since
𝑀𝑈 𝑥 𝑎(𝑥 ∗ − 𝑚) 𝑎−1 (𝑦 ∗ − 𝑛) 𝑏 𝑎(𝑦 ∗ − 𝑛) 3(𝑦 ∗ − 𝑛)
= = = ,
𝑀𝑈 𝑦 ( 𝑥 ∗ ,𝑦 ∗ ) 𝑏(𝑥 ∗ − 𝑚) 𝑎 (𝑦 ∗ − 𝑛) 𝑏−1 𝑏(𝑥 ∗ − 𝑚) 2(𝑥 ∗ − 𝑚)
the equation above can be reduced to (using the fact that 𝑝 𝑥 /𝑝 𝑦 = 2/3)

𝑦∗ − 𝑛 4
= ,
𝑥∗ − 𝑚 9

29
 from which we obtain
4 ∗
𝑦∗ = (𝑥 − 𝑚) + 𝑛.
9
(b) The optimal consumption bundle is characterized by the system (using the result of (a))

∗ ∗
 𝑝𝑥 · 𝑥 + 𝑝𝑦 · 𝑦 = 𝐼




4
 𝑦 ∗ = (𝑥 ∗ − 𝑚) + 𝑛


 9
Solving the system one obtains

9𝐼 − 𝑝 𝑦 (9𝑛 − 4𝑚)
𝑥∗ = .
9𝑝 𝑥 + 4𝑝 𝑦

Thus, when 𝐼 is sufficiently large, the cross-price elasticity of demand for 𝑋 with respect to 𝑝 𝑦

𝜕𝑥 ∗ 𝑝 𝑦 𝑝 𝑦 [9(4𝑚 − 9𝑛) 𝑝 𝑥 − 36𝐼]

𝜀 𝑋𝑌 = · ∗ = < 0,
𝜕 𝑝𝑦 𝑥 (9𝑝 𝑥 + 4𝑝 𝑦 ) [9𝐼 − 𝑝 𝑦 (9𝑛 − 4𝑚)]

since for 𝐼 large, the numerator is negative while the denominator is positive. In other words,
good 𝑋 and good 𝑌 are complements when income is sufficiently large.
(c) When income is so low that it is impossible to have both 𝑥 > 𝑚 and 𝑦 > 𝑛 for every feasible
consumption bundle (𝑥, 𝑦) (i.e., those satisfying 𝑝 𝑥 · 𝑥 + 𝑝 𝑦 · 𝑦 6 𝐼), the utility function could
be illy-behaving. In particular, it might violate the axiom of monotonicity. For example, when
𝑚 = 𝑛 = 10 and 𝑎 = 𝑏 = 2, the consumption bundle (5, 5) will be strictly better than (10, 10)
(since 𝑈 (5, 5) = (5 − 10) 2 (5 − 10) 2 > (10 − 10) 2 (10 − 10) 2 = 𝑈 (10, 10)), which might be
absurd in many plausible applications.
(d) We may interpret the utility function as capturing a situation in which the agent must consume
each good at least up to some minimum level (𝑚 and 𝑛, respectively), that is, both goods are
necessities.

32. (a) For each citizen of East, the marginal rate of substitution must equal the price ratio at the optimal
consumption bundle (since it is interior by assumption), and so

𝑀𝑈𝑋 𝑝𝐸
𝑋 5 1
MRS𝐸
𝑋𝑌 = = = = .
𝑀𝑈𝑌 ( 𝑥 ∗ ,𝑦 ∗ ) 𝑝𝑌𝐸 10 2

Similarly, the marginal rate of substitution for a citizen of West at the optimal consumption
bundle is
𝑀𝑈𝑋 𝑝𝑊
𝑋 100
MRS𝑊 𝑋𝑌 = = = = 2.
𝑀𝑈𝑌 ( 𝑥 ∗ ,𝑦 ∗ ) 𝑝𝑌𝑊 50

(b) Sun Wukong will trade 𝑋 for 𝑌 and Wonder Woman will trade 𝑌 for 𝑋. To see this, let (𝑋 𝑇 , 𝑌 𝑇 )
represent the volume of goods 𝑋 and 𝑌 that are traded between Sun Wukong and Wonder Woman,

30
 where 𝑋 𝑇 , 𝑌 𝑇 > 0. Note that before trading Sun Wukong’s marginal rate of substitution equals
1/2, that is, one unit of 𝑌 is worth two units of 𝑋; in the meantime, Wonder Woman’s marginal
rate of substitution equals 2, that is, one unit of 𝑋 is worth two units of 𝑌 . Thus, for Wukong to
be willing to trade 𝑌 for 𝑋, we must have 𝑋 𝑇 > 2𝑌 𝑇 , while for Wonder Woman to be willing
to trade 𝑋 for 𝑌 , we must have 𝑌 𝑇 > 2𝑋 𝑇 , which is impossible. Since there will be trade, our
claim at the beginning is justified.
Indeed, it is easy to see that the following trade will be mutually favorable: Sun Wukong trades
a “small” amount of 𝑋 for the same amount of 𝑌 with Wonder Woman. Thus, the condition that
“there will be trade” is valid.
(c) Wonder Women’s utility must be (weakly) higher after trading. This is because the trade was
voluntary and so Wonder Woman will never accept an outcome that renders her worse off after
trading.
(d) Yes. The reason is that after trading with Wonder Women, Wukong’s marginal rate of substitution
must be greater than 1/2 (since now he has more 𝑌 but less 𝑋 than before), while Bajie’s marginal
rate of substitution equals 1/2 (as an East citizen), which leaves a room for Wukong to trade
some 𝑌 for 𝑋 with Bajie.

31
 2

Producer Theory and Competitive

Supply

2 Producer Theory and Competitive Supply

32
Exercise 1. Consider a production technology with one variable input (labor 𝐿). Some data about the
production technology is given in the table below. Fill in the gaps (a)-(d).

Quantity of Labor 𝑇 𝑃𝐿 𝐴𝑃 𝐿 𝑀 𝑃𝐿
0 0 - -
1 45
2 180 (a)
3 (b) 90
4 (c) 135
5 675 (d)

Exercise 2. A firm has the production function 𝑄 = 19𝐿 + 10𝐿 2 − 𝐿 3 in the short run, where the only
variable input is labor 𝐿. Assume that 𝐿 changes continuously (i.e., you may consider 𝐿 to be labor hours).

(a) What is the marginal product of labor when 𝐿 = 1?

(b) Find the level of labor input 𝐿¯ at which the average product is maximized.

Exercise 3. A bakery’s production capacity of bread loaves, given its capital equipment, depends on the
number of workers as described by the following table:

Number of Workers 1 2 3 4 5 6 7
Bread Loaves (thousands) 1 1.8 2.4 2.8 3 2.7 2.3

(a) Calculate the marginal product of each worker and the average product for each level of labor input.
(b) Can we determine the types of returns to scale the bakery exhibits, based on the data above? Justify
your answer.
(c) What reasons might make the marginal product of a worker negative, as the data suggest above?
What are the implications of this fact over the slope of the isoquants?

Exercise 4. A group has chartered a bus to Yosemite. They have paid in advance the driver costs $100,
the bus costs $500, and tolls $75. At the same time they started to sell tickets at the price of $15 (per
ticket). The selling will last for one week and then the group must choose either or not to cancel the trip.
If the trip is canceled, then people who have bought tickets will be fully refunded, but the fees paid in
advance are not fully refundable. In particular, the driver’s fee is nonrefundable, but the rest fees can be
returned at a cost of $50.

(a) Among all fees already paid, which are sunk costs when deciding whether to cancel the ticket?
Explain.
(b) How many people must buy tickets so that the trip need not be canceled?

Exercise 5. Answer the following questions for firms with each of following three production functions:

33
 • 𝑄(𝐾, 𝐿) = 𝐿𝐾 2 ;
• 𝑄(𝐾, 𝐿) = 3𝐾 + 7𝐿;
• 𝑄(𝐾, 𝐿) = 2 min{2𝐾, 𝐿}.

(a) Graph the isoquants 𝑄 = 2 and 𝑄 = 10.

(b) Calculate the marginal rate of technical substitution (i.e., 𝑀 𝑃 𝐿 /𝑀 𝑃𝐾 ) at (𝐾, 𝐿) = (2, 4). If the
marginal rate of technical substitution is not well-defined for some production function, briefly
explain the reason.
¯ 𝐿) for 𝐾¯ = 6.
(c) Find the short-run production function 𝑓 (𝐿) = 𝐹 ( 𝐾,

Exercise 6. Suppose in short run, a firm’s capital is fixed at 25, i.e., 𝐾 = 25, but the input of labor 𝐿 is
variable. The firm has the production function

𝑄(𝐿, 𝐾) = 2𝐿 0.5 𝐾 0.5 .

(a) In the short run, in order to produce 60 units of output, how many labor inputs are needed?
(b) If the wage rate 𝑤 = 2 and the capital rental rate 𝑟 = 2, find the short-run cost function.

Exercise 7. Suppose a firm uses labor and capital to produce good 𝑍. The production function is given by

𝑄(𝐿, 𝐾) = 𝐿 1/3 𝐾 2/3 .

The labor wage rate 𝑤 = 1 and the capital rental rate 𝑟 = 2.

(a) If we draw the iso-cost line (with 𝐿 on the horizontal axis and 𝐾 on the vertical axis), what is the
slope of the iso-cost line?
(b) Find the cost-minimizing input combination when the firm wants to produce 16 units of 𝑍.
(c) Derive the cost function.

Exercise 8. Determine and prove whether each of the following production functions exhibits increasing
returns to scale, decreasing returns to scale, constant returns to scale, or none of the three.

(a) 𝑄( 𝐴, 𝐿, 𝐾) = 𝐴𝐿 0.3 𝐾 0.7 ;

(b) 𝑄(𝐿, 𝐾) = 7𝐿 0.5 𝐾 0.2 ;
(c) 𝑄(𝐿, 𝐾) = min{𝐿 2 , 𝐾 0.5 };
(d) 𝑄(𝐿, 𝐾) = 2𝐿 + 𝐾;

34
 (e) 𝑄(𝐿, 𝐾, 𝐹) = 2𝐿 0.5 + 3𝐾 0.75 + 𝐹.

Exercise 9. A company produces easy-to-use home computers and has fixed costs of $250. The marginal
cost of producing computers is U-shaped. In particular, the marginal cost is $700 for the first computer,
$250 for the second, $300 for the third, $350 for the fourth, $400 for the fifth, $450 for the sixth, and $500
for the seventh.

(a) Create a table that shows the company’s output, total cost, marginal cost, average cost, variable
cost, and average variable cost, using the information given above. Use “−” to fill the cells where
the corresponding quantities are not well-defined. Round your answers up to 2 decimal places, if
applicable.
(b) At what price is the shut-down point? At what price is the zero-profit point?
(c) If the company sells the computers for $500, is it making a profit or a loss? What is the size of the
profit or loss?
(d) If the company sells the computers for $300, is it making a profit or a loss? What is the size of the
profit or loss?

Exercise 10. Assume that a firm’s total cost is given by

𝑇𝐶 (𝑞) = 32 + 2𝑞 2 + 16𝑞,

where 𝑞 is the level of output.

(a) What is the fixed cost? What is the variable cost when 𝑞 = 9? What is the marginal cost when 𝑞 = 5?
(b) Find the output level 𝑞 1 at which the average cost is minimized.
(c) Find the output level 𝑞 2 at which the average variable cost is minimized.
(d) Determine whether the firm has economies of scale when 𝑞 = 1 and 𝑞 = 8, respectively.

Exercise 11. Consider the following cost function:

𝐶 (𝑄) = 144 + 3𝑄 + 𝑄 2

(a) What are the formulas for the fixed cost, variable cost, average cost, average variable cost, and
marginal cost?
(b) At what level 𝑄 1 is average cost lowest? What is the lowest average cost?
(c) At what level 𝑄 2 is average variable cost lowest? What is the lowest average variable cost?

Exercise 12. Is each of the following statements true or false? In either case, briefly justify your answer.

(a) If a firm has economies of scale, then its average cost is decreasing and lower than its marginal cost.

35
 (b) A firm whose total cost function is 𝐶 (𝑄) = 𝑄 2 + 𝑄 must have economies of scale.
(c) A firm with total cost function 𝐶 (𝑄) = 3𝑄 + 2 must have constant returns to scale.

Exercise 13. Suppose that you own a firm operating in a perfectly competitive market. The total cost
function is
𝐶 (𝑄) = 200 + 𝑄 2 .

(a) Is your firm operating in the short run or long run? Why?
(b) Derive the marginal cost function.
(c) If the market price is 60, how much should you produce and what is your profit?

Exercise 14. There is a competitive industry with free entry and a U-shaped average cost curve (you may
assume a fixed cost and increasing marginal cost). Suppose the government imposes a tax of 𝑇 on each unit
of good that is produced in the industry. Analyze what happens to each of the following in the long run.

(a) The price of the good;

(b) The output of each firm that stays in the market;
(c) The total output of the industry;
(d) The number of firms in the industry.

Exercise 15. A business incurs an economic loss of $10, 000 per year.

(a) If the business has a fixed cost of $15, 000 per year. To maximize profit, should it produce or shut
down in the short run? Should it stay in the industry or exit in the long run? Briefly explain.
(b) Suppose instead that this business has a fixed cost of $6, 000 per year. To maximize profit, should it
produce or shut down in the short run? Should it stay in the industry or exit in the long run? Briefly
explain.

Exercise 16. There are two firms, 𝐴 and 𝐵, on a competitive market. Firm 𝐴’s total cost is given by

𝑇𝐶 𝐴 (𝑄) = 1 + 0.5𝑄 2 ,

and firm 𝐵’s total cost is given by

𝑇𝐶 𝐵 (𝑄) = 0.5𝑄 2 + 6𝑄.

The market demand curve is 𝑄 𝑑 = 6 − 𝑝. We denote by 𝑄 𝑠 the market supply (i.e., the sum of firm 𝐴’s
supply and firm 𝐵’s supply).

(a) Find the shutdown prices of the two firms, respectively.

36
 (b) What is the slope of the market supply curve (i.e., d𝑄 𝑠 /d𝑝) when 𝑝 = 3? What is the slope of the
market supply curve when 𝑝 = 7?
(c) Find the market equilibrium price 𝑝 ∗ and quantity 𝑄 ∗ .
(d) Calculate the economic profit of firm 𝐴 and firm 𝐵, respectively, at the market equilibrium.

Exercise 17. There are 20 identical firms on a competitive market. Each firm is operating with the
marginal cost
𝑀𝐶 (𝑄) = 3𝑄 2 − 8𝑄 + 6,

where 𝑄 is the quantity of output. A firm’s total cost equals 40 when the output level is 2 (i.e., 𝑇𝐶 (2) = 40).

(a) Are the firms in the long run or short run? Explain.
(b) Find the average cost for 𝑄 = 3.
(c) Determine the level of output level 𝑄 0 that minimizes the average variable cost. What is the lowest
average variable cost?
(d) Fill out the gaps (i)-(iv) (i.e., find the market supply and the individual economic profit under the
corresponding market price).

Market price 1.5 9

Market supply (i) (ii)
Each firm’s economic profit (iii) (iv)

Exercise 18. Summarized in the table below are prices for washing and ironing shirt according to a survey
of dry cleaners in New York State.

Name of dry cleaner City Price ($)

Todt Cleaners Albany 1.95
Number one Tailor Albany 1.95
Weekend Cleaners Albany 1.95
Marco Cleaners Buffalo 2.00
Rockwell Tailors Albany 2.00
Alberto the Tailor Buffalo 2.00
Pressed 4 Time Albany 2.00
Effie the Tailor Buffalo 2.00
Friday Cleaners Albany 2.00
Ablitt’s Fine Cleaners New York 2.10
Tailor Jay New York 2.10
All the Best Cleaners Albany 2.25
Master Cleaners Buffalo 2.25
Magnolia Too Albany 2.25
Queen’s Tailor Buffalo 2.50
Casitas Cleaners Buffalo 2.50

37
 (a) What is the average price per shirt washed and ironed in Albany? In Buffalo? Round your answers
up to 2 decimals if applicable.
(b) Draw typical marginal cost and average cost curves for Master Cleaners in Buffalo, assuming it is a
perfectly competitive firm but is making a profit on each shirt in the short run. Make the short-run
equilibrium point and shade the area that corresponds to the profit made by the dry cleaner.
(c) Assume $2.25 is the short-run equilibrium price in Buffalo. Draw a typical short-run demand and
supply curve for the market. Label the equilibrium point.
(d) Observing profits in Buffalo, another dry cleaning service provider, Green Cleaners, enters the
market. It charges $1.95 per shirt. What is the new average price of washing and ironing a shirt in
Buffalo? Illustrate the effect of entry on the average Buffalo price by a shift of the short-run supply
curve, the demand curve, or both.
(e) If the dry cleaning industry is perfectly competitive, what does the average difference in price
between Albany and Buffalo imply about costs in the two cities in the long run?

Exercise 19. In a market there are three firms, numbered 1, 2, and 3, that behave in a perfectly competitive
fashion (that is, each is a price-taker). The total cost functions of them are, respectively,

𝐶1 (𝑄) = 𝑄 2 + 2𝑄 + 36
𝐶2 (𝑄) = 2𝑄 2 + 2𝑄 + 10
𝐶3 (𝑄) = 𝑄 2 + 6𝑄 + 6

We know that, for the current price on the market, firm 1 is producing at the output level that minimizes its
average cost.

(a) What is the current market equilibrium price?

(b) Calculate each firm’s output level and profit in the market equilibrium.

***********

Exercise 20. Complete the following blanks for a firm in the short run, using the information provided
below. Your answers should be expressions about 𝑄, 𝑄 being the output level.

Fixed cost Marginal cost Variable cost Total cost Average fixed cost Average cost

100 1

38
Exercise 21. There are 100 similar firms in a perfectly competitive industry. Their costs are given by the
expression 𝐶 (𝑞) = 𝑞 2 + 120𝑞 + 𝐴, where 0 < 𝐴 < 1000 and 𝐴 differs from firm to firm.

(a) What is the marginal cost of a firm with 𝐴 = 900? Your answer should be an expression about 𝑞.
(b) If the market price is 200, find the profit-maximizing quantity for a firm with 𝐴 = 360.
(c) What is the short-run economic profit of a firm with 𝐴 = 250 if the market price 𝑝 = 100?
(d) Fill in the blanks: The short-run market supply is 𝑄 𝑠 = when price is at least
; otherwise 𝑄 𝑠 = 0. Your answer to the first blank should be an explicit function of
market price 𝑝, while that to the second should be a number.

Exercise 22. The long-run cost function for firms in a perfectly competitive industry with constant costs is

𝐶 = 0.5𝑄 3 − 10𝑄 2 + 60𝑄,

where 𝑄 is the output level.

(a) How much will a single firm produce in the long run?
(b) What is the market price in the long run?

Exercise 23. A competitive firm sells its products at a price of 0.1 dollar per unit. Its total cost function
(including a 0.5 dollar per unit subsidy) is

𝑇𝐶 (𝑄) = 5 − 0.5𝑄 + 0.001𝑄 2 ,

where 𝑇𝐶 is total cost and 𝑄 is output rate (units per time period).

(a) What is the economic profit if the firm produces at 𝑄 = 500?

(b) Determine the short-run profit-maximizing level of output.

Exercise 24. Abed’s telecommunication firm production function is given by 𝑄 = 400𝐿𝐾 0.5 , where 𝐾 is
the number of internet servers and 𝐿 is the number of labor hours he uses.

(a) In the short run, Abed has fixed contracts for 9 servers. What is the marginal product of labor?
(b) In the short run with 9 servers, is Abed’s marginal product of labor increasing, constant, or
decreasing?
(c) In the short run with 9 servers, what is the minimum cost of producing 𝑄 = 1, 000 if the cost of
labor is $120 per hour and the cost per server is 10, 000?
(d) The cost of labor is $120 per hour and the cost per server is 10, 000. In the short run with 9 servers,
what is Abed’s short-run cost function?
(e) The cost of labor is $120 per hour and the cost per server is 10, 000. In the long run, what is the
slope of the isoquant at the cost-minimizing levels of 𝐿 and 𝐾 (𝐿 is on the horizontal axis)?

39
 (f) In the long run, does this production function exhibit increasing, constant, or decreasing returns to
scale?

Exercise 25. A firm is characterized by the production function 𝑄(𝐿, 𝐾) = 2𝐿 2/3 𝐾 1/3 .

(a) If the wage rate 𝑤 = 10 and the capital rental rate 𝑟 = 20, how many units of 𝐿 will be used to
produce 𝑄 = 100 at the cost-minimizing level? (Round your answer to 2 decimal places.)
(b) (Ignore your answer to (a)) Suppose the cost-minimizing input combination is (𝐿 ∗ , 𝐾 ∗ ) = (5, 7)
when 𝑄 = 100, 𝑤 = 10, and 𝑟 = 20. What is the cost of producing at 𝑄 = 100?
(c) How are costs and the cost-minimizing input combination affected if input prices double to 𝑤 = 20
and 𝑟 = 40?

Exercise 26. Suppose that a firm’s production technology is described by the Cobb-Douglas function

𝑞(𝐿, 𝐾) = 𝐴𝐿 𝑎 𝐾 𝑏 ,

where 𝐴, 𝑎, and 𝑏 are strictly positive constants, 𝐿 is the quantity of labor input and 𝐾 is the quantity of
capital input.

(a) If 𝐴 = 1, 000, 𝑎 = 2/3, and 𝑏 = 2/3, determine and prove the returns to scale of the technology.
(b) If 𝐴 = 100, 𝑤 = 1, 𝑟 = 100, 𝑎 = 0.5, 𝑏 = 0.5, and 𝑞 = 10, 000. What is the cost-minimizing levels
of labor and capital, respectively?
(c) The long-run production function is Cobb-Douglas with 𝐴 = 100, 𝑎 = 0.5, and 𝑏 = 0.5. If 𝑤 = 𝑟 = 1,
what is the long-run total cost function?
(d) Suppose the long-run total cost function for typical firms in an industry is 𝐶 = 0.05𝑄 0.5 . Explain
both the temptation for, and danger of, government regulation in this market. You should refer to
this specific cost function in your explanation.

Exercise 27. You are a fledging entrepreneur and app developer with a degree (with all sorts of honors)
from NYU Shanghai. After many summers of hard work, not to mention lost internship opportunities, you
have finished an app that is really cool. Through the app, users can share their videos with people around
the world. Before launching your app, you need to understand your costs. Give an example for each type
of cost below.

(a) Opportunity cost

(b) Fixed cost
(c) Sunk cost
(d) Variable cost

40
Chapter 2 Answer Key

1. For (a), the marginal product of the second unit of labor input

𝑀 𝑃 𝐿 (2) = 𝑇 𝑃 𝐿 (2) − 𝑇 𝑃 𝐿 (1) = 180 − 45 = 135.

For (b), the average product of labor when 𝐿 = 3

𝑇 𝑃 𝐿 (3) 𝑇 𝑃 𝐿 (2) + 𝑀 𝑃 𝐿 (3) 180 + 90

𝐴𝑃 𝐿 (3) = = = = 90.
3 3 3
For (c), the total output when 𝐿 = 4

𝑇 𝑃 𝐿 (4) = 𝐴𝑃 𝐿 (4) × 4 = 135 × 4 = 540.

For (d), the marginal product of the fifth unit of labor input

𝑀 𝑃 𝐿 (5) = 𝑇 𝑃 𝐿 (5) − 𝑇 𝑃 𝐿 (4) = 675 − 540 = 135.

2. (a) The marginal product when 𝐿 = 1

d𝑄
𝑀 𝑃(1) = = (19 + 20𝐿 − 3𝐿 2 ) 𝐿=1 = 19 + 20 × 1 − 3 × 12 = 36.
d𝐿 𝐿=1

(b) Approach 1: The marginal product and the average product must be the same when the latter is
maximized. Thus, 𝐿¯ satisfies 𝑀 𝑃 𝐿 ( 𝐿)
¯ = 𝐴𝑃 𝐿 ( 𝐿),
¯ that is

19 𝐿¯ + 10 𝐿¯ 2 − 𝐿¯ 3
19 + 20 𝐿¯ − 3 𝐿¯ 2 = ,
𝐿¯

from which we solve 𝐿¯ = 5.

¯ = 0, that is,
Approach 2: Since the marginal cost function is differentiable, we must have 𝐴𝑃 0 ( 𝐿)

d 19𝐿 + 10𝐿 2 − 𝐿 3
 
= 10 𝐿¯ − 2 𝐿¯ 2 = 0,
d𝐿 𝐿 𝐿= 𝐿¯

and so 𝐿¯ = 5.

3. (a) Note that the marginal product of the 𝑛-th worker (here it is clear that the change of labor input
is discrete)
𝑀 𝑃 𝐿 (𝑛) = 𝑇 𝑃 𝐿 (𝑛) − 𝑇 𝑃 𝐿 (𝑛 − 1), 𝑛 > 1.

Thus, 𝑀 𝑃 𝐿 (1) = 𝑇 𝑃 𝐿 (1) − 𝑇 𝑃 𝐿 (0) = 1 − 0 = 1, 𝑀 𝑃 𝐿 (2) = 𝑇 𝑃 𝐿 (2) − 𝑇 𝑃 𝐿 (1) = 1.8 − 1 = 0.8.

The marginal product for each level of labor input is given in the table below.

41
 𝑛 1 2 3 4 5 6 7
𝑀 𝑃 𝐿 (𝑛) 1 0.8 0.6 0.4 0.2 −0.3 −0.4

(b) We are not able to determine the type of returns to scale of the bakery based on the data. Indeed,
to determine the type of returns to scale, we need to learn how the output level changes as
the quantities of all inputs double. Here, the production of bread loaves, according to the
description, requires at least two different factors: labor and capital equipment. The data only
contain information about how the output varies in the change of labor input when the capital
input is fixed. So we lack sufficient information to determine the returns to scale of the bakery.
(c) There are many reasons for the marginal product becoming negative. Too many workers may
cause disorganization or other types of production inefficiency, if the quantities of some of
the other production factors (e.g., rooms, machines, etc.) are fixed. For example, too many
workers may bump into each other in a narrow workplace, each being unable to move around.
For another example, in many cases a bulky group of workers may try to spread the tasks across
everyone, leaving production process less efficient than, and the total output less than, that with
fewer workers.

4. (a) Since by cancelling the trip the group can get back only $500 + $75 − $50 = $525, the sunk
costs are the rest of all costs that have been paid, that is, $100 + $500 + $75 − $525 = $150.
(b) At the moment of decision, if the trip is not canceled, the total revenue will 15𝑁, 𝑁 being the
total number of tickets sold out, while the relevant variable costs will be the difference between
total costs and sunk costs, that is $100 + $500 + $75 − $150 = $525. For not cancelling to be
favorable, we must have 15𝑁 > 525, and so 𝑁 > 35. That says, at least 35 people must buy
tickets.

5. (a) The two isoquants for each of the three production functions are graphed below in three axes.

𝐾 𝐾 𝐾

𝐾 = 0.5𝐿
𝑄 = 10
𝑄 = 10 𝑄=2 𝑄 = 10
𝑄=2 𝑄=2
𝐿 𝐿 𝐿
𝑂 𝑂 𝑂
𝑄(𝐾, 𝐿) = 𝐿𝐾 2 𝑄(𝐾, 𝐿) = 3𝐾 + 7𝐿 𝑄(𝐾, 𝐿) = 2 min{2𝐾, 𝐿}

(b) For the first production function, notice that 𝑀 𝑃 𝐿 = 𝐾 2 and 𝑀 𝑃𝐾 = 2𝐾 𝐿. Thus,

𝑀 𝑃𝐿 22 1
𝑀 𝑅𝑆𝑇 (2, 4) = = = = 0.25.
𝑀 𝑃𝐾 (2,4) 2×2×4 4

42
 For the second production function, note that 𝑀 𝑃 𝐿 = 7 and 𝑀 𝑃𝐾 = 3, and so at all feasible
input bundle (including (2, 4))
𝑀 𝑃𝐿 7
𝑀 𝑅𝑆𝑇 = = .
𝑀 𝑃𝐾 3
For the third production function, notice that the marginal rate of technical substitution is not
well-defined as for Leontief production functions different input factors cannot substitute each
other.
(c) The short-run production function is obtained by letting 𝐾 equal 6 in each of the three production
functions. That is, for the Cobb-Douglas production function we have 𝑓 (𝐿) = 𝐿 · 62 = 36𝐿;
for the linear production function we have 𝑓 (𝐿) = 3 × 6 + 7𝐿 = 18 + 7𝐿; and for the Leontief
production function we have
(
2𝐿, 𝐿 6 12
𝑓 (𝐿) = 2 min{2 × 6, 𝐿} = 2 min{12, 𝐿} = .
24, 𝐿 > 12

6. (a) We must have 60 = 𝑄(𝐿, 25), that is, 60 = 2𝐿 0.5 250.5 , and so 𝐿 = 36. That is, 36 units of labor
are needed.
(b) Let 𝑄 > 0 be the required level of output. Recall that cost function is the minimum cost that a
firm gets to pay in order to produce a certain level of output. Here, in order to produce 𝑄, given
that 𝐾 is fixed at 25, we must have 𝑄 = 2(𝐿 𝑅 ) 0.5 250.5 (𝐿 𝑅 is the required level of labor input),
and so 𝐿 𝑅 = 𝑄 2 /100. Therefore, the short run cost function

𝑄2
𝐶 (𝑄) = 𝐹𝐶 + 𝑉𝐶 (𝑄) = 2 × 25 + 2𝐿 𝑅 = 50 + 2 · = 50 + 0.02𝑄 2 .
| {z } |{z} 100
Fixed Cost Variable Cost

7. (a) A typical iso-cost line can be expressed as 𝑐 = 𝑤𝐿 + 𝑟𝐾 = 𝐿 + 2𝐾, 𝑐 being the corresponding
level of cost (a constant). Thus, the slope of the iso-cost line equals
 
d𝐾 d 𝑐−𝐿 1
= = − = −0.5.
d𝐿 d𝐿 2 2

(b) Since the marginal rate of technical substitution

𝑀 𝑃𝐿 (1/3)𝐿 −2/3 𝐾 2/3 𝐾

MRTS𝐾 𝐿 = = 1/3 −1/3
=
𝑀 𝑃𝐾 (2/3)𝐿 𝐾 2𝐿

The cost-minimizing input combination (𝐾 ∗ , 𝐿 ∗ ) is characterized by the system

∗ 1/3 ∗ 2/3
 (𝐿 ) (𝐾 ) = 16




𝐾∗ 𝑤 1 ,
 ∗ = =


 2𝐿 𝑟 2

43
 from which we solve (𝐾 ∗ , 𝐿 ∗ ) = (16, 16).
(c) Let 𝑄 denote the required level of output. Similar to the situation in (b), the optimal input
combination (𝐾 ∗ , 𝐿 ∗ ) solves the system

∗ 1/3 ∗ 2/3
 (𝐿 ) (𝐾 ) = 𝑄




𝐾∗ 1
 ∗ =


 2𝐿 2
which yields that 𝐾 ∗ = 𝐿 ∗ = 𝑄. Therefore, the (long-run) cost function of the firm is

𝐶 (𝑄) = 𝑤𝐾 ∗ + 𝑟 𝐿 ∗ = 𝑄 + 2𝑄 = 3𝑄.

8. (a) Pick an arbitrary 𝜆 > 1. It is clear that (note that here 𝐴, 𝐿, 𝐾 are all production factors)for
𝑄( 𝐴, 𝐿, 𝐾) > 0

𝑄(𝜆𝐴, 𝜆𝐿, 𝜆𝐾) = 𝜆𝐴(𝜆0.3 𝐿 0.3 ) (𝜆0.7 𝐾 0.7 ) = 𝜆2 ( 𝐴𝐿 0.3 𝐾 0.7 ) = 𝜆2 𝑄( 𝐴, 𝐿, 𝐾) > 𝑄( 𝐴, 𝐿, 𝐾).

Therefore, the production function exhibits increasing returns to scale.

(b) Pick an arbitrary 𝜆 > 1. For 𝑄(𝐿, 𝐾) > 0 we have

𝑄(𝜆𝐿, 𝜆𝐾) = 7(𝜆0.5 𝐿 0.5 ) (𝜆0.2 𝐾 0.2 ) = 𝜆0.7 (7𝐿 0.5 𝐾 0.2 ) = 𝜆0.7 𝑄(𝐿, 𝐾) < 𝑄(𝐿, 𝐾).

So the production function exhibits decreasing returns to scale.

(c) Notice that

𝑄(0.5, 0.5) = min{0.52 , 0.50.5 } = 0.25

𝑄(1, 1) = min{12 , 10.5 } = 1
√
𝑄(2, 2) = min{22 , 20.5 } = 2

Thus, 𝑄(1, 1) > 2𝑄(0.5, 0.5) while 𝑄(2, 2) < 2𝑄(1, 1), which implies that the production
function does not exhibit a consistent pattern of returns to scale.
(d) Pick an arbitrary 𝜆 > 1. We have straightforwardly

𝑄(𝜆𝐿, 𝜆𝐾) = 2𝜆𝐿 + 𝜆𝐾 = 𝜆(2𝐿 + 𝐾) = 𝜆𝑄(𝐿, 𝐾),

which implies that the production function exhibits constant returns to scale.
(e) Pick an arbitrary 𝜆 > 1. For 𝑄(𝐿, 𝐾, 𝐹) > 0 (with max{𝐿, 𝐾 } > 0) one has

𝑄(𝜆𝐿, 𝜆𝐾, 𝜆𝐹) = 2𝜆0.5 𝐿 0.5 + 3𝜆0.75 𝐾 0.75 + 𝜆𝐹

< 2𝜆𝐿 0.5 + 3𝜆𝐾 0.75 + 𝜆𝐹

44
 = 𝜆(2𝐿 0.5 + 3𝐾 0.75 + 𝐹)
= 𝜆𝑄(𝐿, 𝐾, 𝐹).

As a result, the production function exhibits decreasing returns to scale.

9. (a) We summarize different types of cost in the table below. The following formulae are used
to compute the quantities: 𝑇𝐶 (𝑄) = 𝑇𝐶 (𝑄 − 1) + 𝑀𝐶 (𝑄), 𝐴𝐶 (𝑄) = 𝑇𝐶 (𝑄)/𝑄, 𝑉𝐶 (𝑄) =
𝑇𝐶 (𝑄) − 𝐹𝐶, and 𝐴𝑉𝐶 (𝑄) = 𝑉𝐶 (𝑄)/𝑄, where 𝐹𝐶 represents the fixed cost.

𝑄 𝑇𝐶 (𝑄) 𝑀𝐶 (𝑄) 𝐴𝐶 (𝑄) 𝑉𝐶 (𝑄) 𝐴𝑉𝐶 (𝑄)

0 250 − − 0 −
1 950 700 950 700 700
2 1200 250 600 950 475
3 1500 300 500 1250 416.67
4 1850 350 462.5 1600 400
5 2250 400 450 2000 400
6 2700 450 450 2450 408.33
7 3200 500 457.14 2950 421.43

(b) Since the company is operating in the short run, the shut-down price is the price that equals
the lowest average variable cost, while the lowest average variable cost is reached when
the average variable cost equals the marginal cost. It is easy to inspect from the table that
𝑀𝐶 (5) = 𝐴𝑉𝐶 (5) = 400. Thus, the shut-down price is 400.
The company receives a zero profit when the price equals the average cost. However, since the
profit-maximizing level of output equates the price and the marginal cost, it follows that the
zero-profit price is equal to the average cost that is achieved at the level of output that equates the
average cost and the marginal cost. Inspecting the table, one sees that 𝑀𝐶 (6) = 𝑀𝐶 (6) = 450.
Therefore, the zero-profit price is 450.
(c) Since 500 is higher than the zero-profit price, the company is making a profit rather than a loss.
When 𝑝 = 500, it is optimal to produce the quantity 𝑄 ∗ such that 500 = 𝑀𝐶 (𝑄 ∗ ). Inspecting
the table, we can see that 𝑄 ∗ = 7. Thus, the total profit

𝜋 = 𝑝𝑄 ∗ − 𝑇𝐶 (𝑄 ∗ ) = 500 × 7 − 3200 = 300.

(d) Since 300 is lower than the zero-profit price, the company is making a loss. When 𝑝 = 300, it is
optimal to produce the quantity 𝑄 ∗ such that 300 = 𝑀𝐶 (𝑄 ∗ ). The table tells us that 𝑄 ∗ = 3.
Thus, the loss
𝜋 = 𝑝𝑄 ∗ − 𝑇𝐶 (𝑄 ∗ ) = 300 × 3 − 1500 = −600.

10. (a) The fixed cost equals 32 (i.e., the cost that does not vary in the quantity of output). The
variable cost is 𝑉𝐶 (𝑞) = 2𝑞 2 + 16𝑞, and so 𝑉𝐶 (9) = 2 × 92 + 16 × 9 = 306. The marginal cost

45
 𝑀𝐶 (𝑞) = d𝐶 (𝑞)/d𝑞 = 4𝑞 + 16, and so 𝑀𝐶 (5) = 4 × 5 + 16 = 36.
(b) 𝑞 1 will equate the average cost and the marginal cost. Thus,

32
+ 2𝑞 1 + 16 = 𝐴𝐶 (𝑞 1 ) = 𝑀𝐶 (𝑞 1 ) = 4𝑞 1 + 16,
𝑞1

from which we solve 𝑞 1 = 4.

(c) 𝑞 2 will equate the average variable cost and the marginal cost. Thus,

2𝑞 2 + 16 = 𝐴𝑉𝐶 (𝑞 2 ) = 𝑀𝐶 (𝑞 2 ) = 4𝑞 2 + 16,

which implies that 𝑞 2 = 0.

(d) Recall that a firm has economies of scale if its average cost is decreasing in output level. Thus,
to determine if the firm has economies of scale at given output levels it suffices to determine the
sign of the derivative of average cost at each of the output levels. Since

d 32 + 2𝑞 2 + 16𝑞
 
0 32
𝐴𝐶 (𝑞) = = 2− 2,
d𝑞 𝑞 𝑞

it follows that 𝐴𝐶 0 (1) = −30 < 0 and 𝐴𝐶 0 (8) = 1.5 > 0. Therefore, the firm has economies of
scale when 𝑞 = 1 while it does not have economies of scale when 𝑞 = 8.

11. (a) The fixed cost 𝐹𝐶 = 𝐶 (0) = 144, the variable cost 𝑉𝐶 (𝑄) = 𝐶 (𝑄) − 𝐹𝐶 = 3𝑄 + 𝑄 2 , the average
cost 𝐴𝐶 (𝑄) = 𝐶 (𝑄)/𝑄 = 144/𝑄+3+𝑄, the average variable cost 𝐴𝑉𝐶 (𝑄) = 𝑉𝐶 (𝑄)/𝑄 = 3+𝑄,
and the marginal cost 𝑀𝐶 (𝑄) = d𝐶 (𝑄)/d𝑄 = 3 + 2𝑄.
(b) The average cost is minimized when the average cost equals the marginal cost. Thus, 𝑀𝐶 (𝑄 1 ) =
𝐴𝐶 (𝑄 1 ), that is,
144
3 + 2𝑄 1 = + 3 + 𝑄1,
𝑄1
from which we solve 𝑄 1 = 12. Plugging this back to the expression for 𝐴𝐶, we obtain

144
𝐴𝐶min = 𝐴𝐶 (𝑄 1 ) = + 3 + 12 = 27.
12
(c) The average variable cost is minimized when the average variable cost equals the marginal cost.
Thus, 𝑀𝐶 (𝑄 2 ) = 𝐴𝑉𝐶 (𝑄 2 ), that is,

3 + 2𝑄 2 = 3 + 𝑄 2 ,

from which we solve 𝑄 2 = 0, and so

𝐴𝑉𝐶min = 𝐴𝑉𝐶 (𝑄 2 ) = 3.

12. (a) False. By definition, a firm has economies of scale if the average cost is decreasing (in the output

46
 level). However, the average cost is decreasing if and only if it is higher than the marginal cost.
(b) False. The average cost of the firm 𝐴𝐶 (𝑄) = 𝐶 (𝑄)/𝑄 = 𝑄 + 1, which is increasing in the output
level. Thus, the firm has diseconomies, rather than economies, of scale.
(c) False. The firm is clearly in the short run (due to the existence of a fixed cost of 2), but it is
impossible in general to determine the returns to scale based on the short-run cost function, even
when the factor prices are fixed. For example, consider a firm operating with the production
function 𝑄(𝐾, 𝐿) = 𝐾 𝐿 in the short run. The capital input (𝐾) is fixed at 1 and the labor input
(𝐿) is variable. The price of each unit of capital is 2 and that of each unit of labor is 3. It is easy
to derive that the short-run cost function of this firm is 𝐶 (𝑄) = 3𝑄 + 2, but the firm clearly has
increasing returns to scale.

13. (a) It operates in the short run, since the fixed cost 𝐶 (0) = 200 > 0.
(b) The marginal cost
d𝐶 (𝑄)
𝑀𝐶 (𝑄) = = 2𝑄.
d𝑄
(c) We need to first identify the shut-down price. As is well known, such price equals the lowest
average cost function, which is achieved when the marginal cost equals the average variable cost.
Let 𝑄 1 be the output level such that 𝐴𝑉𝐶 (𝑄 1 ) = 𝑀𝐶 (𝑄 1 ), that is,

𝐶 (𝑄 1 ) − 𝐶 (0)
= 2𝑄 1 ,
𝑄1

and so 𝑄 1 = 0 and 𝐴𝑉𝐶min = 0. Thus, when 𝑝 = 60, the firm will not shut down, and the
corresponding optimal quantity 𝑄 ∗ solves 𝑝 = 𝑀𝐶 (𝑄 ∗ ), that is,

60 = 2𝑄 ∗ ,

from which we solve 𝑄 ∗ = 30. The corresponding profit

𝜋 = 𝑝𝑄 ∗ − 𝐶 (𝑄 ∗ ) = 60 × 30 − (200 + 302 ) = 700.

14. (a) In the long run, competitive firms will produce the quantity that achieves the lowest long-run
average cost, and the market price must equal the lowest long-run average cost. Due to the tax,
the lowest long-run average cost will be raised by 𝑇, and so will the price of the good.
(b) In the long run, competitive firms will produce the quantity that achieves the lowest long-run
average cost. Due to the per-unit tax, the long-run average cost at each level of output will
increase by 𝑇. So the quantity of output that minimizes the long-run average cost without the tax
does the same thing with the tax. Therefore, the output of each firm that stays in the market will
not change.
(c) Due to the price increase, the quantity of demand will decrease, and so the equilibrium quantity
of output (and consumption) in the long run will decrease.

47
 (d) The number of firms in the industry will decrease, which is a straightforward implication of (b)
and (c).

15. (a) In the short run, the fixed cost is inevitable. Thus, if the firm shuts down, the total loss will be
$15, 000, which is higher than the current total loss (i.e., $10, 000). So in the short run the firm
should continue to produce. In the long run, however, shut-down will yield a zero profit, which
is better than losing $10, 000. Thus, the firm should exit the industry in the long run.
(b) Now shut-down in the short run becomes profitable: by shutting down the firm incurs a loss of
$6, 000 per year, while by producing the loss will be $10, 000. In the long run, again, since the
firm is not able to make any profit, it will be optimal to exit the industry.

16. (a) Firm 𝐴 is in the short run (as there is a fixed cost of 1), and so its shutdown price equals the
lowest average variable cost. Clearly, 𝐴𝑉𝐶 𝐴 (𝑄) = 0.5𝑄, and so the lowest average variable
cost for firm 𝐴 equals 0 (achieved when 𝑄 = 0). Firm 𝐵, however, is in the long run (as there is
no fixed cost), and so its shutdown price is the lowest average cost. Since 𝐴𝐶 𝐵 (𝑄) = 0.5𝑄 + 6,
the lowest average cost equals 6 (achieved when 𝑄 = 0), which is the shutdown price of firm 𝐵.
(b) For a competitive firm, its supply curve is 𝑝 = 𝑀𝐶 (𝑄) when 𝑝 is no lower than its shutdown
price, and is 0 otherwise. Therefore, firm 𝐴’s supply curve above the shutdown price is 𝑄 = 𝑝
(since 𝑀𝐶 𝐴 (𝑄) = 𝑄), and firm 𝐵’s supply curve above the shutdown price is 𝑝 = 𝑄 + 6 (since
𝑀𝐶 𝐵 (𝑄) = 𝑄 + 6). So firm 𝐴’s and firm 𝐵’s individual supply curves are, respectively,
(
0, if 0 6 𝑝 < 6
𝑄 𝐴 ( 𝑝) = 𝑝 ( 𝑝 > 0), 𝑄 𝐵 ( 𝑝) = .
𝑝 − 6, if 𝑝 > 6

As a result, the market supply

(
𝑝, if 0 6 𝑝 < 6
𝑄 𝑠 ( 𝑝) = 𝑄 𝐴 ( 𝑝) + 𝑄 𝐵 ( 𝑝) = ,
2𝑝 − 6, if 𝑝 > 6

and so the slope of the market supply curve equals 1 when 𝑝 = 3, and equals 2 when 𝑝 = 7.
(c) Let ( 𝑝 ∗ , 𝑄 ∗ ) be the market equilibrium. Then 𝑄 ∗ = 𝑄 𝑠 ( 𝑝 ∗ ) = 𝑄 𝑑 ( 𝑝 ∗ ), i.e.,

𝑄∗ = 𝑝∗ = 6 − 𝑝∗,

from which we solve 𝑝 ∗ = 3 and 𝑄 ∗ = 3. Notice that we should not use the expression for 𝑄 𝑠
when 𝑝 > 6 to characterize the market equilibrium since in that case we would have 𝑝 ∗ > 6 and
2𝑝 ∗ − 6 = 6 − 𝑝 ∗ , causing a contradiction.
(d) Since the equilibrium price 𝑝 ∗ = 3 is below the shutdown price of firm 𝐵, its output is 0 and so
firm B’s economic profit at the market equilibrium is 0. All supply comes from firm 𝐴. Thus,

48
 the economic profit of firm 𝐵 equals

𝑝 ∗ 𝑄 ∗ − 𝑇𝐶 𝐴 (𝑄 ∗ ) = 3 × 3 − (1 + 0.5 × 32 ) = 3.5.

17. (a) According to the functional form of 𝑀𝐶, we can back out the variable cost
∫ 𝑄 ∫ 𝑄  
𝑉𝐶 (𝑄) = 𝑀𝐶 (𝑞) d𝑞 = 3𝑞 2 − 8𝑞 + 6 d𝑞 = 𝑄 3 − 4𝑄 2 + 6𝑄.
0 0

Thus, 𝑉𝐶 (2) = 23 − 4 × 22 + 6 × 2 = 4, which implies that the fixed cost 𝐹𝐶 = 𝑇𝐶 (2) − 𝑉𝐶 (2) =
40 − 4 = 36. Due to the existence of the fixed cost, the firms are in the short run.
(b) Using the result in (a), the total cost

𝑇𝐶 (𝑄) = 𝑉𝐶 (𝑄) + 𝐹𝐶 = 𝑄 3 − 4𝑄 2 + 6𝑄 + 36,

and so
𝑇𝐶 (𝑄) 36
𝐴𝐶 (𝑄) = = 𝑄 2 − 4𝑄 + 6 + .
𝑄 𝑄
As a result, 𝐴𝐶 (3) = 32 − 4 × 3 + 6 + 36/3 = 15.
(c) 𝑄 0 equates the average variable cost and the marginal cost, that is,

𝑄 20 − 4𝑄 0 + 6 = 𝐴𝑉𝐶 (𝑄 0 ) = 𝑀𝐶 (𝑄 0 ) = 3𝑄 20 − 8𝑄 0 + 6,

from which we solve 𝑄 0 = 2. Thus, the lowest average variable cost is

𝐴𝑉𝐶 (𝑄 0 ) = 𝑄 20 − 4𝑄 0 + 6 = 22 − 4 × 2 + 6 = 2.

(d) Since the shutdown price (the lowest average variable cost) for each firm is 2, at the market
price 1.5 all firms will shut down and so the market supply equals 0, in which case each firm’s
economic profit is −𝐹𝐶, namely, −36. When 𝑝 > 2, the optimal quantity of supply for a single
firm satisfies
𝑝 = 𝑀𝐶 (𝑄) = 3𝑄 2 − 8𝑄 + 6,

and so √︁
4 + 3𝑝 − 2
𝑄= .
3
√
Thus, when 𝑝 = 9, the supply of each firm equals (4+ 3 × 9 − 2)/3 = 3, and so the market supply
is 20×3 = 60. Each firm’s economic profit is 𝑝𝑄 −𝑇𝐶 (𝑄) = 9×3− (33 −4×32 +6×3+36) = −18.
The answers are summarized in the table below.
Market price 1.5 9
Market supply 0 60
Each firm’s economic profit −36 −18

49
18. (a) The data about Albany are underlined in the table below, and the data about Buffalo are boxed in
the table below.
Name of dry cleaner City Price ($)
Todt Cleaners Albany 1.95
Number one Tailor Albany 1.95
Weekend Cleaners Albany 1.95
Marco Cleaners Buffalo 2.00
Rockwell Tailors Albany 2.00
Alberto the Tailor Buffalo 2.00
Pressed 4 Time Albany 2.00
Effie the Tailor Buffalo 2.00
Friday Cleaners Albany 2.00
Ablitt’s Fine Cleaners New York 2.10
Tailor Jay New York 2.10
All the Best Cleaners Albany 2.25
Master Cleaners Buffalo 2.25
Magnolia Too Albany 2.25
Queen’s Tailor Buffalo 2.50
Casitas Cleaners Buffalo 2.50

Using the data, it is easy to compute that the average price in Albany

1.95 × 3 + 2.00 × 3 + 2.25 × 2

𝑝¯Albany = ≈ 2.04,
3+3+2
and the average price in Buffalo

2.00 × 3 + 2.25 + 2.50 × 2

𝑝¯Buffalo = ≈ 2.21.
6

(b) The marginal cost curve crosses the average cost curve at the lowest point of the latter. Since
cleaners in Buffalo are making a profit, the price must be higher than the lowest point of the
average cost curve. Given this, cleaners in Buffalo maximizes its profit by producing quantity 𝑄 ∗ ,
at which the market price equals the marginal cost, and the profit is the grey area, as illustrated
in the left panel of Figure 4.
(c) Now the average price in Buffalo

0 1.95 + 2.00 × 3 + 2.25 + 2.50 × 2

𝑝¯Buffalo = ≈ 2.17.
7
The entry of the new firm increases the supply and thus shifts the supply curve to the right, as
indicated by the move from 𝑆1 to 𝑆2 in the right panel of Figure 4. So the new equilibrium
corresponds to a lower equilibrium price, $2.17, and a higher equilibrium quantity 𝑄 ∗2 .
(d) Since, in the long run, firms break even in a perfectly competitive industry, costs have to be

50
 $ 𝑝
𝑆1 𝑆2
𝑀𝐶 (𝑄)
2.25
𝐴𝐶 (𝑄) 2.21
Profit
2.17

0 𝑄∗ 𝑄 0 𝑄 ∗1 𝑄 ∗2 𝑄

Figure 4

higher in Buffalo than in Albany.

19. (a) Let 𝑄 1 be the output level that minimizes firm 1’s average cost. Such 𝑄 1 is characterized by the
condition that 𝑀𝐶1 (𝑄 1 ) = 𝐴𝐶1 (𝑄 1 ), that is,

36
2𝑄 1 + 2 = 𝑄 1 + 2 + ,
𝑄1

from which we solve 𝑄 1 = 6, and so the market price 𝑝 = 𝑀𝐶1 (𝑄 1 ) = 2 × 6 + 2 = 14.

(b) Let 𝑄 2 and 𝑄 3 be firm 2’s and firm 3’s output levels, respectively, in the market equilibrium.
Then we must have 𝑝 = 𝑀𝐶2 (𝑄 2 ) = 𝑀𝐶3 (𝑄 3 ), that is,

14 = 4𝑄 2 + 2, 14 = 2𝑄 3 + 6,

from which we solve 𝑄 2 = 3, 𝑄 3 = 4.

In the equilibrium, the profit of firm 1 is clearly zero (as the market price equals the average
cost); the profit of firm 2 and that of firm 3 are, respectively,

𝜋2 = 𝑝𝑄 2 − 𝐶2 (𝑄 2 ) = 14 × 3 − (2 × 32 + 2 × 3 + 10) = 8,
𝜋3 = 𝑝𝑄 3 − 𝐶3 (𝑄 3 ) = 14 × 4 − (42 + 6 × 4 + 6) = 10.

20. Since the marginal cost is constantly equal to 1, the variable cost
∫ 𝑄
𝑉𝐶 (𝑄) = 1 d𝑞 = 𝑄,
0

and so the total cost

𝑇𝐶 (𝑄) = 𝑉𝐶 (𝑄) + 𝐹𝐶 = 𝑄 + 100,

51
 based on which we have
𝐹𝐶 100
𝐴𝐹𝐶 = =
𝑄 𝑄
𝑇𝐶 (𝑄) 100
𝐴𝐶 (𝑄) = =1+ .
𝑄 𝑄

We summarize the answers in the table below:

Fixed cost Marginal cost Variable cost Total cost Average fixed cost Average cost

100 100
100 1 𝑄 𝑄 + 100 1+
𝑄 𝑄

21. (a) The marginal cost 𝑀𝐶 (𝑞) = d𝐶 (𝑞)/d𝑞 = 2𝑞 + 120, which does not rely on the value of 𝐴 since
𝐴 is the fixed cost.
(b) When 𝐴 = 360, the firm is operating in the short run, and so the shutdown price equals the
lowest average variable cost. Since 𝐴𝑉𝐶 (𝑞) = 𝑞 + 120, the lowest average variable cost is clearly
120. Thus, if the market price is 200, the firm will choose to produce, and the profit-maximizing
quantity of output 𝑞 ∗ solves 𝑝 = 𝑀𝐶 (𝑞 ∗ ), that is

200 = 2𝑞 ∗ + 120,

and so 𝑞 ∗ = 40.
(c) Since the shutdown price is 120, all firms will shut down when the market price 𝑝 = 100. Thus,
the short-run economic profit of a firm with 𝐴 = 250 equals −250, namely, the negative fixed
cost.
(d) Above the shutdown price 120, the short-run individual supply 𝑞 𝑠 solves 𝑝 = 𝑀𝐶 (𝑞 𝑠 ), that is,
𝑝 = 2𝑞 𝑠 + 120, from which we solve

𝑞 𝑠 = 0.5𝑝 − 60.

Thus, the short-run market supply

𝑄 𝑠 = 100𝑞 𝑠 = 50𝑝 − 6000.

Thus, the short-run market supply is 𝑄 𝑠 = 50𝑝 − 6000 when price is at least 120.

22. (a) In the long run, each firm will produce at the lowest average cost. Using the functional form of
the long-run total cost function, the long-run average cost and the long-run marginal cost are,

52
 respectively,

𝐶
𝐿 𝐴𝐶 (𝑄) = = 0.5𝑄 2 − 10𝑄 + 60
𝑄
d𝐶
𝐿 𝑀𝐶 (𝑄) = = 1.5𝑄 2 − 20𝑄 + 60
d𝑄

Let 𝑄 0 minimizes the long-run average cost. Then 𝐿 𝐴𝐶 (𝑄 0 ) = 𝐿 𝑀𝐶 (𝑄 0 ), i.e.,

0.5𝑄 20 − 10𝑄 0 + 60 = 1.5𝑄 20 − 20𝑄 0 + 60,

from which we solve 𝑄 0 = 10. So in the long run, a single firm’s output equals 10.
(b) The market price in the long run, denoted by 𝑝 ∗ , equals the lowest average cost, which is achieved
at 𝑄 0 . Thus,
𝑝 ∗ = 𝐿 𝐴𝐶 (𝑄 0 ) = 0.5 × 102 − 10 × 10 + 60 = 10.

23. (a) The economic profit at 𝑄 = 500 is

𝜋(500) = 0.1 × 500 − [5 − 0.5 × 500 + 0.001 × (500) 2 ] = 45.

(b) Shall we first determine the short-run shutdown price. The average variable cost

𝐴𝑉𝐶 (𝑄) = −0.5 + 0.001𝑄,

whose minimum equals −0.5 (achieved when 𝑄 = 0), which is the shutdown price. Since the
market price (0.1) is higher than the shutdown price, the profit-maximizing level of output 𝑄 ∗
solves 0.1 = 𝑀𝐶 (𝑄 ∗ ), that is,
0.1 = 0.002 × 𝑄 ∗ − 0.5,

from which solve 𝑄 ∗ = 300.

24. (a) By definition, the marginal product of labor when the number of servers is fixed at 9 equals

𝜕𝑄 √
𝑀 𝑃𝐿 = = 400𝐾 0.5 = 400 9 = 1200.
𝐾 =9 𝜕𝐿 𝐾 =9 𝐾 =9

(b) According to our conclusion in (a), it is clear that the marginal product of labor is constantly
equal to 1200 when 𝐾 = 9.
(c) Let 𝐿 𝑅 be the required level of labor input. Then

400𝐿 𝑅 × 90.5 = 1000,

53
 from which we solve 𝐿 𝑅 = 5/6. Thus, the minimum cost of producing 𝑄 = 1000 is

5
10000 × 9 + 120 × = 90100.
6

(d) Let 𝑄 > 0 be the required level of output and 𝐿 𝑅 be the required level of labor input. In order to
produce 𝑄, 𝐿 𝑅 must satisfy
400𝐿 𝑅 × 90.5 = 𝑄,

from which we solve

𝑄
𝐿𝑅 =
.
1200
Therefore, the short-run cost function (let 𝑤 denote the per hour cost of labor and 𝑟 the price of
a server)
𝑄
𝐶 (𝑄) = 𝑟𝐾 + 𝑤𝐿 𝑅 = 10000 × 9 + 120 · = 90000 + 0.1𝑄.
1200
(e) In the long run, the cost-minimizing input combination (𝐿 ∗ , 𝐾 ∗ ) satisfies

𝑀 𝑃𝐿 𝑤
MRS 𝐿𝐾 = = .
(𝐿 ∗ ,𝐾 ∗ ) 𝑀 𝑃𝐾 (𝐿 ∗ ,𝐾 ∗ ) 𝑟

Since 𝐿 is on the horizontal axis, the slope of the isoquant at (𝐿 ∗ , 𝐾 ∗ ) equals

d𝐾 𝑀 𝑃𝐿 𝑤 120
=− =− =− = −0.012.
d𝐿 (𝐿 ∗ ,𝐾 ∗ ) 𝑀 𝑃𝐾 (𝐿 ∗ ,𝐾 ∗ ) 𝑟 10000

(f) Since the production function is Cobb-Douglas, its returns to scale depends on the sum of all
powers of production factors. Here we have two production factors 𝐿 and 𝐾, the power of 𝐿
is 1 and that of 𝐾 is 0.5. Thus, the sum is 1 + 0.5 = 1.5, which is greater than 1, and so the
production function exhibits increasing returns to scale.

25. (a) The cost-minimizing levels (𝐿 ∗ , 𝐾 ∗ ) are characterized by the system

∗ ∗
 𝑄(𝐿 , 𝐾 ) = 𝑄




𝑀 𝑃𝐿 𝑤


 ∗ ∗
=
 𝑀 𝑃𝐾 (𝐿 ,𝐾 ) 𝑟
Using the functional form of the production function, the system can be transformed to

∗ 2/3 ∗ 1/3
 2(𝐿 ) (𝐾 ) = 100




2𝐾 ∗ 1
 ∗ =


 𝐿 2
from which we solve 𝐿 ∗ ≈ 79.37, 𝐾 ∗ ≈ 19.84.

54
 (b) (Note that this is purely a suppositive question) The cost of producing at 𝑄 = 100 equals

𝑟𝐾 ∗ + 𝑤𝐿 ∗ = 20 × 7 + 10 × 5 = 190.

(c) Observe that the system we give in (a) is unchanged when both 𝑤 and 𝑟 double (since only the
ratio between 𝑤 and 𝑟 matters, which is the same if both double), and so the cost-minimizing
input combination will be the same as before. However, since the total cost equals 𝑟𝐾 ∗ + 𝑤𝐿 ∗ , it
is clear that the total cost will also double as 𝑤 and 𝑟 double.

26. (a) Pick an arbitrary 𝜆 > 1 and (𝐿, 𝐾) such that min{𝐾, 𝐿} > 0. Then

𝑞(𝜆𝐿, 𝜆𝐾) = 𝐴(𝜆 𝑎 𝐿 𝑎 ) (𝜆 𝑏 𝐾 𝑏 ) = 𝜆 𝑎+𝑏 ( 𝐴𝐿 𝑎 𝐾 𝑏 ) = 𝜆 𝑎+𝑏 𝑞(𝐿, 𝐾).

Since 𝑎 = 𝑏 = 2/3, 𝜆 𝑎+𝑏 = 𝜆4/3 > 𝜆, and so 𝑞(𝜆𝐿, 𝜆𝐾) > 𝜆𝑞(𝐿, 𝐾). Therefore, the technology
exhibits increasing returns to scale.
(b) The cost-minimizing levels (𝐿 ∗ , 𝐾 ∗ ) are characterized by the system

∗ 0.5 ∗ 0.5
 100(𝐿 ) (𝐾 ) = 10000




𝐾∗ 1 ,
 ∗ =


 𝐿 100
which yields that 𝐿 ∗ = 1000, 𝐾 ∗ = 10.
(c) Let 𝑞 be the required level of output. Then the optimal input combination (𝐿 ∗ , 𝐾 ∗ ) solves the
system
 100(𝐿 ∗ ) 0.5 (𝐾 ∗ ) 0.5 = 𝑞




𝐾 ∗
 ∗ =1


 𝐿
∗ ∗
and so 𝐿 = 𝐾 = 𝑞/100. Therefore, the long-run (total) cost function
𝑞 𝑞 𝑞
𝐶 (𝑞) = 𝑟𝐾 ∗ + 𝑤𝐿 ∗ = + = = 0.02𝑞.
100 100 50

(d) Since the average cost 𝐴𝐶 (𝑄) = 0.05𝑄 −0.5 , which is decreasing in 𝑄, we see that there is
economies of scale in the industry, and so a single producer could provide for the entire market
at the lowest average cost.
The temptation for government regulation, based on economies of scale, is either to have a single
producer (e.g., designate the winner if there are many firms competing) or to legislate a natural
monopoly that arises.
The danger of regulation lies in two main perspectives: incentive distortion and asymmetric
information. More concretely, (i) regulation will for sure exclude competition, but the designated
provider may gain market power and not produce at the socially desired level of average cost; (ii)
there could be rent-seeking behaviors between firms and governments; (iii) the regulator may

55
 not have sufficient information (like that on the production technology) that is needed to regulate
in the right way.

27. All reasonable answers are acceptable. The key points are (i) Opportunity cost must be the value
one can get with one’s degree instead of working on the app (e.g., the value from forfeited internship
opportunities); (ii) Fixed cost is the cost that does not vary in the output level (e.g., rent if applies,
salaries for staff with whom there are long term contracts, the highest salary you would get instead
of working on your own business). Here the output should be measured by the number of customers
the app will serve since the app has been developed. (iii) Sunk cost is the cost that has been paid
but not retractable, like the cost already paid for marketing; (iv) Variable cost in this scenario may
include the cost on server compute time (which can increase when more customers are served) or
data storage.

56
 3

Competitive Market, Market Failure,

and Economic Policies

3 Competitive Markets, Market failure, and Economic Policies

57
Exercise 1. Let 𝑄 𝑑 = 84 − 3𝑝 and 𝑄 𝑠 = 𝑝 − 4 be the demand curve and supply curve for Coke in a city.

(a) Find the total surplus (i.e., the sum of consumer and producer surplus) at the market equilibrium.
(b) If the city government imposes a per unit tax of $8 on Coke, then in the new equilibrium, what is the
percentage tax burden on producers? what is the (actual) price paid by consumers? What is the
deadweight loss caused by the tax?
(c) If the city government, instead of imposing a tax, imposes a supply restriction of 9 on Coke (i.e.,
the total sales of Coke have to be no higher than 9), then in the new equilibrium, what is the price
received by producers? What is the corresponding deadweight loss?

Exercise 2. Suppose demand is 𝑄 𝑑 = 600 − 𝑝 and supply is 𝑄 𝑠 = 𝑝 in the soybean market, where 𝑄
is tons of soybean per year and 𝑝 is the price of a ton of soybean. The government sets a price floor at
𝑝 = 500 and purchases any excess supply at this price. In response, as a long-run adjustment, farmers
switch their crops from corn to soybeans, expanding supply to 𝑄 𝑠0 = 2𝑝.

(a) How does excess supply with the larger supply compare to excess supply prior to the farmers
switching crops?
(b) How much more does the government have to spend to buy up the excess supply, the case without
supply expansion being the reference point?

Exercise 3. Consider an industry where 20 identical firms are operating, each of which has a production
√4
function 𝐹 (𝐾, 𝐿) = 𝐿𝐾. Assume that the prices of both factors are equal to 1.

(a) Determine the total cost function for each of the firms.
(b) Find the supply curve of the industry.
(c) Suppose that the industry demand is 𝑄 𝑑 = 100 − 5𝑝. Find the competitive equilibrium ( 𝑝 ∗ , 𝑄 ∗ ).
(d) The state government wants to finance part of its public expenditure by means of a tax per unit on
the production of the good. Is this feasible if the total expenditure the government wants to finance
is 500? If yes, find the per unit tax that is needed; if no, justify your answer.

Exercise 4. In a city the total number of on-street parking spaces is fixed at 𝑁. The parking spaces are
metered, and the city government has set a uniform charge at $1 per hour, and a car is allowed to stay for at
most 2 hours. According to the estimate of some experts, the equilibrium price for parking spaces is $5
per hour.

(a) What is the direct loss in revenue per hour caused by mispricing to the city government?
(b) Mispricing may cause some other losses in social welfare. State two forms the loss takes.
(c) The city finds that by reducing the maximum stay to 20 minutes the demand for parking spaces can
be reduced so that there are always some available spaces. If the policy is implemented, will it help
eliminate the losses identified in (a)? Explain.

58
 (d) Shopkeepers would oppose raising the price from $1 to $5 per hour, fearing that it would discourage
customers coming to shop. Are their concern justified? Explain.

Exercise 5. There are 100 consumers in a local market of vacuum cleaners, but different consumers may
have different willingness-to-pay (WTP hereafter) for a (new) vacuum cleaner (assume that each consumer
will buy at most one vacuum cleaner). In particular, the distribution of their WTP is given in the table
below:
WTP (in RMB) 0 300 350 375 400 425 450 600
Number of Consumers 16 8 17 34 6 11 5 3

There is a per unit tax of 10 RMB on each purchase of vacuum cleaner (i.e., the tax is paid by consumers).

(a) How many consumers in the market do not need a (new) vacuum cleaner? Explain briefly.
(b) Find the quantity of demand for each of the following prices: 𝑝 = 370, 𝑝 = 410, 𝑝 = 500. Note that
you need to take into account the tax.
(c) If the supply of vacuum cleaner is competitive and the lowest long-run average cost is 370, find the
long-run market equilibrium quantity. What is the (long-run) marginal cost at the equilibrium?
(d) Calculate the consumer surplus in the long-run market equilibrium.
(e) The government wants to increase the local employment via stimulating the consumption of vacuum
cleaners. In particular, it plans to cut the per unit tax by 6 RMB. What will be the new (long-run)
market equilibrium quantity? How will this change the government tax revenue (other things being
equal)?
(f) Now suppose that the supply of vacuum cleaner is still competitive (with the lowest long-run
average cost being 370) , the per unit purchase tax is the same as in (e), but due to some capacity
constraints the maximum quantity of supply is 50. What is the equilibrium quantity? Describe a
simply mechanism that is needed (in addition to the price mechanism as we know it) to achieve the
equilibrium.

Exercise 6. The market demand for natural rubber is given in Table 1. The market is perfectly competitive,
and the costs of each firm are given in Table 2 when it uses its least cost plant in the short run. There are
100 rubber producers in the market.

Table 1 Table 2
Price Quantity demanded Output Marginal cost Average variable cost Average total cost
1.90 1,000 3 2.50 4.00 7.33
2.00 950 4 2.20 3.53 6.03
2.20 800 5 1.90 3.24 5.24
2.91 700 6 2.00 3.00 4.67
4.25 560 7 2.91 2.91 4.34
5.25 400 8 4.25 3.00 4.25
5.50 300 9 8.00 3.33 4.44

59
 (a) Find the (short-run) market equilibrium price 𝑝 𝑆 and quantity 𝑄 𝑆 of rubber.
(b) In the equilibrium, how much rubber does each firm produce?
(c) What is the economic profit made or economic loss incurred by each firm in the equilibrium?
(d) How will the equilibrium quantity change in the long run? Justify your answer. You do not need to
figure out the long-run market equilibrium for this question (but will do for the next question).
(e) What is the market equilibrium price 𝑝 𝐿 and quantity 𝑄 𝐿 , respectively, in the long run? How many
firms will exit the industry in the long run?

Exercise 7. Consider a competitive market in which there are two groups of firms, sector 1 and sector 2,
which use different technologies. The technologies cannot be replicated: entry is impossible and a firm in
one sector cannot adopt the technology of the other sector. The two production sectors and the demand for
the good have the following properties:

• Sector 1: Each firm has the same cost curve, which has a constant marginal cost of 100 but does not
have economies of scale.
• Sector 2: Each firm has the same cost function, which exhibits diseconomies of scale. The aggregate
supply curve for this sector is given in the table below (here “M” means “million”):

Price (in $) 50 75 100 125 150 175 200

Quantity of Supply 1M 2M 3M 4M 5M 6M 7M

The demand curve for this market has the following values:

Price (in $) 50 75 100 125 150 175 200

Quantity of Demand 10M 9M 8M 7M 5M 4M 3M

(a) Suppose sector 2 does not exist (i.e., the market is served purely by sector 1). What is the competitive
equilibrium price? How much is traded in the equilibrium? Does the firms enjoy a positive profit?
Justify your answer.
(b) Suppose, instead, that sector 1 does not exist (i.e., the market is served purely by sector 2). What is
the equilibrium price? How much is traded in the equilibrium? Do the firms earn a (positive) profit?
Justify your answer.
(c) Now suppose that the market is served by both sectors. What are the equilibrium price and quantity?
How much is produced by each sector? Do any of the firms earn a (positive) profit? Explain.
(d) Suppose that the market is served only by sector 1 and that a $25 per-unit sales tax is imposed. What
is the new equilibrium price? How much is traded in the equilibrium? How is the tax burden shared
between firms and consumers?

60
 (e) Suppose that the market is served only by sector 2 and that a $50 per-unit sales tax is imposed. What
is the new equilibrium price? How much is traded in the equilibrium? What is the percentage of the
tax borne by consumers?

Exercise 8. In a market for dry cleaning, the inverse market demand function is given by 𝑝 = 720 − 𝑄,
and the (private) marginal cost of production for the aggregation of all dry-cleaning firms is given by
𝑀𝐶 (𝑄) = 120+𝑄 (𝑄 is the aggregate level of output). The pollution generated by the dry-cleaning industry
leads to external damages, which are captured by the marginal (external) cost curve 𝑀 𝐸𝐶 (𝑄) = 2𝑄. The
market is competitive.

(a) Find the equilibrium price and quantity without government regulation.
(b) What is the socially efficient level of output on the market?
(c) Compute the deadweight loss in the market equilibrium.
(d) The government plans to impose a per unit tax of 𝑇 on dry-cleaning firms. Find the value of 𝑇 at
which the socially efficient level of output on the (competitive) market will be realized.

Exercise 9. A firm has access to five different production processes, each one of which produces the
same amount of daily output but gives off a different amount of pollution per day. The daily costs of the
processes to the firm and the corresponding numbers of tons of smoke are listed in the table below.

Process 𝐴 𝐵 𝐶 𝐷 𝐸
Smoke (tons per day) 4 3 2 1 0
Cost ($ per day) 100 115 140 170 220

(a) If pollution is not regulated, which process will the firm use, and what will be its daily smoke
emission?

The city government wants to cut smoke emissions by half. To accomplish this, it requires a permit for
each ton of smoke emitted. A firm can buy as many permits as it wants and emit smoke up to the quantity
specified by the (total) permits it has purchased.

(b) What is the firm’s willingness to pay for each permit (i.e., the first permit, the second permit, etc.)?
(c) Graph the firm’s demand curve for emission permits, based on your answer to (a).
(d) If the government sets the price for each ton of smoke emission at $20 per day, can the goal of the
government be achieved? Justify your answer.
(e) Assume that the government can only charge a uniform price for each permit. What is the highest
revenue (of selling permits) the government can make, given that the goal (of cutting smoke emission
by half) is achieved? Justify your answer.

61
 ***********

Exercise 10. Many studies have established that charred meat grilled over hot coals may cause cancer.
Since the government cannot easily regulate home cooking methods, an alternative method has been
proposed to discourage the consumption of barbecued meat. The proposal is to place a 100% tax at the
retail level on charcoal briquets (i.e., the price a consumer pays will be 2𝑝 if the price received by the
seller is 𝑝). Suppose the daily demand for charcoal was 𝑄 = 60 − 0.5𝑝 and the supply was 𝑄 = 𝑝 − 30,
where 𝑝 is in dollars per bag and 𝑄 is the number of 20-lb bags of charcoal.

(a) What is the before- and after-tax equilibrium price of charcoal? What is the before- and after-tax
equilibrium quantity of charcoal?
(b) Calculate the percentage of tax that is borne by consumers and producers, respectively.

Exercise 11. Consider demand and supply in the competitive market for xylophones (a kind of musical
instrument)

𝑄 𝑑 = 100 − 𝑝
𝑄 𝑠 = 3𝑝

(a) If the market price is 20, how many units are traded?
(b) What is the total surplus (in the welfare sense) generated by the market at equilibrium?
(c) Using the pass-thru fraction equation (i.e., 𝐸 𝑠 /(𝐸 𝑠 − 𝐸 𝑑 )), calculate the percentage of a (per unit)
tax on xylophones that would be borne by consumers.
(d) Suppose the government, which was previously not in the market for xylophones, commits to buying
20 xylophones on the market, shifting demand. What is the equilibrium quantity after this change?

Exercise 12. The market demand and supply functions are, respectively,

𝑄 𝑑 = 120 − 𝑝
𝑄𝑠 = 𝑝

(a) Calculate the total surplus (in terms of welfare) at equilibrium.

(b) To assist producers, suppose a subsidy of 2 dollars per unit is implemented. How much will the
government spend on the subsidy at the new equilibrium?

Exercise 13. Suppose that the inverse demand function in the market of good 𝑋 is 𝑝 = 200 − 0.1𝑄. The
market for good 𝑋 is competitive, with many small firms. The (aggregate) private marginal cost for all
producers is 𝑀𝐶 = 10 + 0.02𝑄. Unfortunately, the production of good 𝑋 releases small quantities of toxic
chemicals (though 𝑋 is otherwise safe). The marginal external cost is 𝑀 𝐸𝐶 = 0.03𝑄.

(a) What is the efficient quantity in the market?

62
 (b) If the government can charge and enforce a per unit tax on good 𝑋, how large should the tax be in
order to achieve efficiency in the competitive equilibrium?
(c) Why might a per unit tax not be optimal in reality, even if the government could costlessly and
honestly enforce such a policy?

Exercise 14. Total surplus is one of the most important concepts in welfare economics. The purpose of
this exercise is to see if you gave it any thought. To warm up, consider a good that is an experience, like
a football game or a lecture on economics. The marginal cost of providing this experience is zero, but
there is a fixed capacity. There are two types of consumers in this market with different willingness-to-pay
(𝑊𝑇 𝑃 hereafter):

• Fans (𝑊𝑇 𝑃 = 100);

• Non-Fans (𝑊𝑇 𝑃 = 20).

There are 10 members of each group, but we are not able to tell whether a consumer is a fan or not.

(a) Draw the demand curve for the good in a coordinate system (with the convention that the price 𝑝 is
on the vertical axis).
(b) Assume that all prices must be whole numbers. What is the lowest (total) surplus-maximizing price
if the capacity is 10? Carefully justify your answer.
(c) If you were to apply simple supply and demand logic and calculated the area between the curves, the
surplus is potentially greater when the capacity is higher. But is that true here? If the capacity is
11 and the price is set so that exactly 11 people attend, what is the minimum surplus in this case?
Carefully justify your answer.
(d) Early on we argued against interpersonal comparisons of utility; we plead ignorance when asked if
more utility is gained when food goes to a poor person than to a billionaire. In discussing efficiency,
however, we used the concept of total surplus to evaluate policy or behaviors like price discrimination.
Are these positions inconsistent? How are we rationalizing the use of total surplus?

Exercise 15. The area around the Century Avenue campus in Shanghai hosts many small drink and snack
shops (YiDianDian, Happy Lemon, SunTea, TPlus, GongCha, some HK snack places, etc.). Products differ
slightly, but they share many features in common: similar price and selection, labor force characteristics
(as far as I can tell), ingredients. They are mostly simple stalls, with a few notable exceptions (e.g. HeyTea),
often in mall food courts or along the street, without seating and just a one or two employees to take orders
and make drinks behind the counter. Many of these small shops are not owned by the brand name, but are
franchises. Commonly, a small owner-investor gets the use of the name, the recipes, and other operational
support in exchange for a one-time fee and a fraction of sales revenue. The extent of the support varies,
but the idea is that each shop is a separate small business. Individual locations open and close frequently.
Suppose you have an acquaintance who is thinking about owning and operating a milk tea stand. As
a student of microeconomics at NYU Shanghai, they know that you can provide useful comments on

63
their idea. With this limited information, describe the market structure and make predictions about their
prospects. There is flexibility here, but you should at least explain your analysis of the market structure, as
well as potential for economic profit and its determinants in the short and long run.

64
Chapter 3 Answer Key

1. (a) Let 𝑝 ∗ be the equilibrium price and 𝑄 ∗ be the equilibrium quantity. Straightforwardly, we have
𝑄 ∗ = 84 − 3𝑝 ∗ = 𝑝 ∗ − 4, from which we solve 𝑝 ∗ = 22 and 𝑄 ∗ = 18. In addition, the maximum
price 𝑝¯ on the market (i.e., the price at which the demand equals zero) solves 84 − 3 𝑝¯ = 0,
namely, 𝑝¯ = 28. The minimum price 𝑝 on the market (i.e., the price at which the supply equals
zero) solves 𝑝 − 4 = 0, namely, 𝑝 = 4. ¯As is illustrated in Figure 5, the sum of consumer surplus
and producer¯ surplus equals the¯ gray triangular area.

𝑝¯
𝑄𝑠 = 𝑝 − 4
𝑝∗
𝑄 𝑑 = 84 − 3𝑝

𝑝
¯
0 𝑄∗ 𝑄

Figure 5: Economic Efficiency

Therefore, at the market equilibrium,

1 1
Total surplus = ( 𝑝¯ − 𝑝)𝑄 ∗ = (28 − 4) × 18 = 216.
2 ¯ 2
(b) When the demand curve and supply curve are linear, i.e., 𝑄 𝑑 = 𝑎 − 𝑏 𝑝 and 𝑄 𝑠 = 𝑐 + 𝑑𝑝
(𝑎, 𝑏, 𝑑 > 0), the percentage tax burden on consumers 𝑡 𝑐 and producers 𝑡 𝑝 are, respectively,

𝑑 𝑏
𝑡𝑐 = , 𝑡𝑝 = .
𝑏+𝑑 𝑏+𝑑
Thus, after plugging data (i.e., 𝑏 = 3 and 𝑑 = 1), the percentage tax burden on producers is

𝑏 3
𝑡𝑝 = = = 75%.
𝑏+𝑑 3+1
So given a per unit tax 𝑇 = 8, the actual price paid by consumers is (note that 𝑡 𝑐 = 1 − 𝑡 𝑝 )

𝑝 𝑐 = 𝑝 ∗ + 𝑡 𝑐 · 𝑇 = 22 + 25% × 8 = 24.

65
 For the deadweight loss caused by the tax, notice that the new market equilibrium quantity (i.e.,
under the tax) 𝑄 0 = 𝑄 𝑑 ( 𝑝 𝑐 ) = 84 − 3 × 24 = 12. So the deadweight loss, as is illustrated by the
gray area in Figure 6, is

1 1
DWL = 𝑇 (𝑄 ∗ − 𝑄 0) = × 8 × (18 − 12) = 24.
2 2

𝑄𝑠 = 𝑝 − 4
𝑝𝑐
𝑇 𝑄 𝑑 = 84 − 3𝑝

0 𝑄0 𝑄∗ 𝑄

Figure 6: Deadweight Loss of the Per Unit Tax

(c) It is clear that a supply restriction of 9 will cause a shortage on the market of Coke (as without the
restriction the market equilibrium quantity would be 18). So the new equilibrium quantity will be
exactly 9, and for this to be an equilibrium, the new market equilibrium price 𝑝 00 (which is also
the price paid by consumers and that received by producers) must satisfy 9 = 𝑄 𝑑 ( 𝑝 00) = 84−3𝑝 00,
from which we solve 𝑝 00 = 25. The lowest price producers are willing to accept at this level of
output, 𝑝b, solves 9 = 𝑄 𝑠 ( 𝑝b) = 𝑝b − 4, which yields that 𝑝b = 13. The deadweight loss is the gray
area in Figure 7, which equals

1 00 1
( 𝑝 − 𝑝b) (𝑄 ∗ − 9) = (25 − 13) (19 − 9) = 54.
2 2

2. (a) Before the farmers switching crops, the supply curve is 𝑄 𝑠 = 𝑝. At the price floor 𝑝 = 500, the
quantity of demand is 𝑄 𝑑 (500) = 600−500 = 100 while the quantity of supply is 𝑄 𝑠 (500) = 500.
Thus, the excess supply equals

𝑄 Ex = 𝑄 𝑠 (500) − 𝑄 𝑑 (500) = 500 − 100 = 400.

After the switching, the supply curve becomes 𝑄 𝑠0 = 2𝑝. At the price floor 𝑝 = 500, the quantity
of demand is the same as before but that of supply becomes 𝑄 𝑠0 (500) = 1000. Thus, the excess

66
 𝑝

𝑄𝑠 = 𝑝 − 4
𝑝 00

𝑄 𝑑 = 84 − 3𝑝
𝑝b

0 9 𝑄∗ 𝑄

Figure 7: Deadweight Loss due to Supply Restriction

supply now equals

0
𝑄 Ex = 𝑄 𝑠0 (500) − 𝑄 𝑑 (500) = 1000 − 100 = 900.

(b) Using our result in (a), without the switching the excess supply 𝑄 Ex = 400, and so the government
expenditure on clearing the market

𝐸 = 𝑝 · 𝑄 Ex = 500 × 400 = 200, 000.

0 = 900, which requires the government to

With the switching, the excess supply will be 𝑄 Ex
spend
𝐸 0 = 𝑝 · 𝑄 Ex
0
= 450, 000.

3. (a) Let 𝑞 denote the required level of output. Then each firm solves the following cost-minimization
problem √4
min 𝐾 + 𝐿, s.t. 𝐿𝐾 = 𝑞.
𝐾 ,𝐿>0

As is well-known, the optimality condition is that “marginal rate of technical substitution equals
the factor price ratio”, that is, at the optimal input bundle (𝐿 ∗ , 𝐾 ∗ ),

𝑀 𝑃𝐿 𝐾∗
= = 1,
𝑀 𝑃𝐾 (𝐿 ∗ ,𝐾 ∗ ) 𝐿∗

which, together with the constraint that 𝐿 ∗ 𝐾 ∗ = 𝑞 4 , yields that 𝐿 ∗ = 𝐾 ∗ = 𝑞 2 . Therefore, the
total cost function
𝑇𝐶 (𝑞) = 𝐾 ∗ + 𝐿 ∗ = 2𝑞 2 .

(b) For each firm, when the market price 𝑝 is no lower than its shut-down price, the optimal output
level 𝑞 ∗ solves 𝑀𝐶 (𝑞 ∗ ) = 𝑝; otherwise the output will be zero. Let 𝑞 0 achieves the lowest

67
 average cost. Then 𝐴𝐶 (𝑄 0 ) = 𝑀𝐶 (𝑄 0 ), which, using the expression for the total cost, is

2𝑞 0 = 4𝑞 0 .

Thus 𝑞 0 = 0 and the shut-down price equals 0. Therefore, for all 𝑝 > 0, 4𝑞 ∗ = 𝑝, which yields
the individual supply curve
𝑝
𝑞∗ = ,
4
and so the industry supply curve is

𝑄 𝑠 = 20𝑞 ∗ = 5𝑝.

(c) ( 𝑝 ∗ , 𝑄 ∗ ) is characterized by 5𝑝 ∗ = 𝑄 𝑠 ( 𝑝 ∗ ) = 𝑄 ∗ = 𝑄 𝑑 ( 𝑝 ∗ ) = 100 − 5𝑝 ∗ . Solving the system,

we obtain 𝑝 ∗ = 10 and 𝑄 ∗ = 50.
(d) This is not feasible. To see the logic, let 𝑇 > 0 denote the level of a per-unit tax. With the
tax, the new market equilibrium ( 𝑝𝑇 , 𝑄𝑇 ) is characterized by 5( 𝑝𝑇 − 𝑇) = 𝑄 𝑠 ( 𝑝𝑇 ) = 𝑄𝑇 =
𝑄 𝑑 ( 𝑝𝑇 ) = 100 − 5𝑝𝑇 . Solving the system we get 𝑄𝑇 = 50 − 2.5𝑇. So the total tax revenue
𝑅(𝑇) = 𝑇𝑄𝑇 = 𝑇 (50 − 2.5𝑇). Notice that the highest tax revenue is achieved when 𝑇 = 10
(letting 𝑅 0 (𝑇) = 0 or simply by inspecting) and is equal to 𝑅(10) = 10(50 − 2.5 × 10) = 250.
Thus, it is impossible to finance a 500 expenditure via a per-unit tax in the market.

4. (a) The price is below the equilibrium level by $5 − $1 = $4 (per hour). Thus, the direct revenue
loss per hour to the city government is 4𝑁.
(b) (i) Underpricing causes a shortage in supply of parking spaces, and it is very likely that many
cars, attracted by the low price, are circling and waiting for spaces, which leads to unnecessary
time and fuel loss to their drivers; (ii) Circling and waiting cars may also cause traffic congestion,
which may reduce the welfare of other vehicles (which are not searching or waiting for parking
spaces; (iii) Highest value users may not be able to get sufficient parking spaces; (iv) People
might be forced to park at a distance from their destination; etc..
(c) The policy can only reduce the social value of the parking spaces (i.e., people’s willingness-to-pay
is reduced since now the “quality” of parking spaces are not as good as before). This will for
sure reduce the demand and make it look like that the loss identified in (a) is reduced (since now
the actual price is roughly the equilibrium price). But this does not eliminate the gap between
people’s willing-to-pay for “good” service and the inefficiently low price. In contrast, it causes
dissipation of social value. Therefore, the policy does not essentially reduce the loss as identified
in (a).
(d) Not fully justified. The change of price may not change the real cost of shopping: indeed, with
higher price, the cost due to circling and waiting will be lower. Moreover, a counterargument
might be that higher parking price deters some low income individuals but do not deter high
income individuals, but the latter group is more likely to spend more on shopping.

68
5. (a) Consumers who do not need a new vacuum must have non positive WTP. Thus, according to the
table, there are 16 consumers who do not need a new vacuum since their WTP is zero.
(b) People will purchase if and only if the price is no higher than their WTP. Therefore, when
𝑝 = 370, the actual price paid by consumers is 370 + 10 = 380. The number of consumers whose
WTP is higher than 380 in the market equals 6 + 11 + 5 + 3 = 25, and so the quantity of demand
in this case is 25. Similarly, the quantity of demand at 𝑝 = 410 is 19, and that at 𝑝 = 500 is 3.
(c) In the long-run equilibrium of a competitive market, the price must be equal to the lowest
long-run average cost, that is, 370, and the actual price to consumers is 380. Using our result in
(b), we see that the equilibrium quantity is 25. At the lowest long-run average cost, the marginal
cost coincides with the average cost. Thus, the long-run marginal cost at the equilibrium is also
370.
(d) Since the long-run actual price equals 380, by definition, the (total) consumer surplus

CS = 3(600 − 380) + 5(450 − 380) + 11(425 − 380) + 6(400 − 380) = 1625.

(e) Now the actual price paid by consumers is 370 + 4 = 374. According to the information from
the table, the total quantity of demand is now 34 + 6 + 11 + 5 + 3 = 59 (i.e., consumers whose
WTP is 375 are added to the demand). The tax revenue before is 10 × 25 = 250, while after the
tax reduction it will be (10 − 6) × 59 = 236. Thus, the tax revenue will decrease by 14.
(f) Since the maximum quantity of supply (50) is lower than the equilibrium quantity (59), the new
equilibrium quantity should be 50. Notice that now the price mechanism alone is not able to
achieve the market equilibrium any more: If the price is lower than 371, then the market demand
will be more than 59, while if it is higher than 371, the market demand will be less than 25. Thus,
there must be some auxiliary mechanism that helps to exclude excess demand. One example is
the queuing mechanism, that is, the price should be somewhere between 370 and 371 (so there
will be sufficiently many people queuing), but only the first 50 consumers can buy the vacuum.

6. (a) Note that in equilibrium each firm’s output level must equate its marginal cost with the market
price, while the quantity of demand must be equal to the quantity of supply. It is straightforward
to see from the tables that the quantity of demand and that of supply are all equal to 700 when
𝑝 = 2.91. Hence, the short-run market equilibrium ( 𝑝 𝑆 , 𝑄 𝑆 ) = (2.91, 700).
(b) It can be directly read from Table 2 that each firm produces 7 units of rubber in the equilibrium.
(c) When 𝑞 = 7, the average total cost is 4.34. Thus, the profit of a firm

𝜋 = (2.91 − 4.34) × 7 = −10.01.

Thus, each firm in the short run actually incurs a loss of 10.01.
(d) Since each firm incurs a loss in the short run, some firms will exit the market in the long run and
so the equilibrium quantity will decrease in the long run. The decrease in supply will then drive
the equilibrium price up to a level that is sufficient to cover the average cost for the staying firms.

69
 (e) In the long-run equilibrium, firms produce at the lowest average cost. Let 𝑞 ∗ denote the output
level achieving the lowest average cost for a single firm. It is well-known that 𝑀𝐶 (𝑞 ∗ ) = 𝐴𝐶 (𝑞 ∗ ).
Inspecting Table 2, we see that 𝑞 ∗ = 8, and the corresponding market price equals the lowest
average cost, that is, 𝑝 𝐿 = 4.25. The equilibrium quantity 𝑄 𝐿 = 560, as can be read from Table
1. It is then easy to see that the number of firms in the long run 𝑄 𝐿 /𝑞 ∗ = 560/8 = 70, which
implies that 100 − 70 = 30 firms will exit the industry in the long run.

7. (a) The competitive equilibrium price must equal the marginal cost of a firm. Thus, the equilibrium
price equals 100. At this price, the quantity of demand, which is also the quantity traded, equals
8M. Each firm earns zero profit: since there is no economies of scale while the marginal cost is
constant, it follows that there is no fixed cost (or cost that does not rely on the quantity of output
for all positive output), and so the marginal cost equals the average cost, which implies that the
economic profit is zero.
(b) By inspecting the two tables, it is easy to see that the quantity of demand and that of supply are
all equal to 5M when 𝑝 = 150. Thus, the equilibrium price is 150 and the quantity traded is 5M.
Each firm in the equilibrium must earn a positive profit: Each firm has an increasing 𝐴𝐶 curve
(due to diseconomies of scale), which implies that the 𝑀𝐶 curve must be everywhere above the
𝐴𝐶 curve. Since 𝑝 = 𝑀𝐶 in a firm’s optimum, we deduce that 𝑝 > 𝐴𝐶, which then implies that
each firm earns a positive profit.
(c) The equilibrium price must be 100. If 𝑝 < 100, then sector 1 will not produce anything and
sector 2 will produce no more than 3M while the quantity of demand is at least 8M. If, instead,
𝑝 > 100, then firms in sector 1 would want to produce an infinite quantity while the total demand
is no more than 8M.
Given that 𝑝 = 100, sector 2 will produce 3M, and so sector 1 will produce 8M − 3M = 5M,
where 8M is the quantity of demand.
As explained in (a) and (b), no firm in sector 1 earns a positive profit, while each firm that
produces in sector 2 earns a positive profit.
(d) Note that the price received by firms in sector 1 must be 100: below 100 these firms would
produce nothing while above 100 these firms would want to produce an infinite amount. Hence,
the price paid by the consumers, which is also the equilibrium price, must be 100 + 25 = 125.
The quantity traded in equilibrium simply equals the quantity of demand at 𝑝 = 125, which is
7M. Since the supply has infinite elasticity, all tax burden is borne by the consumers.
(e) Let 𝑝 ∗ be the equilibrium price. Due to the per-unit tax, the price received by firms is 𝑝 ∗ − 50.
Thus, we must have 𝑄 𝑑 ( 𝑝 ∗ ) = 𝑄 𝑠 ( 𝑝 ∗ − 50). Inspecting the tables, we see that 𝑝 ∗ = 175 (note
that the tax must lead the price to increase, and so you only need to try prices above the original
equilibrium price 150). Under the price, the equilibrium quantity is 4M.
The tax leads the price paid by consumers to raise by 175 − 150 = 25. Thus, the percentage of
tax borne by consumers is 25/50 = 50%.

8. (a) Let ( 𝑝 ∗ , 𝑄 ∗ ) be the equilibrium without government regulation. Then firms only take into

70
 accounter the private marginal cost of production when deciding the optimal output level. Thus,
the market supply satisfies 𝑝 = 𝑀𝐶 (𝑄) = 120 + 𝑄. Therefore, 𝑄 ∗ = 𝑝 ∗ − 120 = 720 − 𝑝 ∗ , from
which we solve 𝑝 ∗ = 420 and 𝑄 ∗ = 300.
(b) The socially efficient output level 𝑄 𝐸 equates the social marginal cost (the sum of private
marginal cost and the marginal external cost) and the social marginal benefit (given by the
inverse demand function). Thus, we have

720 − 𝑄 ∗ = 𝑀𝐶 (𝑄 ∗ ) + 𝑀 𝐸𝐶 (𝑄 ∗ ) = 120 + 3𝑄 ∗ ,

and so 𝑄 ∗ = 150.
(c) The deadweight loss equals the area between the social marginal cost curve and the demand
curve over the excess output that is produced inefficiently, that is,
∫ 𝑄∗ ∫ 300
DWL = [𝑀𝐶 (𝑞) + 𝑀 𝐸𝐶 (𝑞) − 𝑝(𝑞)] d𝑞 = (4𝑞 − 600) d𝑞 = 45000,
𝑄𝐸 150

where 𝑝(·) is the inverse demand function (which measures the marginal social value of each
additional unit of product).
(d) After being imposed the per unit tax of 𝑇, the marginal cost of firms becomes 𝑀𝐶 (𝑄) + 𝑇 =
120 + 𝑄 + 𝑇, and the market supply becomes 𝑝 = 120 + 𝑄 + 𝑇. Let ( 𝑝𝑇 , 𝑄𝑇 ) be the new
equilibrium under the tax. Then

𝑄𝑇 = 𝑝𝑇 − (120 + 𝑇) = 720 − 𝑝𝑇 ,

from which we solve 𝑄𝑇 = 300 − 0.5𝑇. To achieve the socially efficient level of output, 𝑇 should
be set so that 𝑄𝑇 = 𝑄 𝐸 , that is, 300 − 0.5𝑇 = 150, and so 𝑇 = 300.

9. (a) Without regulation, the firm will clearly choose the least costly production process, that is,
process 𝐴, and the corresponding daily smoke emission is 4 tons.
(b) With one permit of emission, the firm can choose process 𝐷, which is 220 − 170 = 50 cheaper
than process 𝐸. Thus, if the first permit costs no more than 50 then the firm will buy it. In other
words, the firm’s willingness-to-pay (WTP thereafter) for the first permit is 50. Similarly, with
two permits or emission, the firm can choose process 𝐶, which is 170 − 140 = 30 cheaper than
process 𝐷. So if the second permit costs no more than 30 then the firm will buy it, provided that
it has purchased one permit. That says, the firm’s WTP for the second permit is 30. Its WTP for
other permits can be computed in the same way. The answer is summarized in the table below.

No. of permit 1st 2nd 3rd 4th

WTP 50 30 25 15

(c) The firm’s demand curve is graphed in Figure 8.

71
 𝑝
50

30
25

15

0 1 2 3 4 𝑄

Figure 8: The Firm’s Demand for Emission Permits

(d) When the price is $20, it is can be seen from the demand curve that the quantity of demand for
permits will be 3 (since the firm’s WTP for each of the first three permits is higher than 20 while
that for the fourth unit is less than 20). Thus, the goal of the government cannot be achieved
since now the emission will be 3 tons per day, which is 25%, rather than 50% off the original
level.
(e) The highest price is clearly equal to the WTP for the second permit, that is, 30, as the firm’s
WTP is decreasing. Under this price, the government’s revenue is 30 × 2 = 60.

10. (a) Let ( 𝑝 𝑏 , 𝑄 𝑏 ) and ( 𝑝 𝑎 , 𝑄 𝑎 ) be the equilibrium before- and after-tax, respectively. Then we have
( (
𝑄 𝑏 = 60 − 0.5𝑝 𝑏 𝑄 𝑎 = 60 − 0.5 · 2𝑝 𝑎
, ,
𝑄 𝑏 = 𝑝 𝑏 − 30 𝑄 𝑎 = 𝑝 𝑎 − 30

from which we solve ( 𝑝 𝑏 , 𝑄 𝑏 ) = (60, 30) and ( 𝑝 𝑎 , 𝑄 𝑎 ) = (45, 15).

(b) The tax borne by consumers equals 2𝑝 𝑎 − 𝑝 𝑏 = 30, while that borne by producers equals
𝑝 𝑏 − 𝑝 𝑎 = 15. Since the total tax equals 45, it is then easy to see that the percentage of the tax
borne by consumers is 30/45 = 2/3 ≈ 66.67% and that by producers is 15/45 = 1/3 ≈ 33.33%.

11. (a) Since 𝑄 𝑑 (20) = 100 − 20 = 80 and 𝑄 𝑠 (20) = 3 × 20 = 60, it is clear that there is a shortage
(under-provision) when 𝑝 = 20, and so the units that are traded equal min{𝑄 𝑑 (20), 𝑄 𝑠 (20)} = 60.
(b) Let ( 𝑝 ∗ , 𝑄 ∗ ) denote the market equilibrium. Then 𝑄 ∗ = 100 − 𝑝 ∗ = 3𝑝 ∗ , which yields that
𝑝 ∗ = 25 and 𝑄 ∗ = 75. To find the total surplus, let 𝑝¯ and 𝑝 be the maximum and minimum
¯ = ¯0 and 𝑄 𝑠 ( 𝑝) = 0, that is,
prices, respectively, for the market. By definition, 𝑄 𝑑 ( 𝑝)
¯
(
100 − 𝑝¯ = 0
,
3𝑝 = 0
¯

72
 and so 𝑝¯ = 100, 𝑝 = 0. Therefore, the total surplus at the equilibrium equals
¯
1 1
( 𝑝¯ − 𝑝)𝑄 ∗ = (100 − 0) × 75 = 3750.
2 ¯ 2

(c) Let ( 𝑝 0, 𝑄 0) be the equilibrium when a (per unit) tax is present. Then the supply elasticity at the
equilibrium
d𝑄 𝑠 𝑝 0 3𝑝 0
𝐸𝑠 = · = 0,
d𝑝 ( 𝑝0 ,𝑄0) 𝑄 0 𝑄
and the demand elasticity at the equilibrium

d𝑄 𝑑 𝑝 0 −𝑝 0
𝐸𝑑 = · = 0.
d𝑝 ( 𝑝0 ,𝑄0 ) 𝑄0 𝑄

Thus, using the pass-thru fraction equation, the percentage of a per unit tax on xylophones that
would be borne by consumers is

𝐸𝑠 3𝑝 0/𝑄 0 3
𝑡𝑐 = = 0 0 0 0
= = 75%.
𝐸 𝑠 − 𝐸 𝑑 (3𝑝 /𝑄 ) + ( 𝑝 /𝑄 ) 4

(d) With the additional 20 units of demand from the government, now the demand curve 𝑄
b𝑑 =
𝑄 𝑑 + 20 = 120 − 𝑝. Let ( 𝑝 ∗∗ , 𝑄 ∗∗ ) be the new equilibrium. Then



 b𝑑 ( 𝑝 ∗∗ ) = 120 − 𝑝 ∗∗
𝑄



𝑄 𝑠 ( 𝑝 ∗∗ ) = 3𝑝 ∗∗ ,

 𝑄 ∗∗ = 𝑄 b𝑑 ( 𝑝 ∗∗ ) = 𝑄 𝑠 ( 𝑝 ∗∗ )



from which we solve 𝑄 ∗∗ = 90 (and 𝑝 ∗∗ = 30).

12. (a) To determine the total surplus, we need to pin down the equilibrium ( 𝑝 ∗ , 𝑄 ∗ ), the maximum price
¯ and the minimum price 𝑝, for the market. For the equilibrium, we have 𝑄 ∗ = 120− 𝑝 ∗ = 𝑝 ∗ , and
𝑝,
¯ the maximum and minimum prices, note that they are characterized
so ( 𝑝 ∗ , 𝑄 ∗ ) = (60, 60). For
by (
𝑄 𝑑 ( 𝑝)
¯ = 120 − 𝑝¯ = 0
,
𝑄 𝑠 ( 𝑝) = 𝑝 = 0
¯ ¯
and so 𝑝¯ = 120 and 𝑝 = 0. Therefore, the total surplus equals
¯
1 1
( 𝑝¯ − 𝑝)𝑄 ∗ = (120 − 0) × 60 = 3600.
2 ¯ 2

(b) Let ( 𝑝 0, 𝑄 0) be the equilibrium with the per unit subsidy. Now the market supply 𝑄
b𝑠 = 𝑝 + 2.
0 0 0 0 0
Thus, 𝑄 = 𝑝 + 2 = 120 − 𝑝 , from which we solve 𝑝 = 59 and 𝑄 = 61. So the total expenditure
spent on the subsidy is 2𝑄 0 = 2 × 61 = 122.

73
13. (a) The social marginal cost is 𝑀𝐶 (𝑄) + 𝑀 𝐸𝐶 (𝑄) = 10 + 0.05𝑄. The efficient quantity 𝑄 𝐸 equates
the social marginal cost and the social marginal value (the price consumers are willing to pay),
that is,
10 + 0.05𝑄 𝐸 = 200 − 0.1𝑄 𝐸

and so 𝑄 𝐸 ≈ 1266.67.
(b) A per unit (Pigouvian) tax of 𝑡 will increase the marginal cost by 𝑡. So the new market equilibrium
quantity 𝑄 𝑡 satisfies
10 + 0.02𝑄 𝑡 + 𝑡 = 200 − 0.1𝑄 𝑡 .

For the outcome to be socially efficient, we must have 𝑄 𝑡 = 𝑄 𝐸 . So replacing 𝑄 𝑡 with 𝑄 𝐸 (i.e.,
1266.67) in the equation above, we obtain 𝑡 ≈ 38.
(c) The major problem is that the government may not have sufficient information about the social
marginal cost. Note that you cannot simply put “because the tax will twist the competitive
market and cause a deadweight loss”, since the price mechanism (competitive market) fails to
allocate resources efficient here due to the existence of (negative) externality, and the Pigouvian
tax serves as a means to correct the misallocation (underpricing), which, if well implemented,
enhances rather than jeopardizes the social welfare.

14. (a) Since the highest 𝑊𝑇 𝑃 on the market is 100, the quantity of demand will be zero when 𝑝 > 100.
When 20 < 𝑝 6 100, only the Fans are willing to consume, and so the quantity of demand
equals 10. When 𝑝 6 20, all agents on the market are willing to consume, and so the quantity of
demand becomes 20. The demand curve is graphed in Figure 9 below.
𝑝

100

20

0 10 20 𝑄

Figure 9: The Market Demand

(b) Clearly, for each (feasible) price level, A fan generates higher surplus than a non-fan. Thus, to
maximize the total surplus, fans should have the priority to get the good. Since the total capacity
is 10 while there are 10 fans, the total surplus is maximized when it is exactly those 10 fans
who consume the good. For a price mechanism to implement this outcome, it suffices to set
the price at a level in [21, 100], and so the lowest price that will work is 21. Notice that we

74
 cannot guarantee that all fans are served if the price is set at 20 or at an even lower price, because
then all consumers (fans and non-fans) are also willing to pay the price while we cannot tell
whether a consumer is a fan or not. Moreover, this fails to support an equilibrium (since supply
and demand do not match) if we only consider the price mechanism (i.e., without considering
auxiliary mechanisms like queuing mechanism that can help truncate the demand).
(c) The worst scenario is that the least fans are happened to be served. Provided that the capacity is
11, this is equivalent to saying that all 10 non-fans and 1 fan are served. In this case, the total
surplus equals 10 × 20 + 1 × 100 = 300, which happens only if the price is no higher than 20.
(d) It is rationalized (J.S. Mills style) by assuming that there can be transfers between people, the
winners can compensate the losers by more than the loss. Criticism, as by R. Coase (referred
to in a class post on natural monopolies), might begin by acknowledging how ridiculous that
assumption is in most contexts. The answer to this question is generically open, but you should
touch the idea of transfer/compensating payments if you think the positions are consistent, or
carefully justify your objections to the consistency.

15. Although generically open, the key points are: (i) the market structure is highly competitive; and (ii)
in the long run the economic profit will be very low, even zero.

75
 4

Monopoly and Pricing Strategies

4 Monopoly and Pricing Strategies

76
Exercise 1. A monopolist has a demand function given by 𝑄 = 100 − 𝑝 and a total cost function given by
𝐶 (𝑄) = 𝑄 2 + 16.
(a) Derive the marginal cost and the marginal revenue.
(b) Find the monopolist’s profit-maximizing quantity and price.
(c) How much economic profit will the monopolist earn?

Exercise 2. Suppose the market of carbon fiber is monopolized by a firm. The market demand for carbon
fiber per year is 𝑄 = 100 − 𝑝 while the total cost of production is 𝐶 (𝑄) = 1000 + 20𝑄. The firm can only
charge all customers the same price.
(a) Derive first the firm’s total revenue and marginal revenue as a function of output level 𝑄 (recall that
you need to first express 𝑝 as a function of 𝑄, i.e., derive the so-called inverse demand function);
(b) Derive the firm’s marginal cost;
(c) What is the profit-maximizing level of output for the firm? That is, find the output level that equates
the marginal cost and marginal revenue. What is the corresponding optimal price the firm should
charge?
(d) Now suppose that the firm is, for some reason, faced with a capacity constraint so that it can produce
no more than 30 units of carbon fiber. Find the profit-maximizing price in this case.

Exercise 3. Drybar, a hair salon, faces a weekly demand for blowouts that is 𝑞 = 1120 − 20𝑝. The salon
has weekly total cost of 𝑇𝐶 (𝑞) = 0.05𝑞 2 + 20𝑞 + 500.
(a) What is the minimum scale (i.e., the quantity that achieves the minimum average cost) of the salon?
(b) Derive the weekly marginal revenue for this salon.
(c) Find the salon’s (weekly) profit-maximizing level of output and price.
(d) Find the producer and consumer surplus, respectively, at the profit-maximizing level of output.
(e) Find the producer and consumer surplus, respectively, when the salon is exercising first-degree price
discrimination.

Exercise 4. A monopolist has a cost function given by 𝐶 (𝑄) = 𝑄 2 and faces a demand curve given by
𝑄 𝑑 = 120 − 𝑝. Price discrimination is banned in the market and so the monopolist can only charge all
consumers the same price.
(a) What is the profit-maximizing level of output? What price will the monopolist charge?
(b) If a lump-sum tax of 2000 were imposed on the monopolist, what would be its profit-maximizing
level of output?
(c) If you wanted to choose a price ceiling 𝑝¯ for this monopolist so as to maximize the total surplus (i.e.,
the sum of consumer and producer surplus), what should be the value of 𝑝? ¯ How much would the
monopolist produce under the price ceiling?

77
 (d) If a 40% sales tax was imposed on the monopolist, what would be its profit-maximizing level of
output and price, respectively? Would the tax improve consumer surplus? Justify your answer.

Exercise 5. In the market of good 𝑋, the demand curve is given by 𝑄 𝑑 = 144/𝑝 2 . The production of good
√
𝑋 is monopolized by firm 𝑌 , which is operating with average variable cost 𝐴𝑉𝐶 (𝑄) = 𝑄 and a fixed
cost of 5.

(a) What are its profit-maximizing price and output level? What is the resulting economic profit?
(b) Suppose the government regulates the price to be no higher than $4 per unit. How much will the
monopolist produce (and sell)? What will its profit be?
(c) Suppose the government wants to set a ceiling price that induces the monopolist to produce the
socially optimal level of output. What ceiling price will do this?

Exercise 6. A firm sells its cars in Brazil and India. The demand functions in Brazil and India are,
respectively,
 2  5
𝐴 𝐴
𝑄𝐵 = , 𝑄𝐼 = ,
𝑝𝐵 𝑝𝐼
where 𝐴 > 0 is a constant, 𝑝 𝐵 is the price in Brazil, and 𝑝 𝐼 is that in India. The marginal cost of producing
one car is $10000.

(a) Find the price elasticity of demand for car in Brazil.

Suppose resales across countries are prohibitively expensive due to transportation costs and language
difference so that one can consider the two markets totally separated and the firm (as a monopolist) can
charge different prices in different countries.

(b) Calculate the optimal prices the firm will charge in Brazil and India, respectively. In which country
is the mark-up higher?

Now suppose that each country levies a $1000 per vehicle tax on the firm.

(c) Find the optimal prices the firm will charge in Brazil and India, respectively, when the tax is present;
(d) In which country will the price increase more? Explain why that is the case.

Exercise 7. Consider the market for the H-Jeans. H-Jeans are sold by a single firm that carries the patent
for the design. On the demand side, there are 𝑛 𝐻 = 200 consumers who are willing to pay a maximum
amount of 𝑉 𝐻 = 20 for a pair of H-Jeans, and 𝑛 𝐿 = 300 consumers who are willing to pay at most 𝑉 𝐿 = 10
for a pair of H-Jeans. Each consumer chooses whether to buy one pair of jeans or not to buy it at all.

(a) Draw the market aggregate demand curve facing the firm.
(b) The monopolist can produce each unit at a cost of 𝑐 = 5. Suppose the firm cannot price discriminate
and is therefore constrained to set a uniform price. Find the profit-maximizing price set by H-Jeans,
and the profit earned by the firm.

78
 (c) Compute the profit level made by the firm assuming now that it can price discriminate between the
two groups of consumers. Does the monopolist benefit from price discrimination? Is there any
group of consumers that will benefit from price discrimination? Justify your answer.

Exercise 8. In a market for dry cleaning, the inverse market demand function is given by 𝑝 = 720 − 𝑄.
The (private) marginal cost of production for the aggregation of all dry-cleaning firms is given by
𝑀𝐶 (𝑄) = 120 + 𝑄. The pollution generated by the dry-cleaning business creates external damages that
are captured by the marginal external cost curve 𝑀 𝐸𝐶 (𝑄) = 2𝑄, but there is no regulation on the market.

(a) Find the competitive equilibrium (including price and quantity) for the market.
(b) Suppose the business of dry-cleaning is monopolized by a single firm. Find the profit-maximizing
output level and the corresponding price.
(c) Compute the social welfare (in the sense of the total social surplus) in situation (a) and situation (b),
respectively. Which market structure yields higher level of social welfare?

Exercise 9. A monopolist sells in two (separated) markets. The inverse demand curve in market 𝐴 is

𝑝 𝐴 = 200 − 𝑞 𝐴

while the inverse demand curve in market 𝐵 is

𝑝 𝐵 = 300 − 2𝑞 𝐵 .

The firm’s total cost function is

𝐶 (𝑞) = 150 + 𝑞 2 .

The monopolist is able to exercise third-degree price discrimination.

(a) What quantities will the monopolist sell in the two markets, respectively?
(b) What price will be charged in each market?

Exercise 10. Crazy Harry, a monopolist, has a total cost curve given by 𝐶 (𝑄) = 15 + 5𝑄. He sets two
prices for his product, a regular price 𝑝 𝑟 , and a discount price 𝑝 𝑑 . Everyone is eligible to purchase the
produce at 𝑝 𝑟 . To be eligible to buy at 𝑝 𝑑 , it is necessary to present a copy of the latest Crazy Harry
newspaper ad to the salesclerk. Suppose the only buyers who present the ad are those who would not have
been willing to buy the product at 𝑝 𝑟 . Crazy Harry’s demand curve is 𝑄 𝑑 = 4 − 0.2𝑝.

(a) What are the profit-maximizing values of 𝑝 𝑟 and 𝑝 𝑑 ?

(b) How much economic profit does Harry make?
(c) How much profit would Harry have made if he had been forced to charge the same price to all
buyers?

79
 (d) Are buyers better or worse off as a result of Harry’s being able to conduct price discrimination?
Justify your answer.

Exercise 11. The market of good 𝑋 is monopolized by firm 𝐴. There are two types of consumers on the
market: students and non-students. The demand of students is

𝑄 𝑆 = 100 − 2𝑝,

and that of non-student is

𝑄 𝑁 = 120 − 3𝑝.

Let 𝑄 𝑀 denote the market demand (i.e., 𝑄 𝑀 = 𝑄 𝑆 + 𝑄 𝑁 ). Firm 𝐴 is operating with total cost
𝑇𝐶 (𝑄) = 10𝑄, where 𝑄 is the quantity of output. Due to a capacity constraint, firm 𝐴 can produce at
most 80 units of 𝑋.

(a) Derive the market demand curve.

(b) Suppose that firm 𝐴 can only charge all consumers the same (per unit) price. What is the optimal
price? How many units are sold to students and non-students, respectively?
(c) Suppose that firm 𝐴 can exercise third-degree price discrimination and charge students and non-
students different per unit prices. Calculate the optimal prices and the firm’s economic profit.

Exercise 12. On the market of good 𝑌 there are 20 consumers. Each consumer has a demand function

𝑞 = 180 − 2𝑝.

Good 𝑌 is produced by a monopolist, which is operating with a constant marginal cost of 50 and a fixed
cost of 2000.

(a) Under first-degree price discrimination, what is the quantity of good 𝑌 that are sold on the market?
What is the price for the 8th unit of good 𝑌 to each consumer?
(b) Suppose the monopolist adopts the following pricing strategy: The per unit price is 80 unless one
buys more than 80 units, and for each unit above 80 the per unit price will be 60 (e.g., if one buys 90
units, then he pays 80 × 80 + (90 − 80) × 60 = 7000). What is the quantity of good 𝑌 each consumer
will buy? What is the monopolist’s economic profit?

Exercise 13. Assume there are two consumers on the market of good 𝑍. Consumer 1 has a demand
function
𝑞 1 = 17 − 𝑝

80
and consumer 2 has a demand function
𝑞 2 = 32 − 2𝑝.

The producer on the market is a monopolist whose marginal cost 𝑀𝐶 = 4.

(a) Suppose the monopolist uses a two-part tariff to realize first-degree price discrimination (i.e., the
monopolist can distinguish consumers). Find the two-part tariff.
(b) Suppose that the monopolist offers a plan that charges a lump sum fee of 130 and a per unit price of
4. How many units of good 𝑍 will consumer 1 purchase? What is consumer 2’s surplus?

***********

Exercise 14. A profit-maximizing monopolist offers an amazing new service. The cost function is

𝐶 = 6𝑄 + 500.

The monopolist can classify their potential customers into two categories: The first group has demand
function
𝑄 = 190 − 𝑝.

The second group has inverse demand function

𝑝 = 46 − 0.5𝑄.

What is the price the monopolist should charge customers from the first group? What is the quantity the
monopolist will sell to the second group?
Exercise 15. A small biotech company develops a new treatment for a rare disease. The new treatment is
patented and the company is the sole monopolist in the market. The company can sell the treatment to
private pharmacies and public hospitals. The pharmacies’ demand for the treatment is

𝑄 𝑃 = 84 − 0.4𝑝 𝑃 ,

and the public hospitals’ demand for the treatment is

𝑄 𝐻 = 116 − 0.6𝑝 𝐻 .

The marginal cost of the new treatment is 𝑀𝐶 (𝑄) = 20 + 2𝑄. Suppose the current laws bans price
discrimination and so the biotech company can only charge all buyers the same price.
(a) Find the company’s optimal price.
(b) How many units of treatment are sold to pharmacies and public hospitals, respectively?
Now suppose that the legislature passes a new Health Costs Relief Act (HCRA) that allows biotech
companies to exercise price discrimination.

81
(c) Once the new law is enacted, what will be the price for pharmacies? What will be the price for
public hospitals? How many units of treatment will be sold to pharmacies and public hospitals,
respectively?
(d) Who gains and who loses from the enactment of the new HCRA? Justify your answer.

82
Chapter 4 Answer Key

1. (a) It is straightforward to have 𝑀𝐶 (𝑄) = d𝐶 (𝑄)/d𝑄 = 2𝑄. For the marginal revenue, we first find
the total revenue 𝑇 𝑅(𝑄) = 𝑝(𝑄)𝑄 = (100 − 𝑄)𝑄 = 100𝑄 − 𝑄 2 , and then the marginal revenue
follows from differentiating 𝑇 𝑅(𝑄) with respect to 𝑄: 𝑀 𝑅(𝑄) = d𝑇 𝑅(𝑄)/d𝑄 = 100 − 2𝑄.
(b) Let ( 𝑝 ∗ , 𝑄 ∗ ) be the optimal price and quantity. The optimal quantity equates the marginal cost
and the marginal revenue, that is, 𝑀𝐶 (𝑄 ∗ ) = 𝑀 𝑅(𝑄 ∗ ), which is equivalent to

2𝑄 ∗ = 100 − 2𝑄 ∗ ,

and so 𝑄 ∗ = 25, 𝑝 ∗ = 𝑝(𝑄 ∗ ) = 100 − 𝑄 ∗ = 75.

(c) The monopolist’s economic profit

𝜋 = 𝑝 ∗ 𝑄 ∗ − 𝐶 (𝑄 ∗ ) = 25 × 75 − (252 + 16) = 1234.

2. (a) The total revenue 𝑇 𝑅(𝑄) = 𝑝(𝑄)𝑄 = (100 − 𝑄)𝑄 = 100𝑄 − 𝑄 2 , and the marginal revenue
𝑀 𝑅(𝑄) = d𝑇 𝑅(𝑄)/d𝑄 = 100 − 2𝑄.
(b) The marginal cost 𝑀𝐶 (𝑄) = d𝐶 (𝑄)/d𝑄 = 20.
(c) Let ( 𝑝 ∗ , 𝑄 ∗ ) be the optimal price and quantity. Then we have

20 = 𝑀𝐶 (𝑄 ∗ ) = 𝑀 𝑅(𝑄 ∗ ) = 100 − 2𝑄 ∗ ,

from which we solve 𝑄 ∗ = 40, and so 𝑝 ∗ = 𝑝(𝑄 ∗ ) = 100 − 𝑄 ∗ = 60.

(d) Notice that the marginal cost is constant while the marginal revenue is decreasing in the quantity
of sales. Since the marginal cost and marginal revenue are equal at 𝑄 ∗ = 40, it is clear that the
marginal revenue is higher than the marginal cost when 𝑄 < 40. As a result, given that the
output level is no higher than 30, it follows that it is optimal to produce 30, and so the optimal
price 𝑝 ∗ = 100 − 30 = 70.

3. (a) Let 𝑞 0 be the minimum efficient scale. Then 𝑞 0 equates the marginal cost and the average cost,
that is,
500
0.1𝑞 0 + 20 = 0.05𝑞 0 + 20 + ,
𝑞0
from which we solve 𝑞 0 = 100.
(b) The inverse demand function is

1120 − 𝑞
𝑝(𝑞) = = 56 − 0.05𝑞.
20

83
 Thus, the weekly total revenue is 𝑇 𝑅(𝑞) = 𝑞 · 𝑝(𝑞) = 𝑞(56 − 0.05𝑞) = 56𝑞 − 0.05𝑞 2 , and so the
marginal revenue
d𝑇 𝑅(𝑞)
𝑀 𝑅(𝑞) = = 56 − 0.1𝑞.
d𝑞
(c) The profit-maximizing level of output 𝑞 ∗ equates the marginal revenue and the marginal cost,
that is,
0.1𝑞 ∗ + 20 = 56 − 0.1𝑞 ∗ ,

from which we solve 𝑞 ∗ = 180. Plugging this to the inverse demand function, the optimal price

𝑝 ∗ = 56 − 0.05𝑞 ∗ = 47.

(d) By definition, the producer surplus

∫ 𝑞∗ ∫ 180 180
∗
𝑃𝑆 = [ 𝑝 − 𝑀𝐶 (𝑞)] d𝑞 = (27 − 0.1𝑞) d𝑞 = (27𝑞 − 0.05𝑞 2 ) = 3240.
0 0 0

The consumer surplus (where 𝑝(𝑞) is the inverse demand function)

∫ 𝑞∗ ∫ 180 180
𝐶𝑆 = [ 𝑝(𝑞) − 𝑝 ∗ ] d𝑞 = (9 − 0.05𝑞) d𝑞 = (9𝑞 − 0.025𝑞 2 ) = 810.
0 0 0

(e) When exercising the first-degree price discrimination, the price of the last unit of output 𝑝 0
equals the marginal cost of the last unit of output. Since consumers are required to pay the
highest price they are willing to pay, the consumer surplus clearly equals 0, and so producer
surplus equals the total surplus. Concretely, using the inverse demand function, we have (𝑞 0 is
the optimal output level)
56 − 0.05𝑞 0 = 0.1𝑞 0 + 20,

from which we solve 𝑞 0 = 240. Thus, the producer surplus in this case is
∫ 𝑞0 ∫ 240 240
𝑃𝑆 = [ 𝑝(𝑞) − 𝑀𝐶 (𝑞)] d𝑞 = (36 − 0.15𝑞) d𝑞 = (36𝑞 − 0.075𝑞 2 ) = 4320.
0 0 0

4. (a) Let ( 𝑝 ∗ , 𝑄 ∗ ) be the optimal price and quantity. The total revenue 𝑇 𝑅(𝑄) = 𝑝(𝑄)𝑄 = (120−𝑄)𝑄
and so the marginal revenue 𝑀𝐶 (𝑄) = 120 − 2𝑄. The marginal cost 𝑀𝐶 (𝑄) = 2𝑄. Thus,

120 − 2𝑄 ∗ = 𝑀 𝑅(𝑄 ∗ ) = 𝑀𝐶 (𝑄 ∗ ) = 2𝑄 ∗ ,

from which we solve 𝑄 ∗ = 30, and so 𝑝 ∗ = 𝑝(𝑄 ∗ ) = 120 − 𝑄 ∗ = 90.

(b) A lump-sum tax changes neither the marginal cost nor the marginal revenue. Thus, the optimal
level of output will unchanged provided that the monopolist can still obtain a positive profit
after the lump-sum tax. Using the result in (a), the monopolist can obtain 𝑝 ∗ 𝑄 ∗ − 𝐶 (𝑄 ∗ ) =

84
 90 × 30 − 302 = 1800 before the tax. Thus, the profit after tax will be 1800 − 2000 = −200,
which is worse than not producing. Therefore, the profit-maximizing level of output is 0 in this
case.
(c) Let 𝑄¯ be the output level associated with the efficient price ceiling. It is clear that 𝑄¯ must
equate the marginal cost and the marginal WTP, where the latter is given by the demand function.
Therefore, 120 − 𝑄¯ = 2𝑄, ¯ from which we solve 𝑄¯ = 40 and so 𝑝¯ = 120 − 𝑄¯ = 80. Notice that
the firm will optimally produce 𝑄¯ under 𝑝¯ since the marginal cost is increasing while it equals 𝑝¯
¯
at 𝑄.
(d) In this case the total revenue would be 40% off at each price level , so would the marginal revenue.
Thus, let ( 𝑝 0, 𝑄 0) be the optimal price and quantity in this case. Then 𝑄 0 is characterized by

0.6(120 − 2𝑄 0) = 𝑀 𝑅(𝑄 0) = 𝑀𝐶 (𝑄 0) = 2𝑄 0,

and so 𝑄 0 = 22.5, 𝑝 0 = 97.5. Clearly, this would not improve consumer surplus, as the
monopolist would produce less and charge an even higher price than before the tax.
√︁ √
5. (a) The total revenue at output level 𝑄 is 𝑇 𝑅(𝑄) = 𝑝(𝑄)𝑄 = ( 144/𝑄)𝑄 = 12 𝑄, and so the
√
marginal revenue 𝑀 𝑅(𝑄) = d𝑇 𝑅(𝑄)/d𝑄 = 6/ 𝑄. The total cost 𝐶 (𝑄) = 𝐴𝑉𝐶 (𝑄) · 𝑄 + 𝐹𝐶 =
√ √
𝑄 𝑄 + 5, and so the marginal cost 𝑀𝐶 (𝑄) = d𝐶 (𝑄)/d𝑄 = 1.5 𝑄. Let ( 𝑝 ∗ , 𝑄 ∗ ) be the optimal
price and quantity. Then the optimality condition 𝑀 𝑅(𝑄 ∗ ) = 𝑀𝐶 (𝑄 ∗ ) gives that

6 √︁
√ ∗ = 1.5 𝑄 ∗ ,
𝑄

and so 𝑄 ∗ = 4. Plugging this back to the demand function, we obtain

√︄ √︂
∗ 144 144
𝑝 = = = 6.
𝑄∗ 4

At ( 𝑝 ∗ , 𝑄 ∗ ), the monopolist obtains a profit

√
𝜋 = 𝑝 ∗ 𝑄 ∗ − 𝐶 (𝑄 ∗ ) = 6 × 4 − 4 4 − 5 = 11 > 0,

which indicates that ( 𝑝 ∗ , 𝑄 ∗ ) is indeed the optimum (since the monopolist will choose to shut
down if the profit is negative).
(b) Now we write the marginal cost and marginal revenue as functions√︁ about price. Since
√︁
𝑇 𝑅( 𝑝) = 𝑝𝑄( 𝑝) = 144/𝑝 and 𝐶 ( 𝑝) = 𝑄( 𝑝) 𝑄( 𝑝) + 5 = (144/𝑝 ) 144/𝑝 2 + 5 = 1728/𝑝 3 + 5,
2

we arrive at 𝑀 𝑅( 𝑝) = d𝑇 𝑅( 𝑝)/d𝑝 = −144/𝑝 2 and 𝑀𝐶 ( 𝑝) = d𝑇𝐶 ( 𝑝)/d𝑝 = −5184/𝑝 4 . Thus,

d𝜋 144  
= 𝑀 𝑅( 𝑝) − 𝑀𝐶 ( 𝑝) = 4 36 − 𝑝 2 .
d𝑝 𝑝

Since d𝜋/d𝑝 > 0 for 𝑝 ∈ [0, 6], it is clear that the monopolist should optimally charge 𝑝 = 4

85
 under the price ceiling. The corresponding level of output is 𝑄 = 144/42 = 9. In this case, the
profit √
𝜋 = 𝑝𝑄 − 𝐶 (𝑄) = 4 × 9 − 9 9 − 5 = 4.

(c) The socially optimal price ceiling 𝑝¯ will equate the marginal cost and the marginal WTP (i.e.,
𝑝¯ is chosen so that the sum of consumer and producer surplus is maximized). Let 𝑄¯ be the
corresponding output level. Then we have
√︄
144
√︃
= 1.5 𝑄,¯
𝑄¯
√︁
which yields that 𝑄¯ = 8, and so 𝑝¯ = 144/82 = 1.5.

6. (a) The price elasticities of demand for car in Brazil and India are, respectively,
  !
d𝑄 𝐵 𝑝 𝐵 𝐴 𝐴 𝑝𝐵
𝜀𝐵 = · =2 · − 2 · = −2
d𝑝 𝐵 𝑄 𝐵 𝑝𝐵 𝑝𝐵 ( 𝐴/𝑝 𝐵 ) 2
 4 !
d𝑄 𝐼 𝑝 𝐼 𝐴 𝐴 𝑝𝐼
𝜀𝐼 = · =5 · − 2 · = −5
d𝑝 𝐼 𝑄 𝐼 𝑝𝐼 𝑝𝐼 ( 𝐴/𝑝 𝐼 ) 5

(b) Since we have known the elasticities of demand in both countries, the simplest way to figure out
the optimal prices is to employ the formula (𝑝 ∗ is the optimal price)

𝑀𝐶
𝑝∗ = .
1 + (1/𝜀)

Thus, the optimal prices for Brazil and India are, respectively,

10000
𝑝 ∗𝐵 = = 20000
1 + 1/(−2)
10000
𝑝 ∗𝐼 = = 12500
1 + 1/(−5)

The mark-up in Brazil is higher. In particular, in Brazil, the mark-up equals

𝑝 ∗𝐵 − 𝑀𝐶 20000 − 10000 1
= = = 0.5,
𝑝∗ 20000 2

while in India, the mark-up equals

𝑝 ∗𝐼 − 𝑀𝐶 12500 − 10000 1
= = = 0.2.
𝑝 ∗𝐼 12500 5

86
 Indeed, we can deduce from the pricing formula that the mark-up equals
𝑀𝐶
𝑝 ∗ − 𝑀𝐶 1+(1/𝜀) − 𝑀𝐶 1
= =− .
𝑝∗ 𝑀𝐶
1+(1/𝜀)
𝜀

That is, when monopolized, markets with lower elasticity of demand (in the sense of absolute
value) will have higher mark-ups.
(c) The tax leads to higher marginal cost to the firm, which is now 𝑀𝐶 + 10. Plugging this back to
the pricing formula in (b), we obtain the optimal prices with tax:

10000 + 1000
𝑝 ∗𝐵 = = 22000;
1 + 1/(−2)
10000 + 1000
𝑝 ∗𝐼 = = 13750.
1 + 1/(−5)

(d) Due to the tax, the price will increase by 22000 − 20000 = 2000 in Brazil and by 13750 − 12500 =
1250. Thus, the price in Brazil will increase more. This is also a result of the pricing strategy,
where markets with lower elasticities have higher mark-ups, and hence more significant change
in prices as a result of changes in marginal cost. Since the Brazilian demand is less elastic, the
increase of price in Brazil is higher.

7. (a) When 𝑝 > 20, no consumer will buy H-Jeans and so the quantity of demand is 0. When
10 < 𝑝 6 20, there will be 𝑛 𝐻 = 200 consumers who will each purchase a pair of H-Jeans, and
so the quantity of demand in this case is 200. When 𝑝 6 10, all consumers on the market will be
happy to buy H-Jeans, which says that the quantity of demand now is 200 + 300 = 500. Thus,
the demand function is
0, 𝑝 > 20






𝑄 𝑑 ( 𝑝) = 200, 10 < 𝑝 6 20 .


 500, 𝑝 6 10


We plot the aggregate demand curve in Figure 10 below.
(b) Using the demand function derived in (a), we can derive the profit for each price level:

0, 𝑝 > 20






𝜋( 𝑝) = 200( 𝑝 − 5), 10 < 𝑝 6 20 .


 500( 𝑝 − 5),
 𝑝 6 10

Clearly, the monopolist will choose between 𝑝 = 20 and 𝑝 = 10. Note that 𝜋(20) = 200(20−5) =
3000 and 𝜋(10) = 500(10 − 5) = 2500. Since 𝜋(20) > 𝜋(10) > 0, it follows immediately that
the profit-maximizing price 𝑝 ∗ = 20 and the associated profit is 𝜋(20) = 3000.
(c) When price discrimination is possible, it is clear that the best the monopolist can do is to
charge the WTP for each group of consumers. Thus, it will set two prices 𝑝 𝐻 and 𝑝 𝐿 , where

87
 𝑝

20

10

0 200 500 𝑄

Figure 10: The Market Demand

𝑝 𝐻 = 𝑉 𝐻 = 20 and 𝑝 𝐿 = 𝑉 𝐿 = 10, and charge consumer of value 20 (10, respectively) the price
𝑝 𝐻 (𝑝 𝐿 , respectively). The profit level is

𝜋 = 200 × 20 + 300 × 10 = 7000.

The monopolist clearly benefits from price discrimination. No group of consumers benefit from
price discrimination, as the surplus for each consumer is zero, which is the same as the case
without price discrimination.

8. (a) Let ( 𝑝 ∗ , 𝑄 ∗ ) be the competitive equilibrium. Without regulation, firms only take into account
their private marginal when deciding the optimal level of output. Thus, the supply curve is
𝑄 = 𝑝 − 120. Thus, the competitive equilibrium is characterized by

𝑄 ∗ = 𝑝 ∗ − 120 = 720 − 𝑝 ∗ ,

and so ( 𝑝 ∗ , 𝑄 ∗ ) = (420, 300).

(b) Let ( 𝑝 𝑀 , 𝑄 𝑀 ) be the optimal price and output level for a monopolist on the market. Using the
inverse demand function, we can derive the total revenue 𝑇 𝑅(𝑄) = 720𝑄 − 𝑄 2 , based on which
we further derive the marginal revenue 𝑀 𝑅(𝑄) = 𝑇 𝑅 0 (𝑄) = 720 − 2𝑄. The output level of
output equates the marginal revenue and the marginal cost, that is,

120 + 𝑄 𝑀 = 720 − 2𝑄 𝑀 ,

from which we solve 𝑄 𝑀 = 200, and so 𝑝 𝑀 = 720 − 𝑄 𝑀 = 520.

(c) In the competitive equilibrium, the consumer surplus is (𝑝(𝑞) denotes the inverse demand
function)
∫ 𝑄∗ ∫ 300
∗
𝐶𝑆 = [ 𝑝(𝑞) − 𝑝 ] d𝑞 = (720 − 𝑞 − 420) d𝑞 = 45000;
0 0

88
 while the producer surplus is
∫ 𝑄∗ ∫ 300
𝑃𝑆 = [ 𝑝 ∗ − 𝑀𝐶 (𝑞)] d𝑞 = (420 − 120 − 𝑞) d𝑞 = 45000.
0 0

Due to the existence of externality, the external social cost is

∫ 𝑄∗ ∫ 300
𝐸 𝑆𝐶 = 𝑀 𝐸𝐶 (𝑞) d𝑞 = 2𝑞 d𝑞 = 90000.
0 0

Thus, the total surplus in the competitive equilibrium equals 𝐶𝑆 + 𝑃𝑆 − 𝐸 𝑆𝐶 = 0.

In the monopolistic equilibrium, the sum of consumer and producer surplus is
∫ 𝑄𝑀 ∫ 200
𝐶𝑆 + 𝑃𝑆 = [ 𝑝(𝑞) − 𝑀𝐶 (𝑞)] d𝑞 = [720 − 𝑞 − 120 − 𝑞] d𝑞 = 80000
0 0

while the external social cost is

∫ 𝑄𝑀 ∫ 200
𝐸 𝑆𝐶 = 𝑀 𝐸𝐶 (𝑞) d𝑞 = 2𝑞 d𝑞 = 40000,
0 0

and so the total surplus in the monopolistic equilibrium equals 𝐶𝑆 + 𝑃𝑆 − 𝐸 𝑆𝐶 = 40000.

Therefore, the social welfare is higher in the monopolistic equilibrium.

9. (a) Based on the information provided, the total revenue of market 𝐴 is 𝑇 𝑅(𝑞 𝐴) = 𝑝(𝑞 𝐴)𝑞 𝐴 =
200𝑞 𝐴 − 𝑞 2𝐴, and the total revenue of market 𝐵 is 𝑇 𝑅(𝑞 𝐵 ) = 𝑝(𝑞 𝐵 )𝑞 𝐵 = 300𝑞 𝐵 − 2𝑞 2𝐵 . Thus,
the marginal revenues are 𝑀 𝑅(𝑞 𝐴) = 200 − 2𝑞 𝐴 and 𝑀 𝑅(𝑞 𝐵 ) = 300 − 4𝑞 𝐵 . The marginal cost
𝑀𝐶 (𝑞) = 2𝑞. Let ( 𝑝 ∗𝐴, 𝑞 ∗𝐴) and ( 𝑝 ∗𝐵 , 𝑄 ∗𝐵 ) be the optimal price and quantity in market 𝐴 and
market 𝐵, respectively. In optimum, we must have 𝑀 𝑅(𝑞 𝐴) = 𝑀 𝑅(𝑞 𝐵 ) = 𝑀𝐶 (𝑞 𝐴 + 𝑞 𝐵 ), that
is,
200 − 2𝑞 𝐴 = 300 − 4𝑞 𝐵 = 2(𝑞 𝐴 + 𝑞 𝐵 ).

Solving this system, we get 𝑞 ∗𝐴 = 30, 𝑞 ∗𝐵 = 40.

(b) Plugging 𝑞 ∗𝐴 = 30 and 𝑞 ∗𝐵 = 40 to the inverse demand functions, we obtain 𝑝 ∗𝐴 = 200 − 𝑞 ∗𝐴 = 170
and 𝑝 ∗𝐵 = 300 − 2𝑞 ∗𝐵 = 220.

10. (a) Given 𝑝 𝑟 , the quantity of demand that is served at 𝑝 𝑟 is 4−0.2𝑝 𝑟 , and the quantity of demand from
people we are not willing to pay 𝑝 𝑟 at 𝑝 𝑑 < 𝑝 𝑟 is 𝑄 𝑑𝐴 = [4− (4−0.2𝑝 𝑟 )] −0.2𝑝 𝑑 = 0.2𝑝 𝑟 −0.2𝑝 𝑑 .
Thus, at ( 𝑝 𝑟 , 𝑝 𝑑 ), the total profit is

𝜋( 𝑝 𝑟 , 𝑝 𝑑 ) = 𝑝 𝑟 (4 − 0.2𝑝 𝑟 ) + 𝑝 𝑑 (0.2𝑝 𝑟 − 0.2𝑝 𝑑 ) − 5[(4 − 0.2𝑝 𝑟 ) + (0.2𝑝 𝑟 − 0.2𝑝 𝑑 )] − 15.

89
 Let ( 𝑝 𝑟∗ , 𝑝 ∗𝑑 ) be the profit-maximizing prices. The first-order condition for optimality is

𝜕𝜋
= 4 − 0.4𝑝 𝑟∗ + 0.2𝑝 ∗𝑑 = 0,
𝜕 𝑝𝑟 ( 𝑝𝑟∗ , 𝑝𝑑∗ )
𝜕𝜋
= 0.2𝑝 𝑟∗ − 0.4𝑝 ∗𝑑 + 1 = 0.
𝜕 𝑝𝑑 ( 𝑝𝑟∗ , 𝑝𝑑∗ )

Solving the system, we obtain 𝑝 𝑟∗ = 15 and 𝑝 ∗𝑑 = 10. Thus, the quantities sold at 𝑝 𝑟∗ is
𝑄 𝑟∗ = 4 − 0.2𝑝 𝑟∗ = 1 and that at 𝑝 ∗𝑑 is 𝑄 ∗𝑑 = 0.2( 𝑝 𝑟∗ − 𝑝 ∗𝑑 ) = 1. Therefore, the economic profit at
( 𝑝 𝑟∗ , 𝑝 ∗𝑑 ) is
𝜋 = 𝑝 𝑟∗ 𝑄 𝑟∗ + 𝑝 ∗𝑑 𝑄 ∗𝑑 − 5(𝑄 𝑟∗ + 𝑄 ∗𝑑 ) − 15 = 0,

which indicates that the ( 𝑝 𝑟∗ , 𝑝 ∗𝑑 ) are indeed the optimal prices to charge.
(b) The economic profit is 0, as we have computed in (a).
(c) At a uniform price level 𝑝, the profit equals

𝜋( 𝑝) = (4 − 0.2𝑝) ( 𝑝 − 5) − 15,

and optimality prescribes, 𝑝 ∗ being the optimal price, that

d𝜋( 𝑝)
= 4 − 0.2𝑝 ∗ − 0.2( 𝑝 ∗ − 5) = 0,
d𝑝 𝑝= 𝑝 ∗

from which we solve 𝑝 ∗ = 12.5, and the corresponding quantity of output 𝑞 ∗ = 4 − 0.2𝑝 ∗ = 1.5.
Notice that 𝜋( 𝑝 ∗ ) = (4 − 0.2 × 12.5) (12.5 − 5) − 15 = −3.75 < 0. Thus, ( 𝑝 ∗ , 𝑞 ∗ ) is worse than
not producing. That says, Harry would choose to shut down and earns a zero profit if he is forced
to charge the same price to all buyers.
(d) Buyers are clearly better off as a result of Harry’s being able to conduct price discrimination, as
the consumer surplus under price discrimination is positive (as there are consumers who buy at
a price lower than their WTP) while that without price discrimination is 0, since in the latter
case Harry will be forced out of business.

11. (a) It is easy to see that the maximum price (i.e., the price at which the quantity of demand equals 0)
for students is 50 and that for non-students is 40. So the market demand function

100 − 2𝑝, if 40 6 𝑝 < 50







𝑄𝑀 = 220 − 5𝑝, if 0 6 𝑝 < 40


 0,
 otherwise


(b) If the monopolist charges a price 𝑝 > 40 (i.e., it abandons the market of non-students), its profit
𝜋( 𝑝) = 𝑝(100 − 2𝑝) − 10(100 − 2𝑝) = −2𝑝 2 + 120𝑝 − 1000. Since d𝜋/d𝑝 = −4𝑝 + 120, which
is negative when 𝑝 > 40, and so the maximum profit it can obtain by charging a price 𝑝 > 40

90
 is 𝜋(40) = 600 (the corresponding quantity of output is 𝑄 𝑀 = 100 − 2 × 40 = 20, within the
capacity constraint).
If, however, the monopolist charges a price 𝑝 < 40, then the inverse market demand function
is 𝑝 = 44 − 0.2𝑄 𝑀 , and so the marginal revenue 𝑀 𝑅(𝑄 𝑀 ) = 44 − 0.4𝑄 𝑀 . If we ignore the
capacity constraint, the optimal quantity of output 𝑄 ∗𝑀 would solve 𝑀 𝑅(𝑄 ∗𝑀 ) = 𝑀𝐶 (𝑄 ∗𝑀 ), that
is, 44 − 0.4𝑄 ∗𝑀 = 10, and so 𝑄 ∗𝑀 = 85, which violates the capacity constraint. As a result, the
capacity constraint must be binding at optimum. Thus, the optimal quantity of output 𝑄 ∗𝑀 = 80,
which implies that the optimal price 𝑝 ∗ = 44 − 0.2𝑄 ∗𝑀 = 28. The quantity sold to students
𝑄 ∗𝑆 = 100 − 2𝑝 ∗ = 44, and that to non-students 𝑄 ∗𝑁 = 120 − 3𝑝 ∗ = 36.
(c) The inverse demand function for students and non-students are 𝑝 = 50−0.5𝑄 𝑠 and 𝑝 = 40−𝑄 𝑁 /3,
respectively, based on which we find the marginal revenue from students and non-students are,
respectively,

𝑀 𝑅𝑆 (𝑄 𝑆 ) = 50 − 𝑄 𝑆
2𝑄 𝑁
𝑀 𝑅 𝑁 (𝑄 𝑁 ) = 40 −
3
Ignoring the capacity constraint, the optimality condition prescribes that

∗
 50 − 𝑄 𝑆 = 𝑀𝐶 = 10




2𝑄 ∗𝑁
 40 − = 𝑀𝐶 = 10


 3
which yields that 𝑄 ∗𝑆 = 40 and 𝑄 ∗𝑁 = 45. Since 𝑄 ∗𝑆 + 𝑄 ∗𝑁 = 85 > 80, the capacity constraint
must be binding, and so the optimal quantities are instead characterized by the system
 2𝑄 ∗𝑁
 50 − 𝑄 ∗𝑆 = 40 −



3
 ∗ ∗
 𝑄 𝑆 + 𝑄 𝑁 = 80


from which we solve 𝑄 ∗𝑆 = 38, 𝑄 ∗𝑁 = 42. The corresponding optimal prices are 𝑝 ∗𝑆 =
50 − 0.5𝑄 ∗𝑆 = 31 and 𝑝 ∗𝑁 = 40 − 𝑄 ∗𝑁 /3 = 26, and the firm’s profit

𝜋 = 𝑝 ∗𝑆 𝑄 ∗𝑆 + 𝑝 ∗𝑁 𝑄 ∗𝑁 − 𝑇𝐶 (𝑄 ∗𝑆 + 𝑄 ∗𝑁 ) = 1470.

12. (a) A (straight) implementation of first degree price discrimination refers to charging the highest
price an consumer is willing to pay for each unit of the good, and so the optimal quantity of
good must equate the marginal cost and a consumer’s willingness-to-pay. For each consumer,
we know that the willingness-to-pay is given by the inverse demand function 𝑝 = 90 − 0.5𝑞.
So the optimal quantity that should be sold to a single consumer, denoted by 𝑞 ∗ , must solve
90 − 0.5𝑞 ∗ = 50, and so 𝑞 ∗ = 80. The total quantity of good 𝑌 that are sold on the market is

91
 hence 20𝑞 ∗ = 1600.
The price of the 8-th unit of good 𝑌 to each consumer equals a consumer’s willingness-to-pay
(i.e., the inverse demand function) for the 8th unit of good, and so 𝑝 8th = 90 − 0.5 × 8 = 86.
(b) We visualize the pricing strategy (which is basically an exercise of second-degree price
discrimination) in Figure 11 (in red), and then it is easy to see that one’s consumer surplus is
maximized when 𝑞 ∗ = 20. That is, each consumer will buy 20 units of good 𝑌 . As a result, the
monopolist’s economic profit equals

(80 × 20 − 50 × 20) × 20 − 2000 = 10000.

80
Pricing strategy
60

𝑝 = 90 − 0.5𝑞

0 20 60 𝑞

Figure 11: Visualization of the Pricing Strategy

13. (a) As we know, when using a two-part tariff to realize first degree price discrimination, it is optimal
for a monopolist to charge a per unit price at the marginal cost and set the lump sum fee at the
total consumer surplus at the per unit price. Thus, the monopolist should set the per unit price
𝑝 ∗ = 𝑀𝐶 = 4 for each consumer, and for consumer 2, the lump sum fee equals the gray area in
Figure 7, i.e.,
1
× (16 − 4) × 24 = 144.
2

(b) Under the pricing scheme, if consumer 1 ever accepts the plan, it would be optimal for him to
buy 𝑞 ∗1 = 17 − 4 = 13, and the consumer surplus, before paying out the lump sum fee, is

1
(17 − 4) × 13 = 84.5.
2
So after paying out the lump sum fee of 130, consumer 1 would end up with a surplus
84.5 − 130 = −45.5, which is worse than buying nothing. Thus, consumer 1 will optimally
choose to buy 0.

92
 𝑝

16

4 𝑀𝐶 = 4
𝑞 2 = 32 − 2𝑝
0 24 𝑞2

Figure 12: Optimal Two-part Tariff

For consumer 2, as we have computed in (a), under the per unit price 4, it is optimal to buy
32 − 2 × 4 = 24, by which he enjoys a surplus of 144 before paying out the lump sum fee. Thus,
after paying out the lump sum fee of 130, consumer 2’s surplus will be 144 − 130 = 14, which is
positive and so consumer 2 will buy 24 units of good 𝑍 and obtain a surplus of 14.

14. Let 𝑄 ∗1 and 𝑄 ∗2 be the optimal quantities sold to group 1 and group 2, respectively, and 𝑝 ∗1 , 𝑝 ∗2 be the
corresponding optimal prices. The total revenues from the two groups are 𝑇 𝑅(𝑄 1 ) = 𝑝(𝑄 1 )𝑄 1 =
190𝑄 1 − 𝑄 21 and 𝑇 𝑅(𝑄 2 ) = 𝑝(𝑄 2 )𝑄 2 = 46𝑄 2 − 0.5𝑄 22 . Thus, the corresponding marginal revenues
are (
𝑀 𝑅1 (𝑄 1 ) = 190 − 2𝑄 1
𝑀 𝑅2 (𝑄 2 ) = 46 − 𝑄 2
The optimality condition for third-degree price discrimination prescribes that

𝑀𝐶 (𝑄 ∗1 + 𝑄 ∗2 ) = 𝑀 𝑅1 (𝑄 ∗1 ) = 𝑀 𝑅2 (𝑄 ∗2 ),

that is, (
6 = 190 − 2𝑄 ∗1
6 = 46 − 𝑄 ∗2
from which we solve 𝑄 ∗1 = 92, 𝑄 ∗2 = 40, and so 𝑝 ∗1 = 190 − 𝑄 ∗1 = 98, 𝑝 ∗2 = 45 − 0.5𝑄 ∗2 = 25.

15. (a) Based on the information provided, we can derive the market demand curve

84 − 0.4𝑝, if 580/3 6 𝑝 < 210







𝑄𝑀 = 𝑄𝑃 + 𝑄𝐻 = 200 − 𝑝, if 0 6 𝑝 < 580/3


 0,
 otherwise

If the company charges a price 𝑝, then it effectively prices the public hospitals out, and
the corresponding economic profit 𝜋( 𝑝) = 𝑝(84 − 0.4𝑝) − 20(84 − 0.4𝑝) − (84 − 0.4𝑝) 2 =

93
 (84 − 0.4𝑝) (1.4𝑝 − 104). Thus, d𝜋/d𝑝 = 159.2 − 1.12𝑝, which is negative when 𝑝 > 580/3,
and so in this case the maximum profit equals 𝜋(580/3), which indicates that the optimal price
is no higher than 580/3, in which case the optimal quantity 𝑄 ∗ equates the marginal revenue and
the marginal cost, that is,
20 + 2𝑄 ∗ = 200 − 2𝑄 ∗ ,

from which we solve 𝑄 ∗ = 45, and so the optimal price 𝑝 ∗ = 200 − 𝑄 ∗ = 155.
(b) Plugging the optimal price 𝑝 ∗ = 155 to the demand curves, we find
(
𝑄 ∗𝑃 = 84 − 0.4𝑝 ∗ = 22
𝑄 ∗𝐻 = 116 − 0.6𝑝 ∗ = 23

(c) Let (𝑄 ∗𝑃 , 𝑄 ∗𝐻 , 𝑝 ∗𝑃 , 𝑝 ∗𝐻 ) be the optimal plan. It is straightforward to derive that the marginal
revenue from pharmacies is 𝑀 𝑅 𝑃 (𝑄 𝑃 ) = 210 − 5𝑄 𝑃 and that from public hospitals is
𝑀 𝑅 𝐻 (𝑄 𝐻 ) = 580/3 − 10𝑄 𝐻 /3. Then the optimality condition prescribes that

𝑀𝐶 (𝑄 ∗𝑃 + 𝑄 ∗𝐻 ) = 𝑀 𝑅 𝑃 (𝑄 ∗𝑃 ) = 𝑀 𝑅 𝐻 (𝑄 ∗𝐻 ),

that is (note that production is cumulative)

∗ ∗ ∗
 20 + 2(𝑄 𝑃 + 𝑄 𝐻 ) = 210 − 5𝑄 𝑃




∗ ∗ 580 10𝑄 ∗𝐻
 20 + 2(𝑄 𝑃 + 𝑄 𝐻 ) = −


 3 3
from which one solves 𝑄 ∗𝑃 = 20, 𝑄 ∗𝐻 = 25, 𝑝 ∗𝑃 = 210 − 2.5𝑄 ∗𝑃 = 160, and 𝑝 ∗𝐻 = 580/3 −
10𝑄 ∗𝐻 /3 = 110.
(d) The biotech company clearly gains since it decides to exercise price discrimination although
it does not have to do it, which must be because doing so is profitable. The public hospitals
also gain as now they can buy more treatment at a lower price. The only side who loses is the
pharmacies since now they are faced with higher price and consume less.

94
 5

Choice under Uncertainty

5 Choice under Uncertainty

95
Exercise 1. Your current wealth level is 𝑥 0 = 49 and you are forced to make the following wager: If a
fair coin comes up heads, you get 15; you lose 13 if it comes up tails. Your Bernoulli utility function
√
is 𝑢(𝑥) = 𝑥, where 𝑥 represents the final wealth level. Round your answers up to 2 decimal places if
applicable.

(a) Find the expected value of this gamble.

(b) Find the expected utility of this gamble.
(c) How would your answers to (a) and (b) change if the payoff for tails fell to a loss of 15?
(d) What is the most you would pay to get out of the gamble described in (c)?

Exercise 2. Refer to Figure 13 below, where 𝑢 is the Bernoulli utility function of an agent (an expected
utility maximizer) and 𝑤 represents his wealth. Assume that the agent’s initial wealth is 𝑤 0 > 0.

Utility

𝑢(𝑤)

𝑐
𝑏
𝑎

𝑤
0 1 2 4 7 10

Figure 13: An agent’s Bernoulli utility function

(a) Consider two lotteries. Lottery 𝐴: one wins $10 with probability 0.5 and 0 with probability 0.5;
Lottery 𝐵: one wins $5 for sure. Which lottery will the agent choose (if he has to choose one from
the two)? Explain.

(b) Consider the following two lotteries:

Lottery 1 Lottery 2
Prize −$4 −$1 $5 Prize −$5 −$1 $2
Prob. 1/3 1/3 1/3 Prob. 1/4 1/4 1/2

Assume 𝑤 0 = $5. Write the expected utility of each of the two lotteries as an expression of 𝑎, 𝑏, and
𝑐. When is lottery 2 better than lottery 1 to the agent?

96
(c) If 𝑤 0 = $4, and the agent is faced with the following uncertainty: there is a 50% chance that he loses
all, what is his expected utility? What is the certainty equivalence of this uncertainty to the agent?

(d) The agent is still faced with the uncertainty described in (c) and 𝑤 0 = $4, but there is a property
insurance that pays back all the losses (e.g., it pays the agent $4 if he happens to lose $4). What is
the highest price the agent is willing to pay for the insurance? Explain.

Exercise 3. Kate has a Bernoulli utility function given by

1
𝑢(𝑤) = 1 − ,
𝑤
where 𝑤 is the present value of her lifetime income. If Kate becomes a teacher, she will make 𝑤 = 5 for
sure. If she becomes an actress, she will make 𝑤 = 400 if she becomes a star, but only 𝑤 = 2 if she fails to
become a star. The probability of becoming a star is 0.01.

(a) What is the expected income of becoming a teacher? What is the expected income of becoming an
actress?
(b) What is the expected utility of becoming a teacher for Kate? What is the expected utility of becoming
an actress for Kate? Round your answers up to 2 decimals if necessary.

Jake is an infallible judge of acting talent. After a brief interview, he can state with certainty whether Kate
will become a star if she chooses to pursue a career in acting.

(c) Find the most Kate would be willing to pay for the information (from Jake), assuming that Kate has
to pay before the interview.

Exercise 4. You are endowed with the following gamble. A fair coin (i.e., it shows tail (𝑇) with probability
1/2 and head (𝐻) with probability 1/2) is flipped twice and the following payoffs are assigned to each of
the four possible outcomes:

Outcome 𝐻, 𝐻 𝐻, 𝑇 𝑇, 𝐻 𝑇, 𝑇
Prize +10 +5 −7 −16

where, for example, “𝐻, 𝑇” means that it shows 𝐻 in the first flipping and 𝑇 in the second.

(a) Find the probability for each of the four outcomes.

(b) What is the expected prize of this gamble?
(c) Suppose your initial wealth level is 20 and your Bernoulli utility function is 𝑢(𝑤) = 𝑤 2 , where 𝑤
represents the (terminal) wealth level. What is your expected utility from the gamble? Will you be
better off if the gamble is replaced with the following one?

Prize −10 +10

Probability 1/2 1/2

97
 (d) What is the maximum amount for which you would be willing to pay to avoid the (original) gamble
if you are risk neutral?

Exercise 5. Assume that the Bernoulli utility function of Eric is given by

√
𝑢(𝑥) = 𝑥,

where 𝑥 is the wealth level. Initially, Eric’s wealth level is 𝑥 0 = 6.

(a) According to the Bernoulli utility function, what is Eric’s risk preferences (i.e., risk averse/loving/neutral,
or none of the three)?
(b) Consider the following two lotteries (note that Eric’s eventual wealth level equals his initial wealth
level 𝑥 0 plus the realization of the prize):

Lottery 1 Lottery 2

Prize −1 0 1 Prize −1 0 1
Probability 1/3 1/3 1/3 Probability 0.1 0.8 0.1

Which one is better for Eric? Justify your answer.

(c) Consider the following lottery:

Prize −5 3 19
Probability 0.4 0.3 0.3

If Eric is faced with an uncertainty represented by this lottery (i.e., Eric’s eventual wealth level
will be 6 − 5 = 1 with probability 0.4, etc.), what is Eric’s expected utility? What is the certainty
equivalent of this lottery to Eric? What is the risk premium? Round your answers up to 2 decimal
places.

Exercise 6. Suppose you have $10, 000 (your initial wealth) to invest. A broker gives you some information
about certain junk bonds. If the company issuing the bonds posts a profit this year, it will pay you a
40% interest rate on the bond. If the company files for bankruptcy, you will lose all you invested. If the
company breaks even, you will earn a 10% interest rate. Your broker tells you that there is a 50% chance
that they will break even and a 20% chance that the company will file for bankruptcy. Your other option is
to invest in a risk-free government bond that will guarantee 8% interest for one year.

(a) What is the expected interest rate for the junk bond investment?
(b) Which investment will you choose if your Bernoulli utility function is given by 𝑢(𝑤) = 𝑤 2 , where 𝑤
represents the final wealth level?
√
(c) Which investment will you choose if your Bernoulli utility function is given by 𝑢(𝑤) = 𝑤, where
𝑤 represents the final wealth level?

98
Exercise 7. We will consider a generic risk-averse agent, with the Bernoulli utility function being
𝑢(𝑥) = 𝑥 0.5 , where 𝑥 is the wealth of the agent. The agent’s initial wealth is $10, 000. Suppose that the
agent owns a lottery ticket with a 50/50 chance of winning $2, 000.

(a) Would this agent sell the lottery ticket for $1, 000? Carefully justify your answer.
(b) What is the minimum price at which the agent would be willing to sell the lottery ticket?
(c) Sketch the scenario on axes, with 𝑢(𝑥) on the vertical axis and wealth on the horizontal axis. You
should clearly and correctly label the wealth levels and utility levels.

Now suppose instead of a gamble the agent is stuck with a bad investment, with a 25% chance of a $2, 000
loss (otherwise no loss). The initial wealth level is still $10, 000.

(d) Sketch on separate axes this scenario, with 𝑢(𝑥) on the vertical axis and wealth on the horizontal
axis. Again, you should clearly and correctly label the wealth levels and utility levels.
(e) Would this agent be willing to pay someone $500 to take over the investment? Carefully justify your
answer.

Exercise 8. Consider a lottery with three possible outcomes: $80 will be received with probability 0.1,
$50 with probability 0.2, and $10 with probability 0.7.

(a) What is the expected value of the lottery?

(b) What is the variance of the outcome of the lottery?
(c) What would a risk-neutral person pay to play the lottery?

Exercise 9. Consider five different lotteries, numbered 1 through 5. For each 𝑛 = 1, 2, . . . , 5, lottery 𝑛 is
given by

Prize $(5 − 5𝑛) $(5 + 5𝑛)

Probability 0.5 0.5

For example, lottery 1 yields a zero prize with probability 0.5 and a prize of $10 with probability 0.5.

(a) What is the expected prize of lottery 𝑛?

(b) Argue that a risk-neutral agent must be indifferent with all of the five lotteries.
(c) Recall that a twice differentiable function is concave if and only if its first-order derivative is
decreasing. Suppose we have a risk-averse agent whose Bernoulli utility function is 𝑢(·) and whose
initial wealth is zero, which is twice differentiable. How will this agent rank the five lotteries, from
worst to best? Carefully justify your answer.
(d) Rank the variances of the five lotteries, from the highest to the lowest.

99
 (e) You might think that you have learned something from (c) and (d). But you should be careful in
reaching a conclusion about the relationship between the ranking of variance and the risk-preferences
of a risk-averse agent. Consider the following lottery, called lottery 6:

Prize $3 $35
Probability 15/16 1/16

First show that lottery 6 yields the same expected prize as lottery 1, and then compute the variance of
lottery 6 and compare it with that of lottery 1. Finally, determine which lottery, 1 or 6, is preferred
√
by an agent (with zero initial wealth) whose Bernoulli utility function is 𝑢(𝑥) = 𝑥. What do you
learn from this?

Exercise 10. Consider a risk-averse investor with (initial) wealth 𝑤 0 , who must allocate her wealth between
two assets. The safe asset returns 𝑅 for every unit invested while the risky asset returns 𝐻 > 𝑅 with
probability 𝑝 and 𝐿 < 𝑅 with probability 1 − 𝑝, where 0 < 𝑝 < 1. Both 𝐻 and 𝑅 are gross returns, that is,
the sum of principal and net return. Suppose her Bernoulli utility function is 𝑢(·), which is differentiable
and strictly increasing.

(a) What is the expected value of her wealth if she invests all her initial wealth in the safe asset?
(b) What is the expected value of her wealth if she invests all her initial wealth in the risky asset?
(c) Suppose first that 𝑝𝐻 + (1 − 𝑝)𝐿 < 𝑅. How much wealth should she invest in the risky asset? Justify
your answer.
(d) Now suppose that 𝑝𝐻 + (1 − 𝑝)𝐿 > 𝑅 and 𝑢(𝑥) = ln 𝑥. What is the fraction of her initial wealth that
should be optimally allocated to the risky asset?

Exercise 11. Given a choice between 𝐴 (a sure win of $100) and 𝐵 (an 80 percent chance to win $150
and a 20 percent chance to win 0), Sarah picks 𝐴. When when she is given a choice between 𝐶 (a 50
percent chance to win $100 and a 50 percent chance to win 0) and 𝐷 (a 40 percent chance to win $150 and
a 60 percent chance to win 0), she picks 𝐷. Show that Sarah’s choices are not consistent with expected
utility maximization, provided that her initial wealth level is 0.
Exercise 12. Frank, who was injured by a defective product, is weighing the decision whether to sue the
manufacturer. His Bernoulli utility function is given by

1
𝑢(𝑤) = 1 − ,
𝑤
where 𝑤 is his total wealth. His total wealth is 7 if he does not sue. If, however, he sues, he would win
with probability 0.5, in which case he would receive a damage award of 5. If he loses, he will receive
nothing in damages. Suppose the opportunity cost of the time required for a lawyer to file a suit on Frank’s
behalf is 2.

(a) Will Frank file a suit if he has to pay the lawyer a fee of 2? Justify your answer.

100
 (b) Can a risk-neutral lawyer offer Frank a fee schedule whose expected value is sufficient to cover
the lawyer’s opportunity cost (i.e., whose expected value is at least 2 to the lawyer) and that
simultaneously will induce Frank to sue? If so, describe such a fee schedule and justify your answer.

***********

Exercise 13. There is an enticing investment that pays

• 300 with probability 0.4;

• 250 with probability 0.2;
• 200 with probability 0.4.

(a) Compute the expected value of the investment.

(b) Khan has preferences over risk characterized by 𝑢(𝑥) = 𝑥 0.5 . Calculate Khan’s certainty equivalent
for this investment. Round your answer up to 2 decimal places.

Exercise 14. Jim’s attitude towards risks can be described as following: For each lottery, he only cares
about the worst possible outcome, and he prefers more money than less. His initial wealth is assumed to
be 0.

(a) Suppose for this question that Jim is an expected utility maximizer (although he is apparently not)
whose Bernoulli utility function is 𝑢(𝑥), 𝑥 here being the monetary prize he finally receives from a
lottery. What is the implication of fact that “he prefers more money than less” on 𝑢(𝑥)?
(b) Consider the following three lotteries. Which one is the best for Jim? Why?

Lottery 1 Lottery 2 Lottery 3

Prize Probability Prize Probability Prize Probability

$1 0.2 0 0.35 0 0.35
$50 0.75 $50 0.6 $40 0.6
$100 0.05 $100 0.05 $100 0.05

(c) Jim is not an expected utility maximizer. Argue this formally. Hint. Show that Jim would have a
“weird” Bernoulli utility function were he an expected utility maximizer.
(d) Jim’s buddy Jack, however, is an expected utility maximizer who prefers more money than less.
Which of the three lotteries above is the best for Jack? Justify your answer carefully.

Exercise 15. (Allais Paradox) Wendy employed an experiment to investigate people’s choices under
uncertainty. The experiment has two stages, each of which consists of a choice between two gambles. The
prizes for each gamble in each stage are as follows:

101
 Stage 1

Gamble 1A Gamble 1B

Prize $1 million Prize $1 million 0 $5 million

Probability 1 Probability 0.89 0.01 0.1

Stage 2

Gamble 2A Gamble 2B

Prize $1 million 0 Prize 0 $5 million

Probability 0.11 0.89 Probability 0.9 0.1

In each stage, a participant is asked to choose from the two gambles the one he/she prefers. Assume that
each participant has zero initial wealth.

(a) The first participant, Andrea, has chosen gameble 1B in stage 1 and gameble 2B in stage 2. If we
know that Andrea is either risk averse or risk loving, can we conclude, based on the data, that she is
risk loving? Justify your answer carefully.

The second participant, Bob, has chosen gamble 1A in stage 1 and gamble 2B in stage 2.

(b) If Bob is an expected utility maximizer whose Bernoulli utility function is 𝑢(𝑤), write down the
two inequalities about 𝑢(0), 𝑢(1 million), and 𝑢(5 million) that must be satisfied, in order to be
consistent with his choices.
(c) Based on the two inequalities, show that Bob is NOT an expected utility maximizer. Hint. Merge
like terms in your inequalities if necessary.
(d) According to the result of the experiment, most participants actually made the same choices as Bob
did. What would you choose were you a participant of the experiment? Do you think this is a serious
challenge to expected utility theory? Why?

102
Chapter 5 Answer Key

1. (a) The expected value of this game

𝐸 [𝑥] = 0.5(49 + 15) + 0.5(49 − 13) = 50.

(b) The expected utility of this game, given the Bernoulli utility function, is
√ √
𝐸 [𝑢] = 0.5 49 + 15 + 0.5 49 − 13 = 7.

(c) Now the expected value

𝐸 0 [𝑥] = 0.5(49 + 15) + 0.5(49 − 15) = 49,

and the expected utility

√ √
𝐸 0 [𝑢] = 0.5 49 + 15 + 0.5 49 − 15 ≈ 6.92.

(d) Let 𝑝 be the price you pay to get out of the gamble in (c). After paying the price, your expected
√︁
utility will be 49 − 𝑝. Thus, for you to be willing to pay the price, we must have
√︁
49 − 𝑝 > 𝐸 0 [𝑢] = 6.92,

from which we can solve 𝑝 6 1.11. Thus, you will be happy to pay at most 1.11 to get out of the
gamble.

2. (a) Note that lottery 𝐴 and lottery 𝐵 have the same expected value (both equal 5 + 𝑤 0 ). Since the
agent is risk averse (because the Bernoulli utility function is concave, as can be seen from the
figure), he will prefer 𝐵 over 𝐴 since 𝐴 is risky while 𝐵 is risk-free.
(b) The expected utility of Lottery 1 equals

1 1 1
𝐸 1 [𝑢] = 𝑢(5 − 4) + 𝑢(5 − 1) + 𝑢(5 + 5)
3 3 3
𝑢(1) + 𝑢(4) + 𝑢(10)
=
3
0.5𝑢(4) + 𝑢(4) + 𝑢(10)
=
3
𝑎 𝑐
= + .
2 3

103
 Similarly, the expected utility of Lottery 2 equals

1 1 1
𝐸 2 [𝑢] = 𝑢(5 − 5) + 𝑢(5 − 1) + 𝑢(5 + 2)
4 4 2
𝑎 𝑏
= + .
4 2
For Lottery 2 to be better than Lottery 1 to the agent, we must have 𝐸 2 [𝑢] > 𝐸 1 [𝑢], that is,
𝑎/4 + 𝑏/2 > 𝑎/2 + 𝑐/3, which can be simplified to 6𝑏 − 4𝑐 > 3𝑎. That is, when 6𝑏 − 4𝑐 > 3𝑎,
Lottery 2 will be better than Lottery 1 to the agent.
(c) The expected utility
𝐸 [𝑢] = 0.5𝑢(4) + 0.5𝑢(4 − 4) = 0.5𝑎.

As we can see from the figure, 𝐸 [𝑢] = 0.5𝑎 = 𝑢(1), which implies that the certainty equivalence
equals 1.
(d) Let 𝑝 be the price of the insurance. Then the agent’s expected utility after buying the insurance
is 𝑢(4 − 𝑝). Thus, for buying insurance to be profitable, we must have 𝑢(4 − 𝑝) > 𝐸 [𝑢]. Since
𝐸 [𝑢] = 𝑢(1) and 𝑢(𝑤) is increasing, the inequality is equivalent to 4 − 𝑝 > 1, and so 𝑝 6 3.
Hence, the highest price the agent is willing to pay for the insurance is 3.

3. (a) The expected income of becoming a teacher is 5, and the expected income of becoming an
actress is 0.01 × 400 + (1 − 0.01) × 2 = 5.98.
(b) The expected utility of becoming a teacher is 𝑢(5) = 1 − 1/5 = 0.8, and that of becoming an
actress is
   
1 1
0.01𝑢(400) + (1 − 0.01)𝑢(2) = 0.01 1 − + 0.99 1 − ≈ 0.50.
400 2

(c) Let 𝑝 be the price Kate pays for the information from Jake. Without the interview, we know that
Kate will find it better to become a teacher, and the highest expected utility without the interview
is 0.8. If she has paid for the interview, her expected utility will be

0.01𝑢(400 − 𝑝) + (1 − 0.01)𝑢(5 − 𝑝)

since she will choose to become an actress if Jake tells her that she can become a super star and
will choose to become a teacher otherwise. For having the interview to be profitable, we must
have
0.01𝑢(400 − 𝑝) + (1 − 0.01)𝑢(5 − 𝑝) > 0.8.

Applying the functional form of 𝑢(·) and solving the inequality, we get 𝑝 6 0.05. That is, Kate
will be willing to pay at most 0.05 for the information.

4. (a) With a fair coin, each of the four outcomes will be realized with probability (1/2) 2 = 1/4 = 0.25.

104
 (b) The expected prize of the game

1
𝐸 [𝑃] = (10 + 5 − 7 − 16) = −2.
4
(c) The expected utility from the gamble equals

1
[𝑢(20 + 10) + 𝑢(20 + 5) + 𝑢(20 − 7) + 𝑢(20 − 16)] = 427.5.
4
The expected utility from the alternative gamble is

1
[𝑢(20 − 10) + 𝑢(20 + 10)] = 500.
2
Clearly, you will be better off with the alternative gamble.
(d) Let 𝑝 be the price you pay to avoid the gamble and 𝑤 be your initial wealth level. Then for
avoiding the gamble (at a cost 𝑝) to be better, given that you are risk-neutral, we must have

𝑤 − 𝑝 > 𝑤 + 𝐸 [𝑃].

Since 𝐸 [𝑃] = −2, we can solve that 𝑝 6 2.

5. (a) Since the Bernoulli utility function of Eric is concave, we know by definition that he is risk
averse.
(b) We need to compare the expected utility Eric derives from each of the two lotteries. From lottery
1, he obtains an expected utility

1√ 1√ 1√
𝐸 1 (𝑢) = 6−1+ 6+0+ 6 + 1 ≈ 2.44
3 3 3
and his expected utility of lottery 2 is
√ √ √
𝐸 2 (𝑢) = 0.1 6 − 1 + 0.8 6 + 0 + 0.1 6 + 1 ≈ 2.45.

Since 𝐸 2 (𝑢) > 𝐸 1 (𝑢), lottery 2 is better for Eric than lottery 1.
(c) Eric’s expected utility
√ √ √
𝐸 (𝑢) = 0.4 6 − 5 + 0.3 6 + 3 + 0.3 6 + 19 = 2.8.

The certainty equivalent to this uncertainty, denoted by 𝐶𝐸, is the (certain) level of wealth that
yields to Eric the same level of expected utility as the lottery does.Thus we have
√
𝑢(𝐶𝐸) = 𝐶𝐸 = 𝐸 (𝑢) = 2.8,

from which we solve 𝐶𝐸 = 2.82 = 7.84. The corresponding risk premium is the difference

105
 between the expected wealth and certainty equivalent. Since the expected wealth from the lottery
is
0.4(6 − 5) + 0.3(6 + 3) + 0.3(6 + 19) = 10.6,

the risk premium equals 10.6 − 7.84 = 2.76.

6. (a) Let 𝑟 be the interest rate. According to the description, the distribution law of 𝑟 is

𝑟 40% 10% −100%

Probability 0.3 0.5 0.2
Notice that “lose all you invested” amounts to an interest rate of −100% as your final wealth
equals (1 + 𝑟)𝑤 0 , 𝑤 0 being the initial wealth. Therefore, the expected interest rate

𝐸 [𝑟] = 0.3 × 40% + 0.5 × 10% − 0.2 × 100% = −3%.

(b) The expected utility of choosing the risk-free bond is

𝐸 𝑅𝐹 [𝑢] = [10000(1 + 8%)] 2 = 116, 640, 000.

The expected utility of the junk bonds is

𝐸 𝐽 𝐵 [𝑢] = 0.3[10000(1+40%)] 2 +0.5[10000(1+10%)] 2 +0.2[10000(1−100%)] 2 = 119, 300, 000.

Clearly, 𝐸 𝐽 𝐵 [𝑢] > 𝐸 𝑅𝐹 [𝑢]. So the junk bonds will be chosen.

(c) Now you are risk-averse. Notice that the junk bonds yield a lower expected wealth than the
risk-free bonds according to (a), while the former are riskier than the latter. Thus, it follows
immediately that the risk-free bonds will be chosen. Alternatively, you may reach the same
conclusion via computing the expected utility for each of the two financial assets. The expected
utility of the risk-free bond is
√︁
𝐸 𝑅𝐹 [𝑢] = 10000(1 + 8%) ≈ 103.92.

The expected utility of the junk bonds is

√︁ √︁ √︁
𝐸 𝐽 𝐵 [𝑢] = 0.3 10000(1 + 40%) + 0.5 10000(1 + 10%) + 0.2 10000(1 − 100%) ≈ 87.93.

Clearly, 𝐸 𝐽 𝐵 [𝑢] < 𝐸 𝑅𝐹 [𝑢]. So the risk-free bond will be chosen.

7. (a) It is easy to see that both options yield the same expected wealth. Since the agent is risk-averse and
the lottery ticket is riskier (because selling the ticket will bring to the agent a deterministic wealth
of $11,000), the agent will be happy to sell the ticket for $1,000. Alternatively, you can reach the
same conclusion by computing the expected utility of each of the two options. If the agent sells
the lottery ticket for $1,000, her expected utility will be (10, 000 + 1, 000) 0.5 ≈ 104.88. However,

106
 √ √
if she keeps the ticket, your expected utility will be 0.5 10, 000 + 0.5 10, 000 + 2, 000 ≈ 104.77.
As a result, selling the ticket for $1,000 will yield a higher expected utility, which is thus the
optimal choice for the agent.
(b) Let 𝑝 denote the price at which the agent sells the ticket. Then she obtains an expected utility of
(10, 000 + 𝑝) 0 .5 by selling the ticket for 𝑝. For this to be the optimal choice, we must have
√︁ √︁
(10, 000 + 𝑝) 0.5 > 0.5 10, 000 + 0.5 10, 000 + 2, 000,

from which we solve 𝑝 > 977.23 (if you use the approximation 104.77 for the RHS, then the
answer will be 𝑝 > 976.75).
(c) The situation is sketched in Figure 14 below.

Utility 𝑢(𝑤)

𝑢(11, 000)
𝐸 [𝑢]

𝑤 (in K$)
10 11 12

Figure 14

(d) It is easy to see that both options yield the same expected wealth level to the agent. Since the
investment is riskier and the agent is risk-averse, the optimal choice is clearly to pay $500 to
avoid the investment.
(e) The situation is sketched in Figure 15 below.

Utility
𝑢(𝑤)
𝑢(9, 500)
𝐸 [𝑢]

𝑤 (in K$)
8 9.5 10

Figure 15

107
8. (a) The expected outcome

𝐸 [𝑉] = 0.1 × 80 + 0.2 × 50 + 0.7 × 10 = 25.

(b) By definition, the variance

Var[𝑉] = 0.1(80 − 25) 2 + 0.2(50 − 25) 2 + 0.7(25 − 10) 2 = 585.

(c) A risk-neutral person will pay at most the expected outcome of the lottery to play it. Thus, the
highest price a risk-neutral player is willing to pay is 𝐸 [𝑉] = 25, as computed in (a).

9. (a) The expected prize of lottery 𝑛 is

0.5(5 − 5𝑛) + 0.5(5 + 5𝑛) = 5.

(b) A risk-neutral agent only cares about the expected prize. Since we have seen in (a) that all the
lotteries yield the same expected prize, a risk-neutral agent must be indifferent with all of them.
(c) The expected utility of lottery 𝑛 is

𝐸 𝑛 [𝑢] = 0.5𝑢(5 − 5𝑛) + 0.5𝑢(5 + 5𝑛).

To learn how the expected utility depends on 𝑛, consider the function 𝑓 (𝑥) = 0.5[𝑢(5 − 5𝑥) +
𝑢(5 + 5𝑥)]. Then it is easy to obtain that 𝑓 0 (𝑥) = 0.25[𝑢 0 (5 + 5𝑥) − 𝑢 0 (5 − 5𝑥)]. Since 𝑢 0 (·) is
decreasing, 𝑓 0 (𝑥) is decreasing in 𝑥; since 𝑓 0 (0) = 0, it follows that 𝑓 0 (𝑥) < 0 for all 𝑥 > 0. As
a result, the expected utility is decreasing in 𝑛, and so the agent will rank the five lotteries as
5 ≺ 4 ≺ 3 ≺ 2 ≺ 1.
(d) The variance of lottery 𝑛 is

Var𝑛 [𝑉] = 0.5[(5 + 5𝑛) − 5] 2 + 0.5[(5 − 5𝑛) − 5] 2 = 25𝑛2 .

Therefore, we have Var5 [𝑉] > Var4 [𝑉] > Var3 [𝑉] > Var2 [𝑉] > Var1 [𝑉].
(e) The expected prize of lottery 6 is

15 1
𝐸 6 [𝑉] = 3 × + 35 × = 5,
16 16
which is clearly equal to 𝐸 1 [𝑉]. The variance of lottery 6 is

15 1
Var6 [𝑉] = (3 − 5) 2 + (35 − 5) 2 = 60,
16 16
which is higher than Var1 [𝑉] (recall that Var1 [𝑉] = 25). Note that
√ √
𝐸 1 [𝑢] = 0.5 0 + 0.5 10 ≈ 1.58,

108
 15 √ 1√
𝐸 6 [𝑢] = 3+ 35 ≈ 1.99.
16 16
Since 𝐸 6 [𝑢] > 𝐸 1 [𝑢], lottery 6 is preferred to lottery 1 by the agent, although the risk-averse
agent is faced with two lotteries with the same expected prize and lottery 6 is “riskier” in terms
of variance.
This exercise tells us that riskiness is in general not properly measured by variance, even when
we are restricted to the case in which lotteries have the same expected prize.

10. (a) The expected value of her wealth if she invests all her initial wealth in the safe asset is
𝐸 𝑠 [𝑤] = 𝑤 0 𝑅.
(b) The expected value of her wealth if she invests all her initial wealth in the risky asset is

𝐸𝑟 [𝑤] = 𝑤 0 [ 𝑝𝐻 + (1 − 𝑝)𝐿].

(c) When 𝑝𝐻 + (1 − 𝑝)𝐿 < 𝑅, the expected wealth from investing in the risky asset is dominated by
that from investing in the safe asset. Since the agent is risk-averse, it follows immediately that
she will invest all her wealth in the safe asset. Thus, her investment in the risky asset will be 0.
(d) Let 𝜃 ∈ [0, 1] be the fraction of her wealth invested in the risky asset. Then her expected utility
for a given 𝜃 is

𝐸 𝜃 [𝑢] = 𝑝 ln[𝜃𝑤 0 𝐻 + (1 − 𝜃)𝑤 0 𝑅] + (1 − 𝑝) ln[𝜃𝑤 0 𝐿 + (1 − 𝜃)𝑤 0 𝑅].

To maximize 𝐸 𝜃 [𝑢], the optimal 𝜃 ∗ should satisfy the first-order condition

d𝐸 𝜃 [𝑢] 𝑝(𝐻 − 𝑅) (1 − 𝑝) (𝐿 − 𝑅)
= + ∗ = 0,
d𝜃 𝜃=𝜃 ∗ 𝜃∗𝐻 + (1 − 𝜃 )𝑅 𝜃 𝐿 + (1 − 𝜃 ∗ )𝑅
∗

from which we solve

𝑅[ 𝑝𝐻 + (1 − 𝑝)𝐿 − 𝑅]
𝜃∗ = .
(𝐻 − 𝑅) (𝑅 − 𝐿)

11. Proof. We show that there is no Bernoulli utility function that can justify Sarah’s choices. Suppose
Sarah’s Bernoulli utility function is 𝑢(·). Then we must have

𝑢(100) > 0.8𝑢(150) + 0.2𝑢(0)

and
0.5𝑢(100) + 0.5𝑢(0) < 0.4𝑢(150) + 0.6𝑢(0).

After merging like terms, the second inequality can be rewritten as

0.5𝑢(100) < 0.4𝑢(150) + 0.1𝑢(0),

109
 which, after multiplying 2 on both sides, can be equivalently stated as

𝑢(100) < 0.8𝑢(150) + 0.2𝑢(0).

Therefore, the two inequalities are contradicting each other, telling us that there is no Bernoulli
utility function 𝑢(·) that can explain Sarah’s decisions, as was to be shown. 

12. (a) If Frank does not sue the manufacturer, his expected utility is 𝐸 𝑁 [𝑢] = 1 − 1/7 = 6/7 ≈ 0.86.
If he files a suit and pays the lawyer a fee of 2, his expected utility will be
   
1 1
𝐸 𝑆 [𝑢] = 0.5 1 − + 0.5 1 − = 0.85.
7+5−2 7−2

As a result, Frank will not file a suit in this case.

(b) As we have seen in (a), a fee schedule that is outcome-independent cannot make it. Thus, we can
restrict our attention to outcome-dependent fee schedules. In particular, let 𝐹1 be the lawyer’s
fee if Frank sues and wins and 𝐹2 be the fee if Frank sues and loses. Then a schedule satisfying
the requirement exists if and only if the following system of inequalities has a solution:
   
1 1 6
0.5 1 − + 0.5 1 − >
12 − 𝐹1 7 − 𝐹2 7

0.5𝐹1 + 0.5𝐹2 > 2

It is more or less easy to see that the system has a solution. For example, (𝐹1 , 𝐹2 ) = (4, 0)
satisfies the two inequalities, that is, Frank pays the lawyer a fee of 4 if and only if he wins.

13. (a) The expected value of the investment is

𝐸𝑉 = 0.4 × 300 + 0.2 × 250 + 0.4 × 200 = 250.

(b) In order to find the certainty equivalent, we need to first determine the expected utility.
Straightforwardly, we have
√ √ √
𝐸 (𝑢) = 0.4 300 + 0.2 250 + 0.4 200 ≈ 15.7473.

Then the certainty equivalent 𝐶𝐸 satisfies

𝐶𝐸 0.5 = 𝑢(𝐶𝐸) = 𝐸 (𝑢) = 15.75,

from which we solve 𝐶𝐸 ≈ 247.98.

14. (a) The implication is that 𝑢(𝑥) is strictly increasing in 𝑥.

(b) It is clear that Lottery 1 is the best for Jim, since the worst outcome of Lottery 1 (i.e., 1)

110
 dominates that of the other two lotteries (i.e., 0). Also, Lottery 2 and Lottery 3 are regarded as
equally good by Jim since they have the same worst outcome.
(c) Proof.Suppose Jim is an expected utility maximizer whose Bernoulli utility function is 𝑢(·).
Since Lotteries 2 and 3 are equally good to Jim, we must have

0.35𝑢(0) + 0.6𝑢(50) + 0.05𝑢(100) = 0.35𝑢(0) + 0.6𝑢(40) + 0.05𝑢(100),

which implies that 𝑢(50) = 𝑢(40), contradicting the property of “the-more-the-better”. Therefore,
Jim cannot be an expected utility maximizer. 
(d) Lottery 1 will be the best for Jack. Let 𝑣(·) be Jack’s Bernoulli utility function. Then it is clear
that lottery 2 is better than lottery 3 because

𝐸 2 [𝑣] = 0.35𝑣(0) + 0.6𝑣(50) + 0.05𝑣(100) > 0.35𝑣(0) + 0.6𝑣(40) + 0.05𝑣(100) = 𝐸 3 [𝑣].

To show that 𝐸 1 [𝑣] > 𝐸 2 [𝑣], notice that this is equivalent to

0.2𝑣(1) + 0.75𝑣(50) + 0.05𝑣(100) > 0.35𝑣(0) + 0.6𝑣(50) + 0.05𝑣(100),

which can be simplified to

0.2𝑣(1) + 0.15𝑣(50) > 0.35𝑣(0).

Since 𝑣(50) > 𝑣(1) > 𝑣(0), the LHS of the inequality above is greater than 0.2𝑣(0) + 0.15𝑣(0) =
0.35𝑣(0), and so our claim proves true.
√
15. (a) No. For example, if Andrea’s Bernoulli utility function is 𝑣(𝑤) = 𝑤, we still get

𝑢(1 million) = 1000 < 1113.61 ≈ 0.89𝑢(1 million) + 0.1𝑢(5 million)

0.11𝑢(1 million) = 110 < 223.61 ≈ 0.1𝑢(5 million)

which is consistent with the choices we have observed. However, since 𝑣 is concave, this example
illustrates the possibility that Andrea is risk averse.
(b) Since Bob has chosen gamble 1A in stage 1, then it holds that

𝑢(1 million) > 0.89𝑢(1 million) + 0.01𝑢(0) + 0.1𝑢(5 million).

Similarly, since he has chosen gamble 2B over gamble 2A in stage 2, we must have

0.11𝑢(1 million) + 0.89𝑢(0) < 0.1𝑢(5 million) + 0.9𝑢(0).

(c) Notice that the first inequality in (b) can be equivalently written as

0.11𝑢(1 millon) > 0.01𝑢(0) + 0.1𝑢(5 million),

111
 while the second inequality in (b) can be transformed to

0.11𝑢(1 million) < 0.01𝑢(0) + 0.1𝑢(5 million).

Clearly, the two inequalities are contradicting each other, based on which we conclude that Bob
cannot be an expected utility maximizer.
(d) The answer is generically open. The result does show that people may have reasoning other than
maximizing expected utility when deciding in situations with uncertainty, which the expected
utility theory fails to capture. In a sense, it enriches expected utility theory rather than simply
challenging it.

112
 6

Introductory Game Theory and

Oligopoly

6 Introductory Game Theory and Oligopoly

113
Instruction. We follow the convention throughout this document: In a payoff matrix with two players, in
each entry the first number represents the payoff to the row player and the second represents that to the
column player.

Exercise 1. Consider the two-player game described by the payoff matrix below.

Belle
𝐿 𝑅
𝑈 1, 1 0, 0
Alan
𝐷 0, 0 3, 3

(a) There are two pure-strategy Nash equilibria for this game. Find them.
(b) Among the two pure-strategy Nash equilibria, which do you think will be played more likely? Why?
(c) This game also has a mixed-strategy Nash equilibrium; find the probabilities the players use in this
equilibrium, together with an explanation for your answer.

Exercise 2. Imagine a couple who agreed to meet this evening, but cannot recall if they will be attending
the opera or a football game. Suppose that they cannot communicate with each other to fix the problem.
The husband (player 𝐻) would prefer to go to the football game but the wife (player 𝑊) would rather go to
the opera. Both would prefer to go to the same place rather than different ones. In particular, if both end
up with going to the opera, 𝐻’s utility will equal 1 and 𝑊’s will equal 2; if both end up with going to the
football game, 𝐻’s utility will equal 2 and 𝑊’s will equal 1; and each gets zero utility in all other cases.

(a) Represent the game above in its normal form. Make sure that you have clearly and correctly specified
the players, their strategies, and each player’s payoff at every possible outcome of the game.
(b) Is there any dominant strategy for either player? Explain.
(c) Find all pure-strategy Nash equilibria for this game.
(d) Find the mixed-strategy Nash equilibrium for this game.

Exercise 3. Consider the game between an employer and an employee. The employer has to decide first,
whether to award a high salary (𝑤 = 2) or a low salary (𝑤 = 0). Then, the decision of the employee about
whether to put extra effort follows. The employee, after seeing the salary, chooses either to put extra effort
(𝑒 = 5) or not (𝑒 = 0). If the employee chooses effort level 𝑒 under the salary 𝑤, the payoff to the employer
equals 𝑒 − 𝑤 and that to the employee equals 𝑤 − 𝑒.

(a) Draw the decision tree for the sequential game. Clearly label each node and branch.
(b) Argue that the employee has a dominant strategy.
(c) Find all Nash equilibria of the game. Is there any Nash equilibrium in which the employer puts extra
effort and enjoys a high salary?

114
 (d) Suppose the game is repeated 10 times. Find the Nash equilibrium that survives backward induction.

Exercise 4. Consider the rock-paper-scissors game that is played once. That is, there are two players 𝐴
and 𝐵, each has three possible handsigns: rock, paper, and scissors. Players each chooses a handsign and
the winner is determined via the following rule (as is well-known): Paper wins rock, rock wins scissors,
and scissors wins paper, and ties are reached if both choose the same handsigns. Assume that the utility
of the winner is 1 and that of the loser is −1, while each gets zero utility if a tie is reached.

(a) If both players move simultaneously, represent the game in its normal form.
(b) Is there any pure-strategy Nash equilibrium for the game as you represented in (a)? Explain.
(c) Now consider another version of the game. Assume that 𝐵 can “cheat” and always decide after
seeing 𝐴’s decision (but 𝐴 can neither do the same thing nor stop 𝐴 from cheating). Draw the
decision tree for this game and find player 𝐵’s best reaction.
(d) Is there any pure-strategy Nash equilibrium for the game as described in (c)? If yes, find all
pure-strategy Nash equilibria for it; if no, carefully justify your answer.
(e) Find a Nash equilibrium for the game (as described in (c)) in which at least one player randomizes.
Is it the unique Nash equilibrium for the game that has this property? Briefly explain.

Exercise 5. A traveler dines at a restaurant along a highway. Both he and the waiter who serves him are
rational and self-interested in the narrow sense. The waiter must first choose between providing good
service (𝐺) or bad service (𝐵), whereupon the diner must choose between leaving the customary tip (𝑇) or
leaving no tip at all (𝑁). The payoffs for their interaction are summarized by the game tree below, where
in each parenthesis the first entry represents the payoff to the waiter and the second represents that to the
diner.

(80, 30)
𝐵

Waiter (100, 100)

𝑇
𝐺
Diner
𝑁 (20, 200)

(a) According to the game tree, which player moves first, which player moves second? Can the second
mover observe the action of the first mover?
(b) Find the diner’s best reaction when the waiter chooses to provide good service.
(c) Find the Nash equilibrium of this sequential game using backward induction. Your answer to (b)
should be useful here.

115
 (d) Is there any strategy profile that will make both the waiter and the diner better off as compared with
the Nash equilibrium you have identified in (c)? Justify your answer.
(e) What is the most the diner would be willing to pay for the right to make a binding commitment
(visible to the waiter) to leave the customary tip at the end of the meal in the event of having received
good service?

Exercise 6. Consider the following normal-form game:

Bob
𝐿 𝑅
𝑈 2, 3 4, 2
Ann
𝐷 1, 2 3, 3

(a) Is there any dominant strategy for either player? If yes, find all of them; if no, justify your answer.
(b) Find all Nash equilibria of this game.

Now consider a variant of the game. Assume that the game is played sequentially: Ann moves first, and
then Bob moves, before which he can see Ann’s move.

(c) Represent this game in its extensive form.

(d) Find all dominant strategies if there are any, or show that there is no dominant strategy otherwise.
(e) Find all Nash equilibria that survive backward induction for this sequential game. Compare your
answer with that to (b) and interpret the result if the two are different.

Exercise 7. Two firms are in the chocolate market. Each can choose to go for the high end of the market
(high quality) or the low end (low quality). Resulting profits are given by the following payoff matrix

Firm 2
Low High
Low −20, −30 200, 300
Firm 1
High 100, 100 50, 50

(a) Is there any dominant strategy? Explain.

(b) Find all pure-strategy Nash equilibria.
(c) Find the mixed-strategy Nash equilibrium.

116
Exercise 8. (One-shot bargaining) Consider the following game between two (fully rational) players.
Player 𝐴 (she) moves first and proposes a plan (𝑎, 𝑏) to split $100, where 𝑎 > 0 is the money player 𝐴 will
get, 𝑏 > 0 is the money player 𝐵 (he) will get, and 𝑎 + 𝑏 = 100. For example, (50, 50) means a plan that
gives each player $50. For simplicity, assume that both 𝑎 and 𝑏 must be whole numbers. Then, player 𝐵
can accept or reject the plan. If player 𝐵 accepts the plan, then each player gets the money as specified in
the plan and the game ends. If, however, player 𝐵 rejects the plan, then the game ends and each player gets
$0. Assume that more money is better than less to each player.

(a) If player 𝐵’ strategy is to reject any plan with probability 0.4 and accept with probability 0.6 (no
matter what the plan is), find player 𝐴’s best reaction.
(b) If player 𝐵’s strategy is to reject any plan that gives him less than $50 (and accept otherwise), what
is player 𝐴’s best reaction?
(c) Denote by 𝛽 player 𝐵’s strategy in (b), and by 𝛼 player 𝐴’s best reaction against 𝛽 (as you have
obtained in (b)). Is (𝛼, 𝛽) a Nash equilibrium of the bargaining game? Justify your answer.
(d) Find all pure-strategy Nash equilibria that survive backward induction.

Exercise 9. (Incredible Threat) Consider the following extensive form game between Lily and Mark,
where in each parenthesis the first number is the payoff to Lily and the second is that to Mark:

(3, 4)
𝑈

Lily (1, 2)
𝑅
𝐷
Mark
𝐿 (4, 3)

(a) According to the game tree, who moves first? Who moves second? Can the second mover observe
the action of the first mover?
(b) Find the Nash equilibrium that survives backward induction.
(c) Verify that the following strategy profile constitutes a Nash equilibrium: Lily plays 𝑈 and Mark
plays 𝑅 once Lily plays 𝐷. Denote this Nash equilibrium by 𝐼.
(d) Can 𝐼 survive backward induction? Justify your answer.

Exercise 10. (Super Golden Balls) Two agents, Andrew and Betty, are playing the following game. Each
player has two feasible actions, split (𝑃) and steal (𝑇), and they make their decisions simultaneously. If
both choose 𝑃, then each obtains a payoff of 5. If exactly one player chooses 𝑇, then the player gets a

117
payoff of 10 and the other player (i.e., the one who chooses 𝑃) gets 0. If, however, both choose 𝑇, then
each gets 0.
(a) According to the verbal description, find the values for (i)-(v) in the payoff matrix below to complete
the normal form of this game. Note that by convention the first number in each entry of the matrix
should stand for the payoff to the row player and the second for that to the column player.

Betty
𝑃 𝑇
𝑃 5 , (i) (ii) , (iii)
Andrew
𝑇 (iv) , 0 0 , (v)

(i) = , (ii) = , (iii) = , (iv) = , (v) = .

(b) Argue that there is no mixed-strategy Nash equilibrium in which both players randomize. In other
words, show that no player would ever find it optimal to randomize between 𝑃 and 𝑇 regardless the
strategy of his/her opponent.
(c) Find all Nash equilibria of this game. Also, show that in each Nash equilibrium at least one player
will choose 𝑇.
Now consider a repeated version of this game. Suppose that the game is played up to 100 rounds. The
payoff matrix for each round is the same as above. In addition, if either player ever chooses 𝑇 in one round,
then the game will be terminated at the end of that round (and each gets zero forever starting from the next
round); otherwise the game will be played for 100 rounds. Assume that both players want to maximize the
sum of the payoff in each of the 100 rounds.
(d) Suppose Andrew’s strategy is to always play 𝑃 until Betty plays 𝑇, in which case he will always play
𝑇 from the next round on. Find Betty’s best reaction against Andrew’s strategy.
(e) Argue that it is a Nash equilibrium if both players always play 𝑇. Moreover, this Nash equilibrium
can survive backward induction.
(f) (A bit harder) Consider the following strategy profile for the two agents: Andrew plays 𝑇 in all odd
rounds (i.e., the first round, the third round, the fifth round, etc.) and plays 𝑃 in all even rounds (i.e.,
the second round, the fourth round, the sixth round, etc.); while Betty plays 𝑇 in all even rounds and
𝑃 in all odd rounds; if either player deviates from the strategy, then both will play 𝑇 in each of the
rest rounds. Show that this is a Nash equilibrium that survives backward induction.

Exercise 11. Suppose that a good is produced by two firms (i.e., the industry has a duopoly structure),
denoted 1 and 2. Firm 1’s cost function is

𝐶1 (𝑞) = 𝑞 2 ,

118
and firm 2’s cost function is
𝐶2 (𝑞) = 12𝑞.

The demand function for the good is given by 𝑞 = 100 − 𝑝.

(a) Derive the best reaction function for each firm.

(b) Find the output level that will be produced in a Cournot-Nash equilibrium, and the price level in
such an equilibrium.
(c) Illustrate your solution in (b) above in a suitable diagram.

Exercise 12. Consider two firms that are the sole producers of mineral water in a market. Suppose the
market demand is given by 𝑄 𝑑 = 24 − 𝑝, and that each firm can produce mineral water from its own spring
at zero marginal and average cost.

(a) Find the monopoly output level (i.e., suppose the production of mineral water is monopolized by a
single provider).
(b) If the two firms are considering a collusive agreement under which each produces half the monopoly
output and offers it for sale at the monopoly price. Find each firm’s economic profit in this case (i.e.,
when both firms follow the agreement).
(c) Show that each firm has incentive to deviate from the agreement if the other firm abides by the
agreement. Hint. You need to show that the best reaction of a firm is different from the output level
prescribed by the collusive agreement, given that the other firm abides by the agreement.
(d) Find the Cournot-Nash equilibrium for this market.
(e) Suppose there is a third party that can implement the agreement by using punishment. That is, it can
issue a firm a fine of 𝑝 ∗ if the latter fails to abide by the agreement, assuming that each firm’s action
is publicly observable and verifiable. Find the minimum punishment (i.e., the minimum value for
𝑝 ∗ ) that suffices to implement the collusive agreement.

Exercise 13. Defendo has decided to introduce a revolutionary video game. As the first firm in the market,
it will have a monopoly position for at least some time. In deciding what type of manufacturing technology
to employ, it has two alternatives. Technology 𝐴 is publicly available and will result in annual total cost
𝑇𝐶 𝐴 (𝑞) = 10 + 8𝑞. Technology 𝐵 is proprietary technology developed in Defendo’s research labs. It
involves a higher fixed cost of production but lower marginal cost: 𝑇𝐶 𝐵 (𝑞) = 60 + 2𝑞. Market demand for
the new product is 𝑝 = 20 − 𝑄, where 𝑄 is total industry output.

(a) Suppose Defendo were certain that it would maintain its monopoly position in the market for the
entire product lifespan without threat of entry. Which technology would Defendo adopt? Explain.
What would be Defendo’s profit given the choice?
(b) Suppose Defendo expects another firm, Offendo, to enter the market shortly after Defendo introduces
its new product. Offendo will only have access to Technology 𝐴. If Offendo does enter the market,

119
 the two firms will play a Cournot game (competing in quantities) and arrive at the Cournot-Nash
equilibrium. If Offendo does not enter the market, its profit will be zero. Find the profit for each
firm in each of following two cases. (i) Defendo adopts technology 𝐴 and Offendo chooses to enter
the market. (ii) Defendo adopts technology 𝐵 and Offendo chooses to enter the market.
(c) Assume that the Defendo moves first and its decision is observed by Offendo when the latter is
deciding whether to enter or not. Using your results in (a) and (b), represent this game in its extensive
form. Clearly labor each node and branch with the decision maker, the action, or the payoffs.
(d) Find a Nash equilibrium of this game using backward induction.
(e) How does the threat of entry affect social welfare (in terms of the sum of consumer and producer
surplus) in the market? Why is this the case?

Exercise 14. The state has announced its plans to license two firms to serve a market whose demand curve
is 𝑄 𝑑 = 100 − 𝑝. The technology is such that each can produce any given level of output at zero marginal
and average cost, but once each firm’s output is chosen, it cannot be altered any more.

(a) What is the most you would be willing to pay for one of these licenses if you knew you would be
able to choose your output level first (assuming your choice was observable by the rival firm)?
(b) How much would your rival be willing to pay for the license (you rival, after getting the license, will
choose second)?

Exercise 15. Consider two firms that compete in quantities (i.e., Cournot competition). The inverse
demand function is given by 𝑝 = 3 − 𝑄, where 𝑄 is the total output. Firm 1 makes an observable investment
decision before the firms set quantities. If firm 1 decides not to invest, it pays nothing and its marginal and
average cost of production is 1. If, instead, firm 1 decides to invest, it pays a lump-sum cost of 𝐹 but its
marginal and average cost of production will be lowered to 0. In any event, firm 2’s marginal and average
cost is 1.

(a) Find the equilibrium quantities and price if firm 1 does not invest.
(b) Find the equilibrium quantities and price if firm 1 invests.
(c) Given your answer in (a) and (b), when will firm 1 invest?
(d) How does the investment decision affect firm 2’s output and profit levels? Briefly explain. You do
not need to do algebra here. A qualitative analysis will be enough.

Exercise 16. The market demand for good 𝑀 is given by 𝑄 𝑑 = 50 − 𝑝. Good 𝑀 is produced by two firms,
𝐴 and 𝐵. The cost function of firm 𝐴 is 𝑇𝐶 𝐴 (𝑞) = 2𝑞 and that of firm 𝐵 is 𝑇𝐶 𝐵 (𝑞) = 4𝑞. Firm 𝐴 is the
Stackelberg leader and firm 𝐵 is the follower.

120
 (a) Find firm 𝐵’s best reaction when firm 𝐴 has chosen to produce 30.
(b) To drive firm 𝐵’s market share (i.e., the ratio between its output and the total output on the market)
down to 0, how much should firm 𝐴 produce at least?
(c) Find the output level for each firm and the market price in the Stackelberg equilibrium.

Exercise 17. Two firms, 1 and 2, are producing the same good and competing by choosing their prices 𝑝 1
and 𝑝 2 (simultaneously). The market demand is 𝑄 𝑑 = 60 − 𝑝. The firm with a lower price gets the entire
market, and each firm gets one half of the market if 𝑝 1 = 𝑝 2 . Both firms are operating in the long run with
a constant marginal cost that equals 36.

(a) What is the market price (i.e., the price at which trade happens) in the Bertrand equilibrium?
(b) Now suppose that firm 1 is still operating with the same marginal cost (i.e., 36) but firm 2 can reduce
its (constant) marginal cost via technology upgrade. For simplicity, we also assume that all prices
must be whole numbers. Find the highest possible (constant) marginal cost for firm 2 at which it
affectively becomes a monopolist on the market.

Exercise 18. To produce a certain homogeneous final good, 𝑛 manufacturers need two complementary
technologies, whose patents are owned by two firms 𝐴 and 𝐵. Each of the two firms separately licenses
the technologies at a unit royalty fee 𝑤 𝑖 (𝑖 = 𝐴, 𝐵). That is, if a manufacturer produces, then for each unit
of the good it produces, it has to pay firm 𝐴 a fee of 𝑤 𝐴 and firm 𝐵 a fee of 𝑤 𝐵 . The game goes as follows.
In the first stage, the patentholders independently and simultaneously decide the royalty fee 𝑤 𝐴 and 𝑤 𝐵 .
In the second stage, the manufacturers are engaged in a Bertrand competition, and each incurs a constant
marginal and average cost 𝑐 + 𝑤 𝐴 + 𝑤 𝐵 , where 0 < 𝑐 < 1 is the real production cost. The market demand
curve is given by 𝑄 𝑑 = 1 − 𝑝 (as usual, if several firms all charge the same lowest price, then the quantity
of demand is equally shared among them; zero demand goes to firms having higher prices).

(a) When 𝑤 𝐴, 𝑤 𝐵 are given, what is the price in the Bertrand equilibrium in the second stage of the
game?
(b) Given your answer to (a), find the values of royalty fees in the Nash equilibrium of the game (between
the two patentholders). Also, calculate each patentholder’s equilibrium profit in the equilibrium.

Now consider another scenario. Assume that the two patentholders assign the right of exploitation of their
patents to a patent pool. It is now the pool which sets the value of both royalty fees.

(c) Find the optimal royalty fees for the patent pool. Here you need to use your result in (a) to formulate
the patent pool’s profit.
(d) Do consumers benefit from forming the patent pool? Do patentholders benefit from that? Justify
your answer.

121
Exercise 19. There are two firms in a market, denoted by firm 1 and firm 2, respectively. Firm 𝑖
produces good 𝑖 (𝑖 = 1, 2). Firm 1’s total cost function is 𝐶1 (𝑞) = 𝑞 + 𝑞 2 ; firm 2’s total cost function is
𝐶2 (𝑞) = 𝛼𝑞 + 𝑞 2 (𝛼 > 0). Good 1’s demand is given by

𝑄 1𝑑 = 2 − 2𝑝 1 + 𝑝 2 ,

and good 2’s demand is given by

𝑄 2𝑑 = 2 − 2𝑝 2 + 𝑝 1 .

The two firms compete in prices simultaneously (i.e., they set prices simultaneously).

(a) According to the demand functions, are good 1 and good 2 substitutes or complements? Explain
briefly.
(b) Find each firm’s best reaction function.
(c) Given 𝛼 = 1, calculate the equilibrium prices.
(d) Suppose 𝛼 increases. How will this affect the firms’ best reaction functions and the equilibrium
prices?

***********

Exercise 20. Boeing and Airbus both decide whether to produce a new model of airplane. Each firm’s
profits depend on all firms’ decisions. In particular, their payoffs are given in the following matrix:

Airbus
Produce Not produce
Produce −5, −5 100, 0
Boeing
Not produce 0, 100 0, 0

Suppose Boeing and Airbus decide simultaneously.

(a) Does any firm have a dominant strategy? Justify your answer.
(b) Find all pure-strategy Nash equilibria of this game.
(c) Find the mixed-strategy Nash equilibrium of the game.
(d) To boost manufacturing industry, the US government decides to give a subsidy of 25 to Boeing if
Boeing produces the new model of airplane. Find the Nash equilibrium in this case.

From now on suppose that Airbus decides first (i.e., Boeing can observe Airbus’ decision when deciding)
and there is no subsidy available to either company.

(e) Draw a game tree for this sequential game.

122
 (f) Find the Nash equilibrium that survives backward induction for the game. Is there a first-mover
advantage? Explain.

Exercise 21. There are 𝑛 firms that are the sole producers on a market (𝑛 > 2). Each firm has the cost
function 𝐶 (𝑄) = 30𝑄. The inverse demand function for the firms’ products is 𝑝 = 120 − 𝑄, where 𝑄 is
the total output (i.e., the sum of each firm’s output).

(a) What is the best reaction of firm 𝑖 (𝑖 = 1, 2, . . . , 𝑛), denoted by 𝑄 ∗𝑖 , if the total output of the other
firms is 𝑄 −𝑖 ?
(b) Formally argue that all firms must have the same output level in the Cournot equilibrium. Hint. Let
(𝑄 1𝑁 , 𝑄 2𝑁 , · · · , 𝑄 𝑛𝑁 ) be the Cournot equilibrium. One approach is to investigate 𝑄 𝑖𝑁 − 𝑄 𝑖+1
𝑁 using

the best reactions. You may start by investigating 𝑄 1 − 𝑄 2 .

𝑁 𝑁

(c) Let 𝑄 ∗ denote each firm’s output level in the Cournot equilibrium. Find 𝑄 ∗ . Also, find the total
output and the market price in the Cournot equilibrium.
(d) How will the market price behave when 𝑛 is very large? Interpret your finding. Hint. How is the
market price compared with each firm’s marginal cost when 𝑛 gets sufficiently large?

Exercise 22. When a product patent is infringed (i.e., a patent is used without the permission of its holder),
the court can mete out damages. The infringer must either pay the patentholder’s lost profit or return its
unjust enrichment. These remedies serve two purposes: compensate the infringed patentholder and deter
infringement. Assume that the demand curve for the proprietary good is 𝑄 𝑑 = 1 − 𝑝, while its marginal
and average cost of production equals 𝑐 > 0. If an infringer enters the market, both firms compete in
quantities (i.e., Cournot competition). The good cannot be produced without using the patent.

(a) Find first the patentholder’s profit under monopoly (i.e., the patentholder is the only producer in the
market and there is no infringement).
(b) Assume that both firms (the patentholder and an infringer) are competing in the market. Find the
profit level for each of the two in the Cournot equilibrium, given that the infringer is not punished.
(c) Find the patentholder’s lost profit (i.e., the profit that is lost due to infringement) and the infringer’s
unjust enrichment (i.e., the profit that is obtained by the infringer using the patent).
(d) Now consider punishment. Provided infringement, does the patentholder prefer to get back the lost
profit or the unjust enrichment of the infringer? Justify your answer.
(e) Does the lost-profit penalty (i.e., the infringer pays the patentholder the lost profit) deter infringement?
Justify your answer.

123
Chapter 6 Answer Key

1. (a) It is easy to check that both (𝑈, 𝐿) and (𝐷, 𝑅) are pure-strategy Nash equilibria for this game.
(b) Intuitively and empirically, (𝐷, 𝑅) is played more likely than other strategies. A direct explanation
is the people tend to assume that the their opponents are very likely to play 𝐷 or 𝑅. The answer
should be open as there might be psychological or cultural background for this tendency (search
for “Focal point” to learn more about this if you are interested), but it is generally beyond the
scopes that are captured by simple but definite solution concepts like Nash equilibrium.
(c) Let 𝑝, 𝑞 ∈ (0, 1) be the probability that Alan plays 𝑈 and Belle plays 𝐿, respectively, in the
mixed-strategy Nash equilibrium. Then both players’ indifference prescribes that

𝑞 = 3(1 − 𝑞)
𝑝 = 3(1 − 𝑝)

from which we solve 𝑝 = 𝑞 = 0.75. That is, the following constitutes a mixed-strategy Nash
equilibrium: Alan plays 𝑈 with probability 0.75 and 𝐷 with probability 0.25, Belle plays 𝐿 with
probability 0.75 and 𝑅 with probability 0.25.

2. (a) The normal form of the game is given below.

Wife
Opera Football Game
Opera 1, 2 0, 0
Husband
Football Game 0, 0 2, 1

(b) There is no dominant strategy for either player since “Opera” is the unique best reaction to
“Opera” but “Football Game” is the unique best reaction to “Football Game”.
(c) There are two pure-strategy Nash equilibria: (Opera, Opera) and (Football Game, Football Game).
(d) Let 𝑝, 𝑞 ∈ (0, 1) be the probability that Husband plays Opera and Wife plays Opera, respectively,
in the mixed-strategy Nash equilibrium. Then both players’ indifference prescribes that

𝑞 = 2(1 − 𝑞)
2𝑝 = 1 − 𝑝

from which we solve 𝑝 = 1/3 and 𝑞 = 2/3. That is, the following constitutes a mixed-strategy
Nash equilibrium: Husband plays Opera with probability 2/3 and Football Game with probability
1/3, Wife plays Opera with probability 1/3 and Football Game with probability 2/3.

3. (a) See Figure 16 below for the decision tree. In each parenthesis, the first entry stands for the payoff
to the employer and the second for the employee.

124
 𝑒=5 (3, −3)
Employee
𝑤=2
𝑒=0 (−2, 2)
Employer
𝑒=5 (5, −5)
𝑤=0
Employee
𝑒=0 (0, 0)

Figure 16

(b) For all levels of salary (high or low), the best strategy for the employee is not to put extra effort
(𝑒 = 0). Therefore, it is a dominant strategy for the employee to play 𝑒 = 0.
(c) Since the employee will always play 𝑒 = 0, it is clear that the employer’s best reaction is to play
𝑤 = 0 (if the employer plays 𝑤 = 2, then he will end up with a payoff of −2). Thus, there is a
unique Nash equilibrium for this game, in which the employer awards a lower salary (𝑤 = 0)
and the employee does not put extra effort (𝑒 = 0). As a result, there is no Nash equilibrium in
which the employer puts extra effort and enjoys a high salary.
(d) In the last round, it is clear that the agents will play the unique Nash equilibrium (𝑤 = 0, 𝑒 = 0)
irrespective of what happened before. Given this, the Nash equilibrium will also be played
in the second last round. Inductively, if in each of the rest rounds the agents will play the
Nash equilibrium (𝑤 = 0, 𝑒 = 0) no matter what happened before, then they will also play
(𝑤 = 0, 𝑒 = 0) for the current round. Therefore, we conclude that the unique Nash equilibrium
for the finitely repeated game is the strategy profile in which the employer always plays 𝑤 = 0
and the employee always plays 𝑒 = 0.

4. (a) The normal form of the game is the following 3 × 3 payoff matrix, where we use 𝑅 to represent
“Rock”, 𝑃 to represent “Paper”, and 𝑆 to represent “Scissors”.
𝐵
𝑅 𝑃 𝑆
𝑅 0, 0 −1, 1 1, −1
𝐴 𝑃 1, −1 0, 0 −1, 1
𝑆 −1, 1 1, −1 0, 0

(b) There is no pure-strategy Nash equilibrium. The reason is that for each strategy profile of
pure strategies, there will be at least one player who has profitable and unilateral deviation. In
particular, if there is a loser at the strategy profile, then the loser has incentive to deviate to the
best reaction to the other’s strategy. The same is true if there is a tie, in which case each has an
incentive to deviate to the best reaction to the other’s strategy.
(c) The decision tree is drawn in Figure 17 below. Cheating gives player 𝐵 an absolute advantage.

125
 (0, 0)
𝑅
𝐵 𝑃
(−1, 1)

𝑆
(1, −1)
𝑅

(1, −1)
𝑅
𝑃 𝐵 𝑃
𝐴 (0, 0)

𝑆
(−1, 1)

𝑆
(−1, 1)
𝑅
𝑃
(1, −1)
𝐵
𝑆
(0, 0)

Figure 17

In particular, player 𝐵 will always win. The best reaction of player 𝐵 in this case, denoted by
𝐵𝑅 𝐵 (·) is the following contingent plan:

if 𝑥 = 𝑅



 𝑃,


𝐵𝑅 𝐵 (𝑥) = 𝑆, if 𝑥 = 𝑃 .



 𝑅,
 if 𝑥 = 𝑆

(d) Notice that player 𝐴 will always lose in this case, and so player 𝐴’s choice does not matter.
Thus, every strategy of player 𝐴 is a best reaction. So we have three Nash equilibria: (𝑅, 𝐵𝑅 𝐵 ),
(𝑃, 𝐵𝑅 𝐵 ), and (𝑆, 𝐵𝑅 𝐵 ), where 𝐵𝑅 𝐵 is the strategy derived in (c).
(e) It is clear that player 𝐵 has a unique optimal reaction that is pure and so 𝐵 will not randomize.
However, since player 𝐴 is indifferent with all pure strategies, she is happy to randomize over the
set {𝑅, 𝑃, 𝑆}. For example, player 𝐴 can play each of the three pure strategies with probability
1/3, which, together with 𝐵’s strategy profile 𝐵𝑅 𝐵 , constitutes a Nash equilibrium. Clearly,
there are infinitely many Nash equilibria of this type since player 𝐴 is happy to randomize over
{𝑅, 𝑃, 𝑆} in an arbitrary way, given 𝐵𝑅 𝐵 .

5. (a) The waiter moves first and the diner moves second, according to the game tree, since the game
always starts from the origin, where the decision maker is the waiter, and then moves to the
immediate successor of the origin, where the decision maker is the diner, and so on. By the

126
 setup, it is clear that the diner can see the waiter’s move when deciding.
(b) The diner’s best choice is “leaving no tip (i.e., 𝑁)”, as this will give the diner a payoff of 200
while the payoff would be 100 if the diner chose to leave the customary tip.
(c) Given the diner’s best reaction, if the waiter chooses 𝐵, then his payoff will be 80; if he chooses
𝐺, his payoff will be 20 (since in this case the diner will choose 𝑁 rather than 𝑇). Therefore, the
Nash equilibrium that survives backward induction is for the waiter to play 𝐵 and the diner to
play 𝑁 once the waiter ever plays 𝐺.
(d) The Nash equilibrium in (c) yields the payoff (80, 30) to the players, which is dominated by the
payoff (100, 100) (i.e., each player is better off as compared with that in the Nash equilibrium).
The latter payoff is achieved by the strategy profile (𝐺, 𝑇). Thus, (𝐺, 𝑇) is such a strategy
profile.
(e) If the diner pays 𝑝 to achieve the binding commitment, the waiter will choose to provide good
service, and so the diner will end up with a payoff of 100 − 𝑝. Without the commitment, we have
known that the agents will end up with playing the Nash equilibrium and the diner’s payoff will
be 30. Thus, paying to make commitment is preferred by the diner if and only if 100 − 𝑝 > 30,
from which we solve 𝑝 6 70. So the diner will be willing to pay at most 70 to make a binding
commitment.

6. (a) 𝑈 is the dominant strategy for Ann since it is the best reaction to every pure strategy of Bob.
Bob, to the contrary, does not have any dominant strategy, as 𝐿 is the unique best reaction against
𝑈 while 𝑅 is the unique best reaction against 𝐷.
(b) Since 𝑈 is Ann’s dominant strategy, she will play 𝑈 is any Nash equilibrium. Given this, Bob
will play the best reaction against 𝑈, which is 𝐿. Thus, the unique Nash equilibrium of this game
is (𝑈, 𝐿).
(c) The extensive form of this game is shown in Figure 18 below. Note that in each parenthesis the
first entry is the payoff to Ann and the second is that to Bob.

𝐿 (2, 3)
Bob
𝑈
𝑅 (4, 2)
Ann
𝐿 (1, 2)
𝐷
Bob
𝑅 (3, 3)

Figure 18

(d) Now Ann does not have a dominant strategy, too, in the sequential game. To see this, note
that Ann’s unique best reaction is 𝑈 if Bob always plays 𝐿, while that is 𝐷 if Bob’s strategy

127
 is to play 𝐿 upon 𝑈 and to play 𝑅 upon 𝐷. Note the difference originates from that fact that
Bob can condition his play on Ann’s play in the sequential game, which is impossible in the
simultaneous-move game.
Interestingly, now Bob does have a (weakly) dominant strategy: play 𝐿 upon 𝑈 and 𝑅 upon
𝐷, which is simply because it is always a best reaction against Ann’s play. This can be seen
straightforwardly from the normal-form representation of this game:
Bob
(𝐿|𝑈, 𝐿|𝐷) (𝐿|𝑈, 𝑅|𝐷) (𝑅|𝑈, 𝐿|𝐷) (𝑅|𝑈, 𝑅|𝐷)
𝑈 2, 3 2, 3 4, 2 4, 2
Ann
𝐷 1, 2 3, 3 1, 2 3, 3

(e) Bob’s best reaction is clearly to play 𝐿 upon 𝑈 and to play 𝑅 upon 𝐷, which is denoted by
𝐵𝑅 𝐴. Given this, Ann’s best reaction is 𝐷. Therefore, the unique Nash equilibrium that survives
backward induction is (𝐷, 𝐵𝑅 𝐴). This is clearly different from the Nash equilibrium in (b).
Again, the difference is caused by the change of timing: since Bob can condition his choice
upon Ann’s decision and thus becomes more “flexible”, he can play 𝑅 rather than 𝐿 upon 𝐷, as
compared with the simultaneous-move game. This gives incentive for Ann to play 𝐷 and the
equilibrium is hence changed.

7. (a) None of the two firms has a dominant strategy. To see this, notice that for each firm, “High”
is the unique best reaction when the opponent plays “Low”, while “Low” is the unique best
reaction when the opponent plays “High”.
(b) There are two pure-strategy Nash equilibria: (Low, High) and (High, Low), the reason is simple:
as we have explained in (a), “Low” is the best reaction to “High” while “High” is the best
reaction to “Low” to both firms.
(c) Let ( 𝑝, 𝑞) denote the mixed-strategy Nash equilibrium for this game, where 𝑝 ∈ (0, 1) represents
the probability that firm 1 plays “Low” and 𝑞 ∈ (0, 1) represents the probability that firm 2 plays
“Low”. In a mixed-strategy equilibrium a player must be indifferent between actions over which
he/she randomizes. Thus, we have
(
−20𝑞 + 200(1 − 𝑞) = 100𝑞 + 50(1 − 𝑞)
−30𝑝 + 100(1 − 𝑝) = 300𝑝 + 50(1 − 𝑝)

from which we solve

5 5
𝑝= , 𝑞= .
38 9
That is, in the mixed-strategy Nash equilibrium, firm 1 chooses “Low” with probability 5/38
and “High” with probability 33/38; firm 2 chooses “Low” with probability 5/9 and “High” with
probability 4/9.

8. (a) If player A’s strategy is to demand 𝑥 (i.e., propose that 𝐴 gets 𝑥 and 𝐵 gets 100 − 𝑥), then 𝐴’s

128
 expected utility is equal to 0.4𝑥. In order to maximize the expected utility, it is clearly that 𝐴
should choose 𝑥 = 100, that is, player 𝐴’s best reaction will be to always propose (100, 0) and
leave nothing to player 𝐵.
(b) If player 𝐴 proposes to give player 𝐵 less than $50, then the proposal is doomed to be rejected
according to player 𝐵’s strategy. If, instead, 𝐴 proposes (𝑥, 100 − 𝑥) such that 𝑥 6 50, then the
proposal will be accepted. Thus, to maximize the money he will have, player 𝐴 should choose
𝑥 = 50. In other words, player 𝐴’s best reaction is to propose (50, 50).
(c) It is a Nash equilibrium of this game. To see it, notice that we have checked in (b) that 𝛼 is the
best reaction against 𝛽. Meanwhile, it is easy to see that given 𝛼, 𝛽 is a best reaction, since
player 𝐵 cannot get more than $50 upon the plan (50, 50), while playing 𝛽 will deliver exactly
$50 to 𝐵.
(d) In the second stage, player 𝐵 has received player 𝐴’s proposal and is deciding whether to accept
or reject. If 𝐵 rejects, her payoff will be zero, and so she should never reject a proposal such that
𝑏 > 0. When the plan leaves nothing to her, 𝐵 is indifferent between accepting and rejecting.
Thus, “accepting all plans” is a best reaction for 𝐵 in the second stage, and so we have a Nash
equilibrium in which 𝐴 proposes (100, 0) and 𝐵 accepts all proposals. However, due to the
indifference, player 𝐵 has another best reaction: “accept all plans that give her some money and
reject the one that leaves nothing to her”, given which player 𝐴’s best reaction is clearly (99, 1)
(i.e., leave the minimum amount of money to 𝐵 in order to get her approval). To summarize,
there are two Nash equilibria that survive backward induction: (i) Player 𝐴 proposes (100, 0)
and player 𝐵 (kindly) accepts all plans; (ii) Player 𝐴 proposes (99, 1) and player 𝐵 accepts all
plans but the one that leaves nothing to her.

9. (a) Lily moves first, which is followed by Mark’s move. According to the setup, Mark can observe
Lily’s move when he is deciding.
(b) Once Lily plays 𝐷, Mark’s best reaction is to 𝐿. Given this, Lily’s best reaction is clearly 𝐷.
Therefore, the unique Nash equilibrium that survives backward induction is (𝐷, 𝐿 | 𝐷), where
𝐿 | 𝐷 means “play 𝐿 upon 𝐷”.
(c) We check that both are playing best reactions against the other’s strategy. Given that Lily plays
𝑈, what Mark will play is payoff-irrelevant (since Mark will not have a chance to move), and
so 𝑅 | 𝐷 (i.e., play 𝑅 upon 𝐷) is a best reaction. Given that Mark will play 𝑅 | 𝐷, Lily’s best
reaction is clearly 𝑈, since otherwise she will end up with a payoff of 1, which is lower than
what she can secure by playing 𝑈 (i.e., 3). Since both are playing best reactions, (𝑈, 𝑅 | 𝐷)
constitutes a Nash equilibrium of the game.
(d) As we have shown in (b), there is only one Nash equilibrium that survives backward induction,
which is not 𝐼. Indeed, 𝑅 | 𝐷 is not a best reaction for the second stage (i.e., given that Mark
gets a chance to move). This is an example of “incredible threat”, that is, players move later can
threaten early movers by “saying” that they will take certain actions with bad consequences to
deter the latter from some actions, but the former actually do not have incentive to carry out the

129
 threats since they are not good for both the early players and the later players themselves. Such
threats can create new Nash equilibria, but are not robust to backward induction (i.e., are not
credible if all players are optimizing everywhere, not matter one gets a chance to move or not).

10. (a) (i) = 5, (ii) = 0, (iii) = 10, (iv) = 10, (v) = 0.

(b) It is easy to see that 𝑇 is strictly better than 𝑃 when the opponent chooses 𝑃 while it is as good
as 𝑃 when the opponent chooses 𝑇. Thus, 𝑇 is the unique best reaction if the opponent plays 𝑃
with a positive probability. Therefore, there is no mixed strategy Nash equilibrium in which
both players randomize.
(c) There are infinitely many Nash equilibria for this game. First, there are three pure-strategy
Nash equilibria: (𝑃, 𝑇), (𝑇, 𝑃), and (𝑇, 𝑇). Second, there are infinitely many mixed-strategy
Nash equilibria. Indeed, once one plays 𝑇 for sure, the other player is indifferent between 𝑃
and 𝑇 since either yields a payoff of 0. Thus, a mixed-strategy equilibrium is in the following
form, one player, either Andrew or Betty, plays 𝑇 for sure, and the other plays 𝑇 with probability
𝜆 and 𝑃 with probability 1 − 𝜆, where 𝜆 can be any number in (0, 1).
(d) The key for Betty is clearly to determine the first time to play 𝑇, since that will trigger Andrew’s
“revenge” and deliver a zero payoff to Betty from then on. Concretely, if Betty play 𝑇 for the first
time in round 𝑛 (1 6 𝑛 6 100), then her payoff will be

5(𝑛 − 1) + 10.

Since her payoff is increasing in 𝑛, Betty will find it optimal to delay 𝑇 as long as possible.
Therefore, the optimal time to play 𝑇 is in the 100 round. Therefore, Betty’s best reaction is to
play 𝑃 in each of the first 99 rounds and play 𝑇 in round 100.
(e) Given that, say, Andrew will always play 𝑇, it is clear that what Betty plays will have not impact
on her payoff (since her payoff will be 0 for each round no matter her strategy). Thus, a best
reaction for Betty is to also play 𝑇 constantly. Hence, it is a Nash equilibrium if both players
always play 𝑇.
To show that such a Nash equilibrium can survive backward induction, let us start from round
100, i.e., the last round. Now this is like a one-shot game, where we know that it is a (weakly)
dominant strategy for a player to play 𝑇. Therefore, it is a Nash equilibrium for both players
to play 𝑇 in round 100. Suppose both players will play 𝑇 starting from period 𝑛 no matter the
history, then the play in round 𝑛 − 1 is like a one-shot game, and so it is optimal for both to play
𝑇 for round 𝑛 − 1. Therefore, this is a Nash equilibrium that can survive backward induction.
(f) Given Andrew’s strategy, Betty cannot benefit from any one-shot deviating: if she deviates in an
even round, then her total payoff from that round on will be 0, while her payoff is at least 10 if
not deviates; if she deviates in an odd round, then her total payoff from that round on will be 0,
while her payoff is at least 0 if not deviates. As a result, Betty has no incentive to deviate from
the strategy, provided the strategy of Andrew. Similar argument for Andrew will establish the
result that he has no incentive to conduct any one-shot deviation. Also, once there has been any

130
 deviation, then as we have shown in (e), both play 𝑇 forever is a Nash equilibrium surviving
backward induction. Therefore, the given strategy profile constitutes a Nash equilibrium that
survives backward induction.

11. (a) When firm 2’s output is 𝑞 2 , firm 1’s profit at output level 𝑞 1 is

𝜋1 (𝑞 1 ; 𝑞 2 ) = (100 − 𝑞 1 − 𝑞 2 )𝑞 1 − 𝑞 21 .

Thus, the best reaction 𝑞 ∗1 solves

𝜕𝜋1 (𝑞 1 ; 𝑞 2 )
= 100 − 𝑞 2 − 4𝑞 ∗1 = 0,
𝜕𝑞 1 𝑞1 =𝑞1∗

and so we get
𝑞2
𝑞 ∗1 (𝑞 2 ) = 25 −
.
4
Similarly, when firm 1’s output is 𝑞 1 , firm 2’s profit at output level 𝑞 2 is

𝜋2 (𝑞 2 ; 𝑞 1 ) = (100 − 𝑞 1 − 𝑞 2 )𝑞 2 − 12𝑞 2 .

The best reaction 𝑞 ∗2 solves

𝜕𝜋2 (𝑞 2 ; 𝑞 1 )
= 88 − 𝑞 1 − 2𝑞 ∗2 = 0,
𝜕𝑞 2 𝑞2 =𝑞2∗

and so
𝑞1
𝑞 ∗2 (𝑞 1 ) = 44 − .
2
(b) Let (𝑞 1𝑁 , 𝑞 2𝑁 ) be the Cournot-Nash equilibrium. Then we must have


 𝑞𝑁
𝑁 = 𝑞 ∗ (𝑞 𝑁 ) = 25 − 2
 𝑞
 1
 1 2
 4
.
 𝑞𝑁
 𝑞 𝑁 = 𝑞 ∗ (𝑞 𝑁 ) = 44 − 1


 2 2 1 2

Solving the system, we obtain 𝑞 1𝑁 = 16, 𝑞 2𝑁 = 36.

(c) See Figure 19 below for the illustration.

12. (a) The total revenue at output level 𝑄 is 𝑇 𝑅(𝑄) = 𝑝(𝑄)𝑄 − 𝐶 (𝑄) = (24 − 𝑄)𝑄, and so the
marginal revenue 𝑀 𝑅(𝑄) = d𝑇 𝑅(𝑄)/d𝑄 = 24 − 2𝑄. Since the monopoly output level 𝑄 ∗
equates the marginal cost and marginal revenue, we have

24 − 2𝑄 ∗ = 𝑀 𝑅(𝑄 ∗ ) = 𝑀𝐶 (𝑄 ∗ ) = 0,

from which we solve 𝑄 ∗ = 12.

131
 𝑞1
88

𝑞 ∗2

(𝑞 1𝑁 , 𝑞 2𝑁 )

25
𝑞 ∗1
𝑞2
𝑂 44 100

Figure 19: An Illustration of the Cournot Equilibrium

(b) When the total output is equal to the monopoly level 𝑄 ∗ = 12, the market price 𝑝 ∗ = 24−𝑄 ∗ = 12.
Since each firm produces one half of 𝑄 ∗ , the profit of each firm is
∗
 ∗
∗ 𝑄 𝑄
𝜋𝑀 = 𝑝 · −𝐶 = 12 × 6 − 0 = 72.
2 2

(c) When firm 1’s output level is 𝑄 1 , firm 2’s profit at 𝑄 2 is

𝜋2 (𝑄 2 ; 𝑄 1 ) = (24 − 𝑄 1 − 𝑄 2 )𝑄 2 ,

and the best reaction 𝑄 ∗2 solves

𝜕𝜋2 (𝑄 2 ; 𝑄 1 )
= 24 − 𝑄 1 − 2𝑄 ∗2 = 0.
𝜕𝑄 2 𝑄2 =𝑄2∗

As a result, we have
𝑄1
𝑄 ∗2 (𝑄 1 ) = 12 −
.
2
Therefore, when firm 1 produces the collusive level of output (i.e., 𝑄 1 = 6), the best reaction of
firm 2 will be 𝑄 ∗2 (6) = 9 ≠ 6, which implies that firm 1 has an incentive to deviate from the
collusive output level. By symmetry, our argument also applies to firm 1, and so each firm has
an incentive to unilaterally deviate from the agreement.
(d) Let (𝑄 1𝑁 , 𝑄 2𝑁 ) be the Cournot-Nash equilibrium. Then it characterized by


 𝑄𝑁
𝑁 = 𝑄 ∗ (𝑄 𝑁 ) = 12 − 1
 𝑄
 1
 1 2
 2
,
 𝑄𝑁
 𝑄 𝑁 = 𝑄 ∗ (𝑄 𝑁 ) = 12 − 1


 2 2 1 2

132
 from which we solve 𝑄 1𝑁 = 𝑄 2𝑁 = 8.
(e) The fine 𝑝 ∗ must be chosen so that a firm will find it optimal to produce 6 if the other firm abides
by the agreement. Given that the other firm’s output level is 6, the most beneficial deviation is
𝑄 ∗ (6) = 9 (see (c)), and the corresponding profit is

𝜋(9; 6) = (24 − 6 − 9) × 9 = 81.

Thus, the benefit due to deviation is at most 81 − 72 = 9 (see (b)). Therefore, to deter deviation,
the fine should be set so that it suffices to counteract the benefit from deviation, that is, 𝑝 ∗ > 9.
As a result, the minimum punishment is 𝑝 ∗ = 9.

13. (a) Let us calculate Defendo’s monopolistic profits with the two types of technology, respectively.
Using the inverse demand function, the total revenue 𝑇 𝑅(𝑄) = 𝑝(𝑄) · 𝑄 = (20 − 𝑄)𝑄, and so
the marginal revenue
d𝑇 𝑅(𝑄)
𝑀 𝑅(𝑄) = = 20 − 2𝑄.
d𝑄
If Defendo adopts technology 𝐴, then the marginal cost 𝑀𝐶 𝐴 = 8. The optimal output level 𝑄 ∗𝐴
equates 𝑀 𝑅 and 𝑀𝐶 𝐴, that is,
20 − 2𝑄 ∗𝐴 = 8,

from which we solve 𝑄 ∗𝐴 = 6, and the corresponding optimal monopolistic price 𝑝 ∗𝐴 = 20 − 𝑄 ∗𝐴 =

14. Thus, the profit

𝜋 𝐴 = 𝑝 ∗𝐴𝑄 ∗𝐴 − 𝑇𝐶 𝐴 (𝑄 ∗𝐴) = 6 × 14 − (10 + 8 × 6) = 26.

If Defendo, however, adopts technology 𝐵, then the marginal cost 𝑀𝐶 𝐵 = 2. The optimal output
level 𝑄 ∗𝐵 solves
20 − 2𝑄 ∗𝐵 = 2,

from which 𝑄 ∗𝐵 = 9, and the corresponding optimal monopolistic price 𝑝 ∗𝐵 = 20 − 𝑄 ∗𝐵 = 11.

Thus, the profit

𝜋 𝐵 = 𝑝 ∗𝐵 𝑄 ∗𝐵 − 𝑇𝐶 𝐵 (𝑄 ∗𝐵 ) = 9 × 11 − (60 + 2 × 9) = 21.

Since 𝜋 𝐵 < 𝜋 𝐴, Defendo would adopt technology 𝐴, and the profit would be 26.
(b) (i) If Defendo adopts technology 𝐴 while Offendo enters the market, then the marginal cost for
both firm equals 8, and so Defendo’s profit at the strategy profile (𝑞 𝐷 , 𝑞 𝑂 ) is

𝜋 𝐷 (𝑞 𝐷 , 𝑞 𝑂 ) = [20 − (𝑞 𝐷 + 𝑞 𝑂 )]𝑞 𝐷 − (10 + 8𝑞 𝐷 ),

where 𝑞 𝐷 is the output of Defendo, and 𝑞 𝑂 is the output of Offendo. The best reaction of

133
Defendo 𝑞 ∗𝐷 (𝑞 𝑂 ) solves

𝜕𝜋 𝐷 (𝑞 𝐷 , 𝑞 𝑂 )
∗ (𝑞 )
= 20 − 2𝑞 ∗𝐷 − 𝑞 𝑂 − 8 = 0,
𝜕𝑞 𝐷 𝑞𝐷 =𝑞𝐷 𝑂

that is,
12 − 𝑞 𝑂
𝑞 ∗𝐷 = .
2
Similarly, the best response of Offendo is

∗ 12 − 𝑞 𝐷
𝑞𝑂 = .
2
𝑁 , 𝑞 𝑁 )) solves
The Cournot-Nash equilibrium (𝑞 𝐷 𝑂


 𝑁
12 − 𝑞 𝑂
𝑁
 𝑞𝐷 =


 2

 𝑁 =
12 − 𝑞 𝐷
𝑁
 𝑞𝑂

 2

from which we solve (𝑞 𝐷 𝑁 , 𝑞 𝑁 ) = (4, 4). Thus, the market price in the Cournot-Nash equilibrium
𝑂
is 𝑝 (𝑖) = 20 − (𝑞 𝐷
𝑁 + 𝑞 𝑁 ) = 12. The corresponding profits are
𝑂

(𝑖) (𝑖)
𝜋𝐷 = 𝜋𝑂 = 4 × 12 − (10 + 8 × 4) = 6.

(ii) If Defendo adopts technology 𝐵 while Offendo enters the market, the best reaction for
Offendo 𝑞 𝑂∗ is the same as in (i), but due to the change of marginal cost, Defendo’s best reaction

will change accordingly. Concretely, now Defendo’s profit at (𝑞 𝐷 , 𝑞 𝑂 ) is

𝜋 𝐷 (𝑞 𝐷 , 𝑞 𝑂 ) = [20 − (𝑞 𝐷 + 𝑞 𝑂 )]𝑞 𝐷 − (60 + 2𝑞 𝐷 ),

and so Defendo’s best reaction 𝑞 ∗𝐷 (𝑞 𝑂 ) solves

𝜕𝜋 𝐷 (𝑞 𝐷 , 𝑞 𝑂 )
= 20 − 2𝑞 ∗𝐷 − 𝑞 𝑂 − 2 = 0,
𝜕𝑞 𝐷

and so
18 − 𝑞 𝑂
𝑞 ∗𝐷 (𝑞 𝑂 ) = .
2
The Cournot-Nash equilibrium (𝑞 𝐷
𝑁 , 𝑞 𝑁 ) solves the system
𝑂



 𝑁 =
18 − 𝑞 𝑂
𝑁
 𝑞
 𝐷
 2

 𝑁 =
12 − 𝑞 𝐷
𝑁
 𝑞𝑂

 2

134
 𝑁 , 𝑞 𝑁 ) = (8, 2). The market price 𝑝 (𝑖𝑖) = 20 − (𝑞 𝑁 + 𝑞 𝑁 ) = 10 and so the
and so (𝑞 𝐷 𝑂 𝐷 𝑂
corresponding profits are
(𝑖𝑖) (𝑖𝑖)
𝜋𝐷 = 8 × 10 − (60 + 2 × 8) = 4, 𝜋𝑂 = 2 × 10 − (10 + 8 × 2) = −6.

To summarize, in (i), the firms’ profits are (6, 6), and in (ii) they are (4, −6).
(c) The game tree is given below, where 𝐷 represents Defendo, 𝑂 represents Offendo, 𝐴 and 𝐵
stand for technology 𝐴 and technology 𝐵, respectively. For each pair at a terminal node, the first
component is the payoff to Defendo and the second to Offendo.

𝐷
𝐴 𝐵
𝑂 𝑂

Enter Not Enter Enter Not Enter

(6, 6) (26, 0) (4, −6) (21, 0)

Figure 20: Extensive Form of the Game

(d) It is easy to see from Figure (20) that Offendo’s optimal reaction is “Not enter” if Defendo
chooses 𝐵, and is “Enter” if Defendo chooses 𝐴. Provided the best reaction of Offendo, Defendo
will obtain a profit of 6 if it chooses technology 𝐴 and that of 21 if it chooses technology 𝐵.
Thus, the optimal chose of Defendo is technology 𝐵. Thus, there is a unique Nash equilibrium
that will survive backward induction: Defendo chooses technology 𝐵, and Offendo chooses to
enter if Defendo chooses technology 𝐴 and not to enter if Defendo chooses technology 𝐵.
(e) Without the threat of entry, as we have computed in (a), Defendo will adopt technology 𝐴 and
the monopolistic equilibrium ( 𝑝 ∗𝐴, 𝑄 ∗𝐴) = (14, 6). Thus, the total surplus equals
∫ 𝑄 ∗𝐴 ∫ 𝑄 ∗𝐴 ∫ 6
[ 𝑝(𝑞) − 𝑀𝐶 (𝑞)] d𝑞 = (20 − 𝑞 − 8)d𝑞 = (12 − 𝑞) d𝑞 = 54.
0 0 0

When there is the threat of entry, Defendo will adopt technology 𝐵 and the monopolistic
equilibrium ( 𝑝 ∗𝐵 , 𝑄 ∗𝐵 ) = (11, 9), and so the total surplus equals
∫ 𝑄 ∗𝐵 ∫ 𝑄 ∗𝐵 ∫ 9
[ 𝑝(𝑞) − 𝑀𝐶 (𝑞)] d𝑞 = (20 − 𝑞 − 2)d𝑞 = (18 − 𝑞) d𝑞 = 121.5.
0 0 0

Thus, the threat of entry improves the social welfare, since it incentivizes the incumbent to adopt
more efficient technology to deter entry, which, at the same time, reduces the marginal cost and
improves the social welfare.

135
14. (a) Let us start with the case without the license fee. Given the output level of the first mover 𝑄 1 ,
the second mover’s profit at output 𝑄 2 is

𝜋2 (𝑄 2 ; 𝑄 1 ) = (100 − 𝑄 1 − 𝑄 2 )𝑄 2 ,

and the best reaction 𝑄 ∗2 solves

𝜕𝜋2 (𝑄 2 ; 𝑄 1 )
= 100 − 𝑄 1 − 2𝑄 ∗2 = 0.
𝜕𝑄 2 𝑄2 =𝑄2∗

So we have
𝑄1
𝑄 ∗2 (𝑄 1 ) = 50 − .
2
Thus, the first mover’s problem is
  
𝑄1
max 100 − 𝑄 1 − 50 − 𝑄1.
𝑄1 2

The optimal output level 𝑄 ∗1 solves the first-order condition

50 − 𝑄 ∗1 = 0,

and so 𝑄 ∗1 = 50. In the Stackelberg equilibrium, the profit of the first mover is thus

𝜋1 = [100 − 𝑄 ∗1 − 𝑄 2 (𝑄 ∗1 )]𝑄 ∗1 = (100 − 50 − 25) × 50 = 1250.

So you are willing to pay at most 1250 for one of the license, given that you will be the first
mover.
(b) The second mover’s profit in the Stackelberg equilibrium is

𝜋2 = 𝑝 ∗ 𝑄 ∗2 (𝑄 ∗1 ) = 25 × 25 = 625.

Thus, your rival is willing to pay at most 625 for the license.

15. (a) If firm 1 does not invest, then firm 1’s profit at the output profile (𝑄 1 , 𝑄 2 ) is

𝜋1 (𝑄 1 ; 𝑄 2 ) = (3 − 𝑄 1 − 𝑄 2 )𝑄 1 − 𝑄 1 .

The best reaction 𝑄 ∗1 (𝑄 2 ) solves

𝜕𝜋1 (𝑄 1 ; 𝑄 2 )
= 2 − 𝑄 2 − 2𝑄 ∗1 = 0.
𝜕𝑄 1 𝑄1 =𝑄1∗

Therefore, 𝑄 ∗1 (𝑄 2 ) = 1 − 0.5𝑄 2 . By symmetry, 𝑄 ∗2 (𝑄 1 ) = 1 − 0.5𝑄 1 . Thus, the Cournot

136
 equilibrium (𝑄 1𝑁 , 𝑄 2𝑁 ) is characterized by the system
(
𝑄 1𝑁 = 1 − 0.5𝑄 2𝑁
,
𝑄 2𝑁 = 1 − 0.5𝑄 1𝑁

from which we solve 𝑄 1𝑁 = 𝑄 2𝑁 = 2/3.

(b) If firm 1 invests, then firm 1’s profit at the output profile (𝑄 1 , 𝑄 2 ) is

𝜋1 (𝑄 1 ; 𝑄 2 ) = (3 − 𝑄 1 − 𝑄 2 )𝑄 1 − 𝐹.

The best reaction 𝑄 ∗1 (𝑄 2 ) solves

𝜕𝜋1 (𝑄 1 ; 𝑄 2 )
= 3 − 𝑄 2 − 2𝑄 ∗1 = 0.
𝜕𝑄 1 𝑄1 =𝑄1∗

Therefore, 𝑄 ∗1 (𝑄 2 ) = 1.5 − 0.5𝑄 2 . By our result in (a), the best reaction of firm 2 is still
𝑄 ∗2 (𝑄 1 ) = 1 − 0.5𝑄 1 . Thus, the Cournot equilibrium (𝑄 1𝑁 , 𝑄 2𝑁 ) is characterized by the system
(
𝑄 1𝑁 = 1.5 − 0.5𝑄 2𝑁
,
𝑄 2𝑁 = 1 − 0.5𝑄 1𝑁

from which we solve 𝑄 1𝑁 = 4/3, 𝑄 2𝑁 = 1/3.

(c) To answer this question, we need to compare firm 1’s profit levels with and without the investment.
When firm 1 does not invest,
 
𝑁𝐼 2 2 2 2 4
𝜋1 = 3 − − · − = ;
3 3 3 3 9

when firm 1 invests,  

4 1 4 16
𝜋1𝐼 = 3− − · −𝐹 = − 𝐹.
3 3 3 9
Making investment is profitable if and only if 𝜋1𝐼 > 𝜋 𝐼𝑁 𝐼 , that is,

16 4
−𝐹 > ,
9 9
which implies that 𝐹 6 4/3. Thus, firm 1 will invest if and only if the lump-sum cost is no
greater than 4/3.
(d) In Cournot competition, the output of one firm is decreasing in that of the other. Since the
investment can help reduce marginal cost, firm 2 should rationally expect that firm 1’s output
will be higher than without the investment, and so firm 2’s output level will decrease, so will its
profit.

137
16. (a) For each 𝑞 𝐴 > 0, by choosing 𝑞 𝐵 > 0, firm 𝐵’s profit
(
[50 − (𝑞 𝐴 + 𝑞 𝐵 )]𝑞 𝐵 − 4𝑞 𝐵 , if 𝑞 𝐴 + 𝑞 𝐵 6 50
𝜋 𝐵 (𝑞 𝐵 ; 𝑞 𝐴) =
−4𝑞 𝐵 , otherwise
(
[46 − (𝑞 𝐴 + 𝑞 𝐵 )]𝑞 𝐵 , if 𝑞 𝐴 + 𝑞 𝐵 6 50
=
−4𝑞 𝐵 , otherwise
Clearly, the best reaction of firm 𝐵 is 0 when 𝑞 𝐴 > 46. When 𝑞 𝐴 < 46, firm 𝐵’s best reaction
𝑞 ∗𝐵 (𝑞 𝐴) satisfies the first-order condition

𝜕𝜋 𝐵 (𝑞 𝐵 ; 𝑞 𝐴)
∗ (𝑞 )
= 46 − [𝑞 𝐴 + 𝑞 ∗𝐵 (𝑞 𝐴)] − 𝑞 ∗𝐵 (𝑞 𝐴) = 0,
𝜕𝑞 𝐴 𝑞𝐵 =𝑞𝐵 𝐴

from which we solve

46 − 𝑞 𝐴
𝑞 ∗𝐵 (𝑞 𝐴) =.
2
Thus, when 𝑞 𝐴 = 30, the best reaction of firm 𝐵 is (46 − 30)/2 = 8.
(b) Since firm 𝐵’s best reaction is 0 when 𝑞 𝐴 > 46, it is clear that firm 𝐴 should produce at least 46
in order to drive firm 𝐵 out of the market.
(c) Taking into account the best reaction of firm 𝐵, by choosing 𝑞 𝐴 > 0, firm 𝐴’s profit

𝜋 𝐴 (𝑞 𝐴) = 50 − [𝑞 𝐴 + 𝑞 ∗𝐵 (𝑞 𝐴)] 𝑞 𝐴 − 2𝑞 𝐴

  
46 − 𝑞 𝐴
= 48 − 𝑞 𝐴 + 𝑞𝐴
2
= (25 − 0.5𝑞 𝐴)𝑞 𝐴 .

The maximum of the first component is achieved at 𝑞 ∗𝐴 = 25, at which the value equals 312.5 > 0.
Thus, in the Stackelberg equilibrium, firm 𝐴’s output level equals 25, and firm 𝐵’s output level
equals (using the best reaction function derived in (a)) 0.5(46 − 25) = 10.5. As a result, the
market price equals 50 − [𝑞 ∗𝐴 + 𝑞 ∗𝐵 (𝑞 ∗𝐴)] = 50 − (25 + 10.5) = 14.5.

17. (a) In the Bertrand equillibrium, given that the two firms have the same marginal cost, there must be
one firm which sets it price equal to the marginal cost, since otherwise the other firm will have
incentive to slightly undercut its price to win the whole market share. Thus, the market price at
the Bertrand equilibrium is 36. Notice that it is not necessarily true that both firms will set the
price at 36, since a firm will be indifferent amongst t all prices no lower than 36 (but strictly
prefers a price greater than or equal to 36) once the other firm sets its price at 36.
(b) For firm 2 to become a monopolist, its marginal cost must be low enough so that the corresponding
monopolistic price is lower than the (constant) marginal cost of firm 1. Let 𝑀𝐶 be a level of
marginal cost for firm 2 that meets this requirement. When firm 2 is a monopolist, the marginal

138
 revenue equals 60 − 2𝑄, and so the corresponding output level 𝑄 ∗ solves 60 − 2𝑄 ∗ = 𝑀𝐶, which
implies that 𝑄 ∗ = (60 − 𝑀𝐶)/2, and the monopolistic price 𝑝 ∗ = 60 − 𝑄 ∗ = (60 + 𝑀𝐶)/2. To
meet the requirement, we must have 𝑝 ∗ 6 35, and so 𝑀𝐶 6 10. As a result, firm 2’s marginal
cost must be no higher than 10.

18. (a) In the Bertrand equilibrium, each firm charges a price that is equal to the marginal cost. Thus,
the price in the second stage here must equal 𝑐 + 𝑤 𝐴 + 𝑤 𝐵 .
(b) Since the total output in the Bertrand equilibrium equals 1 − 𝑐 − 𝑤 𝐴 − 𝑤 𝐵 , firm 𝑖’s problem is
(𝑖 = 𝐴, 𝐵)
max (1 − 𝑐 − 𝑤 𝐴 − 𝑤 𝐵 )𝑤 𝑖 .
𝑤𝑖

The best reaction 𝑤 ∗𝑖 solves the first-order condition

1 − 𝑐 − 𝑤 𝑗 − 2𝑤 ∗𝑖 = 0, 𝑗 ≠ 𝑖,

and so
1 − 𝑐 − 𝑤𝑗
𝑤 ∗𝑖 (𝑤 𝑗 ) = .
2
Thus, the Nash equilibrium (𝑤 𝑁
𝐴 , 𝑤 𝐵 ) solves
𝑁


 𝑁 =
1 − 𝑐 − 𝑤 𝐵𝑁

 𝑤
 𝐴
 2

 1 − 𝑐 − 𝑤𝑁
𝐴
 𝑤𝑁 =

 𝐵 2
and so
1−𝑐
𝑤𝑁 𝑁
𝐴 = 𝑤𝐵 = .
3
In the Nash equilibrium, each firm obtains a profit

(1 − 𝑐) 2
𝜋 = (1 − 𝑐 − 𝑤 𝑁 𝑁 𝑁
𝐴 − 𝑤 𝐵 )𝑤 𝐴 = .
9

(c) If the (joint) royalty fee is 𝑝 (now only the joint royalty fee matters), then the second stage market
price will be 𝑐 + 𝑝, and so the total quantity of production will be 1 − 𝑐 − 𝑝. Thus, the patent
pool’s problem is
max (1 − 𝑐 − 𝑝) 𝑝,
𝑝

and it is easy to obtain the optimal royalty fee

1−𝑐
𝑝∗ = .
2

139
 The patent pool’s profit is

1 − 𝑐 1 − 𝑐 (1 − 𝑐) 2
 
0
𝜋 = 1−𝑐− = .
2 2 4

(d) Consumers benefit from forming the patent pool since now the price is lower (originally the
price is 𝑐 + 𝑤 𝑁
𝐴 + 𝑤 𝐵 = 𝑐 + 2(1 − 𝑐)/3 while the price with patent pool equals 𝑐 + (1 − 𝑐)/2).
𝑁

Patentholders will benefit from forming the patent pool under some “fair” plan to split the total
profit. For example, if both firms agree to share the total profit equally, then each firm’s profit
will be 𝜋 0/2 = (1 − 𝑐) 2 /8, which is higher than the profit each obtains without forming the
patent pool (i.e., (1 − 𝑐) 2 /9, see (b)).

19. (a) The two goods are substitutes, since the demand of either good is increasing in the price of the
other good.
(b) Firm 1’s problem is

max (2 − 2𝑝 1 + 𝑝 2 ) 𝑝 1 − (2 − 2𝑝 1 + 𝑝 2 ) − (2 − 2𝑝 1 + 𝑝 2 ) 2 .
𝑝1

The best reaction 𝑝 ∗1 solves the first-order condition

12 + 5𝑝 2 − 12𝑝 ∗1 = 0,

from which we solve

5
𝑝 ∗1 ( 𝑝 2 ) = 1 + 𝑝2 .
12
Firm 2’s problem is

max (2 − 2𝑝 2 + 𝑝 1 ) 𝑝 2 − 𝛼(2 − 2𝑝 2 + 𝑝 1 ) − (2 − 2𝑝 2 + 𝑝 1 ) 2 .
𝑝2

Similarly, the best reaction 𝑝 ∗2 solves the first-order condition

10 + 2𝛼 + 5𝑝 1 − 12𝑝 ∗2 = 0,

from which we solve

5+𝛼 5
𝑝 ∗2 ( 𝑝 1 ) = + 𝑝1 .
6 12
(c) When 𝛼 = 1, the Nash equilibrium ( 𝑝 1𝑁 , 𝑝 2𝑁 ) is characterized by the system

5 𝑁
 𝑝 1 = 1 + 12 𝑝 2

 𝑁


.
 5 𝑁
 𝑝𝑁 = 1 +
 𝑝
 2 12 1

Solving the system, we get 𝑝 1𝑁 = 𝑝 2𝑁 = 12/7.

140
 (d) The value of 𝛼 does not affect firm 1’s best reaction, but firm 2’s best reaction is increasing in
𝛼, as can be seen from our result in (b). In general, the Nash equilibrium ( 𝑝 1𝑁 , 𝑝 2𝑁 ) solves the
system
 5 𝑁
 𝑝 1 = 1 + 12 𝑝 2

 𝑁


,
 5+𝛼 5 𝑁

 𝑁
 𝑝2 = + 𝑝
 6 12 1
which yields that
194 + 10𝛼 180 + 24𝛼
𝑝 1𝑁 = , 𝑝 2𝑁 = .
119 119
This implies that the equilibrium prices are increasing in 𝛼.
Alternatively, notice that each firm’s optimal price is increasing in the other firm’s price. Thus,
as 𝛼 increases, firm 1 will rationally expect that firm 2 will charge a higher price (other things
being equal), which gives firm 1 an incentive to charge a higher price. This will, in turn,
incentivize firm 2 to charge an even higher price. Such a reaction circle will clearly generate
higher equilibrium prices.

20. (a) No. This is because to each company, the unique best reaction against “Produce” is “Not
produce”, while the unique best reaction against “Not produce” is “Produce”.
(b) Employing our observation in (a), it is clear that there are two pure-strategy Nash equilibria:
(Produce, Not produce) and (Not produce, Produce).
(c) Let the mixed-strategy Nash equilibrium be denoted by ( 𝑝, 𝑞), where 𝑝 ∈ (0, 1) is the probability
that Boeing produces, and 𝑞 ∈ (0, 1) is the probability that Airbus produces. To make each
company indifferent between “Produce” and “Not produce”, we must have
(
−5𝑞 + 100(1 − 𝑞) = 0
−5𝑝 + 100(1 − 𝑝) = 0

from which we solve 𝑝 = 𝑞 = 20/21 ≈ 0.95. That is, in the mixed-strategy Nash equilibrium,
each firm will choose to produce with probability 20/21.
(d) Now the payoff matrix becomes
Airbus
Produce Not produce
Produce 20, −5 100, 0
Boeing
Not produce 0, 100 0, 0

That is, when both companies choose to produce, the payoff to Boeing will be 25 − 5 = 20
rather than −5. Notice that in this case “Produce” is a strictly dominant strategy for Boeing,
and so in all Nash equilibria Boeing will play “Produce”. As a result, to react optimally, Airbus
will choose “Not produce”. Hence, now the game admits a unique Nash equilibrium, in which
Boeing chooses to produce while Airbus chooses not to produce.

141
 (e) The game tree is given in Figure 21 below, where 𝐴 means Airbus, 𝐵 means Boeing, 𝑃 stands
for Produce, and 𝑁 stands for Not produce.

𝐴
𝑃 𝑁
𝐵 𝐵

𝑃 𝑁 𝑃 𝑁

(−5, −5) (100, 0) (0, 100) (0, 0)

Figure 21: Extensive Form of the Game

(f) Employing the game tree, it is easy to see that in the second stage, Boeing will choose to produce
if Airbus has chosen not to produce, and vice versa. Provided this, in the first stage, Airbus will
clearly choose to produce. This constitutes the unique Nash equilibrium that survive backward
induction. As we can see, the first mover has an advantage since it can occupy an advantageous
position that renders “Produce” unprofitable to the other firm.

21. (a) Given the total output of other firms, firm 𝑖’s problem is

max (120 − 𝑄 −𝑖 − 𝑄 𝑖 )𝑄 𝑖 − 30𝑄 𝑖 .

𝑄𝑖

The best reaction 𝑄 ∗𝑖 solves the first-order condition

90 − 𝑄 −𝑖 − 2𝑄 ∗𝑖 = 0,

which yields the best reaction

𝑄 −𝑖
𝑄 ∗𝑖 (𝑄 −𝑖 ) = 45 − .
2
(b) Let (𝑄 1𝑁 , 𝑄 2𝑁 , · · · , 𝑄 𝑛𝑁 ) be the Cournot equilibrium. Notice that
! 𝑁 !
𝑁 𝑄 −(𝑖+1)
𝑄 −𝑖 1
𝑄 𝑖𝑁 𝑁
− 𝑄 𝑖+1 = 𝑄 ∗𝑖 (𝑄 −𝑖
𝑁
) − 𝑄 ∗𝑖+1 (𝑄 𝑖+1
𝑁
) = 45 − − 45 − = (𝑄 𝑖𝑁 − 𝑄 𝑖+1
𝑁
),
2 2 2

𝑁 . Since 𝑖 is arbitrary, this tells us that 𝑄 𝑁 = 𝑄 𝑁 = · · · = 𝑄 𝑁 .

which then implies that 𝑄 𝑖𝑁 = 𝑄 𝑖+1 1 2 𝑛

(c) 𝑄 ∗ satisfies the condition that

(𝑛 − 1)𝑄 ∗
𝑄 ∗ = 45 − ,
2
and so
90
𝑄∗ = .
𝑛+1

142
 The total output
90𝑛
𝑄𝑇𝑁 = 𝑛𝑄 ∗ = ,
𝑛+1
and the market price
120 + 30𝑛
𝑝 𝑁 = 120 − 𝑄𝑇𝑁 = .
𝑛+1
(d) As 𝑛 gets large, it is clear that the market price will get closer and closer to the limit

120 + 30𝑛
lim = 30.
𝑛→∞ 𝑛+1
Note that 30 is the marginal cost of each firm. Thus, as 𝑛 gets larger, the competition becomes
more fierce, which will eventually drive the market price down to the competitive level (recall
that in a competitive market the price equals the marginal cost).

22. (a) The monopolist’s total revenue 𝑇 𝑅(𝑄) = 𝑝(𝑄)𝑄 − 𝑐𝑄 = (1 − 𝑄)𝑄, and the marginal revenue
𝑀 𝑅(𝑄) = 1 − 2𝑄. The optimal output level 𝑄 ∗ solves 𝑀𝐶 (𝑄 ∗ ) = 𝑀 𝑅(𝑄 ∗ ), that is,

𝑐 = 1 − 2𝑄 ∗ ,

and so
1−𝑐
𝑄∗ = .
2
Thus, the monopolist’s profit

1 − 𝑐 (1 − 𝑐) 2
 
𝑀 1−𝑐 1−𝑐
𝜋 = 1− −𝑐· = .
2 2 2 4

(b) The patent holder’s problem is

max (1 − 𝑄 𝐻 − 𝑄 𝐼 )𝑄 𝐻 − 𝑐𝑄 𝐻 ,
𝑄𝐻

where 𝑄 𝐼 is the infringer’s output level. The best reaction 𝑄 ∗𝐻 solves the first-order condition

1 − 𝑐 − 𝑄 𝐼 − 2𝑄 ∗𝐻 = 0,

and so 𝑄 ∗𝐻 = (1 − 𝑐 − 𝑄 𝐼 )/2. Since the infringer’s problem is the same as that of the patent
holder if the former is not punished, it follows from symmetry that 𝑄 ∗𝐼 = (1 − 𝑐 − 𝑄 𝐻 )/2.
Therefore, the Cournot equilibrium (𝑄 𝐻𝑁 , 𝑄 𝑁 ) is characterized by the system
𝐼


 𝑁 =
1 − 𝑐 − 𝑄 𝐼𝑁

 𝑄
 𝐻
 2
,

 1 − 𝑐 − 𝑄𝐻
𝑁
 𝑄 𝐼𝑁 =

 2

143
 from which we solve
𝑁 1−𝑐
𝑄𝐻 = 𝑄 𝐿𝑁 = .
3
each obtains a profit

1 − 𝑐 (1 − 𝑐) 2
 
𝐶 1−𝑐 1−𝑐 1−𝑐
𝜋 = 1− − −𝑐· = .
3 3 3 3 9

(c) The patentholder’s lost profit is the difference between its monopolistic profit and the profit it
receives in the Cournot equilibrium. Thus, the lost profit

(1 − 𝑐) 2 (1 − 𝑐) 2 5(1 − 𝑐) 2
𝜋 𝐿 = 𝜋 𝑀 − 𝜋𝐶 = − = .
4 9 36
The infringer’s unjust enrichment is simply the profit it obtains in the Cournot equilibrium, that
is,
(1 − 𝑐) 2
𝜋𝑈 𝐸 = 𝜋𝐶 = .
9
(d) Since 𝜋 𝐿 > 𝜋𝑈 𝐸 , it is clear that the patentholder prefers to get back the lost profit.
(e) The lost-profit penalty does deter infringement, as the profit of the infringer after paying out the
patentholder’s lost profit equals

(1 − 𝑐) 2
𝜋𝐶 − 𝜋 𝐿 = − < 0.
36
Thus, infringing is even worse than not entering the market.

144
 7

Information Economics and

Mechanism Design

7 Information Economics and Mechanism Design

145
Exercise 1. There are two groups of people, each with a Bernoulli utility function given by

𝑢(𝑤) = 𝑤 0.5 ,

where 𝑤 denotes the final wealth level, and the initial wealth level for each individual is 144. Each member
of group 1 faces a loss of 44 with probability 0.5; each member of group 2 faces the same loss with
probability 0.1.
(a) What is the most a member of each group would be willing to pay to insure against this loss (i.e., in
all cases one’s final wealth equals the initial wealth)?
(b) If it is impossible for outsiders to discover which individuals belong to which group, how large a
share of the potential client pool can the members of group 1 be before it becomes impossible for
a private company with a zero-profit constraint to provide insurance for the members of group 2?
Explain.

Exercise 2. On a second-hand car market, there are two types of cars: High quality and low quality. By
statistics, 30% of used cars are of high quality and 70% are of low quality. To each buyer, a high quality
car is worth $30, 000 while a low quality car is worth $20, 000, and in each case the value of a used car to
a seller is 80% of that to a buyer. Each seller knows the quality of his/her car, but buyers do not know the
quality of cars. Buyers and sellers are all risk neutral. Each seller is a monopolist of his/her own car and
can always sell it at the maximum price a buyer is willing to pay.
(a) Without considering the motive of sellers, if buyers trust the statistics, what is the maximum price
each buyer is willing to pay for a used car?
(b) Eventually, taking into account the motives of both sellers and buyers, what will be the price for a
used car on the market? Will high-quality cars be crowded out from the market? Explain.

Exercise 3. Imagine that you are the only used-car dealer in a town. All used-car owners in the town must
either sell their used cars to you or keep them. At the time you buy a used car, you are not able to tell
whether it is a good one or a lemon. However, between the time you buy the car and the time you resell it,
you will find out the real quality of the car. You will resell all of the cars you buy, but you are required to
truthfully reveal the quality of the cars to your buyers when making a deal. A lemon can be resold for
$500 while a good car can be resold for $3500. Your profits equal the difference between the revenue you
get from reselling cars and the money you pay for cars. You must post a single price at which you are
willing to purchase each used car that is brought to you. Owners will bring their cars to you if the price
you post is higher than their reservation prices. The reservation value of a good car is $1600 while that of
a lemon is $100. Suppose in your town there are 6 used car that are good and 6 lemons.
(a) What is the lowest price at which lemon-owners would sell their cars? What is your profit if you
offer this price?
(b) What is the lowest price at which all used-car owners will sell their cars? What is your profit if you
offer this price?

146
 (c) What is the profit-maximizing price you should offer to used-car owners?
(d) Suppose there is a service that can reveal the true quality of a car when you buy it. What is the
highest price you would be willing to pay for this service?

Exercise 4. A monopolistic bank has 70 units of funds available for lending. There are 100 potential
risk-neutral borrowers in the credit market. They each needs 1 unit of funding for investment. 90 of them
have safe projects to invest in, which will generate a gross return of 1.5 for sure. The rest borrowers only
have access to risky projects, which will generate a high gross return of 4 with probability 0.5 and a lower
gross return of 0 with probability 0.5. The bank, however, is not able to tell whether a borrower has a safe
or a risky project. The bank can only set a uniform gross interest rate 𝑅. A borrower has to repay 𝑅 to the
bank if the project is successful; otherwise a borrower will default and repay nothing to the bank. Whether
a project is successful or not is publicly observable and verifiable. Each borrower receives a zero utility if
he/she does not borrow. For simplicity, we assume that 𝑅 must be a multiple of 0.05 (i.e., 𝑅 = 1 is allowed
since 1 = 0.05 × 20, but 𝑅 = 1.03 is not allowed).

(a) If the bank sets 𝑅 = 1.6, what is the quantity of demand of borrowing (in terms of units of funds) in
the market? What if the bank sets 𝑅 = 1.3?
(b) What is the expected profit-maximizing 𝑅?

Now suppose that the bank can offer two gross interest rates 𝑅𝑟 and 𝑅𝑠 , with 𝑅𝑟 > 𝑅𝑠 . Borrowers are
allowed to apply for at most one of the two interest rates. A borrower will be granted a unit of fund for
sure if he/she applies for 𝑅𝑟 , but the fund will be granted with probability 0.6 if one applies for 𝑅𝑠 .

(c) Derive the conditions on 𝑅𝑟 and 𝑅𝑠 that guarantee that borrowers with safe projects will apply for
𝑅𝑠 and those with risky projects will apply for 𝑅𝑟 (for simplicity, we assume that a borrower will
choose to borrow and invest if he/she is indifferent between doing and not doing so. ). Graph all
(𝑅𝑟 , 𝑅𝑠 )’s that satisfy the conditions in axes where 𝑅𝑠 is on the horizontal axis.

Exercise 5. Once upon a time a game was played between a princess and a frog. The frog could say he
was a “prince” or a “frog”. Upon the word the frog said, the princess could either kiss the frog or eat the
frog. If the frog was actually a prince, then he would turn into a prince after being kissed by the princess.
It was well-known that 10% of the frogs in the kingdom were actually princes who had had a spell cast
upon them. All frogs could say the word “frog”; frogs who were actually princes could also say the word
“prince”; frogs who were not princes were able to say “prince” only after taking costly lessons in elocution,
but whether a frog had taken lessons was not observable. The payoff to a frog (a real one or a prince)
equals 𝑣 − 𝑐, where
( (
10, if the princess chooses to kiss 𝑥, if a frog chooses to take lessons
𝑣= , 𝑐= .
0, if the princess chooses to eat 0, otherwise

147
The princess receives a utility of 𝑢, where

5, if she chooses to eat







𝑢 = −10, if she kisses a real frog .


 100,
 if she kisses a frog that is actually a prince

We first assume that 𝑥 = 8.

(a) Suppose the princess believed that real frogs would never take lessons and say “prince”. What is her
best strategy, provided that all frogs play optimally?
(b) Is the princess’s belief in (a) correct (i.e., consistent with the actual play of the game)?
(c) Suppose the strategy of the princess is to kiss a frog if it says the word “prince” (and eat it otherwise).
What is the probability that she turns out to have kissed a real frog if frogs are playing optimally?
(d) What is the princess’s expected utility if she plays the strategy in (c)? Is this strategy optimal for
her? Justify your answer.

Now we instead only assume that 𝑥 > 0.

(d) What is the lowest value for 𝑥 that can guarantee the optimality of the strategy in (c)? Justify your
answer.

Exercise 6. A principal (she) is contracting with an agent (he), whom she wants to hire to work on a
project. The project can turn out to be either a success or a failure. A success generates a value of $50, 000
and a failure generates a value of $25, 000. The agent must choose between two levels of effort: low effort
(𝑒 = 0) or high effort (𝑒 = 1). The distribution of the outcome relies on the effort level and is given in the
table below.

Success Failure
𝑒=0 0.5 0.5
𝑒=1 0.75 0.25

The agent’s utility function is given by

√
𝑢(𝑤, 𝑒) = 𝑤 − 20𝑒,

where 𝑤 > 0 represents the agent’s wage. The principal’s utility function is given by

𝑈 (𝑣, 𝑤) = 𝑣 − 𝑤,

where 𝑣 is the value generated by outcome. The agent’s outside option value (in terms of expected utility)
equals 120, that is, the agent will decline a contract if the maximum expected utility he can derive from
the contract is less than 120, in which case the value to the principal is 0.

148
 (a) What is the principal’s best salary plan (i.e., the plan that maximizes the principal’s expected utility)
if the effort level is observable (so the principal can make the agent’s wage conditional on the effort
exerted)?

From now on suppose that the effort level is NOT observable to the principal and so she can only specify
the agent’s wage for each possible outcome.

(b) If the principal offers a flat salary of $14, 000 (called package 1) to the agent, how will the agent
behave?
(c) Consider the following package (called package 2): The agent pays the principal $15, 000 and keeps
all the value generated by the project as his wage. If the principal offers package 2 to the agent, how
will the agent behave?
(d) Consider a third package (called package 3): The principal pays the agent a flat salary of $12, 100,
and ther is a bonus of $16, 800 to the agent if the project succeeds. If the principal offers package 3
to the agent, how will the agent behave?
(e) Find from among 1,2, and 3 the best package for the principal and compute the principal’s maximum
level of expected utility.

Exercise 7. An insurance company sells car insurance to a Tom. Tom’s Bernoulli utility function is
√
𝑢(𝑤) = 𝑤 − 𝑐, 𝑤 being his final wealth level and 𝑐 being the cost ever incurred. Tom’s initial wealth
level 𝑤 0 = 10, 000. It will cost Tom 1, 900 to fix the car if a traffic accident occurs. The odds of a traffic
accident depends on Tom’s attention. In particular, the odds 𝑝 𝑁 = 0.4 if Tom does not pay attention,
while the odds are reduced to 𝑝 𝐴 = 0.1 if Tom drives carefully. Paying attention is costly to Tom: the cost
is 2 if he drives carefully, and 0 otherwise. For simplicity, we assume that the insurance is one-shot, that
is, it covers at most one accident.

(a) If there was no insurance available, what would be the best choice (i.e., pay attention or not when
driving) of Tom?
(b) An insurance plan is called a full insurance if it covers all losses when an accident occurs. What
will be Tom’s best choice (pay attention or not) if he has a full insurance? Justify your answer.
(c) What is the highest price Tom is willing to pay for a fully insurance if he has to choose between full
insurance and no insurance?
(d) An insurance plan is said to be partial if it only covers a fraction 𝜃 of the loss, that is, it repays 𝜃𝑐 to
the insurance holder if the actual loss if 𝑐, where 0 < 𝜃 < 1. Identify the highest value of 𝜃 that will
induce Tom to pay attention, if the price of the partial insurance is 199.

***********

Exercise 8. Consider an auction with two bidders, 𝐴 and 𝐵, whose values of the auctioned item are
independent and private. In particular, for each 𝑖 = 𝐴, 𝐵, the value 𝑉𝑖 follows the distribution

149
 𝑉𝑖 4 8
Probability 0.6 0.4

Each bidder submits a sealed bid. The highest bidder wins and pays his/her own bid. If the two bids are
the same, then each bidder wins the item with probability 0.5.

(a) Fill in the blank: This is a price sealed bid auction (enter either “first” or “second”).
(b) If 𝐵 always bids 8 and 𝐴 always bids 60% of 𝑉 𝐴, what is 𝐴’s expected utility? What is 𝐵’s expected
utility when 𝑉𝐵 = 4?
(c) Argue that if 𝐴 always bids 𝑉 𝐴, then it is not optimal for 𝐵 to always bid 𝑉𝐵 .

Now consider an alternative version of the auction. Assume now that the highest bidder wins and pays the
loser’s bid. Everything else is the same as before.

(d) Fill in the blank: This is a price sealed bid auction (enter either “first” or “second”).
(e) Compute the expected revenue of the auction.

Exercise 9. Eric would like to buy a used car from Ann. The car is either of high quality (𝐻) or of low
quality (𝐿). Ann knows the quality of her car but Eric does not. Instead, according to reliable statistics, a
car in similar conditions is 𝐿 with probability 𝑞 (and 𝐻 with probability 1 − 𝑞). The worth (in RMB) of
different types of cars to Eric and Ann, denoted by 𝑊𝐸 and 𝑊 𝐴, respectively, is given in the table below.

𝑊𝐸 𝑊𝐴
𝐻 60, 000 48, 000
𝐿 40, 000 32, 000

Ann’s utility is 𝑝 − 𝑊 𝐴 and Eric’s utility is 𝑊𝐸 − 𝑝 if Eric buys the car at price 𝑝, while both equal zero if
Eric and Ann fail to make a deal. There is a trustworthy certifier that can potentially mitigate the adverse
selection problem. The certifier can reveal the quality of a car truthfully and publicly, and it charges 500
RMB for the service.
The trade between Eric and Ann is a sequential game. Ann first decides whether to have her car
certified and then sets a price 𝑝, after which Eric decides whether to buy the car (at price 𝑝) or not (in
which case they fail to make a deal). Ann gets to pay the price of certification if she decides to certify (and
so her utility will be subtracted by 500).

(a) If Ann’s car is of type 𝐻 and she has had it certified, then what is the optimal price she should set?
(b) If for some (irrelevant) reason Ann forgets to certify, which Eric knows, and 𝑞 = 0.5, what is the
optimal price Ann should set? Given this scenario (i.e., Ann forgets to certify), what is the highest
value for 𝑞 such that Eric and Ann can always make a deal (no matter the quality)?
(c) In the Nash equilibrium wherein Eric regards a car as of type 𝐿 if he does not see a certificate, will
Ann have her car certified? What price will Ann set?

150
Now suppose that the certifier only provides service upon buyers’ requests, and so the trade game between
Eric and Ann becomes the following: Eric first decides whether to have the car certified. After seeing the
certificate (if any), Ann sets a price 𝑝 and then Eric decides whether to buy (at price 𝑝) or not (fail to
make a deal). Eric gets to pay the price for certification if he decides to certify (and his utility will be
subtracted by 500).

(d) Given that Ann will always set her price optimally, what will Eric’s expected utility be if he chooses
to certify?
(e) Will Eric find it optimal to certify? If yes, briefly state your logic; if no, find a strategy for Eric that
will always do a better job.

Exercise 10. Suppose that there are 100 students in a class. 50 would each be willing to pay 100 RMB for
high-quality lecture notes, and the other 50 would only be willing to pay 20 RMB each. The instructor can
produce such notes at a total cost of 𝐶 RMB (maybe due to psychological pain and suffering). However,
those notes, once produced or purchased, can be cheaply and easily shared.

(a) Fill in the blank: In this scenario, “not to produce the notes” is an efficient outcome if and only if
𝐶> .

The instructor knows the importance of the notes and thus tries to figure out a way to realize the efficient
outcome. Also suppose that the high-quality notes are not too expensive (i.e., 𝐶 is lower than your answer
to (a)).

(b) Describe a simple way to achieve an efficient outcome if the instructor knows precisely every
student’s willingness to pay.

From now on, suppose that the instructor does not know the willingness to pay of any specific student (but
he does know that there are 50 students each being willing to pay 100 RMB and 50 students each being
willing to pay 20 RMB).

(c) If 𝐶 = 2000, describe a simple way to realize an efficient outcome.

(d) If 𝐶 = 4000, and the instructor employs the following way to extract students’ private information:
Each student bids a price simultaneously. If the sum of the bids is greater than 4000 then each
student is required to pay the price he/she bid and the instructor will produce the notes and share
with every student. Argue that there is a Nash equilibrium of this game in which every high-value
student (those who are willing to pay 100 RMB) has a surplus of 40 and the outcome is efficient.

Exercise 11. Suppose that you and your roommate are considering buying a TV for your apartment. The
TV costs $500. You are willing to pay at most 𝑣 1 and your roommate is willing to pay at most 𝑣 2 , where
𝑣 1 , 𝑣 2 < 500. Each of you knows his own willingness-to-pay but not the other’s. If the TV is not bought,
then each gets zero utility, otherwise one’s utility equals the difference between his willingness-to-pay and
the price he pays.

151
 (a) When is it efficient to buy the TV (i.e., buying the TV will maximize the sum of your utility and
your roommate’s utility)? Your answer should be a condition about 𝑣 1 and 𝑣 2 .
(b) Suppose the way to split the cost of TV is that each pays $250 (i.e., the cost is shared evenly between
you and your roommate) once the TV is bought. If both agree on the split rule then the TV will be
bought, otherwise the TV is not bought. Argue that this will guarantee that the outcome is efficient
if the TV is bought.
(c) Will the split rule in (b) always lead to an efficient outcome? If yes, carefully justify your answer; if
no, identify all situations in which the outcome will not be efficient.

Now consider an alternative situation. Assume that 𝑣 1 and 𝑣 2 are commonly known by you and your
roommate. Consider an alternative mechanism (or game): Your roommate first states the price he is
willing to pay (denote by 𝑝 2 ), and then you decide whether to buy the TV (yes) or not (no). If you say yes,
then your roommate pays 𝑝 2 and you pay max{0, 500 − 𝑝 2 } (i.e., you pay the price that is needed to cover
the total cost if 𝑝 2 < 500), and the TV is bought. If, instead, you say no, then nobody pays anything and
the TV is not bought.

(d) What is your best reaction if your roommate has announced 𝑝 2 ?

(e) Argue that in each Nash equilibrium of this game you will always pay 𝑣 1 if the TV is bought.
(f) Can this game always lead to an efficient outcome? If yes, carefully justify your answer; if no,
identify all situations in which the outcome will not be efficient.

Exercise 12. Consider a “society” consisting of two agents, Sarah and Ben. Sarah holds an antique that
is valuable to both her and the other agent, Ben. The value of the antique to Sarah is 𝑣 𝑆 > 0 and that
to Ben is 𝑣 𝐵 > 0, where 𝑣 𝑆 and 𝑣 𝐵 are independent. Each of them knows his/her own valuation with
certainty, but does not know the other’s value. Suppose the lowest possible value for 𝑣 𝐵 is 𝑣 𝐵 and the
highest possible value for 𝑣 𝑆 is 𝑣 𝑆 . Ben reaches out to Sarah and tells her that he is interested in buying
the antique. If they make a deal at price 𝑝, then Ben receives a payoff of 𝑣 𝐵 − 𝑝 and Sarah receives a
payoff of 𝑝 − 𝑣 𝑆 . Each receives a zero utility if there is no trade.

(a) If 𝑣 𝐵 > 𝑣 𝑆 , what is the socially efficient allocation of the antique? Explain.
(b) Describe a trade mechanism (or game) between the two agents that can guarantee that the outcome
is always socially efficient, given 𝑣 𝐵 > 𝑣 𝑆 .

Consider the following trading mechanism: Each of Sarah and Ben secretly submits a price to a third
party. Denote the two prices by 𝑝 𝑆 and 𝑝 𝐵 , respectively. If 𝑝 𝐵 > 𝑝 𝑆 , then Ben pays Sarah 𝑝 𝑆 to get the
antique; otherwise there is no trade, Sarah keeps the antique, and Ben pays nothing.

(c) Interpret the meaning of the two prices.

(d) Argue that a weakly dominant strategy for Ben is to submit 𝑝 𝐵 = 𝑣 𝐵 .
(e) Is it optimal for Sarah to submit 𝑝 𝑆 = 𝑣 𝑆 ? Justify your answer.

152
(f) Will the trading mechanism guarantee an efficient outcome? If yes, carefully justify your answer, if
no, find an example in which the mechanism will not achieve an efficient outcome.

153
Chapter 7 Answer Key

1. (a) Let the price of insurance be 𝑝. Then each member of group 1 obtains an expected utility of
(144 − 𝑝) 0.5 , while his/her expected utility will be 0.5 · 1440.5 + 0.5(144 − 44) 0.5 = 11 if he/she
does not buy the insurance. For the insurance to be worth buying, we must have

(144 − 𝑝) 0.5 > 11,

and so 𝑝 6 23. Thus, each member from group 1 will be willing to pay at most 23 to insure
against the loss. For each member of group 2, since the expected utility without insurance is
0.9 · 1440.5 + 0.1(144 − 44) 0.5 = 11.8, for him/her to prefer the insurance, we must have

(144 − 𝑝) 0.5 > 11.8,

from which we solve 𝑝 6 4.76. Thus, the highest price a member in group 2 is willing to pay is
4.76.
(b) To provide insurance for members of group 2, the price 𝑝 must be no higher than 4.76, as we
have computed in (a). Let 𝜃 ∈ [0, 1] be the share of people in group 1. Then the company’s
expected profit at price 𝑝 equals 𝑝 − [𝜃 (0.5 × 44) + (1 − 𝜃) (0.1 × 44)]. Using the zero-profit
condition, we have
𝑝 − [𝜃 (0.5 × 44) + (1 − 𝜃) (0.1 × 44)] = 0.

Therefore, we deduce that

𝑝 − 4.4 4.76 − 4.4

𝜃= 6 ≈ 2%.
17.6 17.6
That says, in order to make a private company provide insurance for all members of group 2 for
a zero profit, the share of group 1 should be no higher than 2%.

2. (a) The maximum price a buyer is willing to pay equals the expected value of a car to the buyer.
Without considering the motive of sellers, the distribution of quality should follow the statistics,
and so the expected value of a car to a buyer equals

30% × $30, 000 + 70% × $20, 000 = $23, 000.

(b) Knowing that a buyer is willing to pay at most $23,000, sellers of high-quality cars will exit
the market (as the value of a high-quality car is 80% × $30, 000 = $24, 000). Anticipating this,
buyers will rationally expect that there will not be high-quality cars sold on the market albeit the
statistics, and so each buyer is willing to pay at most $20,000, which is the actual market price,
under which high-quality cars will be crowded out from the market.

154
3. (a) The lowest price a lemon owner can accept is the reservation value of a lemon, that is, $100. At
this price, it is clear that only lemon owners will sell their cars to you. Thus, the total price you
will pay the owners will be 6 × 100 = 600 and the total revenue equals 6 × 500 = 3000. Thus,
the profit will be 3000 − 600 = 2400.
(b) Such price equals the reservation value of a good car, namely. $1600. At this price, all used
car owners will sell and so the total price you will pay to them equal 10 × 1600 = 16000.
The total revenue, however, equals 6 × 500 + 6 × 3500 = 24000. Thus, the profit will be
24000 − 16000 = 8000.
(c) Comparing the results in (a) and (b), we can conclude that the profit-maximizing price that
should be offered is $1600.
(d) Without the service, the maximum profit you can earn, as computed in (b), is 8000. With the
service, you can calibrate the price according to the real quality of a car when buying it. In
particular, now it is optimal to offer $100 to a lemon owner and $1600 to a good car owner. Thus,
the service will increase the profit by the money saved on buying lemons: 6(1600 − 100) = 8000.
Therefore, you are willing to pay at most $8000 to get the service.

4. (a) Since the gross return from a safe project is lower than the gross interest rate of borrowing (i.e.,
1.5), it is clear that borrowers with safe projects will refrain from borrowing. For people with
risky projects, however, the expected return equals 0.5 × 4 + 0.5 × 0 = 2, which dominates the
interest rate. Therefore, only the latter group of borrowers has a borrowing demand, the market
demand being 10.
If 𝑅 = 1.3, then the interest rate is lower than the gross return from a safe project. Therefore, all
people in the market now have borrowing demand, and quantity of demand being 100.
(b) Notice that the expected profit (or net return) from lending to a borrower with a safe project
equals 𝑅 − 1, while that to one with a risky project equals 0.5(𝑅 − 1), given that a borrowing
happens. Therefore, for 𝑅 < 1.5 (in other words, 𝑅 6 1.45), the highest possible expected profit
equals 70(𝑅 − 1), which is no more than 70 × 0.45 = 31.5. For 𝑅 = 1.5, the lowest possible
profit equals 10 × (𝑅 − 1) × 0.5 + 60 × (𝑅 − 1) = 32.5. For 2 > 𝑅 > 1.5, the highest possible
profit equals 10 × (𝑅 − 1) × 0.5 6 5. For 𝑅 > 2, the profit will be zero. Thus, it is clear that the
expected profit-maximizing 𝑅 equals 1.5.
(c) First, for both types of agents to borrow, each must enjoy a nonnegative expected payoff from
borrowing, and so we must have

0.5(4 − 𝑅𝑟 ) + 0.5 × 0 6 0
𝑅𝑠 − 1.5 6 0

and so

𝑅𝑟 6 4
𝑅𝑠 6 1.5

155
 Second, for each type of agents not to apply the interest rate targeted on the other type, we must
have

0.5(4 − 𝑅𝑟 ) + 0.5 × 0 > 0.6[0.5(4 − 𝑅𝑠 ) + 0.5 × 0] + 0.4 × 0

0.6(1.5 − 𝑅𝑠 ) + 0.4 × 0 > 1.5 − 𝑅𝑟

which can be simplified to

3 6 5𝑅𝑟 − 3𝑅𝑠 6 8.

All (𝑅𝑟 , 𝑅𝑠 )’s that satisfy the conditions are illustrated in Figure 22 by the grey area.

𝑅𝑠 5𝑅𝑟 − 3𝑅𝑠 = 3 5𝑅𝑟 − 3𝑅𝑠 = 8

1.5

𝑅𝑟
𝑂 0.6 1.6 4

Figure 22: All Admissible Interest Rate Plans (Grey Area)

5. (a) The best strategy of the princess is to kiss a frog who is able to say “prince”. This is because
a prince is perfectly signaled by the ability to say “prince”, according to the belief. Since the
princess prefers to kiss a frog who is actually a prince, she should kiss a frog who is able to say
“prince”.
(b) The belief is not correct, or not self-confirming. This is because real frogs will also find taking
costly lessons and saying “prince” more favorable, since the payoff to a real frog will be 0
without taking the lesson while that will be 10 − 8 = 2 otherwise. Therefore, the belief in (a) is
not consistent with the real play of the game.
(c) As we have analyzed before, the best reaction to the princess’s strategy for princes is to say
“prince” and that for real frogs is to take costly lessons and say “prince”. Therefore, the
consequence of the princess’s strategy is to have all frogs say “prince”, and so the probability
that she turns out to have kissed a real frog equals 0.9 since 90% of the frogs are real ones.
(d) With probability 0.9 she will have kissed a real frog, and with probability 0.1 she will have

156
 turned a frog back to a prince. Thus, her expected utility

𝐸 [𝑢] = 0.9(−10) + 0.1(100) = 1.

This strategy is clearly suboptimal, since by choosing to eat all frogs her expected utility will be
5, which is higher than the expected utility the strategy in (c) yields.
(e) We then need taking lessons to be sufficiently costly to real frogs so they will not do that and say
“prince”, which then validate the princess’s strategy. Given the princess’s strategy, the payoff to
a real frog will be 10 − 𝑥 if he takes lessons and 0 otherwise. To deter real frogs from being
able to say “prince”, we thus need 10 − 𝑥 6 0, and so 𝑥 > 10. In other word, to guarantee the
optimality of the strategy in (c), 𝑥 should be at least 10.

6. (a) In order to incentivize the agent to exert high effort, the principal gets to pay a salary of 𝑤 such
√
that 𝑤 − 20 > 120, and so 𝑤 > 19, 600. If, however, the target is to induce the agent to exert
√
low effort, then principal gets to pay a salary of 𝑤 such that 𝑤 > 120, and so 𝑤 > 14, 400.
Therefore, the highest expected utility to the principal given that the high effort is incentivized
(notice that since the agent is risk-averse, it is not optimal to make the salary dependent on the
outcome) is
0.75 × 50, 000 + 0.25 × 25, 000 − 19, 600 = 24, 150,

while that will be

0.5 × 50, 000 + 0.5 × 25, 000 − 14, 400 = 23, 100.

given that the low effort is incentivized. Therefore, the best salary plan with observable effort
level is to award high effort by 140 and low effort by 0.
(b) A flat salary will not provide any incentive for costly action, and so the CEO will exert low effort
√
if he ever accepts the contract, which gives the CEO an (expected) utility of 14, 000 ≈ 118.32.
Since this is lower than the CEO’s outside option value (120), the CEO will actually decline the
contract.
(c) If the CEO ever accepts the contract, his expected utility equals
√︁ √︁
0.5 50, 000 − 15, 000 + 0.5 25, 000 − 15, 000 ≈ 143.54

if he exerts low effort, and equals

√︁ √︁
0.75 50, 000 − 15, 000 + 0.25 25, 000 − 15, 000 − 20 ≈ 145.31

if he exerts high effort. Thus, the CEO will accepts the contract and exert high effort (as the
maximum expected utility he can derive from the contract is higher than his outside option value
120).

157
 (d) If the CEO ever accepts the contract, his expected utility equals
√︁ √︁
0.5 12, 100 + 16, 800 + 0.5 12, 100 ≈ 140

if he exerts low effort, and equals

√︁ √︁
0.75 12, 100 + 16, 800 + 0.25 12, 100 − 20 ≈ 135

if he exerts high effort. Thus, the CEO will accepts the contract and exert low effort (as the
maximum expected utility he can derive from the contract is higher than his outside option value
120).
(e) The principal’s expected utility is 0 if she offers package 1; is 15, 000 if she offers package 2; and
is 0.5(50, 000 − 12, 100 − 16, 800) + 0.5(25, 000 − 12, 100) = 17, 000. Therefore, the optimal
contract for the principal is package 3. Note that the principal is worse off than with observable
actions (the situation in (a)) due to moral hazard.

7. (a) If there was no insurance available, Tom’s expected utility would be

√︁ √︁
𝐸 𝑁 [𝑢] = 0.6 10, 000 + 0.4 10, 000 − 1, 900 = 96

if he did not pay attention, and that will be

√︁ √︁
𝐸 𝐴 [𝑢] = 0.9 10, 000 + 0.1 10, 000 − 1, 900 − 2 = 97,

if he paid attention. Therefore, it would be optimal for Tom to pay attention when no insurance
was available.
(b) Given that Tom has a full insurance, his expected utility from the final wealth always equals
√︁
10, 000 − 𝑝 no matter he pays attention or not, where 𝑝 is the price of the insurance. Since
paying attention is costly, it is clear that Tom will optimally choose not to pay attention.
(c) If the price of a full insurance is 𝑝, then Tom’s expected utility will be (recall that he will not
√︁
pay attention provided a full insurance) 10, 000 − 𝑝 if he buys the insurance, and that will be
𝐸 𝐴 [𝑢] = 97 if he chooses no insurance, according to our conclusion in (a). For Tom to find it
optimal to buy the full insurance, we hence must have
√︁
10, 000 − 𝑝 > 97,

from which we solve 𝑝 6 591. That says, Tom is willing to pay at most 591 for the full insurance.
(d) Given a 𝜃-partial insurance, Tom’s best choice is to pay attention if and only if
√︁ √︁
0.9 10, 000 − 199 + 0.1 10, 000 − 199 − 1900(1 − 𝜃) − 2
√︁ √︁
>0.6 10, 000 − 199 + 0.4 10, 000 − 199 − 1900(1 − 𝜃)

158
 Solving the inequality we obtain
281
𝜃6 ≈ 33%.
855
Thus, for such a partial insurance to induce Tom to pay attention, its coverage should be no
higher than (roughly) 33%.

8. (a) Since the highest bidder wins and pays the his/her own bid, this is by definition a first price
sealed bid auction.
(b) If 𝐵 always bids 8, 𝐴 never wins if 𝐴 always bids 60% of the true value since 𝑉 𝐴 6 8. Thus, the
expected utility of 𝐴 equals 0 under this strategy profile. Since 𝐵 will win for sure, the expected
utility of 𝐵 when 𝑉𝐵 = 4 equals 4 − 8 = −4.
(c) Notice that in a first-price auction with private values one’s expected payoff is always zero if
bidding truthfully. Thus, if 𝐴 always bids 𝑉 𝐴, then 𝐵 can obtain a strictly positive expected
utility by bidding less than 𝑉𝐵 . For example, if 𝐵 always bids 60% of 𝑉𝐵 and 𝐴 always bids 𝑉 𝐴,
then 𝐵’s expected utility will be

0.6(8 − 0.6 × 8) + 0.4 × 0 = 1.92 > 0

if 𝑉𝐵 = 8. Thus, always bidding 𝑉𝐵 is suboptimal if 𝐴 always bids 𝑉 𝐴.

(d) Since the highest bidder wins and pays the loser’s bid while the loser’s bid is effectively the
second-highest bid when there are only two bidders, this is a second price sealed bid auction.
(e) In the second-price sealed bid auction, as is well-known, each bidder will bid his/her true value
and the second-highest value is paid eventually. So the expected revenue equals the expected
value of the second highest value, i.e., max{𝑉 𝐴, 𝑉𝐵 }. Since max{𝑉 𝐴, 𝑉𝐵 } = 8 if and only if
𝑉 𝐴 = 𝑉𝐵 = 8, which happens with probability 0.4 × 0.4 = 0.16 (using independence), and the
probability for max{𝑉 𝐴, 𝑉𝐵 } = 4 equals 1 − 0.16 = 0.84. Thus, the expected revenue equals

0.16 × 8 + 0.84 × 4 = 4.64.

9. (a) Since both Eric and Ann know for sure that the car is of type 𝐻, Eric will be willing to pay at
most 60, 000 (i.e., the worth of such a car to him). Thus, the optimal price Ann should set is
60, 000.
(b) Since Eric does not know precisely the quality of the car, he form an expected value of car based
on the statistics, that is,

𝐸𝑊 = 60, 000 × 0.5 + 40, 000 × 0.5 = 50, 000.

Thus, he will decline all trade that charges him more than 50, 000, and so the optimal price for
Ann is 50, 000. Notice that this is the optimal price for Ann no matter the quality of the car
because the price is higher than her reservation value in each case.

159
 For there to be always a trade, the expected value of Eric must be higher than Ann’s reservation
value even when her car is of type 𝐻. This amounts to requiring

60, 000(1 − 𝑞) + 40, 000𝑞 > 48, 000,

from which we solve 𝑞 6 0.6. Thus, in this scenario, the highest value for 𝑞 such that Eric and
Ann can always make deal is 0.6.
(c) Ann will certify her car if and only if the car is of type 𝐻. When the car is certified, she will set
the price 𝑝 = 60, 000; when not certified, her price will be 𝑝 = 40, 000. To see this, notice that
if the car is of type 𝐻, then the highest price Eric is happy to pay will be 40, 000 if there is no
certification, in which case Ann’s payoff will be 40, 000 − 32, 000 = 8000, while by certifying
she can make Eric pay 60, 000 and receive payoff 60, 000 − 500 − 48, 000 = 11, 500, which is
clearly better than not certifying. To the contrary, when the car is of type 𝐿, the price she can
charge will be 40, 000 no matter she has the car certified or not. Since certification is costly, the
best choice is not to certify if the car is of type 𝐿.
(d) Once certified, the quality of the car will be publicly known, and so Ann will charge 40, 000 for
type 𝐿 and 60, 000 for type 𝐻, in either case Eric’s expected payoff will be −500.
(e) It is clearly suboptimal for Eric to certify. As we have shown in (d), the payoff of doing so
is −500. But by choosing not to certify and committing to paying no more than 40, 000 his
expected payoff is nonnegative, which, albeit suboptimal in general, dominates the strategy of
certifying (there are many strategies that also dominate the strategy of certifying and it suffices
to show one of them).

10. (a) The total social value of the notes is 100 × 50 + 20 × 50 = 6000. Thus, “not to produce the notes”
is efficient if and only if the marginal cost exceeds the social value, that is, 𝐶 > 6000.
(b) The simplest way is to ask each student to pay a fixed fraction 𝑟 of his/her willingness-to-pay so
that 5000𝑟 + 1000𝑟 = 𝐶. That is, 𝑟 = 𝐶/6000. Since 𝐶 < 6000, 𝑟 ∈ (0, 1) and this is feasible
(i.e., the total money is enough to cover the cost while each student pays no more than his/her
willingness-to-pay).
(c) Since 𝐶 = 2, 000 while there are 100 students, each being willing to pay at least 20 RMB, so the
simplest way is to tax every student 20 RMB, which will be enough to cover the cost. Note that
Eric cannot "price discriminate" since he does not know precisely the willingness-to-pay for a
student (a student who is willing to pay 100 RMB does not look different from one who is only
willing to pay 20 RMB) by assumption.
(d) Note that the following strategy profile is a Nash equilibrium: each high-value student (i.e.,
student who is willing to pay 100 RMB) bids 60 RMB and each low-value student (i.e., student
who is willing to pay 20 RMB only) bids 20 RMB. To see this, notice that for a high-value
student, given other’s strategy, if he/she bids less than 60 RMB, the sum of bids will be less than
4000 RMB and the notes will be not be supplied, yielding to him/her a zero utility; if he/she bids
exactly 60 RMB, his/her utility/surplus will be 40 (as he/she is willing to pay 100 RMB but

160
 only pays 60 RMB) since the sum of bids will be exactly equal to 4000; if he/she bids more
than 60 RMB, the notes will be produced as well, but his/her surplus will be less than 40 as
he/she gets to pay the bid. Similar analysis applies to low-value students. Clearly, the outcome
is efficient (i.e., the notes are produced and everyone has it, while no one pays more than his/her
willingness-to-pay and the total revenue can cover the cost).

11. (a) It will be efficient to buy the TV if the total willingness to pay dominates the cost, that is ,
𝑣 1 + 𝑣 2 > 500.
(b) Clearly, we must have 𝑣 1 > 250 and 𝑣 2 > 250 if the TV is bought, since otherwise at least one
of you and your roommate will not agree on the split rule. Thus, 𝑣 1 + 𝑣 2 > 250 + 250 = 500 if
the TV is bought, which is an efficient outcome according to our conclusion in (a).
(c) No. To see this, imagine the case in which 𝑣 1 = 400 and 𝑣 2 = 200. In this case your roommate
will not be happy to pay $250 since it exceeds his willingness-to-pay, and so the TV will not be
bought. However, we have 𝑣 1 + 𝑣 2 = 600 > 500 in this case, and the corresponding efficient
outcome is to buy the TV. Thus, the split rule cannot always yield the efficient outcome.
(d) If you say yes, then the TV will be bought, and you will have to pay max{0, 500 − 𝑝 2 } and
your utility will be 𝑣 1 − max{0, 500 − 𝑝 2 }. Therefore, it is optimal to say yes if and only if
𝑣 1 > max{0, 500 − 𝑝 2 }. The best reaction is illustrated in Figure 23 below. For all pairs of
( 𝑝 2 , 𝑣 1 ) that are located in the red region, your best reaction is no, while for all pairs that are
located in the blue region, your best reaction is yes.

𝑣1

500
Yes

No

𝑝2
𝑂 500

Figure 23: An Illustration of Your Best Reaction

(e) Your roommate knows that you are willing to pay at most 𝑣 1 , which means that he gets to pay at
least 500 − 𝑣 1 if he wants TV. Since he always wants to minimize his payment, he will always
state the price 𝑝 2 = 500 − 𝑣 1 if 𝑣 2 > 500 − 𝑣 1 , in which case the TV will be bought; and he
will state some arbitrary price 𝑝 2 < 500 − 𝑣 1 so that the TV is not bought when 𝑣 2 < 500 − 𝑣 1 .

161
 Given this and your best reaction as identified in (d), you will pay 500 − (500 − 𝑣 1 ) = 𝑣 1 if the
TV is bought (i.e., you say yes) in each Nash equilibrium of the game.
(f) This game always leads to an efficient outcome because you always pay 𝑣 1 if the TV is bought,
in which case your roommate gets to pay 500 − 𝑣 1 , and he is willing to pay that price if and only
if 𝑣 2 > 500 − 𝑣 1 . That says, the TV is bought if and only if 𝑣 1 + 𝑣 2 > 500, which implies that
the outcome is always efficient.

12. (a) The condition means that Ben always has a higher value than Sarah. Thus, it is socially optimal
to allocate the antique to Ben.
(b) The most straightforward trade mechanism is the one with a fixed price. That is, for some
arbitrary 𝑝 ∈ (𝑣 𝑠 , 𝑣 𝐵 ), simply ask whether Sarah is happy to sell the antique to Ben at the price
𝑝 and whether Ben is happy to buy the antique from Ben at 𝑝. If either of the two disagrees
then the game is over and no deal is made, otherwise Ben pays Sarah 𝑝 and gets the antique.
Importantly, the game is played only once. Neither can refuse now and come back to renegotiate
in the future. Then it is obvious that both sides will accept the deal (recall that we have 𝑣 𝐵 > 𝑣 𝑆 ),
and so the outcome is always socially efficient.
(c) Sarah should consider 𝑝 𝑆 to be the lowest price at which she is willing to sell the antique, while
Ben should consider 𝑝 𝐵 as the highest price at which he is willing to buy the antique.
(d) First, it should be clear that Ben should never overstate his willingness-to-pay: if he does
overstate, then his payoff will be negative if 𝑝 𝐵 > 𝑝 𝑆 and 0 otherwise, and so his expected
payoff will be nonpositive. However, his payoff is nonnegative if he always submits 𝑝 𝐵 = 𝑣 𝐵 as
trade only happens when 𝑝 𝐵 > 𝑝 𝑆 , in which case his payoff is 𝑣 𝐵 − 𝑝 𝑆 = 𝑝 𝐵 − 𝑝 𝑆 > 0. Second,
we argue that understating is also suboptimal. To see this, let us fix an arbitrary price 𝑝 𝑆 for
Sarah. If 𝑝 𝑆 > 𝑣 𝐵 , Ben’s payoff will be the same (0) no matter he submits 𝑝 𝐵 = 𝑣 𝐵 or 𝑝 𝐵 < 𝑣 𝐵 ;
if 𝑝 𝑆 < 𝑣 𝐵 , then submitting 𝑝 𝐵 = 𝑣 𝐵 or any 𝑝 𝐵 ∈ [ 𝑝 𝑆 , 𝑣 𝐵 ) makes no difference (Ben’s payoff
equals 𝑣 𝐵 − 𝑝 𝑆 > 0), but submitting some 𝑝 𝐵 < 𝑝 𝑆 is suboptimal as there will be no deal, which
leaves to Ben with a zero payoff. As a result, submitting 𝑝 𝐵 = 𝑣 𝐵 is a best choice in all cases, it
must be a weakly dominant strategy for Ben.
(e) Sarah will not find it optimal to submit 𝑝 𝑆 = 𝑣 𝑆 . It is clear that this strategy guarantees
a zero payoff to Sarah, while her expected utility could be positive if she overstates her
willingness-to-sell.
(f) No. For example, assume that 𝑣 𝑆 = 3 for sure and 𝑣 𝐵 is identically distributed by

𝑣𝐵 4 12
Probability 0.6 0.4
Then, it is always optimal to allocate the antique to Ben. However, given that Ben will truthfully
state his value (which is weakly dominant as we know), Sarah’s best strategy is to submit
𝑝 𝑆 = 12. Indeed, by submitting 𝑝 𝑆 6 3 or 𝑝 𝑆 > 8 her expected payoff will be at most zero; by
submitting 𝑝 𝑆 ∈ (3, 4] her expected payoff will be 𝑝 𝑆 , the highest being 4 (achieved at 4), while

162
by submitting 𝑝 𝑆 ∈ (4, 8] her expected payoff will be 0.4𝑝 𝑆 (achieved at 12), the highest being
4.8. Thus, when 𝑣 𝐵 = 4, there will be no trade, which is inefficient.

163